Riscos Fiscais UK - 2019

Fiscal risks report
July 2019
Office for Budget Responsibility
Fiscal risks report
Presented to Parliament by the

Exchequer Secretary to the Treasury
by Command of Her Majesty
July 2019
CP 131
© Crown copyright 2019
This publication is licensed under the terms of the Open Government

Licence v3.0 except where otherwise stated. To view this licence, visit
nationalarchives.gov.uk/doc/open-government-licence/version/3
Where we have identified any third party copyright information you will
need to obtain permission from the copyright holders concerned.
This publication is available at www.gov.uk/official-documents
Any enquiries regarding this publication should be sent to us at

obr.enquiries@obr.uk
ISBN 978-1-5286-1497-9
CCS0719538076 07/19
Printed on paper containing 75% recycled fibre content minimum
Printed in the UK by the APS Group on behalf of the Controller of
Her Majesty’s Stationery Office

Contents
Foreword ...................................................................................... 1
Executive summary ........................................................................ 3
Chapter 1 Introduction ................................................................................ 17
Chapter 2 Macroeconomic risks ................................................................... 31
Chapter 3 Risks from the financial sector ...................................................... 57
Chapter 4 Revenue risks .............................................................................. 73
Chapter 5 Primary spending risks ............................................................... 125
Chapter 6 Balance sheet risks .................................................................... 169
Chapter 7 Debt interest risks ...................................................................... 197
Chapter 8 Policy risks ................................................................................ 221
Chapter 9 Climate change ........................................................................ 239
Chapter 10 A fiscal stress test....................................................................... 255
Index of charts and tables .......................................................... 289

Foreword
The Office for Budget Responsibility (OBR) was established in 2010 to provide independent and
authoritative analysis of the UK’s public finances. In the October 2015 update to the Charter for
Budget Responsibility, Parliament required us to produce a fiscal risks report at least once every two
years. The Government has committed to responding formally to each report within a year. We
produced our first Fiscal risks report (FRR) in July 2017. The Government responded to it in
Managing fiscal risks in July 2018. We continue the conversation in this report.
We have always placed considerable emphasis on the risks and uncertainties around any
assessment of the outlook for the public finances. In our Economic and fiscal outlooks (EFOs), we
illustrate the risks to our medium-term forecasts by drawing on the pattern of past forecast errors,
estimates of their sensitivity to changes in key parameters, and scenario analysis. We also subject the
long-term projections in our Fiscal sustainability reports (FSRs) to sensitivity analysis, as well as
highlighting specific fiscal risks revealed by the Whole of Government Accounts.
In our Fiscal risks reports we draw together and expand on these analyses. We hope that this
provides a valuable addition to the material that we produce to help promote an informed public
debate about the sustainability of the public finances. Much of that debate focuses on our central
medium-term forecasts and long-term projections, despite the wide range of uncertainty that
surrounds those central conclusions. By focusing on identifiable risks to the public finances, the FRR
builds on the sensitivity and scenario analysis that we already present in our EFOs and FSRs.
The analysis and conclusions presented in this document represent the collective view of the three
independent members of the OBR’s Budget Responsibility Committee. We take full responsibility for
the judgements that underpin them. We have been hugely supported in this by the staff of the OBR,
to whom we are as usual enormously grateful.
We have also drawn on the help and expertise of officials across numerous departments and
agencies for which we are very grateful. In addition, we have benefitted from insights shared by
experts from outside government who have spoken to us about the wide range of topics we cover. In
particular, we would like to thank Professor Nicholas Barr, Anita Charlesworth, Adam Corlett, Stuart
Adam, Richard Hughes, Professor Andrew Scott, Sir Paul Tucker and Dimitri Zenghelis. We are also
grateful to Sarah Breedon and her team at the Bank of England for taking us through their work on
climate-related macroeconomic scenario analysis. Finally, we are grateful to staff at the International
Monetary Fund (IMF) for their assistance in understanding their no-deal, no- transition Brexit
scenarios from the April 2019 World Economic Outlook, which we have built upon to produce the
fiscal stress test in Chapter 10. We would also emphasise that despite that assistance, all
judgements underpinning that stress test are our own and should not be attributed to the IMF.
1 Fiscal risks report

Foreword
We provided the Chancellor of the Exchequer with a summary of our main conclusions on 12 July.
Given the breadth and depth of the report, we provided exceptional pre-release access to a near-
final version of the full report to a named list of Treasury officials on 15 July. We then provided a full
and final copy 24 hours prior to publication. This is in line with pre-release access arrangements set
out in the Memorandum of Understanding between the Office for Budget Responsibility, HM
Treasury, Department for Work and Pensions and HM Revenue & Customs. In accordance with this
MoU, emerging findings and draft material were discussed with officials in the Treasury under the
auspices of a liaison group set up for the purpose. At no point in the process did we come under
any pressure from Ministers, special advisers or officials to alter any of our analysis or conclusions.
We hope that this report is of use and interest to readers. We consider it to be a work-in-progress
that will be refined and modified over time. We would therefore be pleased to receive feedback on
any aspect of the content or presentation of the analysis. This can be sent to feedback@obr.uk.
Robert Chote Sir Charles Bean Andy King
The Budget Responsibility Committee
Fiscal risks report 2

Executive summary
Overview
1 In 2017 we produced our first Fiscal risks report (FRR), an assessment of the shocks and
pressures that could threaten our forecast for the public finances over the medium term and
fiscal sustainability over the longer term. In this, our second report, we revisit and broaden
that assessment in light of recent economic and fiscal developments, including subsequent
policy decisions and the Government’s welcome and substantive response to our first FRR.
2 Many of the fiscal risks we discussed two years ago remain. That is not surprising, as in
many cases the Government can only seek to manage and mitigate them, not to eliminate
them. It is also important to remember that the whole point of much government activity is to
pool or manage risks that confront society. So taking on fiscal risk can be welfare-improving
and attempts to transfer some fiscal risks back to the private sector may simply create
different ones that are less imminent, less transparent and potentially more costly.
3 That said, the Government has taken useful steps since 2017 to improve the monitoring and
management of fiscal risks, including better management of new contingent liabilities, more
transparent reporting on the balance sheet and, thanks to the Office for National Statistics,
plans to address the pervasive fiscal illusions in the way student loans are captured in the
public finances. It has also further reduced the budget deficit and has begun to lower public
sector net debt as a share of GDP. In addition, it is reducing its exposure to inflation
surprises by relying less on issuing index-linked debt to finance government borrowing.
4 But policy risks to the public finances in the medium term are significant and look greater
than they were two years ago. In his recent statements the current Chancellor has all but
abandoned the Government’s legislated objective to balance the budget by the mid-2020s.
And the £27 billion a year NHS settlement announced in June 2018 – unfunded,
unaccompanied by detailed plans for reform and outside the normal timetable for spending
decisions – has cast doubts over the Treasury’s usually firm grip on departmental spending.
5 Medium-term policy decisions will of course depend on the incoming Prime Minister and
Chancellor, rather than on the current incumbents. The remaining Conservative leadership
contenders have made a series of uncosted proposals for tax cuts and spending increases
that would be likely to increase government borrowing by tens of billions of pounds if
implemented. So all the signs point to a fiscal loosening and less ambitious objectives for the
management of the public finances. (The current Chancellor has urged the candidates at
least to commit to keeping net debt falling as a share of GDP, which, all else equal, would
allow additional borrowing of approaching £25 billion a year over the medium term.)

Executive summary
6 To be clear, it is not the role of the OBR to say what the Government’s fiscal targets should
be nor how much budgetary loosening or tightening it should undertake. But it must be
understood that additional tax cuts or spending increases would push government borrowing
and debt up from the levels expected in our forecasts and that there is no war-chest or pot of
money set aside that would make them a free lunch. The Government does have room for
manoeuvre against its ‘fiscal mandate’ for structural borrowing next year, but that does not
provide an anchor for medium term tax and spending decisions. And the headroom is
measured against our central forecast in March – over the next 300 pages we outline some
of the many risks that might raise borrowing and debt relative to those figures.
7 These decisions will of course need to be taken against the unusually uncertain economic
and fiscal backdrop created by different possible outcomes to Brexit. Our March forecast was
conditioned on the UK securing a deal and exiting smoothly. Given the leadership
contenders’ willingness explicitly to countenance a ‘no-deal’ exit on October 31, we use the
‘stress test’ in this FRR to illustrate the potential fiscal impact of a no-deal, no-transition Brexit
scenario set out by the International Monetary Fund (IMF) in its April 2019 World Economic
Outlook. This scenario is not necessarily the most likely outcome and it is relatively benign
compared to some (for example, assuming limited short-term border disruptions). But it still
adds around £30 billion a year to borrowing from 2020-21 onwards and around 12 per
cent of GDP to net debt by 2023-24, compared with our March forecast baseline.
8 A more disruptive or disorderly scenario, closer to the stress test we considered two years
ago, could hit the public finances much harder. (It is important to remember that the
economic and fiscal developments over the past three years – as well our and the IMF’s
baseline forecasts – already incorporate some impact from the referendum vote, although it
is impossible to isolate that from other surprises relative to our pre-vote forecasts. The impact
of Brexit itself – once it happens – would also continue to unfold for many years beyond the
end of the stress test horizon.)
9 As we noted in our previous report, history tells us that the biggest and most frequent fiscal
risks in peacetime relate to the economy – even in the absence of specific shocks like Brexit.
At the time we finalised this report, these risks appeared to be rising – the latest survey data
suggested that growth paused at best in the second quarter. In part, this is likely to be an
unwinding of stockpiling by businesses ahead of the previous proposed Brexit date of 29
March, but a more general weakness may persist and intensify as 31 October nears.
10 While this may not signal that the economy is currently entering a recession, these occur
roughly once a decade in the UK, and are almost always unexpected when they do. Policy
can reduce the likelihood of these risks crystallising and their fiscal impact when they do, but
the underlying risk cannot be eliminated. In its response to our first report the Government
acknowledged the need to seek to reduce debt during more favourable times to ensure that
there is room to let it rise when shocks hit, without interest costs rising to undesirable levels.
11 Looking at specific risks to receipts and spending points to many ongoing pressures that
governments must deal with, while also preparing for inevitable future shocks. There are
long-term pressures on revenue from some tax bases, from trends in smoking, drinking and

Executive summary
car efficiency, and from the digitalisation of economic activity. And policy is always a source
of risk, given repeated decisions to cancel fuel duty increases, although manifesto
commitments on the income tax personal allowance and higher rate threshold have now
been met. The large, rising and poorly understood cost of tax reliefs poses risks too.
12 As ever, there are significant upward pressures on public spending. Over the longer term,
the biggest arise from non-demographic cost pressures in health and social care, and the
impact of an ageing population on them and the state pension. The near-term fiscal risk
from health spending that we flagged in our previous report crystallised with the June 2018
announcement of an NHS settlement more consistent with historical trends in spending. But
long-term pressures remain and this might be seen as just the first down-payment on them.
And the Government has chosen to retain a triple lock on pensions uprating that ratchets
spending higher as a share of GDP – despite acknowledging the fiscal risk that that poses.
13 One issue that we did not address in any detail in our first FRR was climate change, which
has the potential to inflict both sudden shocks and slower-building pressures on the public
finances. Their nature and cost will of course depend hugely on how the climate itself
evolves. If global mitigation efforts fail, the risks posed by conflict, mass migration and
catastrophic weather events could be severe. But if the targets agreed to in Paris in 2015 are
met and there is only modest further warming, the risks would be less severe. In such a
world, and when viewed individually, the fiscal risks associated with climate change do not
appear especially large relative to others we cover in this report. The effects of extreme
weather are not likely to be on the same scale as a major recession – although they might be
more frequent. And the cost of adaptation and mitigation measures will probably not be as
large as those related to ageing or the cost pressures in health care. But such conclusions
might simply reflect the difficulty we all have in foreseeing the full systemic consequences of
significant global warming. We intend to draw on the Bank of England’s forthcoming
scenario analysis to develop our quantitative assessment of climate-related fiscal risks.
14 Throughout this report, we look at the way fiscal risks have evolved partly in the light of the
Treasury’s response to our first report – Managing fiscal risks (MFR) – which it published a
year ago. Several countries produce fiscal risk reports, but the UK is unusual in having it
prepared by an independent body and is, we think, unique in having a Government that has
promised in legislation to respond. We welcome the substantive nature of that response and
hope that over time both our reports and the Treasury’s responses will help improve fiscal
risk management both directly and by facilitating more informed debate and discussion.
Our approach
15 Chapter 1 sets out our approach. Our goal is to identify risks to the outlook for the UK public
finances over two horizons: to our March forecast over the next five years; and to fiscal
sustainability over the next 50. We are interested primarily in ‘downside’ risks that would
make things look worse rather than better. They are a bigger challenge to policymakers.
16 When discussing fiscal risk, it is important to remember that the whole purpose of much
government activity is to pool risks that society has decided would better be taken on by the

Executive summary
state than by individuals. For example: the NHS takes on the risk that people would
otherwise face not knowing when or how they might fall ill; state pensions remove the risk of
pensioners’ incomes falling below a certain level, perhaps due to their living longer than
expected or receiving lower returns on pension savings than anticipated; and universal credit
reduces the financial risks associated with unexpected periods out of work or on low pay.
17 Where possible, we try to evaluate the probability that particular risks will crystallise over the
medium and long term, and the potential impact if they do. Occasionally these can be
estimated with a degree of precision, but more often broad judgements must suffice.
18 Finally, we consider what governments do in light of these risks, either managing and
mitigating them through different means or tolerating them as developments that will
occasionally push the public finances off course. At the end of each chapter we list some
issues that the Government may wish to address in its formal response to this report.
Macroeconomic risks
19 In Chapter 2 we look at various ways in which macroeconomic risks can affect the public
finances. History suggests that these are the high impact risks most likely to crystallise over
the medium and longer term. Two years ago, we identified potential output growth as the
most important long-term macroeconomic risk. We also noted that the chance of a recession
is around one in two over any five-year horizon and virtually certain over 50 years. We also
discussed the fiscal risks posed by fluctuations in the relatively tax-rich housing market, and
by the fact that employment income is more highly taxed than profits. We highlighted that
Brexit-related uncertainties overlay many of these risks.
20 The Government’s MFR response generally agreed with the assessment we provided. It
recognised the importance of long-run productivity growth and listed steps it was taking to try
to raise it. On cyclical shocks it pointed to previous reforms to the monetary and fiscal policy
frameworks that enable macroeconomic policy to play a supportive role in recessions. And it
highlighted specific steps taken by the Ministry of Housing, Communities and Local
Government to manage the credit risk arising from its exposure to the housing sector.
21 As regards Brexit, MFR declared that the UK would seek a “deep and comprehensive
economic partnership” with the EU, to maximise the benefits and minimise the potential
disruption. This would include seeking the establishment by the UK and the EU of a free
trade area for goods and the phased introduction of a facilitated customs arrangement.
22 Our updated assessment is that the majority of medium-term, non-Brexit related risks are
little changed over the past two years in terms of probability and impact. That said:
• Two years ago we assigned a medium likelihood to trend productivity growth

persisting at the lower rates seen post-crisis rather than picking up as we then forecast,
but noted that the latest data at the time might suggest a gloomier prognosis. So it has
proved. In November 2017 we revised productivity growth down further, having
reviewed the many candidate explanations for the post-crisis malaise and decided that

Executive summary
it was not central to assume that their effects would pass over the coming five years.
But there are still considerable risks. We still assume some pick-up, which might not
materialise. Then again, productivity could recover more quickly than we forecast,
though that seems unlikely in the near term if Brexit-related uncertainty persists.
• Short-term cyclical risks appear to have risen this year. The latest data and surveys
suggest the economy flatlined at best in the second quarter. Some of this is likely to be
a ‘pay-back’ after Brexit-related stock building in the first quarter. But surveys were
particularly weak in June, suggesting that the pace of growth is likely to remain weak.
This raises the risk that the economy may be entering a full-blown recession. The fiscal
risks posed by recessions depend on their depth and persistence, the sectors most
deeply affected, and the pace at which the economy subsequently recovers. The fiscal
risk posed by a cyclical downturn is increased by the proximity of Bank Rate to its
effective lower bound and by doubts that some have expressed about the effectiveness
of unconventional monetary policy, both of which suggest that the authorities may rely
more heavily on fiscal stimulus measures than would previously have been the case.
23 We also take a more in-depth look at the fiscal risks from mismeasurement of the output gap
– the extent to which activity in the economy lies above or below the sustainable level
consistent with stable inflation. Uncertainty surrounding the output gap – which would be
particularly acute if Brexit were to disrupt economic activity – means that our central estimate
could provide the Chancellor with a misleading steer as to the size of the structural budget
deficit. In principle this risk should be symmetric, but evidence around recessions is that
estimates can worsen significantly over time as forecasters tend to conclude that they were
over-optimistic ahead of the downturn. To put the size of this risk in context, in real-time the
Treasury, IMF and OECD all estimated output to be close to potential just before the onset of
the 2008-9 recession. Current estimates of the output gap are now in the range 1 to 3 per
cent. This represents an average change in the estimated size of the structural deficit in 2008
of a little over 1 per cent of nominal GDP – around £26 billion in today’s terms.
Financial sector risks

24 In Chapter 3 we consider the fiscal risks associated with the financial sector. Last time we
focused not only on the potential costs of financial crises, but also how the public finances
might be affected if this tax-rich sector were to shrink as a share of the economy. We also
presented a fiscal stress test that modelled a period of synchronised domestic and global
economic weakness coupled with financial market stress. This increased public sector net
debt by 34 per cent of GDP by the stress test horizon, relative to the baseline forecast.
25 The Government’s MFR response focused on post-crisis changes to the regulatory policy
framework, notably the new macro- and micro-prudential regulators, to reduce financial
stability risks, and steps to reduce taxpayer exposure in the event of an institution failing. In

Executive summary
respect of Brexit and the ‘passporting regime’,1 it pointed to the time afforded by the
Withdrawal Agreement ‘implementation period’ for firms to plan and to steps it had taken to
allow European firms temporary permission to access the UK market in a no-deal Brexit.
26 Given that many of the reforms the Government cited preceded our previous FRR, and with
the Financial Policy Committee’s own assessment of stability risks little changed, our
assessment of the fiscal risks posed by the financial sector is little changed too. So we see a
low, but not negligible, risk of another crisis over the next five years. Over the longer term,
based on the Independent Commission on Banking’s estimate of the historical probability of
crises occurring, we might expect the UK to suffer one around once in every 20 years.
27 One lesson from history is that when risk is suppressed in one part of the financial system, it
often migrates to some other less heavily regulated part. So in this report we take a closer
look at the shadow banking sector and the fiscal risks it might pose. Identifying and
measuring shadow banking activity is not straightforward, but, on the measures available,
the recent rapid growth in shadow banking is somewhat reminiscent of the build-up to the
global financial crisis. While steps have been taken to increase the oversight and regulation
of shadow banking entities, some former policymakers have warned that the present
‘monitor and respond’ approach to shadow banking risks is likely to prove insufficient. For
these and other reasons, we believe it is appropriate to remain cautious in our assessment of
the potential risks from shadow banking. Moreover, experience suggests that a financial
sector that appears to be adequately regulated may turn out not to be so. Vulnerabilities that
appear obvious with hindsight are often much harder to discern in real time.
28 The fiscal risks that a shadow banking-led crisis might pose include calls on the public purse
to bail out institutions and to compensate retail investors, as well as absorbing the fiscal
consequences of any economic fallout. Contagion to the core banking system is possible too.
Revenue risks
29 In Chapter 4 we consider specific risks to revenues. Two years ago, we highlighted several
sources of risk to tax bases (from trends in smoking, drinking and driving, as well as the
depletion of North Sea oil and gas reserves) and to the amount of tax raised from a given
base (from non-compliance and from people choosing self-employment or working for their
own company over employee status). We also discussed tax policy risks (aspirations not yet
factored into forecasts, non-implementation of stated policies, and the disparity between the
relatively certain costs of giveaways and the relatively uncertain yields from measures to
tackle tax avoidance and evasion). And we discussed the increasing concentration of receipts
among a small number of taxpayers (notably for income tax and stamp duty land tax).
30 The Government’s MFR response generally concurred with our assessment of where the risks
lay, but had less to say about its view of those risks or how it manages them. It recognised
1
The ‘passporting’ regime allows firms authorised in one country within the Single Market to sell certain financial products into any other
country within the Single Market. Outside the Single Market, UK regulation would need to be recognised as equivalent by the European
Commission for UK-based firms to continue selling particular types of financial services into the Single Market.

Executive summary
pressures on tax bases, while noting other objectives (for example, the public health gains
from reduced smoking), and set out more concrete steps in respect of managing and
reducing North Sea decommissioning costs. It heralded the declining trend in the measured
‘tax gap’ and highlighted the many measures taken to try to reduce it further, including
through digitalisation. On policy risks, it set out its rationale for policy decisions taken, but
did not address the fiscal risks associated with the approach of stating one policy but
following another. And it had little to say about the concentration of receipts.
31 Our assessment of revenue-related fiscal risks is little changed from two years ago. Long-
term pressures on excise duty tax bases and the North Sea continues. The tax gap has ticked
up, while Brexit is placing greater burdens on HMRC. Fuel duty continues to be frozen year
by year, although manifesto commitments on the income tax personal allowance and higher
rate threshold have now been met. And the concentration of tax receipts remains an issue,
with the large rise in the personal allowance exacerbating the trend in income tax.
32 Our deeper look at the issues of tax reliefs and digitalisation raises some new issues. Tax
expenditures – those reliefs designed to promote another policy objective, like investment in
R&D – cost around 8 per cent of GDP, a cost that has been rising and which receives less
scrutiny than conventional public spending. A modest upside surprise to growth in their cost
could be fiscally material. Risks posed by digitalisation are more balanced, particularly over
a longer horizon. While digitalisation of the economy poses challenges in terms of what can
be taxed and where, the potential gains in terms of tax administration could be material.
Primary spending risks

33 In Chapter 5 we consider risks to primary spending – i.e. everything other than debt interest.
This is spending over which governments have some direct control – for example, via what
they choose to spend on a public service or the way they structure the welfare system.
34 The two largest medium-term spending risks we highlighted two years ago were the
unquantifiable pressure posed by ‘austerity fatigue’ and the more quantifiable shortfall in
prospective health spending growth relative to history. We also noted the declining
proportion of spending subject to departmental expenditure limits (and the spending control
risk that might pose), the risks of local authorities tapping their reserves and their commercial
property investments, and welfare spending risks from ongoing reforms and court cases.
Over the long term, primary spending is the largest risk to fiscal sustainability – thanks to
non-demographic cost pressures in health and social care, and the impact of an ageing
population on them and the state pension. The long-term cost of the triple lock is large too.
35 Before the Government had published MFR it had already announced a multi-year, multi-
billion pound settlement for the NHS that crystallised the largest medium-term risk we had
identified. This was not accompanied by a detailed reform plan explaining how the extra
money would be used or by the announcement of any measures to help pay for it.
36 In MFR the Government noted the risks to health and social care spending, promising a
delayed green paper on the latter by autumn 2018, though in the event it is still pending. Its

Executive summary
approach to the risks from litigation included a strategy to address the rising costs of clinical
negligence, also slated for publication in autumn 2018 and also still pending, and to
improve the management and reporting of legal risks within the welfare system, which has
now been reflected in DWP’s accounts. On the triple lock, it noted the ratchet effect it had on
spending as a share of GDP, but nonetheless committed to retain it “for the rest of this
Parliament”. And with respect to local authorities’ borrowing for commercial purposes, it
pointed to new guidance and that it would “monitor how local authorities respond to the
revised guidance, and take appropriate further action if this is necessary”. Given continued
growth in such investments, the Treasury has told us that further steps are being considered.
37 Risks to our medium-term forecast have probably declined relative to two years ago, largely
because the forecast now includes stronger real growth in health spending. But there has
been little change in many other sources of risk. The spending control framework seems to
be under pressure, with major announcements being made outside fiscal events, and the
Conservative leadership candidates making pledges that would prove expensive if pursued.
38 The likely cost of the BBC’s recent decision to means-test free TV licences for the over-75s by
linking it to pension credit – thereby potentially prompting a material number of those
currently not taking it up to do so – poses a fiscal risk that we had not previously envisaged.
It is unusual for a government to delegate parameters of welfare policy to a broadcasting
company to save money. The unintended consequence is likely to be that the link to pension
credit receipt will raise welfare spending by more than bearing the remaining cost of free
licences reduces BBC spending, so the budget deficit will rise not fall.
39 Other developments include the Government’s aspiration to increase the National Living
Wage further after 2020, and the intention to impose a more restrictive immigration system
after Brexit, both of which would generate additional costs for the health and social care
sectors. By contrast, HMRC’s victory in the landmark Littlewoods tax litigation over the
application of compound interest to repayment of past taxes has removed one major risk.
Balance sheet risks

40 In Chapter 6 we look at risks that could affect the balance sheet directly, via balance sheet
transactions (e.g. lending to the private sector or issuing debt to purchase assets), balance
sheet transfers (when government assumes the liabilities of a private sector entity, either in
the real world or through statistical reclassification) and valuation effects (e.g. the effect of
currency movements on the foreign exchange reserves). In our 2017 report, we noted a wide
range of balance sheet risks, including those arising from: an incomplete understanding of
the balance sheet and the associated fiscal illusions; the management of the government’s
assets and liabilities (including uncertainty about the scale and timing of financial asset
sales); the growing use of government guarantees; and the potential for reclassifications.
41 The Government’s MFR response showed more progress in respect of balance sheet
management than in many other areas. It outlined plans to improve measures of the
balance sheet (all of which have since been delivered or are about to be), steps being taken
to increase transparency around asset sales (with guidance since published) and to manage

Executive summary
guarantees and contingent liabilities (with a new approval regime in place). Its ‘Balance
Sheet Review’ aims to improve the information on assets and liabilities held across
government, identify opportunities for the disposal of assets, improve the return on retained
assets, and reduce associated risks. It also encouraged the ONS to publish clearer
information on the pipeline of classification decisions that flow from its own work plan.
42 Overall, our assessment is that balance sheet risks are largely unchanged, though they are
now better monitored. The greatest risk remains the possibility that the Government will feel
compelled to respond to some unexpected future event or crisis, triggering a step change in
the level of public debt. Fiscal illusions, especially with respect to public sector net debt, and
the incentives they create for policymakers wishing to hide public sector activity off the
balance sheet remain a risk. Changes to the accounting treatment of student loans will
remove a particularly obvious illusion, but the steps taken by the Government to secure the
reclassification of housing associations to the private sector reinstates another.
43 Concrete steps have been taken to improve how the public sector’s assets (including
intangible ones) and liabilities are recorded or understood, which should improve the
Government’s management of them and facilitate our scrutiny of them. We will survey any
examples of this transparency affecting decision-making in future editions of this report.
Debt interest risks

44 In Chapter 7 we consider risks associated with debt interest spending and debt dynamics.
These are affected by the scale and composition of public sector debt – its maturity and the
balance between inflation-linked and conventional government bonds (‘gilts’) – and the
interest rates paid. In our 2017 report, we highlighted the increased medium-term risks since
the crisis arising from a higher debt stock, including a much higher stock of index-linked
gilts, and the increase in the Bank of England Asset Purchase Facility’s gilt holdings, which
are financed by creating reserve deposits on which commercial banks earn only Bank Rate,
so making net payments to the private sector more sensitive to short-term interest rates. We
also highlighted the risk of interest rates rising relative to nominal GDP growth.
45 The Government’s MFR response largely concurred with this assessment. It discussed the
actions being taken to address various risks, including reducing the extent to which the
Government’s financing requirement would be met by issuing index-linked gilts and setting
fiscal policy to ensure that debt continued to fall relative to GDP.
46 Overall the picture is little changed from two years ago, as might be expected with risks to a
stock that is typically slow-moving. The reduction in index-linked gilt issuance has reduced
the sensitivity of future spending to RPI inflation somewhat, but – with interest rates at historic
lows, and having fallen further since our March forecast – interest rates still seem more likely
to overshoot our forecast materially than they are to undershoot it over the medium term.
Market participants also appear to be pricing in some chance of a no-deal Brexit and the
likely monetary policy response, so a smooth outcome could see interest rates rise again.

Executive summary
47 Our deeper look at the ‘growth-corrected interest rate’ (or ‘R-G’ for short), a key factor
driving the evolution of the debt-to-GDP ratio, highlights its variability over the past 120
years, and hence the highly uncertain outlook. While a continuation of the current unusually
low R-G would make it less risky to allow increases in debt in response to adverse shocks, it
is important not to overstate its importance to future fiscal sustainability. The more salient
challenge is how demographic trends and other cost pressures in health and social care
provision will affect spending over the longer run. The favourable effect of a lower R-G
would make the sustainability challenge posed by these pressures less daunting, but the
chance of it being sufficiently large or persistent to offset them altogether is negligible.
Policy risks
48 In Chapter 8 we look at potential sources of fiscal policy risk and the drivers of fiscal policy
over the past decade. We consider the extent to which governments (of different hues) tend
to stick to their stated fiscal objectives and policy announcements and how they respond to
changes in our underlying forecast of the fiscal position. The risk here is that self-imposed
rules are revised to accommodate changes in the forecast, rather than policy being adjusted
in order to meet them. In such a scenario, the profile of debt would be expected to be higher
than if governments adhered to their rules rather than altering them when convenient. This
seems to be borne out by recent history, notably following the EU referendum.
49 At the same time, the Government has engaged in sometimes remarkably precise medium-
term fine-tuning of its policy measures to meet informal objectives for the public finances –
borrowing falling on a particular measure in a particular year, the precise headroom against
the formal fiscal rules, or the profile of the debt-to-GDP ratio to one decimal place. The
margins by which these objectives were met were rarely fiscally or statistically significant.
50 Unsurprisingly, there is a similar pattern of accommodative revisions to already announced

policy, particularly to spending plans that lie outside Spending Review periods. These have
tended to contain fiscal tightening when first announced, but have been relaxed as time
proceeds and plans have to be firmed up. Given that the bulk of departmental spending
plans for 2020-21 onwards have not yet been set, past behaviour would suggest this is a
material source of policy risk to our central forecast. The spending pledges being made
during the Conservative Party leadership contest only make that more likely.
51 The Government’s current formal fiscal objective is to balance the budget by the mid-2020s.
But the current Chancellor’s recent statements suggest that the Government has all but
abandoned that goal in favour of the less stretching target of ensuring that debt falls relative
to GDP. He told the Treasury Select Committee that balancing the budget was “a choice”,
while asking Conservative Party leadership candidates to commit only to “keeping our
national debt falling”. Despite that, both remaining candidates have set out proposals with
potential costs in the low tens of billions of pounds, with few accompanying suggestions of
ways to cut other spending or to raise revenue. It remains to be seen what the winner and his
Chancellor will implement in future fiscal events, which will presumably depend in part on
near-term Brexit developments. But from their statements, it seems highly likely that the new
administration will loosen fiscal policy and adopt less ambitious fiscal objectives.

Executive summary
52 In this context, we look at how far primary borrowing could rise and still keep debt falling on
the assumptions underpinning our March forecast – which include assuming a smooth path
to a Brexit deal. Abstracting from the uneven effect on net debt of Bank of England schemes
(notably the £121 billion Term Funding Scheme), primary borrowing would have to rise by
around £22 billion in 2020-21 for debt not to fall relative to GDP in that year. Given the
£11.6 billion loss of headroom that will follow forthcoming changes to the accounting
treatment of student loans, this would lead to the current fiscal mandate being missed by a
significant margin. Between 2021-22 and 2023-24, primary borrowing would have to rise
by between £21 billion and £25 billion for debt not to fall in any of those years.
Climate change
53 In Chapter 9 we survey some of the climate-related risks to the economy and the nature of
climate-related fiscal risks. We noted this issue in 2017, but did not analyse it in any depth.
Understandably, the Government did not address it in its MFR response. Here we note that:
• The scale of the risks associated with climate change will depend hugely on the extent
to which global temperatures rise. If the most ambitious of the goals agreed in Paris in
2015 are met and temperatures stabilised at 1.5oC above pre-industrial levels – and
therefore only around 0.5oC higher than temperatures witnessed in recent years – then
the scale and frequency of climate-related events is likely to increase, but perhaps not
significantly compared to what we have already experienced. By contrast, if global
mitigation efforts were to fail and temperatures reach 4oC above pre-industrial levels,
the risks could be much greater and more difficult to assess, with mass international
migration and induced periods of conflict, as well as extreme weather events.
• A useful framework for the analysis of climate-related fiscal risks is provided by the
approach adopted by the Bank of England (and other central banks and financial
regulators) for analysing climate-related risks to financial stability. This splits them into
physical risks – those related to extreme weather events and gradual global warming –
and transition risks – those related to the shift to a low-carbon economy. And it
proposes four high-level scenarios that capture different settings along two important
dimensions: the strength of the greenhouse gas mitigating policy response – whether
Paris targets are met or not; and how smoothly and foreseeably those actions are
taken – whether the transition to either end point is orderly or disorderly.
• Climate-related risks to the economy are not particularly well understood or modelled,
but they will clearly be an important indirect source of fiscal risk. Extreme weather
events could disrupt economic activity. Diverting investment to adaptation needs could
impinge on investment in productive capital. The same could be true of investment in
lowering greenhouse gas emissions, although as green technologies advance, the
greater risk may be in foregone output growth if the economy is not well placed to
capitalise on those advances or is locked into outdated infrastructure.

Executive summary
• Direct fiscal implications from changes to tax revenues and spending can arise through
all the channels we consider. Extreme weather events can generate calls on public
spending to repair damage to private and public assets. Investment in adaptation
measures – for example, in flood defences and to manage the consequences of
coastal erosion – require public spending. And mitigation policies will typically have
either fiscal costs in terms of public spending or fiscal gains where taxes are used to
discourage particular activities. In a world in which global warming proceeds as
envisaged by the Paris targets, these direct effects currently appear relatively modest.
A no-deal Brexit stress test scenario

54 In Chapter 10 we carry out a fiscal ‘stress test’ that quantifies the impact on the public
finances of a particular no-deal, no-transition Brexit scenario, namely the less disruptive of
the two presented by the IMF in its April 2019 World Economic Outlook. While it is a
scenario, rather than a forecast, it is nevertheless useful for exploring the channels through
which the public finances might respond in the case of a no-deal Brexit.
55 Heightened uncertainty and declining confidence deter investment, while higher trade
barriers with the EU weigh on exports. Together, these push the economy into recession, with
asset prices and the pound falling sharply. Real GDP falls by 2 per cent by the end of 2020
and is 4 per cent below our March forecast by that point. Higher trade barriers also slow
growth in potential productivity, while lower net inward migration reduces labour force
growth, so potential output is lower than the baseline throughout the scenario (and beyond).
The imposition of tariffs and the sterling depreciation raise inflation and squeeze real
household incomes, but the Monetary Policy Committee is able to cut Bank Rate to support
demand, helping to bring output back towards potential and inflation back towards target.
56 Borrowing is around £30 billion a year higher than our March forecast from 2020-21
onwards. Lower receipts – in particular income tax and NICs (due to the recession) and
capital taxes (due to weaker asset prices) – explain most of the deterioration. These are partly
offset by lower debt interest spending (thanks to lower interest rates and RPI inflation) and the
revenue raised customs duties (which are treated as EU rather than UK taxes in the baseline).
Higher borrowing and the assumed rollover of Term Funding Scheme loans leave public
sector net debt around 12 per cent of GDP higher than our March forecast by 2023-24.

Executive summary
Chart 1: Stress test vs March forecast: sources of differences in PSNB

60
Other receipts and spending elements
Income tax and National Insurance contributions
50
Capital taxes
Customs duties
40
Debt interest spending
Difference
30
£ billion
20
10
-10
-20
2017-18 2018-19 2019-20 2020-21 2021-22 2022-23 2023-24
Source: ONS, OBR
57 The fiscal mandate is met with a slightly smaller margin of £22.7 billion, but the debt target
is missed. Meeting the mandate by only a slightly smaller margin than in March reflects two
features of the scenario: first, the near-term hit to the economy is largely cyclical; and
second, higher customs duties generate a near-term structural improvement to receipts. The
Government’s objective of returning the budget to balance by 2025-26 is further out of
reach than in our March forecast, with borrowing standing at almost £40 billion in 2023-24.
58 The stress test is by no means a worst-case scenario under a no-deal, no-transition Brexit.
Neither the cyclical downturn nor the medium-term loss of potential output are as large as
those considered in the Bank of England’s disruptive and disorderly Brexit scenarios that
were published last year. Most important from the perspective of fiscal sustainability would
be if lower trade intensity were to generate adverse dynamic effects on productivity and
potential output. These would be relevant in any Brexit scenario.
Summary of medium and long-term fiscal risks

59 Figures 1.1 and 1.2 show a stylised illustration of the main risks to our medium-term fiscal
forecasts and to long-run sustainability, by size and likelihood. For the medium-term risks,
because we aim to factor into our central forecast any event or trend that we consider more
likely than not to crystallise, most forecast risks are deemed to be medium or low probability
almost by definition. The exceptions are policy risks, since our forecasts are conditioned on
the Government’s current stated policy rather than a judgement about the most likely path
for policy. For the long run, there are more risks that we consider highly likely to occur and
to be relatively high impact if they do. This is true of both potential shocks and pressures.

Executive summary
60 Alongside this report we are also publishing a comprehensive risk register on our website,
which lists all the fiscal risks discussed in this report, our assessment of their size and
likelihood, and, for those identified in our 2017 report, any changes since then.
Figure 1: Sources of fiscal risk over the medium term

Impact on public sector net debt by 2023-24 (per cent of GDP)
Change since 2017

HIgh (Over 10)
Fiscal risks report
• 2017 stress test: severe New risk for 2019

recession
• Financial crisis
• 2019 stress test: no-deal Brexit • Typical recession
• Weak productivity growth

Medium (1-10)
• Output gap mismeasurement • Discretionary fiscal loosening

• Lower consumption and
labour income shares
• Higher interest rates (1ppt)
• Major statistical • Low migration
reclassifications • Higher inflation (1ppt a year)
• Higher cost of tax reliefs
• Lower incomes for high
• Lower consumption share
Low (Less than 1)
earners
• Faster growth in self-employ- • No fuel duty RPI increases
ment and incorporations
• Lower prices for expensive • Triple lock: higher cost • Additional health spending
• Higher tax litigation costs houses
• Lower dutiable consumption • Additional adult social
care spending
Medium Very low High Low

Very high
(40-60) (Less than 10)(60-90) (10-40)
(Over 90)
Probability of crystallisation (per cent)
Note: The ‘Income tax policy commitments’ risks related to the 2017 Conservative Party manifesto pledges we included in 2017 has crystallised and
therefore been removed.
Source: OBR
Figure 2: Sources of risk to fiscal sustainability

Change since 2017
Impact on public sector net debt by 2068-69 (per cent of GDP)
HIgh (Over 100)
Possible major fiscal Fiscal risks report

risks that we have not
quantified:
• Major wars
• Climate change
• Higher growth-corrected • Health spending: other
interest rate cost pressures
Medium (10-100)
• Health spending: ageing

• Lower productivity growth
• A typical recession • Financial crisis
• Triple lock: high cost • Low migration every decade
• Adult social care: other
cost pressures
• Triple lock: central
• Greater fuel efficiency

• Continued growth in
Low (Less than 10)
incorporations • Lower smoking
• Adult social care: ageing
Very low Low Medium High Very high

(Less than 10) (10-40) (40-60) (60-90) (Over 90)
Source: OBR

1 Introduction
1.1 The OBR is tasked with producing a biennial report on “the main risks to the public finances,
including macroeconomic risks and specific fiscal risks”. Several countries produce regular
fiscal risk assessments, but in most cases these are undertaken by finance ministries or
cabinet offices; the UK is unusual in outsourcing it to an independent fiscal institution,
thereby boosting transparency around the Government’s management of those risks.
1.2 Fiscal risk assessment is a potentially huge subject. There are few activities in the economy
or in the public sector without some implications for the public finances – and each may be
subject to risks and uncertainties. In this report, we look first at fiscal risks related to
developments in the macroeconomy and the financial sector, and then at a variety of
specific revenue, spending and balance sheet risks, before pulling several of them together
in a fiscal ‘stress test’. We also consider fiscal risks associated with climate change and the
operation of fiscal policy. Throughout the report we look at how well our 2017 risk
assessment has held up against developments over the past two years, and how the
Government’s formal response to that assessment addressed the issues we raised.
1.3 This chapter introduces the framework we use to analyse fiscal risks. It starts by stepping
back to consider the role of government in managing the risks faced by society. It then
explains what we consider as fiscal risks and how they affect the public finances, our
approach to analysing specific fiscal risks, and the Government’s approach to managing
them. It ends by raising some overarching issues for the Government’s next response.
Government’s role in managing risks society faces

1.4 The definition of fiscal risk we use in this report (described below) focuses on surprises
relative to forecasts and pressures on fiscal sustainability. But it is important to remember
that the whole purpose of much government activity is to pool risks that society has decided
(via the political process) would better be taken on by the state than by individuals, either
directly or through participation in private insurance markets. For example: the NHS takes
on the risk people would otherwise face not knowing when or how they might fall ill; state
pensions remove the risk of pensioners’ incomes falling below a certain level, due to their
living longer than expected or receiving disappointing returns on pension savings; and
universal credit reduces the risks associated with periods out of work or on low pay.
1.5 One way to split the risks that society confronts is into individual and collective risks. The
rationale for government intervention is different in each case. For example, take pensions:
• Individual risks: in principle, risks to an individual’s retirement income can be insured

against, with financial markets allowing them to be spread through pooling. For

Introduction
example, providers of annuities can offset the higher cost of payments to those who
live longer than average with the lower cost of payments to those who die earlier than
average. The probability attached to different outcomes, and the cost were they to
materialise, can be calculated – making them quantifiable risks rather than
unquantifiable uncertainties. But other information problems mean that markets of this
type do not function well – for various reasons many people are unwilling to buy an
annuity voluntarily. This ‘market failure’ creates a rationale for government
intervention. Until reforms announced in Budget 2014, it was in effect compulsory for
most individuals with a defined-contribution pension to buy an annuity on retirement.
• Collective risks: unlike individual risks, it is impossible even in principle to insure

against collective risks – risks that affect everyone if they crystallise. This type of risk
cannot be shared within generations, at best only across them (even that may not be
possible), and doing that requires government intervention. These risks are often
compounded in practice by unquantifiable uncertainty – for instance, in the case of
pensions, while past trends can provide a guide, prospects for longevity a generation
hence are largely unknowable. In theory, the government could try to establish a
market (for example, by issuing longevity bonds as it does with inflation-linked bonds),
but this would not be straightforward. In practice, the Government has chosen to
provide a state pension, currently uprated by the ‘triple lock’, and tax relief on
individuals’ pension contributions. As a result, government holds the inflation risk in
respect of the state pension. But it has passed some longevity risk back to individuals
by linking the state pension age to projections of life expectancy.
1.6 Given this distinction, when considering long-term risks to health spending posed by
population ageing and other cost pressures, we are talking about risks to the cost of
providing health care, taking as given society’s choice that such costs should be shared
within and across generations via government, rather than falling to the individuals affected.
1.7 In some cases, the appropriate allocation of responsibility between the individual and the
state is highly contentious. That is illustrated by the continuing discussions around who
should meet the costs of adult social care. The appropriate balance between individual and
state has been reviewed repeatedly and another Green Paper is pending. Moreover, this is
an area where unquantifiable uncertainty is rife: for instance, regarding the changing
probability of needing care as life expectancy rises; and the scope for future medical and
technological advances to affect care costs.
What do we consider a fiscal risk?

When is a risk a fiscal risk?
1.8 The International Monetary Fund (IMF) defines a fiscal risk as “the possibility of deviations of
fiscal outcomes from what was expected at the time of the Budget or other forecast”.1 On
this basis, we would define a fiscal risk as a potential deviation from the 5-year-ahead
1
IMF Fiscal Affairs Department, Fiscal risks – sources, disclosure and management, 2009.

Introduction
central forecasts for public sector spending, receipts, borrowing and debt contained in our
Economic and fiscal outlooks (EFO), and from the corresponding 50-year-ahead projections
in our Fiscal sustainability reports (FSR). We are required by Parliament to base these
forecasts and projections on current stated Government policy, although in most cases
current policy is much less clearly defined over the long term than over the medium term.
Where appropriate, we consider policy risks – areas where government statements or past
behaviour point to likely policy changes (see also Chapter 8 in this report).
1.9 On this definition, however, what constitutes a fiscal risk depends crucially on which
developments in the public finances are incorporated into the central projection and which
are regarded as potential deviations. Given the sensitivity of long-term projections to these
sorts of judgements, we focus on risks around our central forecast over the medium term,
but on risks to fiscal sustainability (rather than to our latest central projection) over the
longer term. This ensures that we do not end up ignoring some of the most important –
notably pressures on health spending – simply because we already assume they crystallise.
1.10 Our focus on risks to sustainability also implies an asymmetry of approach – we are more
(although not exclusively) interested in potential ‘bad news’ than ‘good news’. Experience
over time and across countries suggests that shocks to the public finances (especially big
ones) are more likely to be adverse than beneficial – the cost of wars being the most
dramatic example – and that governments are usually quicker to spend unexpected
windfalls from good news than they are to address unexpected costs from bad news.
Fiscal risks and the public finances

Stock and flow measures of the public finances
1.11 Once we have decided what to treat as a fiscal risk, we need to assess how likely it is to
crystallise and how big an impact it would have on the public finances. For the latter, we
employ the fiscal metrics that are reported by the Office for National Statistics (ONS) in the
National Accounts, supplemented with other information. We are interested both in flows of
spending and receipts, and in the stocks of assets and liabilities on the balance sheet.
1.12 Starting with the flows, governments spend money every year on things like public services,
capital investment, pensions and benefit payments, and they raise money, mainly from
taxes. Governments also have to pay interest on their financial liabilities, while they receive
interest and dividend income from their financial assets. Public sector net borrowing (PSNB)
– the headline measure of borrowing – is the difference between total spending and total
receipts.2 The ‘primary’ balance excludes interest and dividend payments and receipts.
1.13 Turning to the balance sheet, the National Accounts recognise a variety of public sector
financial liabilities and assets. The former include ‘debt liabilities’ (notably gilts), plus other
liabilities (such as accounts payable). The latter include ‘liquid assets’ (notably the foreign
exchange reserves), plus illiquid ones (such as student loans or accounts receivable). Public
2
Specifically, ‘public sector net borrowing excluding public sector banks’.

Introduction
sector net debt (PSND) – the headline summary measure of the public sector balance sheet
– is the difference between the government’s debt liabilities and its liquid assets.3 Public
sector net financial liabilities (PSNFL) is a newer and broader measure than PSND, including
all financial assets and liabilities in the National Accounts. On both measures, the
government’s liabilities exceed its assets by a considerable margin. But they both exclude
the government’s fixed assets (such as roads and buildings) and by far its greatest financial
asset – its ability to levy future taxes – which some broader measures do reflect.4
1.14 When looking at the evolution of both stock and flow measures of the public finances over
time, it usually makes most sense to look at them relative to GDP. For, as the economy
grows over time, so too does the pool of potential tax revenue that governments can draw
on to finance public spending.
How fiscal risks can have both stock and flow effects
1.15 Viewed through this stock-and-flow accounting framework, we can think of most fiscal risks
as potential events or trends that would result in:
• a one-off or persistent increase in spending (such as the cost of fighting a war or the
need to spend a higher proportion of GDP on health because of cost pressures);
• a one-off or persistent loss of revenue (such as the sharp falls in stamp duty when
house prices fall or a structural decline in excise duty as a result of reduced smoking);
• a balance sheet transaction, in which the government issues debt to buy an asset or to
lend to the private sector (such as the purchase of shares in RBS during the crisis);
• a balance sheet transfer, in which the government directly absorbs assets and liabilities
from the private sector (this can be a real-world event, like the transfer of Royal Mail’s
historic pension liabilities and associated assets to the public sector in 2012, or a
purely statistical one, as when the ONS reclassified English housing associations from
the private to the public sector in 2015 and back again in 2017); or
• a change in the value of existing assets and liabilities (such as the impact of exchange
rate changes on the sterling value of the UK’s unhedged foreign exchange reserves).
These last three developments are referred to together as ‘stock-flow adjustments’.
1.16 Most balance sheet transactions or transfers between the public and private sectors have a
persistent impact on public sector spending and/or revenue flows, via the income that the
assets generate or the interest or other payments that must be made on the liability.
3
Unless otherwise stated, when we refer to PSND in this report we are referring to ‘public sector net debt excluding public sector banks’.
4
See, for example, Box 1.1 of our 2017 Fiscal risks report on broader measures of the public sector balance sheet.

Introduction
1.17 When we think about fiscal sustainability, stocks and flows can both be sources of fiscal risk,
but it is ultimately the flows that matter. A risk threatens sustainability if its crystallisation
would move the public finances onto, or closer to, a trajectory in which the government
would eventually be unable or unwilling to raise sufficient revenue to deliver core public
services and to meet its financial obligations. If a government does find itself on a trajectory
of this sort, eventually a fiscal crisis will result.5
1.18 Governments will typically seek to take action that gets them off – or avoids them getting
onto – an unsustainable trajectory before a crisis breaks. Indeed, they may deem it prudent
to act even if the outlook appears sustainable on a central projection, for example if debt
reaches a share of GDP where a government feels vulnerable to a shift in market sentiment
that would raise its borrowing costs and/or result in a disruptive currency depreciation.
1.19 There is no consensus in the academic literature or policy world as to what levels of debt are
optimal or, relative to some subjective view of tolerable risk, safe. Even if there were, there is
no reason to believe that these would be constant over time or consistent across countries.
Some studies suggest that policymakers should aim to have the debt-to-GDP ratio falling in
normal times so as to make room to accommodate the cost of large adverse future shocks.
The evolution of public sector debt and interest payments

1.20 Changes in the debt-to-GDP ratio reflect the size of the primary budget balance (and
therefore any revenue and spending shocks), the impact of any stock-flow adjustments and
the difference between the interest rate on the government’s debt and the growth rate of
output (the ‘growth-corrected interest rate’). The last matters because interest payments add
to debt, raising the debt-to-GDP ratio, while growth adds to GDP, lowering it. We explore
the role of the growth-corrected interest rate in Chapter 7.
1.21 As Chart 1.1 shows, the historical experience in the UK has been one of sharp rises in
public debt as fiscal risks crystallise – typically due to wars, but also as a result of the
financial crisis and ensuing recession a decade ago – followed by long periods of gradual
declines.6 Up to the Second World War, debt reduction typically relied on the running of
primary surpluses. Since then, securing a favourable growth-corrected interest rate, thanks
to unanticipated inflation and/or financial repression, has played a more prominent role.
5
A fiscal crisis could involve explicit default, resort to an IMF ‘bailout’, implicit default on domestic liabilities via unanticipated inflation or
accumulation of arrears, or loss of access to capital markets. A study published by the IMF in 2017 estimated that on this definition 15 out
of 35 advanced economies experienced at least one fiscal crisis between 1970 and 2015, including the 1976 UK crisis in which the then
government borrowed $3.9 billion (around 1.7 per cent of GDP at the time) from the IMF. Gerling et al, Fiscal crises, IMF, 2017.
6
Compiling very long time series inevitably requires judgements to be made about how to splice together different data sources and how
to fill any gaps in the available data. We have used the Bank of England’s ‘a millennium of macroeconomic data’ to produce these charts.

Introduction
Chart 1.1: Public sector debt and borrowing since 1800

300
Post-Napoleonic Post-Napoleonic wars WWI WWII Post-war Post-crisis
depression decades
250
Public sector net debt including banks
Public sector net debt
200
Per cent
150
100
50
35
Primary deficit
30
Net borrowing
25
Net interest spending
20
15
Per cent
10
-5
-10
-15
1800-01 1820-21 1840-41 1860-61 1880-81 1900-01 1920-21 1940-41 1960-61 1980-81 2000-01
Source: Bank of England, ONS
1.22 In some respects, the challenge facing governments in reducing debt is greater now than in
the past: the population is ageing at a time when public spending has already been tilted
towards the old; financial repression is harder to achieve when inflation is low, the central
bank is independent and capital flows freely across borders; and expectations for public
services and the welfare state – plus resistance to higher taxation – make primary surpluses
harder to sustain. Against that, the growth-corrected interest rate is historically low and likely
to remain so for at least the immediate future, making debt burdens more sustainable.

Introduction
Identifying the characteristics of specific fiscal risks

1.23 When identifying and assessing a source of fiscal risk, we start by asking:
• What is the nature of the risk? Is it discrete (i.e. an unexpected event or shock) or
continuous (i.e. an unexpected trend or pressure). Is it isolated or is it correlated with
other risks? Is it exogenous (i.e. largely unaffected by policy) or endogenous (i.e. if the
likelihood or impact of crystallisation is affected by policy). Was it taken on by choice?
• How likely is it to crystallise? This will depend to a significant degree on the time
horizon – most are more likely to crystallise over 50 years than they are over five years.
The probability of a risk crystallising also depends on how it is defined – recessions are
fairly common, but deep ones like that of 2008 to 2009 happen only infrequently.
• What impact would crystallisation have on the public finances? We are interested in
the potential impact on both stock and flow measures of the public finances. We noted
above that adverse shocks to flows can be either one-off or persistent. The latter tend
to be a greater threat to sustainability than the former. Persistent sources of risk can
also be divided into those that strike abruptly and those that mount gradually.
1.24 Answering these questions about individual risks to our forecasts and to fiscal sustainability
allows us to assess their relative importance. Quantifying this is not straightforward. Rare
events or emerging sources of risk offer little hard evidence from which to estimate
probability or impact. And for many, the likelihood and impact of crystallisation interact. So
we draw on what evidence is available, complemented by our own broad judgements, to
place individual risks into one of five categories for likelihood and one of three for impact.
This provides a stylised illustration of their relative importance. Figure 1.1 shows the grid
used for this purpose. Populated versions were presented in the executive summary.
Figure 1.1: A stylised fiscal risk matrix

HIgh
Greater impact if it crystallises

Impact on public sector net debt
risk
fi s cal
f
eo
urc
n t so
orta
Medium
re imp
Mo
Low
More likely to crystallise
Very low Low Medium High Very high

(Less than 10) (10-40) (40-60) (60-90) (Over 90)

Introduction
1.25 To complete our risk assessment, we also ask:
• How is the risk currently recognised in official data and forecasts (if at all)? Some fiscal
risks considered in this report are regularly discussed and quantified in our EFOs and
FSRs, either in the central forecasts and projections or in our analysis of the uncertainty
around them (including sensitivity analysis and alternative scenarios). Some are
recognised in the Treasury’s Whole of Government Accounts.
• What is the Government doing to manage the risk? Is it a risk the Government has
chosen to take on? Does it wish to limit its exposure or is it taking on the ‘tail risk’?
Can it influence the likelihood of crystallisation or just the impact were it to crystallise?
These and other issues are discussed in the next section.
How does the Government manage fiscal risks?

1.26 In Managing fiscal risks (MFR), the Government’s 2018 response to our 2017 Fiscal risks
report (FRR), the Government described its strategy for managing fiscal risk as follows:
“…to be aware of the risks it is facing, to reduce risks where this can be done in a
cost-effective way and without detracting from its wider policy objectives, and to
ensure the overall position of the public finances is resilient to risks that remain. In
doing so the government seeks to set clear expectations regarding the limits of the
state’s responsibility if and when risks materialise. However, it also recognises that
actively taking on risk at certain times can minimise overall costs in the future.”
1.27 It stated that its approach to managing fiscal risks was based on the five-stage approach
advocated by the IMF: identifying the source of risks; disclosing them to Parliament and the
public; mitigating them where that can be done cost-effectively; provisioning against those
that remain and can be quantified with reasonable certainty; accommodating those that
cannot.7 It deploys a range of tools to manage risks at different stages in the process.
1.28 MFR did not provide a statement of risk appetite: a baseline against which one could assess
whether changes in fiscal risk move the Government closer to or further from its desired
position. What it did say is that the Government believes that public debt is currently “too
high”. It argued that this leaves the public finances more vulnerable to economic shocks,
unfairly passes financial burdens onto the next generation and increases spending on debt
interest. We asked again about risk appetite when gathering background material for this
report, but the Government did not provide any further information.
1.29 As regards its fiscal targets, the Government set out its objective in the Charter for Budget
Responsibility: to “return the public finances to balance at the earliest possible date in the
next Parliament”.8 In practice, in line with the Conservative Party’s 2017 manifesto, this
objective has been thought of in terms of balance by 2025-26. But more recently, the
7
The IMF’s approach is set out in IMF, Analyzing and managing fiscal risks: best practices, June 2016.
8
The most recent version of the Charter was passed by Parliament in January 2017. This and all previous versions are available on the
‘Legislation and related material’ page of our website.

Introduction
Chancellor has stated that “it will be a policy decision at successive fiscal events how to
balance whatever available fiscal headroom there is between reducing the deficit, reducing
taxes [or] increasing spending”.9 We explore fiscal policy as a source of risk in Chapter 8.
1.30 The Charter also sets out targets for borrowing, debt and welfare spending that require:
• the structural deficit to be below 2 per cent of GDP by 2020-21;
• public sector net debt to fall as a percentage of GDP in 2020-21; and
• selected welfare spending items to lie below a ‘welfare cap’ set for 2022-23.
1.31 In terms of individual risks, the Treasury’s recently updated ‘Orange Book’ guidance states
that “risk evaluation should involve comparing the results of the risk analysis with the nature
and extent of risks that the organisation is willing to take to determine where and what
additional action is required.” The options it sets out include:
• avoiding the risk, if feasible, by stopping or not starting whatever gives rise to the risk;
• taking or increasing the risk in order to pursue an opportunity;
• retaining the risk by informed decision;
• changing the likelihood, where possible;
• changing the consequences, including planning contingency activities; and
• sharing the risk (e.g. through commercial contracts).
1.32 In Chapter 2 of our 2017 FRR, we outlined the Treasury’s risk groups and reporting
processes. Box 1.1 describes developments in fiscal risk management since then.
Box 1.1: Developments in fiscal risk management since our previous report
The Treasury’s approach to risk management remains broadly the same as two years ago: a
number of internal management groups sit beneath its ‘Fiscal Risks Group’ and a spending
control framework makes accounting officers responsible for remaining within spending limits.
But several initiatives, including those pursued as part of the Treasury’s ‘Balance Sheet Review’
(BSR), have deepened these arrangements. These include:
• The Treasury has issued new guidance on disclosing the fiscal impact of asset sales on the
deficit and a variety of balance sheet measures. This might reduce the focus on PSND,
which suffers from fiscal illusions in respect of asset sales because the proceeds net off
whereas the asset being sold does not, so the balance sheet appears to improve.
9
Oral evidence on Budget 2018 from the Chancellor to the Treasury Committee, 5 November 2018.

Introduction
• BSR work to identify government assets, dispose of those no longer required and increase
the utilisation or returns on those that remain.
• Stricter controls on the creation of new contingent liabilities have been introduced, with
greater scrutiny of new liabilities and greater emphasis on government being
compensated for taking on risk. The Treasury plans to introduce greater oversight of the
stock of existing contingent liabilities as its next step in this process.
• Steps have been taken to reduce the growth in clinical negligence claims, which the
Whole of Government Accounts identifies as the fastest growing contingent liability. One
such step is the production of a new patient safety strategy by the NHS.
• BSR work on intangible assets, including the publication of a report at Budget 2018 and
the establishment of a knowledge assets team in the Treasury. We look more closely at
intangible assets in Chapter 6.
• The Government publishes its risk management concepts and principles in the ‘Orange
Book’. The Government Finance Function has published an updated version in July 2019.
The Government wants risk management to become more integral to departmental
planning, including via revised Single Departmental Plan guidance, the design of
Spending Review guidance and enhanced risk disclosures in Annual Reports.
We have a particular interest in the Treasury’s work to investigate the stock of contingent
liabilities. It would be useful for the Whole of Government Accounts to include more systematic
information of the nature of the stock of contingent liabilities and the public sector’s ultimate
exposure to different sources of underlying risk.
Forecast revisions and our 2017 fiscal risk assessment

1.33 When we published our first FRR, our most recent medium-term forecast was from our
March 2017 EFO. Since then, we have updated our forecasts four times. What do the
revisions over that period tell us about the risk assessment we made two years ago? Did we
correctly identify the risks that have crystallised? Did we miss any important ones?
1.34 Chart 1.2 compares our March 2017 five-year forecast against outturns for 2017-18 and
2018-19, and our March 2019 forecast for 2019-20 to 2021-22. Borrowing came in lower
than expected in outturn, but we have revised up our forecast materially in 2019-20 and by
modest amounts in 2020-21 and 2021-22. (We cover debt briefly at the end of the section.)
1.35 As the chart shows, classification changes have lowered measured borrowing in all years. In
outturn, unexpectedly strong tax receipts pushed borrowing below our March 2017 forecast.
This better starting point fed through to our forecasts in the short term, but was eventually
offset by other factors (described below). Pre-measures spending forecast changes were
generally small, although lower interest rates pulled down our debt interest forecast in the
medium term. These favourable forecast and classification changes are outweighed by
policy giveaways from 2019-20 onwards, dominated by the June 2018 NHS settlement.

Introduction
Chart 1.2: Changes to the OBR’s borrowing forecast since our March 2017 EFO
35
Classification changes
30
Tax forecast
25
Primary spending forecast
20
Debt interest spending forecast
15
Measures
10
Pre-measures
£ billion
5
Total
0
-5
-10
-15
-20
-25
2017-18 2018-19 2019-20 2020-21 2021-22
Source: OBR
1.36 The effect of classification changes on borrowing over the past two years is dominated by
the reclassification of housing associations back into the private sector. We flagged this as
the largest medium-term balance sheet transfer risk, since at the time of our previous report
the Government was engaged in a legislative process to give up just enough of its control
over them to justify their reclassification back into the private sector.
1.37 Table 1.1 breaks down the difference between our March 2017 tax forecast and the latest
outturns and forecasts on a like-for-like basis, but excluding the effects of subsequent policy
measures. This shows positive news on effective tax rates in the early years, but the negative
effect of a lower trend productivity assumption building over time and eventually
outweighing the good news from outturns. As regards weaker productivity growth, we
assigned such a development a medium likelihood in our previous report, although noted
that the latest data at the time “might suggest a gloomier prognosis”. In the event, we
revised our assumption down materially in the November 2017 EFO. As regards the positive
news, this has included income tax and NICs being stronger than expected as a share of
labour income and corporation tax being stronger as a share of profits. On the labour
income side, one factor has been strength in pay growth among high earners – a positive
crystallisation of the concentration risk we discussed. On the corporate side, many factors
have played a role, including unexpectedly strong outturns and technical modelling issues.

Introduction
Table 1.1: Like-for-like pre-measures changes in receipts since our March 2017 EFO
£ billion
2017-18 2018-19 2019-20 2020-21 2021-22
Total change 9.5 14.3 7.8 4.4 -0.8
Lower productivity growth -9.3 -14.6 -18.1 -22.8 -26.5
Higher average hours worked 2.6 4.1 5.1 6.4 7.6
Lower unemployment 0.5 3.6 4.1 4.9 5.6
Other economy forecast changes 6.6 10.5 3.6 1.4 -0.5
Higher effective tax rate on labour income 5.1 7.6 6.2 8.6 8.2
Higher effective tax rate on corporate profits 0.9 2.1 3.4 4.2 4.8
Other changes in effective tax rates 3.0 1.1 3.6 1.8 0.0
1.38 Presented with an improved underlying fiscal position, Government policy measures have
added to borrowing in all years – particularly from 2019-20 onwards. As Table 1.2 shows:
• The largest of these was the boost to health spending announced in June 2018 and
factored into our forecast in Budget 2018, added to in Spring Statement 2019. We
flagged higher health spending as a high likelihood medium-term risk in FRR 2017.
• Other departmental spending was also topped up, particularly at Budget 2018.
• Universal credit giveaways – largely completing the reversal of cuts announced in

Summer Budget 2015 – have come at a significant cost. This was not a risk we
flagged. Instead, we focused on risks to rollout plans, which have indeed been pushed
back further since our previous report, but with little effect on public spending.
• Continued fuel duty freezes and further increases in the income tax personal allowance
and higher rate threshold were deemed a near certainty in our previous report – due
to past experience and manifesto commitments respectively. This has proved to be the
case. Alcohol duties were also frozen again.
• The aggregate effect of a range of other decisions follows a familiar profile: near-term
giveaways followed by takeaways from 2020-21 onwards (the fiscal mandate year).
Table 1.2: The effect of policy measures on borrowing since our March 2017 EFO
£ billion
2017-18 2018-19 2019-20 2020-21 2021-22
Total effect of Government decisions 0.7 4.3 21.3 15.2 16.4
Higher health spending 0.0 0.0 7.7 11.5 16.9
Higher other current and capital departmental spending 0.5 3.7 7.3 4.7 1.2
Universal credit giveaways and policy reversals 0.0 0.3 1.0 1.4 1.7
Fuel and alcohol duty freezes 0.0 1.1 2.1 2.1 2.2
Income tax manifesto commitments 0.0 0.0 2.8 1.9 1.4
Avoidance and evasion measures 0.0 -0.2 -1.3 -1.8 -1.8
Other 0.2 0.8 5.3 -0.9 -1.0
Indirect effects -0.1 -1.4 -3.6 -3.8 -4.2
Note: Consistent with the presentation of measures' effects in our EFOs, we have reported the October 2018 pensions contributions
measure on a net basis, amongst other adjustments. We have also presented the PSNB effects of business rates pilot measures on a net
basis as they have large, and largely offsetting, gross effects on DEL and AME, but a small net effect on PSNB.

Introduction
1.39 Overall, most of the large risks that crystallised – reclassification of housing associations,
extra health spending and lower productivity growth – were identified in our first report, but
we did not discuss universal credit giveaways. These same factors also represent most of the
significant drivers of changes in PSND. But in addition to these, it has been revised up due
to the amount of lending under the Term Funding Scheme (TFS) being higher than we had
assumed. We discussed some risks relating to the TFS, but not higher initial lending.
For the Government’s response

1.40 In the rest of this report, we set out various issues the Government might wish to note in its
fiscal risk management strategy and, if it agrees, ask what approach it takes to them. In
doing so we update our assessment of the 57 issues set out in our first report, and add a
handful of new ones.10 It is not for us to recommend particular policy responses, but, as in
2017, this hopefully provides an opportunity for the Government to explain its choices.
1.41 The discussion in Box 1.1 about steps the Government has taken to improve its fiscal risk
management prompts two issues that the Government may wish to consider in its response:
• How to embed planned risk management changes in departments’ decision-making?
• How to ensure that large or complex risks are escalated and managed appropriately?
1.42 We also raised several overarching issues in our 2017 report:
• The need to review risks that governments choose to expose themselves to.
• The need to prepare for near-inevitable future shocks.
• The need to deal with many slow-building pressures.
• The challenges of dealing with those needs while negotiating Brexit.
• The challenges of doing so in an environment of apparent ‘austerity fatigue’.
• The more vulnerable starting fiscal position from which all of this is faced.
• Sources of fiscal risk that we had not analysed – such as wars and climate change.
1.43 In Managing fiscal risks, the Government noted the need to review its exposure to risk, and
set out how it assesses individual risks, but was less concrete about assessing aggregate
exposure. It explained that it would deal with future shocks via macroeconomic stabilisation
policy, and claimed that productivity-boosting measures would help tackle slow-building
pressures, such as the costs of ageing and non-demographic cost pressures in health.
10
These are documented in an online ‘risk register’ published on our website.

Introduction
1.44 The Government provided less assurance on how it would address these issues while
negotiating Brexit (although it discussed its Brexit plans themselves in some detail). On the
issue of apparent ‘austerity fatigue’, it argued that it had “reformed its fiscal framework to
enable it to more actively support the economy” and pointed to the slower planned pace of
consolidation from 2015-16 onwards. It had more to say on the fiscal position from which
all this is faced, which it used to motivate its plans to reduce public sector debt. Perhaps
understandably, it had almost nothing to say on sources of risk we had not analysed.
1.45 With the exception of climate change, which we now cover in Chapter 9, the overarching
issues we raised two years ago remain pertinent to the Government’s next response.
Structure of the report

1.46 We use the analytical framework set out above to structure the report as follows:
• Chapter 2: considers macroeconomic risks;
• Chapter 3: assesses financial sector risks;
• Chapter 4: analyses specific revenue risks;
• Chapter 5: discusses specific non-interest spending risks;
• Chapter 6: looks at the balance sheet;
• Chapter 7: discusses debt interest spending and its relationship with economic growth;
• Chapter 8: discusses the potential risks from policy changes;
• Chapter 9: considers how we might analyse fiscal risks from climate change; and
• Chapter 10: details the results of an illustrative fiscal stress test.

2 Macroeconomic risks
Introduction
2.1 Macroeconomic developments constitute one of the largest and highest probability sources
of fiscal risk. Economic shocks come in many shapes and sizes and propagate through the
public finances in complex ways. As elsewhere in this report, our main focus is the various
downside risks to our latest medium-term forecast and to longer-term fiscal sustainability.
2.2 This chapter sets out our current macroeconomic risk assessment, updating our 2017
assessment in the light of new economic data and revisions to historical data, our latest
forecast judgements, and policy changes. It covers:
• risks to the outlook for potential output growth;
• risks associated with cyclical shocks;
• a deeper review of output gap mismeasurement and the public finances;
• risks associated with the composition of GDP;
• risks from sector net lending positions and balance sheets; and
• a summary of our main conclusions and issues for the Government’s response.
2.3 First, we look back at our 2017 discussion and the Government’s 2018 response.
Summary of previous FRR discussion and the Government’s response

2.4 In our 2017 Fiscal risks report (FRR), we highlighted several issues that we felt the
Government might wish to consider when managing its fiscal risks, including:
• the sources of weak post-crisis productivity growth and the risk of this continuing;
• the inevitability of future recessions and the risk of persistent effects from them;
• the different effective tax rates that apply to the various components of GDP;
• the Government’s fiscal exposure to the housing sector in particular; and
• the economic risks associated with Brexit.

Macroeconomic risks
2.5 In Managing fiscal risks, the Government:
• Recognised the importance of long-run productivity growth and listed several policy
measures to raise it, including: providing additional investment through the National
Productivity Investment Fund to areas viewed as important to economic growth;
launching a ‘modern’ industrial strategy committed to increasing R&D investment; and
improving technical education through apprenticeships.
• Highlighted previous reforms to the monetary and fiscal frameworks, enabling policy
to play a more supportive role in future recessions. It also pointed to the new capital
and income framework agreed with the Bank of England, which it felt would “ensure
[the Bank’s] policy credibility even in the most stressed environment”.
• Noted actions taken to address macroeconomic vulnerabilities from high household

and government debt and a large current account deficit. This included use of macro-
prudential policy by the Financial Policy Committee.
• Recognised its growing exposure to the housing sector and noted that the Ministry of
Housing, Communities and Local Government (MHCLG) had taken steps to manage
the credit risk arising from its loan, guarantee and equity loan products.
• Declared that, to maximise the benefits and minimise the potential disruption from
Brexit, the UK would seek a “deep and comprehensive economic partnership” with the
EU. This would include seeking the establishment by the UK and the EU of a free trade
area for goods and the phased introduction of a facilitated customs arrangement.
Risks to potential output growth

2.6 The path of potential output – the amount that can be produced sustainably, abstracting
from the temporary ups and downs of the economic cycle – is the ultimate driver of living
standards and a central determinant of the sustainability of any given set of tax and
spending plans. Potential output can in turn usefully be decomposed into the available
labour and how much output each unit of that labour can produce (i.e. labour productivity).
2.7 It is important to recognise that potential output is a theoretical construct that cannot be
directly observed. Instead, it must be inferred indirectly from other indicators. The same is
necessarily true of the output gap – the difference between output and potential output –
which is our primary measure of the cyclical position of the economy. Risks associated with
its mismeasurement are discussed from paragraph 2.33 onwards.
2.8 Potential labour input is defined as the total hours of work available when the economy is
operating at a sustainable capacity. It is determined by: the size of the adult population; the
fraction of those adults that are participating in the labour market; the proportion of those
that in turn can be employed sustainably; and the average hours worked by those in
employment. There are risks and uncertainties surrounding projections of each of these.

Macroeconomic risks
Adult population growth

2.9 Growth in the adult population increases the potential size of the economy. It has a clear
positive relationship with total tax revenues, but its effect on revenues per head or as a share
of GDP is ambiguous (dependent on the characteristics of the population). The population’s
contribution to revenues is greatest in age groups where employment rates are highest – i.e.
the ‘working-age’ population, aged between 16 and the state pension age, and within that
‘prime age’ adults, since employment rates are lower for young adults (many of whom are
students) and older adults (some of whom leave the labour market before the state pension
age). But population growth also increases the demand for public services. And in an
ageing society like the UK’s, this place upward pressure on spending as a share of GDP.
2.10 The risks to our medium-term forecast from working-age population growth are relatively
small. Excluding the effects of migration, ‘natural change’ is relatively slow-moving and
predictable, with the lower growth over the next five years reflecting a fall in the birth rate in
the late 1990s and early 2000s (Chart 2.1). There are greater uncertainties surrounding our
projection of net migration, especially in relation to the consequences of the UK’s
prospective exit from the EU and related changes to migration policy (see Chapter 10).
2.11 The fiscal consequences of net migration depend on the age and, if working, productivity of
migrants. Recent net migration to the UK has been concentrated among those of working
age, boosting employment and tax revenues proportionately more than spending. We
assume that migrants are on average as productive as natives, although this will not be true
of every migrant. This assumption seems appropriate for the type of net migration to the UK
seen over the past decade, but might not be the case if all inward migrants are subject to a
migration regime focused on skills or earnings.1
Chart 2.1: Contributions of adult population growth to potential output growth

Average contributions to potential output growth (per cent)
1.0
Net migration March
2019 FSR 2018 projection
Natural change forecast
0.8
Total
0.6
0.4
0.2
0.0
-0.2
1979-1983 1989-1993 1999-2003 2009-2013 2019-2023 2029-2033 2039-2043 2049-2053 2059-2063
Source: ONS
1
We tested the assumptions underpinning how we factor the fiscal effects of net migration into our long-term projections in Annex A of
our 2013 Fiscal sustainability report, and returned to the subject in Box 3.4 of the following year’s report. We looked at potentially
relevant considerations for any post-Brexit changes to the migration regime in Brexit and the OBR’s forecasts, OBR Discussion Paper No.3,
October 2018. For external analysis, see Oxford Economics for MAC, Fiscal Impact of Immigration in the UK, 2018.

Macroeconomic risks
Employment rate
2.12 The outlook for the employment rate depends on trends in participation rates, the
equilibrium unemployment rate and the state of the business cycle. Participation rates for
older people have been rising. For people aged 65 or over, the rate has almost doubled
over the past 15 years, from 6.0 to 11.1 per cent. But participation remains much lower
than among working-age adults (for whom the rate is currently 79.2 per cent), so ageing
puts downward pressure on the average participation rate for the whole adult population.
This compositional effect is expected to dominate over the medium term.
2.13 The proportion of those active in the labour force that would be able to find employment
sustainably is governed by the equilibrium unemployment rate – another unobservable
theoretical construct that we need to form a judgement about. Our medium-term
assumption is informed by an assessment of past trends in the actual unemployment rate,
as well as other labour market developments. In October 2018, with wage growth still
muted but actual unemployment having fallen below our previous assumption, we lowered
our estimate to 4.0 per cent of the labour force. We have not changed it since.
2.14 The proportion of the labour force that actually find work depends on the state of the
business cycle. Chart 2.2 shows that fluctuations in the employment rate have typically
followed those in the output gap quite closely, although the post-crisis experience has seen
unusually strong employment growth given our estimate of the more modest pace at which
the output gap closed. This pattern reflects the fact that as output grows, demand for goods
and services increases and thus so does the demand for labour, raising the employment
rate.
Chart 2.2: The employment rate versus the output gap

63 8
March
Employment rate (left-hand axis) 2019
forecast
62
6
Output gap (right-hand axis)
61
4
Per cent of adult population
Per cent of potential output
60
59 2
58 0
57
-2
56
-4
55
54 -6
1971 1975 1979 1983 1987 1991 1995 1999 2003 2007 2011 2015 2019 2023
Source: ONS, OBR

Macroeconomic risks
2.15 One source of risk to both the potential and actual employment rate – both to the upside
and downside – is the impact of government policy: changes in taxes, in- or out-of-work
benefits, active labour market policies, and minimum wages could all affect the proportions
of people that participate in the labour market and find employment.
2.16 Policies announced over the past two years are unlikely to have had a material effect in this
respect. These include rises in the income tax personal allowance and higher-rate threshold
and to universal credit. A potentially more significant announcement relates to the National
Living Wage, where the Chancellor has indicated an ambition to raise it from the planned
60 per cent of median earnings by 2020 to two-thirds at an unspecified date beyond that.
We discussed possible effects of this in Box 3.3 of our October 2018 EFO, while in March
2019 the Government asked Arindrajit Dube to review international evidence in this area.
Average hours worked

2.17 In 1860, full-time workers on average worked more than 60 hours a week. They now work
less than 40.2 In our medium-term forecasts, we typically assume that the historical
downward trend in average hours will continue. But, over the past few years, average hours
have been relatively flat. In November 2017, we moved to assuming that will continue
across the forecast. So, our forecasts are subject to the risk that the downward trend does
reassert itself – or alternatively that average hours start to rise.
2.18 Taken in isolation, the effect of average hours rising rather than remaining flat would be to
improve the public finances. If, however, this reflected a supply response to continued
weakness in real income growth, then the fiscal effect would be less clear. Income tax
receipts are more sensitive to changes in total hours worked when they reflect changes in
average hours than changes in the number of people employed. That is because an extra
hour worked is taxed at the marginal rate – 20, 40 or 45 per cent – whereas an extra
person employed is taxed at their average or effective rate, which will be lower mainly
because of the tax-free personal allowance that currently stands at £12,500.
Potential productivity growth

2.19 Potential productivity is the amount of output produced from each hour of work when
businesses are operating at a sustainable capacity. This can be split into contributions from
capital deepening (investment in more equipment and technology per unit of labour) and
‘total factor productivity’ (the efficiency with which labour and capital are combined to
produce output). Productivity growth is usually the biggest risk we highlight in each EFO.
2.20 Chart 2.3 shows how actual productivity growth (as opposed to the unobservable growth in
potential productivity) has varied over time. While there have been other periods of
weakness, productivity growth has been persistently weak since around the time of the
financial crisis. Since 2008, the annual growth rate of productivity has averaged just 0.3 per
cent, compared to around 2 per cent beforehand. It edged up to 1 per cent in 2017, but fell
back to just 0.5 per cent in 2018.
2
Speech given by Martin Weale, External Member of the MPC, Bank of England: ‘What’s in a week’s work?’, January 2016.

Macroeconomic risks
Chart 2.3: Productivity growth

8
March
Output per hour FSR 2018 projection
2019
7
forecast
Period average
6
4
Per cent
-1
-2
1972 1976 1980 1984 1988 1992 1996 2000 2004 2008 2012 2016 2020 2024 2028 2032 2036
Note: Output data is on a non-oil GVA basis
Source: ONS, OBR
2.21 There are several competing explanations for the post-crisis weakness in productivity
growth. Most commentators assume that it reflects a combination of factors, with views
differing on their relative importance.3 Until November 2015, our medium-term forecasts
assumed that potential productivity growth would return to its pre-crisis average rate of 2.2
per cent by the end of the forecast. In March 2016, we put more weight on the post-crisis
weakness, taking our medium-term assumption down to 2.0 per cent. And in November
2016, we revised it down further to 1.8 per cent to incorporate the expected effects of the
Brexit vote on investment. In November 2017, we revised it down further to reflect the
persistent weakness in productivity growth following the crisis, with trend productivity growth
assumed to recover more gradually, reaching 1.3 per cent at the five-year forecast horizon.
2.22 Despite this much weaker medium-term productivity growth forecast, there are still
considerable risks. It may become apparent that the current period of low productivity
growth is permanent, or, alternatively, productivity could recover more quickly than we
forecast, though that seems unlikely in the near term if Brexit-related uncertainty persists.
Implications for the public finances

2.23 We have identified risks around each of the components of potential output: population,
employment, hours and productivity – with the prospects for productivity growth particularly
uncertain. But what are the implications for the public finances? That depends on how
revenues and public spending respond as GDP and GDP per head rise.
2.24 In our long-term projections, we assume that most thresholds in the tax and benefit systems
rise in line with living standards (specifically, average earnings). Other things being equal,
3
For a fuller discussion see Barnett et al, The UK productivity puzzle, Bank of England Quarterly Bulletin, June 2016.

