Bachner 2019

Energy Strategy Reviews 26 (2019) 100373
Contents lists available at ScienceDirect
Energy Strategy Reviews

journal homepage: www.elsevier.com/locate/esr
Costs or benefits? Assessing the economy-wide effects of the electricity T

sector's low carbon transition – The role of capital costs, divergent risk
perceptions and premiums
Gabriel Bachnera,∗, Jakob Mayera, Karl W. Steiningera,b
a
Wegener Center for Climate and Global Change, University of Graz AT, Austria
b
Department of Economics, University of Graz AT, Austria
ARTICLE INFO ABSTRACT
Keywords: To mitigate climate change, societies strive to transform the energy sector towards greenhouse gas emission
Climate change mitigation neutrality, a move which assessment studies often indicate incurs large macroeconomic costs. In this context the
Electricity weighted average costs of capital (WACC) are especially important, as renewables are highly capital intensive. In
Europe particular, investors' perceptions and expectations of risks are fundamental determinants of WACC and thus
Risk
strongly influence the macroeconomic outcome of transition analyses. For the case of Europe's electricity sector
Capital costs
transition, we analyze this sensitivity by choosing different WACC settings, driven also by different policy set-
tings redirecting expectations. First, we find that when differentiating WACC across regions and technologies
more accurately than usually done in the literature, immediate and substantial macroeconomic benefits from the
transition emerge. We thereby reveal a systematic overestimation of low-carbon transition costs in the literature.
Second, we find that when pricing-in increasing trust in renewables, these benefits get significantly larger,
outweighing possible negative macroeconomic effects from the risk of stranding of fossil-based assets. We also
demonstrate that in developed regions such as Europe, de-risking renewables is an effective lever for reaching
climate targets, which indicates the relevance of green macroprudential regulation.
1. Introduction Irrespective of how BU and TD models are combined, the results of

such integrated BU-TD assessments are very sensitive to the underlying
The target of limiting global temperature rise “well below 2 °C” [1] assumptions on technology costs and their development over time. One
requires a fundamental transformation of the global economic system parameter that is particularly relevant in energy transition analysis is
towards carbon neutrality, with the energy sector – currently a core capital cost, due to the high cost share of capital in energy technologies
emitter – playing a key role [2]. From a policy-makers’ perspective it is in general and an even higher one in renewables in particular. More
crucial to understand the macroeconomic effects of such a transfor- explicitly, it is the weighted average costs of capital (WACC) parameter
mation, including indirect effects to other sectors or the labor market and its development over time that strongly drives results [11,12]. Yet,
and eventually implications on economic growth and economy-wide with respect to capital cost assumptions we identify three major
welfare. For assessing such effects a rich spectrum of integrated or shortcomings in the literature, which we address in this paper.
“hybrid” energy-economic models has been developed in recent dec- The first shortcoming is that WACC parameters are typically chosen
ades [3]. These models typically use information from detailed bottom- without much differentiation across technologies and regions, even
up (BU) energy sector models (such as the well-known TIMES model though it has been shown that there are substantial differences [13–15].
[4]) and feed it into top-down (TD) macroeconomic models (such as While it is uncontroversial that technologies with high capital in-
EPPA [5]). This allows for making use of the strengths of both model tensities (and large upfront investment costs) benefit stronger from
types [6], enabling researchers to study the macroeconomic effects of lower interest rates than less capital intensive technologies [12], most
interventions in the energy system. In many cases the link from BU to empirical work still applies uniform interest rates, not just across
TD is unidirectional (see e.g. [7]), however stronger integration is also technologies, but also across regions and other domains of risks (for an
possible (e.g. via iteration [8], or via full integration [9,10]). overview see Table A 1 and references thereof). Interest rates are
∗
Corresponding author.
E-mail address: gabriel.bachner@uni-graz.at (G. Bachner).
https://doi.org/10.1016/j.esr.2019.100373
Received 24 September 2018; Received in revised form 5 April 2019; Accepted 23 June 2019
2211-467X/ © 2019 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/BY/4.0/).
G. Bachner, et al. Energy Strategy Reviews 26 (2019) 100373
usually set between 5% and 10% and are mostly not further motivated. conversely, even a reduction of stringency (as exemplified by the US).
For instance, Pfenninger and Keirstead [16, p.307] indicate “an interest Investors thus face a risk from climate policy instability. These fossil-
rate of 10% is assumed for [technologies], so no assumptions about different fuel-related risks are connected to the literature of the so called “carbon
investment attractiveness or project financing models are made.” Most re- bubble” [33,34]: if humanity wants to meet the 2°C-target, large frac-
presentative TD policy evaluation studies do not even mention the tions of currently known fossil reserves need to remain in the ground
underlying interest rate explicitly, adding substantial uncertainty to the [10] and reserves, as well as infrastructure, could end up as stranded
validity of the qualitative insight of results. Although some BU and TD assets [35]. If markets do not adequately price in this risk of stranding,
assessments discuss sensitivities regarding uniform interest rate as- a bubble might emerge due to overvaluation of fossil fuel assets. Recent
sumptions, only few distinguish interest rates by technology. For in- studies indicate that this risk is already materializing on stock [36] and
stance, Pleßmann and Blechinger [17] attach an interest rate of 6% to on capital markets [35,36] as a fossil fuel risk premium. However, as
mature technologies (e.g. gas fired electricity generation or hydro- policy has only declared a temperature target, but no explicit emission
power) and 7% to (in these authors' perception) riskier technologies trajectory to reach this target,4 whether as well as when this risk will
(e.g. photovoltaics (PV) or wind power). However, the values are materialize (and to which degree) remains unknown.
chosen arbitrarily and without further motivation, and differentiation In addressing the three gaps identified, we contribute to a more
across regions is not taken into account. The most comprehensive study robust and adequate basis of information for the decision-making pro-
to our knowledge is by García-Gusano et al. [18], who deploy tech- cess in climate change policy. Due to the first gap (no differentiation of
nology-specific interest rates1 for 17 different electricity generation WACC across regions and technologies) the economy-wide costs of
technologies, with changes over time motivated by a decreasing im- mitigation might be substantially over- or underestimated for specific
plicit technological risk. While their results demonstrate the importance regions, depending on the regional characteristics (such as political
of using differentiated interest rates, they do not go beyond the energy stability, distortions, capital scarcity and factors from other risk do-
sector's domain and only give aggregated results for the whole of mains [39]). Possibly of more importance, the presence of the latter two
Europe. To account for a such required differentiation across electricity shortcomings in the literature (neglecting the potential of de-risking
generation technologies and regions in a transparent and consistent renewables in developed regions as well as neglecting the risk of
way, we build upon the empirical analysis of Steffen [19] who identifies stranding of fossil fuel-oriented assets) implies that the change in risk
differences in financing schemes as main drivers of WACC differences. perceptions has so far been overlooked in the analysis of low carbon
The second gap we identify concerns WACC reduction for renew- transition pathways. Consequently, the costs of mitigation from a shift
ables via “de-risking”. Besides civil society initiatives such as the to renewables might have been (possibly even substantially) over-
“Divestment” movement [20], the greening of finance is also on the estimated. We thus argue for explicit accounting of such changes in risk
agenda of corporate initiatives such as “Climate Action in Financial (perceptions) and corresponding risk premiums in economy-wide nu-
Institutions”, a global collaboration amongst various public and private merical analysis and simulations. This will allow for gaining insights
banks.2 While de-risking and greening is discussed in the literature, it into magnitudes and ranges of effects, acknowledging these refinements
focuses on developing regions, most notably Africa and the MENA re- of fundamental parameters. We propose to do so in two ways: First, the
gion (see e.g. refs [14,21–23]) and on the role of development banks “reference” pathways, to which transition pathways are usually com-
[24]. However, de-risking could also be an effective leverage point for pared, need to include at least the technological-change-driven risk for
climate change mitigation policy in developed regions [25]. Moreover, fossil fuel investments, as technological change happens independently
it has been shown that the stand-alone risk3 of variable renewables of climate policy. Second, the mitigation scenario, e.g. a low carbon
decreases with higher market shares [26] and that project risk for PV transition pathway, should include different WACC settings. On the one
and wind has fallen in the recent years [27]. Yet, for Europe and its hand, the mitigation scenario should include a lower risk premium
regions an economy-wide modeling study on the potential effects of de- (WACC) for renewables than in the reference, due to strengthened trust
risking of renewables is still missing (with the notable exception of first in renewables via the policy(-signal) itself. On the other hand, the mi-
estimates by [28]). tigation scenario should also include a higher WACC for fossil-fueled
The third shortcoming is that there is a strong focus on the WACC of technologies than the reference scenario, due to investors’ perception of
renewables only (see e.g. refs [12,25]), as these technologies seem to be likely further increases in the speed/stringency of climate policy (i.e.
more risky and offer de-risking potentials [21,29]. In contrast, the as- "ratcheting up"); on top of what has been already declared by govern-
sumptions on WACC for fossil fuels are left unattended and fixed to ments. In our analysis, we include these different risks by refining
historic or arbitrary “standard” values from the literature. This is pro- WACC assumptions in a stepwise manner. This allows us to analytically
blematic, as there is a substantial technological-change-driven risk that isolate the respective effects occurring simultaneously in the real world.
fossil fuel assets become less competitive or even stranded due to the This logic of including different risk premiums (due to divergent risk
rapid cost decline for renewables, especially for PV panels and batteries perceptions and expectations) in different narratives has so far not been
– a decline both already observed and expected to continue – even carried over into numerical model analysis and we believe we are the
without any further climate policy [30]. Moreover, current and future first to do so. In general, the peer-reviewed literature on the de-risking
(climate) policy signals add another uncertainty dimension to this of renewables and stranded fossil fuel assets is relatively5 scarce, even
technological-change-driven risk [31]. Even though governments de- more so from an economy-wide perspective. Also, in the domain of the
clare targets (or pathways), these are typically of long-term character finance literature there is only very little peer-reviewed literature on
and thus might simply not materialize in a world dominated by short- climate change mitigation issues [40], making it even harder to esti-
term political cycles. The associated possible adjustment of targets on mate economy-wide effects. Nevertheless, the topic of stranded fossil
short notice can be either an increase in the speed and/or stringency of fuel assets is analyzed from an economy-wide perspective in some
climate policy (e.g. by the “ratcheting-up”-mechanism [32]) or, studies. Mercure et al. [30] analyze the macroeconomic implications of
stranded fossil fuel assets from a demand perspective (globally, using
the demand constrained model E3ME coupled with an energy sector
1
They denote them as being “hurdle” rates.
2
https://www.mainstreamingclimate.org/.
3 4
Stand-alone risk means that an asset is considered in isolation, resembling a Note that many different emission trajectories can lead to the same level of
project finance structure. Alternatively, a portfolio approach would assume that greenhouse gas concentrations (and respectively of global warming).
5
the asset is part of a larger corporate structure, resembling a corporate finance Compared to the vast literature on the economics of climate change miti-
structure. gation.
2
model) and demonstrate that even without additional climate policy, (referring to a long-run macroeconomic state). We assume a fixed
the risk of stranding is substantial. They also show severe GDP effects savings rate, meaning that a fixed part of regional income is spent for
for countries that do not comply with international climate policy investment, which accumulates over time to build up the generic ca-
agreements. Further, Bauer et al. [41] investigate the opposing effects pital stock. Labor supply follows (working-age) population growth.
on greenhouse gas emissions coming from the “green paradox” effect Regarding the general socio-economic development, we follow the
(i.e. selling-out fossil fuels as long as they are still profitable to avoid shared socio-economic pathway (SSP) framework [78] and use SSP2.