Macroeconomic risks
this implies no long-term upward or downward trend in receipts or welfare spending as a

share of GDP, which is in line with long-run historical experience though, of course, there
are likely to be plenty of transitory cyclical, policy-related and other swings. The amount the
Government spends on public services is a political choice, but a reasonable benchmark is
to assume that this too will be a roughly constant fraction of GDP (adjusted for the changing
age structure of the population, and particular cost pressures affecting health and adult
social care services, as discussed in Chapter 5).
2.25 All this means that if a downside risk to potential GDP per head were to crystallise over the
long term – say because of continued weak productivity growth – this would reduce both
receipts and spending in cash terms, but would have a smaller effect on them (and on the
gap between them) relative to GDP. The impact on the quality and quantity of public
services would depend on whether the productivity shortfall across the whole economy was
mirrored in those services. This explains why the long-term projections in our FSRs are
relatively insensitive to different productivity growth assumptions. We are all poorer if the
downside risk materialises, in both the private and public goods we consume, but this does
not translate into a significant threat to fiscal sustainability.
2.26 The impact of unexpectedly weak potential GDP growth is greater over our medium-term
forecast, because the Government has set its policy parameters over this horizon and most
are not linked to changes in earnings and GDP per head. For example, public services
spending totals are planned in cash terms and many tax allowances and thresholds rise with
inflation. In this setting, weaker GDP growth reduces cash revenues significantly and
increases cash spending on debt interest and means-tested benefits somewhat. Receipts fall
less as a share of GDP (with both lower, but receipts more so), but spending rises more, as
unchanged cash plans for public services spending are higher as a share of that lower GDP.
Risks from cyclical shocks

2.27 In addition to the fiscal risks arising from the path of potential GDP, it is highly likely that
actual GDP will diverge from potential – from time to time, significantly so – creating further
fiscal risks.
2.28 Cyclical fluctuations in GDP matter fiscally because of their impact on both revenues and
spending. When economic activity weakens, this reduces tax revenues (because tax bases
are smaller), increases welfare spending (with higher unemployment and more households
on low incomes) and also increases the share of GDP devoted to spending on public
services (because it is largely fixed in cash terms, so as a share of unexpectedly weak GDP it
rises). The opposite happens when activity strengthens. Chart 2.4 shows the fluctuations of
the economic cycle over the past 40 years, based on our estimates of the output gap.

Macroeconomic risks
Chart 2.4: The economic cycle and its impact on the budget balance
6
March
Output gap 2019
forecast
4
2
Per cent
-2
-4
-6
10
Cyclical deficit
Structural deficit
8
Public sector net borrowing
6
Per cent of GDP
-2
-4
1975-76 1979-80 1983-84 1987-88 1991-92 1995-96 1999-00 2003-04 2007-08 2011-12 2015-16 2019-20 2023-24
Source: ONS, OBR
2.29 Based on the average relationship between the output gap and the budget balance since
the 1970s, we assume that for each 1 per cent that activity falls below potential, the budget
deficit worsens by 0.5 per cent of GDP in the same year and by an extra 0.2 per cent in
following year.4 Most of the deterioration comes about through spending rising as a share
of GDP (because reasonably stable cash spending rises relative to weaker GDP), with
receipts falling slightly as a share of GDP (because they weaken slightly more than GDP).
4
Helgadottir et al, Cyclically adjusting the public finances, OBR Working Paper No. 3, 2012.

Macroeconomic risks
2.30 The average cyclical balance in Chart 2.4 is a deficit of 0.3 per cent of GDP – reflecting the
fact that, almost by definition, cyclical surpluses and deficits balance out over time. But the
average absolute cyclical balance is 0.9 per cent of GDP. This suggests that cyclical
movements in the economy can pose significant risks to the fiscal position over a medium-
term horizon, but not over longer periods when they tend to wash out. In practice, however,
effects on the public finances can and do persist as:
• Cyclical shocks can have a lasting impact on the future path of potential output
through their impact on investment or scarring effects in the labour market, so could
be associated with any of the risks described in the previous section of this chapter.
• Cyclical deficits during downturns generate persistent additions to public debt. Chart
2.4 also shows that the cyclical component of the deficit persists. For example,
cumulative cyclical borrowing in the decade from 2008-09 to 2017-18 totalled 8.4
per cent of GDP. The effect on the debt-to-GDP ratio will be amplified if cycles are
skewed to the downside or if the fiscal benefits of booms are not as big as the costs of
recessions. Non-linearities in the tax and benefit system can also play a role. A striking
example is the triple lock on state pensions, which means that spending rises as a
share of GDP in a recession but is very unlikely to fall as a share of GDP in a boom.
• A cyclical shock can have permanent effects on prices and nominal GDP even if real
GDP returns to the same potential path after the shock has dissipated. This can have
long-term implications for the public finances, with the extent influenced by the type of
inflation caused by the shock. A shock that generates higher inflation that then feeds
through into wages will be better for the public finances than one that has little effect
on wages. This is due to ‘fiscal drag’ on income tax receipts, as wage growth increases
income tax receipts by more than the indexation of thresholds reduces them.5
Cyclical shocks in UK post-war history

2.31 The ONS publishes consistent quarterly real GDP data from 1955. Defining a recession as
at least two consecutive quarters of falling output, the latest vintage shows seven recessions
in the past 63 years. That implies the chance of being in recession at some point in any
given five-year period – the horizon for our medium-term forecasts – is around one in two.6
The recessions differed in length, depth and duration of the subsequent recovery (Table
2.1). There have been four ‘major’ recessions in the past 50 years: in the 1970s (with the
two periods in the 1970s that meet the ‘two consecutive quarters’ definition considered
together), the early 1980s, the early 1990s and the late 2000s.
5
See Chapter 5 of our March 2018 Economic and fiscal outlook, in which we considered two scenarios that generated higher inflation.
6
This is the cumulative probability of a recession occurring in at least one of the five years. This is based on the probability of a recession
in any given year (11 per cent) and a Bernoulli distribution, assuming that probability of a recession is independent in each year. The
probability would be sensitive to changes in the average growth rate, since, for a given output variability, lower average growth would
increase the probability of it falling below zero.

Macroeconomic risks
Table 2.1: UK recessions since 1955

Number of quarters unless otherwise stated
Mid Early Early Mid Early Early Late
1950s 1960s 1970s 1970s 1980s 1990s 2000s
First quarter of falling output 1956Q2 1961Q3 1973Q3 1975Q2 1980Q1 1990Q3 2008Q2
Number of quarters till trough 2 2 3 2 5 5 5
Peak-to-trough fall in output (per cent) 0.4 0.7 4.1 2.0 4.2 2.0 6.3
Quarters for output to regain pre-
1 2 11 2 8 6 16
recession peak from trough
2.32 In three of those four recessions, the consequences for tax revenues and spending were
sufficient to push the deficit above 6 per cent of GDP. The exception was in the early 1980s,
when fiscal tightening was itself a factor contributing to the recession. Other than in the
most recent recession, the effect on public debt relative to nominal GDP was relatively
modest. One reason was that higher inflation boosted nominal GDP around the other three
recessions (indeed tightening macroeconomic policy to squeeze inflation out of the system
was a contributor to two of them), whereas nominal GDP fell in the late-2000s recession.
Box 2.1: Is the UK entering recession?

History suggests there is around a one-in-two chance of a recession in any five-year period. The
ONS estimates that GDP fell on a monthly basis in April, after weak growth in March, and only
partly rebounded in May. The National Institute of Economic and Social Research’s (NIESR)
monthly GDP tracker published on 10 July, which uses survey indicators and past trends in the
data, suggests that the economy shrank by 0.1 per cent in the second quarter of 2019.a The all-
sector purchasing managers index (PMI) readings for April to June are also consistent with GDP
contracting by 0.1 per cent in the second quarter according to IHS Markit (Chart A).b So, does
this foretell a full-blown recession? Or might transient factors be to blame?
If the ONS data released in August do show the economy shrinking in the second quarter, it is
likely to be in part a ‘pay-back’ from the strength of Brexit-related stock building in the first
quarter. Indeed, the 0.5 per cent growth in that quarter was more than explained by inventory
accumulation. Manufacturing also grew strongly, which is consistent with businesses increasing
stocks as a precaution against any potential disruption to supply chains. Higher stock building
would have been growth-neutral if the goods being stockpiled were all imported, but it is
possible that it boosted domestic output too, for example in the distribution and warehousing
sectors, and exports might have benefited if EU firms stockpiled UK goods too.c The volatility of
monthly GDP growth in April and May was driven by car production, as the car industry had
brought forward planned annual maintenance shutdowns to coincide with Brexit and minimise
any related disruptions, causing output to plummet in April and rebound in May. But
manufacturing has also weakened globally, reflecting factors such as the introduction of tariffs
by the US and China.

Macroeconomic risks
Chart A: Monthly and quarterly GDP growth

1.0
Monthly GDP growth Monthly growth implied by PMIs Quarterly GDP growth
0.8
0.6
0.4
Per cent
0.2
0.0
-0.2
-0.4
-0.6
Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2
2016 2017 2018 2019
Source: ONS, OBR
If Brexit-related factors were the only ones behind the relative strength of the first quarter and the
possible contraction that followed, the economy might be expected to return to growth in the
third quarter. But the PMI surveys were particularly weak in June, suggesting that the pace of
growth is likely to remain weak. NIESR’s initial estimate for the third quarter is weak GDP growth
of 0.2 per cent. Still, this suggests there is a risk that the economy could enter a technical
recession (i.e. two consecutive quarters of falling output).
The fiscal risks associated with recessions depend on their depth and persistence, the sectors
most deeply affected, and the pace at which the economy subsequently recovers. The stress test
presented in Chapter 10 illustrates the fiscal consequences of a recession following a no-deal
Brexit, consistent with a scenario published by the IMF earlier this year.
a
NIESR Monthly GDP tracker: July 2019, press release.
b
IHS Markit/ CIPS UK services PMI, July 2019 press release.
c
See Box 3 in the Bank of England, May 2019 Inflation Report.
Output gap mismeasurement and the public finances

2.33 Judgements about the current size, and prospects for, the output gap are important drivers
of our economic forecasts, as well as determining our assessment of the structural fiscal
position. But, as noted earlier, the output gap is a theoretical construct that cannot be
observed directly, and must instead be inferred from other indicators. This leaves much
uncertainty surrounding output gap estimates, with revisions frequent and sometimes large.
2.34 Broadly speaking, approaches to measuring the output gap fall into two camps:

Macroeconomic risks
• The first involves estimating potential output first, allowing the output gap to be
calculated by residual from data on actual output. The simplest approach assumes that
potential output evolves smoothly and then uses statistical filters to separate actual
output into a smooth potential component and a residual, often volatile, output gap. A
more sophisticated approach relates potential output to its underlying drivers.
• The second involves estimating the output gap directly using either measures of slack
(such as unemployment and survey measures of capacity utilisation), or else
observable consequences of an output gap (such as movements in inflation). In this
approach, potential output can then be deduced using data on actual output.
We employ measures utilising different variants of both approaches in forming our overall
judgement regarding the paths of potential output and the output gap over the past.7
2.35 Uncertainty surrounding estimates of the output gap arises from three main sources:
• End-point uncertainty. Some estimation methods, notably statistical filters that

assume potential output evolves smoothly, employ data at earlier and later dates
when calculating potential output at any given date. This renders them less reliable
at the beginning and end of the sample. But, of course, the size of the output gap at
the end of the sample is precisely what we are most interested in.
• Data uncertainty. Data are imperfect and are prone to revision as more information
becomes available and measurement methods evolve.8 Output gap estimates
derived from some approaches will be more likely to change than others, depending
on their underlying source of information. For instance, output gap measures that
rely exclusively on survey data on economic slack will be less prone to revision.
• Model uncertainty. The estimation models themselves will evolve in line with changes
to our understanding of how the economy functions. Generally speaking, this is
more likely to be an issue for methods that embody a richer economic structure.
Earlier work showed that while data revisions did play a part in changes in our assessment
of the output gap, rather more was down to the inclusion of new data.9
2.36 Revisions to output gap estimates can be substantial, as Table 2.2 illustrates. The IMF’s real-
time estimate of the output gap in 2007 from its April 2008 World Economic Outlook
(WEO) was 0.4 per cent, but by the time of its April 2019 WEO it had been revised to 3.8
per cent. Similarly, in June 2008 the OECD estimated that there was an output gap of 0.2
per cent in 2007, but by its May 2019 publication it had been revised to 3.1 per cent. The
Treasury’s estimate in Budget 2008 was 0.3 per cent, in contrast to our March 2019
7
See OBR Working paper No.5: Output gap measurement: judgement and uncertainty.
8
This is such a common theme in explaining changes in our economic forecasts that we have devoted part of our website to the boxes
included in our various reports on the theme of economic data revisions. See the ‘box sets’ area of our website.
9
See OBR Working paper No.5: Output gap measurement: judgement and uncertainty.

Macroeconomic risks
estimate of 1.4 per cent. Chart 2.5 shows that size of revisions to output gap estimates can
be particularly large in downturns.
Table 2.2: Real-time and latest output gap estimates ahead of the 2008 recession
Estimates of the output gap in the UK in 2007 (per cent of potential output)
IMF 1 OECD2 OBR3
Real-time estimate -0.1 -0.5 0.3
Latest estimate 2.3 1.3 1.4
Revision (percentage points) 2.4 1.8 1.1
Note: numbers may not sum due to rounding.
1
IMF real time estimate from April 2008 World Economic Outlook.
2
OECD real time estimate from June 2008 Economic Outlook.
3
HM Treasury real time estimate, as of March Budget 2008.
Chart 2.5: Real-time and latest output gap estimates

3
0
Per cent of potential output
-1
-2
-3
-4
-5
Real-time estimate
-6
Latest estimate
-7
1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018
Source: OBR
2.37 The sign of the output gap estimate can change as well as its magnitude. When this
coincides with a material change in terms of size too it can be particularly challenging for
policymakers, as it implies a switch in the appropriate direction of macroeconomic policy –
shifting decisively from the accelerator to the brake, or vice versa. It may be more
appropriate to base an assessment of whether policymakers’ decisions were pro- or
counter-cyclical (i.e. amplifying or attenuating the economic cycle) on the vintage of data
available to policymakers at the time (i.e. the real-time data).10
2.38 Revisions to output gap estimates can affect views about the outlook for future potential
growth. When a recession occurs, it typically prompts a reassessment of the sustainability of
10
See Orphanides, Monetary Policy Rules Based on Real time Data, American Economic Review, 2001, and Cimadomo, Fiscal Policy in
Real Time, European Central Bank Working paper series No.9, 2008.

Macroeconomic risks
pre-recession growth and a lowering of estimates of past and future growth in potential
output. For some estimation methods, this effect can follow as a direct result of the way in
which they are constructed. For example, statistical filtering methods are particularly
sensitive to changes in recent data. So, if GDP growth weakens markedly, the filter will imply
that the structural component of recent growth has weakened too. It is partly for this reason
that we typically place less weight on filter estimates, and, given that no estimate is perfect,
we use a range of evidence.
2.39 Chart 2.6 shows three vintages of Treasury and OBR estimates and forecasts for potential
output, each starting five years before the financial crisis broke. It shows successive
downward revisions to both the level and the slope (i.e. potential growth) over time. This
general pattern is not unusual after recessions, but its scale and persistence has been.
Chart 2.6: Selected potential output growth estimates and forecasts

140
135
130
125
2003-04 = 100
120
115
110
March 2008 (HM Treasury)
105
March 2010 (HM Treasury)
100
June 2010 (OBR)
95
March 2019 (OBR)
90
2003-04 2005-06 2007-08 2009-10 2011-12 2013-14 2015-16 2017-18 2019-20 2021-22
Source: OBR
Output gap estimates and the structural fiscal position

2.40 The extent to which real-time output gap estimates could influence a government’s intended
fiscal stance would depend on the constraints it has set itself, and the weight it places on the
output gap. The Government’s current fiscal mandate target measure is cyclically-adjusted
public sector net borrowing (CA-PSNB), so it has assigned a significant role to our estimate
of the output gap and the sensitivity of borrowing thereto.
2.41 CA-PSNB is constructed using PSNB and output gap estimates.11 Both are uncertain and are
revised, and both play a significant role in revisions to the structural fiscal position. Overall,
the real-time and latest estimates of CA-PSNB move broadly in line with each other, though
11
CA-PSNB=PSNB + 0.5*Gap t + 0.2*Gapt-1, for more on this see OBR Working Paper 3, Cyclically adjusting the public finances.

Macroeconomic risks
the two lines do not match entirely (Chart 2.7).12 (Cyclical adjustment coefficients – the
sensitivity of borrowing to the output gap – could change too, but did not over this period.)
Chart 2.7: Real-time and latest estimates of cyclically adjusted PSNB

10
Real-time estimate
Latest estimate
8
Per cent of nominal GDP
-2
1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 2015 2017
Source: OBR
2.42 Breaking down CA-PSNB revisions by source, the part owing to revisions to output gap
estimates are sizeable, though revisions owing from changes to PSNB played their part,
especially following the financial crisis (Chart 2.8). PSNB can be revised as, for example,
departmental spending forecasts are eventually replaced with outturn data. It is also subject
to methodological and classification changes that can happen much later. And, expressed
as a share of nominal GDP, it will also change as nominal GDP is revised for similar
reasons.
12
Real-time estimates are based on the information policymakers had at the time. For example, in 2007-08, CA-PSNB would be
calculated using estimates for the output gap as at 2007-08. CAPSNBt2007-08=PSNBt2007-08 + 0.5*Gap t2007-08 + 0.2*Gapt-12007-08.

Macroeconomic risks
Chart 2.8: Revisions to cyclically adjusted PSNB by source

4
Revision owing to output gap revisions Revision owing to PSNB revisions
Revisions owing to nominal GDP revisions Total revisions
3
2
Percentage points
-1
-2
-3
1995-96 1997-98 1999-00 2001-02 2003-04 2005-06 2007-08 2009-10 2011-12 2013-14 2015-16 2017-18
Source: OBR
2.43 Given that our estimate of the output gap is liable to change in the light of new information,
it is clear that estimates of the structural position of the public finances are similarly
uncertain, and will be especially so around recessions.
Structural deficit uncertainty and the fiscal policy stance

2.44 A policy maker may choose to respond to sufficiently large positive or negative output gaps
by setting policy to counteract the cycle. Assuming that the automatic fiscal stabilisers are
insufficiently strong to achieve the desired stabilisation on their own, an additional
discretionary fiscal impulse would be necessary. So when there is spare capacity (a negative
output gap), we might expect to see expansionary discretionary fiscal policy (top left
quadrant of Charts 2.9 and 2.10). And when the economy is operating above potential (a
positive output gap) then the policy maker would set contractionary fiscal policy (bottom
right quadrant of Charts 2.9 and 2.10). In other words, if the fiscal stance is ‘appropriate’,
the points should lie in the bottom right or top left quadrants.
2.45 Policy makers unfortunately do not have the luxury of hindsight when making policy. So the
response of the fiscal stance to the cycle should really be assessed based on the real-time
information policy makers had at the time. Chart 2.9 shows the data on a real-time basis.
In this case, almost all the points are in the top left or bottom right quadrant, suggesting
that discretionary policy was indeed generally counter-cyclical in its intent.

Macroeconomic risks
Chart 2.9: Cyclically adjusted PSNB and the output gap on the real-time data
10
Spare capacity, Above potential,
Expansionary fiscal policy Expansionary fiscal policy
8
CAPSNB (per cent of nominal GDP)
-2
Contractionary fiscal policy Contractionary fiscal policy
-4
-7 -6 -5 -4 -3 -2 -1 0 1 2
Source: OBR, ONS Output gap (per cent of potential output)
2.46 Our analysis suggests output gap revisions can be substantial, however, particularly in the
later stages of the business cycle when there is a tendency to overestimate the amount of
slack in the economy. Consequently, a policy setting that seems counter-cyclical in real time
may look less counter-cyclical when judged with historical hindsight. That is illustrated in
Chart 2.10, which shows a considerable number of points lying in the top right quadrant,
where a discretionary fiscal expansion was applied when the economy – with hindsight –
was already operating above potential.
Chart 2.10: Cyclically adjusted PSNB and the output gap on the latest data
10
Expansionary fiscal policy Expansionary fiscal policy
8
CAPSNB (per cent of nominal GDP)
-2
Contractionary fiscal policy Contractionary fiscal policy
-4
-7 -6 -5 -4 -3 -2 -1 0 1 2
Source: OBR, ONS Output gap (per cent of potential output)

Macroeconomic risks
What fiscal risks arise from output gap uncertainty and mismeasurement?
2.47 It is clear that output gap mismeasurement can have consequences for Chancellors’ fiscal
policy choices. What looks like wise counter-cyclical fiscal policy in real time might not do so
with the benefit of hindsight. In particular, it is a feature of some models used to estimate
output gaps that recessions tend also to lead to downward revisions in the pre-recession
path of potential output (and thus also its prospective future path). This matters as it is the
outlook for potential output that underpins judgements about long-run fiscal sustainability.
2.48 Uncertainty surrounding current estimates of the output gap means our central estimate
could provide the Chancellor with a misleading steer as to both the margin of spare
capacity in the economy and the size of the structural deficit, affecting any decisions about
how much to adjust the fiscal stance. We try to address this risk by presenting fan charts
around our forecasts for the CA-PSNB, illustrating the extent to which history would suggest
outturns might differ from our central forecast. We also place our central estimate in the
context of those produced by other organisations – in our March 2019 EFO estimates for
2019 ranged from minus 1.7 per cent to plus 0.8 per cent.
2.49 In principle this risk could be symmetric, but evidence around recessions is that estimates
can worsen significantly as forecasters tend to conclude that they were over-optimistic ahead
of the recession. To put the size of this risk in context, in real-time the Treasury, IMF and
OECD all estimated there to be an output gap relatively close to zero immediately prior to
the 2008 recession. The average of latest estimates is now 2.7 per cent. This represents a
change in the estimated size of the structural deficit in that year of around 1.2 per cent of
nominal GDP – around £26 billion in today’s terms.
Risks from the composition of GDP

2.50 The composition of GDP can be important to the fiscal forecast, because some components
generate more tax receipts per pound than others (i.e. they are more ‘tax rich’). Chart 2.11
illustrates the tax-richness of different income and expenditure components of GDP by
assigning various taxes to each: for example, income tax to labour income or VAT and
excise duties to consumer spending. Not all taxes relate to income or expenditure
components of GDP – in particular, those that relate to disposals or transfers of assets
(capital gains tax or inheritance tax) or balance sheets (the bank levy). These have been
excluded. The chart illustrates the particular fiscal importance of labour income and
consumer spending, both of which are large components of GDP and relatively tax-rich.13,14
13
We have split onshore corporation tax between a negative capital allowance element assigned to investment and a notional pre-capital
allowance element assigned to corporate profits. We have not included use of North Sea capital allowances in this estimate. Almost all
North Sea investment is subject to immediate 100 per cent capital allowances, but the effect on tax receipts depends on the proportion of
that investment undertaken by firms with tax liabilities that can be offset. This is subject to significant uncertainty and varies across years.
14
The effective tax rate calculations underpinning this chart reflect static, average effects in one year. They do not attempt to capture
interactions between components or longer-term dynamic effects – like the effect of higher business investment on potential output.

Macroeconomic risks
Chart 2.11: Selected components of GDP and associated effective tax rates
30
Labour income
25
20
Corporate profits Consumer spending

15
Effective tax rate
10
-5
Business investment
-10
0 10 20 30 40 50 60 70
Share of nominal GDP (per cent)
Source: OBR
Risks associated with the expenditure composition of GDP

2.51 Key risks from the expenditure composition of GDP include:
• Household consumption makes up 66 per cent of nominal GDP by expenditure, and it

is also relatively tax-rich. It accounts for around 70 per cent of VAT receipts. Owing to
its size, relatively small differences between forecast and actual consumption growth
can be fiscally material. Our latest ready reckoners suggest that a 1 per cent shortfall
in consumption relative to forecast would reduce receipts by £¾ billion, while a 1
percentage point fall in the consumption share of GDP – offset by a rise in business
investment – would lower the tax-to-GDP ratio by 0.2 percentage points.
• Business investment makes up a much smaller share of GDP than private

consumption, but is more volatile. In the medium term, higher business investment
reduces tax receipts due to the use of capital allowances. Our ready reckoners suggest
that the direct effect of a 1 per cent rise would reduce receipts by around £0.1 billion
by the end of the forecast. The sensitivity in the short term is greater due to the annual
investment allowance having been temporarily set at £1 million. But the indirect effect
of higher business investment is likely more important in the longer term, boosting
receipts via its effect on potential output growth.
• The UK does not impose export taxes and the customs duties it currently collects are
treated as EU taxes. They will become UK taxes after Brexit, but until the terms of Brexit
or any changes to customs policies are known, any receipts sensitivities are unknown.

Macroeconomic risks
Risks associated with the income composition of GDP

2.52 The two most important income components for the public finances are labour income and
corporate profits, with the former significantly larger and more tax-rich than the latter:
• Wages and salaries make up the largest component of household income. Income tax
and National Insurance contributions (NICs) are the largest taxes on that income – and
the largest sources of total tax receipts. A 1 per cent fall in wages and salaries would
reduce PAYE income tax and NICs receipts by about £3¾ billion in the first year. We
have typically assumed that wages and salaries will be broadly stable as a share of
GDP over the medium term, as they have been in recent decades (Chart 2.12). If the
share were to fall, the tax-to-GDP ratio would fall too. A 1 percentage point fall in the
labour share of income (weighted equally between earnings and employment),
compensated by a corresponding rise in the profit share, would be associated with a
0.1 percentage point drop in the tax-to-GDP ratio. The distribution within labour
income also matters fiscally because of the progressivity of the income tax system.
• A 1 per cent fall in self-employment income15 would reduce self-assessment receipts by

£¼ billion, with a one-year lag. This is much smaller than an equivalent hit to wages
and salaries, which are over six times the size of self-employment income, and also
more heavily taxed (especially through NICs). There has been a long-term trend of a
rising share of workers being self-employed (and a related sharp rise in the number of
people setting themselves up as a single-director company, as discussed in Chapter 4).
We assume the share of self-employment continues to rise in our medium-term
forecast, but at a slightly slower pace than over the past couple of decades.
• Corporate profits are subject to a lower effective tax rate than labour income. Evidence
suggests that profit margins are positively correlated with the economic cycle,16 and
our forecast of the path of the profit share of GDP is partly informed by our output gap
forecast. Positive (negative) cyclical shocks may therefore be associated with a higher
(lower) profit share, reducing (increasing) the aggregate effective tax rate, depending
on how other elements of income evolve. So, an economic shock that mainly affects
the corporate sector will have smaller fiscal consequences than a similar one that
mainly affects the household sector – as the results of our latest stress test illustrate.
15
Recorded in the household income category of ‘mixed income’ in the National Accounts.
16
Macallan et al, The cyclicality of mark-ups and profit margins for the United Kingdom: some new evidence, Bank of England Working
Paper No. 351, 2008.

Macroeconomic risks
Chart 2.12: Labour share of GDP by income

70
March
Labour income Wages and salaries and mixed income 2019
forecast
65
60
Per cent of GDP
55
50
45
40
1970 1974 1978 1982 1986 1990 1994 1998 2002 2006 2010 2014 2018 2022
Source: ONS
Risks associated with the housing sector

2.53 The housing sector accounts for more than two thirds of stamp duty land tax (and equivalent
taxes in Wales and Scotland), more than a third of inheritance tax and a sixth of capital
gains tax. It also accounts for small – but still significant – shares of many other tax bases.
2.54 Housing-related risks to receipts can be illustrated by what happened during and after the
financial crisis: SDLT receipts from residential properties fell from £10.0 billion in 2007-08
to £4.8 billion in 2008-09, as the number of property transactions halved. The level at
which property transactions are taxed also depends on the house price and the type of
purchaser, with residential SDLT revenue highly dependent on a small number of high-
priced transactions (see Chapter 4).
2.55 Housing market shocks are often correlated with other shocks in the economy. House prices
fell in real terms around each of the last four recessions (Chart 2.13). Housing shocks also
have wider indirect effects. For example, consumption and house prices are highly
correlated because they are affected by common factors, including income expectations and
credit conditions. There may also be causal effects from housing to consumption, for
example because housing wealth can be used as collateral for borrowing.17
17
Benito et al, House prices and consumer spending, Bank of England Quarterly Bulletin, June 2006.

Macroeconomic risks
Chart 2.13: Growth of real GDP and real house prices

40 12
Real GDP growth (RHS) Real house price inflation (LHS)
30 9
Percentage change on a year earlier

Percentage change on a year earlier
20 6
10 3
0 0
-10 -3
-20 -6
-30 -9
1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 2011 2014 2017
Source: ONS, OBR
2.56 Other features of the housing market, such as the proportion of owner-occupiers, also affect
public spending. While less volatile from year to year, home ownership rates were on a
downward trend from 2005 to 2016, before apparently stabilising.18 The proportion of
households renting their home – and within that the proportions in the private- and social-
rented sectors – is a key driver of spending on housing benefit, which we estimate will cost
£22.6 billion in 2018-19 (including equivalent payments made via universal credit).
2.57 There are several government-backed schemes to encourage housing supply and home
ownership, which are individually and collectively a source of fiscal risk as they increase the
exposure of the public sector balance sheet to housing market risks (see Chapter 6).
Risks associated with sectoral lending and balance sheets

2.58 The National Accounts framework underpinning our economic forecast allows us to forecast
each sector’s net lending or borrowing from the other sectors. In principle, these should sum
to zero – for each pound borrowed there must be a pound lent. As each sector’s net lending
follows arithmetically from our forecasts of income and expenditure, the profiles provide an
important overall diagnostic on the coherence of our economic forecast. They can also
highlight risks around it:
• The overall position across the four sectors shows how we expect the public sector
deficit to narrow slightly, with that narrowing offset by a modest decline in the rest of
the world surplus (a narrowing current account deficit) while the corporate and
household sectors are expected to run persistent but stable deficits (Chart 2.14).
18
MHCLG, Live tables on dwelling stock (including vacants): Table 102, May 2019. Excludes Northern Ireland.

Macroeconomic risks
• The historically large current account deficit, which means overseas investors remain
significant net lenders to the UK, poses risks. If investors’ confidence in the UK
economy were damaged, this could lead to a sharp fall in sterling causing an abrupt
demand-led narrowing of the current account deficit and a spike in inflation. That
said, the UK’s net international investment liabilities are relatively modest as a share of
GDP, mitigating these risks somewhat. Estimates of the current account balance, which
on the current data vintage have averaged a deficit of 4.5 per cent of GDP since
2012, are also subject to frequent revision.
• A persistent household deficit has implications for household debt – high debt could
cause households to retrench spending more than usual if there were a negative
shock, which would amplify the effects of a slowdown. Household debt remains high
by historical standards relative to income, though below pre-crisis levels (Chart 2.15).
We expect it to rise modestly relative to income over our forecast, but this is largely
driven by student loans (as currently recorded in the National Accounts).19
Chart 2.14: Sectoral net lending

8
March 2019 forecast
6
4
Per cent, rolling annual average
-2
-4
-6
-8
-10
Corporate Public Household Rest of world Statistical discrepancy
-12
2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022 2024
Source: ONS, OBR
19
See Box 3.4 of our October 2018 Economic and fiscal outlook for further discussion

Macroeconomic risks
Chart 2.15: Household gross debt to income

180
March 2019
forecast
170
160
150
140
Per cent
130
120
110
100
90
80
1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 2015 2017 2019 2021 2023
Source: ONS, OBR
Conclusions
2.59 This chapter has highlighted various ways in which macroeconomic risks can affect the
public finances. History suggests that these are some of the highest impact fiscal risks most
likely to crystallise over the medium and longer term.
2.60 Our assessment is that the likelihood of, and impact from, most of these macroeconomic
risks crystallising is little changed relative to our 2017 FRR. There is still a one-in-two chance
of a recession in any five-year period, though of what size is uncertain. The timing of any
recession is also uncertain, with the latest data suggesting the UK economy may have
contracted in the second quarter of 2019 but implications for the third quarter unclear.
Downside risks to productivity growth remain a concern given the continuation of its
weakness since we published our 2017 FRR. We revised down our medium-term
assumption in November 2017, but we have not changed our long-term one. But the longer
the weakness in productivity growth persists, the more likely we are to revise that down too.
2.61 In this report we have looked more closely at the potential fiscal risks arising from
uncertainty around and mismeasurement of the output gap. This shows that real-time
estimates of the output gap are prone to significant revisions, particularly in the lead-up to
recessions. That in turn can lead to inappropriate fiscal policy choices.

Macroeconomic risks

2.62 In this chapter we have highlighted several issues that the Government is likely to wish to
consider when managing its fiscal risks. Among them:
• the sources of weak post-crisis productivity growth and the risk of this continuing;
• the inevitability of future recessions and the risk of persistent effects from them;
• the uncertainty around real-time output gap estimates and its policy implications;
• the persistent current account deficit and households’ financial position; and
• the Government’s fiscal exposure to the housing sector in particular.
2.63 When assessing the macroeconomic outlook and its interaction with fiscal risk over the
medium and long term, does the Government regard these or other issues as important for
its risk management strategy and, if so, how does it intend to address them?

3 Risks from the financial sector
Introduction
3.1 The financial sector is a prominent part of the UK economy and has a significant impact on
the public finances. It is both a direct source of employment and tax revenue and also
facilitates transactions, savings and investment, and insurance/risk transfer in the wider
economy. But financial crises can have a significant adverse effect on the public finances. In
our 2017 Fiscal risk report (FRR), we noted several fiscal risks arising from the financial
sector. In this chapter, as well as revisiting those risks, we also look at the risks posed by
‘shadow banking’ – in broad terms, those entities that act like banks – undertaking maturity
and liquidity transformation and extending credit – but which are not regulated as banks.
3.2 This chapter:
• describes the key characteristics of the financial sector;
• reviews evidence on financial crises and their fiscal impact;
• assesses current and future fiscal risks from the financial sector;
• looks in more depth at risks posed by shadow banking; and
• identifies some issues for the Government’s response.
Characteristics of the financial sector

3.3 Despite shrinking somewhat in the aftermath of the financial crisis, the financial sector
remains prominent in the UK economy and contributes significantly to the public finances:
• Banks’ assets relative to nominal GDP have risen more than fourfold since the 1970s
and in 2017 stood at around 5½ times GDP (Chart 3.1).1 In absolute terms, this
makes the UK’s banking sector the third largest among major OECD countries after
the US and Japan. But, relative to the size of the economy, it is more than four times
bigger than that in the US and somewhat larger than that in Switzerland.2
• Value added by the finance and insurance industry, as defined in the National
Accounts, accounted for around 7 per cent of national output in 2018, down from a
peak of 9 per cent in 2009 but still above the EU average of around 5 per cent.3
1
OECD Stats, Financial balance sheets non-consolidated and GDP output approach, data extracted in May 2019.
2
To the extent that the domestic banking is now insulated from international banking through ring-fencing, that may somewhat overstate
the UK’s vulnerability to a crisis.
3
ONS, UK GDP(O) low level aggregates, May 2019 for the UK and OECD, Value added by activity, 2018 for the EU average.

Risks from the financial sector
• As of December 2018, finance and insurance accounted for 1.1 million jobs or 3 per
cent of the total.4
• The UK trade surplus in financial and insurance services stood at £61 billion in 2018,
equivalent to around 3 per cent of GDP. The UK recorded an overall trade deficit in
2018 of £31 billion – illustrating the sector’s importance to the balance of payments.5
Chart 3.1: Banks’ assets relative to nominal GDP

Luxembourg
United Kingdom
Switzerland
Japan
France
Ireland
Denmark
Netherlands
Sweden
Germany
Spain
Portugal
Italy
Belgium
Austria
Finland
Norway
Greece
Latvia
Estonia
United States
Slovak Republic
Slovenia
Turkey
Hungary
Poland
0 250 500 750 2250 2500
Per cent of GDP in 2017
Note: Data for the United States are as a per cent of nominal GDP in 2016.
Source: OECD
3.4 Chart 3.2 illustrates the composition of financial systems across developed economies on
the left and the composition of the UK financial system over the past 15 years on the right.
3.5 Banks are the largest component of the UK financial sector by assets (at almost half in
2017), followed by other financial intermediaries (OFIs, which accounted for almost a third
of assets in that year). Insurance corporations, pension funds and the central bank are all
much smaller by comparison. And while banks’ assets roughly doubled in size relative to
GDP between 2002 and 2017, OFIs’ assets more than trebled over the same period.6
4
ONS, Workforce jobs by industry, March 2019.
5
ONS, UK trade: May 2019, July 2019.
6
The jump in banks’ assets in 2008 relates to the recording of derivative assets.

Chart 3.2: Assets of financial institutions in advanced economies

Advanced economies in 2017 United Kingdom
2,000 2,000
Other financial intermediaries
1,800 1,800
Pension funds
1,600 1,600
Insurance corporations
1,400 1,400
Per cent of GDP
Public financial institutions

1,200 1,200
Central banks
1,000 1,000
Banks
800 800
600 600
400 400
200 200
0 0
IE NL HKUKCHSG JP EA CA FR BE US AU DE KR ES IT 2002 2004 2006 2008 2010 2012 2014 2016
Note: We have excluded the Cayman Islands and Luxembourg due to high values.
Source: FSB
Direct implications for the public finances

3.6 The financial sector accounts for a disproportionately large share of total receipts, though
less so than in the pre-crisis period. This reflects both the high average pay rates in the
sector and differences in how financial sector companies are taxed relative to others:
• In 2017-18, the sector accounted for only 3 per cent of total jobs,7 but for around 8
per cent of total pay8 and 12 per cent of PAYE income tax and NICs receipts.
• In 2016-17, the sector accounted for 25 per cent of total corporation tax receipts.9
• According to a report produced for the City of London Corporation, total tax paid by
financial sector companies and their employees was £75 billion in 2017-18,
equivalent to 10.9 per cent of total UK tax receipts.10 On that metric, reliance on the
financial sector for receipts has fallen from 13.9 per cent of total in 2007-08.
3.7 Looking specifically at the banking sector, in 2017-18 it accounted for 6.7 per cent of PAYE
receipts and 12.1 per cent of corporation tax receipts (including the bank surcharge). As
Chart 3.3 shows, the banking sector is currently somewhat less important as a source of
receipts than it was before the financial crisis, but its share of receipts has increased steadily
in recent years: from 3.4 per cent of the total in 2013-14 to 3.8 per cent in 2017-18.11
7
ONS, Workforce jobs by industry.
8
ONS, gross annual pay, Annual Survey of Hours and Earnings 2018. Total pay estimated using mean earnings and total number of jobs.
9
This reflects cash receipts, as shown in HMRC’s Corporation Tax Statistics 2018.
10
The City of London Corporation in association with PwC, The total tax contribution of UK financial services in 2018. The estimates in this
report are partly based on scaling up the tax paid or collected by 50 financial sector companies.
11
HMRC, Pay-As-You-Earn and Corporate Tax Receipts from the Banking Sector, 2018.

Chart 3.3: Sources of tax receipts from the banking sector

5.0
PAYE Corporation tax Bank payroll tax Bank levy Bank surcharge
4.5
4.0
Per cent of public sector current receipts
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0.0
2005-06 2006-07 2007-08 2008-09 2009-10 2010-11 2011-12 2012-13 2013-14 2014-15 2015-16 2016-17 2017-18
Source: HMRC, OBR
Financial crises and their fiscal impact

Types and frequency of financial crises
3.8 Financial crises come in many forms and have multiple dimensions. It is impossible to
characterise them by a single indicator, but there are some recurring common themes. Most
notably, financial crises have frequently been associated with:
• substantial booms and subsequent busts in credit volumes and asset prices;
• severe disruptions to financial intermediation and the supply of external finance to

households and/or businesses;
• forced deleveraging by firms, households, financial intermediaries and sovereigns; and
• large scale official sector support (in the form of both liquidity support, typically from
central banks, and recapitalisation, typically by governments).
3.9 Based on a survey of studies of past financial crises compiled by the Basel Committee on
Banking Supervision, the UK’s Independent Commission on Banking (ICB) estimated in
2011 that the probability of a crisis occurring in the UK in any given year was nearly 5 per
cent.12 On this basis, one might expect the UK to experience a significant financial crisis
roughly every 20 years, though of course it need not be as severe as the most recent one.
12
HM Treasury and the Department for Business, Innovation and Skills, The Government response to the Independent Commission on
Banking, 2011.

For instance, the secondary banking crisis in the 1970s and the small banks crisis of the
early 1990s were both instances of UK bank failures, but at a smaller scale.13
Direct fiscal costs

3.10 Direct fiscal costs associated with financial crises include the issuance of debt to finance
capital injections into financial institutions (a ‘bail out’), plus the impact of bringing
institutions wholly onto the public sector balance sheet (nationalisation).14
3.11 The concentration of the UK banking sector remains a potential source of direct fiscal risk.
The top four banks (RBS, Lloyds, HSBC and Barclays) together account for more than 85 per
cent of business current accounts and 90 per cent of business loans.15 Firm level analysis by
the Bank of England suggests that the probability of a bank receiving public assistance if it
gets into difficulty increases with its size relative to the system.16
3.12 Table 3.1 is drawn from our March 2019 Economic and fiscal outlook (EFO). It shows the
cash flows associated with the financial sector interventions undertaken in the UK during the
late-2000s crisis, plus the sums subsequently recovered, those outstanding, and the market
value of the Government’s remaining stake as of mid-February. Finally, it estimates the
financing costs associated with these interventions. This is not an attempt to quantify their
overall effect on the public finances relative to a counterfactual where the Government had
not intervened as the crisis unfolded. The costs of the crisis would almost certainly have
been very much greater in the absence of interventions to restore financial stability.
3.13 In total, the Government’s cash outlays during and after the crisis reached £136.6 billion.
The snapshot estimate of the eventual net cost is much smaller at £27.3 billion, 1.7 per cent
of GDP in 2008-09 (the year most of the outlays were made). But this figure is uncertain
and prone to revision. The estimate has ranged from £10.3 billion to £38.4 billion since we
started reporting it in our November 2011 EFO, as share prices have fluctuated and
financing costs have risen with the passage of time. The final figure will not be known until
the Government has sold all its remaining holdings (notably its 62 per cent stake in RBS).
3.14 The gross outlay attracts considerable public and political attention, but it overstates the
direct cost of the interventions because it does not take account of the (admittedly uncertain)
long-term value of the assets purchased. Most of the gross outlay raises public sector net
debt, at least initially, as few of the assets are liquid. The impact on a broader balance sheet
measure like public sector net financial liabilities – which was not published at the time of
the most recent crisis – would be smaller as it nets off all financial assets.
13
See, for example, The Secondary Banking Crisis, 1973-75: Inside Story of Britain’s’ Biggest Banking Upheaval, Margaret Reid, 2003 and
The United Kingdom’s small banks’ crisis of the early 1990s, Andrew Logan, 2001.
14
We discussed the international evidence on the direct costs from financial crises in Chapter 4 of our 2017 Fiscal risks report.
15
Financial Conduct Authority, Competition report 2013-16, 2016.
16
Too big to fail: some empirical evidence on the causes and consequences of public banking interventions in the United Kingdom, Rose
and Wieladek, Bank of England Working paper No.460, 2012.

Table 3.1: Gross and net cash flows of financial sector interventions
£ billion
1
Lloyds RBS UKAR FSCS 2 CGS3 SLS4 Other Total
Cash outlays -20.5 -45.8 -44.1 -20.9 0.0 0.0 -5.3 -136.6
Principal repayments 21.1 6.3 41.4 20.9 0.0 0.0 5.3 95.0
Other fees received5 3.2 4.3 4.4 3.5 4.3 2.3 0.3 22.2
Net cash position 3.8 -35.2 1.8 3.5 4.3 2.3 0.2 -19.4
Outstanding payments 0.0 0.0 2.3 0.0 0.0 0.0 0.1 2.3
Market value 6 0.0 18.2 8.5 0.0 0.0 0.0 0.0 26.8
Implied balance 3.8 -17.0 12.6 3.5 4.3 2.3 0.3 9.7
Exchequer financing -4.0 -14.0 -12.1 -7.7 1.1 0.3 -0.5 -37.0
Overall balance -0.2 -31.0 0.4 -4.2 5.4 2.6 -0.2 -27.3
1
Holdings in Bradford & Bingley and Northern Rock Asset Management plc are now managed by UK Asset Resolution.
2
Financial Services Compensation Scheme.
3
Credit Guarantee Scheme.
4
Special Liquidity Scheme.
5
Fees relating to the asset protection scheme and contingent capital facility are included within the Lloyds and RBS figures.
6
The RBS share price is an average over the 10 days to 14 February, consistent with other market-derived assumptions in our forecast.
UKAR is book value of equity derived from its accounts published 31 March 2018 (value up to date to 26 February 2019).
Indirect fiscal costs

3.15 Indirect costs reflect the fiscal consequences of the damage that financial crises inflict on the
economy. For example, a decline in tax revenues from weakness in nominal GDP and an
increase in public spending due to higher unemployment.
3.16 In September 2014, we published a working paper that looked in detail at the differences
between the Treasury’s March 2008 Budget forecast – the last official forecast before the
collapse of Lehman Brothers sparked meltdown in the world’s financial markets – and
subsequent outturns. This provides a useful benchmark for the effect of a large multi-
dimensional shock, as the economy fell into a deep recession and the authorities stepped in
to restore stability to the financial system. Public sector net borrowing ballooned to almost
10 per cent of GDP and public sector net debt more than doubled to 74 per cent of GDP
over five years.17 But a full assessment of the costs of the crisis depends heavily on the extent
to which the slowdown in productivity growth over the past decade is deemed to be a direct
consequence of the crisis or instead reflects forces that were already in train – a much
debated question.18
3.17 In our 2017 FRR, we presented the results of a fiscal stress test that modelled a period of
synchronised domestic and global economic and financial market stress. In that scenario
public sector net debt was 34 per cent of GDP higher than our baseline forecast by the
scenario horizon, owing to a sharp recession-induced rise in public sector borrowing.
17
Crisis and consolidation in the public finances, Riley and Chote, OBR working paper No.7, 2014.
18
We discussed possible causes of the productivity slowdown in our November 2017 EFO, when revising down our productivity forecast.

Assessing fiscal risks arising from the financial sector

3.18 In our 2017 FRR, based on the Financial Policy Committee’s (FPC’s) judgement, we
concluded that there was a relatively low risk of another financial crisis over the five-year
horizon of our medium-term forecasts. But, based on the ICB’s analysis of the historical
probability of crises occurring, we felt there was a very high probability of one or more
crises within the 50-year horizon over which we assess fiscal sustainability. The potential
impact of any future crisis is, however, difficult to assess and will depend to a significant
extent on the effectiveness and durability of the Government’s post-crisis regulatory
changes. We also judged there to be a moderate risk that financial sector receipts would be
significantly lower than in our latest projections, in both the medium and long term.
3.19 In Managing fiscal risks (MFR), the Government’s formal response to our 2017 report, it:
• Set out post-crisis changes to the regulatory and policy framework intended to ensure
greater resilience to future shocks. These included the provisions in the Financial
Services Act 2012. Key institutional reforms include the establishment of the Prudential
Regulation Authority (PRA) to regulate the safety and soundness of firms, introduction
of the FPC as the UK’s macroprudential authority and the creation of the Financial
Conduct Authority (FCA) to regulate conduct across financial services and markets.
• Confirmed its continued support for the Basel regulatory framework, which embodies
common regulatory and supervisory standards agreed by international prudential
authorities. The Government has also taken a range of actions to mitigate taxpayer
exposure to the financial sector, including introducing a resolution framework.
• Welcomed the diversification of the financial sector since the financial crisis. This
includes an increased role for market-based finance, a wider range of funding
sources, and a reduced reliance on the UK banking system to intermediate funds. We
explore non-bank finance – ‘shadow banks’ – in more detail later in this chapter.
• Recognised the risks from Brexit to the financial sector, in particular from the potential
loss of market access through the ‘passporting regime’.19 The Government addressed
these risks in the content of the Withdrawal Agreement negotiated between the EU and
UK, but which has not received Parliamentary approval. The Government has also
unilaterally put in place some measures to address these risks were the UK to leave the
EU without a deal. UK legislation allows for temporary permissions to enable EEA firms
currently passporting into the UK to continue to operate after exit day while seeking full
UK authorisation – but the EU has not matched that offer for UK firms operating in the
EEA, which would hit UK firms in the event of a no-deal exit.20
19
The ‘passporting’ regime, allows firms authorised in one country within the Single Market to sell certain financial products into any other
country within the Single Market. Outside the Single Market, UK regulation would need to be recognised as equivalent by the European
Commission for UK-based firms to continue selling particular types of financial services into the Single Market.
20
FCA, The temporary permissions regime for inbound passporting EEA firms and funds, 2019.

• Highlighted actions to ensure enhanced cyber resilience, including working with the
G7 and Financial Stability Board (FSB) to explore the scope for cross-border
collaboration to mitigate risks and a simulation by the Public Finances Business
Continuity (PFBC) group of the consequences of a failure in critical banking
infrastructure.21
3.20 We asked the Government several follow-up questions about developments since the
publication of MFR, in particular where the Government had identified gaps to be filled. It
noted several with a bearing on assessing the fiscal risks from the financial sector:
• In relation to the remaining elements of the Basel III framework, the UK remains
committed to implementing these in any EU exit scenario, and supported the recent
G20 finance ministers’ commitment to the full, timely and consistent implementation of
the agreed financial reforms. Several elements of the Basel III framework, such as a
binding leverage ratio and a new long-term liquidity requirement (the ‘net stable
funding ratio’), have been adopted in the EU. The remaining elements of the Basel III
framework, the package agreed in December 2017 and often referred to as ‘Basel
3.1’, has an implementation deadline of 1 January 2022. Work is underway in the UK
and EU to meet that timeline.
• In relation to Brexit, the FPC updated its assessment of the associated financial stability
risks in its March 2019 Financial Policy Summary and Record. It judged that “most risks
to UK financial stability that could arise from disruption to cross-border financial
services in a ‘no deal’ scenario have been mitigated”, although “some disruption to
cross-border services is possible and, in the absence of other actions by EU authorities,
some potential risks to financial stability remain”.22
• In relation to cyber resilience, the Government continues to work with other UK

financial authorities and the National Cyber Security Centre to improve cyber resilience
in the financial sector. The UK also works closely with the G7 and the FSB to
understand evolving cyber risks and coordinate action in response.
• The Government has identified potential risks from fintech. The FPC noted that,
alongside benefits, fintech may introduce new risks to the financial system or contribute
to the evolution of existing risks, for example the cloud services market is at present
highly concentrated, therefore disruption at one provider, due say to a cyber incident,
could interfere with the provision of vital services by several firms.23
• The Government has also established a ‘Cryptoassets Taskforce’ to explore the risks
and potential benefits of cryptoassets and the underlying distributed ledger technology
(DLT), with the most immediate priority being to mitigate the risks posed by
cryptoassets to consumers and markets, and preventing their use for illicit activity.24
21
The PFBC comprises representatives from HM Treasury, the Debt Management Office, the National Audit Office, Government Banking,
the Bank of England and HMRC, and has established a sub-group (the Information Security Group) which includes the National Cyber
Security Centre (NCSC) to focus on current and emerging cyber threats that could disrupt public finances.
22
Financial policy summary and record of the FPC meeting on 26 February 2019.
23
Financial Stability Report, Bank of England, November 2018.
24
Cryptoassets Taskforce: final report, HM Treasury, FCA and Bank of England, October 2018.

Updated risk assessment

Risks to our medium-term forecast
3.21 In line with the FPC’s judgement, we continue to believe that there is a relatively low, though
not negligible, risk of another UK financial crisis over the next five years, as well as a
moderate risk that underperformance of the sector will lead receipts to fall short of our
central forecast. As in our 2017 report, several related vulnerabilities remain relevant:
• First, the current account deficit remains large by both historical and international
standards. The UK has had the largest deficit in the G7 in eight of the past 10 years. In
its November 2018 Financial Stability Report, the Bank of England noted that
investment in UK assets by foreign investors has risen over recent years, financing the
current account deficit. The heavy reliance on these capital inflows made the UK
vulnerable to a shift in foreign investors’ appetite for UK assets. It also noted that
foreign investors have a large presence in the UK commercial real estate and
leveraged loan markets. The consequences of a reduction in foreign investment could
lead to a tightening in credit conditions for UK households and businesses.
• Second, in recent years private consumption growth has outpaced income growth,
leading to the household sector moving into deficit. Were this to continue, it would
probably be associated with a further increase in the household debt to income ratio,
a key vulnerability indicator.
• Finally, the government’s reliance on the financial sector for tax revenues remains
significant (see paragraph 3.6).
Risks to fiscal sustainability

3.22 In our Fiscal sustainability reports (FSR), our projections do not incorporate any effects from
future recessions or financial crises. Based on the ICB’s estimated probabilities (paragraph
3.9), we might expect the UK to suffer a financial crisis around once in every 20 years. Post-
crisis reforms including tighter regulation and the actions taken by financial institutions have
reduced the immediate risk (and potential impact) of further crises, but not eliminated it.
And when risk is suppressed in one part of the system, it often migrates to some other less
heavily regulated part. This is especially relevant when considering the rapid growth in
credit intermediation through non-bank financial institutions, discussed in the next section.
Conclusions
3.23 The financial sector is a significant part of the UK economy, and a key contributor to the
public finances. This means that risks associated with the financial sector can have
significant implications for the public finances if they crystallise.
3.24 Our assessment of most risks is little changed from two years ago. In particular, the risk that
post-crisis tightening of regulation could subsequently be reversed remains. Since the crisis,
numerous regulatory changes have sought to ensure greater resilience to future shocks,

including the provisions in the Financial Service Act 2012. As we discussed in our previous
report, history has shown that the tightness of regulation typically ebbs and flows, being
ramped up in the aftermath of a crisis and then eased when memories fade.
Shadow banking
What is shadow banking?
3.25 There is no single accepted definition of shadow banking. Some institutions prefer not to use
it at all, given the pejorative connotations of shadowy activity. But in broad terms, it refers to
the set of financial entities that, although they do not take deposits, act in certain respects
like banks – undertaking maturity and liquidity transformation and extending credit – and so
have some or all the inherent fragilities of banks, but are not regulated as banks.
3.26 The essence of a bank is that it finances its investments in illiquid loans and risky marketable
securities largely by taking in deposits that can be withdrawn with little or no notice or else
by borrowing short-term funds wholesale from other institutions. It is thus potentially at risk if
deposits are withdrawn en masse or maturing funding is not rolled over, i.e. it faces a ‘run’.
A run may be triggered by concerns that the value of a bank’s assets is insufficient to meet
its obligations to its creditors. Deposit insurance – the cap on which was significantly
increased after the crisis – can greatly reduce the likelihood of a run by depositors.
3.27 Institutions accepting deposits in the UK have to be licensed by the PRA and comply with a
variety of regulations, such as liquidity and capital requirements and in some cases ring-
fencing from other operations.25 In return, these institutions will usually have access to
emergency liquidity support from the central bank. But entities that depend largely on funds
raised in the wholesale markets or from supposedly sophisticated investors usually have
greater freedom of action, although they are still subject to supervisory oversight (by the
FCA). Such entities can end up behaving like a bank in some respects, but without being
subject to the constraints of being supervised as one. Moreover, the passage of funds
through to the final borrower may pass through a chain of specialised intermediaries that
individually do not appear much like a bank, but the whole chain together in effect
undertakes the maturity and liquidity transformation and credit extension characteristic of a
bank.
3.28 By providing borrowers with an alternative source of funding and liquidity, such institutions
can help to facilitate the efficient allocation of resources and risk transfer. But because
shadow banks are potentially susceptible to a run like a conventional bank, they can also
become a source of systemic risk. This can arise both directly from the shadow banking
system and through its interconnectedness with the regular banking system.
3.29 The potential vulnerabilities that can arise from such entities were well illustrated by the
activities of securitisation vehicles during the crisis. These off-balance sheet entities financed
25
The largest UK banks are required by UK law to separate core retail banking services from their investment and international banking
activities.

their holdings of mortgage-backed securities (MBS) by continually rolling over short-term

wholesale funding, which dried up early in the crisis. Another example from the crisis is the
role played by securities lending in the near-collapse of American International Group (AIG)
in 2008. AIG owned a significant quantity of MBS, which it then repeatedly lent out in
exchange for short-term funds (i.e. in similar fashion to a repo agreement). As AIG had also
sold a significant quantity of insurance against the risk of a default on MBS, it was
particularly exposed to the problems in the US housing market, necessitating its rescue by
the US authorities after the collapse of Lehman Brothers.
3.30 More recently in the UK, the suspension in June 2019 of withdrawals from the Woodford
Equity Income Fund, an opened-ended ‘UCITS’ fund,26 illustrates the problems that can
arise when withdrawable funds are invested in illiquid long-term assets. The Fund was
subject to regulations limiting its holdings in unlisted securities, but it struggled to meet that
requirement following a run of redemptions because it could not realise the value of some
of its assets quickly enough.27 Financial institutions that either hold entirely liquid assets, or
where investors’ funds cannot be redeemed quickly, do not pose the same risks.
The scale of shadow banking

3.31 The financial sector comprises both banks and non-bank financial institutions. Non-banks
include insurance corporations, pension funds, investment funds, other financial
intermediaries (OFIs) and financial auxiliaries. In the UK, the size of the non-bank sector has
increased over the past ten years, from 42 per cent of financial sector assets in 2007 to 51
per cent in 2018.28 However, only a subset of these entities are exposed to bank-like risks
and should be thought of as shadow banks.
3.32 The FSB monitors a variety of different measures of non-bank financial activity and
publishes the results in their annual publication Global Monitoring Report on Non-Bank
Financial Intermediation. The different measures include:
• Monitoring universe of non-bank financial intermediation (MUNFI). This is a broad

measure referring to all non-bank financial intermediation. MUNFI accounted for 48
per cent of the total global financial sector assets in 2017.
• OFIs comprise all financial institutions that are not central banks, banks, insurance
corporations, pension funds, public financial institutions or financial auxiliaries. The
largest OFI sub-sectors are investment funds, captive financial institutions and money
lenders, and broker-dealers. In 2017, OFIs accounted for 63 per cent of all non-bank
financial intermediation and 31 per cent of total global financial assets.
• The narrow measure focuses on non-bank financial institutions that authorities have
assessed as being involved in credit intermediation activities that may pose bank-like
26
UCITS refers to ‘Undertakings Collective Investment in Transferrable Securities’, a regulatory framework of the European Commission
that creates a harmonised regime throughout Europe for the management and sale of mutual funds.
27
Financial Stability Report, Bank of England, July 2019.
28

financial stability risks. It is a subset of MUNFI and comprises some, but not all, OFIs.
In 2017, the narrow measure accounted for 28 per cent of all non-bank financial
intermediation and 14 per cent of total global financial assets. This is equivalent to 64
per cent of world GDP in 2017.29
3.33 The narrow measure provides a reasonable metric for what we refer to as shadow banks.
Although, as the AIG example illustrates, other entities can on occasion undertake bank-like
activities that create significant risks to financial stability and so to the public finances.
3.34 The FSB has developed a high-level policy framework – endorsed by the G20 – for
authorities to strengthen oversight of shadow banking. It comprises a range of economic
functions that enable authorities to identify potential sources of systemic risk.30 These
collectively form the narrow measure of non-bank financial intermediation. As Table 3.2
shows, it includes institutions such as money market mutual funds (MMMFs), structured
investment vehicles and hedge funds.
Table 3.2: Economic functions within the FSB’s narrow measure of non-bank
financial intermediation
Economic function Typical entity types
Management of collective investment vehicles with features MMMFs, fixed income funds, mixed funds, credit
that make them susceptible to runs. hedge funds, real estate funds.
Finance companies, leasing/factoring companies,
Loan provision that is dependent on short-term funding.
consumer credit companies.
Intermediation of market activities that is dependent on short-
Broker-dealers, securities finance companies.
term funding or on secured funding of client assets.
Credit insurance companies, financial guarantors,
Facilitation of credit creation.
monolines.
Securitisation-based credit intermediation and funding of
Securitisation vehicles, structured finance vehicles.
financial entities.
3.35 Chart 3.4 shows that the management of collective investment vehicles is by far the largest
source of shadow banking activity in the UK. This is consistent with global trends. There has
been a decline in the intermediation of market activities dependent on short-term funding
since 2008 – though this measure is quite volatile from year to year. Securitisation-based
credit intermediation has also declined over the same period.
29
World Economic Outlook (April 2019), IMF, 2019.
30
The inclusion of such entities or activities does not constitute a judgement that they definitively pose systemic risks, or that authorities
regulation is necessarily adequate, Financial Stability Board (2017).

Chart 3.4: FSB estimates of the narrow measure of shadow banking in the UK
2.5
Collective investment vehicles
Loans with short-term funding
Intermediation from secured funding
2.0
Credit creation
Securitisation-based credit intermediation
Unallocated
1.5
USD (trillion)
1.0
0.5
0.0
2010 2011 2012 2013 2014 2015 2016 2017
Note: Economic function-based measure. Based on time series data included in jurisdictions' 2018 submissions.
Source: FSB
3.36 Focus on leveraged lending – high interest loans extended to firms that are already highly
indebted – as a potential source of financial stability risk has intensified recently. For
example, the FSB report that they are monitoring developments closely.31 This is especially
the case where it reflects loosening underwriting standards and strong creditor risk appetite.
3.37 The Bank of England’s November 2018 Financial Stability Report noted that the share of
‘covenant-lite’ loans – where investors do not require borrowers to maintain certain
financial ratios – had reached unprecedented highs. Other traditional investor protections in
loan documentation have also been relaxed. Consequently, leveraged lending reached pre-
crisis levels in 2018, both globally and in the UK. The Bank estimates that the global stock
of leveraged loans reached an all-time high of $3.2 trillion. This rapid growth in leveraged
lending has also been associated with increased securitisation through collateralised loan
obligations (CLOs).32 This has some echoes of the build-up to the previous financial crisis.
Financial stability risks from shadow banking

How might shadow banking pose financial stability risks?
3.38 Shadow banking is a potential source of risks to UK financial stability, and thus also poses
indirect fiscal risks. The financial stability risks could operate through several channels:
• More stringent regulation of banks since the crisis has increased the incentives to
undertake similar activities within the more lightly regulated shadow banking sector.
The risk is that such entities will generate higher market, credit and liquidity risk, but
with less capacity to withstand losses than their conventional banking counterparts.
31
FSB Chair reports to G20 Leaders ahead of Osaka Summit, Financial Stability Board, June 2019.
32

• Shadow banking could become a source of systemic risk through its

interconnectedness with the banking system. The European Systemic Risk Board
recently highlighted significant interconnections between European banks and the
shadow banking system, together with some associated vulnerabilities – for instance,
that large-scale redemptions by investors in money-market and other investment funds
could prompt fire sales of bank debt and an increase in the cost of bank funding.33
• The interconnectedness with the banking system could also lead to a contagion effect,
whereby institutions undertaking similar activities are tarred with the same brush as
those that are in trouble.
• Substantial growth in leveraged lending, driven by high demand by CLO funds, is

reminiscent of activities preceding the 2008 financial crisis when sub-prime lending
increased sharply.34 A sudden stop to this type of lending would have adverse
consequences for businesses that rely on it as a source of funds.
• A lack of resilience in the shadow banking system could have more general adverse
economic consequences if critical services that households and businesses rely on were
to be abruptly withdrawn.
Have the risks from shadow banks risen or lessened?

3.39 Since the crisis, a considerable internationally coordinated effort by governments, regulators
and supervisors has led to a notable strengthening of banking regulation. Some would
argue that regulation needs to be tightened even further (for instance, by calculating capital
ratios on the basis of the market value of bank capital as well as the book value)35, but it is
difficult to argue that the panoply of reforms has not made the conventional banking system
more robust than it was on the eve of the crisis. Can the same be said of shadow banking?
3.40 There have certainly been major efforts to improve the regulation of the shadow banking
sector and regulators believe they have significantly reduced the major risks, including those
posed by large funding mismatches, high leverage and opaque off-balance sheet
arrangements.36 Steps have also been taken to make runs on collective investment vehicles
less likely, including reforms to MMMFs.37 Regulators are continuing to work on closing
regulatory gaps – for instance, the FSB underscored the importance of effective
operationalisation and implementation of policies agreed to address risks from liquidity
transformation in certain investment funds.38 To this effect, the International Organisation of
33
EU Shadow Banking Monitor, European Systemic Risk Board, September 2018.
34
The rise of leveraged loans: a risky resurgence?, Bank of International Settlements, BIS Quarterly Review, September 2018.
35
John Vickers, The case for market-based stress tests, June 2019.
36
Some reforms are still being implemented – such as the FSB’s recommendations to address structural vulnerabilities from asset
management activities, focused on liquidity mismatch and leverage. See Policy Recommendations to Address Structural
Vulnerabilities from Asset Management Activities, Financial Stability Board, 2017.
37
“In the US, which has the largest MMF market, the Securities and Exchange Commission (SEC) adopted initial rules in 2010 to increase
the resilience of MMFs to economic stresses and reduce the risks of runs on the funds by tightening the maturity and credit quality
standards and imposing new liquidity requirements.” Financial Stability Board, 2017. The EU implemented reforms to MMMFs in full on
21 March 2019.
38
Assessment of shadow banking activities, risks and the adequacy of post-crisis policy tools to address financial stability concerns,
Financial Stability Board, 2017.

Securities Commissions (IOSCO) has updated its recommendations for liquidity risk
management for investment funds.39 IOSCO has also proposed a framework to improve the
measurement of leverage in these funds.
3.41 Nonetheless, the general sense is that regulators believe the risks from this source are now
sufficiently well contained. For example, the FSB in its 2017 assessment of shadow banking
declared that it had “not identified other new financial stability risks from shadow banking
that would warrant additional regulatory action at the global level”.40
3.42 Others, including some former regulators, are less persuaded, believing that more needs to
be done if the system is to made not just safer, but safe enough. Regulatory reform has
made conventional banking more involved and has raised the cost of compliance,
increasing the incentive for riskier activities to migrate into more lightly regulated shadow
banks.41 As a consequence, shadow banking comprises a constantly evolving and largely
unrelated set of intermediation activities pursued by different types of financial
institutions, requiring continual vigilance.42
3.43 The Systemic Risk Council (SRC) has argued that “relying on monitoring developments,
which for the moment seems to be the default approach, is a recipe for failure given the
obstacles to flexible, timely regulatory initiatives”. This might be the case if the sector has
grown big enough by the time financial stability risks manifest themselves to be deemed
systemically important and to have acquired sufficient lobbying power to resist legislation.
The SRC sees the reliance on monitoring as “more or less… exactly the mistake of the early-
2000s”. It therefore argues that “A clear substantive policy on shadow banking—focusing on
liquidity mismatches and leverage, and so distinguishing between different asset-
management activities and structures” is necessary to fill “a glaring gap in the regimes of
every major jurisdiction”.43 This suggests that regardless of the degree of financial stability
risk posed by shadow banking at present, the economic and fiscal impact of a crisis
originating in the sector under the current regulatory arrangements could still be large.
3.44 For all these reasons, we believe it is appropriate to remain cautious in our assessment of
the potential risks from shadow banking. Moreover, experience suggests that a financial
sector that appears to be adequately regulated may turn out not to be so. Vulnerabilities
that appear obvious with hindsight are often much harder to discern in real time.
Fiscal risks from shadow banking

3.45 Fiscal consequences arising as a result of problems in the shadow banking system could
come through several channels:
39
IOSCO issues recommendations and good practices to improve liquidity risk management for investment funds, 2018.
40
Assessment of shadow banking activities, risks and the adequacy of post-crisis policy tools to address financial stability concerns,
Financial Stability Board, 2017.
41
Sound at last? Assessing a Decade of Financial Regulation, Centre for Economic Policy Research, 2019.
42
Thinking Critically about Nonbank Financial Intermediation, Daniel K. Tarullo, 2015.
43
Policy statement to G20 leaders, Systemic Risk Council, 2017.

• Direct support to systemically important institutions within the shadow banking sector.
During the last crisis, the US Federal Reserve was extremely liberal in its lending
policies under the “unusual and exigent circumstances” clause of the Federal Reserve
Act, while the US Government’s ‘troubled asset relief program’ (TARP) was used to
support several non-banks such as AIG. (Total support extended to AIG by the
Treasury Department and the Federal Reserve eventually reached $182 billion44).
• Indirect effects on the core of the banking sector, necessitating public intervention. To
the extent that regulatory reform has enhanced the resilience of the banking system,
this is less likely now than during the last crisis.
• General macroeconomic effects, leading to lower tax revenues and higher public
spending. During the crisis, the potential macroeconomic consequences of inaction
were a prime reason for rescuing the US investment bank Bear Sterns. The
macroeconomic effects of allowing Lehman Brothers to fail were certainly significant.
• The Government may come under political pressure to compensate investors who have
suffered losses. This is more likely when retail investors have been lured into placing
funds in shadow banking entities, believing that their investments are safe.

consider when managing its fiscal risks. And those that we highlighted in our previous
report remain relevant. Among them:
• The comparatively large and highly concentrated UK banking system;
• Cross-country evidence on the frequency of crises and their fiscal cost;
• The tendency for post-crisis tightening of regulation to be loosened over time; and
• Potential effects of Brexit on the financial sector and the tax receipts it generates.
3.47 And specific to shadow banking:
• The challenge of identifying shadow banking activity in the non-bank financial sector;
• The potential risks in a regulatory approach of monitoring risks, then responding; and
• The nature of the fiscal risks that might arise in a shadow banking-led crisis.
3.48 When assessing financial stability and its interaction with fiscal risk over the medium and
long term, does the Government regard these or other issues as important for its risk
management strategy and, if so, how does it intend to address them?
44
US Treasury Notes Blog on AIG, Timothy G. Massad, 2012.

4 Revenue risks
Introduction
4.1 In 2018-19, public sector receipts totalled £786 billion, equivalent to around £27,700 per
household or 36.8 per cent of GDP (on the latest data). Taxes made up 94 per cent of the
total, with income tax and National Insurance contributions (£330 billion) and value added
tax (VAT, £132 billion) the largest revenue raisers. Public sector income from non-tax
sources (largely the interest received on its stock of assets) made up the remainder.
4.2 Our latest medium-term forecast assumes that total receipts relative to GDP – the most
relevant metric for analysing fiscal sustainability – will rise by 0.4 per cent of GDP between
2018-19 and 2023-24 (a similar path to that set out in our March 2017 Economic and
fiscal outlook (EFO), which underpinned our 2017 Fiscal risks report (FRR)). The rise is more
than explained by ‘fiscal drag’ (as real earnings grow and drag more income into higher
tax rates) as well as our assumption that interest rates will rise (raising the returns on public
sector assets). Our long-term sustainability analysis is predicated on a broadly stable
receipts-to-GDP ratio, as historically other policy changes have tended to offset fiscal drag.
4.3 The outlook for receipts is always clouded by risks and uncertainties, as one can see by
comparing the latest outturn estimates to the forecasts produced by the Treasury and (since
2010) the OBR (Chart 4.1). The differences reflect statistical adjustments and revisions,
policy changes, unexpected developments in the economy and unexpected developments
that affect the amount of revenue raised in any given state of the economy.
Chart 4.1: Successive forecasts for total receipts

1000 38
Treasury forecasts
900
OBR forecasts 37
800
Per cent of GDP
Outturn 36
700
£ billion
600 35
500
34
400
300 33
2000-01 2005-06 2010-11 2015-16 2020-21 2000-012004-052008-09 2012-132016-172020-21
Note: Per cent of GDP forecasts have been restated to remove the effects of subsequent revisions to the ratio in the starting year of the
forecast – these can be large and typically reflect methodological changes that forecasters would not have been able to anticipate.
Source: ONS, OBR

Revenue risks
4.4 Looking over a two-year horizon, the differences between our forecasts since 2010 and the
subsequent outturns have been spread across several taxes, but are dominated by weaker-
than-expected income tax and NICs receipts (Chart 4.2). This is the result of the productivity-
related weakness in real earnings growth as well as a lower-than-expected effective tax rate
on earnings that is likely to reflect movements in the earnings distribution (including among
the self-employed) and subsequent policy changes (such as raising the income tax personal
allowance). More recently, onshore corporation tax receipts have performed more strongly
than we anticipated. Box 3.2 of our 2018 Forecast evaluation report described the drivers of
this, in particular the impact of falling use of loss reliefs partly as a result of policy changes.
Chart 4.2: Two-year ahead forecast differences from successive OBR forecasts
50
Other
Property transaction taxes
40
Fuel duty
30
Oil and gas revenues (cash)
VAT
20
Onshore corporation tax (cash)
Income tax and NICs
10
PSCR
£ billion
-10
-20
-30
-40
-50
Jun 10 Nov 10 Mar 11 Nov 11 Mar 12 Dec 12 Mar 13 Dec 13 Mar 14 Dec 14 Mar 15 Jul 15 Nov 15 Mar 16 Nov 16 Mar 17
Note: For comparability, 'in-year' is assumed to be 2009-10 and 2014-15 for the June 2010 and July 2015 forecasts respectively.
Forecast errors have been adjusted for major classification changes.
Source: ONS, OBR
4.5 In this chapter, we review risks we set out in our 2017 FRR. These include:
• the concentration of receipts at the top of particular tax distributions and sectors;
• the falling effective tax rates from changing employment patterns, including the shift
from employee status to self-employment, and from both to incorporated businesses;
• risks from behavioural or technological change, such as improving fuel efficiency;
• specific revenue policy risks, including non-implementation of announced policy,

aspirations yet to be fully specified and incorporated into our forecasts and the
reliance on the relatively uncertain yield from avoidance and operational measures;
• non-payment of taxes due, as a result of avoidance, evasion and tax planning; and
• risks associated with the UK oil and gas industry, including depletion of the tax base
and the uncertain future costs associated with decommissioning oil and gas fields.

Revenue risks
4.6 We also discuss risks arising from two further areas:
• the uncertain costs of reliefs and tax expenditures; and
• the increasingly digitalised economy and the opportunities and challenges this presents
for tax policy design and administration.
4.7 We end by raising issues for the Government’s next response.
Concentration of tax receipts

4.8 We identified two key areas in our previous FRR where a large proportion of revenue is
being generated from a relatively small number of taxpayers. Greater concentration of
receipts can pose a fiscal risk as it might make the tax base more volatile and sensitive to
shocks that affect particular (usually high-income) taxpayers. These were:
• income tax, where the share of revenue from the top percentile of earners had risen to
over a quarter of total revenues over the recent past; and
• taxes on property transactions and capital gains – where a small number of

transactions were generating a very large proportion of total revenues.
4.9 In its Managing fiscal risks (MFR) publication, the Government recognised that successive
increases in the personal allowance meant that the proportion of adults paying income tax
had fallen, and that a growing proportion of receipts came from higher earners. It
highlighted the number of people whose income tax payments had fallen as a result of
these decisions, but said nothing about its view of the fiscal risks that they might pose.

Income tax
4.10 Since our previous report, income tax receipts have continued to become more concentrated
at the top of the distribution. The latest HMRC statistics suggest that the share of receipts
from the top 1 per cent of taxpayers will rise to 29.6 per cent in 2019-20, the highest level
over the recent past. As Chart 4.3 shows, the increase in the proportion of revenues from
the top 1 and top 10 per cent over the past five years has reflected the structure of the tax
system, rather than their share of taxable income rising. This is probably the result of further
increases to the personal allowance and higher rate threshold, as well as successive
increases in dividend tax, which is disproportionately paid by higher income taxpayers.

Revenue risks
Chart 4.3: Income tax contributions from the top percentile

Shares of taxable income Shares of income tax revenues
60 60
Bottom half
50th to 90th percentile
50 50
90th to 99th percentile
Top 1 per cent
40 40
Per cent
30 30
20 20
10 10
0 0
1999-00 2003-04 2007-08 2011-12 2015-16 2019-20 1999-00 2003-04 2007-08 2011-12 2015-16 2019-20
Note: Data for 2008-09 unavailable.

Source: HMRC
Taxes on property transactions

4.11 Revenues from stamp duty land tax are also highly concentrated, reflecting both the
concentration of property value at the top-end of the market, as well as the tax schedule
(which applies higher rates for higher property prices). Chart 4.4 shows that while properties
worth over £500,000 made up roughly 10 per cent of transactions in 2017-18, they
generated nearly 60 per cent of residential SDLT revenues. More striking still, just 7,000
property transactions in the London boroughs of Westminster and Kensington (0.6 per cent
of the total) accounted for over £1 billion in receipts (11 per cent of the total).
4.12 Over the past few years, the trend towards greater concentration has stabilised somewhat,
which may reflect weakness in the prime London property market, and the new 3 per cent
surcharge on second homes and buy-to-lets from April 2016, receipts from which are much
less concentrated in expensive properties. While the concentration of SDLT on high-value
properties has stabilised, dependence may have increased on high-wealth individuals who
purchase expensive properties as well as second homes and investment properties.
4.13 Since our previous FRR the Government’s decision to cut SDLT for first-time buyers that took
effect in November 2017 will have concentrated the tax base further as this only applies to
properties under £500,000. Its effects are not fully reflected in the latest detailed SDLT data.

Revenue risks
Chart 4.4: Composition of residential SDLT tax base and revenues by property value
35 35 35
Share of transactions Share of transactions Share of receipts
30 30 30
(number) (value)
25 25 25
£500k to £1m
20 20 20
Per cent
£1m to £2m
15 15 15
£2m plus
10 10 10
5 5 5
0 0 0
2013-14 2015-16 2017-18 2013-14 2015-16 2017-18 2013-14 2015-16 2017-18
Source: HMRC
Capital gains tax

4.14 Capital gains tax is levied on profits from the sale of assets. Roughly 40 per cent of receipts
come from sales of property and 60 per cent come from sales of financial assets,
particularly shares. CGT is only paid on annual gains above a threshold (£11,700 in 2018-
19) and sales of primary residences are exempt. This means that CGT is typically paid only
by relatively high earners with high value assets on which they can realise substantial gains.
4.15 Chart 4.5 shows that CGT liabilities are highly concentrated in a small number of taxpayers
and that this concentration has been rising. In 2016-17, only 3 per cent of taxpayers
disposed of assets worth at least £1 million, but these generated over 60 per cent of all
CGT liabilities in that year (up from around 50 per cent of liabilities five years ago).
Chart 4.5: Composition of CGT tax base and revenues by value of overall disposals
Share of taxpayers Share of receipts (value)
100 100
Below £250k
80 80
Between 250k and £1m
Above £1m
60 60
Per cent
40 40
20 20
0 0
2012-13 2013-14 2014-15 2015-16 2016-17 2012-13 2013-14 2014-15 2015-16 2016-17
Source: HMRC
Conclusions
4.16 Developments over the past two years leave our assessment of risks arising from the
concentration of receipts little changed. Our forecasts for all three of these taxes continue to
assume that concentration will rise further over time, given the current parameters of the tax
system. While the Government has recognised these trends, the substantial rise in the

Revenue risks
personal allowance and higher rate income tax threshold announced in Budget 2018 will
exacerbate them. So the income tax concentration risk has risen since our previous report.
Trends in self-employment and incorporations

4.17 Our previous FRR discussed the significant consequences for tax receipts of an individual
switching the way they choose to be taxed on their income from work. On the same
earnings, employees will pay more tax than self-employed individuals, who in turn pay
more than individuals who are incorporated (managing their work via a limited company of
which they are the sole director). Both self-employment and single-director companies have
been increasing in number in recent years, which is likely in part to be a result of these tax
discrepancies. Our forecast assumes that these trends continue, with the risk to our forecast
being that the trend is faster or slower than expected.
• Identified several policy changes it had taken to attempt to reduce the fiscal costs from
incorporation, including reforms to dividend taxation and off-payroll working rules.
• Cited the Taylor Review (July 2017), which highlighted a lack of clarity for employment
status rules, among other issues, which may have contributed to incentives for self-
employment or incorporation. Since MFR the Government has published a Good Work
Plan (December 2018) setting out its next steps in responding to the review, but these
remain ‘aspirational’ at this stage and insufficiently firm for us to reflect in our forecast.

4.19 Two tax changes announced in Budget 2018 will slightly alter the incentive to incorporate
relative to the rules in place at the time of our 2017 report:
• First, an extension to the private sector of more onerous compliance with off-payroll
working rules (‘IR35’) with effect from April 2020. This is expected to reduce the
capacity to incorporate to avoid tax and so should reduce the risk, though as we noted
when certifying the costing of this measure its effects are highly uncertain.
• Second, raising the income tax personal allowance and higher rate threshold shift the
tax burden for employees and the self-employed towards NICs. NICs are not paid by
those who have incorporated, so this slightly increases the incentive to incorporate.
4.20 The effect of these changes can be seen in Chart 4.6, which sets out illustrative examples of
tax paid by an individual earning £70,000 if they are an employee, self-employed or

Revenue risks
incorporated.1 In 2017-18 an individual switching from being an employee to being

incorporated reduced their effective tax by 11.1 percentage points (paying £7,777 less tax).
In 2019-20 their effective tax rate was reduced by 11.6 percentage points (saving £8,147).
4.21 A larger change in the incentive to incorporate would have resulted from the 1 percentage
point rise in Class 4 NICs announced in Spring Budget 2017, but this was reversed shortly
after being announced and ahead of our 2017 FRR. Since addressing the imbalance in the
tax system illustrated in Chart 4.6 is only likely to be possible via raising tax paid by the self-
employed, this episode suggests that it is likely to persist for some time.
Chart 4.6: Tax due on £70,000 of income

30
Income tax
Employee NICs
Employer NICs
25
Self-employed NICs
Corporation tax
Dividend tax
Total tax due (£ thousand)
20
15
10
0
Employee Employee Self-employed Self-employed Single director Single director
2017-18 2019-20 2017-18 2019-20 2017-18 2019-20
Source: OBR
4.22 As Chart 4.7 shows, growth in self-employment and incorporations has slowed over the
past two years.2 Our forecasts continue to assume that both continue on their long-run
upward trends over the medium term, so if the recent slowing were to persist it would
represent an upside rise to our receipts forecast.
1
These calculations assume the individual has only one source of income. The deduction of employer NICs means that less of an
employee’s total compensation is made up of their wage, thereby paying less income tax but more NICs than the self-employed.
Company directors are assumed to withdraw profits in the most tax efficient way, paying themselves a salary up to the primary threshold
for NICs, and taking the rest as dividends, all in the same year. The change also includes the reduction in the dividend allowance from
£5,000 to £2,000 announced in Spring Budget 2017 and implemented from April 2018. These examples all reflect taxpayers outside
Scotland. In Scotland higher tax rates at the top-end of the distribution create a slightly larger incentive to incorporate.
2
There are several issues surrounding the quality of these data. See our previous FRR for more detail. While the broad trends are likely to
be reliable, the precise levels should be treated with some caution. HMRC’s historical estimates of the number of single-company directors
has been revised upwards since our previous report, due to further cleaning of the administrative data.

Revenue risks
Chart 4.7: Trends in self-employment and incorporations

16
14
12
Share of workforce (per cent)
10
Self-employed (ONS definition)
8
Single director companies (HMRC definition)
6
Total company population (HMRC definition)
4
0
2000-01 2002-03 2004-05 2006-07 2008-09 2010-11 2012-13 2014-15 2016-17
Note: Though there appear to be very few single director companies before 2008, many companies defined as having two director s
were in effect acting as single director companies.
Source: HMRC, ONS
4.23 Another issue that poses a risk to our medium-term forecasts over the next two years is the
forthcoming revision to the share of corporation tax receipts paid by quarterly instalment
payments (QIPs). As we set out in our March 2019 EFO, HMRC has identified an issue in the
algorithm that splits its monthly administrative data on cash payments and repayments
between larger quarterly payers and smaller annual payers. Most incorporations will not be
paying tax via the QIPs system. Changes in the historical split of this data might alter our
view of trends in incorporations and so have forecast implications, but it is not possible to
anticipate what those implications might be on the information currently available.
Conclusions
4.24 With little change in the tax differential on different ways of working and modest movements
in the share of total employment they make up, our assessment of the fiscal risks from
employees switching to self-employment or incorporation has not changed materially. The
Government retains control over some of the drivers of these trends, such as the tax rates
and thresholds, but has less control over wider changes in the labour market. As we discuss
later in this chapter, the digitalisation of the economy is likely to have been one factor
contributing to the trends towards self-employment and working via a company structure.
But there is likely to be a limit to which employees choose to switch employment status. Self-
employed and incorporated workers have fewer employment rights and less security, so for
some individuals these downsides will outweigh the tax advantages. And changes to off-
payroll working rules will limit opportunities for purely tax-motivated switching.

Revenue risks
Risks from behavioural or technological change

4.25 Behavioural and technological change can pose a variety of risks to the sustainability of
underlying tax bases, the average tax rates applied to those bases or the government’s
ability to collect those revenues. In this section we provide an update on the risks
surrounding the erosion of some excise duty bases over time due to these changes. Later in
the chapter, we look at the wider risks from the digitalisation of the economy.
4.26 In our 2017 FRR we highlighted several risks related to behavioural and technological
change that pose downside risks to the tax-to-GDP ratio. These included:
• Risks arising from technological change, in particular the impact of rising vehicle
efficiency on fuel duty and vehicle excise duty.
• Risks arising from behavioural change, including the impact of declining alcohol and
tobacco consumption.
• Other potential risks, arising from the impact of technological change on the prices of
goods and services, the impact of globalisation on returns on skills and the impact of
climate change targets on environmental taxes.
• Recognised that rising fuel efficiency may affect tax revenues, but stated that it believes
fuel duty will continue to have an important role in the tax system.
• Stated that it is committed to reducing smoking prevalence and that it will continue to
review tobacco duty rates to ensure they meet revenue and public health objectives.
Updated risk assessment: erosion of excise duty tax bases

4.28 Our latest forecast shows that fuel, alcohol and tobacco duties are expected to raise around
£50 billion in 2019-20. We continue to expect receipts to fall over the medium-term by
0.1 per cent of GDP, reflecting continued erosion of each tax base. There have been no
material developments to change our assessment of this risk over the past two years.
4.29 Chart 4.8 shows that:
• Fuel consumption per adult has remained relatively flat over the past two years despite
sustained growth in overall mileages. This implies that the average efficiency of the
vehicle stock has continued to rise (albeit at a slower pace than over the 2000s). In our
October 2018 EFO we introduced a new forecasting model that better reflects changes
in the composition of fuel usage, in particular due to the strong growth in mileages
from light vans. Over the medium term, we are now slightly less pessimistic (from a

Revenue risks
receipts perspective) about overall consumption, although in the longer term the
continued trend toward alternatively fuelled vehicles will weigh on receipts. The
Government’s 2017 decision to ban the sale of petrol and diesel cars by 2040 would,
under a continuation of the current tax system, ultimately reduce receipts to zero.
• Tobacco clearances have evolved broadly as we expected at the time of our previous
report. Over the recent past, regulatory and technological changes (such as the
introduction of plain packaging and the rise of e-cigarettes) may have helped to
sustain this fall, but overall the underlying decline in smoking prevalence appears to
be the key factor. The latest ONS data suggest that just under 17 per cent of adults in
Great Britain smoke cigarettes, down by half from 34 per cent in 1984.3
• Overall alcohol clearances have held up a bit better than expected. In particular, beer
consumption over the past two years has been much stronger than anticipated. It is
unclear at this stage whether this reflects short-term factors (such as rising popularity of
craft beer) or a wider behavioural shift. We are in the process of reviewing our alcohol
forecasting models and aim to reflect any resulting changes in our next forecast.
Chart 4.8: Trends in fuel, tobacco and alcohol consumption

1300 1800
Litres of fuel per adult Cigarettes per adult
1250 1600
1200 1400
Number of cigarettes
1150 1200
Litres of fuel
1100 1000
1050 800
1000 600
March 2017
950 400
March 2019
900 200
Outturn
850 0
1999-00 2004-05 2009-10 2014-15 2019-20 1999-00 2004-05 2009-10 2014-15 2019-20
34 260
Litres of wine per adult Pints of beer and cider per adult
32 240
220
Pints of beer and cider
30
200
Litres of wine
28
180
26
160
24
140
22 120
20 100
1999-00 2004-05 2009-10 2014-15 2019-20 1999-00 2004-05 2009-10 2014-15 2019-20
Note: Reflects total taxable clearances.
Source: HMRC, OBR
3
Adult smoking habits in Great Britain, Office for National Statistics, July 2019.

Revenue risks
Conclusions
4.30 Overall, nothing has materially changed our view of these risks. While the medium-term
prospects for fuel duty revenues have improved since our previous report, we continue to
expect them to trend towards zero over the longer term under current policy settings. We will
continue to monitor alcohol consumption to see whether the decline in prevalence at
younger ages will put downward pressure on revenues at a longer time-horizon. But, as the
Government sets out in Managing fiscal risks, it has decided to accommodate some of these
risks for other policy reasons – i.e. its commitment to reduce smoking prevalence.
Revenue policy risks

4.31 Parliament requires our forecasts to be based on current government policy, or on our
interpretation of it where it is not clearly defined. Changes in tax policy therefore constitute a
further source of risk to our revenue forecast.
4.32 We identified several specific areas of policy risk in FRR 2017, including:
• The risk associated with stated policy not being implemented, notably the repeated
decisions not to raise fuel and alcohol duties in line with inflation, which the
Government has stated to be its default policy position for our baseline forecasts.
• Policy commitments and aspirations not yet captured in our forecasts. Unless the
Government specifies a new policy with reasonable precision, and for each year of our
medium-term forecast, we are unable to include it in our central forecast. Instead we
consider it to be an aspiration and note it as a risk. This applies to manifesto
commitments, conference announcements and leadership campaign pledges.
• The increasing dependence on the uncertain yield from anti-avoidance and

operational measures. The policy costings for these types of measures are subject to
higher levels of uncertainty compared to the relatively certain cost of tax cuts, such as
the aforementioned freezes in fuel and alcohol duties.
• Estimated that freezes to the headline fuel duty rate over 2010-11 to 2018-19 will
have raised net debt by around £84 billion by 2022-23 (just over 3 per cent of GDP).
The Government stated that its policy is that “all duties continue to be uprated in line
with inflation” but that “final decisions on tax rates are taken at fiscal events”.
• Recognised the uncertainty surrounding our central estimates of operational and

compliance measures. But it did not address the risk posed by policy packages that
pair relatively certain costs with relatively uncertain yields.

Revenue risks
4.34 The Government did not discuss how it manages the risks posed by policy aspirations that
have been described publicly, but not in sufficient detail to allow them to be reflected in our
central forecast. We asked the Treasury for further information about its risk management in
this area when preparing this report. It pointed to the processes it has in place to allow the
Chancellor to consider policy options ahead of fiscal events, including the one by which it
updates us on policy aspirations that we have previously identified as risks to our forecasts.

Non-implementation of stated policy
4.35 Despite the Government’s MFR statement that “all duties continue to be uprated in line with
inflation”, three months later its final decision on the fuel duty rate for 2019-20, taken in
Budget 2018, was to freeze it again (at a cost of around £0.9 billion a year). The cost of
real-terms cuts to fuel duty rates announced since June 2010 is now around £10 billion a
year in 2019-20. Budget 2018 measures also froze beer, cider and spirits duty again in
2019-20 (at a cost of around £0.2 billion a year). The cost of cuts and freezes in alcohol
duty rates announced since 2012-13 is now around £1¼ billion a year in 2019-20.
4.36 Chart 4.9 shows the drivers of growth in fuel and alcohol duty receipts in our March 2019
forecast.4 Of the £6.3 billion rise in receipts between 2018-19 and 2023-24, £5.3 billion
reflects the policy assumption that duty rates will rise in line with RPI inflation. The remaining
£1.0 billion reflects our forecast for underlying growth in fuel and alcohol consumption.
Chart 4.9: Fuel and alcohol receipts: drivers of growth over the medium term
7
of which: alcohol duty rates
of which: fuel duty rates
6
of which: alcohol clearances
of which: fuel clearances
5
Cumulative growth in alcohol and fuel receipts
4
£ billion
0
2019-20 2020-21 2021-22 2022-23 2023-24
Source: OBR
4
This represents the estimated static contribution of the Government’s duty uprating policy. If the upratings were not implemented, the
actual impact on receipts would differ slightly due to the behavioural responses of taxpayers to those lower duty rates.

Revenue risks
4.37 In the March 2016 Budget, the Government announced that it would abolish Class 2 NICs
from 2018 at a cost of around £0.4 billion a year. The abolition was then delayed to April
2019 at Autumn Budget 2017, before being cancelled at Budget 2018, following political
pressure generated by the fact that some taxpayers would lose out from the measure (by
voluntarily paying the more expensive Class 3 rate when Class 2 was abolished) despite the
overall effect being a giveaway. Similarly, the decision to increase Class 4 NICs rates at
Spring Budget 2017 was almost immediately reversed in the face of political pressure
generated by the cost to those who would pay more. This highlights once more the potential
risk that non-implementation of stated policy may present to our central forecast.
Policy commitments and aspirations not yet captured in our forecasts

4.38 In our March 2019 EFO we listed many policy ambitions that the Treasury confirmed did not
yet represent firm Government policy and so should not be reflected in our central forecast.5
Those policies included:
• Multiple policy uncertainties regarding Brexit, including the UK’s participation in the EU
emissions trading system (ETS) and the changes to accounting arrangements for
traders facing the cash flow implications of import VAT being applied on goods
imported from the EU.
• The Government’s announced intention to introduce a tax on plastic packaging from

April 2022, and the consultation on its design, including the tax rates to be applied.
• The consultation on the surcharge on stamp duty land tax for non-resident buyers
acquiring residential property in England and Northern Ireland.
4.39 Since March there have been developments in several policy areas, including:
• The Department of Health and Social Care and the Cabinet Office announced a
consultation on potential changes to the NHS pension scheme to help members who
are affected by the ‘annual allowance taper’ – primarily GPs and consultants. For
those on incomes above £150,000 the taper progressively reduces the amount an
individual can contribute to pension pots from pre-tax income. Estimating how much
will be deemed to have been contributed to a defined benefit pension pot is not
straightforward, so some scheme members have faced unexpectedly large tax bills.
• In June 2019, the BBC announced that it intends to end universal provision of free TV
licences to the over-75s and instead means-test entitlement with reference to receipt of
pension credit. This would raise total licence fee revenue, but, as we describe in Box
5.1, the overall effect of the change is likely to raise rather than reduce the deficit.
4.40 The candidates in the current Conservative leadership campaign have advanced various net
revenue-reducing tax proposals, which represent a policy risk to our central forecast.
5
We discuss these in greater depth in Annex A of our March 2019 EFO.

Revenue risks
Reliance on yield from highly uncertain revenue-raising measures

4.41 In our 2017 FRR, we highlighted the increasing reliance that the Government has placed on
raising money via anti-avoidance and operational measures that attempt to increase the
effectiveness of HMRC’s compliance activity. We tend to assign these costings a higher
uncertainty rating, because the measures generally target a subset of taxpayers who are
already changing their behaviour to reduce their tax liabilities.6 The higher uncertainty also
reflects the less reliable data used to estimate the yields, since the measures are directed at
uncollected tax. (It is important to note that when we describe a costing as highly uncertain,
we see risks lying to both sides of what we nonetheless judge to be a central estimate.)
4.42 Chart 4.10 sets out the cumulative impact of tax giveaways and takeaways since December
2014 (when we first began assigning uncertainty ratings to each measure).7 It shows that:
• The cumulative net impact of Government decisions since December 2014 has been to
raise tax receipts by £13.8 billion (0.5 per cent of GDP) by 2023-24. That reflects a
cumulative gross takeaway of £42.2 billion (1.7 per cent of GDP) partly offset by a
cumulative gross giveaway of £28.4 billion (1.1 per cent of GDP).
• In value terms, around two-thirds of the gross takeaway has been assigned a ‘higher’
uncertainty ranking. In stark contrast, only around one third of the gross giveaway in
value terms has been assigned a ‘higher’ uncertainty ranking.
• The net tax takeaway since December 2014 is almost exactly equal to the total value of
anti-avoidance revenue raisers. Other takeaways (such as the doubling of the standard
rate of insurance premium tax) have been broadly offset by other giveaways (such as
the successive rises in the personal allowance and higher rate thresholds).
6
See our Policy costings uncertainty ratings database online for more information on the framework we use to assign these ratings.
7
The figures in the chart reflect the cumulative sum of policy costings, as they were estimated at the time (i.e. no adjustments have been
made for subsequent recostings). For years outside the forecast period, we have assumed that the impact of the costing remains flat as a
share of nominal GDP, consistent with the methodology used in the Policy measures database available on our website.

Revenue risks
Chart 4.10: Relative uncertainty of tax measures announced since December 2014
50
No rating
'Lower' uncertainty
40
'Higher' uncertainty
Anti-avoidance revenue raisers
30
Net Takeaway
20
£ billion
10
-10
Giveaway
-20
-30
2015-16 2016-17 2017-18 2018-19 2019-20 2020-21 2021-22 2022-23 2023-24
Note: 'Lower uncertainty' includes our three lowest ratings ('low', 'medium-low' and 'medium') while 'Higher uncertainty' covers our three
highest ratings ('very high', 'high' and 'medium-high'). 'Anti-avoidance' also includes operational and compliance measures.
Source: OBR
4.43 Over the past two years, the Government has again introduced several revenue-raisers that
we have assigned higher uncertainty ratings to. Many of these measures are inherently
uncertain to cost because they typically affect taxpayers who are subject to an already
complex tax system and who are actively attempting to reduce their tax liabilities.
4.44 The underlying risk here reflects the consistent use of uncertain revenue-raisers to fund other
objectives with more certain costs, such as tax rate cuts or spending commitments. Over the
past two years, the most notable higher-uncertainty revenue-raisers have included:8
• ‘Avoidance and evasion: additional compliance resource’, announced at Autumn

Budget 2017 and estimated to raise £0.7 billion a year by 2022-23. We assigned this
a ‘very high’ uncertainty rating. As well as the usual difficulties outlined above, this
type of compliance measure is challenging because it does not map directly onto the
National Accounts receipts definitions that we forecast. Another challenge is
determining whether the yield is additional to that already captured in similar previous
measures or HMRC’s existing compliance effort. Compliance measures are also
subject to uncertainty around the timing of operational delivery.
• At Autumn Budget 2018, the Government introduced a digital services tax, which was
estimated to raise £0.4 billion a year by 2023-24. We assigned this measure a ‘very-
high’ uncertainty rating. All aspects of the costing were deemed uncertain, in particular
the reliance on complex, multi-stage modelling.
• Also at Autumn Budget 2018, the Government extended ‘off-payroll’ working reforms
to the private sector, which were estimated to raise £0.7 billion a year by 2023-24.
8
Each of these is discussed in more detail in the relevant Economic and fiscal outlook.

Revenue risks
This measure also received a ‘very-high’ uncertainty rating, reflecting both the quality
of data and the uncertainty of the behavioural response, as well as the challenges in
modelling the impact of the measure.
4.45 Following a familiar pattern, all three were contained within Budget packages where their
uncertain yield was used to fund giveaways with relatively more certain fiscal impacts,
notably increases in departmental spending, freezes to fuel duty and increases in the
income tax personal allowance and higher rate threshold.
Conclusions
4.46 Overall, the continued freezing of fuel and some alcohol duties, as well as the dropping of
various NICs measures, suggests that the risk of non-implementation of stated policy still
poses a material risk to our central forecast. As regards policy commitments and
aspirations, while their number has risen over the past two years, their overall value is likely
to have fallen thanks to the crystallisation of the personal allowance and higher rate
threshold manifesto commitments at Autumn Budget 2018. The tax cuts proposed by the
Conservative leadership candidates pose new risks on top of those we identified in March.
4.47 The Government’s continued behaviour over the past two years to help fund increased
departmental spending and straightforward tax cuts with risky revenue-raisers suggests that
this risk remains (particularly given the nature of governing via a minority and the continued
challenges that Brexit poses to the ability of the Government to enact more certain tax rises).
Non-payment of taxes due

4.48 One important risk to all taxes is that some of those who should pay them do not. That
could be for a number of reasons – legal or illegal. In our previous report, we:
• Highlighted the stark difference in levels of non-compliance across different forms of

tax collection (as measured by HMRC as ‘tax gaps’).
• Discussed the complexity of the UK tax system and how this may have created
opportunities for taxpayers to challenge legal interpretations or exploit boundaries.
• Noted that HMRC’s estimate of the tax gap had fallen in recent years, but that these
estimates were uncertain and that HMRC continues to improve its methodologies.
• Identified a range of measures it had taken to tackle tax avoidance and evasion,
including measures to digitise tax collection through ‘making tax digital’.

Revenue risks
4.50 The Government argued that “the length of tax legislation and the number of tax reliefs” –
two proxies we cited in our 2017 report – “are not necessarily good measures of
complexity”. But it did not say what it does regard as good measures.

Tax gaps
4.51 We do not generally estimate tax gaps explicitly when we produce our forecasts, as they are
usually based on forecasts of growth rates from the latest outturn data collected by HMRC.
This means that each forecast contains an implicit assumption about the tax gap – usually
that it remains flat in proportional terms. There are two key exceptions to this:
• we include the estimated yield from anti-avoidance and compliance measures; and
• our VAT forecast includes explicit estimates of the theoretical liability and a VAT gap,
which we typically set for the initial year of the forecast and hold flat thereafter, other
than factoring in the effect of VAT gap-reducing policy measures.9
4.52 In our 2017 report we discussed how difficult it is to estimate tax gaps and looked at the
main conclusions from HMRC’s 2016 Measuring tax gaps (MTG) report covering 2015-16,
in particular the stark difference in rates of non-compliance between different taxes.
HMRC’s latest published analysis covers 2017-18. It shows the tax gap rising by 0.3
percentage points since 2015-16 to 5.6 per cent. This increase appears to have been driven
by a rise in the VAT gap, partly reflecting the fact that VAT debt owed to HMRC has risen
over the past two years. As HMRC recognises in MTG, there is a relatively wide range of
uncertainty around the central estimates of the individual and overall tax gaps, so they
provide a better indication of longer-term trends than of precise small year-on-year
changes. Chart 4.11 shows HMRC’s latest estimate of the tax gap since 2005-06.
9
More information on how our VAT forecast is produced is available in the ‘forecasts in-depth’ section of our website.

Revenue risks
Chart 4.11: HMRC estimate of the tax gap since 2005-06

8
5
Per cent
0
2005-06 2006-07 2007-08 2008-09 2009-10 2010-11 2011-12 2012-13 2013-14 2014-15 2015-16 2016-17 2017-18
Source: HMRC
4.53 To understand the drivers of changes in the tax gap and the associated fiscal risks, we can
split movements in the aggregate tax gap into:
• Trends in the tax gap for each specific tax head. The drivers of these changes will
reflect a multitude of factors, including estimation error, levels of HMRC compliance
activity and underlying changes in the levels of avoidance and evasion.
• The overall composition of the tax-take. The government may decide to raise more or
less revenue via different streams of taxation through discretionary policy changes, or
changes in the composition of economic activity could affect the composition of the
tax-take. This will affect the overall measure of the aggregate tax gap.
4.54 Chart 4.12 shows movements over the past decade in the level of the tax gap for each of
the main revenue streams (on the vertical axis) against the share of those revenue streams in
the overall HMRC tax-take (on the horizontal one). It shows that the key drivers of the 0.4
percentage point fall in the tax gap since 2007-08 have been:
• A 1.6 percentage point fall in the corporation tax gap. While the gap is substantially
higher for smaller businesses, HMRC analysis suggests that it has fallen across all sizes
of businesses over this period.
• A 2.3 percentage point fall in the excise tax gap, largely driven by declines in the fuel
and tobacco duty gaps.

Revenue risks
• A 0.4 percentage point fall in the PAYE tax gap. PAYE income tax and NICs are the
largest sources of government revenue, so relatively small changes in their level of
compliance can have substantial effects on the overall tax gap.
• Offsetting those falls, the composition of revenue since 2007-08 has shifted away from
income tax (partly reflecting successive real-terms rises in the personal allowance and
higher rate threshold) towards VAT (reflecting the rise in the standard rate to 20 per
cent in January 2011). Because the tax gap for PAYE is much lower than for VAT
(1.0 per cent versus 9.1 per cent in 2017-18) this has raised the overall tax gap.
Chart 4.12: Changes in the tax gap: 2007-08 to 2017-18

20
2007-08
Self-assessed income tax 2017-18
15
Size of tax gap (per cent)
Corporation tax
10
VAT
Excise duties
5
Pay as you earn income tax and
National Insurance contributions
Stamp duties
0
0 10 20 30 40 50 60
Share of HMRC cash receipts (per cent)
Source: HMRC
4.55 We asked the Treasury whether any steps taken on tackling tax gaps had been delayed
because of Brexit-related priorities. It told us that it has allocated over £650 million to
HMRC since 2017-18 for EU exit preparations, including £350 million in 2019-20, which
has been ringfenced from the ‘business as usual’ budget.
Conclusions
4.56 Over the past two years, HMRC’s estimate of the tax gap has risen by 0.3 percentage
points. Given the uncertainty and difficulty in generating these estimates, our view is that the
underlying risk arising from non-payment of taxes due is little changed from our previous
assessment. One source of this risk is the misuse of tax reliefs, which are discussed later in
this chapter. As we set out last time, this risk is endogenous to government activity and it is
important to understand that other policy changes that may influence the composition of the
tax-take will also affect the total level of non-compliance.

Revenue risks
Risks associated with the UK oil and gas industry

4.57 In our previous FRR, we reported on the likely long-term decline of oil and gas revenues –
largely down to exhaustion of resources remaining in the UK continental shelf (UKCS). To a
large extent, the risk to tax payments has crystallised already – with the remaining resources
in the UKCS becoming increasingly expensive to extract. Revenues have already fallen from
£10.6 billion in 2008-09 to £1.2 billion in 2018-19. Our medium-term forecast predicts
that they will stay at low levels given current oil price expectations. Over the longer term, the
key risk reflects the cost of decommissioning infrastructure on the UKCS and how much of
these costs will be borne by government via both tax repayments and foregone taxes.
• Acknowledged a June 2018 Oil & Gas Authority (OGA) report setting out its central
estimate of the cost of decommissioning infrastructure (including platforms, wells,
pipelines and terminals) on the UKCS (£58.3 billion in 2017 prices) as well as the
uncertainty surrounding this estimate.10 HMRC has since estimated that the
Government may bear around 40 per cent of these costs via tax repayments as well as
tax receipts foregone.11
• Stated that it was working with the OGA to monitor progress in reducing
decommissioning costs and had set up a ‘Decommissioning Costs Board’ to “embed
expertise and accountability for supporting the OGA’s target across government”.12
Updated risk assessment and conclusion

4.59 The oil price and expectations of future prices have risen since our 2017 report and our
forecast for net receipts is therefore higher than it was then. But it is still small in cash terms
at around £2 billion a year on average over the medium term. As we set out in our previous
report, the revenue risk from declining production has already been largely realised.
4.60 The latest OGA estimate of total decommissioning costs is 17 per cent lower on a like-for-
like basis than its original 2017 estimate. The OGA said that the reduction had been
“primarily driven by continued improvement in planning and execution practices” although
the underlying detail was not available when we finalised this publication.13 Neither was the
updated HMRC estimate of the total exchequer impact of decommissioning costs.
10
UKCS Decommissioning 2018 Cost Estimate report, Oil and Gas Authority, June 2018
11
Statistics of Government revenues from UK oil and gas production, HMRC, June 2018
12
The target is measured against the 2017 estimate of £59.7 billion (reflecting total spending from 2017 onwards) in 2016 prices.
13
Cost of UKCS oil and gas decommissioning continues to fall, OGA announcement, June 2019.

Revenue risks
4.61 Following a recent Public Accounts Committee hearing into the public cost of
decommissioning oil and gas infrastructure, the Government agreed with the Committee’s
recommendations in a number of areas, including:
• That the OGA should set out how it is making its estimate more certain and what the
expected impact of new and as-yet uncosted projects could be.
• That BEIS and the OGA should report annually on the direct impact it has had on
reducing decommissioning costs, as well as reporting outturn costs against forecast.
4.62 Given the uncertainty surrounding the overall cost of decommissioning as well as its
implications for the public finances, nothing material has changed over the past two years
to suggest that this risk has been substantially mitigated.
Tax reliefs
4.63 Tax reliefs are a part of every tax system and help define what is and what is not to be taxed
– the tax base. But tax reliefs are often introduced with the intention of achieving other
policy objectives. In this section, we discuss those tax reliefs and expenditures for which
HMRC has published estimates, and we use those estimates to illustrate some of the ways
they might constitute a fiscal risk. For example, in some instances, they are used as
disguised and non-transparent alternatives to conventional public spending, getting far less
scrutiny as a result. But it is worth noting at the outset that while we discuss several different
reliefs and expenditures, we are not recommending their wholesale removal – this would
beyond our remit in any case. Reliefs are clearly an integral part of every tax system.
4.64 The estimates HMRC produces are what is known as ‘static’ estimates. That is, they answer
the question ‘Given the activity that took place in the economy in a particular year, if this
relief did not exist and the same activity took place how much additional tax would have
been raised?’. This is a very different question to ‘How much additional tax would be raised
if this relief did not exist?’ – to answer that you would need to consider how activity might
change as taxpayers responded to the different tax incentives now in place. And it is an even
more different question to ‘How would the public finances be affected if this relief did not
exist?’ – which would require you to think about knock-on implications to other elements of
the public finances, for example the welfare spending implications of changing the income
tax personal allowance when the universal credit means test is measured after tax or, where
the figures involved are large, how would economic growth more generally be affected.
4.65 Finally, and beyond our remit, there might be a public policy question of ‘How would the
policy objective met by this relief be met if the relief did not exist, and at what cost?’. So, for
example, governments seek to promote personal saving for retirement. One means of
doing so is by ensuring that pension saving and income is only taxed once. In the UK that is
currently achieved by allowing people to make pension contributions out of pre-tax income
(subject to some limits), not taxing returns on pension savings, but then taxing income in

Revenue risks
retirement (with some exceptions).14 The estimated cost of the former is large, but since the
long-term fiscal cost of inadequate private pension saving might be much higher, one might
reasonably assume that any reform of the system would retain the basic objective.15 So the
public policy question is one of how effectively do different approaches achieve that goal.
Structural reliefs versus tax expenditures

4.66 HMRC splits reliefs into two broad categories:
• ‘Structural reliefs’ are considered “an integral part of the tax system”. These include tax
thresholds, such as the personal allowance for income tax and the primary and
secondary thresholds for National Insurance contributions (NICs).
• ‘Tax expenditures’ are designed to “to help or encourage particular types of

individuals, activities or products for economic or social objectives”.16 This is a very
broad definition, spanning everything from income tax relief for pension contributions
to the exemption that ensures pet cemeteries are not liable for landfill tax.
4.67 Some reliefs fall neatly into one category or the other. For example, HMRC classes R&D tax
relief as a tax expenditure, and it is a clear example of a relief designed to meet a specific
policy objective – to stimulate additional R&D investment by partially or wholly offsetting the
cost against corporation tax liability or providing a payable tax credit. But many include an
element of both tax expenditure and structural relief. For example, the annual investment
allowance is more generous than standard capital allowances and is designed to “support
investment by small- and medium-sized firms”.17 HMRC classes it as having both tax
expenditure and structural components, highlighting one of the challenges in categorising
reliefs. Neither HMRC nor the Treasury maintains a list of reliefs by policy objective.
4.68 In our 2017 FRR we highlighted that the UK tax system, whether measured by the length of
the tax code or the number and size of tax reliefs and expenditures, is one of the most
complex in the world. HMRC currently identifies 1,171 structural reliefs and tax
expenditures, although it is still trying to confirm the precise number in place. Based on past
behaviour, it expects this number to rise as “Governments and Parliament add more reliefs
than they take away”.18 The close to 100 new tax policy measures that the Government has
announced in the two years since our previous report is consistent with a continuation of
that pattern. Despite this, the Government stated in MFR that it “agrees on the need for
simplification of the tax system”.
14
See, for example, House of Commons Library Briefing Paper Number CBP-07505, Reform of pension tax relief, October 2018.
15
See, for example, the ‘principles for reform’ identified in HM Treasury, Strengthening the incentive to save: a consultation on pension tax
relief, July 2015.
16
HMRC, Estimated costs of tax reliefs, 2019. HMRC’s definition of tax expenditures is in line with that proposed by the NAO in its 2014
Tax reliefs report. The NAO also splits non-tax expenditure-type reliefs into five categories: reliefs to correctly measure income or profits;
reliefs to ensure the scope of the tax is as intended; reliefs to improve the progressivity of tax; reliefs to create simplicity; and reliefs
introduced by international agreements.
17
The description used by the Government in Summer Budget 2015.
18
HMRC evidence to the Public Accounts Committee, September 2018.

Revenue risks
Why do tax reliefs and expenditures constitute a fiscal risk?

4.69 There are several reasons why tax reliefs and expenditures could pose fiscal risks, including:
• The Government does not know the overall cost even of the tax reliefs and
expenditures it can identify. HMRC only publishes the costs of around one in six of
them, although these would be likely to account for the majority of the overall cost if
that were known. The National Audit Office (NAO) has previously said that “HMRC has
an inconsistent approach to collecting, using and publishing data on tax reliefs”.19
There is also little information about reliefs affecting non-HMRC taxes.
• The cost of the policy motivated tax expenditures that HMRC has identified is large in
absolute terms – approaching 8 per cent of GDP – and also by international
standards. The cost has also risen significantly over the past decade.
• It is not clear that the Government gives tax reliefs and expenditures adequate scrutiny
to control their cost. This contrasts with the very high degree of scrutiny it applies to
what can be much smaller spending settlements it reaches with departments. To give
some sense of scale, the combined £4 billion annual cost of the ‘tied oils scheme’ and
the ‘rebated rate for gas oil’ is more than the annual cost of running HMRC.
• There is a lack of transparency around tax reliefs and expenditures. HMRC’s annual
data release contains many numbers, but almost no commentary on them.20
• The Government does not seem to have a systematic way of evaluating the
effectiveness of those tax reliefs and expenditures with a stated policy objective. The
NAO has said that “HMRC does not collect the data that would allow it to conclude on
the effectiveness of tax reliefs” and that there is “little evidence that HMRC evaluates
reliefs to see if their objectives are being met”.21
• Tax reliefs and expenditures add complexity to the tax system, which may encourage
more avoidance activity as taxpayers are given the opportunity to exploit new
boundaries or to challenge legal interpretations. The Public Accounts Committee has
recommended that “HMRC should regularly monitor variances between its forecasts of
what tax reliefs will cost and what they actually cost. Where costs significantly exceed
forecasts, it should seek positive evidence that the relief is working as intended and not
being targeted for tax avoidance”.22
4.70 In Managing fiscal risks, the Government said it “agrees on the need for simplification of the
tax system” and that it “recognises the need to monitor and evaluate existing tax reliefs, and
to ensure that any new reliefs introduced are justified and appropriately targeted”.
Somewhat more concretely, it pointed to the role of the Office of Tax Simplification on the
former and new internal HMRC processes that had been shared with the NAO on the latter.
19
National Audit Office, The effective management of tax reliefs, 2014.
20
See HMRC, Estimated costs of tax reliefs, annual estimates, 2019. The single page of this release devoted to ‘commentary on changes
in the published figures’ is dominated by an explanation of corrections to the estimated cost of the income tax marriage allowance.
21
National Audit Office, The effective management of tax reliefs, 2014.
22
Public Accounts Committee, The effective management of tax reliefs, 2015.

Revenue risks
The cost of tax reliefs and expenditures

All reliefs and expenditures for which estimates are available
4.71 HMRC’s most recent publications list 430 reliefs, and it estimates the cost of 195 of them.23
Of the 235 that it does not quantify, the explanation given for most is that the “information
on the usage of this relief is not required in tax returns and cannot be reliably estimated from
other data sources, and the cost of collection for statistical purposes is disproportionate.”24
HMRC does not report on the remaining 741 reliefs it identifies, so there are no estimated
costs of these, but HMRC is currently working to expand the number of estimated reliefs in
future publications. Of the 195 that are estimated, 57 are classed as tax expenditures, 25
as structural reliefs with 33 as combining elements of both. The remaining 80 reliefs are
classed as ‘minor’ – where the estimated cost of each relief is less than £50 million a year.
4.72 The estimated sum of all identified and costed reliefs for 2018-19 is £441 billion (21 per
cent of GDP). To place this in context, it is around half the latest ONS estimate for total
managed expenditure in 2018-19. But, as noted at the start of this section, it is important to
recognise that while aggregating the costs of reliefs in this way gives some sense of the
overall scale, it is not the amount that the Exchequer would gain from abolishing them all,
and it is not directly comparable to the size of public spending. To give another example,
the estimates do not consider interactions across reliefs – if the annual investment allowance
did not exist, then take-up of the remaining capital allowances would be higher as a result.
And if capital allowances were removed from the corporate tax system entirely, then
investment would be lower thanks to the higher post-tax cost of capital, which in turn would
ultimately have adverse second round effects on the economy and tax receipts.
4.73 Chart 4.13 shows that just under half the cost of the identified and costed reliefs relates to
structural reliefs, just over a third to tax expenditures and around a fifth to reliefs that
combine elements of both. The reliefs with the largest costs are structural ones: the income
tax personal allowance (£107 billion, 5.0 per cent of GDP) and the primary and secondary
thresholds for NICs (£57 billion, 2.7 per cent of GDP). Their cost has risen significantly since
2010-11 (from 3.2 per cent GDP for the personal allowance and 1.9 per cent of GDP for
the NICs thresholds), largely due to policy decisions, particularly the near doubling of the
personal allowance. These costs and their evolution are reasonably well understood.
23
HMRC splits the list across three publications: Estimated costs of principal tax reliefs, 2019; Estimated cost of minor tax reliefs, 2019;
and Tax reliefs in force in 2017-18 or 2018-19: Estimates of cost unavailable, 2019.
24
HMRC, Estimated costs of tax reliefs, 2019.

Revenue risks
Chart 4.13: Estimated total cost of tax reliefs in 2018-19

250
Other structural relief Other tax expenditures
NICs primary and NICs pensions relief for employers
secondary threshold
Personal allowance IT pensions relief
200
CGT private residence relief
VAT zero and reduced rate
£ billion
150
Other both
IHT nil rate band
100
Capital allowances
50
0
Structural reliefs Tax expenditures Elements of both
Source: HMRC
Tax expenditures
4.74 The three largest tax expenditures accounted for 85 per cent of the total cost of such policy-
motivated reliefs in 2018-19. These are:25
• Tax relief for registered pension schemes totalled £54.7 billion in 2017-18 (2.6 per
cent of GDP).26 HMRC records this cost in two parts: income tax relief accounted for
£38.4 billion (equivalent to 21 per cent of income tax receipts) and NICs relief the
remaining £16.3 billion (12 per cent of NICs receipts). As discussed above, abolishing
these would lead to large behavioural responses, as people switched to alternative tax-
efficient savings vehicles, so the yield of such a change would be much lower than
these estimated costs. And the Government would be left with the policy question of
how to encourage people to save. We discuss pensions tax relief in more detail below.
• The reduced and zero-rating of VAT, which amounted to £52.7 billion in 2018-19
(equivalent to 2.5 per cent of GDP and 40 per cent of total VAT receipts collected). In
2005-06 these cost 2.0 per cent of GDP, so their cost has risen. The two largest
components are the zero-rating of food (£18.6 billion) and the construction of new
dwellings (£13.6 billion). The reduced 5 per cent rate for domestic fuel and power
adds a further £4.9 billion. Again, applying the standard 20 per cent rate to all
reduced and zero-rated items would not yield the Exchequer the full £52.7 billion. But
this is one example where narrow behavioural responses might be relatively muted – it
is difficult to avoid consuming food, or, for those who need them, children’s clothing
or prescription drugs. For those who chose to switch from bus travel to cycling due to
25
The two reliefs for registered pension schemes – against income tax and NICs – are combined into a single bullet.
26
HMRC, Personal pension statistics, 2019.

Revenue risks
the imposition of VAT on bus tickets, their helmet would now incur VAT too. But, of
course, a £50 billion tax rise would have wider implications for economic growth.
• Private residence relief from capital gains tax (CGT), which was worth £27.2 billion in
2018-19 (1.3 per cent of GDP and three times greater than CGT receipts collected). In
2005-06 this relief cost 0.9 per cent of GDP, so its cost has risen too. Any capital gain
from the sale of a primary residence is, on the whole, exempt from CGT. House price
inflation explains why the cost has risen. This is another relief where there would be
significant behavioural effects were it to be abolished. By lowering the post-tax return
from selling a house and crystallising a tax liability when the transaction took place, its
effects would be equivalent to those of a very large increase in stamp duty: reducing
house prices and deterring transactions on a significant scale. Such consequences
would reduce the amount of CGT it would raise, and would also hit receipts from
property transactions taxes and on the spending associated with house moves.
4.75 Chart 4.14 shows the estimated cost of tax expenditures relative to GDP from 2005-06 to
2018-19, and a projection consistent with our March 2019 forecast up to 2023-24.27 The
cost has generally risen over time, although it fell sharply in 2008-09 and 2009-10 thanks
to falling house prices and the collapse in residential property transactions. It reached a
recent peak of 7.8 per cent of GDP in 2015-16 and stands at 7.6 per cent in 2018-19. Our
forecast suggests that the cost of reliefs will fall back in 2019-20, and then edge higher
relative to GDP over the remainder of the period.
4.76 The cost of zero and reduced rates in VAT has increased because the January 2011
increase in the main rate to 20 per cent made the per-unit cost of the relief greater. The cost
of income tax reliefs has fallen as pensions tax relief has been made less generous. Both the
growth and the year-to-year variation in the cost of relief from capital taxes is largely due to
CGT private residence relief, plus rising costs of inheritance tax reliefs and entrepreneurs’
relief. Greater relief from corporation tax is largely due to policy decisions relating to R&D
expenditure, the ‘creative’ industries and the introduction of the ‘patent box’.
27
Our forecast methodology is explained in more detail in the supplementary tables accompanying this report. For those treated explicitly
in our forecast we use those estimates. For the remaining cases we make the simple assumption that expenditure grows in line with the
relevant tax forecast.

Revenue risks
Chart 4.14: Tax expenditures as a percentage of GDP

10
Other Corporation tax
Forecast
Capital taxes NICs
9
Income tax VAT
Total
8
6
Per cent of GDP
0
2005-06 2007-08 2009-10 2011-12 2013-14 2015-16 2017-18 2019-20 2021-22 2023-24
Source: HMRC, OBR
International comparisons
4.77 The IMF considers HMRC’s coverage of tax expenditures to be ‘good’ by international
standards, but adds that “there is no control on, or budgetary objectives for, the size of tax
expenditures, which are relatively high by international standards.”28 Chart 4.15 is drawn
from the IMF’s 2016 Fiscal transparency evaluation of the UK. It shows that the revenue cost
of tax expenditures in the UK is relatively high when compared to other countries, at around
7 per cent of GDP at the time of this snapshot, and 7.6 per cent now.
28
IMF, United Kingdom fiscal transparency evaluation, 2016.

Revenue risks
Chart 4.15: International comparison of the cost of tax expenditures in 2010

Australia
Italy
Guatemala (2009)
UK (2018-19)
United States
UK (2010-11)
Dominican Rep.
Chile
Uruguay
Poland
Mexico
Spain
Greece
Brazil
Austria
Denmark
Norway
France
Peru
Argentina
Canada
Netherlands
Turkey
Switzerland
Germany
Korea
Portugal
0 1 2 3 4 5 6 7 8 9
Note: Estimates are for 2010, unless otherwise stated. Per cent of GDP
Source: IMF, HMRC, OBR
4.78 There are some thresholds that are not captured within HMRC’s definitions of either
structural reliefs or tax expenditures, but where the level set is within the Government’s
control. The VAT threshold is a good example where, as Chart 4.16 shows, the UK has
chosen to set the highest rate across the OECD.
Chart 4.16: VAT threshold cross-country comparison

90
80
70
60
£ thousand
50
40
30
20
10
0
Belgium
Finland
Switzerland
Norway
Ireland
Korea
Lithuania
Australia
Iceland
Luxembourg
United Kingdom
Slovenia
Sweden
Germany
Latvia
Canada
Czech Republic
France
Austria
Slovak Republic
Denmark
Poland
Portugal
Italy
Estonia
Japan
Greece
Netherlands
Hungary
Israel
New Zealand
Note: Thresholds are for 2018. The exchange rates in the OECD analysis for conversion are purchasing power parity rates (PPPs) for
GDP. The implied exchange rate is £1=$1.40 USD.
Source: OECD

Revenue risks
4.79 Researchers at the Council on Economic Policies in Switzerland recently published a

comprehensive comparative assessment of tax expenditure reporting in 43 G20 and OECD
countries.29 They use nine criteria to judge ‘best practice’, including the frequency of
reporting, whether there is a legal requirement to report, whether reports are integrated into
budgets, the number of estimations and the quality of accompanying descriptions.
4.80 They conclude that the UK falls into a group of 26 countries they deem to produce only
“basic” reporting of tax expenditures. As with the IMF report, the authors praise the UK for
producing a regular annual publication on their estimated cost. But they note that the
proportion of expenditures reported is relatively low, and that failing to produce the
estimates alongside budgets or fiscal statements is a weakness. For nine countries –
Australia, Austria, Canada, France, Germany, Italy, Netherlands, Korea and Sweden – the
authors conclude that the reporting of tax expenditures is “detailed and comprehensive”.
Tax reliefs and expenditures: six case studies

4.81 We have selected six reliefs that highlight some of the risks identified earlier: pensions tax
relief, R&D tax credits, entrepreneurs’ relief, inheritance tax agricultural and business
property reliefs, creative sector reliefs and the ‘patent box’. Comparing them, we note that:
• The cost has grown in five of the examples, as shown in Chart 4.17, but in most cases,
the reasons behind that growth are not fully understood. This suggests a stronger role
for monitoring to enable the Government to manage the fiscal risk more effectively.
• Only pensions tax relief has seen the cost decline. Policy changes have reduced its
generosity, helping to manage the narrow fiscal risk it poses. But this has been
partially offset by introducing more generous tax treatment in the savings regime.
• The R&D tax credit scheme is the only one to have been evaluated by HMRC,30 though
the evaluation does not cover the period after the scheme was made more generous.
For the remaining case studies there is little evidence about whether policy objectives
are being met. There is a risk that, for some, the cost is purely ‘deadweight’, in other
words that it is offering a tax break for something that would have happened anyway.
• Some represent areas where the Government has chosen to tighten the rules after
evidence mounted that the reliefs in question were being abused.
29
Redonda and Neubig, Assessing tax expenditure reporting in G20 and OECD economies, 2018.
30
HMRC, Evaluation of research and development tax credit, 2015.

Revenue risks
Chart 4.17: Tax expenditure cost estimates: six case studies

3.5 0.25
Pensions Forecast R&D Forecast
3.0
0.20
2.5
Per cent of GDP

Per cent of GDP
0.15
2.0
1.5
0.10
1.0
0.05
0.5
0.0 0.00
2010-11 2013-14 2016-17 2019-20 2022-23 2010-11 2013-14 2016-17 2019-20 2022-23
0.25 0.25
Entrepreneurs' relief Forecast Inheritance tax Forecast
0.20 0.20
Per cent of GDP
Per cent of GDP
0.15 0.15
0.10 0.10
0.05 0.05
0.00 0.00
2010-11 2013-14 2016-17 2019-20 2022-23 2010-11 2013-14 2016-17 2019-20 2022-23
0.25 0.25
Creatives sector Forecast Patent box Forecast
0.20 0.20
Per cemt of GDP
Per cent of GDP
0.15 0.15
0.10 0.10
0.05 0.05
0.00 0.00
2010-11 2013-14 2016-17 2019-20 2022-23 2010-11 2013-14 2016-17 2019-20 2022-23
Note: Pensions reliefs are on a different scale.
Source: HMRC, OBR
4.82 Table 4.1 shows our latest forecasts for each of the six reliefs covered in this section.

Revenue risks
Table 4.1: Tax expenditure cost forecasts: six case studies

£ billion
2017-18 2018-19 2019-20 2020-21 2021-22 2022-23 2023-24
Pensions 54.7 57.9 59.4 62.8 65.2 67.8 70.6
R&D 3.9 4.0 4.1 4.0 4.1 4.2 4.4
Entrepreneurs' relief 2.3 2.4 2.3 2.5 2.5 2.7 3.0
Inheritance tax 1.0 1.1 1.1 1.1 1.1 1.2 1.3
Creative sector 0.9 1.0 1.0 1.0 1.1 1.1 1.2
Patent box 1.1 1.1 1.1 1.1 1.1 1.2 1.2
Pensions tax relief

4.83 The Chancellor has previously described the cost of pensions tax relief as “eye-wateringly
expensive”31 and “extraordinarily generous”.32 The gross amount of income tax and NICs
relief on pension contributions was £54.7 billion in 2017-18 (£38.4 billion in income tax
and £16.3 billion in NICs) or around 2.6 per cent of GDP.33 The cost rose from 2.6 per cent
of GDP in 2005-06 to a peak of 3.1 per cent in 2010-11, before declining to around its
current level in 2013-14 and remaining relatively stable since (see Chart 4.17). The spike in
2015-16 is likely to reflect high-income individuals making additional contributions that
year in anticipation of pre-announced policy changes to tax relief that would adversely
affect them. Employer pension contributions are four times greater than those from
employees, while contributions to occupational pension schemes are three times greater
than those to personal pensions.
4.84 The relatively flat profile in recent years hides some opposing effects. The largest upward
pressure has been from the introduction of automatic enrolment in 2012. Since then, an
estimated 10 million workers have been automatically enrolled, contributing to a year-on-
year increase in the number of individuals contributing to a pension and, where they have
income tax liabilities, receiving tax relief. The rising employment rate will also have played a
role, though this is tempered by relatively slow growth in average earnings.
4.85 The largest downward pressure comes from successive restrictions to the lifetime allowance
and the annual allowance, plus the introduction of the annual allowance taper.34 These
restrictions have had a material impact. The latest statistics show that the value of pension
contributions exceeding the annual allowance was £517 million in 2016-17, compared to
£143 million the year before and just £6 million in 2010-11. The Government has recently
signalled it will seek to introduce a ‘more flexible’ approach to the NHS pension scheme, to
counter some of the difficulties the annual allowance taper has created for affected
31
Speech at the IMF Annual Meeting in Bali, 12 October 2018.
32
Evidence to the Treasury Select Committee, 24 April 2019.
33
For our analysis we use gross income tax relief, that is before tax on payments from pension schemes is deducted. This is available in
HMRC, Personal pension statistics, 2019. HMRC’s Estimated costs of principal tax reliefs only reports net income tax relief, i.e. after the tax
has been deducted.
34
The annual allowance sets how much an individual can contribute each year to pension pots before contributions can no longer be
made out of pre-tax income. It was reduced from £255,000 in 2010-11 to £50,000 the following year, and then to £40,000 from 2014-
15 onwards. The taper, which progressively reduces the annual allowance for income over £150,000 began in 2016-17. It stops at
income of £210,000, by which point the annual allowance falls to £10,000. The lifetime allowance restricts the total amount of tax-
relieved contributions that an individual can accumulate in their pension pots. It has progressively fallen from £1.8 million in 2011-12 to
£1 million in 2017-18. It has been increased in line with CPI inflation since then.

Revenue risks
employees. Breaching the lifetime allowance incurs a charge, and the amount collected
from this has risen too, from £12 million in 2010-11 to £102 million in 2016-17.35
4.86 A smaller downward effect comes from the rise in the number of self-employed individuals
and a reduction in the proportion of them contributing to a pension scheme. In 2010-11,
16 per cent of them did, but this has since halved to 8 per cent in 2016-17.
4.87 It is also worth remembering that while the Government has been making the tax treatment
of pensions contributions less generous for high earners, it has been making the tax
treatment of savings more attractive. For example, the limit for the annual amount that can
be saved in a tax-free individual savings account (ISA) has been increased from £7,200 in
2009-10 to £20,000 in 2017-18. During that time the cost of income tax relief on ISAs
doubled, from £1.6 billion in 2009-10 to £2.9 billion in 2017-18.36 The same high earners
that are affected by the pensions restrictions are the ones most able to gain from the
increased ISA limits.37 The most recent statistics – for 2016-17 – show that 60 per cent of
those ISA subscribers with incomes of £150,000 or more saved the maximum amount.38
4.88 In 2015 the previous Chancellor signalled his intention to introduce major pensions reforms
but in the event decided not to. Since then the Government has not signalled any plans to
revisit them. It did state in MFR that “pensions tax relief encourages people to save for their
future”, suggesting that the policy rationale would require at least some form of tax relief to
be retained even if the system were reformed in future. Restricting tax relief on pension
contributions further, perhaps combined with greater relief when pensions are drawn down
in retirement, was one option mooted under the previous Chancellor. This might, of course,
prove politically unpopular, with the costs for those affected being felt upfront while the
benefits for post-tax retirement incomes would accrue some way in the future.39
Research and development tax credit

4.89 The research and development (R&D) tax credit is a complex set of directly payable and
reduced liability corporation tax credits designed to incentivise expenditure on innovation
activities. R&D tax credits are a long-running programme whose structure has changed
several times, but whose overarching characteristics have remained broadly constant: the
scheme allows companies to deduct their expenditure on R&D-related activities for taxable
income purposes, and gives a more generous incentive for smaller companies. In April
2013 ‘research and development expenditure credits’ were introduced (RDEC, also known
as ‘above-the-line’ credits). This scheme is more generous scheme for large companies than
the R&D tax credits that had previously been available to them.
35
All data from HMRC, Personal pension statistics, 2019.
36
HMRC, Estimated costs of principal tax reliefs, 2019 (and previous editions).
37
Other generous savings measures include the savings allowance, the help-to-buy ISA and the lifetime ISA. See OBR, Private pensions
and savings: the long-term effect of recent policy measures, 2016.
38
This was £15,240 at the time (HMRC, Individual savings accounts statistics, 2019).
39
See HM Treasury, Strengthening the incentive to save: a consultation on pensions tax relief, 2015. For a discussion, see Institute for
Fiscal Studies, Green Budget, 2014.

Revenue risks
4.90 The combined cost of the schemes has increased significantly, from £0.5 billion in 2003-04
to £1.6 billion in 2013-14 (the first year of the RDEC scheme) and to £3.5 billion in 2016-
17. Part of the reason for the most recent jump is HMRC revising its outturn data from
2014-15 onwards, to account for claims that had previously been missed in its analysis. It is
possible that years prior to 2014-15 will also be revised eventually, but this does not detract
from the sharp rise in the cost of these schemes in recent years. An increase in take-up,
through greater awareness and the introduction of more generous schemes, is the most
likely explanation. But in the absence of any firm information to help us determine whether
this rise represents a step change in the level or a trend that will persist, we have assumed a
more modest pace of growth in our forecast – rising to £4.4 billion by 2023-24.
4.91 Chart 4.18 shows total R&D expenditure, as reported by the ONS, and the total expenditure
used to claim R&D tax credits, as recorded by HMRC. Methodological differences mean the
two are not directly comparable,40 but it does demonstrate that the expenditure used to
claim R&D tax credits has been increasing at a faster rate. Even with the methodological
caveats borne in mind, the fact that the HMRC measure exceeds the ONS one by such a
significant amount seems worthy of further investigation. That it moves from below to above
the ONS estimate in 2014 seems more likely to relate to the revisions that HMRC has only
taken back to 2014-15 rather than to the introduction of the more generous RDEC scheme
for larger firms, but a role for the latter cannot be ruled out.
Chart 4.18: Expenditure in R&D tax credit claims versus total R&D expenditure
30
Total R&D expenditure (ONS)
Expenditure used to claim R&D tax credits (HMRC)
25
20
£ billion
15
10
0
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016
Note: Reflects Figure 9 of HMRC's Research and Development Tax Credits Statistics publication. 2016-17 reflects partial data and is
expected to increase as more returns are received by HMRC.
Source: HMRC, ONS
40
One difference is that overseas expenditure is not included in the ONS measure but may qualify for tax relief.

Revenue risks
4.92 Evidence shows that schemes like R&D tax credits can be effective in generating additional
R&D, and can do so cost effectively.41 But as with any tax relief, as it becomes more
generous and/or more complex, it increases incentives to re-badge existing expenditure as
qualifying R&D or to engage in fraudulent claims. An example of an attempt to control the
latter is the reintroduction of a PAYE cap on the amount of payable R&D tax credit than can
be claimed by a company under the smaller companies scheme. The cap was removed in
2012 but was brought back at Budget 2018 “to help prevent abuse”.
Entrepreneurs’ relief
4.93 Entrepreneurs’ relief allows directors of companies with significant stakes in them (as well as
certain other taxpayers) to pay a lower tax rate of 10 per cent on disposals of shares below
a certain threshold, rather than the much higher headline capital gains tax (CGT) rate. It
was introduced by the Labour Government in 2008, at the same time as an 18 per cent flat
rate of CGT. Initially the annual cost was expected to rise to £0.9 billion by 2013-14.42 In
fact, it turned out to be three times that and continued rising to £4.2 billion (0.2 per cent of
GDP) in 2015-16. This figure halved to £2.1 billion the following year, the most recent
outturn. We forecast it to rise steadily from that level to reach £3.0 billion in 2023-24.
4.94 The sharp initial rise is most likely due to an increase in generosity. At the June 2010 Budget
the Coalition Government increased the CGT rate for higher rate income tax payers to 28
per cent, significantly widening the differential to the unchanged 10 per cent entrepreneurs’
relief rate. Between the March 2010 Budget and Budget 2011, its generosity increased
further as the lifetime limit on gains that can benefit from the lower rate was progressively
raised, from £1 million to £5 million (at the June 2010 Budget) and then to £10 million.
4.95 The reason for the cost of entrepreneurs’ relief falling so suddenly in 2016-17 is not known.
It might be related to a new 20 per cent CGT rate that applies to gains not from property or
carried interest. The number of claimants dropped by around 20 per cent that year, and the
gross gains on which entrepreneurs’ relief was claimed fell by 12 per cent. But it is not clear
why these led to a 50 per cent drop in the overall cost of the relief. While CGT receipts can
be volatile, HMRC offered little commentary on the drop when it released the latest estimate.
4.96 HMRC’s statistics show that the gains from this relief are increasingly concentrated on a
small number of wealthier claimants. In 2016-17 there were 5,000 individuals making
gains of greater than £1 million, and, in aggregate, they made up three quarters of the
total gains for which entrepreneurs’ relief was claimed. The comparative figure in 2010-11
was around a half. These 5,000 claimants benefited by an average of £325,000 each.43
4.97 Evidence also suggests that entrepreneurs’ relief plays very little role in entrepreneurial
decision-making. A survey of those that benefited from it showed that, at the point of
investment, only 8 per cent were influenced by it and 84 per cent were not even aware it
41
See: HMRC, Evaluation of research and development tax credit, 2015; Guceri and Lu, Effectiveness of Fiscal Incentives for R&D: a quasi-
experiment, 2015; Dechezleprêtre et al, Do tax incentives for research increase firm innovation? An RD Design for R&D, 2015.
42
NAO, Effective management of tax reliefs, 2014.
43
This calculation is based on assuming the taxpayer pays the 10 per cent entrepreneurs’ relief rate instead of the 20 per cent rate that
applies to gains on everything other than property and carried interest.

Revenue risks
existed. Even at the point of disposal, only 16 per cent said they had been influenced by the
relief, with 27 per cent still blissfully unaware but no doubt pleased to receive the windfall.44
4.98 Entrepreneurs’ relief meets many characteristics of a relief that poses a fiscal risk. It has
increased significantly in cost, and the reasons for the rise (and recent fall) are largely
unknown, raising the possibility that the assumptions underpinning our forecast fail to factor
in something that has affected its cost in the past. At Budget 2018 the Government brought
in two new measures to tighten the rules around eligibility, and to address “identified
abuse”, suggesting this has also been a factor. It does not have a well-specified policy
objective, other than making the UK “a more attractive location for entrepreneurs”.
Inheritance tax: business and agricultural property reliefs

4.99 Estates that qualify for business property relief (BPR) can reduce the amount of chargeable
inheritance tax (IHT) by either 50 or 100 per cent. It applies not only to a business, but also
to certain company shares, and qualifying land and machinery. Agricultural property relief
(APR) is available for qualifying agricultural land and properties.45 The objective of these
reliefs is to protect businesses and farms from being broken up as a means to pay an IHT
liability. They are designed to enable estates to be passed down to the next generation
following the death of the owner. (Although evidence suggests that family-owned firms
being passed from one generation to the next tends to weigh on productivity growth.46)
4.100 The cost of the two reliefs has increased from £0.6 billion in 2010-11 to an estimated £1.1
billion in 2018-19 (0.1 per cent of GDP and 20 per cent of IHT receipts). This is primarily
due to BPR, the cost of which has more than doubled to £0.7 billion during that time. We
expect these two reliefs to cost £1.3 billion by 2023-24, rising in line with IHT liabilities.
4.101 Part of the explanation for the rising cost of these reliefs is simply that IHT receipts have
risen during this period. But it also seems likely to be a consequence of the increased
awareness and popularity of these reliefs for tax planning purposes. Research from Savills
identified that the proportion of agricultural land bought by farmers fell from 60 per cent in
2011 to 40 per cent in 2017, reflecting the increasing popularity of holding agricultural
land for ‘lifestyle’ purchasers.47 Estate agents actively promote APR as an investment
opportunity and a tax-efficient means of sheltering wealth.48 There is probably even greater
risk around the use of BPR, where “anyone can invest money in relevant AIM shares – with
no personal relation whatsoever to those companies”.49 Investment companies are actively
marketing BPR as part of an estate planning strategy. Even the Government’s own ‘Money
Advice Service’ website has a page devoted to the ‘Top 5 ways to cut your Inheritance Tax’.
44
IFF Research, Capital gains tax entrepreneurs’ relief: behaviours and motivations, 2017, as reported in Resolution Foundation,
Entrepreneurs’ relief has cost £22 billion over the past 10 years. Was it worth it?, 2018.
45
A fuller definition is available on HMRC’s website, as well as in Office of Tax Simplification, Inheritance tax review – second report:
simplifying the design of inheritance tax, 2019.
46
See, for example, Bloom and Van Reenen, Measuring and Explaining Management Practices Across Firms and Countries, 2006.
47
Savills, GB agricultural land, 2018.
48
Savills, Global market tips, February, 2016.
49
Resolution Foundation, Passing on: options for reforming inheritance taxation, 2018.

Revenue risks
4.102 Chart 4.19 shows the average effective tax rate paid by estates in 2015-16. It shows that
the wealthiest estates can plan their affairs to reduce their liability. The use of APR and BPR
will have played a part, though there are several other ways to manage IHT liability.50 Data
obtained by ‘Tax Justice’ provides some interesting insight to the value of these reliefs to the
wealthy. It shows that over 70 per cent of the reliefs’ combined cost in 2015-16 (£0.7 billion
out of £0.9 billion) went to just 495 estates benefitting by £1.3 million each on average.51
Of course, the value of these reliefs to the wealthy is also simply a function of the high value
of their estates and the fact that they are more likely to have business assets. They are also
more likely to have the means to reduce their IHT liability via gifts and transfers to charities.
Chart 4.19: Inheritance tax average effective tax rate by net estate value
25
20
Average effective tax rate (per cent)
15
10
0
0-1 1-2 2-3 3-4 4-5 5-6 6-7 7-8 8-9 9-10 10+
Net estate value (£ million)
Source: Office of Tax Simplification
4.103 The Government has not formally evaluated whether the original policy objectives for APR
and BPR are being met. But the current use of both seems far removed from those
objectives, and this appears to have contributed to the rise in their cost.
4.104 In addition, the introduction of ‘pensions flexibility’ over defined contribution pension funds
in 2015 might pose some future risks. This gives individuals the flexibility to withdraw their
pension funds from age 55, subject to tax paid at their marginal rate. Prior to this, they
needed to convert them into an annuity, that would then be subject to income tax. The
pensions flexibility rules allow individuals to leave money in their pension pots (without
annuitising it) and then bequeath any that remains to anyone without an IHT liability.52
50
See Office of Tax Simplification, Inheritance tax review – first report: overview of the tax and dealing with administration, 2019.
51
Tax Justice, In stark relief, 2019.
52
This is discussed in more detail in Institute for Fiscal Studies, Green Budget, 2018.

Revenue risks
Creative sector reliefs

4.105 Creative sector tax reliefs provide eligible companies with a way to increase the amount they
can deduct against corporation tax. This means that a company’s taxable income is reduced
by more than one pound for every pound spent on expenditure qualifying for relief. For
loss-making companies there may be the possibility of benefiting from a payable tax credit.
Relief is available for film production, ‘high-end’ television, animation production, video
games, orchestral concerts, theatrical productions, children’s television, and museum and
galleries exhibitions. Films, television programmes, animations and video games must also
pass a ‘cultural test’, and the Government has outsourced the certification of it to the British
Film Institute. It has also decided that such a test is unnecessary for theatrical productions,
orchestral concerts and exhibitions in museums and galleries.
4.106 Film tax reliefs were introduced in 2007 and the other creative reliefs have been introduced
progressively since 2013-14. Their cost has risen significantly – from £0.2 billion in 2010-
11 to an estimated £0.9 billion in 2017-18. This rise is mostly due to changes in film tax
relief, unexpectedly high take-up and the introduction of new reliefs. We forecast the cost of
creative reliefs to rise to £1.2 billion in 2023-24. Film tax relief alone accounts for over half
of this – 10 years after its introduction, the cost has trebled.
4.107 While the popularity of these creative reliefs is clear, in the absence of a policy evaluation it
remains unclear whether they meet their policy objectives. There are also examples where
HMRC has successfully challenged some film schemes on the grounds that they were being
used as tax avoidance vehicles, suggesting some of the rising cost might be from abuse.
Patent box
4.108 The ‘patent box’ rewards intellectual property (IP) that is commercialised in the UK by
lowering the corporation tax (CT) rate on profits made from those patents. The original
scheme was introduced in April 2013 and applied a 10 per cent CT rate to profits earned
after that date. However, the full benefit of the reduced rate was phased in over four years,
making it progressively more generous each year. The main rate of CT in 2013 was 20 per
cent. It was cut to 19 per cent in 2017 and is set to fall to 17 per cent in 2020.
4.109 The stated policy objective for the patent box at the time of its introduction was to “provide
an additional incentive for companies to retain and commercialise existing patents and to
develop new innovative patented products” in the UK.53 But the Government at the time also
stated that it was “focusing on scientific and high-tech IP because of their particularly strong
link to Research and Development (R&D) and technical innovation activities”.54 When it was
announced in 2010 the IFS described it as “poorly targeted at research” and argued that it
would “add complexity to the tax system and require policing to ensure that both income and
costs are being appropriately assigned to patent”.55
53
HMRC Tax information and impact note, Corporation Tax Reform: Patent Box, December 2011.
54
HM Treasury, Corporate Tax Reform: delivering a more competitive system, November 2010.
55
Institute for Fiscal Studies, The UK will introduce a Patent Box, but to whose benefit?, 2010.

Revenue risks
4.110 In 2016, to comply with new OECD guidelines for tax-favoured IP regimes that were
published in October 2015 as part of its Base Erosion and Profit Shifting (BEPS) project, the
patent box was significantly amended. This ensured it was compliant with the OECD’s new
‘nexus approach’, linking the benefit allowable to the proportion of relevant R&D
expenditure by the company (or unconnected subcontractors) on the patents in question. But
transitional arrangements allow those companies that entered the scheme before July 2016
to continue to benefit from the original rules until July 2021. The latest estimate is that the
patent box cost £1.1 billion in 2018-19. We forecast it to rise to £1.2 billion in 2023-24.
4.111 HMRC statistics show that in 2016-17, the most recent year for which it has outturn data, 96
per cent of the relief claimed was from large companies. This, and the fact that UK-
domiciled companies can, for now, continue to gain a UK tax relief even when conducting
R&D overseas, suggests that some of the costs of the patent box might also be ‘deadweight’.
This is reinforced by the fact that the patent box applies to all IP, rather than just new IP,
subject to a ‘qualifying development’ condition.56
Conclusions
4.112 Managing the fiscal risks around tax reliefs and expenditures is an area where the
Government’s actions do not seem to match its ambition. While keen on simplifying the tax
system, the Government has introduced around 100 new tax policy measures since our
previous report, and it last asked the Office of Tax Simplification to review the topic in 2011.
While the removal of some reliefs has been announced, such as the ‘wear and tear
allowance’ for landlords and NICs relief for termination payments, others have been
introduced, including the ‘first-time buyers’ relief from stamp duty land tax and ‘public
lavatories relief’ from business rates. The Government has not signalled any plans to
develop metrics that might allow it to measure its progress in simplifying the tax system.
4.113 The Government states that “HMRC in partnership with [the Treasury] continuously monitor
tax reliefs” but, despite this, did not offer us any explanation for why, for example, the cost
of R&D tax relief or entrepreneurs’ relief has been rising or why the cost of entrepreneurs’
relief apparently halved in a single year. More generally, few of the reliefs have been
evaluated, either in terms of cost or of effectiveness in meeting stated objectives.
4.114 Following an NAO recommendation, HMRC is undertaking work to identify more reliefs and
expenditures and this will be reflected in a more comprehensive list that it publishes and
also the number of them for which it estimates an annual cost.
4.115 Overall, the IMF’s 2016 recommendation that the Government “control tax expenditures by
subjecting them to numerical limits and enhance reporting through the addition of a sectoral
and functional breakdown in cost publications” appears to remain valid.57
56
The company making the claim must have either contributed to the creation of the patented invention or performed a significant activity
in its development.
57
IMF, United Kingdom fiscal transparency evaluation, 2016.

Revenue risks
Taxation in a digitalised economy

Introduction
4.116 The technological innovations of recent years have had a dramatic impact on people’s lives.
Personal computing, the internet, smartphones and social media networks have transformed
the way we interact, how and what we consume, and even how we work. A few taps on a
screen and a product could be on its way to you from the other side of the world. The
product is not physical but digital – perhaps an app developed by collaborators in Mumbai
and Berlin. You downloaded the app from the ‘Xiaomi Market’ app store, based in China,
though it is built on Android, which is open-source software, though it is also owned by
Google. And you have ‘purchased’ it for free, as have most of your friends and thousands
of others. Despite that, everyone involved seems to be making money, and it may be not
immediately obvious what, if anything, is taxable and which country has the right to tax it.
4.117 This stylised example highlights some of the tax challenges posed by a digitalised economy,
where the sheer speed of change has led to concerns that traditional tax systems are
lagging behind. In this section we explore some of those challenges but also some of the
opportunities digitalisation offers, particularly for improved administration of the system.
The digitalised economy

4.118 If we take a wide definition of the digitalised economy – all activities that make use of digital
information and data, and all activities with an online presence – then the digitalised
economy is largely indistinguishable from the whole economy. Chart 4.20 shows the
percentage of businesses within OECD countries that had a website in 2018 versus 2009.58
It shows that not only did the percentage increase in almost all countries, but also that in
most countries the majority of businesses have an online presence – indeed, the proportion
is above 80 per cent in 15 countries, including in the UK.
58
Businesses are defined as those with ten or more employees.

Revenue risks
Chart 4.20: The proportion of businesses with a website

100
2018
90
2009
80
70
60
Per cent
50
40
30
20
10
0
Luxembourg
Australia
Belgium
Ireland
Korea
Finland
Norway
Switzerland
Lithuania
Iceland
Turkey
Canada
United Kingdom
Sweden
Germany
Portugal
Austria
Poland
Slovenia
Czech Republic
France
Slovak Republic
Denmark
Colombia
Latvia
Greece
Brazil
Italy
Spain
Netherlands
Estonia
Hungary
Japan
Mexico
United States
New Zealand
Note: For the '2009' series: Canada and United States = 2007; Iceland = 2010; Mexico and New Zealand = 2008.
Note: For the '2018' series: Australia, Brazil, Canada, Japan and Korea = 2017; Colombia and Netherlands = 2016; Mexico Note:
and United States = 2012; Switzerland = 2011.
Source: OECD
4.119 Another way of looking at the growth in the digitalised economy is to look at global internet
traffic since the launch of the world wide web in the early 1990s, which, as Chart 4.21
shows, has been rising at an exponential rate in recent years. While the first commercial
browser – the Netscape Navigator – was launched in October 1994, and music downloads
became popular in the late 1990s, it was not until the mid-2000s, including the births of
YouTube and Facebook, that traffic really took off. Cisco Systems expect this to continue,
forecasting a tripling in traffic over the next five years. The rate of growth is even faster for
mobile internet traffic. The use of mobile devices to access the internet only became popular
in the late 2000s, not least due to the launch of the first iPhone in 2007.
4.120 The ONS estimates that nine out of ten homes in the UK in 2018 had internet access, versus
just one in ten 20 years ago, and that 86 per cent of adults were online daily in 2018, up
from 35 per cent 12 years earlier, with smartphones overwhelmingly the device of choice.59
59
ONS, Internet access – households and individuals, 2018.

Revenue risks
Chart 4.21: Growth in global internet traffic

400
Total internet protocol traffic Cisco Systems'
forecast
Fixed internet traffic
350
Mobile internet traffic
300
Exabytes per month
250
200
150
100
50
0
1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022
Source: Cisco Systems
Why does digitalisation pose a fiscal risk?

4.121 Much of the economic activity that takes place within a digitalised economy seems at odds
with the traditional tax system. Transactions may be global, rather than domestic; goods
may be digital rather than physical; services may be provided remotely rather than locally.
New business models have led to changes in what is consumed, for example internet
searches or social media interactions, and how these are priced, often free at the point of
purchase. Instead, those searches and interactions have been monetised by search engines
and social media networks being able to sell customer information to businesses that value
the power of targeted advertising.
4.122 Value creation is still a function of underlying economic inputs and activities, but it is much
harder to know in which tax jurisdiction the value was created and therefore which
government has the right to tax it. The ‘Bean Review’ of UK economic statistics sums up the
difficulty of measuring activity in the digitalised economy: “The nature of digital products has
led to business models where it is harder for the statistician to observe both transactions and
a corresponding price. The great challenge for economic measurement stems from the fact
that the consumption of digital products often does not involve a monetary transaction that
corresponds to its value to consumers. Digital products delivered at a zero price, for instance,
are entirely excluded from GDP, in accordance with the internationally-agreed statistical
standards.”60 These challenges are equally important to tax administration and compliance.
4.123 The all-encompassing nature of digitalisation means that its effects will be felt to some
extent across the entire tax system. In this section we focus on a relatively small number of
examples that can be grouped under two broad categories:
60
Bean, Independent review of UK economic statistics, 2016.

Revenue risks
• The challenges for tax policy design. The OECD’s base erosion and profit shifting
(BEPS) project identifies issues around taxing multinational profits that threaten the
traditional corporation tax base. Digitalisation exacerbates these issues by making it
easier for companies to be international, and by enabling new business models that
make it harder to define the connection between company and tax jurisdiction and to
determine if there is a taxable presence. The growth of online retailing is largely at the
expense of ‘bricks and mortar’ retailing. Online transactions might be less readily
observable to HMRC, particularly where the seller is based abroad. The primary risk
then is to the VAT base, but there are also consequences for business rates and, post-
Brexit, to customs duties. We discussed the trend towards self-employment and its
associated revenue risks earlier in this chapter. This growth has been aided by the
popularity of online peer-to-peer platforms and the new ways of working they allow.
The main risk arises from the loss of income tax and National Insurance contributions.
• The opportunities for tax administration. HMRC’s “ambition is to become one of the
most digitally advanced tax administrations in the world”.61 While this is partly driven
by the objective to improve the ‘customer experience’, it also reflects that digitalisation
has led to the creation of a wealth of new data and information that might be used to
improve tax compliance and reduce tax gaps. Many of the ‘behaviours’ that HMRC
identifies in constructing its tax gap estimates are facilitated by an information
constraint. Most obviously, tax evasion related to criminal activity and the ‘hidden
economy’ is not easily observed. But tax avoidance, whether through contrived
arrangements or differences in legal interpretation, often builds upon asymmetric
information too. This is borne out by the results of HMRC’s audits, which tend to
increase rather than lower the amount of tax due.62 Better information for the taxpayer
might also reduce errors and ‘failure to take reasonable care’. Again, the evidence
suggests correcting these increases tax revenue on average. Harnessing the
opportunities that digitalisation presents can help relax information constraints.
4.124 One hypothesis that flows from the debate about digitalisation and the measurement of
GDP is that current measurement is not fully capturing the positive effects of digitalisation
and its innovations, for example because some products have a zero price. So real GDP per
head might be higher, and growing faster, than prevailing GDP estimates suggest. From a
tax revenue perspective, this does not necessarily mean that receipts are also being
underreported. If some of the unmeasured benefit from these innovations is already being
realised, perhaps in higher productivity, then it is likely that some positive impacts on tax
revenues are already being captured in the taxes that are being paid. If it were captured in
nominal GDP data too, effective tax rates would be lower than we currently believe. If the
issue is that estimates of nominal GDP are overstating prices and understating real GDP,
this would tell us something about living standards (the volume of goods and services being
enjoyed is higher than measured) but not about taxes (where only nominal GDP matters).
61
HMRC, Overview of making tax digital, 2019.
62
Advani, Elming and Shaw, The dynamic effects of tax audits, 2019.