assets becoming stranded) and the divestment effect (i.e. reducing in- The sectoral resolution in the model distinguishes between sixteen
vestments in fossil fuel projects due to uncertain profitability). The economic sectors. The “Electricity” sector distinguishes eight different
authors make use of the intertemporal optimization (perfect foresight) generation technologies, implemented as separate sub-sectors with
model REMIND and use a globally binding CO2 price as policy instru- Leontief-type production technologies,7 and a remainder representing
ment to stimulate an “announcement effect.” They conclude that the “Collection and Distribution” (CAD). A Leontief-nested electricity mix
divestment effect dominates CO2 emission changes (with net reduc- aggregate (together with CAD) then eventually supplies electricity to
tions), because divestment in coal is relatively strong. Similarly, but in the market. For the spatial resolution, WEGDYN differentiates between
an analytical model, Rozenberg et al. [42] also demonstrate that private sixteen regional aggregates.8 We focus on EU-28 member states (plus
costs of stranding are significant when choosing an optimal CO2 price as Norway, Liechtenstein and Iceland) represented by four regional ag-
policy instrument. gregates: Eastern Europe (EEU), Northern Europe (NEU), Southern
In the present contribution, we add to this literature but follow a Europe (SEU) and Western Europe (WEU). Additionally, we keep Aus-
different approach and demonstrate its relevance for the European tria (AUT) and Greece (GRC) as separate model regions. We do so,
electricity sector's low carbon transition. As compared to other state-of- because Austria and Greece serve as examples with very different initial
the-art analyses, we do not focus on an optimal (and often un- conditions for the transition and thus are expected to react differently.
realistically high and globally uniform) CO2 price as policy instrument, Compared to other regions, Austria is characterized by low interest
but rather assume a top-down second-best policy in the form of re- rates (on debt; reflecting low risks from regional characteristics), high
newable portfolio standards (RPS). This approach is closer to reality as return on equity and a high share of renewable electricity in its mix,
most countries across the world already have such standards in place however with only moderate capacity factors for wind and PV. With
[43] and RPS are increasingly gaining popularity [44]. In addition, RPS respect to these factors, Greece is very different, facing higher interest
can be introduced unilaterally and is thus compatible with the post- rates (on debt) than other regions, lower returns on equity and a rela-
Paris climate policy architecture, with Nationally Determined Con- tively high share of fossil electricity, but promising capacity factors for
tributions (NDCs) at its heart. Another difference to existing studies is wind and PV (many shorelines and high insolation). For more details on
that we refrain from assuming perfect foresight, an assumption which in the structure of WEGDYN, the associated closure rules and background
our view is problematic when it comes to uncertainty and risk analysis. assumptions we refer to Appendix A2, and refs. [45,50].
Instead, we assume myopic behavior, however with changing risk-
perceptions (via different interest rate assumptions – risk premiums
fractions – in the “present”). The innovation of the analysis presented 2.2. Electricity sector pathway implementation
here lies in the implementation of changes in risk perceptions, which
are assumed to be triggered by the introduction of such binding RPS, Regarding the development of the electricity sector, we follow a RPS
interpreted by investors as a strong policy signal. As pointed out by approach. As discussed in section 1, this second-best-approach fits well
Schmidt [25], a credible policy framework – and even more so the re- into the current post-Paris climate policy architecture. We do so by
sulting signaling – is pivotal to increase trust in renewables, thereby exogenously setting a region-specific electricity generation mix, speci-
decreasing their capital costs via lower risk-premiums. However, such fied for each 5-year step of the model, that meets the given electricity
policy comes at a cost, which in our case is the deviation from a demand.
baseline pathway to a (at least from a BU perspective) costlier but re- In WEGDYN the dynamics in supply of electricity is modelled as
newable energy pathway. follows. Each electricity generation technology is represented by a sub-
We address the stated shortcomings for the case of a transition sector of the electricity sector aggregate, or in other words by a tech-
scenario to an (almost) 100% renewable electricity system in Europe. nology's power plant stock (e.g. the PV power plant stock of a region).
Our results demonstrate the importance of WACC differentiation in Each power plant stock develops over time since there are power plant
macro-economic modelling per se and the importance of possible di- additions and shutdowns (given plant lifetime and the RPS over time).
verging risk perceptions. On a European scale, no such analysis is To account for changing investment costs over time, we differentiate by
currently available. We fill this gap and give a first quantification of annual vintages and only then aggregate to a technology power plant
ranges and magnitudes. stock. Particularly, we take care of the changing capital costs (CAPEX)
of each power plant, since additional investments for capacity upscaling
lead to additional annuities at the aggregate level. Since the production
2. Model description, calibration and scenario framework
of a kWh has different unit-costs across technologies, we combine
physical target quantities (kWh) and technology specific generation
2.1. General model description and calibration
costs (i.e. levelized costs of electricity, LCOE), to derive cost mark-up
factors to account for the differences in unit-costs (as for example done
For the economy-wide assessment of a large-scale renewable elec-
in [7]). For further details on the implementation the electricity sector
tricity expansion under different risk perceptions we use the WEGDYN
pathway see Appendices A3 and A4.
model [45,46]. WEGDYN is a global multi-region, multi-sector, re-
cursive-dynamic computable general equilibrium (CGE) model, which
solves in five-year steps, starting in 2011. From a macroeconomic
modelling perspective WEGDYN is supply-side constrained, meaning
that capacities (capital, labor6 and resource endowments) are fully 7
The benchmark monetary output levels of the electricity supply sector (in-
utilized, constraining macroeconomic expansion through scarcity cluding power generation and collection/distribution) in 2011 are based on
GTAPv9 data [47]. For its disaggregation by generation technology we use
physical generation from 2011 (Table A 1) and benchmark LCOE provided by
6
Note that we model “classical” unemployment via a minimum wage. Details [48,49].
8
on WEGDYN labor market modelling are given in [45]. Table A 6 gives details on country-wise aggregation.
3
Fig. 1. Benchmark electricity mix (2011) across EU regions and mixes for 2050 for the reference scenarios (EU-ref) and for the large-scale expansion of renewables
scenarios (RES-e).
2.3. Scenario framework crucial determinant of model results and recommendations for policy
designs. In general, the weighted average costs of capital is defined as
In general, we compare two simulation runs, or scenarios. First, for the sum of the cost of equity and the cost of debt [53]:
the baseline simulation run we choose the regional shares of renewable
electricity generation from the “EU Reference Scenario” (EU-ref, [51]). WACC = iRoE *E/(E+ D)+ iRoD *D/(E+ D) (1)
EU-ref depicts a moderate expansion of renewable electricity, which is
with iRoE being the return on equity, iRoD the return on debt, E the equity
imposed exogenously. Second, we carry out a simulation run with a
part of the investment (in absolute terms) and D the debt part of the
large-scale expansion of renewable electricity, which reaches an almost
investment (in absolute terms). It is thus the return rates themselves as
100% renewables share by 2050, called “Renewable Energy Sources for
well as the respective shares of debt and equity that determine WACC.