Revenue risks
4.125 More fundamentally, there are some that believe the age of transformative technological
innovation has drawn to a close.63 Others take a more optimistic view, seeing the recent
phase of digital technologies as a prelude to a new era of economic prosperity. Their
argument is that “the transformations brought about by digital technology will be profoundly
beneficial ones” but that there is always a lag between major innovations and when their
full economic benefits are realised.64 If the optimistic view proves correct, and the benefits of
innovations in artificial intelligence, 3D printing and unmanned aerial vehicles, etc,
generate longer-term increases in GDP, then that would support tax receipts. But typically,
such effects build slowly, and governments are adept at finding ways to spend the proceeds.
Tax policy design

Taxation of multinational profits
4.126 Debating how to tax the profits of multinational companies that operate across international
tax jurisdictions is not new. Agreements allocating taxing rights between a ‘source’ and
‘residence’ country to avoid ‘double taxation’65 have existed since the 19th century,66 and
model tax treaties were first developed by the League of Nations in the 1920s.
4.127 Many treaties follow the principle that profits are subject to tax in a country only once the
company generating them meets the definition of a ‘permanent establishment’. Only then
can it properly be “regarded as participating in the economic life of that other State to such
an extent that the other State should have taxing rights on its profits”.67 Defining what
constitutes a permanent establishment can be complex, for example in respect of a ‘fixed’
place of business through which the company’s business operations are conducted.
4.128 International tax rules generally state that countries are entitled to tax those profits that are
attributed to activities undertaken through the permanent establishment. However,
determining the correct level of those profits may not be straightforward, partly because
transactions within a multinational company – for example the transfer of royalty and
interest income between establishments in different countries – also need accounting for.
There are further complex rules, such as those around ‘transfer pricing’, that try to ensure
that these internal transactions are based on an ‘arm’s length principle’ – i.e. what the price
of that transaction would be on the open market rather than between related parties.
4.129 Recent years have seen an increasing focus on the strategies that multinational companies
use to minimise their global tax liabilities, facilitated by international tax rules that permit
them to shift profits away from countries with higher corporate tax rates, even when the
underlying economic activity and the ‘value creation’ took place in those countries.
63
Gordon, The rise and fall of American growth: the US standard of living since the civil war, 2016.
64
Brynjolfsson and McAfee, The second machine age: work, progress and technology in a time of brilliant technologies, 2014.
65
‘Source’ refers to where the income is generated and ‘residence’ to the place individuals or businesses are domiciled. Double taxation
occurs if the same income is taxed in both jurisdictions.
66
Jogarajan, Prelude to the international tax treaty network: 1815-1914 early tax treaties and the conditions for action, Oxford Journal of
Legal Studies, 2011.
67
From OECD Commentaries, as reproduced in OECD, Addressing the Tax Challenges of the Digital Economy, 2015.

Revenue risks
4.130 These concerns led the OECD to develop its Action plan on base erosion and profit shifting
(BEPS).68 This initiative seeks to counter tax avoidance by multinational companies through
fostering inter-governmental cooperation. Its ambition was to reform the international tax
system so that the reporting of profits is more closely aligned with the location of the
underlying economic activity and value creation. BEPS was launched in 2013, since when
the UK Government has announced several policy measures in this area, including the
diverted profits tax, changes to the taxation of royalties, rules around hybrid mismatches
and restrictions in the amount of interest that can be relieved against corporation tax.
4.131 The Government also agreed to implement the OECD’s recommendation that multinational
companies file a new annual return, providing HMRC and other tax authorities with country-
by-country reporting. The purpose of these returns is to increase the visibility of
disaggregated, country-level financial results, including profits.
4.132 Digitalisation does not significantly change the nature of these issues, but it does make them
more acute. It makes operating internationally easier – for example, by holding shares in a
foreign company or setting up a subsidiary. It means that even small businesses can reach
consumers across the globe and can do so without a physical presence in the consumer’s
country. Digitalisation also facilitates increasingly complex supply chains and enables them
to be managed from multiple locations.
4.133 Digitalisation is related to, but not limited to, the digital sector. Despite this, there is an
increasing focus on the rapid growth of large technology companies and the additional
challenge this presents to taxing corporate profits. The OECD has concluded that
“digitalisation presents no unique BEPS issues but that some business models can exacerbate
BEPS concerns”.69 Similarly, the Government has stated that the principle of taxing profits in
the country in which the value is generated is “being challenged by business models for
which value creation is in part reliant on the engagement and participation of users”.70
4.134 Digitalised companies have been able to generate value from business models that did not
previously exist. The OECD has identified three factors that are frequently observed:
• ‘Cross-jurisdictional scale without mass’ – digitalisation allows businesses to operate

more easily across national boundaries, allowing them to locate production processes
across different countries and tap into a global consumer base. And they can do so
without a permanent establishment, and in some cases without even a legal presence.
• Importance of intangible assets – digitalised businesses are characterised by high

investment in intangible assets and a heavy reliance on intellectual property, such as
software and algorithms. As discussed above, intra-company transactions involving
such assets make it harder to determine the correct level of profits to be taxed.
• Value created by user participation – a social media network is a good example of a

business model where user participation and user generated content synergises with
data and network effects to create value.
68
OECD, Action plan on base erosion and profit shifting, 2013.
69
OECD, Tax challenges arising from digitalisation – interim report, 2018.
70
HM Treasury, Corporation tax and the digital economy: position paper update, 2018.

Revenue risks
4.135 At Budget 2018 the Government announced a new tax on the revenues of large businesses
in the digitalised sector that derive value from a UK user base. The ‘digital services tax’ will
apply regardless of whether a business has a taxable UK presence. It will affect social media
platforms, search engines and online marketplaces. Rather than defining value or how it is
derived, it will levy a 2 per cent tax on revenues generated by specific business models and
activities that the Government has deemed to meet its definition and to relate to UK users. A
consultation on the detailed design of the tax and its implementation closed in February and
draft legislation was published in July. The tax is due to take effect from April 2020.
4.136 The Government considers the digital services tax to be an interim measure and that the
“ultimate objective is to address the challenges… through reform of the international
corporate tax framework”. It is “optimistic that progress can be made on multilateral reform”
and is committed to working in forums such as the OECD and G20, and with the EU.71
Online retailing
4.137 Chart 4.22 shows that while the value of online retail spending remains small relative to the
total, it is increasing its share year-on-year. In 2018, the ONS estimates that internet sales
made up 18 per cent of total sales by value, compared to just 5 per cent in 2008.
Chart 4.22: Retail spending, store-only versus online-only sales

25 400
Store
Store
Online 350
20
All retailing excluding fuel Online
300
Index 2008=100
Per cent of GDP
15 250
200
10 150
100
5
50
0 0
2008 2010 2012 2014 2016 2008 2010 2012 2014 2016 2018
Source: ONS, OBR
4.138 The growth in online retailing is at the expense of traditional ‘bricks and mortar’ business,
and the fiscal risk this creates is around the erosion of the tax bases for business rates and
VAT. The issue for business rates is straightforward – competition from online retailers, with
lower fixed costs, puts traditional businesses at a competitive disadvantage. Even when an
online retailer is operating at large scale, they can do so via huge fulfilment centres in areas
where business rates are low. The risks for VAT largely centre around compliance, and the
risk that the tax system is struggling to keep pace with the changing nature of the economy.
71
HM Treasury, Digital services tax: consultation, 2018.

Revenue risks
4.139 For domestic traders the compliance issues are those we discussed earlier in this chapter
around the relatively higher ‘tax gaps’ for small businesses. For overseas traders two
examples highlight the types of issues involved:
• In the 1990s and 2000s UK-based online retailers were selling CDs, DVDs and other
goods VAT-free via the Channel Islands. This was perfectly legal and took advantage
of ‘low value consignment relief’ (LVCR), which allows VAT-free transit of imports with
a value not exceeding £15 within EU countries. The rationale is that the administrative
costs of collecting relatively low amounts of VAT make it inefficient to do so. At Budget
2011 the threshold was reduced from £18 to £15 and Autumn Statement 2011 then
withdrew LVCR from the Channel Islands. These changes were expected to generate
additional yield of £115 million a year by 2015-16. LVCR still provides an incentive for
domestic businesses selling such goods to reroute them via an offshore location,
something that digitalisation facilitates. It also creates an incentive to evade VAT by
falsely claiming LVCR. It is reasonable to think significant numbers of parcels that
should not qualify for LVCR could pass through undetected.
• In 2018 the European Commission launched a legal challenge against the UK to

recover what it claims is around €2.7 billion of uncollected customs duties. This
followed an anti-fraud investigation that uncovered significant undervaluation of the
value of imports from China, dating back to 2007. The charge is that the UK failed “to
take appropriate risk control measures… [and]… failed to prevent the fraud”.72
4.140 Neither example is related purely to digitalisation, but the growth of online retailing and the
ability to transact seamlessly with businesses located overseas is likely to increase the risk.
4.141 VAT compliance will also be affected by the outcome of the Brexit negotiations, where
agreements have yet to be reached. Post-Brexit, a UK consumer making a purchase in an
EU country might be entitled to a VAT refund from that country. And the Government has
said it will not charge VAT on such a purchase where it is for ‘personal use’.73 It may or may
not be liable to customs duties, depending on what trade arrangements are agreed.
4.142 Exiting the EU also raises questions around VAT administration. The current system has been
built on a system of EU-wide cooperation and information sharing. It is not clear what the
level of post-Brexit cooperation and information sharing will be. For example, the VAT ‘mini
one stop shop’ is an EU-wide system used by traders to report and pay VAT from sales of
digital services to consumers in the EU.74 Also, while LVCR will no longer apply, the
rationale around the cost effectiveness of compliance activity in such cases will remain.
4.143 The Government has introduced measures to mitigate some of these risks. The Autumn
Budget 2017 measure ‘online VAT fraud: extend powers to combat’ makes online
72
European Commission press release, EU Budget: Commission takes further action to ensure the United Kingdom makes customs duties
fallen due, available to the EU budget, 24 September 2018.
73
HM Treasury, Customs bill White Paper, 2017.
74
The EU has recently announced measures to extend the ‘mini one stop shop’, turning it into a ‘one stop shop’, including bringing goods
as well as services within its scope.

Revenue risks
marketplaces ‘jointly and severally liable’ for unpaid VAT from sales on their platforms from
both UK and overseas sellers. We noted the very high uncertainty around this costing – a
similar measure in Budget 2016 had significantly less impact than first expected. ‘Making
tax digital’ is expected to improve compliance among smaller retailers. The digital services
tax is not targeted at compliance but it will generate some yield from online marketplaces.
Online peer-to-peer platforms

4.144 Digitalisation has facilitated new business models, such as online peer-to-peer platforms,
that virtually connect individuals looking to trade goods and services. A mobile internet
application, for example, can match buyers and sellers, provide a rapid and near costless
payment mechanism, and reduce information asymmetries by providing reputational and
feedback mechanisms. Activities range from full-time contract working, the supply of
irregular or casual labour and small or ‘home’ business transactions. Some of these
activities or transactions might be on a short-term, rental or payment-by-task basis.
4.145 Online platforms have contributed to the growth in the so-called ‘gig’ or ‘sharing’ economy
– the terms often being used interchangeably. This has been a contributing factor to the
trends towards self-employment and working via a company structure that we discussed
earlier in this chapter. Digitalisation also facilitates wider labour market changes by making
remote-working easier, for example through improvements in communication technologies.
4.146 Online platforms do not change the nature of the associated tax risks, but do promote ways
of working that pose a bigger fiscal risk. Both self-employment and incorporation generate
less tax per pound of earnings than working as an employee (as we showed in Chart 4.6).
And as was also discussed earlier, the ‘tax gap’ for self-assessed income tax and small
companies’ corporation tax is proportionally much greater than that for PAYE income tax.
4.147 There is no official estimate of the number of people providing work via an online platform,
but recent survey-based research commissioned by the TUC suggests that almost 10 per
cent of the adult population have worked via an online platform at least once a week.75
Platform working is centred around courier services, task-specific work, transport services
and food delivery, and is disproportionately carried out by those aged under 35.76
4.148 Peer-to-peer platforms also allow individuals to generate income from a range of other
activities, including accommodation services, professional services, financial services and
online sales. These earnings too will be self-assessed for income tax or corporation tax
purposes and therefore subject to the same compliance risks.
4.149 The Government’s main policy response has been measures around disguised employment
(‘off-payroll reform’, commonly referred to as ‘IR35’77), which are expected to reduce the
capacity to incorporate for tax-motivated reasons, (though the effects are highly uncertain,
as we noted when certifying the costing). The rules around joint and several liability for
75
University of Hertfordshire, Platform work in the UK 2016-2019, 2019.
76
Department for Business, Energy & Industrial Strategy, The characteristics of those in the gig economy, 2018.
77
IR35 is actually the original legislation around off-payroll working from 2000.

Revenue risks
online marketplaces, while aimed at VAT, might also have an impact if they provide HMRC
with better information on activities carried out via online marketplaces. Obtaining the
digital records held by platforms is a potential compliance benefit that we discuss below.
‘Making tax digital’, again aimed at VAT, but also available for income tax on a voluntary
basis, is a further source of potential upside risk.
4.150 Any Government policy response to the ‘Taylor Review’ of modern working practices might
also have an effect, though it is worth recalling that tax definitions and employment rights
definitions are not the same thing. Any measures attempting to align them might achieve
the desired simplification, but could also create additional risks to tax revenue if they were to
create new boundaries in the tax system that could be exploited or legal definitions to test.
Tax administration
4.151 The issues discussed in respect of tax policy also represent some of the challenges that
digitalisation poses for tax administration and compliance – establishing international taxing
rights; observing taxable activity online; and the trend toward small owner-manager
businesses and the self-employed, both of which are typically less compliant.
4.152 But digitalisation also offers opportunities to enhance tax administration and compliance by
providing HMRC with better and more timely information and data. It can help close the
information gaps that facilitate tax avoidance and evasion. It could also reduce the number
of taxpayers that HMRC estimates fail to ‘take reasonable care’ in their tax returns, or those
that commit ‘errors’ when doing so. Together these two categories account for almost 30
per cent of HMRC’s estimate of the overall tax gap.78 Better information and data can also
enable HMRC to target its compliance interventions more effectively.
4.153 In Managing fiscal risks, the Government states it “will ensure the tax system keeps pace
with the rise of digital technologies and harnesses innovation to improve the administration
of the tax system”. This had already been set out in HMRC’s 2014 ‘Digital strategy’,
including ambitions to “promote digital take-up and voluntary compliance”, “use data to
help customers avoid errors through pre-population”, “using better data to make…
intervention decisions” and “intelligence-led responses, supported by better data and risk
analysis”.79 In the remainder of this section we consider a few of the ways that digitalisation
can help fill information gaps and the positive fiscal risks that could flow from doing so.
4.154 Few of the risks we discuss can readily be quantified, but to put them in context HMRC’s
latest estimate of the tax gap is £35 billion, representing 5.6 per cent of all taxes due. So,
reducing the tax gap by one percentage point would be worth around £6 billion a year.
Interacting digitally with taxpayers

4.155 HMRC states that well over 90 per cent of self-assessment returns are now completed
online. Its ‘personal tax account’ has been accessed by 15 million taxpayers since its launch
78
HM Revenue and Customs, Measuring tax gaps, 2019.
79
HM Revenue and Customs, HMRC digital strategy: 2014, November 2014.

Revenue risks
in 2015, and its ‘business tax account’ – also launched in 2015 – by over 3 million
businesses.80 While neither is mandatory, and the functions available are relatively basic, it
is not inconceivable that they might be expanded or that the lessons learned from their
introductions might be used to improve HMRC’s future digital offering. This might include
broadening the coverage, improving guidance and engagement, or pre-populating
additional fields in tax returns. The use of artificial intelligence might allow taxpayers to
interact more productively with HMRC, and in doing so improve the understanding of their
tax affairs. It may also help HMRC learn more about taxpayers and allow for more effective
use of behavioural interventions – ‘nudges’ – to improve tax compliance.
4.156 The April 2019 launch of ‘making tax digital’, HMRC’s initiative to use software to interact
with taxpayers, initially for VAT but available on a voluntary basis for income tax, is a more
ambitious and fundamental change. It will require affected businesses to keep digital
records and, for most, to use compatible software to send completed tax returns to HMRC.
In our central forecast, we assume the use of software will reduce record-keeping errors and
will therefore generate additional revenues rising to £300 million a year by 2023-24.
4.157 The process of reaping the tax-administration benefits of digitalisation may not, of course,
be entirely without hiccups. New digital systems need designing and testing, and this might
be challenging and time-consuming. We have previously reported on several HMRC digital
initiatives that, after initial enthusiasm, have been delayed or scrapped.81 Even when a
programme begins as scheduled, such as the collection of ‘real-time information’ (RTI) from
the PAYE income tax system in April 2014, the payoffs can take a long time to be realised.
For example, HMRC did not begin to publish RTI statistics until 2018, and still deems them
to be ‘experimental’ and in a ‘development phase’.82 To date, RTI helps inform some of the
judgements in our income tax forecast but is not yet a fundamental building block.
Use of third-party data

4.158 HMRC already collects vast amounts of information from third parties such as employers
and financial institutions. This is used to improve the accuracy of tax liability calculations, to
guide compliance activity and to pre-populate digital tax returns. Pre-population removes
some of the burden from taxpayers by requiring them simply to verify information rather
than sourcing and validating it themselves. This should improve compliance. Digitalisation
has generated a raft of additional data that might be used to improve tax administration
and compliance. From our earlier discussion, examples include sales data held by online
marketplaces and the income generated by users of platforms in the sharing economy.
HMRC already has significant powers for ‘bulk-data gathering’, but these do not yet apply
to non-UK platforms. The Government has said that it “will continue to explore opportunities
for technological solutions for users to share their own data directly with HMRC”.83
80
HMRC, Annual report and accounts 2017-18.
81
Examples include the ‘digital disclosure service’, plans to put inheritance tax online for customers and agents, and allowing all charities
to register jointly with HMRC and the Charity Commission, as reported in our November 2017 Economic and fiscal outlook.
82
HMRC, Earnings and employment statistics from Pay As You Earn Real Time Information: experimental statistics, 2019.
83
HMRC, The role of online platforms in ensuring tax compliance by their users: summary of responses, 2018.

Revenue risks
Cashless transactions
4.159 In 2008, cash payments made up 60 per cent of all transactions. By 2018, they were just 28
per cent.84 In its place, the number of payments by debit and credit cards has grown rapidly,
driven jointly by the popularity of contactless payment and online shopping. Online banking
and the use of faster payments has also grown considerably in recent years. Cash
transactions can hinder tax compliance by contributing to less accurate record-keeping and
the under- and non-reporting of income.85 Digital payments should leave a more prominent
audit trail that enables better record-keeping and reduces the incentive for evasion.
4.160 HMRC already has the power to access data on debit and credit card transactions, from
electronic payment providers and from ‘business intermediaries’, so the growth of digital
payment mechanisms, and the audit trail they leave, could improve compliance. However,
there are caveats: first, there will continue to be some regular cash users, disproportionately
represented by those on lower incomes and the elderly; and second, those determined to
avoid or evade tax will find other ways to do so, even when using electronic payment – for
example by suppressing the recording of electronic transactions.86
Cryptoassets
4.161 It has been suggested that cryptoassets, alternatively known as cryptocurrencies or virtual
currencies, could be a boon to tax administration.87 They combine elements of a new
payment system with a private currency, one that is not currently supported by the Bank of
England. While the use of cryptocurrencies remains largely outside the mainstream, it is the
underlying technology that is of potential benefit to tax authorities. Distributed ledger
technology (often referred to as ‘blockchain’) is essentially a database that records and
verifies all transactions made using the cryptoasset. In theory then, this might solve the
information constraint faced by taxpayers and tax authorities, providing a complete
historical record of all transactions and transacting parties. In practice, the Government is
still developing its thinking in this area, so we are probably quite some way from any tax
benefits being realised.88 For now, there is as much interest in how to tax the gains from
assets like bitcoin as there is in harnessing the underlying technology.89 But the risks are not
one-sided. With the technology arguably developing faster than the Government’s
response, it may be that in the short- and medium-term the downside risk – an increase in
the number of untaxed activities – is greater than the upside.
International information exchange

4.162 Digitalisation and the additional information and data that it generates can also be shared
across tax jurisdictions to improve compliance. Country-by-country reporting is one example
(see paragraph 4.131). The common reporting standard (CRS) is another that was
84
UK Finance, UK payment markets report, 2019.
85
HM Treasury, Cash and digital payments in the new economy: call for evidence, 2018.
86
Office of Tax Simplification, Technology review: a vision for tax simplicity, 2019.
87
See for example, Krishna et al, Instilling digital trust, blockchain and cognitive computing for government, IMF: Digital revolutions in
public finance, 2017.
88
HM Treasury, Financial Conduct Authority, Bank of England, Cryptoassets taskforce: final report, 2018.
89
HMRC, Cryptoassets for individuals, policy paper, 2018.

Revenue risks
established within the OECD, and then announced in the UK at Budget 2015. CRS requires
financial institutions to disclose their clients’ offshore holdings to tax authorities. Information
is then exchanged, annually and automatically across countries.90
4.163 Over 100 countries have signed up to CRS, and HMRC has made two exchanges to date –
receiving information on 1.5 million accounts in September 2017 and a further 5 million in
September 2018. HMRC is already targeting compliance activity based on the earlier
information exchange, but the huge quantity of data means that considerable time is
needed for processing and analysing before responding. This means there may be a
considerable lag between an exchange and its final payoff.
4.164 This is also true of unauthorised information exchanges, such as the millions of files leaked
as the so-called ‘Panama papers’ and ‘paradise papers’. As a result of work linked to the
Panama Papers, HMRC has opened investigations for suspected tax offences and expect to
bring in more than £190million.91
Data analytics
4.165 Digitalisation combined with stronger powers of collection has allowed HMRC to collect vast
amounts of data – from its own administration of the tax system, and from third parties and
international information exchanges. These data are often unconnected and unstructured,
which combined with their large volume makes the use of traditional data processing and
software unfeasible. Instead, making sense of ‘big data’ requires what has become known
as ‘data analytics’. This is a broad term that covers many different types of statistical
techniques and data analyses that are ultimately designed to reveal insights and patterns
within complex data. Data analytics seeks to describe what has happened in the past and
explain why, and therefore to predict what might happen in the future and suggest an
appropriate course of action. By combining and analysing its different data sources, HMRC
will potentially gain a much fuller picture of a taxpayers’ affairs. This can allow HMRC to
carry out targeted compliance activities, such as risk-based interventions and audits.
4.166 HMRC currently employs around 600 professional analysts and data scientists to work on
‘data exploitation’. In the longer term, with the volume of data continuing to increase, many
expect much of this insight to be achieved via the use of artificial intelligence, and in
particular ‘machine learning’. Indeed, HMRC is already using machine learning in its risk
assessment and compliance work.92
Conclusions
4.167 On balance, it seems likely that a declining share of activity will be taxable on current
policies, and that the downside risks to tax bases, some of which have already crystallised,
will be predominant in the shorter term. This is an area where the Government is actively
90
The OECD has also developed a ‘common transmission system’ to “provide a single, secure connection between tax administrations
through which they can exchange tax information” and can also allow “the secure exchange of other relevant tax information as
necessary” (OECD, Tax and digitalisation, 2018).
91
HMRC, No Safe Havens 2019: responding appropriately, May 2019.
92
Interview with HMRC’s chief digital and information officer, as reported on PublicTechnology.net.

Revenue risks
trying to mitigate the risks. Arguably it is moving faster than other countries, but less quickly
than the underlying technologies. It acknowledges that multilateral policy solutions are more
effective ways to address the risks, and it remains to be seen whether unilateral initiatives
like the digital services tax will help to facilitate those multilateral solutions.
4.168 The risks to effective tax rates from improved tax compliance seem to be on the upside in
the longer term, and this is another area where the Government has ambitious plans. But
the potential upside is constrained while operating within the existing tax system, suggesting
downside risks in the short-to-medium term, as the Government plays catch-up.
4.169 Ultimately, the pace of change means that there is much uncertainty around the fiscal risks
posed by the digitalisation of the economy – a process that probably still has some way to
go. For example, a tax directed at the value created by users might soon be out of date if it
is not also able to tax the value created by machines. And if some of the more dramatic
predictions about the scope for artificial intelligence to replace workers in different
occupations were to prove right, then the resulting consequences could be significant.93

4.170 In this chapter we have discussed several issues that the Government is likely to wish to
consider when managing its fiscal risks, while others that we discussed in greater detail in
our 2017 FRR remain pertinent despite changing little since then. These include:
• Narrowing of the income and capital tax bases, thanks in part to policy measures.
• Loss of revenue as people move to more lightly taxed forms of employment status.
• Pressure on excise duty tax bases from behavioural and technological change.
• Periodic policy reversals and persistent failure to implement some default tax rises.
• The number of policy aspirations not yet costed, including from leadership candidates.
• The stark difference in non-compliance rates across different forms of tax collection.
• Uncertainty around the projected cost of oil and gas infrastructure decommissioning.
• The high cost of tax expenditures, and the poor understanding of how it changes.
• Tax policy challenges from digitalisation in terms of what can be taxed and where.
• Potential tax administration gains from digitalisation that could be significant.
4.171 When assessing the outlook for revenue over the medium and long term, does the
Government regard these or other issues as important for its risk management strategy and,
if so, how does it intend to address them?
93
Baldwin, The globotics upheaval: globalisation, robotics and the future of work, 2019.

5 Primary spending risks
Introduction
5.1 In 2018-19, public spending amounted to £812 billion, equivalent to £29,800 per
household or 38 per cent of GDP (on the latest official data). In our March forecast we
estimated that central and local government would spend £468 billion on the day-to-day
(’current’) running costs of public services and administration, and that government
departments, local authorities and public corporations would spend £83 billion on capital
investment (such as roads, rail and buildings). Cash transfers through the welfare system
are expected to have cost £223 billion and net debt interest payments £37 billion.
5.2 Our latest medium-term forecast assumes that the ratio of total spending to GDP – the most
relevant metric for analysing fiscal sustainability – will fall by 0.2 percentage points over the
next five years (from 38.0 to 37.8 per cent). Our long-term fiscal sustainability analysis
factors in demographic pressures on demand for public services and welfare transfers, plus
non-demographic cost pressures in the health and care sectors. On unchanged policy, these
would place spending and debt on an unsustainable upward path over the long term.
5.3 The outlook for spending is always clouded by risks and uncertainties, as one can see by
comparing latest outturn estimates to the successive official five-year forecasts produced first
by the Treasury and then the OBR (Chart 5.1). The differences reflect methodological
changes and other statistical revisions, policy changes, unexpected economic developments
and unexpected changes in how spending is affected by a given state of the economy. The
charts show that the forecasts more often under-predicted spending than over-predicted it.
Chart 5.1: Successive forecasts for total public spending

1000 48
Treasury forecasts
900 46
OBR forecasts 44
800
Outturn 42
Per cent of GDP
700
£ billion
40
600
38
500
36
400 34
300 32
2000-01 2005-06 2010-11 2015-16 2020-21 2000-01 2005-06 2010-11 2015-16 2020-21
Note: Per cent of GDP forecasts have been restated to remove the effects of subsequent revisions to the ratio in the starting year of the
forecast – these can be large and typically reflect methodological changes that forecasters would not have been able to anticipate.
Source: ONS, OBR

5.4 Looking over a shorter two-year horizon, outturns have differed from our forecasts since
2010 in both directions. Initially, we over-predicted departmental and local authority
spending. We did not foresee the extent to which departments would underspend the limits
they had been set by the Treasury and we underestimated local authorities’ desire to
continue adding to their reserves. More recently, we have over-predicted debt interest
spending (where the interest rates at which the government can borrow have continued to
surprise on the downside) and personal tax credits. But the apparent sizeable under-
prediction of departmental spending in our December 2014 and March 2015 forecasts
reflects subsequent policy decisions by the Government to increase planned spending
significantly. Forecast differences have been smaller in more recent forecasts, but larger
policy-related differences will emerge again soon: the boost to health spending taking effect
this year will cause large differences relative to our two-year ahead forecasts for 2019-20.
Chart 5.2: Two-year ahead forecast differences from successive OBR forecasts
35
Public services and capital spending
Debt interest
25
Welfare
15
Other annually managed spending
Total spending
£ billion
-5
-15
-25
Jun 10 Nov 10 Mar 11 Nov 11 Mar 12 Dec 12 Mar 13 Dec 13 Mar 14 Dec 14 Mar 15 Jul 15 Nov 15 Mar 16 Nov 16 Mar 17
Note: Outturn data have been adjusted for major classification changes, to ensure they are consistent and comparable over time.
Two-year ahead errors for the forecasts from November 2016 onwards are calculated using a mixture of provisional outturn and
our most recent forecast, depending on the availability of data. For comparability, 'in-year' is assumed to be 2009-10 and 2014-15
for the June 2010 and July 2015 forecast respectively.
Source: ONS, OBR
5.5 In this chapter we update our assessment of the broad range of risks to public spending that
were identified in our 2017 Fiscal risks report (FRR), noting how the Government responded
in its 2018 Managing fiscal risks (MFR) publication to the issues we raised. We discuss the
drivers of public spending and how governments seek to control their effects, before
identifying medium- and long-term risks associated with:
• health spending: including the new NHS spending settlement and long-term plan;
• adult social care spending: where a green paper has been pending for some time;
• welfare spending: where the state pension age has been reviewed, while reforms to
means-tested working-age benefits and to disability benefits continue;

• local authorities and public corporations: particularly local authority borrowing;
• devolved administrations: where new fiscal frameworks have begun operating;
• major provisions and contingent liabilities: including those relating to future nuclear
decommissioning costs, clinical negligence claims and tax litigation; and
• Brexit-related spending items: including farm support and the ‘divorce bill’.
5.6 We focus in this chapter on ‘primary’ spending – i.e. spending excluding debt interest, the
risks to which are discussed in Chapter 7. The chapter concludes with a list of issues that the
Government may wish to address in its next edition of Managing fiscal risks.
Drivers of public spending

5.7 When thinking about risks to public spending, it is helpful to think about its underlying
drivers. In most cases, these can be grouped into:
• policy choices: such as which public services to provide, when to upgrade public
infrastructure or what financial support to offer through the welfare system;
• demand-side drivers: the number of people to whom a given service will be provided
or that will be eligible for a particular benefit; and
• unit-cost drivers: in particular the effect of inflation on the cost of providing each unit
of a public service or the average amount awarded to each benefit recipient.
5.8 These drivers vary in importance for different elements of public spending. For example,
state pensions spending is projected to rise as a share of GDP over the long term due to the
ageing population. As Chart 5.3 shows, ageing is also a source of upward pressure on
health and adult social care spending relative to GDP. But, in our latest long-term
projections, other unit-cost drivers are expected to be even more important. Policy choices
can increase or reduce the effects of the various drivers of public spending.

Chart 5.3: Average public spending per person by age

45
Total spending Education
40
Health Adult social care
Receipts/spending in 2023-24 (£ thousand)
35
Other spending Welfare
30
25
20
15
10
0
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
Source: OBR Age
Control of public spending

5.9 The Treasury uses two administrative ‘control totals’ to manage public spending:
• departmental expenditure limits (DELs)1 cover spending on public services, grants,

administration and capital investment, which can be planned over many years; and
• annually managed expenditure (AME) covers categories of spending less amenable to

multi-year planning, such as social security spending and debt interest.
Departments’ DELs are further split into those covering current or ‘resource’ spending
(RDEL) and those covering investment or ‘capital’ spending (CDEL).
5.10 DEL spending is subject to greater control than AME. In particular the Treasury usually
requires departments to offset spending pressures in one area of their budget by bearing
down elsewhere. It therefore tends to be less volatile than AME spending, with the most
significant source of changes being policy choices rather than factors beyond the immediate
control of government. In 2014, the Coalition Government sought to increase control over a
subset of AME spending by introducing a ‘welfare cap’.
1
Our presentation of expenditure only shows those components of DEL and AME that are included in the fiscal aggregates of public
sector current expenditure (PSCE) and public sector gross investment (PSGI), i.e. the elements that affect public sector net borrowing. For
budgeting purposes, the Treasury also includes other components in DEL and AME such as non-cash items (like depreciation and
provisions) and financial transactions (like student loans). The non-cash items do not affect public sector net borrowing or net debt.
Financial transactions affect debt but not borrowing and are discussed in Chapter 6 on risks to the public sector balance sheet.

5.11 In our 2017 FRR, we noted that the Treasury has exhibited a high degree of control over
DEL spending once plans have been finalised, so the main medium-term risk to spending
was not that limits are overspent but that policy decisions are taken to raise them. We also
noted the declining proportion of total spending (total managed expenditure or TME) subject
to relatively firm DEL controls – thanks to larger cuts to DEL than to AME during the post-
crisis consolidation, and to the cost of state pensions being lifted by the triple lock and
ageing. And we highlighted the jump in DEL spending as a share of GDP as a result of the
late-2000s financial crisis and recession, when pre-crisis DEL plans were largely followed
but nominal GDP fell far short of pre-crisis expectations.
5.12 In that context, we highlighted two issues for the Government’s response:
• The declining proportion of total spending subject to relatively firm DEL controls.
• The possibility of cost overruns for major projects like HS2 and universal credit IT.
We also noted the growing reliance on loans and other financial transactions rather than
conventional expenditure to fund new policies.
5.13 In MFR, the Government noted that it has introduced controls in AME spending, such as the
welfare cap, which covers around 15 per cent of TME and adherence to which we monitor
in each Economic and fiscal outlook (EFO). In our March 2019 EFO, we judged that the cap
would be met in 2022-23, the year in which the current one applies. Since it was introduced
in 2014, the Government has changed the level of the cap several times and its design
once. It was lowered substantially in July 2015, alongside a package of welfare spending
cuts, but then raised substantially in November 2016, at which point the previous cap was
set to be exceeded by more than 7 per cent in the year in which it applied. As such, it is not
clear that the welfare cap has any meaningful impact on spending plans and outcomes.
5.14 The Government also pointed to the fact that Network Rail would be brought into DEL in
2019-20 (and shadowed DEL procedures during 2018-19). But Scottish Government
spending has moved the other way, more than offsetting the shift of spending into DEL.

The balance of DEL and AME in total spending
5.15 We now forecast the proportion of TME subject to DEL controls to increase over the period
to 2023-24, largely as a result of policy changes announced over the past two years. The
main contributor is the NHS DEL settlement announced in June 2018, and the subsequent
increases in both NHS and non-NHS DEL totals announced in October 2018 and March
2019. The net effect of switching items between DEL and AME has been to raise AME, with
the Scottish Government move into AME outweighing Network Rail moving into DEL. The
Budget 2018 announcement of higher employer contribution rates for all public service
pension schemes reduced the net cost of those schemes, which are controlled through AME.
But it increased DEL totals to meet the higher gross contributions incurred by departments.

This shifted around 12 per cent of the gross annual cost of public service pensions from
AME to DEL from 2019-20 onwards. Despite now rising over the forecast period, we still
forecast DEL as a share of TME to be a little over 45 per cent in 2023-24, down 9
percentage points relative to 2007-08.
5.16 This suggests that the risk posed by spending being controlled more lightly under AME than
DEL remains elevated relative to its pre-crisis level, but that it is no longer on a rising path.
However, since that is largely the result of health spending risks having crystallised over the
past two years, and given that there are continuing upward risks to health spending, it is not
clear that there has been a material reduction in overall spending control risks. The main
risk remains one of policy decisions raising DELs rather than those DELs being overspent.
Spending Review 2019

5.17 The Chancellor announced in his 2019 Spring Statement that the Treasury would launch a
full three-year Spending Review before the summer. But this was a conditional commitment
“assuming a Brexit deal is agreed over the next few weeks and the uncertainty that is
hanging over our economy is lifted.” On 2 July, the Chief Secretary to the Treasury told
Parliament that “it will be for the new government to decide whether the circumstances make
it appropriate to conduct a full three-year spending review, or a single-year exercise.” So one
medium-term spending control risk is that the Spending Review does not cover three years,
leaving departments outside the NHS uncertain over their share of the DEL envelope
pencilled in at the Spring Statement. Given the above-inflation RDEL settlement for the NHS,
the Spring Statement RDEL totals would have seen non-NHS RDEL spending rise with
inflation (as measured by the GDP deflator), but the Treasury has told us that “these figures
do not represent the final RDEL (…) spending envelopes for [the 2019] Spending Review”.
5.18 While Spending Review controls have generally been tightly adhered to by international
standards, there have been recent examples of large changes being announced outside
Spending Reviews – most obviously the NHS settlement announced in June 2018. The
announcement was not accompanied by a detailed reform plan explaining how the extra
money would be used or by the announcement of any measures to help pay for it.
5.19 In addition, both candidates currently running for the Conservative Party leadership have
made spending pledges that they would pursue as Prime Minister. Boris Johnson has talked
of 20,000 extra police officers, raising per-pupil funding in education and higher public
sector pay. The largest single pledge is Jeremy Hunt’s wish to add 0.5 per cent of GDP to
the existing NATO commitment on defence spending (which would cost around £11 billion
in 2019-20, rising to £12½ billion in 2023-24). Meeting either candidate’s pledges on top
of the NHS settlement would either put enormous pressure on unprotected areas of
spending if the envelope for the Spending Review were held at the Spring Statement totals,
or would require higher spending and borrowing.
5.20 This might suggest that the role of Spending Reviews (and indeed Budgets) as the main
vehicle for fiscally significant spending decisions – where costs can be considered in the
round and trade-offs between priorities made – may be changing. This could have
implications for the effectiveness of Treasury spending control and fiscal risk.

Cost overruns in major procurements

5.21 Over the past two years, there have been several examples of major projects costs rising
beyond original estimates or the latest ones available at the time. These include:
• Crossrail: the funding envelope for this project has increased from £14.8 billion at
Spending Review 2010 to £17.6 billion in the latest estimate (December 2018). This is
a 19 per cent increase relative to the original settlement. Although that is not
proportionately as large as for some other projects, there is still scope for it to escalate
further. Crossrail’s chief executive told Transport for London that it was “not currently
possible to give a date for the opening of the central section” of the line.2
• Ministry of Defence (MoD) equipment plan 2018 to 2028: over this period, the MoD is
expected to allocate over 40 per cent of its total budget to equipment and support
programmes. This includes both equipment already in use, such as Typhoon combat
aircraft, and new equipment, such as four new nuclear-armed submarines. The
programme is budgeted for £187 billion over ten years, but the MoD’s forecast is for it
to cost £7 billion more than that. The National Audit Office (NAO) published an
assessment in November 2018, commenting that the plan “remains unaffordable” and
that even the department’s worst-case scenario – that it is £15 billion over budget –
relies on analysis that is “optimistic and costs could increase further”.3
• Home Office Emergency Services Network (ESN): the Home Office originally intended
to launch ESN, a new system to be used by 107 police, fire and ambulance services in
England, Scotland and Wales, in September 2017, ready to replace Airwave – the
current system – fully in December 2019. In 2017, the Home Office concluded that
that timetable would not be achievable, and it reset the programme in 2018. The
programme delivered has now been delayed by at least three years, and is expected to
cost £3.1 billion more than initially anticipated, a cost overrun of 49 per cent. The
NAO’s most recent report concluded that the latest cost forecast is “highly uncertain”
and that “based on past performance, it seems unlikely that ESN can be delivered by
the target date of 2022”, raising the possibility of further delays and cost escalation.4
The Infrastructure and Projects Authority has also assigned ESN a ‘red’ rating in its
delivery confidence assessment, meaning that in its view “Successful delivery of the
project appears to be unachievable”.5
5.22 In our previous FRR, we highlighted the High Speed 2 (HS2) rail project as a potential
source of upside spending risk from cost overruns. The official HS2 cost estimate of £56
billion (in 2015 prices) has not changed, but the House of Commons Library has derived a
£65 billion (in 2015 prices) estimate based on the NAO review of the project in 2016 and
the July 2017 Department for Transport financial case.6 The project has been subject to
further delays, with Ministerial approval for major construction works on the first phase of
2
Transport for London Board minutes, 21 November 2018.
3
National Audit Office, Ministry of Defence: The equipment plan 2018 to 2028, 5 November 2018.
4
National Audit Office, Progress delivering the Emergency Services Network, 10 May 2019.
5
Infrastructure and Projects Authority, Annual report on major projects 2017-18, 4 July 2018.
6
House of Commons Library, High Speed 2: the business case, costs and spending, Briefing paper number CBP 8601, 26 June 2019.

the project having been pushed back six months to December 2019. The NAO has reported
that land and property acquisitions have so far cost more than the estimated budget, but
within the contingency in the spending envelope.7
Historical performance of Treasury spending control

5.23 Last year, the Institute for Fiscal Studies (IFS) and the National Institute of Economic and
Social Research (NIESR) each published a study of public spending control in the UK.8 The
IFS found that controls generally work relatively well, in the sense that outturns do not
diverge in large amounts from original plans systematically. But they note that the Treasury’s
spending controls formally being adhered to – in the narrow sense that departments mainly
underspend relative to their ultimately binding allocations – masks the fact that allocations
themselves are changed to match political pressures. The IFS cite in particular top-ups to
settlements mid-way through Spending Review periods, and reallocation of capital budgets
to cover resource pressures, examples of which were present across the whole period they
considered (1993 to 2015). Similarly NIESR found that announced multi-year spending
plans were changed significantly in subsequent fiscal events, and that the most significant
factor in explaining such revisions were changes to the expected path of GDP growth.
5.24 The findings from both studies reinforce a risk that we highlighted in our previous report.
Although DEL controls are formally effective, initial settlements in a Spending Review tend to
be revised in future years, especially when political priorities change, and therefore they are
less effective controls on medium-term expenditure than they might initially appear.
Conclusion
5.25 Given developments over the past two years, our assessment of fiscal risk from spending
control (as distinct from spending policy, under the incoming administration) remains
unchanged. Although our forecast now includes a small rise in the proportion of spending
subject to DEL controls by 2023-24, it will remain well below pre-crisis levels. And more of it
is to be spent on health, where (as described in the next section) history points to DEL plans
being topped up periodically. The shift of Scottish Government spending from DEL to AME
seems less of a risk than the operation of fiscal devolution more generally (discussed from
paragraph 5.108 onwards). And the perennial risk of major projects costing more, and
taking longer to complete, than initially planned continues.
5.26 It is difficult to say at this stage whether recent announcements of major spending policies
outside Spending Reviews and Budgets represent a material shift in Treasury’s ability to
control total spending or whether these are merely further examples of political priorities
bypassing the system on occasion, which the historical studies suggest is not unprecedented.
If this does become the norm, then spending risk will have increased.
7
National Audit Office, Investigation into land and property acquisition for Phase One (London – West Midlands) of the High Speed 2
programme, HC 1531, 13 September 2018.
8
IFS, The planning and control of UK public expenditure, 1993–2015, July 2018 and NIESR, Understanding and Confronting Uncertainty:
Revisions to UK Government Expenditure Plans, October 2018.

Health spending
5.27 In our 2017 FRR, we highlighted medium- and long-term pressures on health (and adult
social care) spending from several sources:
• Demographic factors, especially an ageing population. ONS population projections

point to an increase in the proportion of the population at older ages, where per
capita consumption of health and social care is higher than average.
• Income-related drivers, since demand for health and social care rises as income rises.
• Non-demographic cost pressures, such as: the labour intensity of health and social
care provision; technological advances increasing demand and therefore spending
even if they reduce unit costs; and the rise in chronic conditions, increasing demand.
5.28 In FRR 2017, we drew on analysis published by the Health Foundation and the IFS, the
NHS’s assessment of the efficiency savings necessary to live within its settlement to 2020-21,
and our own analysis of contributions to changes in NHS spending relative to GDP, to
conclude that there was a high probability that a moderately large upside spending risk
would crystallise over the medium term. Based on our Fiscal sustainability report (FSR)
projections, we concluded there were much larger long-term risks to fiscal sustainability
from demographic and other cost pressures on health spending being accommodated.
5.29 This led us to raise two specific issues for the Government’s response:
• Significant long-term upward cost and demand pressures on health spending.
• The precedent created by repeated topping-up of initial health spending settlements.
5.30 The medium-term risk we had flagged crystallised before the Government published its
response to these issues. In June 2018, the Prime Minister announced a significantly higher
NHS RDEL spending settlement. When the Government published MFR in July 2018, it
described the settlement as recognising the need for increased resources, while also
referring to five ‘financial tests’ that the NHS would need to meet in its 10-year plan in order
for the increased funding to be confirmed. In the event, the funding was confirmed in the
Budget in October, with The NHS long term plan eventually published in January 2019.9

The new NHS funding settlement and remaining medium-term risks
5.31 The new NHS RDEL settlement amounts to a real-terms increase of 3.4 per cent a year on
average until 2023-24. It covers NHS England’s resource budget, with ‘Barnett
9
NHS England, The NHS Long Term Plan, January 2019.

consequentials’ added to the devolved administrations’ budgets. It does not cover other
current spending by the Department of Health and Social Care in England, for example on
public health, or any capital spending. Still, it is a substantial increase relative to the 1.1 per
cent a year real-terms increase between 2010-11 and 2017-18. It crystallises one of the
main medium-term fiscal risks we identified in our 2017 FRR.
5.32 Despite its increased generosity, it remains uncertain whether the new settlement will stick.
Although the Treasury told us in March 2019 that the updated settlement (topped up to
cover changes in the inflation forecast relative to October 2018) was now final, experience
with previous settlements suggests that March 2019 is unlikely to be the last top-up we see.
5.33 One way to gauge the prospects for further top-ups in the future is to look at the same kind
of external analysis that informed our risk assessment two years ago. The Health Foundation
and the IFS compared the settlement to various scenarios (Chart 5.4). They concluded that it
may only be sufficient to maintain current standards.10 They estimate that real-terms funding
for NHS England would have to rise by 3.3 per cent a year to maintain current standards,
fractionally less than the announced settlement. To improve standards, as the long-term
plan aims to do, real-terms funding would have to rise by 4 to 5 per cent a year, at an
additional cost of £6 billion to £15 billion a year (in 2018-19 prices) by 2023-24.
Chart 5.4: New health settlement versus different Health Foundation/IFS scenarios
220
Historical average (3.7 per cent real terms increase a year)
Invest to modernise (Health Foundation/IFS)
210
Modest improvements (Health Foundation/IFS)
Maintain standards
200
Funding announcement
190
£ billion
180
170
160
150
2018-19 2019-20 2020-21 2021-22 2022-23 2023-24
Source: Health Foundation, IFS, OBR
Productivity improvements
5.34 One of the main assumptions underpinning the NHS long-term plan is that it will be
possible to deliver an increase in productivity of at least 1.1 per cent a year. This appears
challenging. Since 2010-11 productivity growth in the NHS (whether adjusted for quality
10
Health Foundation and IFS, Securing the future: funding health and social care to the 2030s, May 2018.

improvements or not) has outperformed the wider economy, but it has been much lower in
previous periods when funding growth has been higher (Chart 5.5). Indeed, it was only
during the period of financial pressures from 2010-11 to 2017-18 that productivity growth
averaged more than 1 per cent a year. This reflected growth in quality-adjusted output
slowing less than input growth, suggesting that in periods of higher funding growth there is
less pressure on providers to improve productivity to meet rising demand for services.
5.35 The NAO has also highlighted a significant risk that the need to deliver more services will
lead to the use of more expensive agency staff, increasing unit costs and weighing on
productivity.11 Reliance on productivity improvements to help meet increasing demands on
the health service might therefore represent a spending risk over the medium term.
Chart 5.5: Real-terms funding and productivity growth in the health service
10
Average annual real growth in funding
9
Quality-adjusted average annual productivity growth
8
Non-quality adjusted average annual productivity growth
7
Economy-wide average annual productivity growth
6
Per cent
0
2000-01 to 2004-05 2004-05 to 2010-11 2010-11 to 2017-18
Source: NHS Digital, OBR
Cost drivers
5.36 Around half of NHS expenditure relates to staff costs. Among the largest non-pay costs are
clinical supplies, prescribed drugs and services bought from non-NHS suppliers. Historically
slow rates of growth in overall spending since 2010-11 – 2.8 per cent a year on average in
nominal terms – have mostly been achieved through pay restraint. Total staff costs grew by
2.0 per cent a year on average over the period, compared with 3.5 per cent growth in non-
pay costs. But non-pay costs would have risen faster were it not for statins coming off patent
in 2012, allowing the NHS to prescribe them in their generic form. This reduced the annual
cost of statins to the NHS from £310 million in 2011 to £55 million in 2016. With statin
prescriptions dispensed rising from just over 10 million in 2011 to 32 million in 2016, they
would have cost the NHS £1 billion in 2016 if unit costs had remained flat.12
11
National Audit Office, NHS financial sustainability, 18 January 2019.
12
The King’s Fund, The rising cost of medicines to the NHS: what’s the story?, 26 April 2018. See also NHS Digital, Prescription Cost
Analysis – England, 2016, 30 March 2017.

5.37 There is a limit to how long pay growth in health can be held below that in the wider
economy without damaging recruitment and retention. Indeed, the Government has recently
relaxed the NHS pay cap through the ‘Agenda for Change’ multi-year pay deal, noting the
need to recruit, retain, motivate and boost the productivity of its workforce.13 Employment
growth in the public health system has also picked up in recent years, being above 1.5 per
cent a year since 2014-15, after employment fell in 2011-12 and 2012-13. The need to
maintain growth in employment – highlighted in the long-term plan – is likely to put further
upward pressure on pay due to the need to recruit and retain staff. The expected tightening
of the immigration regime after Brexit would add to these pressures.14
5.38 With non-pay costs having increased more rapidly than pay costs in recent years, and the
price of drugs typically rising faster than whole economy inflation, bearing down on them is
seen as an important potential driver of productivity gains and improved services. The NHS
long-term plan sets out several priority areas where it believes it can reduce costs, such as
improving the coordination of procurement and the accuracy and turnaround times on tests
and scans. Were the NHS to succeed in reducing unit costs in these areas, it is not clear
whether it would lead to lower spending or higher volumes of services. Were it not to
succeed, it seems plausible that it would generate pressure for spending to be topped up.
Financial sustainability
5.39 The NHS long-term plan also prioritises returning all components of the NHS in England –
including all NHS trusts – to financial balance by 2023-24. This is likely to be challenging,
despite the additional funding, especially given the improvements in care that are planned.
For example, the NAO argues that “previous funding boosts appear to have mostly been
spent dealing with current pressures rather than making the changes that are needed to put
the NHS on a sustainable footing.” This suggests there will be a trade-off between returning
NHS trusts to financial balance and delivering desired improvements in care. In January
2019, the NAO expected deficits (excluding sustainability and transformation payments) to
reach £3 billion by the end of 2018-19, and concluded that “current support and incentives
will not alone be sufficient to return these trusts to financial balance”.15
Our latest long-term spending projections

5.40 We published our latest long-term fiscal projections in July 2018. By far the largest source
of change to our health spending projection was the higher medium-term starting point
generated by the Prime Minister’s NHS settlement announcement. Unfortunately we were
not able to update our quantification of non-demographic cost pressures as the NHS had
not (and still has not) repeated its analysis of the sources of spending growth that underpin
it.16 Our latest projection shows health spending reaching 13.8 per cent of GDP in 2067-
68, on the assumption that demographic and other cost pressures are accommodated, up
1.0 per cent of GDP from the January 2017 projections that underpinned our previous FRR.
13
Parliamentary Under-Secretary of State for Health, NHS Pay Review Body Report 2018/19 and Agenda for Change Multi-Year pay deal,
House of Lords Written Statement – HLWS777, 27 June 2018.
14
See, for example, Health Foundation, Closing the gap: Key areas for action on the health and care workforce, March 2019.
15
National Audit Office, NHS financial sustainability, HC1867, 18 January 2019.
16
NHS England, NHS Five Year Forward View: Recap briefing for the Health Select Committee on technical modelling and scenarios, May
2016.

Conclusions
5.41 The crystallisation of the main medium-term policy risk to health spending we identified two
years ago means we no longer see a high risk of current plans being increased materially
over the medium term. But, despite what the Treasury has told us about this representing an
absolutely final settlement, external analysis and past experience suggest that further, more
modest, increases in funding may well be politically unavoidable in the medium term.
5.42 Long-term risks to fiscal sustainability from demographic and other cost pressures are little
changed relative to the assessment we made two years ago. But in part this reflects the
absence of new information on non-demographic cost pressures in health spending. It
would help our own analysis of this crucial issue – and no doubt other researchers – if the
NHS were to revisit its analysis of 2015-16 spending growth on an annual basis.
Adult social care spending

5.43 In common with health care, the main drivers of spending on adult social care are
demographic factors (both ageing and health status at given ages), income-related drivers
(as demand rises with incomes) and non-demographic factors (notably for social care its
labour-intensive nature and the rise of chronic conditions in the population). The high share
of staff costs in service provision, and the low average wage paid, means that social care is
particularly exposed to changes in minimum wage policy. And the adult social care system
itself has been the subject of continuing uncertainty thanks to successive reviews, including
the 2011 Dilnot Commission – the recommendations of which were enacted in the 2014
Care Act, but were not implemented and have subsequently been shelved.
5.44 On this basis our 2017 FRR raised two main issues associated with adult social care:
• The potential impact of the NLW and migration reform on health and social care costs.
• Potential pressure to bail out a private social care provider if in financial difficulty.
• Noted the introduction of the adult social care precept in council tax raised by local
authorities with social care responsibilities, and the improved Better Care Fund, as its
main responses to pressures posed by the National Living Wage and other factors.
• Highlighted the statutory role of the Care Quality Commission (CQC) in monitoring
and assessing the financial sustainability of care providers in England, providing
greater support for local authorities in ensuring continuity of care provision.
5.46 With the Dilnot reforms in abeyance, the Government stated that it “will be publishing a
green paper on care and support for older people in Autumn 2018.” It suggested that in any

future reform to the system “there should continue to be a principle of shared responsibility,
and that people should continue to expect to contribute to their care costs when preparing
for later life.” The green paper has yet to be published and no date for it has yet been set.

Cost pressures
5.47 There have been no concrete developments over the past two years in respect of the two
main sources of pressure on the costs of delivering adult social care we identified in our
previous report, but the Government is studying or consulting on possible changes to both:
• The National Living Wage: existing policy will see the effective minimum wage for
employees aged 25 and over – the NLW – rise to 60 per cent of median hourly
earnings by 2020. In Budget 2018, the Government expressed an ambition to
increase that to two-thirds of the median thereafter. At Spring Statement 2019, the
Chancellor asked Arindrajit Dube to undertake a review of the international evidence
on the impacts of minimum wages, to inform how it sets the formal remit of the Low
Pay Commission, which the Government has said it will do by the end of the year.
Further rises would raise the cost of social care because a large proportion of care
workers are affected by the NLW – at a rate of two-thirds of median earnings, over a
third could be directly affected.17 The extent to which the cost would represent a fiscal
risk would depend on whether it were funded through council tax or borrowing.
• Reforms to the immigration system: the Government has announced its intent to
impose a more restrictive immigration system after Brexit, and is currently consulting
on the salary threshold for obtaining a visa. Wherever the threshold is set, it seems
likely to be higher than the median wage for a full-time social care worker (£18,300 in
2018).18 This would make it harder to recruit care workers, leading either to more
vacancies remaining unfilled or to increased labour costs (or both). The Institute for
Public Policy Research estimates that these would leave 400,000 vacancies unfilled in
England by 2028, which would be equivalent to around 20 per cent of estimated
demand for workers in the sector at that point.19
5.48 In addition, pressures on local authority budgets have fed through to adult social care. The
reduction in general-purpose grants from central government has hit at a time when
demand for care has been increasing rapidly. Adult social care spending per adult in
England fell by 3.1 per cent in real terms between 2011-12 and 2017-18, even when
accounting for new funding sources.20 But it is then projected to rise by 2.8 per cent in real
terms between 2017-18 and 2019-20 as council tax and other funds for social care rise.
17
Skills for Care, Pay in the adult social care sector, March 2019, and OBR analysis of ONS, Earnings and Hours, care workers: ASHE
table 26, 25 October 2018.
18
ONS, Earnings and Hours, care workers: ASHE table 26, 25 October 2018.
19
Institute for Public Policy Research, Fair care: A workforce strategy for social care, 25 October 2018. See also Social Care Institute of
Excellence, Forecasting the adult social care workforce by 2035: Workforce intelligence report, 2016.
20
The Government has introduced central funding for adult social care through the Better Care Fund (BCF). The Government’s
expenditure statistics, Public expenditure statistical analyses, classify the BCF as health expenditure, as it forms part of the Department of

5.49 Our 2018 FSR projection shows adult social care spending rising substantially over the long
term, from 1.2 per cent of GDP in 2018-19 to 1.9 per cent of GDP by 2067-68. In our
modelling, only around 10 per cent of this increase is driven by purely demographic factors.
Other factors, such as demand for adult social care increasing with income and the rise in
disability rates due to an ageing population, are more important drivers of demand. As with
health care, our baseline projection assumes these pressures are accommodated.
5.50 In addition to these demand and cost drivers, the Government remains committed to
limiting the cost of care for recipients. Although the Dilnot reforms have now been shelved,
it seems almost certain that any reform finally adopted will seek to reduce the contributions
from users,21 and will therefore increase public spending on social care as a share of GDP.
The fiscal implications of such a reform would depend on whether they were accompanied
by measures to fund them and by how much.
5.51 We discussed several sensitivities around our baseline projection for adult social care
spending in Annex B of our 2018 FSR. These are illustrated in Chart 5.6. Spending would
be higher in a world of higher unit costs, an older population or one where age-specific
rates of disability were higher. Spending would also be higher in a world where local
authorities provided care for more than just those who meet the minimum eligibility
requirement in the Care Act 2014 or one where the Dilnot reforms were implemented. And
it is of course possible that several differences from our baseline assumptions could
materialise at once. If they were all to push in the same direction, rather than offsetting each
other, spending could rise more significantly than shown in any one of these scenarios.
Chart 5.6: Adult social care spending in 2067-68: alternative scenarios
Expanded eligibility scenario
Higher unit costs
Old-age structure
Dilnot reforms included
Higher disability
Baseline
Lower disability
Young-age structure
Lower unit costs
Spending in 2018-19
0 0.5 1 1.5 2 2.5 3

Source: OBR Per cent of GDP
Health and Social Care’s budget, but it is delivered through local authorities for adult social care purposes. We therefore include the BCF
as part of expenditure on adult social care, consistent with its ultimate purpose.
21
House of Commons Library, Social care: forthcoming Green Paper (England), Briefing paper number 8002, 13 May 2019.

Pressure to bail out struggling private social care providers

5.52 In April 2019, the holding companies controlling Four Seasons Health Care Group, the
second largest private social care provider at the time, went into administration.22 The CQC,
in its function of market oversight, stated that it did not expect service provision to be
affected, according to the Minister of State for Care’s written statement to the House of
Commons.23 The operational parts of the group are now in the process of being sold, which
the Government expects to happen by the end of 2019.
5.53 Local authorities in England are required by the Care Act to meet people’s needs for social
care temporarily should a provider fail. While this is not expected to affect provision in this
particular case, local authorities’ legal position as a backstop makes this an ongoing risk.
5.54 Despite the apparently successful handling of Four Seasons Health Care Group’s financial
failure, our assessment of the fiscal risk associated with pressure to meet the obligations of
private sector providers of a public service is little changed. The sector’s financial position
remains a concern among providers – 80 per cent of directors of providers believed that
providers were facing financial difficulties in 2017-18.24 So while the system for managing
failures of providers appears to be working, the fiscal risk associated with local authorities
needing to take over service provision remains due to their position in legislation.
Social care funding reforms

5.55 Since we published our 2017 FRR, the Government has announced that it would no longer
pursue the reforms to social care funding proposed by the Dilnot commission,25 having
previously delayed implementation by four years due to concerns about their cost.26 These
reforms were intended to cap the costs of care for individuals at £72,000 over a lifetime.
Instead, and as mentioned in MFR, a social care green paper was scheduled to be
published in late 2018 (having itself already been delayed more than once).
5.56 With the green paper still not published, uncertainty over any future reform to the funding of
adult social care continues to affect users and providers. The current system, which
potentially sees recipients of care bearing costs limited only by their assets, seems unlikely to
be politically sustainable over the long term. Any moves to limit the costs faced by
individuals would present a fiscal risk relative to our baseline spending projection. And even
on current policies, adult social care spending represents a source of pressure on long-term
fiscal sustainability under all the scenarios we looked at in our 2018 FSR.
5.57 NHS England’s long-term plan looks at the integration of health and social care provision.
Cash transfers from the NHS to local authorities (through the Better Care Fund) are the
22
The two companies, Elli Finance (UK) plc and Elli Investment Limited, announced on 30 April 2019 that they were entering
administration. A statement with the same date was posted on the Four Seasons Health Care website. See also, House of Commons
Library, Four Seasons Health Care Group – financial difficulties and safeguards for clients, Briefing paper number 8004, 21 May 2019.
23
Minister of State for Care, Health and social care provider update, Written ministerial statement HCWS1532, 1 May 2019.
24
Care Quality Commission, The state of health and adult social care in England: 2016/17, 23 November 2017.
25
Parliamentary Under-Secretary of State for Health, Social care, Oral statement to the House of Commons, Hansard Volume 632,
Column 1235, 7 December 2017.
26
Parliamentary Under-Secretary of State for Health, Cap on care costs, Written statement to the House of Lords HLWS135, 17 July 2015.

main channel for that at present. Fuller integration could reduce pressures on the NHS,
freeing resources in hospitals by providing more adequate care in the community. But if
funding for social care remains much tighter than that for health, pressures on social care
could continue to spill over onto the NHS, for example where patient transfers out of
hospital care are delayed while they wait for a social care package to be put in place.
Conclusions
5.58 The fiscal risks associated with adult social care noted in our 2017 FRR remain significant:
• The intention to raise the NLW further once the policy target for 2020 has been met,
and the tighter migration regime that is expected to be put in place after Brexit, could
place even greater upward pressure on the unit costs of providing social care.
• The risk that private providers may need to be bailed out remains, although the system
for ensuring that a resolution occurs without local authorities having to take over
financial obligations seems to be working in the latest case of a large failure.
5.59 More generally, while the Government has decided not to pursue the Dilnot reforms to the
funding of social care, the pressure to limit individuals’ exposure to the costs of social care,
and hence the risk to public spending, remains. Uncertainty over how that will be done –
including the multiple delays to the publication of the latest green paper – may have its own
costs, posing a challenge for users and providers as they try to plan for the future.
Welfare spending
5.60 As defined in our forecasts, welfare spending covers benefits and tax credits that transfer
cash to eligible individuals or households. It is the single largest item of AME spending, and
accounted for 27.5 per cent of total government spending in 2018-19.
5.61 Some welfare spending – particularly on working-age people – rises and falls with the
economic cycle, as the number of claimants and the cash amounts awarded tend to rise
during a recession and fall during the subsequent recovery. But most is little affected by the
economy in cash terms, although it still rises and falls as a share of GDP due to fluctuations
in GDP.27 Spending overall has fluctuated between 9.3 and 12.1 per cent of GDP over the
past 20 years and remains at just above 10 per cent of GDP in our latest forecast.
5.62 Our 2017 FRR discussed medium- and long-term risks to spending on state pensions and
working-age benefits, noting important demographic and health-related drivers of
caseloads and the role of uprating policy – notably the state pension ‘triple lock’ – in driving
average payments. Delivery of major reforms to existing benefits and the introduction of
27
See Box 4.1 in our 2014 Welfare trends report for analysis of the sensitivity of different welfare spending lines to the economic cycle.

universal credit, plus legal challenges to different parts of the system, were identified as
medium-term risks. We highlighted three of these issues for the Government’s response:
• The renewed commitment to the ‘triple lock’, which ratchets pension spending higher.
• Risks surrounding the implementation of the new state pension and universal credit.
• Limited formal reporting of the cost of potential legal challenges to the welfare system.
• Stated that the triple lock had contributed to faster growth in pensioner incomes and to
reduced pensioner poverty, noted our analysis of its ratchet effect on spending and
promised to retain it for this Parliament. It did not set out longer-term objectives, nor
did it discuss, for example, how it deems what costs and risks associated with the triple
lock are appropriate to meet its goals for pensioner incomes and poverty rates.
• Recognised the risks associated with the rollout and forecasting of universal credit,
establishing regular official-level monitoring and assurance procedures (in which OBR
staff play a role). It felt that the new state pension did not pose risks, but noted
continuing monitoring and evaluation of its operation by DWP officials.
• Described the procedures DWP follows in monitoring legal challenges to the welfare
system and reporting on them in its departmental accounts. DWP’s accounts now
separately identify provisions for benefit payments related to legal challenges, bringing
its transparency on this to a level comparable with HMRC. But it records a negligibly
small contingent liability in this area, which seems less transparent than HMRC.

5.64 There have been many relevant developments since we published the 2017 FRR. These
include several policy announcements and further delays to delivery plans, the results of
formal reviews of future state pension age policy, the crystallisation of some legal risks, and
the conclusions from our two most recent Welfare trends reports (WTR), which looked in
depth at universal credit (in WTR 2018) and disability benefits (in WTR 2019).
Overall medium-term risks

5.65 Cyclical and other changes in the economy pose several risks to welfare spending, in
particular on working-age people. With the uprating freeze on most working-age benefits
now ending, cash spending from 2020-21 onwards will be more sensitive to inflation.
Caseloads are also sensitive to trends in the labour market, housing market and health
status, though the latter matters more over longer horizons. The contribution of these and
other factors to past fluctuations in welfare spending were covered in our 2014 WTR.
5.66 Aside from economy-related risks, medium-term risks to welfare spending largely concern
the Government’s reform of the welfare system, reflecting policy risks, legal challenges,
implementation risks and risks to take-up among eligible populations.

Policy risks
5.67 We do not make assumptions about how Government policy might change in the future, but
we can consider the risks posed by:
• Stated policy intentions: these typically lack the detail and certainty necessary to
include their effects in our forecasts, but they do signpost future policy and we can
therefore consider their potential impact on the public finances.
• Existing trends in policy: to the extent that recent history might provide a reasonable
guide to the future, the continuation of such trends can be identified as potential risks
to our forecasts. The fuel duty freezes described in Chapter 4 are a prime example.
5.68 The Government’s March 2019 announcement that it intends to test combining personal
independence payment (PIP) assessments and employment and support allowance (ESA) /
universal credit (UC) work capability assessments poses one such risk.28 DWP envisages that
this would only apply to the small number of claimants who make an application to both
benefits within a 3-month window, and require face-to-face assessments for both. If limited
in this way, the spending consequences would be small, as individuals would still need to
apply for each benefit separately. But bringing some assessments together could result in
higher take-up of both benefits, and particularly PIP, as more people become aware that
they may be entitled to both rather than just one (particularly if the assessments then
become more automatic). If this change were adopted, it could increase welfare spending –
for example, every 10,000 extra PIP claimants would cost around £50 million a year. By
way of illustration, in November 2018 there were around 450,000 people receiving the
support group rate of ESA, or the equivalent rate in UC, who do not receive PIP or disability
living allowance (DLA), showing the potential for significant increases in take-up.29
5.69 Take-up risks also arise with the BBC’s June 2019 announcement that it intends to end
universal provision of free TV licences to the over-75s and instead restrict entitlement to
households where someone receives pension credit. It is accompanying this change with an
awareness campaign, using its broadcast channels and letters to affected households to
promote take-up of pension credit. This is likely to raise welfare spending (see Box 5.1).
28
Secretary of State for Work and Pensions, Health and Disability Announcement, Written statement to the House of Commons, WS1376,
5 March 2019.
29
We focus on the ESA support group here as these claimants are most likely to qualify for PIP as well. In November 2018 a further
335,000 claimants of ESA outside the support group, or the UC limited capability for work component, did not receive PIP or DLA.

Box 5.1: How much might it cost to means-test free TV licences for the over 75s?
Free TV licences for households including someone aged 75 or over were introduced in 2000,
with the BBC being compensated for the foregone revenue by the Department for Work and
Pensions (in effect transferring some of the BBC’s funding from licence fee receipts to general
taxation).a In 2006, the licence fee as a whole was reclassified as a tax, making the free licences
in effect a tax relief. All BBC income and spending is treated as part of the public sector finances.
In the July 2015 Budget the Government announced that compensation from DWP would be
withdrawn progressively, so that from 2020-21 the full cost of free TV licences would be borne
by the BBC. As part of that agreement, the Government gave the BBC responsibility for the policy
beyond the term of the then current Parliament, legislated for in the Digital Economy Act 2017.
We have assumed to date that the BBC would maintain the current system of free TV licences, so
that the reduction in compensation would reduce BBC spending and the budget deficit.
However, the BBC launched a consultation on the future of the concession in November 2018,
and in June 2019 announced its decision not to maintain the current system but to focus
eligibility on households containing someone aged 75 or over who receives pension credit. A
report prepared for the BBC by Frontier Economics estimated that maintaining the current
regime would cost the BBC £745 million in 2021-22, but that means-testing it in this way would
reduce the cost to £209 million, after accounting for additional administration and compliance
costs but assuming no increase in pension credit claims.b Announcing its decision, the BBC Board
estimated the full cost at £250 million, factoring in “implementation costs including compliance
with the new policy and possible increased take-up of pension credit”.c
The scale of any likely increase in pension credit claims is highly uncertain. If the £40 million
increase in costs estimated by the BBC Board were accounted for entirely by higher take-up, this
would imply around an extra 250,000 claimants, costing around £850 million depending on
their characteristics – more than the original move was expected to save the Government. And
since the BBC will spend the £500 million or so it saves by means-testing, the overall cost to the
public finances will be even greater relative to the assumptions in our latest forecast.
DWP estimates there were around 470,000 people aged 75 or over who were entitled to the
guarantee element of pension credit in 2016-17 but who did not receive it, almost 40 per cent of
the total number entitled. These had an average entitlement of £65 a week, resulting in around
£1.6 billion of unclaimed benefit among this age group.d So around half of that group would
need to start claiming to wipe out the expected savings from transferring responsibility to the BBC
and the BBC cutting its domestic spending by a corresponding amount. But if the BBC spends
what it saves via means-testing free licences, that fraction would fall to only a sixth.
Experience from 2003 to 2008, when DWP undertook extensive activity to encourage pension
credit claims, suggests that very large increases in take-up are unlikely, but more than a sixth is
quite possible. The critical difference this time is that potential claimants are facing a potential
loss via the licence fee, whereas then they were only forgoing income they had never claimed.
During this period, take-up of guarantee credit initially increased, but plateaued at between 70
and 80 per cent of those eligible. DWP narrowly missed its 2008 target of 2.2 million guarantee
credit claims and fell further short of its 3.2 million target for pension credit.

Part of this disconnect between the then estimates of entitled non-recipients and the ability for
DWP staff to identify them was thought to be shortcomings in the estimation methodology, in
particular mis-recording of benefit receipt by respondents to the family resources survey (FRS)
and the understatement of savings. But methodological improvements since then, including
matching FRS responses to administrative data, have increased rather than reduced estimated
non-take-up, partly as a result of better identification of disability benefit receipt, which further
increases the numbers entitled to pension credit (and the amounts they receive).
The publicity associated with the latest change will also be different, with more scope for the BBC
to use its channels to advertise pension credit than DWP had with a limited communications
budget. The BBC has stated its aim to encourage take-up of pension credit, saying:
“… We want to raise the visibility of Pension Credit, which Age UK cites as one of the
reasons why people don’t claim, and have already written to charities and older people’s
groups to work together to do this. We have started a public information campaign which
includes using our airwaves and writing to all 4.6 million households setting out the new
scheme. We hope that pensioners will consider claiming as they could then be eligible for
around £2,500 and other benefits as well as a free TV licence.”
DWP took a similar approach, but eventually concluded that “it would not represent value for
money to repeatedly press unwilling eligible people to take up their entitlement.”e
Today’s over 75s might, however, have different experiences and awareness of the benefit
system, and attitudes to claiming, than their counterparts 15 years previously. This may increase
the likelihood of take-up. Between 2012-13 and 2016-17 the proportion of eligible people aged
75 or over taking up the guarantee element of pension credit has fallen by almost 10
percentage points. This is likely to reflect a combination of the lack of proactive take-up
promotion by DWP and the gradual reductions in pension credit as a whole as the savings credit
element has been eroded in value. If this fall in take-up were reversed, there would be around
120,000 extra claimants of pension credit, at an annual cost of around £400 million.
The BBC’s announcement appears already to have had an effect. New pension credit claims
rose from 7,600 in the four weeks to 7 June (immediately prior to the announcement) to 9,300
in the four weeks to 4 July. After allowing for the fact that no new claims were made on the late
May bank holiday, that represents an increase of around a quarter.
Potential knock-on effects of the BBC’s decision extend beyond pension credit. In particular the
act of claiming pension credit might prompt additional claims for attendance allowance,
particularly among those who receive advice from third parties. And for those renting their
homes, claiming pension credit could increase claims for housing benefit, though given the
higher level of take-up of housing benefit among pensioners (perhaps due to it often being paid
through a reduction in rent rather than as a benefit), the risks here appear smaller.
In summary, it is relatively unusual for a government to delegate parameters of welfare policy to
a broadcasting company in an attempt to save money, and it is perhaps not surprising that this
may have unintended consequences. The BBC’s decision to means-test free TV licences via a link
to pension credit receipt may well raise welfare spending by more than it reduces BBC spending,

particularly once the BBC spends the money it saves by means testing. The net effect on the
public finances would therefore be to push the budget deficit up not down.
Government policy towards the licence fee in recent years also highlights the fiscal illusions and
policy risks to which hypothecated taxes or charges of this sort are prone. In principle the licence
fee is a charge that people choose to pay for the right to receive broadcast services, but the link
between the amount that licence holders pay and the money that the BBC spends providing its
share of those services has been weakened first by requiring it to pay for part of foreign policy
(the World Service from 2014) and now part of welfare policy. Given the fate of other attempts
to reduce the generosity of parts of the welfare system in recent years, there is also a risk that the
Government will step in to prevent or ameliorate the losses for those due to lose out.
a
In the public finances, this is shown as reduced licence fee receipts, with BBC spending unaffected. As the reimbursement from DWP
to the BBC is a transaction that moves funds between entities within the public sector, this is not included within welfare spending.
b
Frontier Economics, Review of over-75s funding: A report prepared for the BBC, November 2018.
c
BBC, Age-related TV licence policy decision document, June 2019.
d
Focusing only on the guarantee element of pension credit is appropriate in this instance, as the savings credit element is no longer
open to new claimants, and has gradually been eroded in value since 2010.
e
DWP Departmental Report 2008, May 2008.
5.70 A recent policy trend that poses a medium-term risk to spending relates to reforms to UC.
UC was originally expected to cost more than the benefits it replaced, but a succession of
cuts culminating in the 2015 Summer Budget package resulted in it being expected to save
around £3 billion a year, compared to the system it replaced, once fully rolled out.
5.71 Since then this marginal effect of UC has been revised back to a cost, partly thanks to a
succession of giveaways, including the partial reversal of previously announced cuts to work
allowances and a more generous earnings taper. In our latest forecast, UC costs £1.6
billion a year more than the legacy system once fully rolled out. But there are still substantial
elements that are expected to save money. As greater numbers of people come into contact
with the system, calls for policy giveaways have tended to increase, resulting in some of the
recent reversals we have seen. This could well continue, particularly in respect of elements
that are less generous than the legacy system.30 This risk is heightened by the need, under
current legislation, for DWP to seek approval from Parliament before it can extend
managed migration to a caseload greater than the 10,000 cases in its pilot phase.
5.72 Another notable trend in UC policy is the pairing of measures increasing generosity with
those lowering costs. In the past three fiscal events that included UC policy changes,
increases in generosity were accompanied by delays or alterations to the managed
migration phase that deliver temporary savings, in particular by lowering the cost of
transitional protection paid to claimants who would otherwise lose out financially when
being moved from their legacy benefits onto UC. This typically involves permanent costs
being paired with temporary savings, so it generates longer-term costs.
30
In a recent report, the Institute for Fiscal Studies noted that “77% of those who lose at least £1,000 p.a. are affected by universal credit’s
harsher treatment of one of the following groups: those with financial assets; the low-earning self-employed; couples where one member is
above state pension age and the other below; and some claimants of disability benefits.” See IFS, Universal credit and its impact on
household incomes: the long and the short of it, IFS Briefing Note BN248, April 2019.

Legal challenges
5.73 Legal challenges pose a risk to spending, and these are more likely when there is major
reform of the system creating scope to test the interpretation of new legislation. Over the
past two years, legal challenges, or the potential for them, have led to two special ‘legal
entitlements and administrative practices’ (LEAP) exercises, resulting in additional spending
that is included within our March 2019 forecast:
• In ESA, claimants who were reassessed from incapacity benefit between 2011 and
2014 did not have their entitlement to income-related ESA considered if they were not
in receipt of income support at the time of reassessment, and were only assessed for
contributory benefit. They are now being assessed retrospectively for entitlement to the
income-related component. This exercise is now expected to cost £920 million in
arrears of benefit, with a continuing cost of around £100 million a year.
• In PIP, following two Upper Tribunal judgements,31 and a subsequent judicial review,
an exercise is being undertaken to award arrears of benefit to some claimants who
should have scored more highly in the PIP assessment. Our March forecast assumed
this would cost £450 million in arrears of benefit, with a continuing cost of around
£480 million a year for these claimants and future claimants who will now receive
higher scores. As of mid-June 2019, DWP had reviewed and cleared around 440,000
cases, from which 3,500 arrears payments had been made to qualifying claimants,
totalling around £15 million.32 The very low proportion of reviewed cases giving rise to
arrears suggests that our March forecast assumptions will prove too high, although if
particular types of case are reviewed first then these early success rates may not be
representative of the overall success rate.
5.74 These are two examples of a steady stream of legal challenges in the benefits and tax
credits systems. The larger ones tend to focus on disability and incapacity benefits, but all
parts of the welfare system are subject to them. For example, the bereavement benefits
system was recently successfully challenged in respect of its treatment of unmarried
couples.33 In our previous FRR we noted that DWP’s departmental accounts did not report
the expected or potential cost of current or anticipated legal challenges, in contrast to
HMRC’s treatment of tax litigation risks. DWP has now made explicit provisions for previous
underpayments of benefit – its 2018-19 accounts record just over £1.2 billion covering the
above two exercises and a general provision of £390 million for further legal cases where
there is a probable cost.34
5.75 Following a lengthy campaign by the ‘Women Against State Pension Inequality’ (WASPI)
group, a legal challenge has been brought by the ‘Backto60’ group against the 1995,
2007 and 2011 Pensions Acts that respectively equalised the state pension age for men and
31
MH v Secretary of State for Work and Pensions (PIP) [2016] UKUT 0531 (AAC) and RJ, GMcL and CS v Secretary of State for Work and
Pensions (PIP): [2017] UKUT 105 (AAC). We discussed the first case further in our 2019 Welfare trends report.
32
DWP, PIP administrative exercise: progress on cases cleared, at 14 June 2019, July 2019.
33
In the matter of an application by Siobhan Mclaughlin for Judicial Review (Northern Ireland) [2018] UKSC 48.
34
Department for Work and Pensions Annual Report and Accounts 2018-19, June 2019. The provision in respect of the two LEAP
exercises include in our forecasts is less than the arrears shown above, as these include payments made before 31 March 2019.

women at 65, raised it further to 66 (originally from 2024-26), then accelerated the
timetable so that it is now due to reach 66 by October 2020. The judicial review hearing
took place in June 2019, but the outcome is not yet known. Our forecasts assume the state
pension age continues rising according to the legislated timetable, so any decision to alter
that timetable, or to pay compensation, would crystallise a potentially large fiscal risk.
5.76 The cost of any ensuing policy changes would depend on their precise nature. An illustration
of potential scale is provided by DWP’s estimate that had the Acts not been implemented,
cumulative spending over the period 2010-11 to 2025-26 would be £215 billion higher.35 It
put the difference at £21 billion in 2021-22 (the first full year after the state pension age
reaches 66), rising in real terms thereafter. A range of other proposals would have smaller,
although generally still substantial, costs.36 None of these estimates consider the
implications for tax receipts if some proportion of those affected chose no longer to work.
5.77 DWP’s latest accounts mention the state pension age judicial review in respect of complaints
it receives as a result of the continuing campaign, but not any associated contingent liability
should the judicial review ruling find against the Government. Instead it makes only a
general statement that “The Department has other ongoing legal cases that are not being
disclosed as either contingent liabilities or remote contingent liabilities as any disclosure
could prejudice the Department’s position”. It records a contingent liability of just £82
million in respect of ongoing cases, which clearly does does not include this one.
5.78 Our forecasts do not incorporate the cost of potential future legal cases that are currently
unknown explicitly, but there will be an implicit allowance to the extent that previous legal
cases have affected the observed trends in caseloads and spending that inform our
judgements about future trends. The cost of future legal cases could be greater than we
implicitly assume. It will almost certainly be higher than the £82 million in DWP’s accounts.
Implementation risks
5.79 In our previous FRR, and in our 2018 WTR, we highlighted the repeated delays in rolling out
UC and moving all existing claimants onto the new system. Since then there have been
further delays, with completion now assumed in 2023-24 – six years later than originally
planned. There remain significant obstacles ahead. But rollout delays only represent an
adverse fiscal risk when the policy itself is designed to save money. UC policy changes over
the past two years mean UC is now expected to cost a little more than the legacy system, so
further delays will tend to reduce spending a little, unless they disproportionately affect
aspects of the reform that reduce spending relative to the legacy system.
5.80 The past two years have also seen further delays in completing the reassessment of existing
working-age DLA claimants for transfer to PIP. We now estimate that this will be completed
in early 2021, five years behind the original plan. This date is still subject to uncertainty too,
for example if capacity to undertake the required assessments were to prove inadequate. As
detailed in our 2019 WTR, PIP no longer saves the money it was originally intended to and
35
DWP, Analysis relating to State Pension age changes from the 1995 and 2011 Pensions Acts, June 2019.
36
See House of Commons Library, State pension age increases for women born in the 1950s, Briefing paper CBP-7405, 20 June 2019.

on average those who are reassessed are expected to receive a higher amount of PIP than
they did of DLA. Consequently, delays generate small savings – the March 2019 delay
saved around £150 million in 2019-20, the year of the largest effect. So, as with UC now,
our forecast is relatively insensitive to further changes in this rollout.
5.81 As regards the new state pension, while we have not yet needed to make significant
revisions to our medium-term forecasts, this remains a risk since the first few years’ data
have yet to be fully analysed for any implications for our forecast. Given the rapid growth
and ultimately large scale of spending on the new state pension – it rises from £8.0 billion
in 2018-19 to £35.1 billion in 2023-24 in our latest forecast, but will ultimately rise to well
over £100 billion a year – even relatively small changes in forecast assumptions could
generate fiscally material forecast revisions.
Other take-up risks

5.82 In addition to policy-related take-up risks, the UC rollout itself also poses an upside risk to
PIP take-up, because material numbers of longer-term ESA claimants who have had little
contact with the benefits system for a long period are required to make a claim to UC. It is
possible, particularly if they require advice and assistance from third parties, that awareness
of PIP will increase and therefore also the number of claims.
Overall long-term risks

5.83 In the long run, the size, structure and characteristics of the population are the most
important determinants of welfare spending. Policy choices also play a part, with changes in
the state pension age (SPA) being used as a tool by successive governments to help mitigate
the fiscal effects of increasing life expectancy. The rules for uprating benefits can also have
significant effects on spending when they build up over a long time period.
Population ageing and the state pension age

5.84 The ageing population places pressure on spending as the proportion of the population
eligible for pension-age benefits increases (holding eligibility rules constant). These benefits
now account for over 55 per cent of welfare spending, which is forecast to rise further. Two
key developments have affected our long-term projections of pensioner benefits spending
and our assessment of the associated fiscal risks. Both were incorporated in our 2018 FSR:
• The Government’s response to two reviews of the state pension age: the independent
review led by John Cridland and the review conducted by the Government Actuary’s
Department were both published in March 2017.37 The response consisted of a plan to
bring forward the increase in the SPA to age 68 from 2046 to 2039 and to establish a
presumption that an individual should, on average, spend ‘up to 32 per cent’ of their
adult life in receipt of state pension.38 The Government stated that there would be a
further review before legislating to bring forward the rise to 68, to allow consideration
37
John Cridland CBE, Independent review of the state pension age: Smoothing the transition, March 2017 and Government Actuary’s
Department, Periodic review of the rules about State Pension age: Report by the Government Actuary, March 2017.
38
Bringing forward the increase to age 68 does not, however, achieve the ’32 per cent of adult life’ principle, since the timetable is
constrained by a further principle that the SPA should only increase by one year every ten years.

of updated life expectancy projections, and that it would not formalise policy beyond
2039. Our 2018 FSR projections show that bringing the rise to 68 forward from 2046
to 2039 would ultimately reduce spending by 0.2 per cent of GDP a year. Our
projection based on longevity-linked SPA rises (reaching 70 in 2068) is 0.6 per cent of
GDP lower than that on the currently legislated timetable to 68 in 2046 in 2067-68.
• The ONS’s 2016-based population projections: these included large upward revisions
to mortality rates compared to the previous projections, with 4.6 per cent fewer people
aged above the SPA in 2067 than previously assumed. But downward revisions to
assumptions about fertility and net inward migration meant that the projected old-age
dependency ratio in 2067 was revised up slightly from 46.2 to 46.8 per cent. It is this
ratio that matters most for the effect of ageing on fiscal sustainability.
5.85 Taken together, the accelerated SPA timetable more than offset the effect of a higher old-
age dependency ratio to reduce our estimate of state pension spending in 2067-68 by 0.2
per cent of GDP, relative to our previous projection. But population ageing is still a
significant source of long-term spending pressure in the UK – our baseline FSR projection of
spending on state pensions rises from 5.0 per cent of GDP in 2022-23 to 6.9 per cent of
GDP in 2067-68. And these figures remain highly sensitive to assumptions about the
population structure. Moving from our baseline projection to one based on the ONS old-
age variant raises spending in 2067-68 by 0.5 per cent of GDP.
Benefit uprating
5.86 Our long-term projections assume that the Government’s current uprating policy for the
state pension – the ‘triple lock’ – continues. This raises the value of both the new state
pension and the basic state pension that preceded it by the highest of CPI inflation, earnings
growth or 2.5 per cent each year. It therefore causes spending to rise as a share of GDP if
average earnings growth exceeds nominal GDP growth per person or if average earnings
growth is low relative to CPI inflation or 2.5 per cent, as it has been in recent years.
5.87 We factor in the effect of the triple lock as an average premium relative to earnings growth,
to reflect the assumed likelihood and extent of earnings growth falling short of CPI inflation
or 2.5 per cent. In our 2018 FSR, we estimated this to add 0.36 percentage points to
average uprating (a touch higher than in our 2017 FSR). This translates into state pension
spending being around 1 per cent of GDP higher in 2067-68 than if it were uprated with
earnings growth alone. The triple lock has been triggered once since the March 2017 EFO
that was the most recent forecast at the time of our previous FRR. This was as predicted in
that forecast, but its effect (in April 2018) was 0.6 percentage points larger than assumed,
at 0.8 percentage points above earnings growth. A 0.6 percentage point difference raises
spending on items subject to the triple lock by around £550 million a year by 2023-24.
5.88 In our November 2017 EFO we revised down our medium-term trend productivity growth
assumption by 0.7 percentage points a year, which reduced average earnings growth
commensurately. For the 2018 FSR we assumed this weakness would ultimately fade and
that productivity growth would return to 2.0 per cent a year after an extended period. Were

growth in productivity and average earnings to be lower permanently, the cost of the triple
lock would be expected to be higher as there would be a greater likelihood of earnings
growth in any given year lying below CPI inflation or 2.5 per cent. This would ratchet up
state pension spending relative to GDP more quickly. Over the period we use to calculate
the triple lock premium, if earnings growth is uniformly reduced by 0.7 percentage points,
the premium would almost double. That would add around a further 1 per cent of GDP to
state pensions spending in 2067-68. In this respect, the long-term fiscal risk associated with
the triple lock might be even greater than we had previously assumed.
5.89 The IMF studied the triple lock in its 2018 Article IV consultation on the UK, noting among
other things that it is unique among advanced economies. Some governments uprate
pensions by prices, some by earnings and some by a blend of the two, but “The UK
minimum 2.5 percent lock is an outlier compared to other countries’ indexation policies.”39
5.90 The Government’s response to our 2017 FRR noted that the triple lock supports pensioner
incomes, but did not explain why it thought that the fiscal risks this poses were appropriate.
5.91 Other aspects of benefit uprating create favourable risks for the public finances, compared
to our baseline projections. Under our ‘unchanged policy’ assumptions, our FSR projections
assume that children’s and working-age benefits (along with some smaller pensioner
benefits) are uprated in line with average earnings from the end of our medium-term
forecast. Current government policy is for these benefits to rise in line with CPI inflation, and
if this were to be maintained for 15 years beyond the end of the medium-term forecast, then
welfare spending would be 1.0 per cent of GDP lower. But average incomes of benefit
recipients would also fall by 28 per cent relative to average earnings over that period.
Rising prevalence of disability benefit receipt

5.92 Disability benefits accounted for around 11 per cent of all welfare spending in 2018-19.40
Their cost is sensitive to the number of people that experience disabling conditions. In our
2019 WTR we noted that disability benefits spending had risen by 0.7 per cent of GDP since
1991-92 and that the cost was expected to continue rising over the medium term. The key
driver has been caseloads rising as a share of the population, which has coincided with
survey evidence of rising prevalence of disability in the population.41
5.93 The link between disability prevalence and the prevalence of disability benefit claims in the
population is not one-for-one. Eligibility criteria, administration and take-up behaviour also
play important roles in determining the size of the disability benefits caseload. Receipt of
disability benefits has tended to rise by more than survey measures of disability prevalence
(even after taking account of the ageing population and the fact that prevalence rises
significantly with age).42 Despite that, the number of disability benefit recipients (around 5.1
39
See Box 2 in IMF, ‘Long-term fiscal challenges in the UK’, United Kingdom Selected Issues, November 2018.
40
By disability benefits, we mean those benefits that aim to provide support to the extra costs incurred by those with disabling conditions.
In terms of current benefits, this includes disability living allowance, personal independence payment and attendance allowance.
41
See Chapter 2 of our 2019 WTR for a more detailed discussion of trends in disability prevalence over time.
42
The prevalence of receipt rises for each year above the age of 65, with around 12 per cent of the 65-year-old population in receipt of a
disability benefit in 2018. This figure rises to around 17 per cent of 75-year-olds and over 50 per cent of those aged 90+. See Chapter 3
of our 2019 WTR for a more detailed discussion of trends in prevalence of receipt by age.

million in 2016-17) remains well below the number of people reporting disabling conditions
in DWP’s annual survey of family circumstances (around 14.4 million in 2016-17). This
suggests that significant scope remains for further increases in disability benefit receipt even
were there to be no further changes in the prevalence of disabling conditions.
5.94 Age-specific prevalence of benefit receipt has evolved over time. It has tended to increase
for children and working-age adults, while for pension-age adults it increased continuously
in the four decades to 2010 but has fallen since then. The recent trend for pensioners is
consistent with an increase in disability-free life expectancy, although there may be other
shorter-term drivers. However, even if the likelihood of receiving a benefit is falling at some
older ages, increasing life expectancy is likely to result in individuals receiving a disability
benefit for longer in future. The extent to which additional years of life are spent with or
without a disability, and how this translates into disability benefit receipt, is very uncertain
and could change substantially. The adverse fiscal risks associated with the disability
benefits system would be greater were receipt at any given age to rise faster than assumed
or the ageing of the population to be more rapid than projected.
5.95 This discussion assumes the continuation of current Government policy. In reality, of course,
it is likely to change, and history points to the scope of disability benefits widening over time,
for example as society recognises a broader range of (particularly mental health) conditions.
Conclusions
5.96 Our assessment of the fiscal risks posed by welfare spending is little changed from 2017. By
far the largest remains the long-term pressure associated with an ageing population,
though this is being mitigated through increases in the SPA. It is also relatively slow-moving,
affording time for policy to respond to these pressures. Our latest analysis of health and
disability trends, and their implications for disability benefits spending, suggests that is a
greater source of risk than we recognised in 2017. The triple lock remains a significant
source of risk that the Government has chosen to take on – and the chance and cost of it
crystallising is greater when productivity and real earnings growth are lower. Longer-term
policy towards state pension uprating is unclear. Fiscal risks associated with the rollout of
universal credit and reforms to the disability benefits system have diminished, because
neither is now expected to save money, which means the fiscal impact of delays is minimal.
But fiscal risks from legal challenges still appear significant, though reporting on those that
are viewed as probable is now more transparent.
Local authorities and public corporations

5.97 In 2018-19, local authorities spent around £179 billion (22 per cent of total public
spending). Around 60 per cent was financed by funding from central government, with the
remainder financed by local sources of income – notably council tax and business rates.

5.98 Our 2017 FRR discussed various signs of pressures on local authorities’ budgets. These
included the large cuts in central government grant funding (only partly offset by further
devolution of tax-raising and spending powers), pressures on the use of reserves,
overspending against specific budget lines (in particular children’s and adult social services),
and their pursuit of alternative income sources (most notably through commercial
investments). The key medium-term fiscal risks related to their ability to deliver the real per
capita spending cuts implied by our then forecasts and the Government’s spending plans.
5.99 In that context, we highlighted two of these issues for the Government’s response:
• Initial signs that local authorities have started running down their reserves.
• Examples of local authorities undertaking potentially risky commercial investments.
5.100 We also noted:
• Delivery of core services to standard. Should delivery begin to fall short of legally or
politically acceptable levels, pressure for more funding could build.
• At the extreme, one or more local authorities could in effect become insolvent, but that
this was unlikely given systems in place to prevent it and no examples in recent history.
• Reported that it had updated the terms of the Debt Management Account Deposit
Facility to give local authorities greater choice over how to invest their cash, while also
helping it monitor local authorities’ financial resources.
• Pointed to the updated prudential framework for borrowing that is underpinned by

guidance published by MHCLG and the Chartered Institute of Public Finance and
Accountancy, and new recommendations of quantitative indicators for local authorities
to prepare and monitor when making investment decisions. The Government stated
that it would “monitor how local authorities respond to the revised guidance, and take
appropriate further action if this is necessary”.
• In respect of pressures on core services, noted that it had set four-year funding
allocations and had raised them in the Local Government Finance Settlement 2018-19
(albeit by a modest £1 billion over two years). As regards the imposition of Section
114 spending controls at Northamptonshire County Council in January 2018, it
pointed to the independent inspector’s report that found these reflected failings in the
Council’s internal controls rather than funding cuts.43
• As regards the risk of insolvency, it pointed to the existing legal framework governing
balanced in-year budgets and restrictions on borrowing and that no local authority
had defaulted on its borrowing since the Prudential Code was introduced in 2004.
43
Max Caller CBE, Northamptonshire County Council Best Value Inspection, 15 March 2018.

5.102 We also noted risks around public corporations relating to the possibility of excessive
borrowing, shocks to financing costs and persistent loss-making.

Local authorities
5.103 There have been several developments over the past two years that have affected our
assessment of the fiscal risks associated with local authorities:
• The apparent trend towards drawing down from current reserves has reversed. The
proceeds of measures that boost funding – for example, piloting business rates
retention – have seemingly been saved rather than spent. In aggregate, authorities
added to reserves in 2017-18 and we expect them to have done so again in 2018-19.
The situation can be different for individual authorities – the Chartered Institute of
Public Finance and Accountancy estimates that 10 to 15 per cent of local authorities
exhibit at least some signs of risk to their financial stability.44 The scale of the potential
risk posed by reserve drawdowns remains large though – local authorities held £21.9
billion as of the end of March 2018.45 But the likelihood of it crystallising has declined.
• By contrast, we have been surprised on the upside by local authorities’ borrowing for
capital spending. ’Prudential’ borrowing increased from £6.7 billion in 2015-16 (the
latest data at the time of our previous FRR) to £12.2 billion in 2017-18 (fully £6.8
billion higher than we forecast in March 2017). Much of this has been borrowed from
the Public Works Loans Board (PWLB), with Spelthorne Borough Council being the
largest user of PWLB funding to pursue commercial investments (see Box 5.2). The
rapid expansion of prudential borrowing by local authorities suggests that this is a
greater source of risk than we thought two years ago. It also raises questions about the
effectiveness of the revised guidance in the Prudential Framework. We therefore asked
the Treasury whether, as it suggested in MFR, its monitoring of the situation suggested
further action would be appropriate. It told us:
“The Government is aware of a few authorities who are continuing to undertake

significant amounts of borrowing for commercial purposes. Government shares the
concerns of CIPFA and others about the risks that these local authorities are exposing
themselves and local taxpayers to. MHCLG is now engaged in a post implementation
review of the changes to the prudential code, to understand how the updated guidance
and codes have been received by the sector and to evaluate the effectiveness of the
revised framework. Findings from the review will be published in due course.”
• Policy decisions added modest amounts to local authority sources of income – for
example, increasing the cap on annual council tax increases (without being subject to
a local referendum) and various extensions to business rates pilots. But there is still no
clarity about the design of full business rates retention or when it will be implemented.
44
The Chartered Institute of Public Finance and Accountancy, Measured resilience in English authorities, December 2018.
45
Ministry of Housing, Communites and Local Government, Local authority revenue expenditure and financing: 2017-18 final outturn,
England – Revised, 20 March 2019.

• The Government has made a series of interventions in Northamptonshire County

Council. First it appointed two Commissioners to oversee the Council, who
recommended scrapping it and replacing it alongside seven others with two unitary
councils. The Council’s officers then issued a second Section 114 notice banning all
new spending except on statutory services for protecting vulnerable people. The
Government appointed its own children services commissioner, while granting the
council permission to use capital receipts to fund current spending. It also sanctioned a
5 per cent council tax rise in 2019-20 without a local referendum. In July 2019 the
Council reported that it had under-spent its budget by £4.5 million.46 The many and
exceptional steps the Government has been willing to take reinforces our assessment
that the risk of a local authority becoming insolvent is low because central government
would step in first. It also suggests that the scale of fiscal risk here is small.
5.104 In our previous report we noted that local authorities’ decisions about where to invest their
reserves or other surplus funds had generated risks in the past, including the fall-out from
the interest rate swaps litigation in the early 1990s (notably in relation to Hammersmith and
Fulham London Borough Council) and, more recently, when it appeared possible that local
authorities would lose money deposited at high interest rates in Icelandic banks that failed
during the global financial crisis. In a similar vein, it has recently emerged that Kent County
Council had invested £263 million (as of the start of June) in the Woodford Equity Income
Fund, which has barred all investors from withdrawing their funds.
5.105 Our overall assessment of the fiscal risks posed by local authorities is little changed. They
could crystallise via reserves drawdown; top-ups to central government funding; or if central
government chooses to step in if one or more local authorities were unable to service its
borrowing. These risks appear balanced and their magnitude is likely to be relatively small.
We continue to see a low risk of authorities defaulting on their borrowing and that central
government would probably step in before an authority reached the point of insolvency. But
the scale of local authority borrowing is greater than we were expecting two years ago.
5.106 Ultimately, the key driver of such risks crystallising could be the delivery of core services to
standard. Should delivery begin to fall short of legally or politically acceptable levels,
pressure for more funding could build. If that were deficit financed, rather than tax financed,
it would represent a fiscal risk. This assessment is unchanged from our previous report.
46
Northamptonshire County Council, Outturn Report for the financial year ending 31st March 2019, July 2019.

Box 5.2: The rise in ‘prudential’ borrowing – the Spelthorne example

Recent years have seen a marked rise in ‘prudential’ borrowing, driven inceasingly by local
authorities taking advantage of low interest rates offered by the Public Works Loan Board
(PWLB). Local authorities’ PWLB debt has risen from £64 billion in March 2015 to £66 billion in
March 2017 and now stands at £77 billion in March 2019.a There are currently seven authorities
with outstanding balances in excess of £1 billion: Transport for London, the Greater London
Authority, Birmingham City Council, Leeds City Council, Woking Borough Council, South
Lanarkshire Council and Spelthorne Borough Council. Spelthorne Borough Council, home to just
100,000 people, has acquired its £1 billion of PWLB debt in the space of just three years.
Spelthorne’s PWLB borrowing has mainly financed the acquisition of commercial property. The
Council argued that it was using fixed low rates of interest to “help offset the impact of
disappearing general revenue grant support from the Government”.b On its largest investment so
far – the £385 million purchase of the BP campus in Sunbury-on-Thames – the independent
auditor of its accounts cited a “number of significant weaknesses in the Authority’s arrangements
to secure economy, efficiency and effectiveness in its use of resources”.c Since then Spelthorne
has made further high-value property purchases – an office block in the City of London and
another office development in Nine Elms, Battersea – more than doubling its PWLB borrowing.
Woking Borough Council has also taken on £0.6 billion of PWLB debt since March 2017,d
financing the acquisition of retail and property sites, with its largest investment so far the new
Victoria Square commercial and property development (at an estimated cost of £500 million).
In May 2019 the Public Accounts Committee highlighted concerns that while governance
arrangements for the sector as a whole were “generally robust”, some councils have “audit
committees that do not provide sufficient assurance, ineffective internal audit, weak arrangements
for the management of risk in local authorities’ commercial investments, and inadequate oversight
and scrutiny”. It also recommended that MHCLG increases its oversight of the sector urging it to
ensure “concrete actions and outcomes on a timely basis”.e
A number of commentators have highlighted the risks from the concentration of local authority
investments in commercial property, given uncertainty over the future financial prospects of such
investments. Commercial property is often among the hardest hit asset classes during economic
downturns. Prices fell 24 per cent in two years around the 2008 financial crisis.
a
Ministry of Housing, Communities and Local Government, Borrowing and investment live table, Q4 2018 to 2019, July 2019.
b
Spelthorne Borough Council, Statement of Accounts 2016-2017, February 2019.
c
Spelthorne Borough Council, Statement of Accounts 2016-2017, February 2019.
d
Debt Management Office, Public Works Loan Board year end values.
e
House of Commons Committee of Public Accounts, Local Government Governance and Accountability, Ninety-Seventh Report of
Session 2017–19, May 2019.
Public corporations
5.107 Little has changed over the past two years to influence our assessment of the fiscal risks
posed by public corporations. The most significant development has been the delays to the
completion of Crossrail, which have adversely affected Transport for London’s finances and
required additional funding from central government to complete the project (see

paragraph 5.21). The possibility of excessive borrowing, shocks to financing costs and
persistent loss-making remain the most likely sources of future fiscal risks.
Devolved administrations
5.108 Directly elected devolved administrations in Scotland, Wales and Northern Ireland have
controlled substantial public service spending since the late 1990s. Significant tax revenue
has also been devolved since 2015. Funding for the devolved administrations is largely
allocated via a block grant using the Barnett formula, but to account for tax devolution the
Scottish and Welsh Governments’ block grants are additionally adjusted in line with the
fiscal frameworks signed with the UK Government in 2016.47 We saw the core fiscal risk to
the UK public finances in this area being that a devolved administration becomes unable to
fund essential services while servicing any debts it has taken on, prompting the UK
Government to ‘bail out’ a devolved administration rather than allow public services to fail.
5.109 We concluded that the risk was low for two main reasons. First, borrowing powers are
relatively constrained, which should limit the pressure from debt servicing. Second, the
devolved administrations can move resources around within their overall settlements to
manage spending pressures as they occur, with any pressure more likely than not to build
up relatively slowly. But the fiscal frameworks act to insulate the administrations from UK-
wide economic shocks, so more of the risk from moving to greater self-financing is borne by
the UK Government than is the case for local authorities.
5.110 In responding to our report, the Government also noted that the fiscal frameworks require
independent scrutiny of the devolved administrations’ tax forecasts, which should reduce the
risk of overly-optimistic forecasts and in turn large forecast errors and funding shortfalls.
This is carried out by the Scottish Fiscal Commission in Scotland and will be by us in Wales.

5.111 Since our 2017 report there have been two relatively minor developments affecting our
assessment of the fiscal risks associated with the devolved administrations:
• The Scottish Fiscal Commission (SFC) has estimated that income tax forecast-outturn
reconciliations could reduce the Scottish Government’s spending power by several
hundred million pounds over its next three budgets.48 We believe the scale of these
possible adjustments may be unusually large and associated with errors in the initial
estimates of the level of Scottish income tax.49 The largest annual reconciliation in the
47
The agreement between the Scottish Government and the United Kingdom Government on the Scottish Government’s fiscal framework,
February 2016, and The agreement between the Welsh Government and the United Kingdom Government on the Welsh Government’s
fiscal framework, December 2016.
48
See Scottish Fiscal Commission, Scotland’s Economic and Fiscal Forecasts, May 2019 for more detail on reconciliations. The final
amounts will change in light of HMRC estimates of outturn income tax liabilities in Scotland and the rest of the UK in 2017-18, which had
not been published at the time we finalised this publication.
49
We also discussed the error in respect of 2016-17 liabilities in our October 2018 Devolved taxes and spending forecasts publication.

SFC’s latest forecast is £608 million in 2021-22. This would be less than 2 per cent of
Scottish Government current expenditure in that year (of £30.7 billion). As these
reconciliations relate to sums moved between two parts of central government – the
UK and Scottish Governments – they would only represent a fiscal risk to the UK public
finances if the Scottish Government were to borrow to maintain spending (where it is
subject to limits on both reserve use and borrowing) or the UK Government chose to
override the fiscal framework reconciliations to top up Scottish Government funding.
• The Scottish Government has announced proposals for wide-ranging changes to the
design and administration of disability benefits as they are devolved to a new agency –
Social Security Scotland. These may create longer-term pressures on spending, in an
area where the ability to control it is more limited. Major disability benefit reforms have
historically tended to cost more than expected, while the prevalence of disability
reported in surveys is significantly greater than the prevalence of disability benefit
receipt, giving considerable scope for the latter to rise without changes to the former.50
Conclusions
5.112 The nature of the fiscal risks posed by fiscal devolution is largely one of policy choices. We
see a very low probability that a devolved administration could in effect become insolvent.
Much more likely is that pressure from a devolved administration leads to spending
allocations being topped up. There are early signs that this might be true for Scottish and
Welsh devolution, despite the fiscal frameworks. For example, in our March 2019 forecast
the Treasury advised us that outside the Spending Review period its central assumption for
the Scottish Government’s block grant would be that it rises in line with overall DEL
spending, raising our forecast by several hundred million pounds cpmpared to our
expectations based on the Barnett formula and fiscal framework. The size of initial
reconciliations estimated by the SFC could result in further calls for ‘top-ups’ to maintain
spending.
Major provisions and contingent liabilities

5.113 The Government’s accounts record a large number of provisions and contingent liabilities
that reflect future spending incurred by actions to date that are deemed either probable or
possible respectively. The Treasury has established a new control system to manage the flow
of new contingent liabilities that departments take on, with 92 having been subjected to
scrutiny over the past two years – 85 were approved and seven not. But from the perspective
of overall fiscal risk, these new items have generally been small compared with the three
largest sources of provisions and contingent liabilities recorded in the Whole of Government
Accounts: nuclear decommissioning costs, clinical negligence claims and tax litigation cases.
We focus on these major items in this section.
50
See our January 2019 Welfare trends report for a discussion of trends in disability benefits spending and the effects of past reforms.

Nuclear decommissioning

5.114 In our 2017 FRR, we highlighted the risk that the cost of cleaning up the UK’s nuclear sites is
likely to increase over time. Successive reviews had increased estimates of the cost, reflected
in the Nuclear Decommissioning Authority’s (NDA) nuclear provision (for older sites) more
than doubling in real terms in ten years to 2014-15, abstracting from changes to discount
rates.51 Uncertainty about the amount of nuclear waste that will have to be cleared up
(thanks to inadequate record-keeping in the early days of nuclear power generation) and
the technologies that will be available to do so make this particularly uncertain, especially
over the very long term. While the overall impact of nuclear provisions on the Whole of
Government Accounts balance sheet is large, the annual fiscal impact is relatively small
since the costs of the clean-up operations are spread over as much as 120 years.
5.115 We also discussed how the Government’s approach to risk management varies by vintage
of site, tolerating risks around older sites – accepting costs falling entirely to government –
but seeking to transfer risk to the private sector for the second generation and new sites. On
this transfer approach, we highlighted the risk associated with the Nuclear Liabilities Fund
(NLF) – an independent segregated trust with assets intended to meet the future costs of
decommissioning. In the event that the fund’s assets are insufficient to meet its liabilities,
outstanding liabilities fall to government. The NLF’s Trustees warned in its 2015-16 Annual
Report that “expected investment returns may be insufficient to meet the currently projected
nuclear liabilities, based on current assumptions and current investment policy”.
5.116 In light of this, we raised two specific issues for the Government’s response:
• The increase over time in the expected cost of cleaning up the Sellafield nuclear site.
• The Government’s potential exposure to clean-up costs for new nuclear stations.
5.117 In MFR the Government outlined its approach to managing nuclear decommissioning risks,
including identifying and disclosing future costs and provisions, a set of mitigating actions,
such as legislating for new projects to have a ‘Funded Decommissioning Programme’, and
applying best practice across the sector.
Updated risk assessment and conclusions

5.118 Having risen significantly in two of the past three years, the NDA’s 2018-19 annual report
records a significant fall in the provision for nuclear decommissioning – from £234 billion in
2017-18 to £131 billion in 2018-19.52 However, these movements are almost entirely
explained by changes to the discount rate (Chart 5.7), which are adjusted each year to
reflect changes in real government borrowing costs. But this year’s NDA provision estimate
is also affected by a methodological change in the Treasury’s guidance on discount rates.
51
National Audit Office, The Nuclear Decommissioning Authority: progress with reducing risk at Sellafield, June 2018.
52
Nuclear Decommissiong Authority 2018-19 annual report and accounts, 4 July 2019.

This change applies to all provisions and will lead to significant changes in next year’s
WGA. Because the expected outlays from nuclear decommissioning are very long term – up
to 120 years – the discounted estimate is very sensitive to small changes in the discount rate.
5.119 Abstracting from the effect of discount rate changes, provisions are only marginally higher.
After many years of escalating costs, in June 2018, the NAO reported that the NDA’s
estimate of the undiscounted provision “has stabilised … as the NDA’s understanding of the
scope and cost of work has improved”.53 This means that this risk remains broadly
unchanged as the undiscounted provision remains within a 5 per cent range since 2014-15.
Chart 5.7: Nuclear decommissioning provisions

250
Discounted
Undiscounted
200
£ billion (2018-19 prices)
150
100
50
0
2013-14 2014-15 2015-16 2016-17 2017-18 2018-19
Source: NDA, OBR
5.120 The Nuclear Liabilities Fund, set up by the Government as an independent segregated trust
to meet future costs of decommissioning, continues to have around £9 billion in assets as of
31 March 2018.54 As discussed in our 2017 FRR, if the fund’s assets are insufficient to meet
its liabilities, outstanding liabilities will fall to government. This risk here is unchanged.
5.121 To date Hinkley Point C remains the only new generation site officially approved. As regards
the ‘contract for difference’ related to Hinkley Point C’s power generation, BEIS’ best
estimate of its fair value has risen from £29 billion in 2016-17 to £37 billion in 2017-18
(although this has not been recognised in its accounts).55 The NAO’s report on the 2017-18
Whole of Government Accounts highlights the “uncertainties inherent in valuing the Hinkley
Point C contract for difference which has resulted in the liability not being recognised on the
Statement of Financial Position, as it is not considered to meet the recognition criteria set out
in the Conceptual Framework for Financial Reporting.” The contract for difference means
53
National Audit Office, The Nuclear Decommissioning Authority: progress with reducing risk at Sellafield, June 2018.
54
Nuclear Liabilities Fund 2018 annual report, 21 December 2018.
55
Department for Business, Energy, and Industrial Strategy Annual Report and Accounts 31 March 2018, July 2018.

that the immediate bearers of risk are households, who would pay higher prices for
electricity in the event that wholesale prices were lower than assumed. This could become a
fiscal risk if the Government decided to intervene to reduce the extent to which household
energy bills increased, while honouring the Hinkley Point contract.
5.122 While the nuclear decommissioning provision remains large, the fiscal risk associated with it
remains relatively small because it covers such a long period. The fiscal risks associated with
the NLF and Hinkley Point also appear relatively small. On this basis, our overall
assessment of the fiscal risks in this area is unchanged from our 2017 report.
Clinical negligence

5.123 In our 2017 FRR, we discussed increasing clinical negligence pay-outs, highlighting the
growing demand for medical services, as well as growth in the number of negligence cases,
the average damages per successful claim and the associated legal fees. We raised two
issues for the Government’s response:
• The likelihood of higher clinical negligence pay-outs than currently provisioned for.
• The significant proportion of clinical negligence costs still accounted for by legal fees.
5.124 In MFR the Government said it intended to publish a review of cross-government strategy in
autumn 2018, but, like the social care green paper, publication has been delayed.56 The
Government now plans to provide an update on progress on the strategy later this summer.

5.125 Since our 2017 FRR, the number of clinical negligence claims has plateaued, but the
average amount paid out per claim has increased by 16 per cent a year on average since
2010-11. This pushed the annual cost up to £2.1 billion in 2017-18.57 It also means that
pay-outs have risen as a proportion of the health budget in England (Chart 5.8).
56
Letter from the Permanent Secretary of DHSC to the Chair of the Public Accounts Committee of the House of Commons, Managing the
costs of clinical negligence in hospital trusts, Session 2017-2019 (HC 397), 12 October 2018.
57
NHS Resolution annual report and accounts 2017-18, 12 July 2018.

Chart 5.8: Clinical negligence payments as a proportion of the DHSC budget

1.8
1.6
1.4
Per cent of total DHSC DEL
1.2
1.0
0.8
0.6
0.4
0.2
0.0
2006-07 2007-08 2008-09 2009-10 2010-11 2011-12 2012-13 2013-14 2014-15 2015-16 2016-17 2017-18
Source: DHSC, HM Treasury
5.126 NHS Improvement has published a new patient safety strategy, which includes a focus on
reducing negligence.58 It is too soon to say how effective this might be, and any positive
results are only likely to have an effect in the longer run. On limiting legal fees, the
Government is awaiting the Civil Justice Council’s recommendations on legal costs
disproportionate to damages awarded before consulting on next steps.
5.127 One policy variable that can affect costs per claim is the personal injury discount rate (PIDR).
In 2017 the Government reduced it from 2.5 to minus 0.75 per cent. The PIDR is used in
calculating the present value of insurance and indemnity pay-outs, with a lower rate
meaning higher pay-outs. The reduction to -0.75 per cent increased costs for the NHS,
which HM Treasury funded from the reserve. The direct cost of the lower PIDR to the NHS in
England was £400 million in 2017-18, accounting for almost 80 per cent of the £520
million increase in payments compared with the previous year.
5.128 In accordance with the Civil Liability Act 2018, the Lord Chancellor launched a review of the
PIDR on 19 March 2019, which must be completed by 5 August 2019.59 The terms of the
review suggest that the rate will be raised when it is next changed, but the extent of any
change is uncertain. If it were raised, that would lower costs for the NHS.
5.129 NHS Resolution has since April 2019 operated a new state indemnity scheme for general
practice in England – the Clinical Negligence Scheme for General Practice (CNSGP).60 It
provides a fully comprehensive indemnity for all claims within its scope for incidents arising
on or after 1 April 2019, and replaces the previous system, in which GPs had to seek
58
NHS Improvement, The NHS patient safety strategy, 2 July 2019.
59
Letter from the Lord Chancellor and Secretary of State for Justice to the Chair of the Justice Committee of the House of Commons, 18
March 2019.
60
NHS Resolution business plan 2019-20, 7 May 2019.

indemnity cover from private medical defence organisations or other providers. In addition,
DHSC has reached agreement with the Medical Protection Society Limited, one of the
medical defence organisations, in relation to existing in-scope liabilities for general practice
in England for incidents prior to 1 April 2019. These arrangements will add to the
Government’s liabilities, though figures have not been published by NHS Resolution or
DHSC. But it is not clear that this change increases overall fiscal risk. Under the previous
system, if the cost of clinical negligence claims against GPs increased, and the cost of their
indemnity cover did too, it seems plausible that it would have been reflected in public sector
pay decisions.
Tax litigation

5.130 In our 2017 FRR, we discussed the possibility of increasing tax litigation pay-outs if HMRC
lost a significant ‘lead’ case, which could trigger pay-outs in subsequent ‘follower’ cases. In
particular, we highlighted the Littlewoods case over the way interest is calculated on repaid
tax. At the end of 2015-16, HMRC had recorded contingent liabilities of £49.1 billion on
possible pay-outs. In November 2017, HMRC unexpectedly won the Littlewoods case, which
the Government noted in its MFR response to our report. This victory reduced HMRC’s
central estimate of tax litigation losses by £18 billion.

5.131 HMRC’s 2017-18 annual report shows a significant drop in contingent liabilities, from
£49.1 billion in 2015-16 to just £6.0 billion in 2017-18 (Chart 5.9). This mostly reflects
HMRC’s win in the Littlewoods case, which also meant no ‘follower’ cases were brought.
The fall in contingent liabilities and the number of cases HMRC records as being under
dispute that involves large sums has led us to downgrade our assessment of the fiscal risk
posed by large tax litigation pay-outs adding significantly to public spending. New cases will
always emerge – for example, Credit Suisse is seeking to recover £239 million it paid in
‘bank payroll tax’ that was levied briefly after the financial crisis – but there are no cases on
the horizon that would have follower-case implications as large as the Littlewoods one.

Chart 5.9: HMRC contingent liabilities in litigation cases

60 30
Contingent liabilities
Number of cases
50 25
Contingent liabilities (£ billion)
40 20
Number of cases
30 15
20 10
10 5
0 0
2013-14 2014-15 2015-16 2016-17 2017-18
Source: HMRC
Contingent commitments not treated as contingent liabilities

5.132 Over the past two years, we have investigated the assurances that the Government provided
to Nissan in 2016, and we reported on these in our March 2019 EFO. We were particularly
interested in this case because these assurances seemed to constitute a commitment that
was contingent on Nissan making a particular decision about production in the UK, so in
some sense a ‘contingent liability’. But they were not recorded as such and the NAO
confirmed that this was consistent with the standards that govern the Government’s
accounts.61 This posed the question of whether there are other apparently contingent
commitments that do not meet the accounting definition of a contingent liability, but that
when aggregated might represent a fiscal risk that we should still be concerned about.
5.133 We have discussed this wider issue further with the NAO. We understand that the Nissan
case will not be unique. Where departments have the legislative and discretionary power to
issue grants from within their DEL allocations, they can engage with prospective grant
recipients and signal that support could be given, while stopping short of entering into any
commitment or obligation until the point when the grant is finally awarded. As such, there
can be a long period – as with the Nissan case – between the initial stage of signaling that a
grant may be forthcoming, and the final stage, if reached, of awarding it during which
proposals are subject to due diligence and approval. Formally, there is no commitment to
spend until the grant is awarded. It can only be made as and when there are funds
available within a department’s DEL. As such, the initial statements of intent are judged to
be part of departments’ process of exercising their discretionary powers, rather than
creating guarantees. The wider fiscal risk (of which we need to be mindful, but which
61
Letter from Sir Amyas Morse, Comptroller and Auditor General, to Andrew Tyrie, Chair of the Treasury Select Committee, Government’s
commitments and/or assurances to Nissan, 13 December 2016.

departments are required to manage) is the extent of the underlying need that creates the
spending pressure, versus the extent of the funds available.
5.134 Other instances of actual contingent liabilities that are not reported in full include cases
where the liability is deemed to be commercially sensitive or confidential, as with time-
limited guarantees that might be issued to underwrite sales of assets, for instance. Under
accounting standards, if such contingent liabilities are assessed as being both material in
size and with a ‘possible’ likelihood of happening, then they should be reported in the
department’s accounts, as clearly as possible without breaking confidentiality. The Treasury
has assured us that, as far as this can be done, such contingent liabilities are also included
in the WGA. Sensitive contingent liabilities of this type are subject to the Treasury’s new
system for scrutinising new contingent liabilities, and included within the regular update to
us that we use as the basis for reporting on contingent liabilities in each EFO.
Brexit-related spending risks

5.135 In our 2017 FRR, we discussed several Brexit-related risks to spending, including the
financial settlement with the EU (the so-called ‘divorce bill’); potential future commitments to
replace spending relating to EU programmes such as farm support and research funding;
establishing and running UK-specific regulators; carrying out Brexit and other related
negotiations; and any sector-specific interventions to support or compensate companies and
industries adversely affected by the UK leaving the single market and the customs union.
5.136 In MFR, the Government responded to our assessment of these risks by highlighting the
principles underlying the agreement regarding the financial settlement and said that future
decisions with respect to replacing EU expenditure in the UK and the UK’s continued
participation in EU programmes would be considered at the 2019 Spending Review.

The EU financial settlement
5.137 The Government and the EU have now agreed the methodology for calculating the financial
settlement. This was set out in final detail in the Withdrawal Agreement published on 25
November 2018, although it has been broadly settled since the joint report of 8 December
2017. We published a detailed walk-through of the calculation in Annex B of our March
2018 EFO, and have since updated our estimate at each forecast.
5.138 Our March 2019 estimate stood at €41.8 billion, which translated into £37.8 billion at the
exchange rates in our forecast. This was predicated on the UK leaving the EU on 29 March
2019. As this did not happen, the UK continues to contribute as a member, rather than
making payments under the settlement. However, while this technically reduces the size of
the settlement, it does not change the overall amount that the UK transfers to the EU. The
Withdrawal Agreement provides that the UK will contribute to the EU’s annual budgets as if
it were a member until December 2020, and so no changes to payments have occurred.

5.139 Based on our March forecast, but assuming a 31 October 2019 exit date, the financial
settlement would be €36.3 billion, or £32.8 billion, with the remaining £5.0 billion
difference being paid as part of the UK’s normal membership contributions to the EU.
Future expenditure commitments

5.140 As regards future expenditure in place of current EU programmes, the Government has
already committed to maintaining several of them after Brexit, including farm support,
industrial strategy and science programmes. In our forecasts, we have maintained a fiscally
neutral assumption that any reductions in net expenditure transfers to the EU will be recycled
into additional domestic spending. The Government’s statements to date suggest this
provides a reasonable holding assumption until the precise details of those commitments
are settled. Specifically, the following commitments have been made since the referendum:
• Farm support: the Government’s proposed Agriculture Bill in the current Parliamentary
session contains a pledge “to continue to commit the same cash total in funds for farm
support until the end of this parliament, expected in 2022”.62 European Commission
figures suggest that this farm support was around €4 billion in 2018.63
• Shared prosperity fund (SPF): the Government’s November 2017 industrial strategy
white paper set out that “following the UK’s departure from the European Union, we
will launch the UK Shared Prosperity Fund. We intend to consult next year on the
precise design and priorities for the fund.”64 The 2017 Conservative manifesto stated
that “we will use the structural funds that come back to the UK following Brexit to create
a UK shared prosperity fund”. On the definition of structural funds used in the
Treasury’s annual EU finances document, the 2018 figure was around £1 billion.65
• Replacement of Official Development Assistance (ODA) funds: the Government has

legislated to maintain ODA spending of at least 0.7 per cent of gross national income
(GNI).66 At present, some EU ODA (broadly the proportion funded by UK
contributions) counts towards this commitment. The attribution of some EU ODA
spending to the UK accounted for £1.4 billion of the UK’s total £14.1 billion of ODA
spending in 2017.67
• Science and education: the Prime Minister stated in her 2018 Mansion House speech
that “The UK is also committed to establishing a far-reaching science and innovation
pact with the EU, facilitating the exchange of ideas and researchers. This would enable
the UK to participate in key programmes alongside our EU partners. And we want to
take a similar approach to educational and cultural programmes, to promote our
shared values and enhance our intellectual strength in the world – again making an
62
Parliamentary Under-Secretary of State (Department for Environment, Food and Rural Affairs), Answer to written question HL10006 in
the House of Lords, 18 September 2018.
63
European Commission, Agriculture in the European Union and Member States – United Kindgom statistical factsheet, June 2019.
64
Department for Business, Energy and Industrial Strategy, Industrial strategy: building a Britain fit for the future, 27 November 2017.
65
HM Treasury, European Union Finances 2018, 5 June 2019.
66
The International Development (Official Development Assistance Target) Act 2015, 26 March 2015.
67
Department for International Development, Statistics on international development: Final UK Aid spend 2017, November 2018.

ongoing contribution to cover our fair share of the costs involved.”68 It is unclear what
precisely this would apply to, but if it were taken to relate to programmes like
Erasmus+, Creative Europe and Horizon 2020, the cost of participation is currently
around €2 billion a year.69
• Regulatory agencies: the Prime Minister also noted in her Mansion House speech that
“We will also want to explore with the EU, the terms on which the UK could remain part
of EU agencies such as those that are critical for the chemicals, medicines and
aerospace industries: the European Medicines Agency, the European Chemicals
Agency, and the European Aviation Safety Agency. We would, of course, accept that
this would mean abiding by the rules of those agencies and making an appropriate
financial contribution.” Again, it is unclear as to the precise list of agencies that the UK
will seek to remain part of (there are over 30 EU regulatory agencies). The EU
decentralised agencies budget in 2017 was around €350 million, implying a UK
financing share of around €40 million for the full list.70
5.141 These statements were all made by the outgoing administration, so all are subject to
uncertainty as a new one takes up the reins of government and pursues its own priorities.
5.142 We have continued to include the ‘assumed spending in lieu of EU transfers’ as an AME line
in our forecasts, but any expenditures that replace EU schemes might be more suited to
being managed under the DEL spending control system. If that were the case, these
programmes would be considered as part of the forthcoming or subsequent Spending
Reviews. Indeed, this is what the Government indicated in MFR that it would look to do.
Conclusions
5.143 Since our 2017 FRR, the Government and the EU have agreed the methodology for
calculating the financial settlement, which means that the risk can now be quantified. It is
somewhat lower than the estimates that had been put forward by external commentators at
the time of our previous report. Risks regarding post-Brexit replacement expenditure remain,
as does the uncertainty as to how exactly the replacement programmes will be delivered,
although the Government has already committed to a large list of targeted expenditure.
68
Prime Minister’s speech on the UK’s future economic partnership with the European Union, 2 March 2018.
69
House of Commons Library, UK funding from the EU, Briefing paper number 7847, 28 November 2018. Our analysis assumes a 12.3
per cent UK finacing share, consistent with the assumption used in our March 2019 EFO.
70
Draft General Budget of the European Union for the financial year 2019, COM(2018) 600, June 2018.


5.144 In this chapter we have discussed several issues that the Government is likely to wish to
consider when managing its fiscal risks, while others that we discussed in greater detail in
our 2017 FRR remain pertinent despite changing little since then. These include:
• The tendency for major spending policies to be announced outside Spending Reviews.
• The possibility of cost overruns for major projects like HS2 and MoD procurement.
• The plausibility of assumed rates of productivity improvements in health care.
• The risk of further topping-up of health spending settlements, despite more funding.
• The large long-term upward cost and demand pressures on health and care spending.
• The potential impact of the NLW and migration reform on health and social care costs.
• Policy uncertainty around the desire to limit individuals’ exposure to social care costs.
• Potential pressure to bail out a private social care provider if in financial difficulty.
• The precedent set by yielding to pressure to reverse planned cuts to welfare spending.
• The continued commitment to the ‘triple lock’, which ratchets pension spending higher.
• The knock-on effects to benefit take-up of other policy and operational changes.
• The possibility that local authorities will resume running down their reserves.
• The rise in local authorities’ prudential borrowing for commercial property purchases.
• Pressure to top-up devolved administrations’ budgets despite the ‘fiscal frameworks’.
• The scale of nuclear decommissioning costs and uncertainty over Hinkley Point costs.
• The rising share of NHS spending taken up by clinical negligence pay-outs.
• The cost of commitments to replace current EU programmes.
5.145 When assessing the outlook for public spending over the medium and long term, does the
Government regard these or other issues as important for its risk management strategy and,
if so, how does it intend to address them?

6 Balance sheet risks
6.1 The public sector balance sheet provides estimates of the assets and liabilities held by
central government (including the devolved administrations), local government and public
corporations. Several balance sheet measures are published for the UK, differing in
coverage and accounting treatment. All show that the stocks of assets and liabilities are
large relative to flows of spending and receipts and to national income. The asset and
liability sides of the balance sheet can both be sources of fiscal risk.
6.2 The most comprehensive National Accounts balance sheet measure is public sector net
worth (PSNW). This includes financial and non-financial assets and liabilities, but excludes
the present value of future tax revenues and most spending. Ideally, we would use PSNW
throughout this chapter to discuss balance sheet risks as it provides the best National
Accounts measure of fiscal sustainability. But the ONS has not published PSNW data since
2012 due to concerns about the quality of the public corporations’ non-financial assets
data. It aims to address these concerns later in 2019. So instead we use other National
Accounts aggregates: public sector net financial liabilities (PSNFL), which, unlike PSNW,
excludes non-financial assets, and public sector net debt (PSND), which only includes a
narrow set of ‘liquid’ financial assets. PSND is the Government’s target measure of debt.
6.3 The Whole of Government Accounts (WGA) offer an alternative view, based on international
accounting standards. These have wider coverage even than PSNW, particularly of future
liabilities such as the present value of future public service pension costs already accrued
and provisions for items where past activity has generated future expected liabilities.
6.4 Table 6.1 shows how these different measures related to each other at the end of March
2018. PSND stood at 85 per cent of GDP, with nearly £2 trillion of debt liabilities (largely
debt securities, including gilts and Treasury bills) offset by less than £200 billion of liquid
assets (largely the foreign exchange reserves). PSNFL includes a slightly broader set of
liabilities but a significantly larger pool of financial assets (notably student loans, which are
recorded at their nominal value). As such it is somewhat lower than PSND at 67 per cent of
GDP. WGA net liabilities, which include £1.9 trillion of public service pension liabilities
partially offset by £1.2 trillion of non-financial assets, stands much higher at 122 per cent of
GDP. Public service pension liabilities and provisions are both very sensitive to the rate at
which future expected costs are discounted, with discount rate changes driving volatility in
liabilities from year to year. Such a change will materially lower the provisions liability in
2018-19, as we discuss in Chapter 5 with respect to nuclear decommissioning.

Balance sheet risks
Table 6.1: Various measures of the public sector balance sheet in 2017-18
Liability type Asset type Balance sheet aggregate
Per cent of
£ billion £ billion £ billion
GDP
Gilts and T-bills 1505 Foreign exchange reserves 116
Other debt liabilities 457 Other liquid assets 68
PSND liabilities 1962 PSND assets 183 PSND 1779 85
Other financial liabilities 118 Other financial assets 481
PSNFL liabilities 2080 PSNFL assets 664 PSNFL 1416 67
Accrued pension liabilities 1865 Non-financial assets 1208
Provisions 423 Other valuation differences 141
Other valuation differences 211 WGA net
WGA liabilities 4579 WGA assets 2014 liabilities 2565 122
Note: 'Other debt liabilities' in PSND includes the net liability of the Bank of England.
6.5 The WGA, like PSNW and other measures of debt used in this report, exclude the public
sector banks. At their peak, including both RBS and Lloyds Banking Group in most of these
measures would have roughly doubled the size of both assets and liabilities. In PSND,
including them raised liabilities much more than assets, because the banks’ assets – notably
their mortgages and business loans – are not defined as ‘liquid’ for PSND purposes.
6.6 This chapter:
• discusses potential sources of balance sheet risk;
• reviews the risks covered in our previous Fiscal risks report (FRR), grouping them by
potential source and noting how the Government responded in its 2018 Managing
fiscal risks (MFR) publication to the specific issues we raised;
• revisits the issue of ‘fiscal illusions’; and
• takes a more in-depth look at intangible assets.
Sources of balance sheet risk

6.7 The evolution of the debt-to-GDP ratio depends on the primary budget balance (the
difference between non-interest revenues and spending), debt interest spending and ‘stock-
flow adjustments’. The first two are flows recorded within the deficit that add to the stock
year by year. The latter affect the stock directly, without an associated deficit flow. It is these
we investigate in this chapter.
6.8 A variety of stock-flow adjustments are relevant when assessing fiscal sustainability:
• Balance sheet transactions, in which the government issues debt to buy a financial
asset or to lend to the private sector (such as the purchase of shares in RBS and Lloyds
Banking Group during the financial crisis, or, more recently and on a smaller scale,
the loan provided to British Steel to meet its 2018 EU carbon emissions liability).

Balance sheet risks
• Balance sheet transfers, in which the government directly absorbs the assets and
liabilities of a private sector entity (this can be a real-world event, like the transfer of
the Royal Mail’s historic pension liabilities and associated assets to the public sector in
2012, or a statistical one, as when the ONS reclassified English housing associations
from the private to the public sector in 2015 and then back again in 2017).
• Changes in the value of existing assets and liabilities, such as the impact of a
movement in the exchange rate on the sterling value of the UK’s unhedged foreign
exchange reserves and, less materially, its debt denominated in foreign currencies.
• Timing or other accounting differences. Timing differences arise because public sector
net borrowing (PSNB) is recorded on an accruals basis, whereas PSND is largely a
cash measure. This means, for example, that tax receipts are usually recorded in PSNB
when the underlying economic activity that is being taxed took place, but in PSND
somewhat later when the associated tax payments are received.
6.9 Timing and accounting differences are not usually relevant when assessing long-term fiscal
sustainability, because they typically even out in the end. But they are worth scrutiny when
the accounting methodology clouds the ‘true’ picture. Two examples are discussed in the
fiscal illusions section of this chapter: the interest on student loans and the recording of gilts.
6.10 As Chart 6.1 shows, the cumulative effect of stock-flow adjustments was not, in aggregate,
a significant contributor to the level of net debt over the years prior to the financial crisis.
The path of PSND in the decade before the crisis could be explained almost entirely by the
path of borrowing with little effect from stock-flow adjustments. But from 2007-08, stock-
flow adjustments contributed 12 percentage points of the overall 49 per cent of GDP rise in
PSND. Over the medium term we expect PSND to decline slowly relative to GDP, largely
thanks to primary surpluses but also a lower contribution from stock-flow adjustments
Chart 6.1: Contributions to changes in net debt since 1997-98

60
Cumulative stock-flow adjustment Forecast
Cumulative primary deficit
50
Cumulative debt interest
Cumulative PSND
40
30
Per cent of GDP
20
10
-10
-20
1997-98 2000-01 2003-04 2006-07 2009-10 2012-13 2015-16 2018-19 2021-22
Note: Stock-flow adjustments calculated as the residual unexplained by the primary balance or debt interest.
Source: ONS, OBR

Balance sheet risks
Public sector net worth

6.11 In our 2017 FRR we identified risks arising from the deterioration in broad measures of
public sector net worth since the crisis. In MFR, the Government pointed to new measures of
balance sheet performance. These included asking us to forecast the gross stocks underlying
our PSNFL forecast and the Treasury’s own work with the ONS to introduce tables consistent
with the IMF’s Government Finance Statistics Manual (GFSM) into the regular public sector
finances statistical release, which it has now done. In Chapter 1 we looked at the Treasury’s
‘Balance Sheet Review’ and other steps it has taken to manage its balance sheet better.

6.12 The deterioration in public sector and general government net worth are shown in Chart
6.2. Both fell sharply during the financial crisis (largely for the same reasons driving the rise
in PSND) and have been negative since. The direction of travel is the meaningful indicator
of how fiscal sustainability has evolved. There is no particular significance to their move
from positive to negative territory, since the public sector’s balance sheet is backed by its
right to tax this and future generations – an asset not recognised in either measure.
Chart 6.2: Public sector and general government net worth

40
General government net worth
30
Public sector net worth
20
10
Per cent of GDP
-10
-20
-30
-40
-50
1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 2015 2017
Source: ONS
6.13 The IMF has been developing a new database of cross-country estimates of public sector net
worth to complement the narrower measures that are more commonly used in international
comparisons.1 Its estimate of the UK public sector’s net worth stood at minus 125 per cent
1
International Monetary Fund, 2019, Public Sector Balance Sheet Database.

Balance sheet risks
of GDP in 2016, the second weakest out of the 13 advanced countries for which it has full
estimates (Chart 6.3). Total assets were 135 per cent of GDP, the lowest recorded among
these countries, while total liabilities amounted to 260 per cent, the third highest.
6.14 Cross-country analysis of this nature is challenging. Gathering public sector-wide data on
non-financial assets and pensions is particularly difficult and countries will not be using
entirely consistent valuation methodologies. Nevertheless, removing these two elements of
the IMF’s cross-country estimates would not alter the UK’s position relative to the other
countries shown (though it would have larger effects on most of them, and particularly so in
Norway and Australia, as it excludes their natural resources, and in Japan, where the value
of other non-financial assets is higher relative to GDP than in the other countries).
Chart 6.3: Public sector balance sheets in selected advanced economies

800
Financial assets Natural resources
Other non-financial assets Pension liabilities
600
Other liabilities Net worth
PSNW ex pensions and non-financial assets
400
200
-200
-400
-600
Finland
Australia
UK
Norway
US
South Korea
Austria
France
Canada
Germany
Portugal
New Zealand
Japan
Source: IMF
6.15 Given the current hiatus in publishing PSNW data in the UK, we do not currently forecast it.
But we do forecast PSNFL. Chart 6.4 shows how our PSNFL forecast has been revised down
significantly since the March 2017 forecast that underpinned our previous FRR. In part this is
because the ONS has revised down the outturn in 2014-15 relative to the data in 2017 – by
£56 billion (3.9 per cent of GDP). That in turn largely reflects changes in the recording of
pensions liabilities, reducing them materially. The downward revision relative to our March
2017 forecast becomes larger in 2017-18, due to the reclassification of English housing
associations to the private sector that took effect in November 2017 (see Box 6.1).
6.16 We expect PSNFL to decline in each year of our latest forecast, at an average rate of 1.5 per
cent of GDP a year. This is about the same rate we forecast in March 2017, despite the
lower level. But in both these forecasts PSNFL was flattered by an over-valuation of the
student loans stock, which the ONS intends to address in methodological changes to be
implemented this September. The ONS has released provisional estimates of the value of

Balance sheet risks
the stock of student loans under the new methodology. They show PSNFL in 2017-18 would
have been 2.1 per cent of GDP higher, while the drop in PSNFL relative to 2016-17 would
have been 0.3 per cent of GDP smaller. On that basis, PSNFL would continue to improve
each year of the forecast, but at a slower pace of 1.1 per cent of GDP a year.
Chart 6.4: Successive PSNFL forecasts and outturns

80
75
70
Per cent of GDP
65
60
March 2017
55
March 2019
March 2019 after student loan stock revision
50
2014-15 2015-16 2016-17 2017-18 2018-19 2019-20 2020-21 2021-22 2022-23 2023-24
Note: The forecasts of revised student loans are consistent with those presented in Annex B of our March 2019 EFO.
Source: ONS, OBR
Conclusions
6.17 Overall the trend in the various measures of the balance sheet position are little changed
since our previous report. The sharp deterioration since the financial crisis has ceased and a
slow improvement is forecast on current policy settings. But, at the current rate of progress it
would take several decades to get back to a pre-crisis position. And available data suggest
the UK’s public sector net worth is among the worst of its international peers.
Balance sheet transactions

6.18 In the medium term we consider risks as deviations from our central forecast. The risk from
asset sales then is that the annual receipts from them are significantly different from
expectations. For guarantees it is that there are significant calls on them.
6.19 In the longer term, where we are concerned with sustainability, the considerations are
different. Asset sales swap an uncertain flow of future returns on the assets for a known cash
sum up front. Where markets are liquid and the sales process competitive we may assume
that asset sales are broadly neutral for sustainability as the proceeds should reflect what the
asset is worth. For guarantees, if government is able to assess the risk it is taking on and to

Balance sheet risks
price that risk accordingly, fiscal sustainability may be little changed in expectation – i.e. on
average in all possible future scenarios, although of course in those where guarantees are
called it would be worsened. And it is also worth considering whether government is taking
on risks that are correlated with other risks – in this case it would leave itself more
vulnerable to extreme events.
6.20 In our 2017 FRR we identified risks arising from:
• asset sales that could be delayed or raise less than expected;
• asset sales that have not been factored into current forecasts; and
• the growing use of guarantees in infrastructure and housing.
6.21 In MFR the Government drew attention to the progress it had made in disposing of the
assets acquired during the financial crisis and noted that the Public Accounts Committee
and the National Audit Office “have generally been positive on the overall assessment that
sales have proved to be value for money”. It also pointed to the increased transparency
around the fiscal impact of asset sales and the framework put in place by the Ministry of
Housing, Communities and Local Government to manage the risks associated with its loan,
guarantee and equity loan products, with procedures covering both new products as they
are developed, and live products that have already been approved.

Asset sales
6.22 Chart 6.5 shows successive asset sales forecasts. Usually the Government has planned to
raise more through asset sales in the early years of each forecast and little by the end. At
successive forecasts additional sales are then pencilled in for the later years. For example, at
March 2017 the Government had planned for only £0.7 billion to be raised in 2021-22
(and this was only from the natural run-off of UKAR mortgages), but by March 2019 it had
increased this to £6.6 billion, with plans to sell tranches of student loans and RBS shares.
6.23 The largest forecast revisions can be seen in 2019-20: in March 2017 the Government had
plans to raise £3.1 billion, but this has since risen to £16.4 billion. The largest contributors
to this are UKAR and RBS, where there were no plans for active sales in March 2017 but the
Government now plans to raise £13.2 billion from these sources.
6.24 As the Government sells the assets acquired in the financial crisis (after 2019-20 only RBS
shares should remain to be sold) the risks from volatility in the proceeds from asset sales
should subside. That said, Government policy is to seek further sales not currently factored
into our forecasts, which it considers to be “an integral part of the government’s plan to
improve the public finances”.2
2
HM Treasury, Budget 2018, October 2018.

Balance sheet risks
Chart 6.5: Successive OBR forecasts of proceeds from financial asset sales
30
March 2017
Successive forecasts
25
March 2019
20
£ billion
15
10
0
2017-18 2018-19 2019-20 2020-21 2021-22 2022-23 2023-24
Source: OBR
6.25 When considering asset sales, the Government seems to look primarily at the PSND impact
rather than other balance sheet measures. For example, in MFR it pointed to the reduction
in PSND from the sales, but did not discuss PSNFL, PSNW or WGA measures of the balance
sheet. And PSND is the only metric on which such sales could be considered to “improve the
public finances”. Reducing PSND was one of three formal objectives for the sale of student
loans. That said, in recently published guidance for departments on how to report asset
sales to Parliament, the Treasury has sought to address this excessive focus on PSND at least
in presentation.3 It requires disclosures to Parliament to include:
• a qualitative policy rationale for the sale;
• a justification of the format and timing of a sale;
• the proceeds of the sale;
• an explanation of whether the sale was above, at or below the retention value; and
• the impact of the sale on PSNB, PSND, PSNFL and WGA net liabilities.
6.26 It is too soon to assess whether these greater disclosure requirements will affect the weight
of PSND impacts in decision-making. This may depend on the prominence given to changes
in the PSND-to-GDP ratio in the next vintage of fiscal targets. Against the broader measures
of the balance sheet, most asset sales make little difference as assets will be valued at close
to their market price.
3
HM Treasury, Asset sale disclosures: guidance for government, March 2019.

Balance sheet risks
6.27 For student loans, the difference between the effect on PSND and on broader measures is
more stark as they are sold at steep discounts to their value in the Department for
Education’s accounts and in the National Accounts measure PSNFL. The sales clear the
hurdle of ‘value for money’ due to the choice of high discount rates when making that
assessment. This issue was discussed in Box 4.4 of our October 2018 EFO, where we
concluded that in essence these sales mean the Government pays the buyers a higher rate
of return than it would have had to pay if it had retained the assets on its own balance sheet
and financed them conventionally via government bonds. They therefore represent a more
expensive way of financing the loans, so the beneficial effect of the sales on PSND masks a
modest detrimental effect on fiscal sustainability.
6.28 In its evaluation of the value for money of student loans sales the National Audit Office
concluded that the use of different assumptions when issuing and when selling them “risks
government: not knowing with enough certainty the cost to the taxpayer of student loans
when they are issued; and of selling assets too cheaply relative to their long-term value”.4
Loans
6.29 The public sector currently has three lending schemes that are fiscally material: student
loans, Help to Buy equity loans and lending via the Bank of England’s Term Funding
Scheme. The past two years have seen significant changes to all three:
• Changes to the accounting treatment of student loans are discussed later in this
chapter. Partly as a result of these changes there is currently considerable uncertainty
about how the Government will finance higher education in future. Box 6.2 discusses
the Review of post-18 education and funding. The departing Prime Minister made a
statement welcoming the report’s findings and supporting its recommendations, but it
is not clear what view the new administration will take.
• In October 2018 the Government announced a two-year extension to the Help to Buy
equity loan scheme to 2022-23, alongside changes to how the scheme will operate
until then that are expected to lower the flow of new lending. It committed to end the
scheme in March 2023, but it is of course possible that the scheme will be extended
again nearer its planned closure date, as has happened three times already. A recent
NAO report concluded that Government’s investment (£11.7 billion by December
2018) had exposed it to significant market risk as the loans are sensitive to house price
movements and to the timing of buyers repaying the loans.5
• In FRR 2017 we estimated that loans under the Term Funding Scheme (TFS) would
reach £90 billion, whereas the scheme actually lent £127 billion. This has no
implications for fiscal sustainability in any but the most extreme circumstances, and
does not affect the broad balance sheet measures. But, as the loans roll off, it
increases the risks to the path of PSND in the medium term. Were the loans to be
rolled over – rather than being redeemed at the end of their 4-year term as we
4
National Audit Office, The sale of student loans, July 2018.
5
National Audit Office, Help to Buy: Equity Loan scheme – progress review, June 2019.

Balance sheet risks
currently assume – this would reduce the Government’s 3.2 per cent of GDP
headroom against its falling PSND target by 2.2 per cent of GDP.
Guarantees
6.30 The Government has introduced a new approval regime for guarantees and other
contingent liabilities. It aims to ensure that new proposals are evaluated against a range of
criteria (rationale, exposure, risk and return, risk management and mitigation, and
affordability). This gives the Treasury the tools to monitor the Government’s exposure to new
contingent liabilities. We make use of the associated database at each EFO when we review
developments since our previous forecast and their broader fiscal implications.
6.31 The Government’s appetite for extending new guarantees related to housing has yet to be
sated. At Autumn Budget 2017, it announced a further £8 billion of guarantees to support
housebuilding. Of this:
• £1 billion were made available for the British Business Bank’s ENABLE guarantees
programme, which is aimed at small and medium-sized housebuilders; and
• £3 billion were allocated to an Affordable Homes Guarantee Scheme at Spring

Statement 2019, although no further details about this scheme are available yet.
6.32 In preparing this report, we asked the Treasury about how many contingent liabilities
entered into before the new approval regime took effect had crystallised and at what cost. It
was not able to provide complete information to answer this. This information gap could be
important, since it is the stock of all contingent liabilities, rather than just the flow of new
ones, that matters in terms of the Government’s fiscal risk exposure over time.
Conclusions
6.33 Asset sales continue to represent a risk to our medium-term PSND forecast, but this may
subside as the assets acquired during the financial crisis are sold. The new transparency
requirements around asset sales provide useful information that the Government could use
to ensure that asset sales do not worsen fiscal sustainability while improving PSND.
6.34 Risks in the medium term from loans have increased, in particular policy risks in respect of
student loans. We will analyse the impact of any new higher education financing plans in
the relevant EFO and in more detail in next year’s Fiscal sustainability report.
6.35 Risks from guarantees have also risen somewhat thanks to new announcements, but remain
low. The Government’s contingent liability approval regime should help to improve the
quality and management of new guarantees. But information about the stock of existing
contingent liabilities is less complete.

Balance sheet risks
Balance sheet transfers

6.36 In our previous report we drew attention to the risks that arise from the possibility of
reclassifications that expand the public sector balance sheet. In its response the Government
announced that the ONS had published a strategy that aimed to give greater transparency
and predictability to the pipeline of future reclassifications.6
6.37 The reclassification of bodies on or off the public sector balance sheet can happen because
the Government undertakes some action that, in the opinion of the ONS, causes the body to
be moved into or out of the public sector. Or it can be because the ONS reacts to changed
accounting guidance or to new information and retroactively reclassifies a body. On
reclassification the assets and liabilities of the body are added to or removed from the
public sector balance sheet, which can sometimes produce a large change in the size of the
public sector with no corresponding flow recorded in the deficit.
6.38 In the medium term reclassifications change the path of debt and, usually to a lesser extent,
the deficit. They can be highly distorting to the path of debt. This is why the Government
uses fiscal aggregates that exclude the public sector banks for its fiscal targets and why they
are the headline measures in the ONS public finances release and our forecasts.
6.39 In the longer term it is often difficult to judge what if any impact the reclassification has on
fiscal sustainability. For example, the ONS has included Network Rail both on and off the
balance sheet at various times yet this has made no apparent real-world difference to the
extent to which Government supports the rail industry or stands behind the debts of Network
Rail. However, as we discussed in our previous report, concerns about the classification
have influenced past policy decisions with some real-world effects on borrowing costs.
Updated risk assessment: statistical reclassification

6.40 We do not usually anticipate ONS decisions on the classification of bodies. But where it has
indicated a decision and we have a solid understanding of the eventual fiscal consequences
we may include them in our forecast before they are incorporated in the outturn statistics.
6.41 The ONS publishes a Forward work plan spreadsheet detailing future classification cases
and an annual Looking ahead article that gives a longer-term view of plans and potential
impacts. None of the bodies under review in its latest plans would make a large difference
to the balance sheet – indeed, all those listed in the latest ONS plan would, if reclassified,
have an impact assessed as ‘small’ or ‘medium’, defined as no bigger than £1 billion.
Funded public sector pension schemes

6.42 In September 2019 the ONS intends to include funded public sector pension schemes (and
their associated assets and liabilities) in the public sector. Any changes will be purely
6
Office for National Statistics, Looking ahead: developments in public sector finance statistics, July 2018.

Balance sheet risks
statistical: the change will neither increase nor reduce the Government’s exposure to the
schemes. The ONS has estimated that this change will reduce PSND by £31 billion at the
end of 2017-18 and PSNFL by £9 billion. This was not included in our most recent forecast
as estimates of its effect were not available at the time. We will reflect it in our next forecast.
6.43 The inclusion of funded pension schemes on the balance sheet highlights the discrepancy
between the treatment of funded and unfunded pension schemes in the public finances. The
liabilities of unfunded schemes are not reflected on any statistical measure of the balance
sheet and the treatment in the deficit also differs. The ONS will produce estimates of the
liabilities of these schemes (in tables compliant with the IMF’s Government Financial Statics
Manual), which we will use when assessing fiscal sustainability. The WGA includes the
discounted liabilities of unfunded schemes that have been accrued to date. At the end of
2017-18 the WGA estimated the net liabilities of unfunded pension schemes at £1.74
trillion, compared with those of funded schemes at £0.12 trillion.
Pension Protection Fund

6.44 Alongside the other changes to funded pensions schemes, the ONS will include the impacts
of the Pension Protection Fund (PPF). The PPF takes on defined benefit pension schemes that
have become insolvent. As well as a reduction in PSND (included in the estimates above),
the PPF introduces some volatility into the future path of debt and the deficit.
6.45 Income from the scheme (from its investments and a levy on eligible schemes) and
payments to pensioners are likely to be relatively stable, but as the PPF takes on schemes in
the future the deficit will be increased by the level of the shortfall in each one when it is
taken on. Most are small relative to the size of the economy, and the PPF takes on many
each year meaning it is only when a particularly large one is taken on that significant
volatility will be introduced into PSNB. PPF compensation is also less generous to scheme
members than that offered by the failed scheme, which reduces the PPF’s liabilities relative
to the scheme’s. In 2017-18 the PPF recorded £1.2 billion in deficits from new schemes
taken on, up £0.9 billion from the previous year. Much of the increase will have arisen from
the inclusion of the Carillion pension schemes that were estimated to be in deficit by £1
billion.7 The ONS has not released provisional estimates of the PSND impact of the PPF, but
the scale is likely to be relatively small.
Updated risk assessment: Government actions

6.46 While the ONS has worked to improve transparency on forthcoming statistically driven
reclassifications, it is rare that it or we have clarity about those driven by the Government’s
own actions. Unlike statistically-driven reclassifications, these also more often have real-
world consequences and are therefore more likely to affect fiscal sustainability.
7
Letter from Carillion (DB) Pension Trustee Limited to the Chair of the Work and Pensions Committee, 26 January 2018.

Balance sheet risks
Housing associations
6.47 The history of the classification position of housing associations is described in Box 6.1.
Unusually for a Government-driven reclassification there was clarity about the prospect of it.
The Government was candid that it was only legislating to relinquish controls over housing
associations to remove them from the balance sheet. As we concluded in our previous
report, in this case the movements also make no real-world difference to fiscal sustainability
– they have not reduced the chances of a housing association failing, nor, in such
circumstances, the probability of the Government stepping in to continue the provision of
affordable housing. In practice, the Government has simply moved housing associations
back into the realm of fiscal illusions – providing a public service off the balance sheet.
Implicit government liabilities

6.48 Other than housing associations, Government actions have not triggered reclassifications
since our 2017 report. When thinking about potential future actions it is worth considering
the industries in which the Government might choose to intervene. In our previous report we
considered two broad (and overlapping) categories: those industries vital for the economy
and those vital for the delivery of public services. For both categories the implicit recognition
of the liability is illustrated by the level of regulation to which they are subject.
6.49 In the event of another financial crisis the Government may feel forced to intervene because
the consequences of not doing so would be too severe. There are other areas where
governments might similarly feel compelled to act. The Centre for the Protection of National
Infrastructure lists 13 sectors that contain “Those critical elements of national infrastructure
(facilities, systems, sites, property, information, people, networks and processes), the loss or
compromise of which would result in major detrimental impact on the availability, delivery or
integrity of essential services, leading to severe economic or social consequences or to loss of
life”.8 These are: chemicals, civil nuclear, communications, defence, emergency services,
energy, finance, food, government, health, space, transport and water.
6.50 Other heavily regulated industries include housing associations and education, which the
Government substantially finances and relies on to deliver public services. The Office for
Students warns of three key uncertainties: Brexit, the Augar Review (see Box 6.2) and the
decline in the number of 18-year olds in the next few years, as well as pressures from
pensions and other costs.9 The sector’s student recruitment expectations look optimistic. The
OfS found that, in total, providers were aiming to increase full-time UK and EU students by
7 per cent over four years and overseas students by nearly 21 per cent. The OfS warns that
“a provider whose financial viability and sustainability is underpinned by reliance on fee
income based on student recruitment targets which prove to be unrealistic is exposing itself
to significant risk”. Our latest forecast assumes 0.5 per cent growth in student numbers in
the four years up to 2021-22. The number of 18-year olds is projected to fall 5.2 per cent.
Another risk could come from a growing reliance on non-EU students. We only forecast EU
entrants as they are eligible for student loans, but EU students comprise only a quarter of all
8
See Centre for the Protection of National Infrastructure, Critical National Infrastructure.
9
Office for Students, Financial sustainability of higher education providers in England, April 2019.

Balance sheet risks
non-UK students: China alone accounts for a higher proportion.10 Were the attractiveness of
the UK as a destination to change drastically, for example through changes in UK visa
requirements or due to policy changes in students’ origin countries, it would be very difficult
for some providers to make up the lost fee income quickly.
6.51 The chair of the OfS has stated that “the OfS will not bail out providers in financial
difficulty”. It requires all higher education providers to have student protection plans in
place to ensure continuity of study. But it is worth noting that these plans have yet to be
tested and government rarely commits to bailing out institutions in advance as it can
generate risky behaviour on the part of affected institutions. In the event, it might still choose
to rescue a failing institution or to provide assistance to its students when the time came.
Conclusions
6.52 The reclassification of housing associations produced a large statistical change but did not
materially affect fiscal sustainability. Otherwise, the Government’s commitment to support
industry remains strong, as evidenced by the support proffered to Nissan (see Chapter 5)
and extended to British Steel. Brexit may compel the Government to provide higher levels of
support but it is not clear whether that might be sufficient to trigger any reclassifications.
Valuation changes
6.53 The Government’s exposure to revaluation risks lies mostly with the official foreign exchange
reserves. The sterling value of the unhedged portion of these (around a third of the total)
fluctuates with exchange rates. In 2016-17 the large drop in sterling increased the value of
the reserves by £4.4 billion (4.0 per cent). Since then revaluations reduced their value by
£2.3 billion in 2017-18 and increased it by £2.9 billion in 2018-19. The possibility of future
large exchange rate movements – particularly in the context of Brexit – remains significant.
Fiscal illusions
6.54 In our previous report we highlighted risks arising from fiscal illusions – where accounting
treatments obscure the true fiscal position and where they drive policy decisions. In its
response the Government pointed to developments in several areas, including:
• the greater controls in place for contingent liabilities;
• active monitoring of guarantee exposures;
• the disclosure of more information on asset sales; and
• updated guidance for departments on financial transactions.11
10
Migration Advisory Committee, Impact of international students in the UK, September 2018.
11
These guidelines have now been incorporated in the Consolidated budgeting guidance (2019-20).

Balance sheet risks

6.55 Having adopted the IMF term ‘fiscal illusions’ in our previous report, we looked in detail at
the illusions generated by the current recording of student loans in a working paper last
year. The term has also been adopted more widely in the student loans debate and used in
both the House of Lords and Treasury Select Committee reports on the subject.
6.56 Within the OBR we now use the concept more systematically when thinking about instances
where the accounting treatment or presentation disguises the ‘true’ fiscal position and, in
particular, where the attractiveness of the illusion drives Government policy. For example, in
respect of housing associations in our November 2017 EFO and when discussing the
closure of the PF2 private finance scheme in our October 2018 EFO.
6.57 In this section, we discuss three general classes of fiscal illusion: off-balance sheet financing;
the use of financial transactions to replace spending; and valuation of assets and liabilities.
We note different types of illusion elsewhere in this report too. Earlier in this chapter we
discussed asset sales, where the Government emphasises how these reduce PSND but not
that they usually have minimal impacts on broader measures of the balance sheet. In
Chapter 4 we discuss the use of tax expenditures, which often perform a task that could also
be achieved via conventional public spending but do so in a less transparent way – with the
cost being revenue forgone rather than spending voted for in Parliament. And in Chapter 5
we note how shifting the burden of a welfare benefit – free TV licences for the over-75s – to
the BBC to reduce the deficit appears likely to have fiscally costly unintended consequences.
Off-balance sheet financing

6.58 The private finance initiative (PFI and, after 2012, PF2) is a long-term contractual
arrangement usually used for the construction and maintenance of an infrastructure asset. It
has attracted much criticism as a fiscal illusion as in the public sector finances most of the
contracts are off-balance sheet. This means that the construction costs of the assets are
recognised over a long period (often 25 to 30 years) rather than when they are incurred.
6.59 The Whole of Government Accounts recorded a capital value of £57 billion from 704
contracts at the end of 2017-18, down from £59 billion and 715 contracts the year before.
This reflects a lack of new contracts in recent years: there have only been six PF2 contracts
with a capital value of £900 million signed since 2012. At Budget 2018 the Government
announced that there would be no more use of PF2 contracts and the Government has now
closed the PF2 scheme to new projects, so the value will continue declining.12
6.60 Before reclassification the debt of housing associations recorded on the public sector
balance sheet was nearly £70 billion and they were borrowing an average of £3 billion a
year. The spending associated with this borrowing is no longer recorded in the deficit and
the debt is no longer in any balance sheet aggregates. But the fiscal risk it poses remains.
12
The Scottish and Welsh administrations have their own models of public-private partnership, which they may continue to use.