Electricity Scenario” (RES-e), based on generation shares from [17].
The extant literature usually uses commonly accepted WACC values
The RES-e scenario is imposed exogenously as a binding RPS, but as
uniformly across the board (usually in the range of 5%–10%; cf. Table A
compared to the EU-ref is much more ambitious in terms of renewables
1), i.e. without differentiation across technologies and regions. How-
shares. Note that even though this RPS materializes in the model,
ever, capital costs differ strongly across regions, depending on factors
agents are not perfectly informed in reality (approached here with
such as political stability or the business cycle. Also, it is rather plau-
myopic foresight). Hence, they might still have doubts about future
sible that technological characteristics are at the core of assessing the
climate policy stability and the materialization of the RPS. The results
expected return of an investment and – the other side of the coin –
of the two simulation runs (EU-ref and RES-e) are then compared to
evaluating the respective riskiness. Hence, the simplification of a uni-
each other to isolate the effect of enforcing the RPS. This comparison is
form WACC across regions and technologies strongly distorts relative
done several times, but each time we change the underlying WACC
costs and benefit evaluations of any policy.
parameters to increase accuracy and/or reflect different risk perception.
We also ascertain that in the literature there is a strong focus on the
In both simulation runs (EU-ref and RES-e) we use the same total
WACC of renewables for two reasons: First, because they are more ca-
electricity demand trajectories until 2050, based on [17,52], which is
pital intensive than fossil technologies, and second because they are
assumed to increase linearly, with the total electricity demand in
considered through a risk lens. This risk perspective led to a consensus
Europe being about 37% higher in 2050 (compared to 2011),
that renewable technologies are subject to higher WACC than fossils
amounting 4,448 TWh. Until 2050, for the whole EU-28plus3 policy
fired technologies, because they are considered riskier.9 This partial
region the share of renewable electricity generation (biomass, wind,
view is problematic, however, because projects in the fossil industry are
PV, hydro) increases from 22% (2011) to 57% in the EU-ref simulation
also risky undertakings [54], facing even some of the same risks, e.g.
and to 93% in the RES-e simulation (with gas accounting for the re-
from the business cycle or political stability. Moreover, it is important
maining share). Note that generation from solid fuels (coal), nuclear
to note that a high WACC does not necessarily reflect a high risk, but
and petroleum is fully phased out in the RES-e simulation. In Fig. 1 the
rather high profitability in general, i.e. high returns on investment [55],
regional electricity mixes are shown for the benchmark year (2011) as
which are higher in economically strong regions.
well as for 2050, respectively for the EU-ref and the RES-e scenarios. In
Only recently has academia started to call for more attention for
RES-e additional investments in the electricity system accrue from grid
WACC settings and its meaning. Schmidt [25] stresses the need for a
expansion and the implementation of batteries and power to gas facil-
global database on financing costs and Polzin et al. [55] argue for not
ities to the system (taken from [17]). See Appendix A5 for details on the
only looking at risks, but also at differences in returns. Egli et al. [27]
investment modelling.
collected project micro data from Germany and use both WACC de-
terminants (return rates and financing shares) to deduce WACC for PV
3. The weighted average cost of capital and scenario settings
and Wind. For 2017, they find WACC of 1.6% for solar PV and 1.9% for
3.1. WACC revisited

9
For instance, and as noted above, Pleßmann and Blechinger [17] assume
Energy policy evaluation studies show that the costs of capital are a WACC rates of 6% for conventional and 7% for renewable technologies.
4
Fig. 2. WACC rates across regions and technologies of the MAIN setting. (SF=Solid Fossil Fuels (coal); PE=Petrol (oil); GS = Gas; BM=Biomass, WI=Wind,
PV=Photovoltaics; HY=Hydropower; NU=Nuclear).
wind power plants (5.1% and 4.5% on average between 2000 and 2005, Figure A 5). This enables us to construct a technology- and region-
respectively). These are unprecedented low values, driven by high debt specific WACC.11
shares in combination with the current low return on debt rates; a
beneficial financing structure that is also highlighted in [25]. Steffen
[19] isolates the main drivers of this mechanism, revealing that mostly 3.2. WACC settings for scenario simulations
(groups of) individuals and (local) initiatives have used primarily debt
money for financing renewables projects. In order to reflect the recent empirical findings regarding different
Taking up these recent findings, we explore the relevance of using financing structures, returns as well as risk (perception) in the elec-
detailed costs of capital data, considering regional and technological tricity sector, we change/update the underlying WACC in different si-
peculiarities. Following [56], we calculate WACC rates according to Eq. mulation runs in a stepwise manner. More specifically, we carry out the
(1), which are replicable with publicly available data and are con- two simulation runs (EU-ref and RES-e) under different WACC as-
sistently determined. We use return on equity data from IMF [57], and sumptions, or settings, and investigate how these differences in WACC
return on debt data from World Bank [58] and ECB [59] (summarized assumptions alter the macroeconomic effects of taking the RES-e in-
in Figure A 4).10 The economic strength of regions like WEU and stead of the EU-ref pathway.