Balance sheet risks
Box 6.1: Housing associations, classification changes and fiscal risks

Successive reclassifications of housing associations into and then back out of the public sector
have generated large balance sheet transfers and affected our recent forecasts. The move back
into the private sector returns them to the status of a ‘fiscal illusion’, having briefly featured in the
full public finances data. The nature of the illusion stems from their use to deliver a public policy
objective and the related likelihood that the Government ultimately stands behind them.
The 2008 Housing and Regeneration Act gave the Government significant controls over housing
associations in England, but these powers were not used to a fiscally material extent until
Summer Budget 2015, when the Government announced that it would force housing
associations to cut social sector rents by 1 per cent a year for four years, thereby reducing the
housing benefit bill on their properties. In our accompanying Economic and fiscal outlook we
noted that this might prompt the ONS to reconsider their classification in the private sector. In
October 2015, the ONS reclassified English housing associations into the public sector, with
effect from 2008 when the relevant legislation had been enacted. It then reviewed the treatment
of associations in the rest of the UK and took the same decision for them in September 2016.a
We noted at the time that these changes raised PSND, but did not materially affect fiscal
sustainability.
Later, the Government took legislative steps to reduce its control over housing associations. It
was unusually candid in admitting that those steps were precisely calibrated to relinquish just
enough control to allow the ONS to reverse its decision, but no more.b On the back of these
regulatory changes, the ONS reclassified English housing associations back to the private sector
with effect from November 2017.c Scottish and Welsh associations followed a year later,
reflecting slower passage of relevant regulations through the Scottish Parliament and the Welsh
Assembly.
Chart A: Housing associations’ contribution to PSND
80
Northern Ireland
Wales
70
Scotland
England
60
UK
50
£ billion
40
30
20
10
0
2006-07 2007-08 2008-09 2009-10 2010-11 2011-12 2012-13 2013-14 2014-15 2015-16 2016-17 2017-18 2018-19
Source: ONS, OBR

Balance sheet risks
Our 2017 Fiscal risks report highlighted the potential reduction in PSND that would follow the
Government’s desired reclassification, were it to happen. But it also noted that the implications
of housing associations for fiscal sustainability would be little changed either way since their role
as providers of social housing would not change – and neither would the likelihood that the
Government would stand behind them were they to face financial difficulties in future.
This episode highlights the degree to which statistical boundaries can drive regulatory policy
decisions. This is probably not unusual, although this instance was unusual for the Government’s
candour about its motivation in designing the regulatory changes that were enacted.
a
See ONS, Classification announcement: “Private registered providers” of social housing in England, 30 October 2015.
b
Specifically it limited the percentage of officers that a local authority may nominate as board members of a housing association
and removed a local authority’s ability to hold voting rights. See The regulation of social housing (influence of local authorities)
(England) Regulations 2017. Its motivation was set out in evidence to the House of Lords Secondary Legislation Scrutiny Committee -
see 6th Report of Session 2017-19, October 2017.
c
ONS, Statement on classification of English housing associations, November 2017.
Financial transactions that replace spending

6.61 In our previous report we noted that there had been several policies in recent years that
replaced grant funding with loans. These included maintenance grants for students from
low-income families, nursing bursaries and the ‘support for mortgage interest’ benefit.
While these lowered the overall cost to government of providing that support, they pushed
write-off costs beyond the forecast horizon – hence generating a fiscal illusion in PSNB. No
further policies of this type have been announced over the past two years.
6.62 As discussed in our 2018 working paper,13 there are many fiscal illusions associated with
student loans as currently recorded, including the interest accrued, the balance sheet
valuation and the treatment on sale. But the fiscal illusion that may have seduced
government the most is the fact that moving from grant to loan financing allowed it to
increase higher education funding while reducing recorded spending and the deficit. As
discussed in Box 6.2 the ONS intends to remove these illusions by changing the accounting
treatment for student loans, with effect from September this year. In our March 2019
forecast we estimated that these changes would increase the recorded deficit by £10.5
billion in 2018-19 rising to £13.7 billion in 2023-24. Initial ONS estimates of the effect on
the PSNB data (of £10.6 billion in 2018-19) suggest these estimates are reasonable.
Valuation of assets and liabilities

6.63 The yield on gilts has fallen dramatically since the financial crisis and currently the real yield
on index-linked gilts is negative. As the Debt Management Office cannot issue index-linked
gilts with a negative real coupon, the coupon yield is greater than the market demands.
Consequently, at auction index-linked gilts are sold for considerably above ‘par’ – that is,
purchasers lend government more than £1 for each £1 they will receive at redemption. In
13
OBR, Working paper No. 12: Student loans and fiscal illusions, July 2018.

Balance sheet risks
our March forecast, based on market interest rate expectations at the time, we assumed the
Government would raise £140 in cash for each £100 of index-linked gilts issued.14
6.64 PSND (and other statistical balance sheet aggregates) record the liability from auctioning
debt at the face value of the debt rather than the cash raised. This has the counterintuitive
result that when government issues £100 of gilts to raise £140 of cash, PSND actually falls:
measured liabilities rise by £100 but measured assets rise by £140. In our March forecast
this ‘auction premia’ effect reduced PSND by £8.1 billion in 2019-20.
6.65 Auction premia unwind over the lifetime of the gilt, through the higher cash coupon
payments, but that can take several decades. Meanwhile those premia have a significant
impact on PSND. Since 2007-08 they have reduced debt by £97.5 billion (including £18.8
billion in 2016-17) and we forecast premia to average £7.3 billion a year over the forecast.
But the net impact of premia is decreasing: the stock effect is now unwinding by £4.6 billion
a year on average and this effect will get larger as long as premia at auction continue. This
unwinding is sufficient to keep the levels impact on PSND at around 3.5 per cent of GDP in
our latest forecast. These estimates are volatile, as relatively small changes in yields or the
choice of gilts auctioned can have large impacts: at the time of our previous report we
forecast auction premia of £12.7 billion for 2018-19; outturn was only £5.8 billion.
Conclusions
6.66 There have been some actions over the past two years that have reduced (or will reduce)
fiscal illusions. Most importantly, the ONS’s revised accounting treatment for student loans
will remove the most egregious illusions from the accounts. We will look carefully at any
new higher education financing arrangements to see if they introduce new illusions or
appear to exploit the new accounting treatment to disguise their true fiscal effects. The
Government’s decision to end PFI schemes means that this illusion will gradually unwind as
existing contracts mature – and it will no longer affect the flow of new spending decisions.
6.67 But the Government’s actions to secure the reclassification of housing associations show that
the incentive to exploit accounting and classification rules to keep activity off the balance
sheet remains. While the letter of the rules was observed, the spirit of fiscal transparency
was not, with a large quasi-public sector activity shifted back into the realm of fiscal illusion.
14
The equivalent mechanism plays out for conventional gilts, but the effect is currently much smaller than the effect on index-linked gilts.

Balance sheet risks
Box 6.2: Accounting treatment and policy developments affecting student loans
Student loans have become an increasingly important part of our fiscal forecasts, with gross
outlays reaching £18.1 billion (0.8 per cent of GDP) in 2018-19 and forecast to reach £22.6
billion (0.9 per cent) in 2023-24. Flows of this size would make student loans an important
source of medium-term risk at any time, but prospective changes to their treatment in the
National Accounts and potential future policy changes provide additional sources of risk.
Accounting treatment developments since our previous report

In our 2017 Fiscal risks report we discussed the fiscal illusions resulting from recording interest
income on student loans that would never be received as revenue. Subsequently, both the
Treasury Select Committee and the House of Lords Economic Affairs Committee have produced
reports calling for changes to the accounting treatment of student loans in the public finances.a
In July 2018 we issued a working paper discussing the various fiscal illusions associated with
student loans.b The ONS published an article alongside our report setting out potential
improvements.c The illusions to be addressed included those arising from significant write-offs of
unpaid loans expected at the end of their 30-year term and from the sale of loans at deep
discounts to their recorded value in the National Accounts without hitting borrowing.
In December 2018, after consulting international statistical authorities, the ONS published a
follow-up article laying out its plans.d Eurostat also published advice on the recording of income-
contingent loans.e The ONS proposed a new approach under which outlays on student loans
would be split (‘partitioned’) into a portion that was expected to be repaid and would therefore
be treated as a loan accruing interest, and a portion that was not expected to be repaid and
would therefore be written-off at the point of outlay and so recorded as expenditure.
We updated our fiscal forecasts to take account of the ONS’s plans in Annex B of our March
2019 EFO. In June 2019 the ONS then released an article detailing the methods it would use to
calculate the loan-partitioning and how it would approach revisions.f Provisional ONS estimates
of the impact of the new treatment suggest that our March estimates were reasonably accurate:
these ranged from an increase to PSNB in 2018-19 of £10.5 billion (ONS estimate £10.6
billion) rising to £13.6 billion in 2023-24. The ONS intends to introduce a new student loans
time series into the public finances in September this year.
Fiscal risks under the new accounting treatment
By removing fiscal illusions, the new approach constitutes a material improvement. When the
terms on the loans or the economic assumptions underpinning projected cash flows change, the
accounting treatment will henceforth correctly reflect the changes in the underlying cash flows.
(That is not the case with changes in interest rates charged, which will only affect cash flows for
the minority of borrowers that repay in full, but will affect upfront write-offs for the majority that
are not expected to do so.) This much closer alignment of the accounting treatment with
economic reality reduces perverse incentives and promotes fiscal sustainability, so reducing risks.
However, the new treatment increases uncertainty in the medium term by introducing multi-
decade projections into the calculation of outturn data, and thus introduces a new source of risk
to our forecast. The economic assumptions embodied in our student loans forecasts and long-

Balance sheet risks
term projections will be revised routinely at each forecast. Government has also periodically
changed the terms and conditions on the loans. Both affect the expected stream of interest
charged and repayments, which under the new treatment will change the estimated partition of
gross loan outlays into expenditure and loan portions.
We will show the impact of such changes on future loan outlays in our forecasts, regardless of
the underlying source of the change. But all changes to economic assumptions (such as
projections for earnings growth or RPI inflation) and many policy changes will also affect
estimates of how the existing stock of loans should be partitioned. The ONS will not revise
historical estimates of PSNB in the light of these changes, but could change the loan balance
and so PSNB in the year that policy changes come into effect. It has provided guidance about
how these changes to the loan balance should be treated and considered three general cases:
• Where changes derive from revisions to underlying economic assumptions, these will
generally be recorded as a revaluation on the balance sheet, with no associated flow
transaction recorded in PSNB.
• Policy changes “that significantly change the loan stock value through expectations of
future repayments” – i.e. policy changes that affect cash flows significantly – will be
recorded as a flow transaction affecting PSNB and the balance sheet in equal measure.
The ‘significant’ test is designed to avoid making large changes to PSNB due to changes
in terms and conditions that do not genuinely affect economic reality. But by creating a
dividing line in the accounting treatment, it might also generate opportunities for
policymakers to try to avoid their policy decisions affecting the deficit.
• Policy changes “that affect the stock value predominantly through a change in the
discounting factor” – i.e. policy changes that affect the interest rate charged, but not the
amounts repaid – will be recorded as a revaluation, with no associated PSNB transaction.
(Since lowering the interest rate charged reduces future write-offs, if a PSNB transaction
were recorded it would generate a large fiscal benefit in year one offset by decades of
lower interest receipts.) The ONS argues that this would “adversely affect the
interpretation of the fiscal aggregates”, hence the different treatment to other policy
changes. It also removes a potential fiscal illusion that policymakers might feel a strong
incentive to exploit.
The ONS article provides an illustrative scenario with the impact on a single cohort of plan 2
English loans issued in 2018-19 of a decision to link interest rates to CPI, rather than RPI,
inflation from 2024-25 onwards. This is the equivalent of reducing the interest rate charged by
around 1 percentage point. Under the new treatment, this change increases the estimate of the
loan balance for this cohort from £9.6 billion to £10.3 billion (out of total outlays of £16.3
billion).
We can use this example to give an idea of the scale of future revisions. If all future loans paid 1
percentage point less in interest – a little larger than the effect an RPI methodology change might
have (see Box 7.1) – it would reduce spending by a little less than £1 billion a year in our
forecast. If it applied to existing loans (and passed the ONS’s test as a policy change that
significantly affected future repayments) the impact might be around £5 billion in 2019-20.

Balance sheet risks
The Augar Review

In May 2019, an independent review chaired by Dr Philip Augar reported its findings.g The
review made recommendations across both further and higher education, but it is those relating
to higher education financing that would be most fiscally significant were they to be adopted.
The review recommended:
• a reduction (compared to current plans) on spending on higher education by freezing the
fee cap for a number of years and reducing eligibility for financing of ‘foundation years’;
• some giveaways to students, including replacement of part of the loan with a grant, the
reintroduction of means-tested maintenance grants, lower in-study interest rates and a
cap on total repayments; and
• some takeaways from students, including lower repayment and interest rate thresholds
and a longer payment duration so that outstanding amounts would be written off later.
The review suggested that, (under the proposed new accounting treatment), the further and
higher education recommendations would together raise borrowing by £1.2 billion to £1.5
billion in 2024-25. This reflected a mixture of proposals that would increase and reduce
spending. The departing Prime Minister welcomed the recommendations but also noted that
funding decisions needed to be “taken in the round”.h This highlights the most important
consequence of the ONS accounting treatment change: in the future, higher education financing
will have to compete with other tax and spending policies, rather than appearing to be a ‘free
lunch’ as at present.
a
House of Commons Treasury Committee, Student Loans Seventh Report of Session 2017-19, February 2018 and House of Lords
Economic Affairs Committee, Treating Students Fairly: The Economics of Post-School Education Second Report of Session 2017-19,
June 2018.
b
OBR, Working paper 12: student loans and fiscal illusions, July 2018.
c
ONS, Looking ahead: developments in public sector finance statistics, July 2018.
d
ONS, New treatment of student loans in the public sector finances and national accounts, December 2018.
e
Eurostat, Ex ante advice on the recording of the UK student loans in Government Finance Statistics, December 2018.
f
ONS, Student loans in the public sector finances: a methodological guide, June 2019.
g
Review of post-18 Education and Funding, May 2019.
h
PM speech at Augar Review launch, May 2019
Intangible assets
Intangible assets and the public sector
6.68 Intangible assets are those that are neither physical nor financial in nature. This definition
covers a broad spectrum: from software to patents, research and development (R&D), data,
training and know-how. In a recent book, Jonathan Haskell and Stian Westlake argue that
four common characteristics tend to differentiate intangible assets from other ones:15
• Scalability: once the asset is created or acquired, it can be made use of multiple times
at relatively little cost (due to both ‘non-rivalry’ and ‘network effects’).
• Sunkenness: it is harder to recoup the cost of an investment in an intangible asset by

selling the asset, so there are irrecoverable or ‘sunk’ costs.
15
Haskell, J. and Westlake, S., Capitalism without Capital: The Rise of the Intangible Economy, 2017.

Balance sheet risks
• Spillovers: once an intangible investment has been made, it is relatively easy for others
to take advantage of it (and relatively hard to prevent them from doing so).
• Synergies: the value of intangibles is greater when combined with other assets.
These features will generally mean that narrow financial returns to the creator or owner of
intangible assets will be exceeded by broader economic and social returns on those assets.
6.69 Research has traditionally focused on the role of physical capital in the production process,
but the impact that the properties of intangible assets have on productivity, competitiveness
and growth has been receiving growing attention (although this research tends to focus on
the private sector). For instance, a study by Corrado et al shows the importance of
intangible capital deepening in accounting for recent growth trends.16 And, at the micro
level, one by Hall et al demonstrates a strong link between firms’ market valuations and the
number of well-cited patents they hold.17
6.70 Given the scalability, spillovers and synergies present, the public and private sectors might
be expected to take different approaches to intangible investment.18 In particular, the private
sector can be expected to focus narrowly on private financial returns, while the public sector
should take broader factors into account. Indeed, the Treasury’s ‘Green Book’ guidance to
departments is clear that “all impacts – social, environmental, financial etc. – [are] to be
assessed” in cost-benefit analyses carried out by the public sector.
The scale of public sector intangible assets

6.71 A comprehensive valuation of the intangible assets owned by the public sector would need
to consider all the ways these assets affect society. But even estimating only the financial
value of intangibles is difficult. Accounting standards often recognise only the value of
individual intangible assets, and then only if relatively stringent criteria are met, such as the
existence of reliable cost estimates. And, unlike listed private companies, governments do
not have a market-determined equity value, so the total financial value of intangibles
cannot be estimated by subtracting the value of tangible assets from the value of equity.
6.72 The value of the public sector’s intangible assets is thus likely to be substantially
underestimated in the official accounts. For instance, the Whole of Government Accounts
(WGA) is prepared under international financial reporting standards (IFRS), which do not
allow many types of intangible assets to be capitalised and recognised therein. Indeed,
intangible assets represent only 2 per cent of total public sector assets in the WGA,
compared with global private sector estimates of 52 per cent to 84 per cent.19 At end-March
2018, this amounted to £36 billion of intangibles, mostly owned by the Ministry of Defence,
like ‘military equipment’ (£24 billion) and ‘capitalised development expenditure’ (£6 billion).
16
Corrado, C., et al., Intangible investment in the EU and US before and since the Great Recession and its contribution to productivity
growth, 2016.
17
Hall, B., A. Jaffe, and M. Trajtenberg, Market value and patent citations, 2005.
18
Here, as elsewhere in this section, we focus on the public sector’s intangible knowledge assets (this includes assets created by human
intellect but excludes any value derivable from using the government’s sovereign powers, such as the ability to levy taxes).
19
See Ocean Tomo, Intangible Assets Market Value Study, 2017, and Chartered Institute of Management Accountants, Global Intangible
Financial Tracker 2015, 2015.

Balance sheet risks
6.73 Other approaches generate higher – and probably more realistic – estimates of the value of
intangibles, such as that adopted by the SPINTAN (Smart Public Intangible) project, which
looks at the capitalised net present value of past investment in activities such as research,
software and organisational capital. This finds that public sector intangible assets in the UK
amounted to £150 billion in 2015,20 around 8 per cent of total public assets.21 Moreover,
this costs-of-production approach is still likely to underestimate the true value of intangibles,
as it does not take account of the potential streams of financial revenue they might generate
or any wider economic and social returns from spillovers to other parts of the private sector
or synergies with the rest of the capital stock.
6.74 Chart 6.6 shows the increase in public sector intangibles in the UK, as recorded by the
SPINTAN project, over the past two decades. Training activities and R&D accounted for over
80 per cent of total intangibles in 2015. The largest increase has been in training capital,
whose share rose by almost a quarter over two decades. This represents the cumulative cost
of training public sector employees – teachers, nurses, civil servants and so on. (In contrast,
in the private sector in the UK training accounts for a much lower 22 per cent of the total
investment in intangibles, with organisational capital, software and R&D forming the bulk of
the remainder.22)
Chart 6.6: Intangible assets in the UK public sector by type

9
Organisational capital
Design
8
Advertising
7
Market research
Training
6 R&D
Software
0
1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 2015
Source: SPINTAN
6.75 The scale and composition of public sector intangibles also varies significantly across
countries, as shown in Chart 6.7. Of the 21 advanced economies covered by the SPINTAN
dataset for 2015, the UK ranked fifth in terms of the total stock (at 8 per cent of GDP), with
only the US, Sweden, Austria and Finland ranked higher. The UK ranked top in terms of
20
See SPINTAN Project, SPINTAN database on intangibles in the Public Sector (1995-2012) and Real Time data (2013-2015).
21
Different sectoral coverage also partly accounts for this difference: as well as adopting a broader accounting definition of intangible
assets, SPINTAN includes some non-profit entities, such as charities, that WGA does not capture as they are not part of the public sector.
22
See ONS, Experimental estimates of investment in intangible assets in the UK, 2015.

Balance sheet risks
training assets, but only seventeenth in terms of R&D assets, which account for only a
quarter of the UK stock compared with around 70 per cent in the US and Sweden. Different
asset types are likely to generate different financial, economic and social returns.
Chart 6.7: Public sector intangible assets: international comparison

14
Other R&D Training Total assets
12
10
0
SE US AT FI UK IT BE IE CZ DK DE NL PT EL HU FR SI SK PL ES LU
Source: SPINTAN
6.76 There are no estimates of the financial, economic or social values of public sector
intangibles in the UK. The Government reviewed its management of intangible (or
‘knowledge’) assets alongside Budget 2018, recognising that they were often not well
understood. Through the ‘Balance Sheet Review’ (see Box 1.1 in Chapter 1), the
Government plans to do more to measure and manage its intangible and other assets. The
Treasury’s 2018 review of intangibles suggest using new measurement and accounting
approaches, and establishing a central repository for tracking the value of intangibles.23
6.77 As part of the exercise, the Treasury is looking at how to ”improve the management of public
sector knowledge assets, and generate greater social, economic and financial value”.24 Any
attempt to do so could bring with it risks for the public finances, as we discuss next.
What type of fiscal risks might arise from intangible assets?

6.78 The Treasury’s review noted that policymakers face trade-offs between maximising the
revenue stream an asset can generate directly and the public benefit that derives from its
broader economic and social returns. It recognised that alongside the financial value that
could be obtained by the sale or licence of intangible assets (or other exploitation for a
commercial return), intangible assets also held social value (for example by supporting the
provision of public services) and – more relevantly for any assessment of fiscal risks –
economic value (by stimulating innovation and competition in the private sector).
23
HM Treasury, Getting smart about intellectual property and other intangibles in the public sector: Budget 2018, October 2018.
24
HM Treasury, Knowledge asset implementation study: terms of reference, December 2018.

Balance sheet risks
6.79 Potential fiscal risks from the exploitation of intangible assets can also be grouped this way:
• financial returns could be improved as a result of increased government effort; and
• economic returns could be affected positively or negatively in the process of seeking

higher financial returns (generating indirect effects on public finances via tax bases).
6.80 Many of the Treasury review’s recommendations have a financial focus. For instance, it
suggests registering the intellectual property assets with the most commercial potential and
further exploiting the data owned by the public sector. Clearly, the direct impact of this
constitutes an upside risk to the public finances, but the potential indirect implications could
work in either direction. For example, if the Government were to develop a new technology,
it could sell it for profit or alternatively share it freely (for some highly scalable assets like
data, this corresponds to ‘selling’ at marginal cost). Doing the latter removes any potential
financial returns but maximises economic and social impacts because anyone who values
the new technology at all would gain relative to the price they have to pay to access it.
Alternatively, the act of trying to commercialise returns from a public sector intangible asset
might in itself generate positive wider economic and social returns. Estimating the net fiscal
implications of either approach would, of course, be hard.
6.81 One relevant example is CT scanner technology in the 1960s and 1970s, which the
Department of Health and Social Security and the NHS played a key role in developing. If
the public sector had sought to focus narrowly on the financial returns, the economy might
have benefitted less from its further development by commercial players and society might
have benefitted less from the health gains associated with its widespread application. More
recently, a Deloitte study found that giving app developers free access to transport data had
generated £90 million to £130 million worth of benefits, largely from time-savings for
passengers.25
6.82 Distributing intangible assets at marginal cost will not always be appropriate. In July 2017,
the Information Commissioner’s Office found the Royal Free Hospital in breach of the Data
Protection Act for giving data on 1.6 million patients to the Google subsidiary DeepMind. As
part of its National Data Strategy the Government aims to explore how to “unlock the power
of data across government” while “building citizen trust in its use”.26 This may be challenging
– a recent National Audit Office report criticised the Government’s existing capabilities,
noting a “culture of tolerating and working around poor data”.27
6.83 Personal data is a significant source of economic and social value to both businesses and
consumers, with the latter benefitting from personalised goods and services. Personal data
can reveal new findings or insights if aggregated or merged with other personal
information. In the private sector, tech giants such as Google or Facebook have shown that
they can successfully monetise the data they hold on individuals, while the Government is
still in the process of recognising the potential value of the data it collects. The Government
25
See HM Treasury, The economic value of data: discussion paper, August 2018.
26
See Department for Digital, Culture, Media & Sport, National Data Strategy open call for evidence, June 2019.
27
See NAO, Challenges In using data across government, June 2019.

Balance sheet risks
can reap a financial (and wider economic) benefit by selling it in anonymised form to
private companies, but at the same time it could open itself up to legal, fiscal and
reputational risks if it were mismanaged or misused.
How large might those fiscal risks be?

6.84 Given the uncertain scale of financial returns and the uncertain direction and scale of
economic returns, attempting to quantify the potential fiscal risk from attempts to exploit the
existing stock of intangible assets more effectively would be extremely challenging.
6.85 The most comprehensive estimate of the UK public sector stock of intangibles is the cost-
generated one presented by SPINTAN. The Treasury review provided an illustrative
calculation based on the SPINTAN stock, noting that achieving a 3.5 per cent return on it
would generate £5 billion a year in revenue. That would be equivalent to 0.7 per cent of
2018-19 current receipts. But, as described above, half those assets reflect training of the
public sector workforce, which it is hard to see generating additional financial returns
beyond those already embodied in the productivity with which they deliver public services.
6.86 In the process of seeking higher financial returns, economic returns could be affected
positively or negatively. Quantifying potential wider consequences for the economy and
their indirect fiscal implications is not straightforward. It would need to factor in the many
knock-on effects across the economy, not just those on the specific activity affected by use of
public sector intangibles. The scale of the resulting indirect effects on public finances is
therefore even more uncertain than the scale of any financial returns.
6.87 As with many other government policies and activities that affect the economy’s capacity to
grow, we would typically wait to see evidence of any effect in the data rather than
anticipating it in our forecasts. It does seem likely that attempts to secure greater financial
returns to government from deploying an intangible asset will come at some cost in terms of
foregone social (and probably economic) returns. How that might translate into risks to our
medium-term forecast – which makes no explicit assumptions in this area – is unknowable.
Conclusions
6.88 The post-crisis deterioration in the balance sheet has halted, with debt forecast to fall on all
National Accounts measures. But returning to pre-crisis levels at the currently projected rate
would take decades. This is broadly as we had anticipated at the time of our 2017 report.
Our assessment of balance sheet risks is similar to that from two years ago: the greatest risk
remains the possibility that the Government will feel compelled to respond to some
unexpected future event or crisis, triggering a step change in the level of public debt.
6.89 But significant risks still remain from smaller – and more predictable – developments. Some
will genuinely affect long-term sustainability, such as potential changes to the financing of
higher education. Others will have purely statistical impacts, such as moving certain pension
schemes’ balance sheets inside the public sector boundary and the improved way of
accounting for student loans in the public finances.

Balance sheet risks
6.90 Fiscal illusions and the incentives they create for policymakers remain a risk. Changes to the
accounting treatment of student loans will remove a particularly obvious illusion, but the
steps taken by the Government to secure the reclassification of housing associations to the
private sector reinstates another and illustrated policymakers responding to incentives.
6.91 Other changes that are likely to mitigate fiscal risks over time include the steps taken by the
Government, highlighted in its MFR report, where more progress was made in respect of
managing balance sheet than in other areas. (This may be partly because the issues we
raised were more focused on transparency and management than on difficult policy
questions.) Concrete steps have been taken to improve how the public sector’s assets and
liabilities are recorded or understood, which should improve the Government’s
management of them and facilitate our scrutiny of them. We will survey any examples of this
greater transparency affecting decision-making in the next edition of this report.
For the Government response

6.92 In this chapter, we have highlighted several issues that the Government is likely to wish to
consider when managing its fiscal risks. Most raised in our previous report remain relevant:
• The deterioration in broad measures of public sector net worth since the crisis;
• Asset sales that could be delayed or raise less than expected;
• Asset sales that have not been factored into current forecasts;
• The possibility of reclassifications that expand the public sector balance sheet;
• The growing use of guarantees in infrastructure and housing; and
• The impact of fiscal illusions where accounting rules drive policy decisions.
6.93 We have also highlighted several new issues:
• The use of PSND as a fiscal sustainability metric and its suitability for this task;
• The management and oversight of the stock of existing contingent liabilities;
• How the commendable improvements in balance sheet transparency affect decisions;
• Statistical classification driving regulatory policy in respect of housing associations; and
• Trade-offs between exploiting intangible assets’ narrow financial and other impacts.
6.94 When assessing the outlook for the public sector balance sheet over the medium and long
term, does the Government regard these or other issues as important for its risk

7 Debt interest risks
Introduction
7.1 Debt interest is one of the largest elements of public spending not under the direct control of
government. It is determined by the stock of debt – mostly the legacy of past budget deficits
– and the rates of interest that the government must pay on it.
7.2 Earlier chapters looked at risks that could raise future deficits, or the debt stock directly, both
of which would increase debt interest spending. But increases in the cost of new borrowing
are an important additional risk, not just because they would make it more expensive to
service a given debt, but also because they could push the debt-to-GDP ratio towards an
unsustainable trajectory if they were to rise relative to the rate of growth of nominal GDP.
7.3 The public sector paid £38.5 billion (1.8 per cent of GDP) of debt interest to the private and
overseas sectors in 2018-19, with £37.5 billion paid by central government (net of the APF),
£0.4 billion by public corporations and £0.6 billion by local authorities. The public sector, in
its turn, received £8.1 billion of interest payments from the private and overseas sectors,
including accrued interest on student loans and interest on its foreign exchange reserves.
Chart 7.1: Total debt interest spending by government sector

6
Other public corporations Forecast
Local authorities
5
Central government net of the APF
4
Per cent of GDP
0
1973-74 1978-79 1983-84 1988-89 1993-94 1998-99 2003-04 2008-09 2013-14 2018-19
Source: ONS, OBR

Debt interest risks
7.4 Most outstanding public debt in the UK is the liability of central government. So in this
chapter we focus on risks to interest spending on central government gross debt (bearing in
mind that some factors we identify would have partly offsetting effects on interest receipts).
An important complication is that the Bank of England – also part of the public sector – has
bought a substantial quantity of central government debt, financed by the creation of
reserves on which it currently pays an interest rate of just 0.75 per cent – the Bank Rate set
by the Bank’s Monetary Policy Committee (MPC).1 In effect, this has allowed the
Government to refinance some of its past fixed interest borrowing at a lower floating rate,
reducing interest payments for now but leaving it more exposed to the risk of higher debt
servicing costs if the MPC were to decide to raise Bank Rate in the future.
7.5 When considering interest rate risks to fiscal sustainability, it is important to do so relative to
growth rate risks. Changes in the debt-to-GDP ratio depend, among other things, on the
relationship between the effective interest rate on the debt stock and the rate of nominal
GDP growth – increases in the former raise it and in the latter lower it. The difference
between them is known as the ‘growth-corrected interest rate’ (sometimes simply referred to
as ‘R-G’). When the effective interest rate and growth rate are affected to the same extent,
the growth-corrected interest rate is unchanged, with little implication for fiscal sustainability.
But shocks that raise the effective interest rate relative to GDP growth increase spending and
debt faster than GDP, threatening fiscal sustainability.
7.6 This chapter updates those risks identified in our 2017 Fiscal risks report (FRR) relating to:
• the debt stock and sensitivity to inflation and interest rate risk; and
• the impact of the Bank’s Asset Purchase Facility on effective interest rates.
We also take a deeper look at the influence of the growth-corrected interest rate on fiscal
sustainability and risks to the favourable path assumed in our latest central forecast, before
offering some conclusions and raising some issues for the Government’s next response.
Debt stock and sensitivity to inflation and interest rate risk

7.7 In our 2017 FRR, we outlined the risks that arise from:
• the rise in the debt stock and the issuance of index-linked gilts in recent years; and
• the increased sensitivity of debt interest spending to inflation and interest rate risk.
7.8 In Managing fiscal risks (MFR), the Government’s response to our 2017 report, it noted that:
1
We discussed the accounting treatment of this (and the Bank’s other unconventional monetary policies) in a recent explainer: OBR, The
direct fiscal consequences of unconventional monetary policies, March 2019.

Debt interest risks
• debt was forecast to fall as a percentage of GDP over the medium term; and
• the Government was reducing its inflation exposure by progressively reducing the
proportion of index-linked gilt issuance in its overall financing plans.

7.9 In the medium term, risks to our debt interest forecast relate to the level of new debt issued,
the interest rate charged on that debt and the rate of inflation acting on index-linked debt.
7.10 The level of new debt issued reflects a combination of the cash deficit and the size and
structure of existing debt. All else equal, medium-term debt interest risks will be lower when
the initial stock of debt is lower, when that debt is of longer average maturity, and when
more of the debt is at fixed interest rates rather than floating or linked to inflation.
The overall level of debt

7.11 Table 7.1 shows our most recent forecast for general government gross debt (GGGD)
relative to the size of the economy, and compares it with the March 2017 forecast that
formed the basis of our 2017 FRR. Debt has been revised down in all years, but by
diminishing amounts from 2019-20 onwards. All else equal, this has reduced debt interest
risks. But the changes are small and debt remains at more than twice its pre-crisis level.
Table 7.1: General government gross debt forecasts

Per cent of GDP
2017-18 2018-19 2019-20 2020-21 2021-22 2022-23 2023-24
March 2017 EFO 87.7 87.7 86.5 84.8 83.6
March 2019 EFO 85.3 85.5 83.8 82.9 82.2 81.1 80.0
Change -2.4 -2.2 -2.8 -2.0 -1.5
Maturity structure of debt

7.12 Most government debt is issued in the form of gilts, Treasury bills (T-bills) or NS&I products
(such as premium bonds). At the end of 2018-19, these accounted for 94 per cent of
GGGD. The average maturity of the debt stock is around twice that of most G7 economies,
at around 15 years. This reduces medium-term risks to our debt interest forecasts, because
only around 40 per cent of existing debt will be refinanced within a five-year forecast
period. As yield curves generally slope upwards (which means that longer maturity debt
pays higher rates of interest) this lower risk is paid for in higher annual financing costs.
7.13 Even though some debt will not be redeemed within the forecast horizon, significant
amounts will be – and some is issued at rates that can change during that period. Roughly
20 per cent of debt will redeem in 2019-20 or is otherwise subject to interest rate changes
(abstracting from the gilts held by the APF and index-linked gilts, discussed later in the
chapter). This is up slightly from the position ahead of our previous report (18 per cent in
2017-18), as more gilts are coming up for redemption and there has been a big increase in
NS&I financing through floating interest rate products. Over our five-year forecast, 39 per
cent of the stock will be subject to rate changes, unchanged from our previous report.

Debt interest risks
Market interest rates

7.14 We derive our medium-term interest rate forecasts from market expectations embodied in
financial market instruments. Changes in these expectations pose risks to our debt interest
forecast. Chart 7.2 shows the changes in expectations for gilt yields and Bank Rate (the most
important two interest rate assumptions in our forecast) at each of our past five forecasts.
We have revised our gilt rate forecast down three times and up once, while the Bank Rate
forecast has been revised down twice and up twice. The average absolute revision to each
has been 0.18 percentage points across the period, relatively small in historical context. But
it has still yielded some relatively large changes in our debt interest forecasts – adding £1.0
billion a year on average in our March 2018 forecast and subtracting £1.5 billion a year in
March 2019. As of 5 July, market expectations for both Bank Rate and gilt yields had fallen
significantly relative to our March 2019 forecast.
Chart 7.2: Successive forecasts for Bank Rate and gilt rates
2.5 2.5
Bank Rate Gilt rate
2.0 2.0
1.5 1.5
Per cent
Per cent
1.0 1.0
March 2017
Successive forecasts
0.5 0.5
March 2019
5 July 2019
0.0 0.0
2017-18 2019-20 2021-22 2023-24 2017-18 2019-20 2021-22 2023-24
Note: Gilt rates are the weighted average interest rate on conventional gilts.
Source: OBR
7.15 In our 2017 FRR we judged that the risks to our forecast for interest rates were two-sided,
but that with interest rates still close to historic lows the risks to spending from rates rising
above our central forecast were likely to be more significant. Over the past two years, our
gilt rate forecast has fallen on average, and rates have fallen yet further since then, but
large increases still seem more likely than large falls over the longer term. Factors that could
drive UK interest rates higher include faster than expected increases in global interest rates
or increases in the UK risk premium demanded by investors.
7.16 We also discussed how revisions to debt interest forecasts had most often partially offset
revisions to receipts forecasts. And that this was unsurprising as market expectations of
future interest rates tend to rise/fall when expectations of GDP growth are raised/lowered.
7.17 Up to our 2017 FRR, revisions to debt interest and receipts had contributed in the same
direction to borrowing revisions in only three of our preceding 15 forecasts. But in two of the
four forecasts since then, receipts have been revised up but debt interest revised down. The
latter is likely to reflect the market pricing in some probability of a no-deal Brexit and an

Debt interest risks
associated monetary policy easing. So to some extent it is inconsistent with the assumption
of a smooth exit that underpins our economy and receipts forecasts. If a smooth Brexit is
achieved, market interest rates – and our debt interest forecast – could rise again.
Index-linked gilts
7.18 Index-linked bonds are one of the tools governments use to manage the risks faced by
society. Historically, financial markets have struggled to meet investors’ demand for safe
assets protected against inflation (which arises in part because of requirements on defined
benefit pension schemes to uprate payouts in line with inflation). By issuing index-linked
debt, the Government offers the private sector this protection in exchange for lower yields.
7.19 Today, the UK issues a greater share of its debt as inflation-linked securities than any other
G7 government. At the end of 2018-19, index-linked gilts (ILGs) amounted to 29 per cent
of all gilts (and 38 per cent if Bank of England holdings are excluded, as none have been
bought under quantitative easing). The return on ILGs is linked to the Retail Prices Index
(RPI) and uncertainties over the path of RPI inflation are a significant risk to our medium-
term forecasts. But as we discuss in Box 7.1, this risk reflects not just the usual risks around
price movements but also potential changes to the way the RPI is constructed.
Box 7.1: How a change to the RPI would affect our fiscal forecasts
The RPI is a long-standing measure of inflation that is used to uprate a variety of payments and
taxes, notably including those related to index-linked gilts (ILGs). In 2010, a change in the
collection method for clothing prices resulted in a significant increase in the gap between the RPI
and CPI measures of inflation, prompting a recognition that the RPI embodied an unsuitable
formula for aggregating individual price quotes.a Over the past four years, the ONS estimates
that the effect of the unsuitable formula has raised annual RPI inflation by an average of 0.7
percentage points. This box looks at the potential consequences for our fiscal forecast were the
RPI to be changed in a way that removed this formula effect and thus narrowed the gap with CPI
inflation.
The stock of index-linked gilts stood at £429 billion in 2018-19 (19.8 per cent of GDP). Both
coupons and the value of the principal due for ultimate repayment rise with RPI. Both elements
are accrued each year, though investors only receive the principal uplift on redemption. This
means RPI inflation tends to be the most important near-term influence on our accrued debt
interest forecast. What would a change to the formula for RPI mean for the holders of ILGs? The
ready reckoners published alongside our most recent forecast imply that if RPI inflation were 0.7
percentage points a year lower – equivalent to the effect of removing the formula effect – debt
interest spending in 2019-20 would be £3.1 billion (0.1 per cent of GDP) lower. The effect
builds somewhat each year, reaching £4.4 billion (0.2 per cent of GDP) in 2023-24.
Assessing the fiscal implications of such a change is, however, further complicated in that three
of the oldest ILGs outstanding contain a clause requiring the Treasury to offer to redeem them at
(uplifted) par in the event of a change to the RPI that the Bank of England judges to be
“fundamental… and materially detrimental” to bondholders. Given current market prices are far

Debt interest risks
above uplifted par for these bonds, it seems unlikely that many bondholders would exercise this
option. In sum, were such a change to go ahead, it would be likely to lower debt interest
spending by progressively larger amounts over time. The large reduction in the overall return to
investors over the lifetime of each ILG would generate a large drop in market prices for those
bonds, with potentially significant implications for existing investors’ balance sheets. The change
could also prompt wider market instability were it to be seen as a breach of trust, though as the
Government would simply be curtailing an unintended windfall that seems unlikely.
A change to the RPI would have several other effects on our fiscal forecast. Many payments to
the Government and other regulated prices are uprated using the RPI – including the annual
revalorisation of various excise duties and increases in regulated rail fares – and the RPI is also
used in the calculation of the interest accrued on most student loans. The business rates
multiplier was switched from RPI to CPI inflation uprating in April 2018. In contrast, many
payments from the Government to the public, other than those on ILGs, are uprated with CPI –
including many welfare payments, personal tax credits, and, since May 2019, NS&I index-linked
savings certificates. Income tax and NICs thresholds are also linked to CPI inflation.
Looking just at ILGs, excise duties and accrued interest on student loans (under the current
accounting treatment) – the largest elements of the public finances still linked to the RPI – the net
effect of a 0.7 percentage point a year drop in RPI inflation relative to CPI inflation would be to
reduce borrowing by £2.3 billion in 2020-21, falling slightly to £1.9 billion in 2023-24.
In 2013, the UK Statistics Authority (UKSA) acknowledged the problems with the RPI by
withdrawing its status as a National Statistic, though it continues to publish it. The RPI is also
unusual in that it requires the Chancellor’s consent before any major changes can be
implemented, rather than changes being the sole domain of statisticians.b The wider implications
for ILG holders of addressing the shortcomings of RPI represent a disincentive to change. In light
of the continued lack of action in addressing the problems with the RPI, the House of Lords
Economic Affairs Committee recently concluded that UKSA should request “a fix to the clothing
problem” and that the “Chancellor should approve this change regardless of the effects on index-
linked gilt holders”.c That may hasten a methodological change in the RPI.
a
The CPI largely uses the ‘Jevons’ and ‘Dutot’ methods to aggregate prices at the most basic level, while the RPI largely uses the
‘Carli’ and ‘Dutot’ methods. The Carli method tends to inject spurious inflation into the index and is therefore prohibited by Eurostat
regulations except in ‘exceptional cases’. The ONS provides more detail in: ONS, Shortcomings of the Retail Prices Index as a
measure of inflation, March 2018.
b
See The Statistics and Registration Service Act 2007, 2007.
c
House of Lords Economic Affairs Committee, 5th Report of Session 2017–19 HL Paper 246: Measuring Inflation, January 2019.
7.20 Our previous report pointed out that the relatively high issuance of ILGs and their relatively
long maturities meant that in our March 2017 forecast the stock of these gilts continued to
rise, from 20 per cent of GDP in 2016-17 to 24 per cent in 2021-22, despite total debt
falling relative to GDP. This would further increase the sensitivity of accrued debt interest
spending and the deficit to RPI inflation.
7.21 Partly in response, in March 2018 the Government announced a 1 to 2 percentage point
reduction in ILG issuance in 2018-19.2 In October 2018, it then announced that it would
2
HM Treasury, Debt Management Report, March 2018.

Debt interest risks
“look to reduce index-linked gilt issuance in a measured fashion as a share of total issuance
over the medium-term, in line with this planned reduction”.3 Our latest forecast assumes the
mid-point of this range – a 1.5 percentage point a year decline. This takes ILG issuance
down from around 20 per cent of gross gilt issuance in 2018-19 to less than 15 per cent in
2023-24. As a result, the total stock of ILGs is now forecast to peak at 20.4 per cent of GDP
in 2021-22 before falling to 19.3 per cent of GDP in 2023-24. The Government has not
indicated whether the proportion of ILG issuance will continue to decline beyond 2023-24.
7.22 In MFR, the Government stated that it kept the potential issuance of new debt instruments
under review and discussed a previous consultation on CPI-linked gilts (unlike the RPI, the
CPI is a National Statistic that meets international statistical standards). The Government has
not announced any plans to consult again on CPI-linked gilts.
Sensitivity analysis
7.23 Chart 7.3 shows the debt interest ready reckoners we published alongside our March 2017
and March 2019 forecasts. With debt stocks and maturities little changed it is not surprising
that the sensitivity to changes in interest rates and the net cash requirement are little
changed. The sensitivity to inflation has fallen slightly reflecting lower issuance of ILGs. This
reduced sensitivity will only become significant over time if the stock of ILGs decreases.
Chart 7.3: Debt interest ready reckoners

0.30
March 2017 forecast
Effect on debt interest spending in final year of forecast
March 2019 forecast

0.25
0.20
(per cent of GDP)
0.15
0.10
0.05
0.00
1 percentage point 1 percentage point 1 percentage point £5bn increase
increase in gilt rates increase in short rates increase in inflation in CGNCR
Note: All increases are assumed to take effect at the beginning of the first year of the forecast and continue until the final year.
Source: OBR
3
See Annex A of HM Treasury, Budget 2018, October 2018.

Debt interest risks
Conclusions
7.24 The level of debt remains around twice its pre-crisis level, leaving the Government more
exposed to changes in the effective interest rate than it was then. And debt servicing costs
remain more sensitive to changes in interest rates and inflation than they were prior to the
crisis, due to the debt stock’s shorter maturity and a higher proportion of index-linked gilts.
7.25 In many respects the picture is little changed from our previous FRR, as might be expected
with risks to a stock that is typically slow-moving. However, the reduction in ILG issuance
has reduced the sensitivity of future spending to RPI inflation somewhat (and potential
changes to the RPI itself might further reduce the Government’s exposure to inflation). But
market participants appear to be pricing some chance of a no-deal Brexit and the likely
monetary policy response, so a smooth outcome could see rates rise again. And with
interest rates remaining at historic lows, and having fallen further since our March forecast,
over a medium-term horizon interest rates seem more likely to overshoot our forecast
materially than they are to undershoot it significantly.
The Asset Purchase Facility and effective interest rates

7.26 In our previous report, we noted that the Asset Purchase Facility (APF) lowers government
borrowing costs, but that shortens the average maturity of the debt stock at the public sector
level and increases exposure to changes in short-term interest rates.
7.27 In its response, the Government pointed to the oversight and governance arrangements it
had in place for the APF. Looking ahead to a world in which the MPC decides to unwind
quantitative easing by selling the gilts held in the APF, the Government stated that any sales
the Bank of England undertook would be coordinated with the Treasury and the Debt
Management Office to manage the impact on the gilt market.

7.28 The APF purchases gilts as part of the Bank of England’s quantitative easing programme.4
It has purchased around a third of all conventional gilts in issue, costing the Bank a little
under £435 billion at prevailing market prices. This is essentially unchanged from two years
ago. The purchases are financed by the issuance of Bank reserves, which pay interest at
Bank Rate. This has the effect of reducing public sector debt interest payments (by the
difference between the rate paid on the gilts and Bank Rate) but also shortening the average
maturity of the debt stock and so increasing risk. The impact of the APF is to increase the
amount of public sector debt with a maturity of less than one year from 20 to 40 per cent.
7.29 There are two distinct fiscal risks arising from the APF’s gilt holdings:
4
OBR, The direct fiscal consequences of unconventional monetary policies, March 2019.

Debt interest risks
• The difference between the rate it receives on its gilts and Bank Rate may change
unexpectedly. Movements in the effective interest rate it receives on its gilt holdings will
move relatively slowly (only around 33 per cent of its holdings at the end of 2018-19
were due to be redeemed within five years). By contrast, Bank Rate changes affect the
entire stock of reserves immediately. So Bank Rate movements represent the primary
source of fiscal risk. The full-year effect of each 0.25 percentage point rise in Bank
Rate – assuming no change in the gilts it holds and therefore the coupons it receives –
reduces the debt interest saving from the APF by £1 billion.
• The second risk comes from changes in the size of the APF and how that is accounted
for in PSND. The difference between the price that gilts are bought or sold at and their
nominal value is reflected in PSND. We do not forecast increases in APF assets unless
the MPC has announced them, but we do forecast decreases if the terms of the MPC’s
stated policy – that it will not begin to reduce its stock until the Bank Rate reaches 1.5
per cent – are met. This threshold was changed by the MPC in June 2018, having
previously been 2 per cent.5 It currently lies beyond our forecast horizon, so we do not
forecast reductions, but in our October 2018 forecast Bank Rate reached 1.5 per cent
in the final year. On our assumption that sales of gilts began at that point, and further
assumptions about their pace and composition, this reduced PSND by £4 billion.
7.30 While Bank Rate does not currently reach 1.5 per cent within our forecast horizon, the curve
is very flat, which, given normal levels of volatility, suggests that the 1.5 per cent threshold is
likely to move into and out of our forecast horizon on a regular basis. As of 5 July, that risk
had receded as market expectations for Bank Rate in 2023-24 had fallen from 1.1 per cent
at the time of our March forecast to just 0.7 per cent. The nature of any APF run-off is also
uncertain. The Bank has been clear that 1.5 per cent is not a hard threshold at which sales
will definitely start and has given little guidance on their pace or composition. Any forecasts
of this are therefore subject to considerable uncertainty.
7.31 There is also a risk that as the APF sells gilts it increases yields by more than is priced into
our forecast. The Government’s MFR statement that “if and when the MPC decides to
unwind the facility, any sales of APF assets would be coordinated with the Treasury and the
Debt Management Office (DMO) to manage the impact on the gilt market” suggests that this
risk may have some bearing on how any eventual run-off is managed.
Conclusions
7.32 The level of gilts held in the APF has not changed since our previous report and the
sensitivity to changes in interest rates is therefore little changed. So it remains a material
source of risk to our medium-term spending forecast. The MPC’s guidance about when the
rundown of APF assets might commence has changed, with the Bank Rate threshold lowered
from 2 to 1.5 per cent. But market interest rate expectations have also fallen further. Taken
together, the probability of the rundown starting within our forecast horizon is also little
changed. There is no further information about how any APF rundown would be managed.
5
Bank of England, Monetary Policy Summary and minutes of the Monetary Policy Committee, June 2018.

Debt interest risks
The growth-corrected interest rate

7.33 The evolution of the debt-to-GDP ratio reflects three factors.6 First, the size of the primary
budget balance – the difference between government spending on all but debt interest on
the one hand and tax revenues on the other. Second, the ‘growth-corrected’ interest rate –
the difference between the interest rate paid on the government’s debt, which raises the
debt-to-GDP ratio, and the growth rate of output, which reduces it.7 Finally, any ‘stock-flow
adjustments’. As Chapter 1 explains, these are changes in the stock of debt left unaccounted
for by the flows of primary borrowing and debt interest: they can stem either from the net
acquisition of financial assets or from timing, classification and valuation effects.
7.34 The growth-corrected interest rate (or ‘R-G’ for short) is an important factor driving the
dynamics of the debt stock. When R-G is high – for instance when there are fears of default
(either de jure through failure to meet debt payments or de facto through currency
depreciation and unanticipated inflation) – the debt-to-GDP ratio can rise rapidly.
Consequently, the behaviour of R-G can be a key determinant of fiscal sustainability.
7.35 In an important recent contribution, former IMF chief economist Olivier Blanchard, in his
presidential address to the American Economic Association (AEA), revisited the historical
evidence on R-G in the US, concluding that more often than not the interest rate on US
government debt had been less than the growth rate.8 He noted this meant that “historically,
debt rollovers would have been feasible” – in other words that if debt rises the government
could repay the principal and accumulated interest with new borrowing without triggering
an upward spiral in the ratio of debt to GDP. He went on to consider the implications of this
continuing to be the case for the conduct of US fiscal policy, concluding that the costs of
debt are “smaller than is generally taken as given in current policy discussions.”
7.36 In light of this we now:
• document the history of the growth-corrected interest rate in the UK;
• consider how different possible future paths for R-G would affect the future evolution of
public debt relative to GDP; and
• discuss how other assumptions might affect the balance of fiscal risks.
6
This decomposition can be expressed as: dt – dt-1 = pt + st + [(Rt – Gt)/(1+Gt)]dt-1. The change in debt-to-GDP (dt – dt-1) is equal to the
primary deficit (pt), plus any stock-flow adjustments (st), plus the impact of any difference between the effective nominal interest rate (Rt) on
the debt stock and nominal GDP growth (Gt). The growth-corrected interest rate, Rt – Gt, which determines whether this impact is positive
or negative can also be expressed in real terms: the real interest rate (rt) minus real GDP growth (gt).
7
This relationship holds more generally: if the nominal interest received from an asset (or paid to a liability) is larger than the rate of
growth of the economy, then that asset (or liability) will tend to increase relative to the size of the economy (assuming interest costs are
rolled over and any interest received is reinvested). All interest rates discussed in this section are the rates paid on government debt. These
so-called ‘riskless’ rates tend to be lower than the rate of return received by holders of other assets. For instance, Thomas Piketty discusses
the returns on capital in the economy as a whole: Piketty, T., Capital in the 21st Century, 2013.
8
Blanchard, O., Public debt and low interest rates, January 2019.

Debt interest risks
The growth-corrected interest rate in history

7.37 The historical importance of the growth-corrected interest rate to the evolution of the public
finances over the past century can be seen in Charts 7.4, 7.5 and 7.6:9
• During both the First and Second World Wars, the debt-to-GDP ratio rose sharply – by
about 100 percentage points – as conflict pushed up military spending. In both cases,
the effect was partially offset by a favourable growth-corrected interest rate, due to
both higher growth (from high government spending and inflation) and lower interest
rates (due to concessional lending and the issuance of low-coupon war bonds).
• In the 1920s, the Government tightened both monetary policy (in an ill-fated attempt
to return to the gold standard at the pre-war parity) and fiscal policy (in an attempt to
pay off its war debts in real terms). The resulting high interest rates, shrinking real
economy, and deflation combined to generate the greatest upward pressure on debt
from the growth-corrected interest rate in any decade, and ultimately outweighed the
downward effect on debt from a string of primary surpluses.
• After the Second World War, the debt-to-GDP ratio fell by around 200 percentage
points over three decades. Half of this fall came from growth exceeding the effective
interest rate (Chart 7.5).10 Initially this was largely due to ‘financial repression’,11 but as
the 1970s wore on it increasingly reflected unanticipated inflation (Chart 7.6).
• Excluding banks taken onto the public sector balance sheet, debt rose by almost 50
per cent of GDP during the late 2000s financial crisis. This mainly reflected a sharp
rise in the primary deficit, with the cash value of spending rising relative to GDP
(reflecting mainly the unexpected and sharp contraction in nominal GDP) while tax
revenues fell even more sharply than GDP. Although the effective interest rate initially
exceeded the growth rate, this raised the debt-to-GDP ratio only very modestly.
9
Compiling very long time series inevitably requires judgements to be made about how to splice together different data sources and fill
any gaps in the data. We have used the Bank of England’s ‘Millennium of macroeconomic data’ to produce these charts and analysis.
10
To line up with the Bank’s dataset, primary balances are calculated as PSNB minus gross interest spending. Effective interest rates are
therefore calculated as gross interest payments divided by the previous period’s net debt.
11
Policies that have the effect of keeping nominal interest rates below the competitive market equilibrium – e.g. regulatory requirements
for pension funds to hold government debt, or capital controls on foreign exchange. See, for example, Reinhart, C., Kirkegaard, J. and
Sbrancia, B., Financial repression redux, IMF Finance & Development, June 2011, Volume 48, No.1.

Debt interest risks
Chart 7.4: Public sector net debt since 1900

300
WWI WWII Post-war decades Post-crisis
250
200
Per cent
150
100
50
Public sector net debt including banks
0
1900-01 1910-11 1920-21 1930-31 1940-41 1950-51 1960-61 1970-71 1980-81 1990-91 2000-01 2010-11
Chart 7.5: Contributions to changes in the debt-to-GDP ratio by decade

200
Primary borrowing
Growth-corrected interest rate (R-G)
150
Stock-flow adjustments
Total change
100
Percentage points of GDP
50
-50
-100
-150
1900s 1910s 1920s 1930s 1940s 1950s 1960s 1970s 1980s 1990s 2000s 2009-10
to
2018-19
Note: Contributions to changes in PSND are annual changes summed over the decade (with the 1900s defined as the change from
1899-00 to 1909-10 and so on).
Source: Bank of England, ONS, OBR

Debt interest risks
Chart 7.6: Decomposing the growth-corrected interest rate

30
WWI WWII Post-war decades Post-
crisis
20
10
Per cent
-10
Effective nominal interest rate (R)

-20
Year-on-year change in nominal GDP (G)
Growth-corrected interest rate (R-G)
-30
1900-01 1910-11 1920-21 1930-31 1940-41 1950-51 1960-61 1970-71 1980-81 1990-91 2000-01 2010-11
7.38 The mean growth-corrected interest rate since 1900 is -0.3 per cent, while the median is
slightly higher at 0.3 per cent. Confining attention to peacetime, so excluding the very
favourable R-G conditions experienced in some years of each world war, the mean rises to
0.5 per cent and the corresponding median to 0.7 per cent.
7.39 While these averages are all relatively close to zero, the range of R-G outcomes over the
past century is very wide, twice exceeding 10 per cent and falling below -10 per cent in
seven years (Chart 7.7). This variability is crucial when considering risks to fiscal
sustainability. Because the range is so wide, the future path of R-G is necessarily highly
uncertain. For instance, even if we (heroically) assume that the outturns summarised in
Chart 7.7 come from a stable statistical distribution, a 95 per cent confidence band around
the mean R-G across the whole period would nevertheless still span both -1.4 per cent and
+0.7 per cent. A recent IMF study obtained negative estimates of various growth-corrected
interest rates in the UK (and some other G7 countries) using a variety of statistical tests, but,
like us, could not statistically reject the hypothesis that the mean growth-corrected interest
rate is actually a low positive number.12
12
Barrett, P., Interest-Growth Differentials and Debt Limits in Advanced Economies, April 2018.

Debt interest risks
Chart 7.7: Distribution of growth-corrected interest rates: 1900-01 to 2018-19

30
More favourable Less favourable
25
20
Number of years
15
Wars
Peacetime
Total
10
Post-crisis
0
Less than -10 to -7½ -7½ to 5 -5 to -2½ -2½ to 0 0 to 2½ 2½ to 5 5 to 7½ 7½ to 10 More than
-10 10
Difference between effective interest rate and nominal GDP growth (percentage points)
Source: Bank of England, ONS, OBR
7.40 Chart 7.8 repeats Chart 7.7 for the rest of the OECD countries over the period since 1961,
showing frequency distributions based on a dataset collated by Charles Wyplosz in a recent
response to Blanchard’s AEA address.13 From the perspective of fiscal risk, the most
important thing it shows is that the range of R-G outcomes has been very wide in all
countries. The averages over the period range from -2.7 per cent in Ireland to 2.6 in Italy
and 5.1 per cent in Greece. For the UK, it was -0.5 per cent over this period, similar to the
averages in the OECD and US and to the average over the longer period used in Chart 7.7.
Chart 7.8: Growth-corrected interest rates in advanced economies: 1961 to 2017

60
More favourable Less favourable
50
Other advanced economies
Italy
40
Per cent of distribution
USA
UK
30
20
10
0
Less than -10 to -7½ -7½ to 5 -5 to -2½ -2½ to 0 0 to 2½ 2½ to 5 5 to 7½ 7½ to 10 More than
-10 10
Difference between effective interest rate and nominal GDP growth (percentage points)
Source: IMF, OECD, World Bank
13
Due to a lack of available data, some countries and some years are excluded. See, Wyplosz, C., Olivier in Wonderland, June 2019.

Debt interest risks
7.41 As well as a wide variance, R-G also displays significant serial correlation – it goes through
phases rather than fluctuating randomly.14 So, even though debt rollovers may have been
possible ‘on average’ over the past century, there were extended periods when the growth-
corrected interest rate was pushing the debt-to-GDP ratio upwards.
7.42 Chart 7.9 illustrates this. Each scenario assumes debt stood at 100 per cent of GDP at the
start of a decade, and isolates the effect of R-G outturns on it over the subsequent 20 years.
In effect, these scenarios assume no primary borrowing, no stock-flow adjustments, and that
the entire debt stock is rolled over onto the new effective rate each period. In seven of the
12 periods, debt is a smaller proportion of GDP after two decades, but in five it is higher. In
one, based on the 20 years beginning in 1920-21, R-G would have caused debt to double.
Chart 7.9: The effect of R-G on debt in the UK in overlapping 20-year periods
250
200
Per cent of GDP (Start of decade = 100)
150
100
50
0
1900-01 1910-11 1920-21 1930-31 1940-41 1950-51 1960-61 1970-71 1980-81 1990-91 2000-01 2010-11
Source: OBR
7.43 Looking at the history of R-G is informative, but for a fuller picture of potential future risks to
sustainability, we must also ask whether the past is likely to be a good guide to the future.
What might the future hold?

What do recent trends imply for future growth-corrected interest rates?
7.44 In some simple theoretical models, there is a tight relationship between domestic real
interest rates and economic growth.15 But in practice, it is not one that we would expect to
hold in any given year, and, in any case, several other factors are also likely to be relevant.
Empirical studies suggest that for globally integrated economies like the UK, domestic
14
This remains the case even after removing the effect of the roughly 15-year average term to maturity of the debt stock, which means
changes feed through to the effective rate only gradually.
15
For instance, in the Ramsey-Cass-Koopmans growth model with logarithmic preferences, the real interest rate is equal to the real
growth rate plus the rate of time preference.

Debt interest risks
growth is not the most important driver of domestic interest rates, with a high degree of
correlation between UK and US government bond yields, for instance.16 Other important
factors include global interest rates and domestic policy.
7.45 As regards domestic policy, sustained reductions in the debt-to-GDP ratio have often been
associated with financial repression. In some respects, this is harder to achieve today:
inflation is lower and the task of keeping it low has been delegated to independent central
banks, and capital flows more freely across borders. But several interventions have boosted
gilt prices in recent years and so lowered effective interest rates (thereby having financial-
repression-like consequences, even if they were not pursued with that objective in mind).
These include Basel III regulations that require banks to hold enough high-quality liquid
assets (i.e. government bonds) to fund cash outflows for the duration of a 30-day stress
scenario, and the Bank of England’s purchase of £435 billion worth of gilts at prevailing
market prices under its quantitative easing programmes.
7.46 Abstracting from policy interventions, over the past 30 years or so, several structural factors
are thought to have lowered real safe interest rates by affecting the demand for, and supply
of, savings. For instance, ageing populations’ desire to save for retirement increases
demand for safe assets, reducing interest rates. And there are other candidate explanations:
changes in the relative price of capital goods; technological developments favouring greater
use of intangible instead of physical capital; increased income inequality; and the impact of
increased market power. Several of these are likely to persist, in the near future at least.17
7.47 Real growth rates in the UK have also fallen, although by less than interest rates, as
productivity growth has stubbornly failed to return to its pre-crisis trend. Our November
2017 EFO outlined several candidate explanations for this underperformance, including
weak investment, highly accommodative monetary policy, and employment growth
becoming increasingly skewed toward low productivity jobs and industries. Given the
extended period of post-crisis productivity weakness, we revised down our long-run
productivity growth assumption in that forecast rather than continuing to assume that
whatever has depressed growth in recent years will disappear over the coming five years.
7.48 So, downward revisions to risk-free interest rates have outweighed downward revisions to
growth rates. On the face of it, that – coupled with historical evidence on the feasibility of
debt rollovers – would suggest that policymakers can afford to be more relaxed about the
risks to debt sustainability than in the past, echoing Blanchard’s conclusion. (Of course,
relatively little weight should be given to any particular value of R-G in supporting this
conclusion, since history has also shown that any departure from a favourable central
expectation could be both significant and persistent, with large fiscal implications.)
7.49 One concern that could temper that conclusion is that the lower level of long-term risk-free
interest rates at present could in fact foreshadow a period of weaker growth than we (and
16
For example, Chinn, M., and J. Frankel, Debt and Interest Rates: The US and the Euro Area, 2007.
17
See, L. Rachel and L. Summers, On Falling Neutral Real Rates, Fiscal Policy, and the Risk of Secular Stagnation, 2019, for an overview
of several of these factors. The authors go on to argue that because these forces will continue to keep short-term interest rates at or near
their ‘zero lower bound’, fiscal policy may need to become more activist in future.

Debt interest risks
many other forecasters) currently expect. Since the crisis, forecasts conditioned on market-
derived interest rates have often had growth rates exceeding interest rates. But while the
assumed path for interest rates has often proved a good guide to the outturn, growth
forecasts have been repeatedly revised down. So what currently looks like a favourable
outlook for R-G may instead presage a further downgrading of the growth outlook.
What do we and other institutions assume in forecasts and projections?

7.50 In our most recent medium-term forecast, the growth-corrected interest rate averages -1.1
per cent (Table 7.2). As our growth forecast picks up by slightly more than interest rates rise,
debt dynamics become slightly more favourable over the medium term. As we discuss in
paragraph 7.17, this could well reflect market participants pricing in some probability of a
no-deal Brexit and the monetary easing that they expect would ensue (in contrast to the
smooth Brexit on which our economic forecast is conditioned).
Table 7.2: Medium-term forecast for the growth-corrected interest rate

Per cent of GDP
2018-19 2019-20 2020-21 2021-22 2022-23 2023-24 Average
Effective interest rate 2.1 2.3 2.2 2.3 2.4 2.4 2.3
Nominal GDP growth 3.3 3.0 3.4 3.5 3.6 3.6 3.4
Growth-corrected interest rate -1.1 -0.7 -1.2 -1.3 -1.2 -1.3 -1.1
Note: Effective interest rates are calculated as gross interest payments divided by the previous period’s net debt.
7.51 Favourable dynamics in our forecast are in line with the recent past in the UK and with the
outlook in many other advanced economies. Indeed, as Chart 7.10 shows, in the IMF’s
most recent economic and fiscal forecasts, R-G is expected to be favourable to public sector
debt dynamics over the next five years in 27 out of 33 advanced economies.
Chart 7.10: IMF forecasts for the growth-corrected interest rate

3
2
Per cent (average from 2019 to 2024)
-1
-2
-3
-4
-5
Lithuania
Korea
Belgium
Ireland
Australia
Finland
United Kingdom
Norway
Iceland
Luxembourg
Switzerland
Canada
Average
Czech Republic
Slovakia
Sweden
Denmark
Austria
Portugal
France
Slovenia
Germany
Latvia
Cyprus
Italy
Spain
Estonia
Hong Kong
Japan
United States
Malta
New Zealand
Israel
Netherlands
Source: IMF

Debt interest risks
7.52 In making long-term projections, economists and policymakers – including the US

Congressional Budget Office and the European Commission, as well as ourselves – are
wont to assume a modest positive growth-corrected interest rate. (For instance, the CBO
recently used a projection of long-term government bond yields rising to 4.6 per cent
between 2039 and 2048, compared with a nominal growth rate of 4.0 per cent.18) But this
is as much for convenience as on the basis of compelling theory or evidence.
7.53 In our case, we have assumed since our first Fiscal sustainability report (FSR) in 2011 that
the long-run real interest rate on government debt will eventually exceed the growth rate by
0.2 percentage points.19 The latest market expectations for interest rates are consistent with
a slightly more favourable growth-corrected interest rate than in our central projection – on
the assumption that the market holds the same view as us about long-term nominal GDP
growth, which it may not. In its own FSR, the European Commission shows that switching to
a market-expectations-based projection of nominal interest rates would only lower debt-to-
GDP in 2029 by 3.7 percentage points – again, leaving long-run growth rates unchanged.
7.54 Hitherto, we have judged the assumption of a modestly positive growth-corrected interest
rate beyond our medium-term forecast horizon to be relatively neutral for assessing fiscal
sustainability and have maintained it in successive FSRs. Consequently, changes in R-G have
not been a source of variation from one projection to the next. Most importantly, R-G is not
material to our reports’ main conclusions: in the FSR we are less interested in the stability of
debt following a one-off step increase than we are in the primary balance required to
achieve sustainable debt dynamics over the longer run. In that context, whether R-G is a
small positive or small negative number does not make a huge difference.20
7.55 Towards the end of our latest medium-term forecast, debt stands at around 75 per cent of
GDP. Given anticipated growth rates, interest rates and stock-flow adjustments in 2023-24,
the Government would need to run a 0.1 per cent of GDP primary surplus (as defined in
this chapter) to hold debt constant at that level.
7.56 Chart 7.11 shows the primary balance required to stabilise the debt-to-GDP ratio in
2023-24 for all the combinations of interest and growth rates in the period since 1900-01.
Faced with some years’ growth-corrected interest rates, particularly those seen in some
wartime years, the Government could run sizeable primary deficits and still hold debt flat.
But were it to face the hugely unfavourable debt dynamics of 1921-22, for instance, a
primary surplus of almost 20 per cent of GDP would be required.
7.57 Over the long run, demographic and other cost pressures are likely to put significant
upward pressure on spending. Accommodating them raises primary spending by 8.2 per
cent of GDP between 2022-23 and 2067-68 in our baseline FSR projection. Only growth-
18
Congressional Budget Office, The 2019 Long-Term Budget Outlook, June 2019.
19
This is the rate paid on newly-issued debt. In our 2018 FSR, as the existing stock of debt gradually rolls on to these new rates, the
effective rate converges on 0.2 per cent. Like the other interest rates discussed in this section, this does not net off interest receipts. Doing
so would lower the effective rate, due to the 3 per cent real interest rate accrued on the stock of student loans – much of the interest
accrued is never expected to be repaid, while student loan assets do not net off against PSND, hence using gross interest payments here.
As we discuss in Chapter 6, the ONS intends to change the accounting treatment of student loans soon.
20
Specifically, in the absence of any stock-flow adjustments, the debt-to-GDP ratio will be stable if, and only if, pt = - (Rt - Gt)/(1 + Gt)dt-1,
where, as before, dt-1 and pt are the debt and primary deficit as shares of GDP and Gt and Rt are the nominal growth and interest rates.

Debt interest risks
corrected interest rates in the bottom 2 per cent of the historical distribution would be
favourable enough to offset such a rise. This is extremely unlikely to persist over decades.
Chart 7.11: Combinations of R-G levels and debt-stabilising primary balances

20
15
Debt-stabilising primary balance (2023-24)
10
-5
-10 R-G sufficient

to offset long
run spending
-15 pressures
-20
-40 -30 -20 -10 0 10 20 30 40
Source: OBR Growth-corrected interest rate in 2023-24
What does this mean for fiscal risks?

What are the impacts of higher debt levels?
7.58 Historically, large rises in the debt-to-GDP ratio have resulted from the sharp increases in
deficits associated with wars and financial crises. If growth-corrected interest rates were
positive, higher debt ratios leave policymakers less flexibility to accommodate these shocks
in the future. Financing a war or the consequences of a natural disaster then requires the
government to raise taxes, or cut other spending, if fiscal sustainability is to be retained.
Today, large-scale conflicts are hopefully a less likely source of risk than at the turn of the
20th century. But another significant financial crisis would still carry a sizeable fiscal cost.
And the future may contain new pressures, perhaps relating to Brexit or to climate change.
7.59 Higher debt also leaves interest payments more exposed to changes in the growth-corrected
interest rate. This means that following a shock, a larger correction to the primary balance
is required to stabilise the debt – and, in the absence of corrective action, that debt rises
more quickly. Holding all else constant, Chart 7.12 zooms in on Chart 7.11 to show that if
the growth-corrected interest rate were to rise to 2 per cent in 2023-24 (versus -1.3 per cent
assumed in our central forecast), a primary surplus of roughly 2.5 per cent of GDP would
be required to hold debt constant. (This compares to a debt-stabilising primary surplus of
0.1 per cent of GDP given our central assumption for R-G – and to the 1.2 per cent of GDP
primary surplus in our central forecast.) And were debt twice as high as in our central
forecast (at 150 per cent of GDP, instead of 75 per cent), a primary surplus of around 4 per
cent of GDP would be required to hold it stable with R-G at 2 per cent.

Debt interest risks
Chart 7.12: Growth-corrected interest rates and debt-stabilising primary balances at

different starting levels of debt
5
Debt at 2023-24 level R-G greater than or
4 Debt at double 2023-24 level equal to 2 per cent
Debt-stabilising primary balance (2023-24)
2 March 2019 central

forecast
1
-1
-2
-3
-4
-5
-10 -8 -6 -4 -2 0 2 4 6 8 10
Source: OBR Growth-corrected interest rate in 2023-24
7.60 In contrast, if we could be confident that R-G would be negative, then – in the framework we
use to analyse sustainability in our FSRs – one-off increases in debt would not require tax
rises or spending cuts for the government to remain on a sustainable path. Based on the
analysis of past and expected future trends outlined above, some commentators have
suggested that policymakers therefore need to be less worried about high levels of debt in
the current context than previously thought. For instance, Blanchard argues that historically
negative R-G is “at striking variance with the current discussions of fiscal space, which all
start from the premise that the interest rate is higher than the growth rate.” He also notes the
importance of considering other factors beyond R-G. Two such sources of risk associated
with higher debt levels that are not examined in our FSRs include:
• Interactions between public debt levels and the economy. Debt can be used to achieve
economic outcomes – for instance, governments can spread out over time the
economic cost of events like wars or financial crises by allowing public debt to rise to
support economic activity in any given year21 – but these interactions can be complex.
Evidence suggests that, although the debt level itself is not a particularly strong
indicator of the medium-term economic outlook, its trajectory, at least, does matter.22
• Financing risks. This concerns market expectations for the future path of the primary
deficit. Higher primary deficits can be expected to increase the cost of borrowing, and
if markets begin to fear default, this increase can be sharp. This can result in situations
where an initially solvent debtor’s liabilities become unsustainable simply because
creditors begin to think that is so and become unwilling to refinance them.23
21
This theory of optimal debt accumulation is discussed in R. Barro, On the determination of the public debt, 1979.
22
Chudik, A., et al., Is there a debt-threshold effect on output growth?, March 2017.
23
See, Calvo, G., Servicing the public debt: the role of expectations, 1988.

Debt interest risks
A government will be more exposed to such self-fulfilling prophecies when debt is

high. Policymakers may thus increasingly fear that financial markets will demand
progressively higher interest rates as debt rises. Higher debt levels may also constrain
policymakers’ behaviour because the future path of the growth-corrected interest rate
is unknown and policymakers wish to be prudent.
Is there an optimal level of debt-to-GDP?

7.61 In Box 2.F of MFR, using an OECD-supplied framework,24 the Treasury reviewed various
estimates of public debt ratios for the UK. Estimates of ‘debt limits’ (above which the UK
would be likely to lose market access) ranged between 167 and 223 per cent of GDP. The
OECD suggests ‘debt thresholds’ (above which macro-stabilisation and growth are impaired
and sustainability might be threatened) could range between 70 and 90 per cent of GDP.
And estimates of ‘debt targets’ (levels low enough that debt stays below debt thresholds
even following large adverse shocks), range from 40 to 73 per cent of GDP.
7.62 As this highlights, there is little consensus in the academic literature or policy world as to
what levels of debt might be safe or desirable. None of these estimates fully reflect all the
factors that affect the UK’s ability to incur and service its debts. For instance, most public
debt in the UK is owed to British citizens and almost all is issued in sterling. Furthermore, the
average term to maturity of public debt in the UK is long. There is also no reason to suspect
that any estimate would remain constant over time (although a recent IMF study suggested
that debt limits in the UK might have varied relatively little historically25).
7.63 Nevertheless, it is also important to emphasise the credibility and resilience of the UK’s
policy framework as a factor determining the level of debt that can be safely sustained.
Indeed, the most plausible scenario in which interest rates would significantly exceed growth
rates is one in which market participants lose confidence in the institutional framework and
start to expect default and/or monetisation. As the euro area debt crisis highlighted, such a
situation could cause market beliefs to shift quite rapidly, although the fact that the UK
controls the supply of the currency in which its debt is issued is an important difference.
Recap of previous FRR discussion and the Government’s response

7.64 This deeper look into the risks posed by the growth-corrected interest rate reinforces the
conclusions we reached two years ago, when we noted the risks that would be associated
with interest rates rising to more normal levels relative to GDP growth. In MFR, the
Government argued that reducing the deficit below its 2017-18 level of around 2 per cent
would be necessary to get debt falling, after taking “typical economic shocks” into account
(which were assumed to increase debt-to-GDP by 10 per cent of GDP every 9 years), and
that doing so would partly offset projected rises in debt interest spending as a share of GDP.
It also pointed to its strategy for boosting long-run productivity, noting that, all else equal,
higher trend productivity growth would improve debt dynamics.
24
Fall, F., et al., Prudent debt targets and fiscal frameworks, 2015.
25
Barrett, P., Interest-Growth Differentials and Debt Limits in Advanced Economies, 2018..His estimates range between 90 and 95 of GDP
over time, although he suggests that accounting for the maturity structure of UK debt might raise this by 50 percentage points or so.

Debt interest risks
Conclusions
7.65 Debt rose much less after the financial crisis than it did after major wars, but by much more
than after any other peacetime shock. While a lower or negative growth-corrected interest
rate would make it less risky to absorb these sorts of one-off shocks, it is important not to
overstate the importance of the growth-corrected interest rate to the conclusions we draw in
our FSRs. The more important challenge to fiscal sustainability is how demographic trends
and other cost pressures in health and social care provision will affect spending over the
longer run. A lower R-G would make the sustainability challenge posed by ageing and
health cost pressures somewhat less daunting, but the chance of the beneficial effect being
sufficiently large or persistent to offset them altogether is vanishingly small.
7.66 There are inherent difficulties associated with predicting growth and interest rates over the
long run. The actual value of R-G is certain to differ from our baseline assumption and – if
historical variability is any guide – will probably do so by a significant amount at different
points in the coming decades. Any differences are likely to persist for a period of years.
Conclusions
7.67 This chapter has illustrated the sensitivity of debt interest spending to several factors, notably
the interest rate on new borrowing that feeds through to the effective interest rate on the
outstanding stock of debt – in some cases quickly, in others over many years. The most
important downside fiscal risks are those that would push interest rates up relative to
economic growth, thereby raising debt interest spending proportionately more than GDP.
7.68 Over the medium term, there are many factors that could raise debt interest spending
relative to our most recent forecast. As in our 2017 report, higher Bank Rate or RPI inflation
would affect spending quickly, and higher gilt yields or borrowing would affect it more
slowly. The Government’s exposure to inflation risk, which it has begun to reduce, could be
further reduced if methodological changes to the RPI were to be implemented. Brexit poses
a risk to our interest rate forecast, given that markets now appear to be pricing in some
chance of a ‘no deal’ outcome and the monetary loosening that they expect it would bring.
7.69 There has been much debate in recent months over what, if anything, history tells us about
the future growth-corrected interest rate. It remains to be seen if rates will stay ‘lower for
longer’ and whether that would hold if growth rates picked up. But although such an
outcome poses a favourable risk to our baseline debt projection, no plausible future path is
likely to offset fully the building demographic and other pressures on primary spending.

Debt interest risks

consider when managing its fiscal risks. Among them:
• The increase in the debt stock and the issuance of index-linked gilts in recent years.
• The increased sensitivity of debt interest spending to inflation and interest rate risk.
• The balance between inflation risk exposure and other goals in ILG issuance choices.
• The potential fiscal impacts of material changes to the Retail Prices Index.
• The temporary impact of the APF in lowering the Government’s borrowing costs.
• The potential impact if interest rates rise to more normal levels relative to GDP growth.
• The balance of risks around the future path of the growth-corrected interest rate.
7.71 When assessing the outlook for debt interest spending and debt dynamics over the medium
and long term, does the Government regard these or other issues as important for its risk

8 Fiscal policy risks
Introduction
8.1 This chapter:
• discusses potential sources of fiscal policy risk;
• introduces a framework for measuring and analysing fiscal policy; and
• analyses the key drivers of fiscal policy over the past decade.
Sources of fiscal policy risk

8.2 Understanding how governments tend to set fiscal policy and react to events is important
when assessing the future sustainability of the public finances. There are many risks
associated with the setting of fiscal policy, some of which reflect uncertainty surrounding:
• Real-time economic data. Fiscal policy can have unintended consequences if based on
poor data or significant misjudgements about the current state of the economy.1
Judgements about the output gap are particularly uncertain and prone to revision.
• Our underlying public finances forecast, i.e. the state of the world absent any policy
changes. As we set out in each Economic and fiscal outlook (EFO), there is a wide
range of potential outcomes surrounding our central forecasts.
• The economic and fiscal impact of policy changes. In many cases, the fiscal impact of
individual tax and spending measures is hard to estimate accurately. We assess the
uncertainty around the costing of each measure at each fiscal event and report it on
our website. The impact of fiscal policy is also uncertain, in particular in respect of the
multipliers used to incorporate fiscal policy changes into our economic forecasts.
• The behaviour of future policy makers. Changes to plans and fiscal rules in response
to unforeseen events could impart a systematic drift in the public debt. Since we are
required to base our forecast on stated Government policy, any such behaviour is
necessarily absent from our central forecasts.
1
This finding is consistent with the international academic evidence. See, for example, Cimadomo, Real-time data and fiscal policy
analysis: A survey of the literature, European Central Bank working paper No. 1408, 2011.