countries like Austria show up in relatively high returns on equity. In As a point of departure (and to connect to the state-of-the-art lit-
contrast, weaker economies, such as SEU and countries like GRC, face erature) we adopt a uniform WACC rate of 8% across regions and
lower returns on equity. Regarding debt, low values of return on debt technologies (setting UNI). In a second setting, which we will refer to as
reflect low lending risks. We observe that north-western regions and the “Main Setting” (MAIN), we deviate from this uniform WACC as-
Austria are confronted with lower rates as opposed to south-eastern sumption and apply the calculated technology- and region-specific
regions and Greece. Interestingly, the regional means across regions are WACC. The calculated WACC are shown in Fig. 2, with values for wind
11% for return on equity and 4% for return on debt, which correspond ranging between 4.3% and 8.5% (average: 5.6%) and for PV between
roughly to the upper and lower bound of the cost of capital range that is
used in the literature (cf. Table A 1). To account for technological 11
One may argue whether these observed type-of-finance ratios for German
differentiation, we make use of the findings of [19] of different finan-
projects are applicable for other regions and that this share could change over
cing shares (debt or equity) for different technologies (summarized in time (e.g. equity shares for renewables may increase if also large utility firms
strive towards decarbonizing their portfolios). However, we deem these as-
sumptions reasonable for at least two reasons. First, they are at most equally
10
To account for the currently exceptional situation in the aftermath of the incorrect as assuming uniform WACC rates. Quite the contrary, they highlight
2007/2008 financial and economic crises, we take for each measure a medium- the implications of assumptions that are inconsistent with observations. Second,
term median (2003-2017) of country-specific values instead of current values the main objective here is gaining insight and not uncontested numbers. We
and aggregate individual country values using respective gross domestic pro- investigate orders of magnitude effects and the sign of implications, not the
duct weights. exact value of costs and benefits related to the implementation of RPS.
5
Fig. 3. WACC rates across regions and technologies of the Uniform (UNI), Main (MAIN), Fossil Fuel Risk (FRR) and De-risking Renewables (DRR) settings. Whiskers
show the maximum of assumed WACC increase from climate policy instability risk. (UNI=Uniform WACC across all regions and technologies; SF=Solid Fossil Fuels
(coal); PE=Petrol (oil); GS = Gas; PV=Photovoltaics; WI=Wind).
Table 1 and thus WACC for renewables. We assume that WACC rates of re-
Overview of WACC settings. (* = additional settings for sensitivity analysis). newables can be reduced to the rate as observed in the recent years in
Settings Simulation run WACC setting
Germany [27] (1.6% and 1.9% for solar PV and wind, respectively).12
Note that the policy signal from the ambitious RPS only exists in the
UNI EU-ref Uniform RES-e scenario. We thus implement this assumption of lower WACC for
RES-e Uniform renewables only in the RES-e simulation run.
MAIN EU-ref Main Setting
RES-e Main Setting
In a fourth setting, “Fossil Fuel Risk” (FFR), we take into account
DRR EU-ref Main Setting that investors price in carbon-content-related risks for new investments,
RES-e De-Risking Renewables i.e. the risk of assets becoming stranded before their economic lifetime
FFR EU-ref Fossil Fuel Risk (technology driven: 1/2 of lifetime) ends (with the exact point in time being uncertain, though).13 In the
RES-e Fossil Fuel Risk (technology and policy stability
EU-ref simulation run, as a first approximation, we assume a halving of
driven: 1/8 of lifetime)
COMB EU-ref Fossil Fuel Risk (technology driven: 1/2 of lifetime) the expected lifetime for gas, coal and petroleum fired technologies
RES-e De-Risking Renewables and Fossil Fuel Risk (similar as done by [41]), which is assumed to be driven by technolo-
(combined technology and policy stability driven: gical change alone. In the RES-e simulation run, we account for an even
1/8 of lifetime) higher risk and assume a lifetime reduction for fossil fired technologies
FFR_med* EU-ref Fossil Fuel Risk (1/2 of lifetime)
RES-e Fossil Fuel Risk (1/4 of lifetime)
to 1/8 of the respective economic lifetime. This is driven by technolo-
FFR_low* EU-ref Fossil Fuel Risk (1/2 of lifetime) gical change and in addition by the risk from climate policy instability,
RES-e Fossil Fuel Risk (1/2 of lifetime) since a credible RPS might come along with the investor's suspicion that
policy could even speed up the transition, relative to what is signaled by
the RPS (“ratcheting-up”).
2.9% and 7.7% (average: 4.4%). The average across all technologies
and regions is 8%, which lies in the middle of the range of usually used
uniform WACC assumptions. Fig. 2 also clearly indicates the different 12
Compared to MAIN. The WACC of the other technologies remain as in the
financing structures of fossil and renewables. MAIN setting.
In a third setting, called “De-risking Renewables” (DRR) we take 13
The reader can find an explanation of our approach in Appendix A6. Note
into account that the perceived policy signal (i.e. an ambitious RPS), that we keep WACC of PV, wind power, hydropower, biomass and nuclear
elevates investors’ trust in renewables, which reduces risk premiums power as in the main setting.
6
Fig. 4. Change in bottom-up derived unit-costs (RES-e versus EU-ref).
A last setting combines the isolated effects of DRR and FFR, denoted the results qualitatively, with higher unit costs in the first years of the
“Combination” (COMB). The five described settings are summarized in transition. Hence, the UNI setting substantially overestimates the direct
Fig. 3 and Table 1. All of the resulting WACC rates are shown in Table A (and consequentially also indirect) costs of the transition.15 Using the
7. MAIN WACC setting, generation costs of the RES-e pathway are far
below the EU-ref pathway, reaching cost reductions between −15%
(AUT) and −30% (in EEU) by 2050. Second, we find strong cost-re-
4. Results
ducing effects from the assumed de-risking of renewables (DRR). A
credibly designed RPS may increase trust in renewables, driving down
4.1. Direct economic effects
risk perception and risk premiums for renewables. Due to the capital-
intensive nature of renewables, this lever turns out to be very effective.
Fig. 4 shows the change in average unit costs of electricity (EUR/
The green area in Fig. 4 shows the possible benefit of such a de-risking
kWh), when comparing the RES-e to the EU-ref simulation under the
(relative to the MAIN setting without de-risking). Thereby cost reduc-
different WACC settings as presented in Table 1.14 We observe one
tion can reach magnitudes between −20% (in AUT) and −45% (in
common pattern across all model regions: In the long run the average
EEU) below the EU-ref. This rather low WACC level under the DRR
unit costs are lower in the RES-e simulation run compared to the EU-ref,
scenario setting should be regarded as lower bound, though.16 Third,
irrespective of the chosen WACC setting. This reflects the recently ob-
we find interesting effects from including carbon-content related fossil
served, and further expected, decline in investment costs of renewables,
fuel risks (FFR). In the first phase of the transition (RES-e), additional
particularly of PV and wind power. However, there are important dif-
gas capacities are required to bridge the phase out of coal and oil-fired
ferences in the short run and also between the different WACC settings.