Fiscal policy risks
Measurement of fiscal policy

A framework for analysing fiscal policy
8.3 The simplest measure of the change in the fiscal position is the change in public sector net
borrowing (PSNB) – the difference between the changes in total receipts and expenditures.
Movements in PSNB can reflect both temporary and structural factors, each of which can be
either related to policy or exogenous. In our EFOs, we decompose changes in PSNB into:
• The impact of the business cycle through the automatic stabilisers. This is estimated via
the historical sensitivity of the deficit to the output gap and largely reflects the structure
of the tax and benefit system and the scale of revenues and spending.
• ‘Structural’ changes in the public finances, i.e. changes in cyclically adjusted public
sector net borrowing (CA-PSNB). These are calculated by residual, having subtracted
the estimated cyclical borrowing. These ‘structural’ changes can reflect a variety of
exogenous factors, including both temporary and permanent components.
8.4 The split between cyclical and structural movements in the public finances requires
judgements on both the path of the output gap and the impact of the cycle on taxes and
spending. Both are uncertain. Chapter 2 sets out some of the challenges and issues
surrounding estimation of the output gap, while our 2012 Cyclically adjusting the public
finances working paper details our estimates of the impact of the cycle on revenues and
expenditure and the surrounding uncertainty.
8.5 We can further decompose changes in the structural budget balance into:
• Discretionary fiscal policy decisions, i.e. the tax and spending policy decisions
announced at each fiscal event. In each EFO, we show the total impact of discretionary
policy announcements as the ‘effect of Government decisions’ on the budget balance.2
• Non-discretionary, non-cyclical factors. Other factors considered when constructing

our fiscal forecast include those relating to the path of the economy (e.g. trend
productivity growth) as well as others not necessarily related to the macroeconomy
(e.g. the effective VAT rate). This element is calculated by residual, having first
estimated the effect of discretionary policy.
8.6 These changes can also be measured across two dimensions: over time (most commonly
changes relative to the previous year) or relative to previous plans (which is what the
Treasury’s scorecard of policy decisions at each fiscal event represents). The next section
describes how fiscal policy evolved in recent years. The remainder of the chapter then
focuses on how fiscal plans were changed in response to unexpected events.
2
In order to estimate the impact of government decisions at each fiscal event, we have to consider a counterfactual to compare against.
In most cases, our counterfactual simply reflects the default policy parameters that the Government has asked us to assume in our
forecasts, e.g. that excise duties should be uprated in line with RPI inflation each year. In other cases, the counterfactual is less clear. For
example, our forecast for spending on public services (RDEL) is set by the Government and is overlaid by our judgement on
underspending. In the absence of a counterfactual when our forecast is extended by a year, we make the neutral comparison that
spending would otherwise have remained constant as a share of nominal GDP.

Fiscal policy risks
The evolution of fiscal policy since the financial crisis

8.7 The threefold decomposition of changes in the budget balance into endogenous cyclical
movements, discretionary policy changes, and other structural factors provides an
accounting framework for understanding the forces driving the public finances and
assessing the associated risks. Chart 8.1 sets out changes in headline borrowing since
2007-08, decomposed into these three factors.3 It shows that:
• Between 2007-08 and 2009-10, borrowing rose by 7.1 per cent of GDP, of which
2.9 per cent of GDP can be explained by cyclical borrowing caused by the downturn
(based on our latest estimate of the output gap). Non-discretionary structural factors –
notably the decision not to reduce cash spending plans despite the unexpected
weakness of nominal GDP – were the largest source of fiscal deterioration (worth
3.1 per cent of GDP). Discretionary policy changes, most significantly the temporary
cut in the standard rate of VAT to 15 per cent, explained only 1.2 per cent of GDP.
(This reflects the fact that changes in cash spending plans are treated as discretionary
policy changes, rather than changes in planned spending as a share of GDP.)
• Since its peak in 2009-10, borrowing has fallen from 9.9 per cent of GDP to 1.1 per
cent of GDP in 2018-19. That fall is more than explained by discretionary policy,
largely cuts to public services spending and to welfare spending, as well as some tax
rises (including the rise in the standard rate of VAT to 20 per cent). Our latest estimate
of the output gap suggests that around two-thirds of the cyclical rise in borrowing
following the crisis has now unwound. (One would not expect it to unwind fully as
output was above potential in 2007-08, so the actual level of PSNB exceeded the
cyclically-adjusted level). Other factors have raised structural borrowing, largely
reflecting the productivity-driven weakness in GDP growth. This has depressed the tax-
to-GDP ratio (partly because the associated weakness in real earnings growth has
depressed the income tax and NICs effective tax rate4) and has raised the spending-to-
GDP ratio via the denominator effect.
3
The discretionary policy series is derived from the IFS’ Fiscal response to the crisis dataset. It shows the cumulative impact of discretionary
policy measures since Budget 2008.
4
See Box 4.1 of our March 2017 EFO for more detail.

Fiscal policy risks
Chart 8.1: Explaining annual changes in net borrowing since 2007-08

12
Public sector net borrowing: levels

10
8
Per cent of GDP
0
2007-08 2008-09 2009-10 2010-11 2011-12 2012-13 2013-14 2014-15 2015-16 2016-17 2017-18 2018-19
6
Public sector net borrowing: year-on-year change

4
2
Per cent of GDP
-2
Discretionary fiscal policy since Budget 2008 (IFS)
Cyclical factors (automatic stabilisers)
-4
Other structural factors
Year-on-year change in borrowing
-6
2007-08 2008-09 2009-10 2010-11 2011-12 2012-13 2013-14 2014-15 2015-16 2016-17 2017-18 2018-19
Source: ONS, OBR
8.8 In our 2017 Forecast evaluation report, we estimated the effect of discretionary fiscal policy
changes on GDP growth (based on estimates of the consolidation produced by the Institute
for Fiscal Studies (IFS) together with our estimates of fiscal multipliers, which are drawn from
the available empirical literature). The analysis considered how changes in discretionary
fiscal policy compared with our forecast errors for GDP growth. We concluded that the
shortfall in GDP growth since 2010-11, relative to our forecast, was more likely to be
explained by non-policy factors than the application of inappropriate multipliers.

Fiscal policy risks
Fiscal policy over the past 10 years and potential risks

Fiscal objectives
8.9 Since the 1990s, UK governments have adopted ‘fiscal rules’ to guide and constrain fiscal
policy. These have typically come in pairs. One for the budget balance, with different deficit
measures targeted at different times, and one for the debt-to-GDP ratio, sometimes its level
and sometimes its profile. These rules have taken various forms over the past two decades:
• For most of its time in office, the 1997 to 2010 Labour Government followed two fiscal
rules. Its ‘golden rule’ required the Government only to borrow to invest over the
economic cycle, while its ‘sustainable investment rule’ stated that debt should be kept
at a ‘prudent’ level over the economic cycle (later specified at 40 per cent of GDP). The
crisis prompted the replacement of these longstanding rules with a simpler temporary
one and then, briefly, a Fiscal Responsibility Act, passed just before the 2010 election.
• The 2010 to 2015 Coalition Government specified a ‘fiscal mandate’ that aimed to
balance the cyclically adjusted current budget by the end of a rolling, five-year period;
in December 2014, the horizon was shortened to three years ahead. It also adopted a
‘supplementary debt target’ that required the debt-to-GDP ratio to be falling in 2015-
16. For a period, the target date was shifted to 2016-17, but the rule was then further
tweaked to aim for year-on-year falls from 2015-16 onwards.
• From July 2015 to November 2016, the Conservative Government targeted a

headline PSNB surplus by the end of 2019-20, together with a supplementary target
that required the debt-to-GDP ratio to be falling every year from 2015-16 to 2019-20.
This version of the fiscal rules contained an explicit escape clause, allowing the targets
to be overridden in the face of a significant adverse economic shock.5
• The current Government’s rules require cyclically adjusted PSNB to be less than 2 per
cent of GDP by 2020-21 and the debt-to-GDP ratio to be falling in the same year.
Again, these can be overridden in the face of an adverse shock. The Government also
has a ‘fiscal objective’ to return the public finances to balance by the ‘earliest possible
date in the next parliament’ (which in practice has been treated as 2025-26).
8.10 One notable feature over the past decade is the frequency of changes in the rules. Chart
8.2 shows the headroom at each fiscal event since June 2010 against the four fiscal
mandates chosen by successive governments. It is striking that, regardless of the rule in
operation at the time (represented by the solid segments), successive Chancellors have
consistently chosen to maintain headroom of around £20 billion against breaching the rule.
The contemporaneous rule has been on course to be met by between £10 billion and £30
billion in 14 of our 20 forecasts.
5
Defined as real GDP growth of less than 1 per cent on a rolling 4 quarter-on-4 quarter basis, which was to be assessed by the OBR.

Fiscal policy risks
8.11 The first notable exception came in December 2014 and March 2015, when headroom
against the five-year-ahead fiscal mandate far exceeded that, which prompted a tightening
of the cyclically adjusted current balance target. The next came in November 2016,
following the EU referendum, which prompted us to lower our growth forecast, leaving the
balance by 2019-20 target missed by a large margin. In response, the Government
abandoned that target and adopted a far less demanding one for a cyclically adjusted
deficit of 2 per cent of GDP. It also pushed the balanced budget objective to the mid-2020s.
Chart 8.2: Headroom against successive fiscal mandates since June 2010
100
Balance cyclically adjusted current budget at end of 5 years
Balance cyclically adjusted current budget at end of 3 years
80
Balance public sector net borrowing by 2019-20
Cyclically adjusted public sector net borrowing <2% GDP by 2020-21
60
Headroom against contemporaneous rule
40
£ billion
20
-20
-40
-60
Jun Nov Mar Nov Mar Dec Mar Dec Mar Dec Mar Jul Nov Mar Nov Mar Nov Mar Oct Mar
10 10 11 11 12 12 13 13 14 14 15 15 15 16 16 17 17 18 18 19
Note: Solid lines correspond to the operative rule at the relevant date. Dashed lines show the same measure at other dates.
Source: OBR
8.12 This suggests that governments’ fiscal decisions cannot be explained fully by reference to the
fiscal rules, but rather that the choice of rules itself depends in a somewhat predictable way
on the fiscal outlook. Indeed, alongside the formal rules, governments pursued several less
formal (and sometimes surprisingly precise) objectives for the post-measures fiscal position,
which have also moved alongside changes in our forecasts. These include:
• In the March 2013 Budget, the Government wanted to ensure that borrowing fell year-
on-year in cash terms in every year of our central forecast. This looked challenging for
2012-13 on the available data at the time. It achieved its goal by reducing
departmental spending in the final months of the year, by an unusually large amount,
including by pushing some of that spending out into 2013-14.
• In the December 2014 Autumn Statement, the Government reduced its final-year
spending assumption such that total expenditure was set to fall to its lowest share of
national income since the 1930s. In the March 2015 Budget, it wanted to ensure that
this was no longer the case, but also to ensure that the post-measures borrowing
forecast was revised down in every year up to 2018-19. It achieved that by altering its
spending assumption in a way that left a rollercoaster profile for public services

Fiscal policy risks
spending in the years beyond those covered by Spending Review plans. As seemed
inevitable, the profile was smoothed out in time for the 2015 Spending Review.
• In the March 2016 Budget, the Government wanted to ensure that its headroom versus
the fiscal mandate (to balance the budget in 2019-20) remained as close as possible
to the headroom we had predicted at the previous fiscal event (£10.1 billion). Our pre-
measures forecast showed the target being missed, so a significant tightening was
required to achieve this goal. The overall policy package largely delivered that by
shuffling money between years. A measure bringing forward £6.0 billion of
corporation tax instalment payments was delayed so that it flattered the target year
(this was “to give businesses more time to prepare”), while £1.6 billion of departmental
capital spending was shifted out of 2019-20, “accelerating investment plans”.
• In the October 2018 Budget, the Government fine-tuned its policy measures to ensure
that the headroom against the fiscal mandate remained at the same level (to the
nearest £0.1 billion) as it had been in our March 2018 forecast. It decided against
maintaining this unofficial target in the March 2019 Spring Statement.
8.13 Governments have also periodically sought to exploit the fact that public sector net debt only
nets off liquid assets, or taken other steps, to ensure that debt is forecast to fall relative to
GDP in a particular year, sometimes by fiscally insignificant amounts. Examples include:
• In Budget 2016, the Government brought forward several asset sales (in particular
sales of Bradford & Bingley mortgages held by UK Asset Resolution), in an attempt to
ensure the debt-to-GDP ratio fell in 2015-16 (which was almost over by then).
• In Autumn Budget 2017, various spending items were pushed back from 2018-19 to
later years in changes to the policy package that were made too late for us to reflect
them in our economy forecast. This ‘reprofiling’ was sufficient in size to mean that net
debt fell from 86.5 to 86.4 per cent of GDP between 2017-18 and 2018-19 –
although by only 0.03 per cent of GDP in unrounded terms – rather than being flat.
8.14 These were later examples of a trend we highlighted in Box 5.1 of our March 2015 EFO,
noting the increasing impact of non-primary balance factors in explaining the debt profile in
2015-16 (the target year at the time). The Government made several policy decisions
designed to affect the profile of PSND in that year, largely relating to new asset sales
(including some of the UKAR loan book as well as shareholdings in Lloyds Banking Group).
Such decisions exploit what we note in Chapter 6 to be a fiscal illusion to which PSND is
susceptible, namely that selling an illiquid asset for what it is worth lowers PSND (by
swapping an illiquid asset for a liquid one) but has no implications for fiscal sustainability.
Conclusions
8.15 The setting of fiscal rules needs to balance the credibility of the objective against flexibility in
the face of unexpected events. However, repeated moving of the goal posts over the past
decade (typically alongside movements in the underlying outlook for the economy and

Fiscal policy risks
public finances) risks making it more difficult to discern the likely future path for fiscal policy
and the credibility of the constraint the Government has sought to place on itself. (In
Chapter 5, we note a similar moving of the goal posts with the ‘welfare cap’.) At the same
time, the Government has engaged in sometimes remarkably precise medium-term fine
tuning of its policy measures to meet informal objectives for the public finances. The
margins by which these objectives were set to be achieved were rarely fiscally or statistically
significant given the uncertainty around both the pre- and post-measures forecasts.
8.16 The risk is that when revisions to the underlying fiscal position are sufficiently adverse, the
rules are reset to accommodate that, rather than policy being adjusted to meet them.
Relative to a world in which fiscal rules acted as a greater constraint, the path of debt can
be expected to be higher. But that is not to say that any particular vintage of fiscal rule
should be pursued in all circumstances, as some shocks may clearly warrant overriding it.
Policy reactions to unanticipated shocks

8.17 Alongside the reputational demands of being seen to meet fiscal rules and informal
objectives, there are three key factors that governments are likely to consider when
determining discretionary fiscal policy. Empirical evidence suggests that a mix of these
factors are important and that the factors vary across countries and over time:
• Macroeconomic conditions. Discretionary fiscal policy may be used to influence the

business cycle (over and above the impact of automatic stabilisers) and so the size and
direction of the output gap may be a key consideration.
• The level and sustainability of public debt. A government may wish to alter the flow of
annual borrowing in order to target a particular trajectory for the stock of debt given
its judgements about the risks posed by its pre-measures path.
• The effectiveness of monetary policy. A government may be more inclined to use fiscal
policy to influence the economy if monetary policy is constrained (for example, if
interest rates are at their effective lower bound and unconventional monetary policies
such as asset purchases are thought ineffective or undesirable).
8.18 Chart 8.3 shows like-for-like revisions in our forecasts for PSNB since June 2010, split into
discretionary policy changes, the impact of automatic stabilisers and other underlying
changes in our forecasts.6 It shows that, since June 2010:
• Discretionary fiscal policy has responded to changes in the outlook for the public
finances conveyed by our central forecasts. In more than two-thirds of fiscal events
since June 2010, the Government has attempted to offset movements in our
underlying forecast with changes in discretionary policy.
6
Further information on the methodology underpinning the rest of the analysis in this chapter can be found in Annex B of our March
2016 EFO or the Forecast revisions database that we publish online. The methodology takes the diagnostic tables that we have included
in each EFO and decomposes forecast revisions into: classification changes, underlying forecast changes and changes relating to
Government decisions. In this report, we have further decomposed underlying forecast changes into its respective structural and cyclical
components by applying the revisions to our output gap forecasts over time.

Fiscal policy risks
• Discretionary fiscal policy has not been particularly responsive to changes in our view
of the business cycle. Most notably, our prediction of a cyclical deterioration in our
December 2012 EFO and a cyclical improvement in our December 2013 EFO (the
largest movements in our output gap forecasts since the creation of the OBR) were
each met with a modest discretionary tightening by the Coalition.
Chart 8.3: Sources of revision to PSNB forecasts since June 2010
2.0
Discretionary policy
Underlying forecast changes
1.5
Automatic stabilisers
Like-for-like forecast revision
1.0
Per cent of GDP
0.5
0.0
-0.5
-1.0
-1.5
Nov Mar Nov Mar Dec Mar Dec Mar Dec Mar Jul Nov Mar Nov Mar Nov Mar Oct Mar
10 11 11 12 12 13 13 14 14 15 15 15 16 16 17 17 18 18 19
Note: Diamonds reflect the cash revision over the last 5 years of the forecast, rebased to each historic nominal GDP forecast.
Source: OBR
8.19 Plotting discretionary policy changes (the blue bars above) against our underlying structural
forecast changes (the yellow bars above) shows us how governments tend to react to our
forecasts on average. Chart 8.4 shows that, in around three-quarters of the fiscal events
since June 2010, governments have with a surprising degree of consistency aimed to offset
about a third of the change in our underlying structural forecast via discretionary measures.7
8.20 For the remaining five fiscal events (shown in blue on the chart), the relationship does not
follow this pattern, with an average policy response that is around 0.3 per cent of GDP
looser than would be implied by the more common pattern. In stark contrast, there has
been no fiscal event where the policy response has been significantly tighter than this typical
pattern. The context for each of these five policy responses has been unusual in some way:
• In our November 2011 forecast, we revised down trend productivity growth materially,
leaving our pre-measures structural borrowing forecast around £30 billion a year
higher on average. The Government decided to offset only around 15 per cent of this
change over 2011-12 to 2015-16, but pencilled in a consolidation worth 1.6 per cent
of GDP in 2016-17, which entered the forecast period for the first time. This was
7
In Box 4.1 of our March 2019 Economic and Fiscal Outlook, we compared discretionary policy responses to the overall revisions in our
underlying borrowing forecasts. It has been a stated objective of fiscal policy to allow automatic stabilisers to operate freely since Gordon
Brown’s 1998 ‘Code for Fiscal Stability’ and so in this updated analysis, we have excluded the cyclical component of our forecast revisions
below.

Fiscal policy risks
almost entirely comprised of cuts to DEL spending outside the Spending Review period.
(This extra consolidation in 2016-17 is not captured in the methodology underpinning
the chart, because there was no prior forecast for 2016-17 to be revised). The lack of
further up-front tightening is likely to have reflected the wider macroeconomic climate
at the time, such as the intensification of the euro area debt crisis.
• At the March 2015 Budget (the last Coalition Budget before the May 2015 election),
the Government more than spent the improvement in our underlying structural
borrowing forecast, partly reflecting higher departmental spending plans to keep total
spending from falling to a post-war low as a share of GDP (as described above).
• At every fiscal event since the EU referendum, the Government has loosened fiscal
policy (though sometimes only very marginally). In both November 2016 and
November 2017 policy was loosened in spite of a worsened outlook for the pre-
measures structural deficit. In October 2018, the Government spent the underlying
improvement in our structural borrowing forecast on a one-for-one basis, largely by
raising spending on the NHS (in an announcement that preceded the forecast).
Chart 8.4: Discretionary policy responses to our underlying forecast revisions

1.0 Structural improvement, Structural deterioration,
Policy giveaway Policy giveaway
October 2018
March 2015
0.5
Discretionary policy change (per cent of GDP)
November 2017
November 2016
0.0
November 2011
-0.5
Structural improvement, Structural deterioration,

Policy takeaway
-1.0 Policy takeaway
-1.0 -0.5 0.0 0.5 1.0 1.5 2.0
Underlying structural forecast revision (per cent of GDP)
Note: Reflects the cash revision over the last 5 years of the forecast, rebased to each historic nominal GDP forecast.
Source: OBR
Conclusions
8.21 As we set out in each EFO, the uncertainty surrounding our central forecasts is large and
each is therefore liable to substantial revision over time. Clearly the overall impact of
discretionary policy has been to substantially lower overall borrowing since 2009-10. But
the analysis above reveals a pattern in how the Government typically responds to our
underlying structural forecast revisions, but also that the exceptions to this pattern all lie in
one direction – with a looser policy setting – which could generate an upward source of bias
in revisions to the path of debt over time.

Fiscal policy risks
Do governments stick to their policy announcements?

8.22 The analysis above looks at announced plans at each fiscal event. These may reflect
changes to tax and spending as well as revisions to previous plans, so it is important to
track such revisions. Chart 8.5 shows the cumulative effect of announced discretionary
policy on each fiscal year since 2010-11, as it was announced at each fiscal event.8
• The Coalition Government broadly stuck to its original plans during the 2010
Spending Review period (from 2010-11 to 2014-15).
• The Coalition pencilled in increasing amounts of fiscal tightening beyond the 2010
Spending Review period, driven by unspecified departmental spending cuts. This
planned tightening peaked in December 2014, when spending was projected to reach
its lowest level since the 1930s as a share of GDP. In the accompanying EFO, we
showed the implications of the Government’s stated spending assumptions to illustrate
the policy risk that appeared to be building – and that has since largely crystallised.9
• Following the 2015 election, the Conservative Government then reversed a

considerable proportion of these planned cuts, in particular by raising departmental
expenditure plans in the 2015 Spending Review and since. The Government paid for
some of the higher public services spending via tax rises and welfare spending cuts.
• Some of the planned saving from welfare spending cuts has not materialised, for
example due to tax credit cuts announced in July 2015 that were dropped before
being implemented, and to a series of universal credit giveaways that have in effect
reversed much of saving that was expected from July 2015 universal credit cuts.
• Based on our October 2018 forecast, cumulative discretionary tightening in 2018-19

(relative to the baseline of our June 2010 pre-measures forecast) has fallen by nearly a
quarter from 7.0 per cent of GDP in December 2014 to 5.5 per cent of GDP.
8
The analysis in this section is also derived from the numbers underpinning our Forecast revisions database online. The charts reflect the
cumulative cash impact of announced Government decisions, summed over successive fiscal events since the June 2010 ‘pre-measures’
forecast. They have not been adjusted for subsequent recostings, so simply reflect the estimated costs or yields of individual policies as
they were announced. They are shown in the charts as a proportion of outturn nominal GDP.
9
See Box 4.6 of our December 2014 EFO.

Fiscal policy risks
Chart 8.5: Cumulative effect of discretionary policy since June 2010

1
Public sector net borrowing Years in which policy takes effect:
0
2011-12 2012-13 2013-14
-1
2014-15 2015-16 2016-17
Per cent of outturn GDP
-2
2017-18 2018-19
-3
-4
-5
-6
-7
-8
1
Current receipts (Sign is inverted to show effect on PSNB and is plotted on a different scale)
-1
-2
1
Total managed expenditure
0
-1
-2
-3
-4
-5
-6
-7
-8
Jun Nov Mar Nov Mar Dec Mar Dec Mar Dec Mar Jul Nov Mar Nov Mar Nov Mar Oct
10 10 11 11 12 12 13 13 14 14 15 15 15 16 16 17 17 18 18
Fiscal event
Source: OBR

Fiscal policy risks
8.23 The largest component of the consolidation since 2010 has been cuts in departmental
spending as a share of GDP. Chart 8.6 shows that the June 2010 plan to reduce DEL
spending by around 1 per cent of GDP by 2015-16 (relative to the DEL totals inherited from
the previous Labour Government) were broadly stuck to. From November 2011 to
December 2014, the Coalition pencilled in increasingly large cuts, largely in years that were
outside Spending Review periods (i.e. from 2016-17 to 2019-20). From March 2015
onwards, these cuts were largely unwound as detailed plans were set. In our December
2014 forecast, the cumulative effect of DEL announcements totalled an additional reduction
of around 5 per cent of GDP by 2018-19. Successive announcements since then mean that
the cumulative total impact is now roughly half that at 2.6 per cent of GDP.
Chart 8.6: Cumulative effect of DEL policy decisions since June 2010
1
Cumulative DEL policy change (relative to June 2010 pre-measures baseline)
0
-1
-2
Per cent of GDP
-3
-4
Year in which policy takes effect:
-5
2011-12 2012-13 2013-14
-6
2014-15 2015-16 2016-17
-7
2017-18 2018-19
-8
Jun Nov Mar Nov Mar Dec Mar Dec Mar Dec Mar Jul Nov Mar Nov Mar Nov Mar Oct
10 10 11 11 12 12 13 13 14 14 15 15 15 16 16 17 17 18 18
Source: OBR Fiscal event
8.24 Viewed another way, in terms of changes in cash RDEL spending since 2014-15, these
revisions to plans are particularly striking. In our December 2014 EFO, the Coalition’s RDEL
policy assumption would have seen RDEL spending fall by around £38 billion between
2014-15 and 2019-20. In July 2015, the Conservative Government revised that to assume
RDEL would be roughly flat in cash terms over that period. But since the June 2018 NHS
settlement was announced, RDEL has been set to rise significantly. Our March 2019 forecast
shows it rising by around £26 billion between 2014-15 and 2019-20, a turnaround worth
around £65 billion from December 2014. And it continues rising in cash terms thereafter.

Fiscal policy risks
Chart 8.7: Cumulative changes in RDEL since 2014-15

80
March 2013 December 2014
March 2015 July 2015
60
November 2016 October 2018
March 2019
40
20
£ billion
-20
-40
-60
2014-15 2015-16 2016-17 2017-18 2018-19 2019-20 2020-21 2021-22 2022-23 2023-24
Source: OBR
Conclusions
8.25 This analysis shows that previous Governments tended to make substantial revisions to
already announced policy, particularly when plans lay outside Spending Review periods. In
the recent past, such plans have tended to contain fiscal tightening when first announced,
but have been relaxed as the time comes for assumed totals to be replaced with firm plans.
8.26 As we set out in each EFO, our forecasts are conditioned on the Government’s announced
tax and spending policies. Our March 2019 forecast shows overall net borrowing falling
from 1.1 per cent of GDP in 2018-19 to 0.5 per cent of GDP in 2023-24, with the fall
largely driven by discretionary policy. Given that the bulk of DEL plans for 2020-21
onwards have not yet been allocated to departments in a Spending Review, past behaviour
suggests there is a material source of policy risk to our central forecast. Spending pledges
being made during the Conservative Party leadership contest point that way too.
Fiscal policy risks over the medium term

8.27 Our March 2019 forecast shows the fiscal mandate being met with a margin of
£26.6 billion in 2020-21. The supplementary target (for debt-to-GDP fall in 2020-21) is
currently met with 3.2 per cent of GDP to spare (giving headroom of around £75 billion),
although this largely reflects the impact of Term Funding Scheme loans being repaid in the
target year. Excluding that, the debt-to-GDP ratio falls by only 1.0 per cent in that year
(which would reduce the headroom to around £24 billion).

Fiscal policy risks
8.28 As discussed in Chapter 6, forthcoming changes to the accounting treatment of student

loans could raise the measured level of borrowing by between £11.2 billion and £13.7
billion over the next five years. This would reduce headroom against the fiscal mandate in
2020-21 to £15.0 billion.10 But as these accounting treatment changes relate to when the
cost of write-offs and interest income is accrued, they do not affect cash flows, so the
change in the debt-to-GDP ratio in the target year would be unaffected.
8.29 The Government’s formal fiscal objective is to balance the budget by 2025-26 and while
our forecast horizon does not currently extend that far, on past forecast performance it now
has around a 40 per cent chance of doing so by 2023-24. But the current Chancellor’s
recent statements suggest that the Government’s goal might be moving away from
achieving budget balance towards the less stretching target of ensuring that debt is falling:
• In a statement to the Treasury Select Committee in March he said “budget balance is

clearly achievable – but it is a choice. It would be equally sustainable to increase RDEL
spending growth still further; reduce taxation levels; or increase capital spending – or a
combination of these measures. The critical point is that Britain once more has choices
in planning its fiscal future, whilst ensuring that debt will continue to fall.”
• In a letter to Conservative Party leadership candidates in June, he asked them all “to
pledge that if you are the next prime minister your government will, at a minimum, have
a clear commitment to keeping our national debt falling every year, and to maintain the
current limit of the deficit at 2 per cent of GDP at least through 2021-22.”
8.30 Both leadership candidates have set out policy proposals with a potential total cost in the
low tens of billions of pounds,11 with few accompanying suggestions of ways to cut spending
or raise revenue. It remains to be seen what the victorious candidate and his Chancellor will
actually implement in future fiscal events, which will presumably depend in part on near-
term Brexit developments. But it seems highly likely from their statements to date that the
new administration will loosen fiscal policy and adopt less ambitious fiscal objectives.
How far can the primary deficit rise before debt stops falling?
8.31 Between July 2015 and November 2016, the Government aimed for debt to fall in every
year from 2015-16 to 2019-20. Given the recent comments about ensuring debt falls each
year, we now look at how far primary borrowing could rise before this constraint would bite.
8.32 Based on the assumptions that underpin our March 2019 forecast, Chart 8.8 sets out how
much higher primary borrowing would need to be for debt not to fall in any year of the
forecast period as a share of GDP.12 This estimate includes the extra debt interest that would
be incurred by higher primary borrowing, but does not adjust for any other factors. Clearly
10
In the May 2019 Public sector finances release, the ONS published its first provisional estimate of the impact of the new accounting
treatment of student loans. It put the impact on the deficit in 2018-19 as an increase of £10.6 billion, close to the £10.5 billion we
estimated in our March 2019 forecast.
11
For a discussion of some of the proposals made see Jeremy Hunt's tax and spending policies: what would they cost and who would
benefit?, IFS, June 2019, and Boris Johnson's tax policies: what would they cost and who would benefit?, IFS, June 2019.
12
‘Debt falling’ is defined as a year-on-year change of minus 0.1 per cent of GDP on a rounded basis, which is typically the way it has
been viewed by successive governments since the profile of the debt-to-GDP ratio was adopted as a target by the Coalition in 2010.

Fiscal policy risks
the particular drivers of such a rise in borrowing could have other implications for the debt-
to-GDP ratio. If it were an underlying forecast revision, it would probably be driven by a
weaker outlook for GDP growth, which would put further upward pressure on the debt-to-
GDP ratio. If it reflected discretionary policy, then it could have partly offsetting indirect
effects that supported GDP, although the size of the effect would be sensitive to the
composition and timing of the overall package, as well as any monetary policy response.
8.33 Abstracting from those factors, the chart shows that:
• The large impact of Bank of England schemes (particularly the Term Funding Scheme)
on the path of debt in 2020-21 and 2021-22 mean that primary borrowing could rise
by a very large amount in those years before the headline debt-to-GDP ratio would
stop falling (by roughly £70 billion and £95 billion in those years respectively).13
• Ignoring those schemes, primary borrowing would have to rise by around £22 billion
in 2020-21 before debt stops falling in that year. This would also lead to the current
fiscal mandate being missed (under the assumptions of our March 2019 forecast, plus
the assumed impact of student loan accounting treatment changes).
• Between 2021-22 and 2023-24, primary borrowing would have to rise by between
£21 billion and £25 billion before debt (excluding the impact of Bank of England
schemes) stops falling in any of those years. The profile is somewhat lumpy, reflecting
other factors driving the year-on-year change in the debt level (i.e. the flow of net
lending), as well as the starting point of debt in each year.
Chart 8.8: Maximum levels of borrowing consistent with the debt-to-GDP ratio falling
6 88
Primary borrowing Debt-to-GDP ratio
5 86
84
4
82
Per cent of GDP
Per cent of GDP
3
80
2 78
1 76
March 2019
74
0 Keep debt (ex. BoE schemes) falling
72
-1 Keep debt (inc. BoE schemes) falling
70
Outturn
-2 68
2015-16 2017-18 2019-20 2021-22 2023-24 2015-16 2017-18 2019-20 2021-22 2023-24
Note: Primary borrowing equals PSNB less net debt interest payments.
Source: ONS, OBR
8.34 Relative to our average underlying forecast revision, a deterioration of £24 billion a year
(the average annual change in primary borrowing in Chart 8.8) would be large – roughly
as large as that in our November 2016 forecast (following the EU referendum), although
13
See paragraph 4.146 of our March 2019 EFO for more information on these schemes.

Fiscal policy risks
smaller than our largest upward revisions to borrowing, which came in November 2011
and December 2012. In absolute terms, our average two-year ahead underlying forecast
revision is around £9 billion.
8.35 In a world in which debt was falling only very gradually relative to GDP in our central
forecast, there would necessarily be a roughly one-in-two chance that revisions to our
underlying forecasts would result in it being forecast to rise instead. From a fiscal policy risks
perspective, if our central forecast were to show a government running close to whatever
target it had chosen, any future forecast deterioration would need to be offset on at least a
one-for-one basis for the target to remain on track to be met. This no doubt at least partly
explains why Chancellors have aimed for a reasonable amount of headroom against their
fiscal targets. But as we have shown, since June 2010 governments have only attempted to
offset around a third of our underlying forecast movements via discretionary measures –
and this relationship has tended to be weaker where other factors have been at play, such
as impending elections or the change of administration that followed the EU referendum.

consider when managing its fiscal risks. Among them, governments’ tendencies to:
• Revise fiscal rules in line with movements in the forecast.
• Respond asymmetrically to movements in our underlying forecasts.
• Assume cuts outside Spending Review periods, but revise totals up when plans are set.
8.37 When assessing the outlook for the public finances over the medium and long term, does
the Government regard these or other issues as important for its risk management strategy
and, if so, how does it intend to address them?

9 Climate change
Introduction
9.1 Evidence of climate change is clear in rising average temperatures – 17 of the 18 hottest
years recorded globally since 1880 occurred in the past 18 years (Chart 9.1). The global
rise of around 1oC relative to the pre-industrial period has been matched in the UK, with
most of the warming occurring since the 1970s (yielding milder winters and hotter
summers). Sea levels around the UK have risen by 15 to 20 centimetres since 1900, with the
rate of change picking up since 1990. In the UK, the risk of heatwaves and severe flooding
(due to the greater moisture levels that a warmer atmosphere can hold) has risen. By
contrast, cold snaps in winter are now around half as likely as they were in the 1960s.1
Chart 9.1: Global average temperatures

1.4
1.2
Average global temperature relative to 1880-1910
1.0
average (Degrees Celsius)
0.8
0.6
0.4
0.2
0.0
-0.2
-0.4
1880 1890 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010
Source: NASA
9.2 The Intergovernmental Panel on Climate Change (IPCC) has concluded that the increase in
man-made greenhouse gas emissions since the pre-industrial era, driven largely by
economic and population growth, has led to unprecedented concentrations of greenhouse
gases in the atmosphere and that these emissions are extremely likely to have been the
dominant cause of the observed global warming since the mid-20th century.2
1
All the UK-specific evidence in this paragraph has been drawn from the Committee on Climate Change, UK Climate Change Risk
Assessment 2017 Synthesis report: priorities for the next five years, July 2016.
2
International Panel on Climate Change, Fifth Assessment Report of the International Panel on Climate Change, 2014.

Climate change
9.3 The global response to this challenge is embodied in the 2015 Paris Agreement. Its central
aim is to strengthen action to mitigate climate change so as to keep the global temperature
rise this century to below 2oC above pre-industrial levels, and to pursue efforts to limit the
rise further to 1.5oC. The UK Government is party to the Paris Agreement and has recently
adopted a more ambitious target of reaching net zero greenhouse gas emissions by 2050.
9.4 Our 2017 Fiscal risks report (FRR) acknowledged climate change as a potentially significant
source of fiscal risk that we had not analysed. Drawing on the Bank of England’s
assessment that climate-related risks to financial stability arise through two primary
channels – physical effects and the impact of the transition to a low-carbon economy – we
noted that the fiscal risks from climate change are likely to arise through similar channels.
When responding to the FRR in Managing fiscal risks, the Treasury did not mention climate
change. Indeed, the Committee on Climate Change’s Climate change risk assessment
2017: Evidence report stated that, as of mid-2016 when it was finalised, there was “no UK-
wide assessment quantifying indirect or macroeconomic effects of climate risks”.3 The
Government’s 2017 Climate change risk assessment did not discuss fiscal risks (Box 9.1).
9.5 Over the past two years, the Bank of England has further developed its approach to
assessing climate-related risks to financial stability, while the Network for Greening the
Financial System, a collaboration between many of the world’s central banks and banking
supervisors, has published ‘a call for action’ on climate change as a source of financial
risk.4 This work – and the planned next steps – will help us undertake future assessments of
the nature and scale of the fiscal risks associated with climate change.
9.6 Considering climate change from the perspective of fiscal risk, while also respecting
Parliament’s instruction to us not to consider alternative policy paths, requires us to set aside
some important questions. For example, tax and spending decisions that help us adapt to
the changing climate or discourage greenhouse gas emissions will affect the public finances
directly, while also affecting the path of climate change itself (especially where taken in
concert with the governments of other countries). But it will not be for us to consider the
merits of the choices made or that are available. Instead, we can seek to illustrate the risks
associated with projected climate trends and current policies. With this in mind, this chapter:
• sets out a framework for considering climate-related fiscal risks;
• discusses the nature of climate-related fiscal risks and their relative scale;
• sets out next steps we plan to take in collaboration with other institutions; and
• summarises issues raised in the chapter that the Government may wish to address in
its response to this report.
3
See Section 8.3.1 in Committee on Climate Change, UK Climate Change Risk Assessment 2017: Evidence Report, July 2016.
4
Network for Greening the Financial System, First comprehensive report, A call for action: Climate change as a source of financial risk,
April 2019.

Climate change
Box 9.1: The UK Government’s Climate Change Risk Assessment

The Climate Change Act 2008 provides the statutory underpinning for the UK’s approach to
cutting greenhouse gas emissions and adapting to the changing climate. Among other things, it
requires the Government to assess the risks and opportunities arising from climate change once
every five years. The most recent assessment was published in 2017.a It drew on over 2,000
pages of evidence gathered by the Committee on Climate Change (CCC).b In its synthesis of this
evidence, the CCC identified six priority risks where more action and/or further research was
required. The Government accepted almost all the CCC’s recommendations (with the exception
of some related to food security). Those six priority risks were:
• Risks to communities, businesses and infrastructure from flooding and coastal change.
Damages from flooding and coastal change were estimated to cost around £1 billion a
year, with separate but related risks associated with episodic heavy rainfall, river flows,
sea levels, tidal surges and coastal erosion.
• Risks to health, well-being and productivity from high temperatures. With both average
and extreme temperatures expected to rise further, there was a need to adapt existing
buildings to cope with higher temperatures, while the risks associated with overheating in
hospitals, care homes and the like were unknown.
• Risks of shortages in the public water supply, and for agriculture, energy generation and
industry. Scenario analysis suggested that demand for water could materially outstrip
available resources in many areas by mid-century, thanks to changes in rainfall patterns,
increased evaporation and soil aridity, coupled with population growth raising demand.
• Risks to natural capital, including terrestrial, coastal, marine and freshwater ecosystems,
soil and biodiversity. The affinity of animals and plants to different regions would be
altered by environmental change. High-grade agricultural land was projected to
deteriorate – perhaps significantly – due to soil aridity, water scarcity and other factors.
• Risks to domestic and international food production and trade. Increased risk of extreme
weather would affect food production and supply chains. And while temperature rises
could present opportunities for greater domestic food production, such benefits would
most likely be limited by increasing soil aridity and water scarcity.
• Risks from new pests and diseases and from invasive non-native species. Pathogens
already present in the UK at low levels might become more prevalent, while new ones
might arrive from overseas. A warmer UK would also be at greater risk of diseases
carried by mosquitos and ticks. Here, the CCC noted particularly large uncertainties.
Each of these climate-related risks comes, to varying degrees, with associated fiscal risks. Floods
and heatwaves disrupt activity, affecting tax revenues. Taking mitigating action will generate
more calls on the public purse. And the consequent changes in the structure of the economy will
affect both the reliability of tax revenues and the allocation of public spending.
a
HM Government, UK Climate Change Risk Assessment 2017, January 2017.
b
Committee on Climate Change, UK Climate Change Risk Assessment 2017 Evidence Report, July 2016.

Climate change
A framework for assessing climate-related fiscal risks

9.7 Climate change is such a pervasive phenomenon that it potentially affects almost every
aspect of the public finances. And the policies to mitigate and adapt to climate change have
both direct fiscal implications and indirect ones via their effects on the economy (and the
path for climate change). It is therefore helpful to have a framework in which to think about
climate-related fiscal risks to make the task tractable. The Bank of England’s approach to
climate-related financial stability risks provides a relevant framework for our purposes.
9.8 Risks to financial stability and to fiscal sustainability have much in common. The banking
system is exposed to risks facing banks’ customers and counterparties, which in effect make
up the entire economy. Those same actors – and the financial system itself – are the ultimate
source of many fiscal risks. So the approach to assessing climate-related risks to financial
stability and to fiscal sustainability will overlap to a considerable degree. The key difference
is that tax and spending policies can affect the path of climate change – especially when
pursued globally – so the risks are to some extent endogenous, whereas climate change is
exogenous to financial stability. For example, the OECD argues that ‘ambitious’ mitigation
policies will reduce both the damage from, and the probability of, malign outcomes.5
The Bank’s framework for assessing climate-related risks to financial stability

9.9 The Bank of England has developed its response to climate-related financial stability risks
over the past four years, starting with Governor Carney’s September 2015 speech on the
nature of the challenge for central banks posed by climate change.6 He described how its
potentially large costs lie beyond the horizon of most policymakers and set out the channels
through which it can affect financial stability, establishing the framework we will use too:
• Physical risks were described as “the impacts today on insurance liabilities and the
value of financial assets that arise from climate- and weather-related events, such as
floods and storms that damage property or disrupt trade”.
• Transition risks were described as “the financial risks which could result from the
process of adjustment towards a lower-carbon economy. Changes in policy, technology
and physical risks could prompt a reassessment of the value of a large range of assets
as costs and opportunities become apparent.”
In the speech, he also separated out ‘liability risks’ – “the impacts that could arise tomorrow
if parties who have suffered loss or damage from the effects of climate change seek
compensation from those they hold responsible.” These were specific to the insurance sector.
9.10 Alongside the Governor’s speech, the Bank published a more detailed assessment of how
climate change could affect the Prudential Regulation Authority’s statutory objectives with
regard to insurance.7 This assessment was produced at the invitation of Government and
5
OECD, The Economic Consequences of Climate Change, November 2015.
6
‘Breaking the Tragedy of the Horizon – climate change and financial stability’, Speech given by Mark Carney, Governor of the Bank of
England, Chairman of the Financial Stability Board, Lloyd’s of London, 29 September 2015.
7
Bank of England, The impact of climate change on the UK insurance sector, A Climate Change Adaptation Report by the Prudential
Regulation Authority, September 2015.

Climate change
informed the 2017 Climate change risk assessment. It was followed by a report on the risks
from climate change to the banking sector.8 Further reports include a working paper
discussing the impact of climate change and the transition to a low-carbon economy on
monetary policy and financial stability, published in 2016,9 and a Quarterly Bulletin article
outlining the Bank’s response to climate change more broadly.10 This touched on fiscal risk,
noting that “significant uninsured losses from physical risk could also result in economic
disruption at a national level, reducing tax revenues and increasing fiscal expenditures.”
9.11 In January 2018, a Bank working paper reviewed the analysis of macroeconomic risks
deriving from climate change.11 It noted the importance to those responsible for fiscal and
monetary policy, and various other policy areas, of understanding the impact on the
economy of climate change. The paper looked at physical risks due to extreme weather
events and due to gradual global warming, as well as transition risks. The macroeconomic
consequences of these were broken down into those affecting the supply-side and demand-
side of the economy – i.e. affecting potential output and the output gap respectively.
9.12 Table 9.1 is taken from the Bank’s paper. Possible fiscal consequences of such
macroeconomic risks were discussed in Chapter 3 of our 2017 FRR and Chapter 2 of this
report. That discussion would point to the largest risks being those associated with lower
potential output growth reducing growth in tax bases and thus the resources available to
finance future public spending. These pressures typically build slowly, giving governments
time to factor their effects into fiscal policy, but also scope to put off necessary action.
Table 9.1: Examples of macroeconomic risks from climate change

Physical risks
Type of shock/impact From extreme From gradual Transition risks
weather events global warming
Uncertainty about climate 'Crowding out' from climate
Investment
events policies
Increased risk of flooding 'Crowding out' from climate
Demand Consumption
to residential property policies
Disruption to import/export Distortions from
Trade
flows asymmetric climate policies
Loss of hours worked due Loss of hours worked due
Labour supply
to natural disasters to extreme heat
Energy, food and Food and other input
Risks to energy supply
other inputs shortages
Diversion of resources from Diversion of resources from
Damage due to extreme
Supply Capital stock productive investment to productive investment to
weather
adaptation capital mitigation activities
Diversion of resources from Uncertainty about the rate
Diversion of resources from
innovation to of innovation and
Technology innovation to adaptation
reconstruction and adoption of clean energy
capital
replacement technologies
8
Bank of England, Transition in thinking: The impact of climate change on the UK banking sector, September 2018.
9
Bank of England Staff Working Paper No. 603, Let’s talk about the weather: the impact of climate change on central banks, Batten S.,
Sowerbutts R. and Tanaka, M., May 2016.
10
Bank of England Quarterly Bulletin, 2017 Q2, The Bank of England’s response to climate change, Scott M., van Huizen, J. and Jung, C.
11
Bank of England Staff Working Paper No. 706, Climate change and the macro-economy: a critical review, Batten, S., January 2018.

Climate change
Fiscal risks from extreme weather events

9.13 The 2006 Stern Review estimated that the global cost of extreme weather events could reach
0.5 to 1 per cent of world GDP a year by the middle of this century. In developing countries,
these costs primarily reflect income losses for those affected. In advanced economies, the
cost of infrastructure damage and depreciation dominate.12 The Bank estimates that the
number of weather-related insurance payouts around the world has tripled since the 1980s,
while the insurance-related losses associated with these events have increased five-fold.13
And for the UK, the Met Office has estimated that the probability of a summer heatwave like
that of 2018 is now around 30 times higher than under pre-industrial climate conditions.14
9.14 The Bank’s 2018 working paper noted that studies of the economic effects of natural
disasters provide a useful guide to the effects of extreme weather events.15 It concluded that
there was general agreement on their negative short-term effects on GDP, and that while
evidence on their longer-term effects was both scarcer and more mixed, the largest number
of studies pointed to them generating longer-term GDP losses too. But it is important to note
that these studies are partial, often covering only a small number of the transmission
channels (i.e. the impact of temperature on productivity) and do not usually account for the
possibility of climate tipping points accelerating impacts in a non-linear way.16
9.15 Fiscal risks from extreme weather events could take many forms. To the extent that they
reduce economic activity over the medium term – rather than merely shifting it between time
periods – tax receipts would be hit and the social security bill would rise. Were the
Government to choose to meet some of the costs of repairing and rebuilding private
property – or to spend more to restore assets of its own that were damaged – public
spending could be higher. And if critical infrastructure were affected – or its owners placed
at financial risk – private sector assets could be taken onto the public balance sheet.
9.16 The scale of such risks would depend on the precise nature of the extreme weather event.
The Environment Agency estimates that the 2007 summer floods in the UK were associated
with the largest economic costs of recent major floods (£3.9 billion in 2015 prices). The
2015/16 winter floods, which were the most extreme on record by intensity of rainfall,
generated a smaller cost (£1.6 billion). In both cases the estimated cost is dominated by
damage to property and transport infrastructure.17 But the fiscal implications were modest
as they only affected parts of the country and their local economies. Given the concentration
of tax receipts generated by economic activity close to the Thames estuary, a coastal surge
that led to severe flooding in this area could have more significant fiscal consequences.
12
Stern Review, The Economics of Climate Change, October 2006.
13
Bank of England, The impact of climate change on the UK insurance sector: A Climate Change Adaptation Report by the Prudential
Regulation Authority, September 2015.
14
Met Office, Chance of summer heatwaves now thirty times more likely, December 2018.
15
16
As discussed by the NGFS in, A call for action: Climate change as a source of financial risk, April 2019.
17
Environment Agency, Estimating the economic costs of the 2015 to 2016 winter floods, January 2018.

Climate change
Fiscal risks from adaptation to climate change

9.17 The scale of the risks associated with gradual global warming will depend on the extent to
which temperatures rise. If the most ambitious of the Paris goals were met and temperatures
stabilised at 1.5oC above pre-industrial levels – and therefore only around 0.5oC higher
than temperatures witnessed in recent years – the nature of climate-related risks might be a
more severe version of those already being seen. By contrast, if temperatures were to rise to
4oC above pre-industrial levels, consistent with the IPCC’s unmitigated climate change
scenario,18 the risks could be much greater and much more difficult to assess. These could
include much greater international migration flows and induced periods of conflict.19
9.18 In its 2018 working paper, the Bank noted that, from the perspective of the wider economy,
gradual rises in temperatures are generally viewed as a potential source of economic loss
as they tend to reduce the productivity of workers and agricultural crops.20 (For particular
countries at particular times this may not be the case, as the IMF’s cross-country analysis
described below illustrates.) The main channel along which this effect takes place is the
diversion of resources from productive capital and innovation to ‘adaptation capital’. This is
a climate-related risk to what we routinely identify as the most important and uncertain
judgement underpinning our medium-term forecasts: underlying productivity growth.
9.19 The extent to which adaptation to global warming will affect economic growth is subject to
considerable uncertainty. The Bank noted that studies of the economic consequences of
adapting to gradual temperature rises typically deploy models that combine climate-related
assumptions (about greenhouse gas emissions and their consequences for the climate) with
economy-related assumptions (about how climate change affects output). The ‘damage
function’ assumptions that drive the economic results of these models are important and
subject to huge uncertainty.21 In particular, long-term economic costs of higher temperatures
vary greatly depending on whether they are assumed to have static or dynamic effects – i.e.
whether they affect only the level of activity, with a transitory effect on growth, or the growth
of activity, with an ever-increasing effect on the level.22 As the Bank notes, even small
growth effects ultimately dominate large level effects as their impact cumulates over time.
9.20 In a cross-country study, the IMF estimated that the effect of a 1oC average temperature rise
would be particularly harmful in Africa, India and much of South East Asia, but beneficial in
18
The IPCC’s ‘RPC8.5’ representative control path sees greenhouse gas emissions continue to rise leading to high concentrations levels in
the atmosphere. As a result, global temperatures would be projected to rise by 3.7oC (likely range 2.6 to 4.8oC) by the end of the century.
See Table SPM.2 in IPCC, ‘Summary for Policymakers’ in: Climate Change 2013: The Physical Science Basis. Contribution of
Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change, 2013.
19
See, for example, Raleigh, C., Jordan, L. and Salehyan, I., Assessing the Impact of Climate Change on Migration and Conflict, World
Bank Group, 2008, and Sawas A., Workman, M. and Mirumachi, N., Climate change, low-carbon transitions and security, Grantham
Institute Briefing paper No 25, March 2018.
20
21
For example, in Climate Change Policy: What Do the Models Tell Us? Journal of Economic Literature, 2013, Robert Pindyck concludes
that the answer is “very little”, while Lord Stern has warned that models “may be profoundly misleading on the issues of great significance”
(in The Structure of Economic Modeling of the Potential Impacts of Climate Change: Grafting Gross Underestimation of Risk onto Already
Narrow Science Models, Journal of Economic Literature, 2013).
22
See also, for example, Dietz, S. and Stern, N., Endogenous growth, convexity of damages and climate risk: how Nordhaus’ framework
supports deep cuts in carbon emissions, June 2014.

Climate change
Canada, Russia and Scandinavia. The effect on UK GDP would be insignificantly small.23
But this study was based on historical correlations between temperature and per capita GDP
growth. It did not factor in possible effects of rising sea levels, extreme weather events or
spillovers from more vulnerable economies. And even if the economy-wide effect were
limited, that could mask potentially large gains and losses at the sector level. Any effects
might become significant with greater increases in temperature, given the non-linear
relationship between temperature changes and output.24 The OECD’s central projection for
its baseline scenario, where no further action is taken to mitigate climate change, is damage
of 2 per cent of global GDP by 2060. In this scenario, changes in crop yields and labour
productivity are expected to deliver the largest hit to global GDP.25
9.21 As well as the fiscal implications of any effects of climate change on GDP and the tax base,
adaptation can generate more direct fiscal costs. For example, the Environment Agency has
estimated that an average of £1 billion a year will need to be spent on flooding and coastal
infrastructure in order to be resilient to a 4oC rise in global temperatures.26 It projects that a
continuation of current planning outcomes – i.e. the decisions taken by local authorities
about where new properties can be built – would result in the number of properties exposed
to flood risk almost doubling from 2.4 million at present to 4.6 million in 2065. But it
estimates that the associated increase in flood damages would be limited to 4 per cent, due
to planning policy mitigation measures such as raised floor levels.27 Of course, in a 4oC
world, the public spending implications of conflict and mass migration could represent a far
more significant source of fiscal risk than those of flood defences and coastal protection.
Fiscal risks from the transition to a low-carbon economy

9.22 In some ways, the nature of fiscal risks from mitigating emissions matches those associated
with adaptation: i.e. they stem from both the direct fiscal implications of policies designed to
lower emissions and from the indirect effects of structural change in the economy as it
moves to a low-carbon future. The extent of these will be guided by the degree of emissions
reduction being targeted and the efficiency and predictability of the measures adopted. The
extent to which resources in the economy shift to activities that will prosper in a low-carbon
global economy poses a different type of risk. Failure to adjust flexibly could damage
growth prospects by leaving resources in sectors and technologies where growth prospects
have weakened. Such risks become more important as the costs of green technologies fall.28
9.23 The Bank’s 2018 working paper argued that the policy-driven transition to a low-carbon
economy in order to meet the Paris commitments on limiting global warming will entail its
own set of risks.29 Such policies have typically been seen as an economic burden in the
shorter run, diverting resources to climate change mitigation from other productive activities
23
IMF, World Economic Outlook, October 2017. See Annex Figure 3.3.1. Countries for which the estimated effect of temperature increase
on real per capita output was between -0.31 and +0.32 per cent were deemed not to be statistically significant.
24
Burke, M., Hsiang, S. and Miguel, E., Global non-linear effect of temperature on economic production, Nature, October 2015.
25
OECD, The Economic Consequences of Climate Change, November 2015.
26
Environment Agency, Draft National Flood and Coastal Erosion Risk Management Strategy for England, May 2019.
27
Environment Agency, Long-term investment scenarios, 8 May 2019.
28
See, for example, Dechezleprêtre, A., Martin R. and Mohnen, M., Knowledge spillovers from clean and dirty technologies: a patent
citation analysis, September 2013.
29

Climate change
and raising input costs, although in the long run successful mitigation would reduce the
economic costs of gradual global warming. The OECD notes that in a delayed-action
scenario, where climate change mitigation accelerates only after 2025, GDP losses are
estimated to be 2 per cent on average across the G20 after 10 years.30
9.24 As the cost of green technologies falls, and as policymakers around the world look to orient
their economies to capitalise on the growth in their use, mitigation policies are increasingly
seen as a source of potential growth. The same OECD study argues that “bringing together
the growth and climate agendas, rather than treating climate as a separate issue, could add
1% to average economic output in G20 countries by 2021 and lift 2050 output by up to
2.8%. If the economic benefits of avoiding climate change impacts such as coastal flooding
or storm damage are factored in, the net increase to 2050 GDP would be nearly 5%.”
9.25 The macroeconomic risks of climate policy will depend on how it is managed: well-signalled
and orderly policies that allow time for the economy to adjust and for technological
advances to reduce costs might pose little risk; unexpected or inconsistent policy changes
could have more severe consequences in terms of higher energy prices reducing economic
activity or stranded assets depleting wealth and generating financial stability risks.
9.26 The Stern Review argued that reducing greenhouse gas emissions sufficiently to avoid the
worst impact of climate change (an 80 per cent reduction on then current levels) could cost
around 1 per cent of global GDP a year.31 More recently, the CCC estimated that the
annual costs of reducing the UK’s greenhouse gas emissions to net zero by 2050 would be
around 1 to 2 per cent of GDP, but that the structural changes underlying these modest
aggregate costs would be very large.32
9.27 In a letter to the Prime Minister that was leaked to the press, the Chancellor chose to sum
the annual costs, drawing on analysis from the Department of Business, Energy & Industrial
Strategy, to reach a figure of £1 trillion for the total cost – both public and private – of
meeting the target.33 To place this in context, our latest long-term projections suggest that
over the same period the economy will generate around £90 trillion of real GDP and the
state will spend about £40 trillion, so £1 trillion is not as much as it first appears.
9.28 One of the CCC’s advisory groups argued that the macroeconomic costs of deep
decarbonisation are likely to be small globally, and for a fossil fuel importing country like
the UK there could be macroeconomic gains.34 The CCC itself recommended that the
Treasury review how the costs of decarbonising the economy will be met, considering effects
on the public finances and on those affected by the policies that are put in place.
30
OECD, Investing in Climate, Investing in Growth, June 2017.
31
32
Committee on Climate Change, Net Zero: The UK’s contribution to stopping global warming, May 2019.
33
Letter from the Chancellor of the Exchequer to the Prime Minister, 2050 Net Zero Emissions Target, May 2019.
34
Modelling of the macroeconomic consequences of ‘deep decarbonation’ are discussed in an accompanying ‘Report to the Committee
on Climate Change from the Advisory Group on Costs and Benefits of Net Zero’ by Paul Ekins. It illustrates how different assumptions can
generate positive or negative GDP effects, depending in particular on how innovation and technological change are assumed to evolve.

Climate change
9.29 Successful mitigation of greenhouse gas emissions can create its own fiscal risks, to the
extent that tax revenues are currently dependent on emission-generating activities. We have
considered some of these issues in previous reports, particularly in respect of vehicle-related
taxes. For example, our 2014 Fiscal sustainability report considered the effects of improving
fuel efficiency of vehicles on fuel duty revenue and the effect of reducing emissions on
vehicle excise duty (VED) receipts. We updated these analyses in our 2017 FRR, noting how
reforms to VED in the intervening period had reduced this source of fiscal risk. Receipts from
the climate change levy and EU emissions trading system (or UK-equivalent were the UK to
leave the ETS after Brexit) would also fall if carbon emissions were to fall to ‘net zero’. If the
financial system were to find itself overly exposed to ‘old’ sectors and technologies, its
revenues could be hit by the consequences of stranded assets and under-priced risks, which
in turn could hit tax revenues from one of the most tax-rich sectors of the economy.
The nature of climate-related fiscal risks

9.30 From a broad public policy perspective, the reasons for government to address climate
change are simple – climate change results from a market failure, namely that the costs to
society of greenhouse gas emissions are shared widely, including across generations, rather
than being borne by those generating the emissions. Indeed, the Stern Review described
climate change as “the greatest and widest ranging market failure ever seen.”35 As well as
this, governments increasingly seek to ensure that their economies are competitively placed
to take advantage of green growth opportunities and that they are not locked into
infrastructure or institutions that could become devalued or stranded in the future.
9.31 For the UK Government, fiscal risks stem from the consequences of climate change – the
damage wrought by extreme weather events and the cost of adapting to a slowly changing
climate – and from policy measures designed to reduce the UK’s own greenhouse gas
emissions and to ensure that the economy is well placed to thrive in a low-carbon world.
The former are essentially unavoidable, although their scale can be influenced. The latter
reflect policy choices, but are framed by the international agreements to which the UK is
party and its own zero net emissions target for 2050. The more successful the global
reduction of emissions, the lower the cost of weather events and adaptation measures.
9.32 The Bank of England’s physical/transition framework for assessing climate-related risks to
financial stability risks maps well onto the categories of fiscal risk we identified in our 2017
FRR. Extreme weather events represent potential ‘shocks’ with short-term consequences that
are unlikely to be foreseen in our central forecasts because of their uncertain scale and
timing. The fiscal consequences of adaptation costs and mitigation policies represent
‘pressures’ on the public finances that are foreseeable and will typically build slowly,
affording time for policymakers to address them and adjust as new information emerges. In
the case of green growth opportunities, the risk relates to potential output growth foregone.
35

Climate change
9.33 Relative to the two definitions of fiscal risk we use in this report, climate-related risks include:
• Conventional medium-term forecast risks: the Government allocates some spending to

climate change adaptation and mitigation (the Treasury estimates that in 2017-18
such spending amounted to around £10½ billion – £9¼ billion on mitigation and
£1¼ billion on adaptation – £7¼ billion of which went on spending for low carbon
electricity, funded via levies on consumer and business energy bills); it has established
‘Flood Re’ to take on the flood risk element of eligible insurance products (although
unusually this is structured such that tail risk sits with the private sector); and it raises
revenue from environmental taxes like the climate change levy (around £2 billion a
year). It might also be subject to climate-related legal challenges that generate future
costs.36 So medium-term fiscal risks include unexpected costs that raise spending
relative to plans or shocks that cause revenues to fall short of forecast. And were an
extreme weather event to be sufficiently large to cause material disruption to the
economy and tax revenues, this too would be a source of fiscal risk.
• Risks to fiscal sustainability: longer-term climate-related fiscal risks can be considered

from the perspective of the economy – such as the extent to which redirecting resources
to adaptation and mitigation activities weighs on productivity growth or the
opportunities from reorienting the economy towards green technologies boost it – or
from the direct pressures placed on tax revenues and public spending – such as the
declining fuel duty tax base or requirement for extra spending on flood defences.
9.34 Taking each issue in isolation, the material reviewed in this chapter might suggest the fiscal
risks associated with climate change are not large relative to others we have assessed in this
and our previous FRR. Extreme weather events, such as flooding or heatwaves, could have
wider economic costs through damage to property or lost working days that hit tax revenues
or generate calls on public spending. But it seems unlikely that those costs would be on the
same scale as a recession or financial crisis. Redirecting resources to adaptation and
mitigation measures will have longer-term implications for the composition of revenue and
spending – and for the structure of the economy itself. But the fiscal implications are likely to
be small relative to those of population ageing or the cost pressures seen in health care. But
this relatively sanguine conclusion might simply reflect the difficulty in seeing through to the
full systemic consequences of significant global warming – where interconnections and
associated amplifying mechanisms might be more important than is initially apparent. And
of course the severity of any impacts will be correlated with the extent of global warming.
9.35 Others have argued that associated fiscal risks are likely to be manageable. For example:
36
In Global trends in climate change legislation and litigation: 2018 snapshot, May 2018, the LSE’s Grantham Research Institute counted
225 climate-related cases brought against governments or their representatives around the world. These cases include those seeking to
overturn decisions taken on climate-related grounds (for example, denying a licence for a coal-fired power plant) and those challenging
the allocation of allowances under emissions trading schemes. There are also more strategic cases, such as those seeking to force courts
to rule on the consistency of countries’ actions with the Paris Agreement.

Climate change
• The Treasury’s 2009 Long-term public finance report discussed climate change. It
argued that the approach to lowering carbon emissions that the Government of the
time was advocating (and that is little changed today) would “ensure that although
there may be some direct pressure on the public finances (e.g. spending on R&D and
investment, and spending to raise awareness and change behaviour) the effect is likely
to be manageable. Policies will also impact on the economy and therefore indirectly
affect the public finances. Again, however, this effect is likely to be modest.”37
• The CCC’s Advisory Group on Costs and Benefits of Climate Change noted more
recently that “where taxing fossil fuels (such as motor fuels in the UK) provides a
significant source of revenue already, the shift away from them and any other carbon-
based fuels that may have been taxed in the meantime will require, in the medium and
long term, a new tax base to be developed. But these fiscal changes are likely to be
gradual and manageable.”38
Box 9.2: Case study: US federal government on climate-related fiscal risks

In November 2016, the US Office of Management and Budget (OMB), in collaboration with
President Obama’s Council of Economic Advisers, presented a preliminary assessment of some
of the fiscal risks associated with climate change.a Guided by the available climate studies and
economic modelling, it included five detailed risk assessments – covering expenditure risks
related to crop insurance, air quality and health care, wildfire suppression, coastal storm disaster
relief and flood risks to Federal government facilities. It also illustrated the potential revenue
costs of climate change reducing the level of US economic activity.
The bottom-up programme-specific expenditure risk assessments pointed to various ways in
which climate change will raise Federal spending in the US. The overall cost estimate was
dominated by the projected increase in spending on coastal disaster relief. In total it was put at
between $9 billion and $28 billion a year by late in this century (around 0.05 to 0.15 per cent of
US GDP in 2016). But the report stressed that this did not represent a complete assessment of
the spending implications of climate change. Important areas were not quantified due to lack of
reliable inputs to model them – for example, the focus on the health costs of air quality changes
was considered likely to reflect “only a slim component of the full fiscal risk related to health care
and public health”. As well as the scale of the risk, the report noted that the variability of calls on
affected spending programmes was likely to increase, causing greater recourse to safety nets
and challenges for expenditure planning.
The indicative revenue hit was calculated top-down using a commonly utilised economic model
that puts the global economic cost of a four-degree climate scenario at 4 per cent of global GDP
and simple assumptions about the US share of global economic activity by the end of the century
and the revenue share of GDP.b It therefore illustrated the revenue loss associated with an
unmitigated climate change scenario relative to a reference scenario assuming no further climate
change. On that basis, the revenue cost in today’s terms by the end of the century was put at
$56 billion to $111 billion a year (around 0.3 to 0.6 per cent of US GDP in 2016).c
37
See Chapter 3 of HM Treasury, Long-term public finance report: an analysis of fiscal sustainability, December 2009.
38
Ekins, P., Report to the Committee on Climate Change of the Advisory Group on Costs and Benefits of Net Zero, May 2019.

Climate change
A comparable calculation for the UK would yield a range of revenue costs in today’s terms of
£16 billion to £32 billion a year by 2100.d
The OMB’s assessment of flood risk to federal property was not included in its expenditure risk
estimates. Instead, it reviewed a sample of all federal property and identified $83 billion of
federal assets located within the 100-year floodplain, $23 billion within the 500-year floodplain
and $62 billion in coastal assets that would be threatened by inundation or otherwise severely
affected in a scenario in which sea levels rose by six feet. The Congressional Budget Office has
estimated more recently that annual expected losses from hurricane winds and storm-related
flooding total $54 billion, with $12 billion of this expected to fall on the public sector.e
a
US Office of Management and Budget, Climate change: The fiscal risks facing the Federal Government, November 2016.
b
Nordhaus, W., The Climate Casino: Risk, Uncertainty, and Economics for a Warming World, 2013.
c
The range was generated by different assumptions about the US share of global GDP by 2100 – which varied from 10 to 20 per
cent, with the higher figure being roughly the US share of global GDP at the time.
d
This is equivalent to 0.7 to 1.5 per cent of GDP. It is considerably higher than the OMB’s figure for the US because it covers total
receipts (which we forecast to be 36.8 per cent of GDP in 2019-20) whereas the OMB figure relates only to US Federal Government
revenues (i.e. excluding state and local governments), which the CBO estimates to have been 17.6 per cent of US GDP in 2016.
e
Congressional Budget Office, Expected Costs of Damage From Hurricane Winds and Storm-Related Flooding, April 2019.
Next steps
9.36 The Bank’s next steps in this area were discussed by Sarah Breedon, Executive Director for
International Banks Supervision, in April 2019,39 coinciding with the publication of the
Network for Greening the Financial System’s first comprehensive report on financial risk
and climate change (see Box 9.3). These steps include supervisory expectations of the
individual banks and insurance companies the Bank regulates.40 Of most relevance to us on
fiscal risks, the Bank will include climate-related factors in its 2021 ‘biennial exploratory
scenario’ – part of the Bank’s financial stability stress testing infrastructure.41 The NGFS is
also developing its framework for scenario analysis, on which the Bank is leading.
9.37 This work underway at the Bank and the NGFS can provide a foundation upon which we
can test the sensitivity of the public finances to alternative climate pathways. We have
therefore launched two collaborative workstreams:
• With the Bank of England: we will engage with the Bank’s analysts as they consider
possible financial stability stress-testing scenarios. This will allow us to consider how
those scenarios can be adapted for fiscal risk analysis.
• Between the OECD network of fiscal councils and the NGFS: we have proposed
working with the OECD Network of Parliamentary Budget Offices and Independent
Fiscal Institutions to engage with the NGFS on its scenario workstream. This will allow
us to draw on the experience and expertise of dozens of central banks and fiscal
councils around the world.
39
‘Avoiding the storm: Climate change and the financial system’, Speech given by Sarah Breeden, Executive Director, International Banks
Supervision, Official Monetary & Financial Institutions Forum, London, 15 April 2019.
40
Bank of England Prudential Regulation Authority, Policy Statement PS11/19, Enhancing banks’ and insurers’ approaches to managing
the financial risks from climate change, April 2019.
41
Bank of England, Financial Stability Report, July 2019.

Climate change
Box 9.3: Climate-related scenario analysis: the NGFS approach

The Network for Greening the Financial System (NGFS) was established in January 2018 and
builds on the work of the Taskforce for Climate-related Financial Disclosures (TCFD). In April
2019 it issued a ‘call for action’ that uses the physical/transition framework to set out next steps
for assessing climate-related risks to financial stability.a It considers the transmission channels
and feedback mechanisms that relate the physical risks from climate change and the transition
risks associated with reducing greenhouse gas emissions to the financial system.
To facilitate central banks’ and banking supervisors’ work on climate-related financial stability
risks, the NGFS is developing a scenario analysis framework for assessing those risks. It has
proposed using four high-level scenarios that capture different settings along two important
dimensions: the strength of the greenhouse gas mitigating policy response; and how smoothly
and foreseeably those actions are taken. This yields the scenario matrix in Figure A.
Figure A: NGFS scenario analysis framework
Strength of response
Based on whether climate targets are met
Met Not met
Disorderly
Transition pathway
Transition risks
Orderly
Physical risks
In the next phase of the NGFS’s work, it plans to develop data-driven narrative and quantitative
parameters on which to base these scenarios, including proposals for key assumptions about
policy settings and technological change. This work should facilitate our own next steps.
The TFCD’s recommendations have prompted others to look at climate-related scenario analysis.
For example, the Institutional Investors Group on Climate Change has published a guide for
investors on how to select appropriate scenarios and apply them to the analysis of investments.b
The Government itself recently released its Green Finance Strategy setting out how it will build on
the TCFD recommendations, including by ensuring that all UK listed companies and large asset
owners disclose in line with those recommendations by 2022. It will also work to embed the
TCFD reporting proposals in public financial bodies and to foster greater transparency in nature-
related financial disclosures. In respect of the public sector, it will undertake a review, led by the

Climate change
Treasury, of the costs associated with achieving net-zero emissions by 2050, and will consider
risks from climate change and the low-carbon transition in Managing fiscal risks next year. The
Government also said that it would be “Clarifying responsibilities for the Prudential Regulation
Authority, the Financial Conduct Authority and the Financial Policy Committee to have regard to
the Paris Agreement when carrying out their duties.”c
a
Network for Greening the Financial System, First comprehensive report, A call for action: Climate change as a source of financial
risk, April 2019.
b
The Institutional Investors Group on Climate Change, Navigating climate scenario analysis: A guide for institutional investors,
February 2019.
c
Department for Business, Energy & Industrial Strategy, Green Finance Strategy, July 2019.

9.38 In this chapter we have surveyed some of the climate-related risks to economies and the
nature of the fiscal risks that might be posed, and set out some steps we propose to take in
considering these risks. This raises several issues, among them:
• the integration of climate-related risks into the broader management of fiscal risks;
• the appropriateness of the Bank/NGFS scenario framework for assessing fiscal risks;
• the analysis of the sources and transmission channels relevant to the public finances;
• the trade-off between climate and other objectives – for example, around fuel duty;
• the way to manage potential shocks to the public finances from climate change; and
• the trade-off between longer-term climate-related fiscal pressures and other priorities.
9.39 When assessing the outlook for the economy and public finances over the medium and long
term, does the Government regard these or other issues as important for its risk

10 A fiscal stress test
Introduction
10.1 The International Monetary Fund (IMF) recommends that fiscal risk analysis should include a
‘fiscal stress test’, which examines how the public finances would respond to a significant
economic and financial shock. It argues that this can provide a “more comprehensive and
integrated assessment of the potential shocks to government finances” and that it “can help
policymakers simulate the effects of shocks to their central forecasts and their implications
for government solvency, liquidity, and financing needs.”1
10.2 Our latest central forecast was published in our March 2019 Economic and fiscal outlook
(EFO). It was based on broad-brush assumptions about Brexit – that we would leave on 29
March, that trade with the EU would be less free after Brexit, and that the migration regime
would be tighter. While these assumptions are not based on any particular deal, they are
consistent with a smooth transition to a deal of some sort. So leaving without a deal –
perhaps in a disorderly or disruptive manner – is a key risk to our central forecast.
10.3 In this chapter, we present the results of a stress test based on the IMF’s no-deal, no-
transition Brexit scenarios in its April 2019 World Economic Outlook. Specifically, we employ
‘Scenario A’ – the less disruptive of two scenarios it sketched out.2 In doing so, our goal is to
illustrate what could happen to the public finances, not necessarily what is most likely to
happen: it is a scenario, not a forecast. For example, we do not specify how fiscal policy
might respond in such a scenario. But we do discuss Brexit-related risks that either do not
feature in the scenario, or that could be more fiscally detrimental, at the end of the chapter.
10.4 Several other institutions have produced no-deal Brexit scenarios. In November 2018, the
Bank of England produced a number of Brexit scenarios, including two that assumed a
‘disruptive’ and a ‘disorderly’ no-deal, no-transition Brexit. The scale of the shock they
assumed was similar to that assumed in our 2017 fiscal stress test. One advantage of using
the IMF’s scenario as the basis for the stress test in this report is therefore that it allows us to
subject our fiscal forecasts to a different set of assumptions to those used two years ago.
10.5 This chapter:
• summarises the assumptions underpinning the stress test;
• describes the results for receipts, spending and the main fiscal aggregates; and
• discusses the potential wider economic and fiscal risks.
1
IMF, Analyzing and Managing Fiscal Risks – Best Practices, May 2016.
2
See Scenario Box 1.1 in Chapter 1 of the IMF’s World Economic Outlook, April 2019.

A fiscal stress test
Assumptions underpinning the stress test

10.6 The IMF scenario sets out its key assumptions regarding the trading and migration
relationship between the UK and the EU in the event of a no-deal, no-transition exit. It also
quantifies the impact on real GDP growth in annual terms over the period from 2019 to
2023 and the long-term effects of different trade and immigration assumptions on real
GDP. This provides the foundation for the stress test in this chapter. We have shifted
everything back to be consistent with Brexit taking place on 31 October, whereas the IMF
assumed 29 March, the date Brexit was slated to happen when it finalised its scenario.3
Assumptions taken from the IMF scenario

10.7 We use the following policy and economy assumptions drawn from the IMF scenario:
• Imports into the UK face the Government’s announced temporary tariff regime – which
sees 87 per cent of imports exempt from tariffs for a year4 – before reverting to the
current EU ‘most-favoured nation’ (MFN) rates (of around 4 per cent, on average).5
• Implementation of a temporary recognition regime with the EU for some financial

services, and temporary recognition by the UK of multiple EU product standards.6
• Limited physical border disruptions, but trading under WTO rules means that non-tariff
barriers to trade with the EU rise by an additional 14 per cent in tariff-equivalent
terms.
• Net migration into the UK is lower by 25,000 a year out to 2030.
• Potential output growth is therefore lower, due to the combination of higher trade
barriers and weaker labour force growth.
• The economy enters recession and there is a significant cyclical fall in output.
• There is a modest tightening of financial conditions. Interest rate spreads increase by

12.5 basis points for gilts and by 20 basis points for UK corporates, but fall back to
the baseline over the subsequent 18 months.
• There is no discretionary fiscal policy response, but automatic stabilisers are assumed
to operate. Monetary policy is eased according to a Taylor rule-type response.
3
This adjustment affects annual real GDP growth rates relative to those in the IMF’s scenario.
4
HM Government, Temporary tariff regime for no deal Brexit published, 13 March 2019.
5
For more details on EU MFN tariff rates, see Chapter 2 of OBR, Discussion paper No.3: Brexit and the OBR’s forecasts, October 2018.
6
For more details see HM Government, Implications for Business and Trade of a No Deal Exit on 29 March 2019, 26 February 2019.

Other economic assumptions

10.8 We need to make assumptions about several economic variables that are not specified in
the IMF’s scenario before we can use our ready-reckoners to generate the fiscal side of the
stress test. These include: the expenditure and income composition of GDP; employment
and productivity growth; average earnings growth; CPI and RPI inflation; house prices and
property transactions. In making these assumptions, we have sought to produce paths
consistent with the output profile and conditioning assumptions outlined by the IMF. All the
following assumptions are expressed relative to a baseline of our March 2019 forecast.
Financial markets
10.9 Our main financial market assumptions are:
• The nominal sterling effective exchange rate depreciates by 10 per cent immediately,
and is around 5 per cent lower in real terms in the first quarter of 2024, as market
participants judge that a fall in the pound is needed to compensate for the reduced
competitiveness with the EU.
• Despite higher interest rate spreads, gilt yields and mortgage rates are slightly lower,
thanks to a lower path for Bank Rate (see below). In reality, medium-term Bank Rate
expectations and gilt yields have already fallen back further than this since March,
which may reflect a higher probability of a no-deal Brexit being priced in.
• Equity prices fall 5 per cent in the quarter in which the no-deal Brexit takes place,
before rising in line with (the weaker) path for nominal GDP, ending 6.7 per cent
lower than our March forecast by 2023-24.
Real GDP, potential output and the output gap

10.10 Our assumptions about real GDP growth and its cyclical-structural decomposition are:
• The UK enters a year-long recession in the fourth quarter of 2019. Real GDP falls by
2.1 per cent, around the same as in the early-1990s recession but only about a third
of what was seen in the financial crisis. By mid-2021, real GDP is 4.0 per cent lower
than in our March forecast. The economy then picks up and is just 1.6 per cent smaller
than the baseline by the first quarter of 2024 (Chart 10.1).
• A negative output gap opens, peaking at 2.5 per cent of potential output at the end of
2020, before narrowing to 0.5 per cent by the scenario horizon.
• Potential output is 1.1 per cent lower than the baseline by the end of the period.

Chart 10.1: Real GDP: stress test versus March forecast

114
Stress test
112
March 2019
110
108
106
2016Q2 = 100
104
102
100
98
96
94
92
2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024
Source: ONS, OBR
Composition of GDP
10.11 The composition of GDP matters for the fiscal consequences of the stress test given differing
effective tax rates across different components. We assume that:
• The weaker path for GDP is mainly driven by lower business investment, thanks to
heightened uncertainty about the outlook and the increase in trade costs resulting in a
prospective loss of some export markets. Residential investment and consumer
spending also weaken (though proportionately less), as slower nominal earnings
growth and the depreciation of sterling squeeze household real incomes.
• In the near term, the contribution of net trade to GDP growth is similar to our March
forecast. The cyclical downturn reduces import growth and the sterling depreciation
boosts export growth, but these effects are largely offset by the immediate rise in tariff
and non-tariff barriers on UK exports to the EU. Over the medium term, increased
barriers to trade with the EU lower the trade intensity of the UK economy – with the
UK’s export market share and the import intensity of demand hit by similar amounts.
But weaker domestic demand means that imports are hit by more than exports. That
leaves the medium-term net trade contribution somewhat higher.
• The share of labour income in GDP rises initially before falling back. Wages and
salaries are 2.6 per cent lower at the start of 2024.
• Corporate profits initially fall as a share of GDP, but rise above the March forecast in
the medium term. Non-oil, non-financial profits – a key determinant in the fiscal
forecast – are 0.7 per cent below the March forecast in the first quarter of 2024.