First, we find that the uniform WACC assumption (UNI) leads to
substantial discrepancies when compared to the more differentiated
MAIN scenario setting (discrepancies shown as yellow shaded area). For 15
The only exception is Greece, where we find a slight underestimation of
some regions (Austria, EEU and SEU), this discrepancy even changes direct costs, however less pertinent due to similar levels of return on debt and
return on equity.
16
Note that the WACC development in Germany is not directly related to risk-
14
Note that this BU measure is net of any fiscal effects (e.g. a CO2 price) and reduction, but we choose these values as a possible lower bound of how low
also does not acknowledge any macroeconomic feedbacks, yet. WACC can get in reality.
7
Fig. 5. Change in electricity supply price (RES-e versus EU-ref).
capacities. Factoring in fossil fuel related risk17 implies higher financing CO2 pricing, investment costs for storage, macroeconomic and inter-
costs of these additional gas-fired capacities (as compared to the set- sectoral feedbacks as well as cumulative capital stock effects over time.
tings where this risk is not included) and thus of overall unit costs of
electricity supply. In isolation, in some regions (Austria, Greece, EEU
and SEU) this effect leads to unit-costs of the total supply of electricity 4.2.1. Sector level
that are even higher than in EU-ref. However, in the long term they The large-scale renewable electricity expansion (RES-e) in Europe
converge back to the low values as under the MAIN setting. Finally, leads to changes in the market price of electricity relative to the EU-ref
combining the isolated effects of DRR and FFR (COMB) reveals further baseline, as shown in Fig. 5. Generally, the pattern of unit cost effects
interesting dynamics. Although long-term unit-costs are substantially (Fig. 4) recurs in price changes, however with steeper reductions in the
below the EU-ref case, (here driven by the expected decline in invest- second half of the modeling period due to an increasing CO2 price that
ment costs of renewables as shown with UNI and MAIN) and policy supplements the transition. This effect is particularly strong for Greece,
could lower the long-term unit-costs even further, policy makers should where in the EU-ref pathway coal fired power plants are phased out
be aware of fossil-fuel-related risk – which they can partly influence via only by 2050, whereas in the RES-e pathway coal is phased out by
stability signaling – on bridging technologies like gas-fired power 2035. This leads to a strong price advantage from 2035 onwards in the
plants. (The magnitude of the FFR is however subject to large un- RES-e case. For Austria we see that the differentiation of WACC is less
certainty and we thus carry out further sensitivity analysis on this in- important, since Austria's electricity mix already has a relatively high
fluential parameter in section 4.3). share of renewables in the benchmark year (2011) as well as in the EU-
ref scenario. This means that a de-risking of renewables has a weaker
effect (lower WACC only apply for additional capacities), while si-
4.2. Economy-wide feedback effects multaneously the country is less exposed to fossil fuel related risks. In
general, we find that even without de-risking (i.e. MAIN) immediate
In the following sections we present the economy-wide effects from price declines emerge in all regions. Only under a setting where there
integrating the direct effects into WEGDYN. We are now acknowledging would be a fossil fuel risk (FFR), but no de-risking of renewables at all,
electricity supply prices would be slightly higher in the first years of the
17
Driven by both technological change and possible climate policy instability transition (in Austria, Greece, EEU and SEU), but then converge to-
(i.e. “ratcheting up”). wards the MAIN level in the long-run.
8
Fig. 6. Change in gross domestic product (RES-e versus EU-ref).
4.2.2. Macroeconomic level build up the capital stock, which becomes highly effective and gen-
Fig. 6 shows regional GDP effects. Under the MAIN WACC setting, erates capital income in the following years (which is also reflected in
long-term GDP implications are positive in all regions compared to EU- the steep increase in GDP in 2050 in Greece).
ref. This is due to the previously shown lower generation costs and
market prices for electricity that lead to higher economy-wide pro-
ductivity and thus GDP. In 2050 GDP is higher by +1.5% (Austria) to 4.2.3. Labor market effects
+3% (WEU) under the MAIN settings. Again, we find that a uniform Since renewables are much more capital intensive than fossil fuel-
WACC assumption would substantially bias the effects to the negative based technologies, the regional electricity mixes in the RES-e simula-
(yellow areas), potentially blocking the initiation of the transition due tions are characterized by higher capital intensities and capital rents
to the fear of lower productivity and GDP growth rates. When assuming (Figure A 11). Thus, less labor is required which temporarily leads to
de-risking of renewables (DRR) we find that the positive GDP effects higher unemployment rates18 (see Figures A 8 and A 9). However, our
from MAIN become even stronger (green areas show the benefits of de- results show that in the long-term, positive effects on employment le-
risking), especially for regions in which large fossil fractions of the vels emerge in all regions, as the higher economy-wide productivity
electricity mix are replaced by renewables (EEU and Greece). When leads also to higher employment in non-electricity sectors. These results
also including a fossil fuel risk premium, these benefits of the transition hold for all WACC settings, however, we find that an additional fossil
are getting smaller, though (COMB). This is because of the “bridging- fuel risk premium weakens this positive employment effect. Also, note
effect” as explained. that the uniform WACC assumption would again bias the results to-
Contrary to GDP, for welfare the RES-e transition induces some wards negative (or less beneficial) macroeconomic, in this case em-
unfavorable short-term effects as soon as large investments for storage ployment, effects.
become necessary (see Figure A 7, showing consumption possibilities of
the regional household). This effect is particularly strong for Greece,
where additional necessary investments are relatively large (see Figure
A 10), requiring savings to be sharply increased, crowding out con-
sumption accordingly. This leads to a negative welfare effect (−2% in 18
Provided that economies are characterized by full capacity utilization, i.e.
2046 in Greece), however only temporarily, since these investments there is no output gap.
9
Fig. 7. Sensitivity analysis on changes in gross domestic product (RES-e versus EU-ref).
4.3. Sensitivity analysis sectors, we carry out further simulation runs where we – additionally to
the standard FFR scenario setting – increase the global fossil fuel prices
In the previous sections we have shown that the “best-case” setting such that they are increased by 1.5% (FFR_FFP1.5%) and 3%
(in terms of lowest transition costs) would be the DRR setting, i.e. a (FFR_FFP3%) in 2050, as compared to the default model assumption.
relatively strong de-risking of renewables but without any considera- This increase is however only implemented in the RES-e simulation run,
tion of fossil fuel related risk. This DRR should be thus regarded as an as the strong policy signal and respective fossil fuel risk only occur in
upper bound for positive effects of de-risking. The setting that leads to this case. This would lead to relatively strong negative effects from
the “worst-case” is when we only consider fossil fuel related risks, i.e. following the RES-e pathway, as the whole economy would suffer from
the FFR setting, which is associated with a fossil fuel risk premium that higher fossil fuel prices, particularly the gas-fired capacities which are
we approximated by a reduction to 1/8th of economic lifetime. This needed for the bridging-phase of the transition. However, this effect
assumption is, however, highly uncertain and might be overestimating should be interpreted with care, as we do not model any further dec-
the additional risk premium from climate policy instability. Thus, we arbonization options in the non-electricity energy sectors (e.g. mobility,
carry out sensitivity analyses in the FFR scenario setting, where we heating). This means that the economy cannot easily switch to cleaner
reduce lifetime to only 1/4 (i.e FFR_med) and 1/2 (i.e. FFR_low) in the technologies as they simply do not exist in the modelled world.
RES-e simulation run (instead of 1/8, cf. Table 1).19 The red shaded
corridors in Fig. 7 show the range resulting from changing the lifetime 5. Discussion & limitations
reduction-assumptions. We find that when lowering the fossil fuel risk
(i.e. going from FFR to FFR_med and FFR_low), GDP effects are shifted In this article we shed light on the significant bias from uniform
upwards, however only weakly, when compared to the effect of DRR. WACC assumptions, which also has far-reaching implications for
Since climate policy in the electricity sector might also lead to economy-wide assessments. In state-of-the-art climate change mitiga-
changing risk perceptions and risk premiums in the fossil fuel extraction tion assessments, be it via bottom-up or in integrated modelling ex-
ercises, WACC are typically not differentiated across technologies and
regions but set to “standard” values. For the case of Europe's electricity
19
Note that FFR_low means that the fossil fuel risk premium only includes sector, we show that this assumption systematically overestimates the
technological change-driven risk, but no additional risk from climate policy direct costs of a low carbon transition and that also important policy-
instability. relevant macroeconomic indicators (e.g. GDP effects) follow this trend.
10
In some cases, this bias is so strong that anticipated costs even change to 6. Conclusions
benefits when using a more differentiated approach for setting WACC
parameters. The main reason for this bias is the difference in financing Our findings indicate that immediate positive effects emerge at
structures between conventional (fossil) energy and renewable energy macroeconomic scales when using more accurate data on capital costs
technologies. Whereas financing of the former more strongly depends and more up-to-date data on investment costs of renewable electricity
on equity, the latter depends much more on debt. This bias might lead technologies. Consequently, weighted average costs of capital (WACC)
to too cautious climate policy measures, and may even postpone or assumed for both fossil-based and renewable technologies, need to be
prevent measures. Further, we demonstrate in this article that there is a handled with care, communicated transparently and differentiated.
huge potential for economy-wide benefits, employment and stronger Uniform WACC assumptions (across technologies and regions) imply a
economic growth from de-risking renewable electricity technologies. significant bias in results, which by now can be avoided relatively ea-
This effect is significant for all modelled European regions, but parti- sily. A more accurate modelling framework adds important robustness
cularly strong for those regions that currently face large fractions of to techno-economic modelling and possibly even a more ambitious
fossil fuels in their electricity mix. Finally, we reveal interesting dy- climate policy.
namics coming from the need of bridging the phase-out of coal and oil- De-risking renewables further improves the effects of renewable
fired electricity generation by using additional gas-fired capacities. As electricity transition across all regions, particularly in eastern and
there are many reasons for expecting increasing risk premiums for fossil southern Europe, where electricity is produced relatively CO2-intensive
fuel investments (coming from both technological change but also cli- in the reference scenario. The positive effects do not involve any need
mate policy instability) these additional capacities might be confronted for technology costs to become cheaper, but the credible long-term
with higher capital costs, which in turn might reduce the benefits of the framework conditions turn out to be most important. Setting up such
low carbon transition. In total, however, we show that the potential conditions does not necessarily involve large direct costs and can also
benefits of an extreme de-risking scenario by far outweigh the potential be implemented unilaterally.
negative effects from an extreme fossil fuel risk scenario. We also conclude that the bridging-phase needs special attention. If
As a limitation, we have to note here, that the equity shares in the gas capacities are used in the beginning of the transition for phasing-out
financing structure of renewables might increase in the future, if large coal and oil, conflicting goals may emerge, as policy makers are on the
energy suppliers increase their portfolio shares of renewable energy. In one hand interested in fostering the transition, but by sending signals to
a more decentralized energy system this might not be the case, though. eliminate fossil fuel use the transition actually might get more ex-
Further limitations of the analysis relate to both the level as well as pensive. However, we show that the potential benefits of an extreme de-
comprehensiveness of carbon-content-related risk (and corresponding risking renewables setting outweigh the potential negative effects of an
risk premiums). Regarding the level, we choose a simple approach for a extreme fossil fuel risk setting.
first approximation of changing capital costs, by reducing the economic By considering the fact that transitions towards renewables are
lifetime in the calculation for annuities. To that end more sophisticated particularly capital-intensive, the analysis exceedingly emphasizes that
methods and research are needed. Endogenizing risk-perception based investors’ risk perceptions of renewables are of higher relevance than
on market shares could be one way to approach this. Regarding com- those of fossils. Thus, the role of expectations from investors and as-
prehensiveness, we note that in the analysis presented we focus on sociated financial policies and macroprudential financial regulations
power plant investment risk but not on possible investment risk for should be given a more prominent role, going beyond carbon pricing, as
extraction facilities (e.g. offshore oil or gas platforms). However, first discussed by [61]. De-risking renewables provides decision makers with
sensitivity analyses indicate that also acknowledging the existence of an additional game-changing option for staying within long-term cli-
risks in the fossil fuel extraction sector might have additional strong mate policy objectives. The results clearly show the value of more re-
effects. liable financing cost parameters: they are of utmost epistemic im-
A further limitation is the uncertainty regarding investments re- portance.
quirements and the here-imposed assumption of full capacity utilization
[60]. These issues should be explored in more depth in future assess- Acknowledgements
ments. Welfare measures, in terms of consumption possibilities, are
especially sensitive to this assumption, as more investments at the cost We thank Guido Pleßmann for providing data and Keith Williges for
of higher savings (and thus less consumption) might lead to less ben- proofreading. This research was supported by the EU Research and
eficial (or negative) welfare effects. Innovation Framework Program Horizon 2020 (research grant numbers
642260 (“TRANSrisk”) and 776479 (“COACCH”)).