Labour market
10.12 Labour income is the most tax-rich component of GDP. We assume:
• Employment initially falls and unemployment rises, peaking at just over 5 per cent in
2021. This is a smaller rise and a lower peak than in either of the past two recessions.
Employment then recovers and the unemployment rate is only slightly above our
March forecast by the first quarter of 2024.
• Productivity growth is weaker. Output per hour is 2.2 per cent lower in 2021, but the
subsequent cyclical recovery leaves it 1.1 per cent lower in the first quarter of 2024.
• Earnings growth is lower in the first few years, due to weaker productivity growth.
Earnings growth is similar in the medium term, as a rise in productivity growth is not
fully matched by an increase in real wages, consistent with the rise in unemployment.
• The combination of weaker nominal earnings growth and higher CPI inflation leaves
real wages significantly lower – by 2.5 per cent by the start of 2024. Their previous
peak is not regained until 2022, representing 13 years of stagnant real wages.
Prices and nominal GDP

10.13 Real GDP growth is an important driver of the public finances, but ultimately it is nominal
GDP – factoring in price movements too – that matters. We assume:
• CPI inflation is initially higher, due to the weaker pound, on top of the increase in
tariffs on imports from the EU, which raise import prices in two steps – an initial
smaller increase as the ‘no-deal’ tariffs are used, then a second larger rise when MFN
rates are imposed. CPI inflation then drops below the baseline as the effect of higher
import prices fades and spare capacity lowers domestically generated inflation.
• RPI inflation is lower in the near term, thanks to falling house prices and lower
mortgage interest rates, which affect RPI but not CPI inflation. RPI inflation is similar to
that in the baseline forecast towards the end of the period, as the recovery in house
prices and mortgage rates broadly offset the lower path for CPI inflation.
• As with CPI inflation, GDP deflator inflation picks up initially, then falls back to below
the March forecast later in the period.
• Nominal GDP is 1.4 per cent lower than our March forecast by the start of 2024 – a
similar shortfall to that in real GDP.
10.14 Chart 10.2 decomposes the sources of difference between the stress test scenario for
nominal GDP and our March forecast. For most of the period, a significantly more negative
output gap provides the largest contribution to the nominal GDP shortfall. Towards the end
of the scenario, the deterioration in potential output becomes the most important factor.

Chart 10.2: Contributions to difference in nominal GDP from March forecast

1
-1
Percentage points
-2
-3
-4
Potential output Output gap and other Deflator
Real GDP Nominal GDP
-5
Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1 Q3 Q1
2019 2020 2021 2022 2023 2024
Source: OBR
Property markets
10.15 Property transactions and prices often move disproportionately during recessions, with
significant implications for tax receipts. We assume:
• House prices fall by almost 10 per cent between the start of 2019 and mid-2021, to
be around 14 per cent below our March forecast. They recover a little, but remain
almost 13 per cent lower by the end of the period.
• Residential property transactions fall by 20 per cent by the end of 2020. They are 11
per cent below the March forecast at the scenario horizon.
• Commercial property prices are assumed to fall more steeply than those in the
residential market, as was the case in the 2008 recession, while commercial property
transactions move broadly in line with residential transactions.
Monetary and fiscal policy

10.16 In this type of scenario, the response by policymakers would be important. We assume:
• Inflation expectations remain anchored, allowing the MPC to look through the
temporary import-price-driven rise in CPI inflation. As spare capacity lowers domestic
inflation, the MPC cuts Bank Rate to around 0.2 per cent by the end of 2020. As the
negative output gap closes, Bank Rate gradually increases back towards the level
assumed in our March forecast.
• We do not assume that any further monetary easing is delivered through quantitative
easing, so the Asset Purchase Facility remains the same size as in our March forecast.

• But we do assume that lending under the Term Funding Scheme is rolled over rather
than being repaid at its four-year term in 2020 and 2021 (as is the case in our March
forecast), since the economy is in recession in 2020.
• Discretionary fiscal policy is unchanged from that set out in the Spring Statement, but
automatic fiscal stabilisers are assumed to operate freely.
Summary of the stress test scenario and comparison with our March forecast
10.17 The main economic assumptions and how they differ from our March forecast are outlined
in Tables 10.1 and 10.2. The key public finance determinant assumptions and how they
differ from our March forecast are detailed in Tables 10.3 and 10.4.
Table 10.1: Economic assumptions: stress test

Percentage change on a year earlier, unless otherwise stated
2018 2019 2020 2021 2022 2023
UK economy
Gross domestic product (GDP) 1.4 0.9 -1.4 0.8 2.3 2.9
GDP per capita 0.8 0.3 -2.0 0.2 1.8 2.4
GDP level (2018=100) 100 100.9 99.4 100.2 102.5 105.5
Nominal GDP 3.2 2.9 0.7 3.0 4.1 4.8
Non north sea profits 2.9 1.5 -2.7 2.8 5.7 6.6
Output gap (per cent of potential output) 0.2 -0.3 -2.2 -2.4 -1.9 -0.9
Potential output 1.2 1.4 0.5 1.0 1.8 1.9
Expenditure components of GDP
Household consumption¹ 1.7 0.9 -1.4 0.5 2.1 3.0
General government consumption 0.2 2.1 1.7 1.6 1.6 1.7
Business investment -0.9 -2.1 -8.0 -1.3 5.0 7.0
General government investment² 0.5 5.9 1.8 2.2 0.9 2.0
Residential investment² 1.4 0.4 -4.9 -1.4 3.7 5.1
Exports of goods and services 0.2 0.4 -3.5 0.8 0.0 -0.3
Imports of goods and services 0.8 2.2 -3.6 -0.1 0.1 0.9
Inflation
CPI 2.5 2.1 2.4 2.4 1.7 1.8
RPI 3.3 3.0 2.5 2.5 2.4 3.2
GDP deflator at market prices 1.7 2.0 2.2 2.2 1.7 1.9
Labour market
Employment (millions) 32.4 32.6 32.4 32.5 32.7 33.0
Productivity per hour 0.5 0.5 -0.6 0.7 1.5 1.9
Wages and salaries 4.5 3.1 1.1 2.7 3.6 3.9
Average earnings3 3.0 2.9 1.6 2.6 3.0 3.2
LFS unemployment (% rate) 4.1 4.2 5.0 5.2 4.8 4.3
Household sector
Property transactions (000s) 1193 1147 947 965 1087 1185
House prices 3.3 0.5 -5.9 -3.0 3.0 5.7
¹ Includes households and non-profit institutions serving households.
2
Includes transfer costs of non-produced assets.
3
Wages and salaries divided by employees.

Table 10.2: Economic assumptions: stress test versus March forecast

Percentage change on a year earlier, unless otherwise stated
2018 2019 2020 2021 2022 2023
UK economy
Gross domestic product (GDP) 0.0 -0.3 -2.9 -0.8 0.7 1.3
GDP per capita 0.0 -0.3 -2.9 -0.8 0.7 1.3
GDP level (2018=100) 0.0 -0.3 -3.3 -4.1 -3.5 -2.2
Nominal GDP 0.0 -0.3 -2.6 -0.6 0.5 1.2
Non north sea profits 0.0 -0.7 -5.6 -0.4 2.1 3.0
Output gap (per cent of potential output) 0.0 -0.2 -2.1 -2.3 -1.8 -0.9
Potential output 0.0 -0.1 -1.1 -0.6 0.2 0.3
Expenditure components of GDP
Household consumption1 0.0 -0.2 -2.9 -1.2 0.5 1.4
General government consumption 0.0 0.0 0.0 0.0 0.0 0.0
Business investment 0.0 -1.1 -10.3 -3.6 2.7 4.7
General government investment2 0.0 0.0 0.0 0.0 0.0 0.0
Residential investment2 0.0 -0.4 -5.5 -2.2 2.3 3.3
Exports of goods and services 0.0 -1.0 -5.2 0.6 0.3 0.2
Imports of goods and services 0.0 -0.8 -5.7 -0.7 0.0 0.7
Inflation
CPI 0.0 0.1 0.5 0.4 -0.3 -0.2
RPI 0.0 0.0 -0.3 -0.5 -0.7 0.1
GDP deflator at market prices 0.0 0.0 0.3 0.2 -0.2 -0.1
Labour market
Employment (millions) 0.0 0.0 -0.3 -0.4 -0.3 -0.2
Productivity per hour 0.0 -0.2 -1.6 -0.4 0.3 0.6
Wages and salaries 0.0 -0.2 -2.2 -0.9 0.1 0.4
Average earnings3 0.0 -0.2 -1.4 -0.5 -0.1 -0.1
LFS unemployment (% rate) 0.0 0.1 0.9 1.1 0.8 0.3
Household sector
Property transactions (000s) 0 -25 -287 -316 -232 -169
House prices 0.0 -0.4 -7.2 -6.6 -1.1 1.5
¹ Includes households and non-profit institutions serving households.
2
Includes transfer costs of non-produced assets.
3

Table 10.3: Fiscal determinants: stress test

195 45 45 45 45 45
Percentage change on previous year, unless otherwise specified
2019-20 2020-21 2021-22 2022-23 2023-24
GDP and its components
Real GDP 0.3 -1.2 1.2 2.6 2.8
Nominal GDP 1 2.9 0.8 2.9 4.1 4.8
Nominal GDP (£ billion) 1,2 2193 2210 2275 2368 2482
Nominal GDP (centred end-March £bn) 1,3 2229 2249 2316 2410 2526
Wages and salaries4 2.4 1.4 2.9 3.7 3.9
Non-oil PNFC profits4,5 1.5 -2.7 2.8 5.7 6.6
Consumer spending4,5 3.1 0.9 2.9 3.8 4.9
Prices and earnings
GDP deflator 2.1 2.2 2.0 1.7 1.9
RPI 3.0 2.4 2.4 2.6 3.2
CPI 2.2 2.5 2.1 1.7 1.9
Average earnings6 2.5 1.8 2.7 3.1 3.2
'Triple-lock' guarantee (September) 3.5 2.5 2.6 3.0 3.2
Key fiscal determinants
Employment (millions) 32.5 32.4 32.5 32.8 33.0
Output gap (per cent of potential output) -0.8 -2.4 -2.3 -1.6 -0.7
Financial and property sectors
Equity prices (FTSE All-Share index) 3808 3744 3855 4024 4213
HMRC financial sector profits1,5,7 2.2 -5.3 1.2 4.4 5.5
Residential property prices8 -1.2 -6.2 -1.5 4.3 5.6
Residential property transactions (000s) 9 1098 934 991 1118 1200
Commercial property prices9 -3.6 -13.4 -6.1 2.2 4.1
Commercial property transactions9 -5.7 -19.1 4.1 11.6 6.4
Oil and gas
Oil prices ($ per barrel) 5 62.1 61.6 62.0 63.3 64.5
Oil prices (£ per barrel) 5 49.0 51.2 50.1 49.7 49.8
Gas prices (p/therm) 5 50.5 53.1 54.1 55.2 56.3
Oil production (million tonnes) 5 48.4 48.4 46.0 43.7 41.5
Gas production (billion therms) 5 13.7 13.3 12.7 12.0 11.4
Interest rates and exchange rates
Market short-term interest rates (%) 10 0.9 0.5 0.5 0.8 1.2
Market gilt rates (%) 11 1.3 1.4 1.4 1.6 1.7
Euro/Sterling exchange rate (€/£) 1.08 1.02 1.03 1.03 1.02
1 7
Not seasonally adjusted. HMRC Gross Case 1 trading profits.
2 8
Denominator for receipts, spending and deficit Outturn data from ONS House Price Index.
9
forecasts as a per cent of GDP. Outturn data from HMRC information on stamp duty land tax.
3 10
Denominator for net debt as a per cent of GDP. 3-month sterling interbank rate (LIBOR).
4
Nominal. 5 Calendar year. 11
Weighted average interest rate on conventional gilts.
6

Table 10.4: Fiscal determinants: stress test versus March forecast

Percentage change on previous year, unless otherwise specified
2019-20 2020-21 2021-22 2022-23 2023-24
GDP and its components
Real GDP -0.9 -2.7 -0.4 1.0 1.2
Nominal GDP 1 -0.3 -2.6 -0.6 0.5 1.2
Nominal GDP (£ billion) 1,2 -6 -64 -80 -72 -46
Nominal GDP (centred end-March £bn) 1,3 -7 -65 -81 -73 -47
Wages and salaries4 -0.7 -2.1 -0.6 0.3 0.4
Non-oil PNFC profits4,5 -0.7 -5.6 -0.4 2.1 3.0
Consumer spending4,5 -0.2 -2.5 -0.8 0.2 1.2
Prices and earnings
GDP deflator 0.1 0.4 0.1 -0.2 -0.1
RPI 0.0 -0.4 -0.6 -0.5 0.2
CPI 0.2 0.6 0.1 -0.3 -0.1
Average earnings6 -0.4 -1.2 -0.4 -0.1 -0.1
'Triple-lock' guarantee (September) 0.0 -0.4 -0.5 -0.1 0.0
Key fiscal determinants
Employment (millions) -0.1 -0.4 -0.4 -0.3 -0.2
Output gap (per cent of potential output) -0.6 -2.2 -2.2 -1.6 -0.7
Financial and property sectors
Equity prices (FTSE All-Share index) -122 -319 -353 -335 -304
HMRC financial sector profits1,5,7 -0.9 -7.0 -0.6 2.6 3.7
Residential property prices8 -1.4 -8.4 -5.4 0.2 1.4
Residential property transactions (000s) 9 -82 -313 -300 -210 -162
Commercial property prices9 -2.1 -12.5 -8.1 0.2 2.1
Commercial property transactions9 -6.8 -20.6 2.5 10.0 4.8
Oil and gas
Oil prices ($ per barrel) 5 0.0 0.0 0.0 0.0 0.0
Oil prices (£ per barrel) 5 1.2 4.7 3.8 3.0 2.7
Gas prices (p/therm) 5 0.0 0.0 0.0 0.0 0.0
Oil production (million tonnes) 5 0.0 0.0 0.0 0.0 0.0
Gas production (billion therms) 5 0.0 0.0 0.0 0.0 0.0
Interest rates and exchange rates
Market short-term interest rates10 0.0 -0.7 -0.7 -0.4 -0.2
Market gilt rates11 0.0 -0.1 -0.1 0.0 0.0
Euro/Sterling exchange rate (€/£) -0.06 -0.10 -0.08 -0.06 -0.06
1 7
Not seasonally adjusted. HMRC Gross Case 1 trading profits.
2 8
Denominator for receipts, spending and deficit Outturn data from ONS House Price Index.
9
forecasts as a per cent of GDP. Outturn data from HMRC information on stamp duty land tax.
3 10
Denominator for net debt as a per cent of GDP. 3-month sterling interbank rate (LIBOR).
4
Nominal. 5 Calendar year. 11
Weighted average interest rate on conventional gilts.
6

What does the scenario imply for the overall economic impact of Brexit?
10.18 In describing the stress test, we have concentrated on differences from our March forecast.
That was conditioned on the UK securing a deal and exiting smoothly, as well as the same
broad-brush assumptions of the medium and longer-term impact of Brexit that we have
used since our November 2016 forecast. So the comparisons described here represent the
additional cost of leaving without a deal or transition period, not the total effect of Brexit.
10.19 In the IMF scenario used here, the loss of potential output continues to build well beyond
our five-year scenario horizon. In the long run, potential output is 2.9 per cent lower than
the IMF’s own baseline forecast, with 0.6 percentage points coming from lower migration
and 2.3 percentage points from less capital deepening. This compares to potential output
being only 1.1 per cent lower than our March forecast by the end of our stress test based on
this scenario. Some other studies assume that the loss of potential output from a no-deal,
no-transition exit occurs more quickly. For example, the Bank of England argues that “A
range of evidence suggests that the impact of introducing frictions such as tariffs and
customs checks at the border comes through quickly. A substantial number of firms have little
experience with customs checks and, for many, business models and supply chain
management could be significantly disrupted by delays at the border.”7
10.20 The IMF’s latest central forecast already includes a long-term GDP loss of 3 per cent relative
to remaining in the EU – 0.6 percentage points of which comes from lower migration.8
Combining the IMF’s baseline and scenario estimates of the cost of Brexit gives a long-run
estimated hit to GDP from leaving the EU and then trading under WTO rules of 5.9 per cent
(Table 10.5). This compares to the Government’s estimates of between 7.7 and 9.3 per cent
(depending on the migration policy assumption used);9 NIESR’s estimate of 5.5 per cent;10
and a joint study by the UK in a Changing Europe and the LSE Centre for Economic
Performance’s (LSE) which contains estimates of between 4.7 and 9.9 per cent (depending
on whether or not there are additional dynamic effects on productivity).11
Table 10.5: Contributions to long-run impact on GDP from trading with the EU on
WTO terms
HM Government LSE
IMF Unchanged Zero net EEA NIESR Static Dynamic
migration policy immigration productivity productivity
Migration -1.2 -0.2 -1.8 Not stated -1.4 -1.8
Productivity -4.7 -7.5 -7.5 Not stated -3.3 -8.1
Total -5.9 -7.7 -9.3 -5.5 -4.7 -9.9
Note: Estimates are in percentage points.
7
Bank of England, EU withdrawal scenarios and monetary and financial stability, November 2018.
8
IMF, United Kingdom: Selected Issues, November 2018.
9
HM Government, EU Exit: Long-term economic analysis, November 2018.
10
Hantzsche et al, The economic effects of the Government’s proposed Brexit deal, 26 November 2018.
11
CEP, The economic consequences of the Brexit deal, November 2018.

Fiscal assumptions
10.21 We have made several further assumptions on the fiscal side. Among them:
• Asset sales: our March forecast included £55.2 billion of financial asset sales over the
forecast period to 2023-24. These consist primarily of the remaining loan assets held
by UK Asset Resolution (UKAR), sales of RBS shares and sales of Plan 1 student loans.
Around £4.9 billion of the £16 billion of sales in 2019-20 have already occurred. The
remaining sales in 2019-20 and those in subsequent years are assumed to be delayed
by two years. Several sales were delayed after the EU referendum in 2016.
• The cost of interventions in the private sector: we assume the Government uses £10
billion in lending to provide support to private sector companies, for example those
most affected by the introduction of tariffs (in manufacturing or agriculture) or by
higher non-tariff barriers (in business and financial services). This is far less than the
initial cost of the financial sector interventions undertaken in the crisis.
• Spending related to guarantees: we assume that £1 billion is spent to reflect possible

costs related to schemes such as Help to Buy or UK Guarantees.
10.22 As in our post-referendum forecasts, we assume that any savings from EU contributions are
recycled into domestic spending, for example on the Government’s commitments to
maintain farm support and industrial and science programmes after EU withdrawal. As we
currently assume the financial settlement costs are subsumed within this assumption, it
means the scenario is not sensitive to different assumptions about when and how much
would be paid under the terms of any alternative financial settlement. We have therefore not
made any assumptions about whether the amount or terms would be renegotiated.
Results of the stress test

The big picture
10.23 Relative to our March forecast, borrowing is around £30 billion a year higher on average
from 2020-21 onwards. There are four main drivers of the results (Chart 10.3):
• Income tax and NICs receipts are hit by the cyclical downturn, raising borrowing by
around £16½ billion a year from 2020-21 onwards.
• Capital tax receipts fall sharply thanks to the falls in asset prices, especially in the
housing market, adding around £10 billion a year from 2020-21 onwards.
• But debt interest spending benefits more from lower interest rates and RPI inflation
than it suffers from higher borrowing, lowering the deficit by £5 billion to £6 billion in
2020-21 and 2021-22, and by diminishing amounts thereafter.
• And customs duties bring in more than in our March forecast, and are treated as UK
tax receipts rather than being collected on behalf of the EU. This boosts receipts by

£6.3 billion in 2020-21 and by around £10 billion a year thereafter. The effect on
borrowing is around £0.7 billion a year less than that because we lose the amount
currently credited to the UK that notionally covers the cost of collecting tariffs. Box 10.1
gives more detail on how these customs duties figures have been generated.
Chart 10.3: Stress test versus March forecast: sources of differences in PSNB
60
Other receipts and spending elements
50
Capital taxes
Customs duties
40
Difference
30
£ billion
20
10
-10
-20
2017-18 2018-19 2019-20 2020-21 2021-22 2022-23 2023-24
Source: ONS, OBR
Receipts
10.24 The shortfall in receipts is primarily from weaker nominal tax bases and the downturn in
asset markets. The receipts to GDP ratio is around 0.2 to 0.3 percentage points lower over
the period. This drop would have been greater were it not for higher customs duties.

10.25 Receipts are more than £16 billion a year below our March forecast from 2021-22
onwards. The shortfall is dominated by the effects of weaker earnings growth (which takes
over £10 billion a year off receipts by 2022-23) and lower employment (taking £4 billion a
year off at its peak in 2021-22). Other sources of weakness include less fiscal drag, and a
disproportionate adverse effect on some higher-paid sectors.
10.26 As we discuss in Chapter 4, income tax receipts have become more concentrated at the top
of the income distribution. Some non-tariff barriers resulting from a no-deal exit (such as the
end of passporting for financial services and mutual recognition for some business services)
could affect high-paying jobs in these sectors. The motor and pharmaceuticals sectors,
which would be the worst affected by tariffs on exports to the EU, are also well paid
compared to the rest of manufacturing. To allow for these potential effects, we have taken
around £1 billion a year off income tax receipts, in addition to the other effects.

10.27 Self-assessment (SA) income tax receipts are £2 billion lower in 2021-22 than in our March
forecast. We assume that the hit to self-employment and dividend income is in line with that
to employee incomes and profits respectively. Lower short-term interest rates also reduce tax
on savings income collected via the SA system. These effects come through more slowly than
most others because of the lag between activity and tax payment in SA.
VAT
10.28 Relative to our March forecast, VAT receipts are around £6 billion a year lower on average
from 2020-21 onwards, reflecting weaker nominal tax base growth, a hit to the share of
consumer spending subject to the standard rate and a rise in the VAT gap:
• Weaker consumer spending takes around £3 billion a year off receipts from 2020-21
onwards. Higher inflation only partly offsets the decline in real consumption during the
recession. Weakness in the exempt sector (primarily the financial sector, where VAT on
purchases cannot be recovered) and the housing sector together take around another
£1 billion a year off receipts from 2020-21 onwards.
• Consumer durables spending, which is usually standard rated, is hit harder than the
rest of consumer spending. This reduces receipts by £1.6 billion at its peak in 2021-
22, which partly unwinds as durables spending recovers relatively strongly.
• In the late 2000s recession, the VAT gap rose by around 3 percentage points as firms
delayed paying HMRC due to cash flow problems. It then fell back sharply. Similar,
though less severe, issues in the scenario raise the VAT gap by 1 percentage point in
2020-21 and 2021-22. It then returns to the profile in our March forecast.
10.29 A no-deal exit would mean an abrupt move to a UK-only VAT regime from the EU system
that has been in place for more than four decades. The stress test does not include any
additional fiscal losses from such a move, but clearly that would be a risk. We discuss
potential risks around compliance and the operation of a new regime later in the chapter.
Onshore corporation tax

10.30 Receipts of onshore corporation tax are over £4 billion a year lower in 2020-21 and 2021-
22, primarily reflecting weaker profits, only partly offset by lower capital allowance claims
following the drop in business investment:
• Lower profits take around £3 billion off onshore corporation tax receipts in 2021-22
compared with March, narrowing to around £1 billion by 2023-24.
• Business investment falls a cumulative 15 per cent below our March forecast by the
end of 2021, before recovering in the final years of the scenario. Lower use of capital
allowances adds between £0.5 to £0.8 billion a year to receipts.
10.31 We assume that financial sector profits are hit harder than those for non-financial
companies. This takes £0.5 billion a year off receipts on average. A downturn would mean

more trading losses by all firms which can then be used to offset against future profits.
However, with loss restriction measures now in place, the effect on receipts from using these
losses in future would now take place over a longer period.
Stamp duties
10.32 Stamp duty land tax (SDLT) is the most adversely affected tax stream in proportionate
terms, due to the fall in residential and commercial property prices and transactions. We
assume the same proportional effect for the devolved transactions taxes in Wales and
Scotland. Property transaction taxes are around £6½ billion lower than in our March
forecast in the final three years, a shortfall of around 40 per cent on average:
• With SDLT thresholds fixed in cash terms, lower house prices generate reverse fiscal
drag (with a higher proportion of each transaction taxed at lower rates) lowering the
effective tax rate. They take around £2½ billion a year off receipts from 2021-22.
• Residential property transactions fall by around 20 per cent by the end of 2020. This
takes an average of around £2½ billion a year off SDLT receipts.
• Commercial property prices and transactions fall even more steeply, taking around
£1.5 billion a year on average off SDLT receipts.
10.33 Stamp duty on shares is £0.3 billion a year lower on average, reflecting the short-term drop
in equity prices that is only partly reversed by the scenario horizon.
Taxes on capital
10.34 Both inheritance tax and capital gains tax (CGT) are hit by the falls in equity and property
prices. CGT receipts are lower by £2.7 billion a year by 2023-24. Given the lags in
payment, the effect of the short-term fall in asset prices is only felt in 2021-22.
10.35 With housing assets accounting for around half of the value of estates notified for probate,
the downturn in the residential property market is the main driver of the £0.9 billion a year
hit to inheritance tax receipts by 2023-24.
Excise duties
10.36 Duties on fuel, alcohol and tobacco are £1.5 billion lower in 2021-22 and £1.1 billion
lower by 2023-24. These taxes are affected both by lower RPI inflation, which lowers duty
rates, and weaker real GDP and consumption, which lowers demand for fuel and alcohol.
Other taxes
10.37 Oil and gas revenues rise by around £0.3 billion a year from 2020-21 onwards. The
weaker exchange rate against the dollar raises oil prices in sterling terms.
10.38 Business rates and council tax are both little changed from the March forecast and so are
higher as a share of GDP. This was also the case during the financial crisis since the tax
base (the stock of buildings) is very stable. We have assumed no change in council tax

relative to our March forecast. Higher CPI inflation raises the business rates multiplier
applied to the rateable value of non-domestic properties, lifting receipts. We assume only a
modest effect on business rates from the recession – a 1 per cent decline in rateable values
and a temporary rise in empty property relief in 2020-21 and 2021-22.
Other receipts
10.39 Interest and dividend receipts include the interest income on the government’s financial
assets, among them student loans and bank deposits held by the Debt Management Office
and local authorities. Short-term interest rates are 0.6 percentage points lower between
2020-21 and 2022-23, which lowers receipts by around £1½ billion a year in those years.
10.40 Lower interest rates and RPI inflation reduces accrued interest on student loans. This effect
peaks at £0.9 billion in 2021-22. As we discuss in Chapter 6, the ONS intends to change its
accounting treatment for student loans in September 2019. The stress test has been
calculated on the basis of the existing methodology, consistent with the March forecast
baseline.
10.41 With RBS asset sales delayed for two years, the Government holds more RBS shares than in
the baseline. However, RBS’s decisions over dividend payments might be affected in such a
scenario. For simplicity, we have assumed the worst-case scenario from the public finances
perspective of no further dividends being paid over the period. This lowers receipts further.
Box 10.1: Customs duties in a no-deal Brexit

In a no-deal Brexit, the UK would be able to apply its own external tariff. On 13 March 2019,
the Government announced the rates that would apply for the first 12 months after a no-deal
exit from the EU.a In line with this, in our stress test we have assumed that:
• for 12 months from November 2019, the announced ‘no-deal’ tariff rates will apply;
• from November 2020 onwards, the UK applies ‘most-favoured nation’ (MFN) tariff rates
in line with the current EU common external tariff to all incoming goods not covered by
preferential agreements (including goods coming in from the EU); and
• in line with the IMF scenario, the UK secures replacement free trade agreements with third
countries currently covered by EU agreements by the time MFN tariff rates apply.
Customs duties receipts on UK imports from outside the EU totalled £3.3 billion in 2018-19. Our
March forecast was for this to increase to £3.5 billion by 2023-24. But customs duties are at
present remitted to the EU, so do not add to UK current receipts. Their only effect on PSNB
comes via the 20 per cent retained to cover the cost of collection. That reduces net expenditure
transfers to the EU.b Post-Brexit customs duties receipts would score in the same way as any tax.
The ‘no-deal’ tariff schedule contains tariff lines that cover around £36 billion of imports in
2018-19 (around 7 per cent of all goods imports). Based on a simple application of these tariff
rates to detailed goods imports data, we assume that they would raise around £3.8 billion over
the 12 months they would be in force. Three quarters of this revenue would come from tariffs on

EU goods, and around £5 of every £6 of revenue would come from tariffs on finished cars – the
main area to be subject to tariffs outside the agri-food sector.
Application of MFN tariffs from November 2020 to goods coming in from the EU raises more
revenue than is raised on goods from the rest of the world. This is mostly because a higher
proportion of imports from the EU are goods that face higher MFN duty rates (notably agri-food
goods) and goods imports from the EU are worth 20 per cent more than those from the rest of
the world. The average MFN tariff that would apply to goods currently imported from the EU is
3.9 per cent, compared with 2.8 per cent for goods from non-EU countries.
Again, a simple application of MFN tariff rates to goods imports yields revenues on goods from
the EU of around £6.3 billion a year, in addition to the £3.5 billion from non-EU countries in our
March baseline forecast. Removing the 20 per cent of non-EU revenues currently retained to
cover costs of collection, the impact of MFN tariffs would be to reduce borrowing in the scenario
by around £9.2 billion a year relative to the assumptions in our March forecast.
Table A: Impact of customs duties on PSNB under the fiscal stress test scenario
£ billion
2019-20 2020-21 2021-22 2022-23 2023-24
Baseline revenues (£ billion; no effect on PSNB) (a) 3.4 3.4 3.5 3.5 3.5
Baseline retained cost of collection (b) 0.7 0.7 0.7 0.7 0.7
No-deal tariff schedule (to October 2020) (c) 1.5 2.3 0.0 0.0 0.0
MFN tariffs (from November 2020) (d) 0.0 4.1 9.8 9.9 9.9
Scenario retained cost of collection (e) 0.0 0.0 0.0 0.0 0.0
Effect on public sector current receipts (f) = (c) + (d) 1.5 6.3 9.8 9.9 9.9
Effect on total managed expenditure (g) = (b) - (e) 0.7 0.7 0.7 0.7 0.7
Effect on PSNB (h) = (g) - (f) -0.8 -5.7 -9.1 -9.2 -9.2
The figures we have generated for this scenario are illustrative of the potential magnitudes
involved, but they should not be taken as a precise central estimate of what the application of
these tariff schedules would raise. We have not attempted to account for all the intricacies
involved in operating the customs regime or the potentially important behavioural responses of
importers and the ultimate consumers of imports. Nor have we made assumptions about any
operational challenges, for example regarding the land border between Northern Ireland and
the Republic of Ireland.c We have not modelled non-linear effects of the ‘no-deal’ tariffs, such as
tariff-rate quotas, which disproportionately apply to agricultural goods.
Finally, we make no assumptions about the rate of compliance with the regime, or how traders
might change their behaviour in response to the tariff schedule. In particular:
• In most tariff schedules – including the EU’s common external tariff – similar tariff rates
are applied to similar goods to remove incentives for misclassification of goods to avoid
higher tariffs. This is not the case in the no-deal schedule – for example, with an 8 per
cent tariff on frozen freshwater fish but a 15 per cent tariff on other frozen fish. This could
present opportunities for non-compliance that might be difficult to detect.
• The no-deal tariff schedule imposes very different tariffs on certain components and
finished goods, particularly on cars, where a 10 per cent tariff applies to finished

products but parts can be imported tariff-free. The extent to which a car can approach
completion without being deemed finished and hit by the tariff is not clear. This could
prompt producers to import cars in a sufficiently unfinished state to avoid the tariff.
All these and no doubt other factors would need to be considered should we be required to
produce a full medium-term customs duties forecast on a no-deal Brexit basis.
a
Department for International Trade, MFN and tariff rate quotas of customs duty on imports if the UK leaves the EU with no deal, 13
March 2019 (updated 29 March 2019).
b
The retained cost of collection is included in the calculation for the UK’s net contribution to the EU as a negative contribution,
therefore reducing expenditure, rather than as a tax.
c
The Government’s published guidance regarding transitional simplified procedures states that it does not apply to the movement of
goods between Northern Ireland and Ireland, and that new guidance for that border will be issued at a later date. As such, it is not
possible to know what impact that will have on compliance with tariffs and/or with trade volumes between the two.
Table 10.6: Current receipts: the stress test scenario

£ billion
2019-20 2020-21 2021-22 2022-23 2023-24
Income tax (gross of tax credits) 1 193.7 199.0 204.5 212.8 223.6
of which: Pay as you earn 161.9 166.8 171.8 178.4 186.8
Self assessment 34.0 33.7 34.5 36.4 38.9
National insurance contributions 142.2 144.8 149.2 154.9 161.5
Value added tax 135.9 135.1 138.9 144.5 151.4
Onshore corporation tax 2 56.1 52.5 54.0 57.6 61.5
Offshore corporation tax 1.6 2.2 2.5 2.7 2.7
Petroleum revenue tax -0.4 -0.4 -0.4 -0.4 -0.3
Fuel duties 28.3 28.5 29.2 30.2 31.4
Business rates 31.3 31.6 33.3 34.2 34.7
Council tax 36.3 37.5 38.7 39.9 41.1
VAT refunds 18.4 18.9 19.4 20.0 20.6
Capital gains tax 9.1 6.9 6.8 7.8 8.9
Inheritance tax 5.3 4.8 4.7 5.0 5.3
Property transaction taxes3 11.0 7.5 7.3 9.1 11.2
Stamp taxes on shares 3.6 3.5 3.6 3.7 3.9
Tobacco duties 9.1 9.0 8.9 8.8 8.8
Alcohol duties 12.5 12.7 13.2 13.7 14.3
Air passenger duty 3.7 3.7 3.8 4.1 4.3
Customs duties 1.5 6.3 9.8 9.9 9.9
Other taxes 53.0 54.0 54.5 56.0 57.7
National Accounts taxes 752.1 758.2 781.8 814.4 852.5
Interest and dividends 10.2 8.4 9.1 11.2 14.2
Gross operating surplus 43.3 45.3 47.0 48.9 51.0
Other receipts 0.8 0.8 0.6 -0.1 0.0
Current receipts 806.4 812.7 838.6 874.5 917.7
Memo: UK oil and gas revenues 4 1.1 1.8 2.1 2.3 2.4
1
Includes PAYE, self assessment, tax on savings income and other minor components such as income tax repayments.
2
National Accounts measure, gross of reduced liability tax credits.
3
Includes SDLT, ATED and devolved property transaction taxes.
4
Consists of offshore corporation tax and petroleum revenue tax.

Table 10.7: Current receipts: stress test versus March forecast

£ billion
2019-20 2020-21 2021-22 2022-23 2023-24
Income tax (gross of tax credits) 1 -2.0 -9.2 -11.3 -11.9 -10.8
of which: Pay as you earn -2.0 -7.2 -9.3 -10.0 -9.5
Self assessment 0.0 -1.9 -2.0 -1.9 -1.4
National insurance contributions -1.2 -4.9 -6.3 -6.3 -5.9
Value added tax -0.7 -6.4 -7.4 -6.4 -4.2
Onshore corporation tax 2 -0.6 -4.3 -4.3 -3.4 -2.0
Offshore corporation tax 0.1 0.3 0.3 0.3 0.2
Petroleum revenue tax 0.0 0.0 0.0 0.0 0.0
Fuel duties -0.1 -0.7 -1.0 -1.0 -0.7
Business rates 0.0 -0.1 -0.2 -0.1 -0.2
Council tax 0.0 0.0 0.0 0.0 0.0
VAT refunds 0.0 0.0 0.0 0.0 0.0
Capital gains tax 0.0 -2.8 -3.0 -2.9 -2.7
Inheritance tax -0.1 -0.6 -0.9 -0.9 -0.9
Property transaction taxes3 -1.6 -5.9 -7.2 -6.3 -5.6
Stamp taxes on shares -0.1 -0.3 -0.3 -0.3 -0.3
Tobacco duties 0.0 0.0 -0.1 -0.2 -0.1
Alcohol duties -0.1 -0.2 -0.4 -0.3 -0.2
Air passenger duty 0.0 -0.2 -0.2 -0.2 -0.1
Customs duties4 1.5 6.3 9.8 9.9 9.9
Other taxes 0.3 0.1 0.1 0.2 0.2
National Accounts taxes -4.6 -28.9 -32.4 -29.8 -23.5
Interest and dividends -0.4 -2.5 -3.0 -2.2 -0.3
Gross operating surplus 0.0 0.0 0.0 0.0 0.0
Other receipts 0.0 0.0 0.0 0.0 0.0
Current receipts -5.0 -31.4 -35.5 -32.1 -23.8
Memo: UK oil and gas revenues 5 0.1 0.3 0.3 0.3 0.2
1
Includes PAYE, self assessment, tax on savings income and other minor components, such as income tax repayments.
2
National Accounts measure, gross of reduced liability tax credits.
3
Includes SDLT, ATED and devolved property transaction taxes.
4
Comparison against the baseline where no customs duties score as a UK tax
5
Consists of offshore corporation tax and petroleum revenue tax.
Public spending
10.42 Total spending in cash terms is little changed in the stress test compared with March. Lower
debt interest spending is largely offset by higher welfare and local authority spending. But
spending is higher as a share of GDP because the economy is smaller.
Departmental spending
10.43 We have assumed no change in discretionary fiscal policy, including no adjustments to meet
overseas aid and defence spending targets that are set relative to GDP and would therefore
cost less in cash terms. The only departmental spending difference relative to our March
forecast is the £1 billion in spending on guarantees being called, which in the scenario is
assumed to be financed via reduced underspending in 2020-21 and 2021-22.12
12
In our 2017 stress test we assumed departmental spending would increase to reflect larger block grant payments to the Scottish
Government. In 2018 Scottish Government expenditure was reclassified into annually managed expenditure. In this scenario we assume

Welfare spending
10.44 Welfare spending rises by £2.7 billion relative to March in 2021-22. A higher caseload and
higher awards for working-age benefits is partly offset by lower spending on state pensions.
10.45 Higher unemployment adds over £2 billion a year. As modelled, this is primarily through
higher jobseeker’s allowance, housing benefit and out-of-work tax credits in 2020-21 and
2021-22. In reality, much of this would be paid via universal credit. As unemployment falls
back, the effect diminishes to around £0.6 billion a year by 2023-24. Higher inflation raises
working-age average awards from 2021-22 onwards, costing over £1 billion by 2022-23.
10.46 State pensions are uprated by the triple lock (the higher of earnings growth, CPI inflation or
2.5 per cent). In our March forecast, they are uprated by earnings in each year. Earnings
growth is much weaker in the scenario, and state pensions spending is close to £1 billion a
year lower by 2023-24, with the 2.5 per cent minimum being triggered in 2021-22.
10.47 One uncertainty over the welfare spending consequences of a recession hitting in the next
year or so is the rollout of universal credit, which by 2020-21 would be a much greater
source of means-tested benefits than it is today. Universal credit differs from the systems it
replaces – with tax credits-type support being provided through jobcentres rather than by
HMRC and conditionality being applied to many more claimants. We have not attempted to
predict how that might affect caseloads and spending in a downturn.

10.48 Debt interest spending is lower than our March forecast in each year, despite the higher
deficit. This reflects the beneficial effects of lower short-term interest rates and lower RPI
inflation. The debt interest saving peaks at £6 billion in 2021-22.
10.49 Table 10.8 compares the stress test path of debt interest spending with our March forecast.
The main differences are:
• Lower Bank Rate reduces the cost of financing the Bank of England reserves created to
fund the Asset Purchase Facility’s gilt holdings. Bank Rate in 2020 and 2021 is around
0.6 percentage points lower than in March. This reduces debt interest spending by
over £3½ billion in those years.
• Lower RPI inflation reduces accrued spending on index-linked gilts. This reduces debt
interest spending by an average of £2½ billion a year from 2020-21 onwards.
• Gilt rates are marginally lower, reducing debt interest spending a little.
• Higher borrowing raises the stock of debt. This adds £2 billion a year to debt interest
spending by 2023-24.
that Welsh Government expenditure, which remains within departmental spending, remains constant with increased block grant payments
fully offset by lower self-financed expenditure.

Table 10.8: Debt interest spending: stress test versus March forecast
£ billion
2019-20 2020-21 2021-22 2022-23 2023-24
March forecast 40.2 38.9 40.3 41.5 42.3
Stress test 39.5 34.0 34.3 37.7 40.3
Change -0.8 -4.9 -6.0 -3.8 -2.0
of which:
RPI inflation 0.2 -1.7 -3.2 -2.6 -2.7
Interest rates 0.0 0.0 -0.1 -0.1 -0.2
Financing 0.1 0.5 1.0 1.5 2.0
Short-term interest rates (APF) -1.0 -3.6 -3.7 -2.5 -1.1
Other spending
10.50 Other spending items affected materially by the stress test are:
• Public service pensions: higher CPI inflation boosts inflation-linked payments, raising
spending by a maximum of £0.4 billion in 2022-23.
• Net expenditure transfers to the EU: as we note in Box 10.1, customs duties will be
treated as a UK tax after Brexit. This will mean the removal of the 20 per cent of
customs duties that are retained for collection costs while we are still a member of the
EU. This will raise spending by £0.7 billion a year.
• Local and devolved expenditure: we assume that local authorities draw down reserves
to keep total service expenditure unchanged in real terms. Higher inflation pushes
spending up by £0.3 billion in 2021-22. We also assume that Scottish Government
expenditure remains constant.

Table 10.9: Public spending: stress test scenario

£ billion
2019-20 2020-21 2021-22 2022-23 2023-24
Public sector current expenditure (PSCE)
PSCE in RDEL 312.2 325.1 334.8 344.6 356.8
PSCE in AME 440.9 445.9 460.7 479.2 496.9
of which:
Welfare spending 227.8 234.5 242.7 251.9 261.3
Net public service pension payments 6.7 6.4 7.4 8.2 8.4
Expenditure transfers to EU institutions 13.4 10.5 10.4 7.7 4.1
Assumed spending in lieu of EU transfers 3.7 3.7 6.3 10.0
Locally financed current expenditure 54.3 53.3 55.4 57.1 58.7
Central government debt interest, net of APF 39.5 34.0 34.3 37.7 40.3
Other current expenditure 99.2 103.6 106.8 110.3 114.0
Total public sector current expenditure 753.1 770.9 795.5 823.9 853.7
Public sector gross investment (PSGI)
PSGI in CDEL 60.3 65.5 68.8 71.2 75.4
PSGI in AME 27.8 27.5 25.7 25.8 25.8
Total public sector gross investment 88.1 93.0 94.5 97.0 101.2
Less public sector depreciation -41.1 -42.6 -44.2 -45.8 -47.5
Public sector net investment 47.0 50.5 50.3 51.2 53.8
Total managed expenditure 841.2 864.0 890.0 920.9 954.9
Table 10.10: Public spending: stress test scenario versus March forecast
£ billion
2019-20 2020-21 2021-22 2022-23 2023-24
Public sector current expenditure (PSCE)
PSCE in RDEL 0.0 0.5 0.5 0.0 0.0
PSCE in AME 0.4 -1.7 -2.2 -0.1 0.0
of which:
Welfare spending 0.5 2.3 2.7 2.5 0.9
Net public service pension payments 0.0 0.0 0.1 0.4 0.2
Expenditure transfers to EU institutions 0.7 0.0 0.0 0.0 0.0
Assumed spending in lieu of EU transfers - 0.7 0.7 0.7 0.7
Locally financed current expenditure 0.1 0.3 0.3 0.2 0.1
Central government debt interest, net of APF -0.8 -4.9 -6.0 -3.8 -2.0
Other current expenditure 0.0 0.0 0.0 0.0 0.0
Total public sector current expenditure 0.4 -1.2 -1.7 -0.1 0.0
Public sector gross investment (PSGI)
PSGI in CDEL 0.0 0.0 0.0 0.0 0.0
PSGI in AME 0.0 0.0 0.0 0.0 0.0
Total public sector gross investment 0.0 0.0 0.0 0.0 0.0
Less public sector depreciation 0.0 0.0 0.0 0.0 0.0
Public sector net investment 0.0 0.0 0.0 0.0 0.0
Total managed expenditure 0.4 -1.2 -1.7 -0.1 0.0

Financial transactions and the crystallisation of contingent liabilities

10.51 Financial transactions push up public sector net debt by a cumulative £147 billion by 2023-
24, of which £121 billion is down to our assumption on the rolling over of TFS loans that
would otherwise have been repaid in 2020-21 and 2021-22 as in our March forecast.
10.52 Other financial transaction effects include:
• The £10 billion crystallisation of contingent liabilities related to loans for private sector
companies. We assume this support is provided in 2020-21.
• The assumed delay in asset sales for two years. By 2023-24, around £13½ billion of
planned asset sales have not taken place.
• Other smaller changes include those to accruals adjustments to bring the cash
borrowing and PSND elements of the stress test into line with the PSNB elements.
Table 10.11: Financial transactions: stress test scenario

£ billion
2019-20 2020-21 2021-22 2022-23 2023-24
Public sector net borrowing 34.8 51.3 51.4 46.4 37.3
Financial transactions 11.8 41.4 14.8 29.1 20.3
of which:
DEL net lending 6.0 16.5 5.9 6.4 2.3
Other government net lending 16.7 17.6 19.0 18.4 17.8
Sales or purchases of financial assets -4.9 0.0 -11.5 -5.3 -6.6
Bank of England schemes 0.0 0.0 0.0 0.0 0.0
Cash flow timing effects -6.0 7.4 1.5 9.6 6.9
Public sector net cash requirement 46.6 92.7 66.2 75.5 57.5
Table 10.12: Financial transactions: stress test versus March forecast

£ billion
2019-20 2020-21 2021-22 2022-23 2023-24
Financial transactions 11.5 66.8 66.3 3.1 -1.1
of which:
DEL net lending 0.0 10.0 0.0 0.0 0.0
Other government net lending 0.0 0.0 0.0 0.0 0.0
Sales or purchases of financial assets 11.5 5.2 -4.9 2.4 -0.7
Bank of England schemes 0.0 51.1 70.3 0.0 0.0
Cash flow timing effects 0.0 0.4 0.9 0.7 -0.3

Fiscal aggregates
Public sector net borrowing
10.53 The deficit is around £30 billion a year higher from 2020-21 onwards. It rises to 2.3 per
cent of GDP in 2021-22, then falls back gradually to 1½ per cent in 2023-24 as the
economy recovers, but remains 1 per cent of GDP higher than our March forecast.
10.54 The near-term deterioration in output has a significant cyclical element, but the GDP loss at
the five-year horizon is mostly structural. Cyclically adjusted net borrowing is modestly
higher than in our March forecast in 2020-21 (Chart 10.4). There are structural
improvements from higher customs revenues, a more fiscally-favourable composition of
GDP and lower debt interest spending. These largely offset structural deteriorations from
weaker potential output and asset markets. Structural borrowing rises steadily from 2021-22
onwards as the GDP composition and debt interest effects fade.
Chart 10.4: Fiscal aggregates: stress test versus March forecast

2.5 95
Cyclically adjusted PSNB Public sector net debt
2.0 90
Per cent of GDP
Per cent of GDP
1.5 85
1.0 80
Stress test
0.5 75
March 2019
0.0 70
2017-18 2019-20 2021-22 2023-24 2017-18 2019-20 2021-22 2023-24
Source: ONS, OBR
Balance sheet measures

10.55 Table 10.13 shows the impact of the stress test on public sector net debt (PSND). In our
March forecast, PSND falls relative to GDP in each year of the forecast. In the stress test, it
rises each year until 2021-22, before falling slowly in the final two years. This leaves PSND
in 2023-24, 12.1 per cent of GDP higher than in our March forecast. Of this, around 1.4
percentage points reflects lower nominal GDP, with the remainder due to the £272 billion
higher level of cash debt in 2023-24. This in turn is dominated by higher borrowing (which
adds £125 billion) and the assumed rolling over of TFS loans (£121 billion). Other factors,
including loans to private companies and delayed assets sales, have only modest effects.

Table 10.13: Public sector net debt: stress test versus March forecast
Per cent of GDP
2019-20 2020-21 2021-22 2022-23 2023-24
March forecast 82.2 79.0 74.9 74.0 73.0
Stress test forecast 83.2 86.3 86.8 86.6 85.1
Change 1.0 7.4 11.9 12.6 12.1
of which:
Change in nominal GDP 1 0.2 2.3 2.6 2.2 1.4
Change in cash level of net debt 0.8 5.1 9.2 10.3 10.8
£ billion
March forecast 1838 1828 1796 1838 1878
Stress test forecast 1855 1942 2010 2087 2150
Change in cash level of net debt 17 114 214 249 272
of which:
Borrowing 5 36 69 101 125
Crystallisation of contingent liability 0 10 10 10 10
Delay in asset sales 12 17 12 14 13
Term Funding Scheme 0 51 121 121 121
Other factors 0 0 1 2 2
1
Non-seasonally-adjusted GDP centred end-March.
The Government’s fiscal targets

10.56 The fiscal mandate in 2020-21 is met with a slightly smaller margin of £22.7 billion, but the
supplementary target in the same year is missed. Meeting the mandate by only a slightly
smaller margin than in March reflects two main elements of the scenario: first, a significant
element of the near-term hit to the economy is cyclical rather than structural; and second,
higher customs duties generate a near-term structural improvement to receipts, with the
detrimental effect of those tariffs on potential GDP building more gradually. With the target
year for the mandate only one year ahead, it does not provide an anchor for medium term
tax and spending decisions. The Government’s objective of returning the public finances to
balance by 2025-26 is further out of reach in the stress test than in our March forecast, with
borrowing standing at almost £40 billion in 2023-24 and declining at a modest pace.
10.57 None of these figures reflect the forthcoming changes to the accounting treatment of student
loans, which we estimate will raise the measured level of borrowing by between £11.2
billion and £13.7 billion over the next five years. This would reduce headroom against the
fiscal mandate in 2020-21 to £11.1 billion. As these accounting treatment changes relate to
when the cost of write-offs and interest income is accrued, they do not affect cash flows, so
the change in the debt-to-GDP ratio in the target year would be unaffected.

Table 10.14: Fiscal aggregates and other indicators

Per cent of GDP
2019-20 2020-21 2021-22 2022-23 2023-24
Receipts and expenditure
Public sector current receipts 36.8 36.8 36.9 36.9 37.0
Fiscal mandate and supplementary target
Cyclically adjusted net borrowing 1.2 1.0 0.6 0.7 0.8
1 83.2 86.3 86.8 86.6 85.1
Deficit
Current budget deficit -0.6 0.0 0.0 -0.2 -0.7
Cyclically adjusted current budget deficit -0.9 -1.3 -1.6 -1.5 -1.3
Primary deficit 0.2 1.1 1.1 0.8 0.4
Cyclically adjusted primary deficit -0.2 -0.2 -0.5 -0.5 -0.3
Financing
Central government net cash requirement 1.9 4.2 3.1 3.2 2.4
Alternative balance sheet metrics
Public sector net debt exc. Bank of England 74.9 78.1 78.9 79.1 77.9
Public sector net financial liabilities 66.2 67.5 67.4 66.6 65.0
£ billion
Public sector net debt 1855 1942 2010 2087 2150
Memo: Output gap (per cent of GDP) -0.8 -2.4 -2.3 -1.6 -0.7
1
Debt at end March; GDP centred on end March.

Table 10.15: Fiscal aggregates: stress test versus March forecast

Per cent of GDP
2019-20 2020-21 2021-22 2022-23 2023-24
Receipts and expenditure
Public sector current receipts -0.1 -0.3 -0.3 -0.2 -0.3
Fiscal mandate and supplementary target
Cyclically adjusted net borrowing -0.1 0.1 -0.1 0.1 0.3
1 1.0 7.4 11.9 12.6 12.1
Deficit
Current budget deficit 0.2 1.3 1.4 1.3 0.9
Cyclically adjusted current budget deficit -0.1 0.1 -0.1 0.0 0.3
Primary deficit 0.3 1.5 1.6 1.4 1.0
Cyclically adjusted primary deficit -0.1 0.2 0.0 0.1 0.3
Financing
Central government net cash requirement 0.8 2.1 1.4 1.5 0.9
Alternative balance sheet metrics
Public sector net debt exc. Bank of England 1.0 4.9 6.5 7.5 7.3
Public sector net financial liabilities 0.3 3.2 4.9 5.8 5.9
£ billion
Public sector net debt 17 114 214 249 272
Memo: Output gap (per cent of GDP) -0.6 -2.2 -2.2 -1.6 -0.7
1
Debt at end March; GDP centred on end March.
Other economic risks from Brexit

10.58 The stress test we have presented is by no means a worst-case scenario under a no-deal,
no-transition Brexit. Neither the cyclical downturn nor the medium-term loss of potential
output are as large as those considered in the Bank of England’s disruptive and disorderly
Brexit scenarios that were published last year.13 But even so it leaves borrowing around £30
billion a year higher and sees debt rising relative to GDP until 2021-22. In this section we
review some of the more adverse economic risks that a disorderly Brexit could bring.
Short-term risks from a disorderly Brexit

10.59 A disorderly Brexit could take many forms and would depend, in part, on the extent of
mitigating action taken by policy makers in both the UK and EU. Its economic and fiscal
effects would depend on the reactions of market participants, firms and households.
Temporary border disruptions

10.60 A disorderly exit could be expected to result in temporary constraints on the supply of some
imported products and domestic goods that rely on imported components. That might
occur, for instance, if a lack of customs preparedness led to significant delays at the border.
13
Bank of England, EU withdrawal scenarios and monetary and financial stability, November 2018.

If the UK and EU were unable to agree to continued mutual recognition (‘grandfathering’)

of existing product standards and professional qualifications, goods may need to be re-
approved or sale and loss of market access would severely restrict services trade.
10.61 The impact of such effects would depend not only on their extent, but also on their duration.
The IMF scenario that we base our stress test on includes only minimal border disruptions.
Its more disruptive one includes delays in customs processes that lower GDP by an
additional 2.2 per cent in the first two years after Brexit. The Bank’s ‘disorderly’ Brexit
scenario included short-run falls in UK trade and potential productivity of 15 and 5 per cent,
respectively, due to border disruptions. But the UK and the EU would presumably eventually
get the staff and systems in place to cope with the new trading relationship, so these
disruptions would be likely to have only temporary demand and supply consequences.
Distinguishing between supply and demand

10.62 Should we be required to produce a forecast based on a no-deal, no-transition Brexit, we
will need to form a judgement regarding the cyclical and structural components of its impact
on output. That judgement is critical to assessing performance against the Government’s
fiscal mandate, which is expressed in terms of cyclically adjusted borrowing in 2020-21. As
explained in Chapter 2, it is hard enough in normal times to separate in a robust way the
movements in activity into changes in potential output and the output gap. It would be
almost impossible to do so if a disorderly Brexit were underway.
10.63 Indeed, the distinction between supply and demand may not be very helpful at that point in
assessing the state of the public finances. A disorderly exit might well result in supply
capacity falling temporarily, for instance, from temporary border disruptions. If we reflected
that in our forecast for potential output, then any associated increase in borrowing would be
deemed structural rather than cyclical. But if the hit to supply was assumed to be temporary,
then the cyclically adjusted deficit would be assumed to improve as the temporary loss of
potential passed. In such a scenario, the cyclically adjusted balance in 2020-21 would
provide a misleading guide to the medium-term sustainability of the public finances.
Financial conditions
10.64 The stress test involves a small, temporary, tightening of financial conditions, so clearly the
tightening could be larger and more persistent. In the IMF’s more disruptive scenario, it
assumes that sovereign and corporate interest rate spreads rise by 100 and 150 basis
points respectively. In the Bank’s ‘disruptive’ no-deal scenario, the term premium on gilts
rises by 50 basis points, while interest rate spreads on household and business loans rise by
150 basis points. In its ‘disorderly’ scenario, they rise by 100 and 250 basis points
respectively – similar to the increases seen during the financial crisis.
10.65 A more pronounced tightening in financial conditions could result in a larger cyclical
deterioration in the economy as well as weighing on potential output in the medium term. In
addition to the lower receipts from the weaker economy, higher gilt yields would boost
government spending. A 1 percentage point permanent rise in gilt yields adds £0.5 billion
to debt interest spending in the first year, rising to £4.2 billion a year after five years.

10.66 Asset prices – including the sterling exchange rate – could fall sharply in a disorderly Brexit,
reflecting the likely deterioration in financial market participants’ views about the future
economic outlook and heightened risk premia. Such falls could be especially large if a no-
deal Brexit was associated with a loss of trust in the effectiveness of the UK’s political
machinery. Our stress test includes a 5 per cent fall in UK equity prices and an initial 10 per
cent fall in the pound. But these market movements could be more significant. For example,
the Bank of England’s ‘disruptive’ no-deal scenario included an initial 15 per cent
depreciation and its ‘disorderly’ scenario assumed a 25 per cent drop.
10.67 Typically, the most fiscally significant effect of such a depreciation is to push inflation up
relative to earnings. This raises inflation-linked spending (especially debt interest) and
lowers receipts through a reversal of fiscal drag (i.e. inflation-linked tax thresholds rise faster
than earnings, meaning a smaller proportion of income is taxed at higher rates).
Policy response
10.68 The MPC’s response to a no-deal Brexit would depend on the balance of the effects on
demand, supply and the exchange rate. At an appearance at the Treasury Select Committee
on 26 June, Governor Mark Carney stated that in a no-deal Brexit, “it is more likely we
would provide some stimulus” and “we have said we would do what we could in the event of
a no-deal scenario but there is no guarantee on that”. The stress test includes an easing of
monetary policy. That is in line with the Bank’s response in the wake of the 2016
referendum, when it cut Bank Rate and took other measures to support activity. But in the
case of a disorderly Brexit, the inflationary consequences of a fall in supply, a weaker pound
and, potentially, the imposition of new tariffs could limit the MPC’s scope to ease policy.
10.69 In the stress test, Bank Rate troughs at around 0.2 per cent. This is likely to be close to its
effective lower bound so any further monetary stimulus in a more disruptive scenario would
probably occur via unconventional measures. These can have quite large direct effects on
borrowing (quantitative easing via gilt purchases lowers debt interest spending when Bank
Rate is lower than gilt yields) and public sector net debt (if the market price of gilts
purchased is higher than their nominal value or where other forms of unconventional easing
involve the purchase of assets deemed illiquid for PSND purposes).14 But these direct fiscal
effects would probably be less important than the indirect ones via growth and inflation.
10.70 The stress test assumes the automatic fiscal stabilisers operate, but discretionary fiscal policy
is unchanged. If instead the Government eased fiscal policy to support the economy, then
the public finances would obviously deteriorate more significantly as the direct cost of fiscal
measures would only be partly offset by the indirect gain from supporting growth.
10.71 The fiscal risk posed by a cyclical downturn is increased by the proximity of Bank Rate to its
effective lower bound and by doubts that some have expressed about the effectiveness of
unconventional monetary policy, both of which suggest that the authorities may rely more
heavily on fiscal stimulus measures than would previously have been the case.
14
For more information, see OBR, The direct fiscal consequences of unconventional monetary policies, 13 March 2019.

Longer-term Brexit risks to potential output

10.72 The greater risk of Brexit over the longer term relates to the future trading and migration
relationships between the UK and the EU. These risks would remain even if the UK and EU
implemented a withdrawal agreement that includes a transition period. The consequences
of changes in the trading and migration relationship for potential output are particularly
important for the public finances, and are especially uncertain.
Migration
10.73 In the IMF’s no-deal scenarios, net inward migration is assumed to be 25,000 lower than
their baseline scenario each year out to 2030 because of stricter immigration policies. This
lowers potential output in the long term by 0.6 per cent. A variety of studies suggest that net
inward migration is on average positive for the public finances over the medium term.15 So
lower inward migration from the EU might be expected to weaken the public finances. That
said, it is possible that any changes in migration from non-EU countries could offset a fall in
EU migration, as has happened to some extent since the EU referendum.
10.74 Inward migration could also be affected by ‘pull’ factors such as the value of wages in
prospective migrants’ home currency. This would be especially relevant in the case of a
disorderly Brexit or if the UK and EU ended up trading on WTO terms. The Bank’s
‘disruptive’ no-deal scenario included a much larger fall in net inward migration to around
30,000 a year versus the ONS central projection of 165,000 a year, due to both policy and
‘pull’ factors. Even in the Bank’s scenario that assumes a smooth transition to trading on
WTO terms, net inward migration was assumed to fall to 85,000 a year by 2023.
10.75 In our March 2016 EFO, we examined the potential implications of lower net inward
migration of 105,000 a year rather than 185,000 in 2021. By 2020-21, the final year of
our forecast at the time, this increased the budget deficit by £5.9 billion or 0.3 per cent of
GDP (taking into account the impact on cash receipts and spending and the reduction in
GDP). The fiscal impact of reduced migration reflected not just a reduction in the overall size
of the population, but also a shift to a less favourable age structure as the foregone inward
migrants were more likely to be of working age than the native stock.
Productivity
10.76 In the IMF’s no-deal scenarios, the impact on potential productivity over the longer term
comes from the effect that moving to trading on WTO terms has on capital deepening.
Increased trade barriers with the EU are assumed to reduce the returns to capital, the
desired capital stock and business investment, which in turn renders workers less productive
than they would otherwise be. To date, our forecasts for the effect of Brexit on potential
productivity have also concentrated on this capital deepening channel, though operating
through the discouraging effect of heightened uncertainty on business investment.
15
See Chapter 3, OBR, Discussion paper No.3: Brexit and the OBR’s forecasts, October 2018, for more details on the effects of migration
on the public finances.

10.77 The increase in trade barriers between the UK and the EU from trading on WTO terms
would also weigh on the UK’s export market share and the import intensity of demand,
lowering trade as a share of the economy. Economic theory and empirical evidence suggest
that greater trade intensity leads to increases in productivity by allowing economies to
specialise more in their areas of comparative advantage. This is because specialisation
allows an economy to use its resources more efficiently. The IMF’s scenarios did not include
an effect on productivity from lower trade intensity, but several other studies have – although
the range of estimates is relatively wide.
10.78 The effects on productivity discussed so far are ‘static’ ones – a one-off decline in potential
output associated with the less effective exploitation of comparative advantage. These one-
off shifts will affect the growth rate of the economy for several years as the economy adjusts
to the new equilibrium. On top of these static effects, some studies suggest that barriers to
trade and migration are also likely to have adverse dynamic effects on productivity and
potential output, for example, by impeding technology transfer and slowing innovation and
technological progress. There is little consensus regarding the size of such effects, though
some Brexit studies have attempted to allow for them. The Bank’s no-deal scenarios
included a 5 per cent hit over the subsequent five years. The relevant studies that we
covered in our Brexit Discussion paper showed an average total effect (i.e. including the
static impact as well) on long-run productivity from moving to trading with the rest of the EU
on WTO rules of around 8 per cent.
10.79 To calibrate the potential fiscal effects of lower productivity, in our November 2016 EFO, we
produced a ‘weak productivity’ scenario where potential productivity growth was just 0.8 per
cent a year compared to an average of about 1.8 per cent in our central forecast. After five
years, public sector net borrowing was £41 billion or 1.9 per cent of GDP higher than in the
central forecast and net debt was 8.0 per cent of GDP higher.
Other risks to tax receipts from a no-deal Brexit

10.80 A no-deal Brexit would require immediate changes in several tax regimes (for example, VAT
and the EU emissions trading system (ETS)) that are based on common EU rules and would
introduce tax compliance risks associated with the effectiveness of new tax systems and
border controls. We have not attempted here to quantify any receipts losses from such
changes in tax regimes or increased avoidance, evasion or non-compliance.
10.81 Policy intentions have been announced in some areas, including some parameters of a
replacement system for the EU ETS. But to cost such measures we would need more detail
on their implementation. Risks to receipts from non-compliance will depend hugely on the
extent of border disruptions following a no-deal exit. In the medium term, it will also depend
on the extent of cooperation between the EU and UK tax authorities. But it is clear that risks
lie to the downside relative to the assumptions in the stress test.
10.82 It is not possible to quantify these risks, but for the simple reason that it is such a large
source of revenue, risks to VAT receipts are likely to be the most material. By way of context,
every 1 percentage point increase in the ‘VAT gap’ lowers receipts by around £1½ billion.

Non-compliance risks
10.83 The effectiveness of the customs and tax systems following a no-deal Brexit remains
uncertain. We noted in our March EFO that HMRC’s new customs declaration service was
being rolled out at a slower pace than first anticipated, with HMRC prioritising the
upgrading of its existing ‘CHIEF’ system instead. A no-deal exit would mean that customs
duties would be applied to imports from the EU under the one-year temporary tariff regime.
This would challenge the smooth operation of HMRC’s systems and the readiness of
businesses to submit customs declarations.
10.84 The Northern Ireland border is a further fiscal risk. Arrangements for the border following a
no-deal exit remain uncertain. If a future tariff regime were not fully applied at the Northern
Ireland border, either by design or for operational reasons, we would need to consider the
associated loss of customs revenues due to non-compliance. This could be sizeable if tariffs
could be avoided relatively easily by routing goods through Northern Ireland. It could also
have knock-on effects on other taxes collected at the border such as import VAT and excise
duties.
Tax regime changes

10.85 While in the EU, the VAT system is based on European directives applied in national
legislation. A no-deal exit would mean an abrupt move to a UK-only VAT regime. The EU-
based VAT regime currently treats imports from within the EU and those from outside the EU
differently. Goods from outside the EU are liable for VAT immediately on arrival, rather than
just after a sale occurs as is the case for goods and services supplied from the EU. In the
absence of policy changes, a move to a UK-only VAT regime would mean that goods
arriving from the EU would also be liable to import VAT.
10.86 The Government has announced that if the UK left the EU without a deal it would introduce
postponed accounting for import VAT on goods brought into the UK from all countries. This
would mean that import VAT would be paid via a VAT return. This would both mitigate
potential cashflow problems for importers from the EU who otherwise would have had to
pay VAT earlier in the process and improve cashflow for importers from non-EU countries
relative to the current system. This would clearly generate a timing effect on receipts (with
VAT payments made later than now), but we would need to consider whether this might also
represent an opportunity for increased fraud.
10.87 The Government has also announced several other changes to the VAT system in the event
of a no-deal exit. These include the extension of VAT zero-rating to exports of financial
services and that ‘low value consignment relief’ would not be available to goods entering
from the EU. The financial services measure would bring the treatment of exports of such
services to the EU (currently exempt) in line with those to non-EU countries. It would not
change the VAT charged by businesses, as none is charged whether they are exempt or
zero-rated. However, VAT on purchases attributable to zero-rated outputs can be reclaimed.
This would increase the recovery of input VAT by businesses exporting financial services to
the EU and would therefore lower VAT receipts overall. Low value consignment relief allows

for goods of a low value (under £15 in the UK) to be exempt from VAT when they are
imported from outside the EU. In the event of a no-deal exit, this relief would be removed
for all imports, boosting import VAT and possibly removing a vehicle for fraudulent activity
in the current system.
10.88 The Government has also announced that in the case of a no-deal exit, it would replace the
EU emissions trading scheme with a carbon emissions tax which would apply to all
installations participating in the EU scheme. This would apply from the first day following
exit. It stipulated a rate of £16 per tonne of carbon would be applied, which was close to
the EU ETS carbon price at the time of the announcement. Given the large fluctuations in
carbon prices since then, it is unclear what applying such a rate would cost relative to an
updated EU ETS baseline forecast, or whether the rate might change.
Conclusion
10.89 The stress test presented in this chapter illustrates the fiscal consequences of one possible
no-deal Brexit scenario. It is not the most disruptive one we could have chosen. For an
illustration of a more severe variant, the results of our 2017 stress test might be a better
guide. But it still adds around £30 billion a year to borrowing (which would come on top of
the more than £10 billion a year that will be added when the accounting treatment of
student loans is improved later this year). And it sees debt rising rather than falling relative
to GDP over the next three years.
10.90 We also discuss the many other factors about which judgements would have to be made if
we needed to produce a central forecast on the basis of a no-deal exit. These range from
the near-impossible task of distinguishing between supply and demand effects in what
might be substantial fluctuations in economic activity, to knotty questions about tax
compliance at the border. Uncertainty would be the watchword.

Index of charts and tables
Executive summary
Chart 1: Stress test vs March forecast: sources of differences in PSNB .............................. 15
Figure 1: Sources of fiscal risk over the medium term ..................................................... 16
Figure 2: Sources of risk to fiscal sustainability ............................................................... 16
Chapter 1 Introduction
Chart 1.1: Public sector debt and borrowing since 1800 ................................................ 22
Figure 1.1: A stylised fiscal risk matrix ........................................................................... 23
Chart 1.2: Changes to the OBR’s borrowing forecast since our March 2017 EFO ............ 27
Table 1.1: Like-for-like pre-measures changes in receipts since our March 2017 EFO ..... 28
Table 1.2: The effect of policy measures on borrowing since our March 2017 EFO .......... 28
Chapter 2 Macroeconomic risks

Chart 2.1: Contributions of adult population growth to potential output growth ............... 33
Chart 2.2: The employment rate versus the output gap .................................................. 34
Chart 2.3: Productivity growth ....................................................................................... 36
Chart 2.4: The economic cycle and its impact on the budget balance ............................. 38
Table 2.1: UK recessions since 1955 ............................................................................. 40
Chart A: Monthly and quarterly GDP growth.................................................................. 41
Table 2.2: Real-time and latest output gap estimates ahead of the 2008 recession .......... 43
Chart 2.5: Real-time and latest output gap estimates ..................................................... 43
Chart 2.6: Selected potential output growth estimates and forecasts................................ 44
Chart 2.7: Real-time and latest estimates of cyclically adjusted PSNB .............................. 45
Chart 2.8: Revisions to cyclically adjusted PSNB by source .............................................. 46
Chart 2.9: Cyclically adjusted PSNB and the output gap on the real-time data ................ 47
Chart 2.10: Cyclically adjusted PSNB and the output gap on the latest data .................... 47
Chart 2.11: Selected components of GDP and associated effective tax rates .................... 49
Chart 2.12: Labour share of GDP by income ................................................................. 51
Chart 2.13: Growth of real GDP and real house prices .................................................. 52
Chart 2.14: Sectoral net lending ................................................................................... 53

Chart 2.15: Household gross debt to income................................................................. 54
Chapter 3 Risks from the financial sector

Chart 3.1: Banks’ assets relative to nominal GDP .......................................................... 58
Chart 3.2: Assets of financial institutions in advanced economies .................................... 59
Chart 3.3: Sources of tax receipts from the banking sector ............................................. 60
Table 3.1: Gross and net cash flows of financial sector interventions ............................... 62
Table 3.2: Economic functions within the FSB’s narrow measure of non-bank financial
intermediation .................................................................................................. 68
Chart 3.4: FSB estimates of the narrow measure of shadow banking in the UK ................ 69
Chapter 4 Revenue risks

Chart 4.1: Successive forecasts for total receipts ............................................................ 73
Chart 4.2: Two-year ahead forecast differences from successive OBR forecasts ............... 74
Chart 4.3: Income tax contributions from the top percentile ............................................ 76
Chart 4.4: Composition of residential SDLT tax base and revenues by property value ...... 77
Chart 4.5: Composition of CGT tax base and revenues by value of overall disposals ....... 77
Chart 4.6: Tax due on £70,000 of income .................................................................... 79
Chart 4.7: Trends in self-employment and incorporations .............................................. 80
Chart 4.8: Trends in fuel, tobacco and alcohol consumption .......................................... 82
Chart 4.9: Fuel and alcohol receipts: drivers of growth over the medium-term ................ 84
Chart 4.10: Relative uncertainty of tax measures announced since December 2014 ........ 87
Chart 4.11: HMRC estimate of the tax gap since 2005-06 ............................................. 90
Chart 4.12: Changes in the tax gap: 2007-08 to 2017-18 ............................................ 91
Chart 4.13: Estimated total cost of tax reliefs in 2018-19 ............................................... 97
Chart 4.14: Tax expenditures as a percentage of GDP ................................................... 99
Chart 4.15: International comparison of the cost of tax expenditures in 2010 ................ 100
Chart 4.16: VAT threshold cross-country comparison ................................................... 100
Chart 4.17: Tax expenditure cost estimates: six case studies ......................................... 102
Table 4.1: Tax expenditure cost forecasts: six case studies ............................................ 103
Chart 4.18: Expenditure in R&D tax credits claims versus total R&D expenditure ............ 105
Chart 4.19: Inheritance tax average effective tax rate by net estate value ...................... 108
Chart 4.20: The proportion of businesses with a website .............................................. 112
Chart 4.21: Growth in global internet traffic ................................................................ 113
Chart 4.22: Retailing spending, store-only versus online-only sales ............................... 117

Chapter 5 Primary spending risks
Chart 5.1: Successive forecasts for total public spending .............................................. 125
Chart 5.2: Two-year ahead forecast differences from successive OBR forecasts ............. 126
Chart 5.3: Average public spending per person by age ................................................ 128
Chart 5.4: New health settlement versus different Health Foundation/IFS scenarios ....... 134
Chart 5.5: Real-terms funding and productivity growth in the health service .................. 135
Chart 5.6: Adult social care spending in 2067-68: alternative scenarios ....................... 139
Chart 5.7: Nuclear decommissioning provisions .......................................................... 160
Chart 5.8: Clinical negligence payments as a proportion of the DHSC budget .............. 162
Chart 5.9: HMRC contingent liabilities in litigation cases .............................................. 164
Chapter 6 Balance sheet risks

Table 6.1: Various measures of the public sector balance sheet in 2017-18 .................. 170
Chart 6.1: Contributions to changes in net debt since 1997-98 .................................... 171
Chart 6.2: Public sector and general government net worth .......................................... 172
Chart 6.3: Public sector balance sheets in selected advanced economies ...................... 173
Chart 6.4: Successive PSNFL forecasts and outturns ..................................................... 174
Chart 6.5: Successive OBR forecasts of proceeds from financial asset sales ................... 176
Chart A: Housing associations’ contribution to PSND ................................................... 184
Chart 6.6: Intangible assets in the UK public sector by type .......................................... 191
Chart 6.7: Public sector intangible assets: international comparison.............................. 192
Chapter 7 Debt interest risks

Chart 7.1: Total debt interest spending by government sector ....................................... 197
Table 7.1: General government gross debt forecasts .................................................... 199
Chart 7.2: Successive forecasts for Bank Rate and gilt rates .......................................... 200
Chart 7.3: Debt interest ready reckoners ..................................................................... 203
Chart 7.4: Public sector net debt since 1900 ................................................................ 208
Chart 7.5: Contributions to changes in the debt-to-GDP ratio by decade ...................... 208
Chart 7.6: Decomposing the growth-corrected interest rate .......................................... 209
Chart 7.7: Distribution of growth-corrected interest rates: 1900-01 to 2018-19 ............ 210
Chart 7.8: Growth-corrected interest rates in advanced economies: 1961 to 2017 ........ 210
Chart 7.9: The effect of R-G on debt in the UK in overlapping 20-year periods ............. 211
Table 7.2: Medium-term forecast for the growth-corrected interest rate ......................... 213
Chart 7.10: IMF forecasts for the growth-corrected interest rate .................................... 213

Chart 7.11: Combinations of R-G levels and debt-stabilising primary balances ............. 215
Chart 7.12: Growth-corrected interest rates and debt-stabilising primary balances
at different starting levels of debt ..................................................................... 216
Chapter 8 Fiscal policy risks

Chart 8.1: Explaining annual changes in net borrowing since 2007-08 ........................ 224
Chart 8.2: Headroom against successive fiscal mandates since June 2010.................... 226
Chart 8.3: Sources of revision to PSNB forecasts since June 2010 ................................. 229
Chart 8.4: Discretionary policy responses to our underlying forecast revisions ............... 230
Chart 8.5: Cumulative effect of discretionary policy since June 2010 ............................ 232
Chart 8.6: Cumulative effect of DEL policy decisions since June 2010 ........................... 233
Chart 8.7: Cumulative changes in RDEL since 2014-15 ............................................... 234
Chart 8.8: Maximum levels of borrowing consistent with the debt-to-GDP ratio falling ... 236
Chapter 9 Climate change

Chart 9.1: Global average temperatures ..................................................................... 239
Table 9.1: Examples of macroeconomic risks from climate change ............................... 243
Figure A: NGFS scenario analysis framework............................................................... 252
Chapter 10 A fiscal stress test

Chart 10.1: Real GDP: stress test versus March forecast ............................................... 258
Chart 10.2: Contributions to difference in nominal GDP from March forecast ............... 260
Table 10.1: Economic assumptions: stress test ............................................................. 261
Table 10.2: Economic assumptions: stress test versus March forecast ............................ 262
Table 10.3: Fiscal determinants: stress test .................................................................. 263
Table 10.4: Fiscal determinants: stress test versus March forecast ................................. 264
Table 10.5: Contributions to long-run impact on GDP from trading
with the EU on WTO terms ...................................................................... 265
Chart 10.3: Stress test versus March forecast: sources of differences in PSNB ................ 267
Table 10.6: Current receipts: the stress test scenario .................................................... 272
Table 10.7: Current receipts: stress test versus March forecast ...................................... 273
Table 10.8: Debt interest spending: stress test versus March forecast............................. 275
Table 10.9: Public spending: stress test scenario .......................................................... 276
Table 10.10: Public spending: stress test scenario versus March forecast ....................... 276
Table 10.11: Financial transactions: stress test scenario ............................................... 277
Table 10.12: Financial transactions: stress test versus March forecast ............................ 277

Chart 10.4: Fiscal aggregates: stress test versus March forecast .................................... 278
Table 10.13: Public sector net debt: stress test versus March forecast ............................ 279
Table 10.14: Fiscal aggregates and other indicators .................................................... 280
Table 10.15: Fiscal aggregates: stress test versus March forecast .................................. 281

CCS0719538076
978-1-5286-1497-9

Riscos Fiscais UK - 2019

Uploaded by

Copyright:

Available Formats

Riscos Fiscais UK - 2019

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Riscos Fiscais UK - 2019

Uploaded by

Copyright:

Available Formats

Fiscal risks report

Presented to Parliament by the

This publication is licensed under the terms of the Open Government

This publication is available at www.gov.uk/official-documents

Any enquiries regarding this publication should be sent to us at

Printed on paper containing 75% recycled fibre content minimum

Printed in the UK by the APS Group on behalf of the Controller of

Her Majesty’s Stationery Office

Executive summary ........................................................................ 3

Chapter 1 Introduction ................................................................................ 17

Chapter 2 Macroeconomic risks ................................................................... 31

Chapter 3 Risks from the financial sector ...................................................... 57

Chapter 4 Revenue risks .............................................................................. 73

Chapter 5 Primary spending risks ............................................................... 125

Chapter 6 Balance sheet risks .................................................................... 169

Chapter 7 Debt interest risks ...................................................................... 197

Chapter 8 Policy risks ................................................................................ 221

Chapter 9 Climate change ........................................................................ 239

Chapter 10 A fiscal stress test....................................................................... 255

Index of charts and tables .......................................................... 289

1 Fiscal risks report

Robert Chote Sir Charles Bean Andy King

The Budget Responsibility Committee

Fiscal risks report 2

3 Fiscal risks report

Fiscal risks report 4

5 Fiscal risks report

• Two years ago we assigned a medium likelihood to trend productivity growth

Fiscal risks report 6

Financial sector risks

7 Fiscal risks report

Fiscal risks report 8

Primary spending risks

9 Fiscal risks report

Balance sheet risks

Fiscal risks report 10

Debt interest risks

11 Fiscal risks report

50 Unsurprisingly, there is a similar pattern of accommodative revisions to already announced

Fiscal risks report 12

13 Fiscal risks report

A no-deal Brexit stress test scenario

Fiscal risks report 14

Chart 1: Stress test vs March forecast: sources of differences in PSNB

Summary of medium and long-term fiscal risks

15 Fiscal risks report

Figure 1: Sources of fiscal risk over the medium term

Change since 2017

Fiscal risks report

• 2017 stress test: severe New risk for 2019

• Weak productivity growth

• Output gap mismeasurement • Discretionary fiscal loosening

Medium Very low High Low

Figure 2: Sources of risk to fiscal sustainability

HIgh (Over 100)

Possible major fiscal Fiscal risks report

• Health spending: ageing

• Greater fuel efficiency

incorporations • Lower smoking