Appendix. A1: Literature review on the current handling of interest rates in energy modelling
A2: Underlying model and shared socio-economic pathway assumptions
We use SSP2 growth rates for GDP and population to calibrate the model. SSP2 represents a middle-of-the road scenario [78]. It is implemented
through exogenous parameters for regional effective labor force growth (capturing population growth), multi-factor productivity growth, autono-
mous energy efficiency improvements, and capital depreciation rates, such that region-specific GDP growth rates are met. Additionally, a moderate
CO2 tax is included, starting with 5 EUR/tCO2 in 2011 and reaching 46 EUR/tCO2 in 2050. This reflects the IEA [79] “New policies scenario” (given
for EU, but implemented globally to mimic nationally determined contributions as a weak price signal).
A3: Electricity generation technology data
In 2011, the installed capacity for each electricity generation technology is shown in Table A 4. Assumptions on technology-specific investment
costs and economic lifetime are given in Table A 5. For the former, we apply the values from [17] for the benchmark year 2011. Technological
progress in PV and wind power technologies is highly dynamic. We therefore use detailed observations for the benchmark year 2011 and the latest
available information at times of modelling: 1,000 EUR/kW in 2014 for a representative PV system (i.e. module, inverter and components for system
balancing) given in [80] (Fig. 40) and [81] (Fig. ES-1) and 2,050 EUR/kW in 2017 for a representative wind power plant (assuming 75% onshore)
given in [82] (Fig. 5.3 and 5.10). Expected investment costs in 2050 for PV and wind power are taken from [80,83]. Note that we linearly interpolate
11
investment costs until 2050 for all technologies. Economic lifetime assumptions are based on [83].
The reference LCOE for each region is calculated taking the generation-weighted average of technology-specific LCOE in the benchmark (based
on data from [84], cf. Table A 6). This results in reference LCOE of 0.06 EUR/kWh for AUT, of 0.08 EUR/kWh for GRC, NEU and SEU, of 0.09 EUR/
kWh for WEU and of 0.10 EUR/kWh for EEU.
A4: Linking electricity sector investment-cycle model with the top-down CGE model
Based on the benchmark electricity generation quantities in 2011 [kWh] (Table A 4) and target shares in 2050 by technology (Fig. 1), we
calculate technology-specific demand in a spreadsheet module as follows. For PV and wind, we take the benchmark and target quantities and assume
logistic break-through. For all remaining technologies except gas, we linearly interpolate between 2011 and 2050. The resulting residual load is filled
with generation from gas power plants based on the assumption that it represents the most flexible technology in terms of “dispatchability.”
The resulting technology-specific "demand" trajectories (measured in [kWh]), which represent the exogenously imposed RPS, are then compared
to potential output (based on existing capacities) determining whether additional capacities have to be built up. If so, the required investments are
converted to vintage-differentiated annuities, ultimately increasing capital expenditures (CAPEX) in the respective periods of repayment (which is
assumed to coincide with the economic lifetime of the technology). The derivation of annuity payments A follows the common specification
(1 + i )t i
A=S ,
(1 + i )t 1 (A1)
with S being the loan amount, i being the interest rate and t being the financing term.
Based on the specific capacity development of each technology also the development of the operating expenditures (OPEX) can be evaluated
applying unit-OPEX of the benchmark to actual generation throughout 2050.
The bottom-up calculated development of OPEX and CAPEX (which differ for the EU-ref and the RES-e scenario since target shares in 2050 are
different) flows into the macroeconomic top-down assessment in the following way. First we rescale the benchmark inputs of each technology-
specific production function (measured in [EUR]) for a given year in order to reflect the changing input structure over time. For this purpose we use
each input i to electricity generation ELY(i) classified as OPEX (i.e. excluding capital input) to calculate the benchmark share of operating ex-
penditures ox_sh_bench(i) (measured in [%]) in order to update the over time changing OPEX (measured in [EUR]) 20:
ELY (i )
ox _sh _bench (i ) = .
i
ELY (i ) (A2)
Accordingly, vector A(i,t) consists of all updated OPEX multiplying the benchmark share of operating expenditures with the actual OPEX in
period t, and we extend this vector with an entry for technology-specific CAPEX in the respective period which gives us vector A*(i*, t) with i* being
all inputs to electricity generation including capital. Vector A*(i*, t) is then used to calculate a dimensionless factor Q which is weighted with the
inverted input share of the benchmark and subsequently multiplied with the benchmark inputs of the production function for each electricity
generation in the macroeconomic model (measured in [EUR]):
A (i , t ) = ox _sh _bench (i ) OPEX (i , t ), (A3)
A (i , t ) i
ELY (i )
Q (i , t ) =
i
A (i , t ) ELY (i ) (A4)
Note that the updated cost structures for each electricity generation technology still represent unit-cost functions. Those are combined to give an
electricity mix. To account for the changing technology-specific unit-costs we use a cost mark-up cmr which gives the ratio of the technology-specific
updated unit-costs UC and the reference levelized costs of electricity21 LCOEref for each technology tec in period t:22
UC (t , tec )
cmr (t , tec ) = .
LCOEref (A5)
This cost mark-up ratio is then weighted with the inverted technology-specific benchmark generation BG(tec) (measured in [EUR]) and the
technology-specific physical generation target PG(tec) of each period t (measured in [kWh]) representing a dimensionless cost mark-up factor cmf
which is used for scaling the unit costs in the electricity mix:
tec
BG (tec ) PG (t , tec )
cmf (t , tec ) = cmr (t , tec ).
BG (tec ) PG (t , tec )
tec (A6)
This approach warrants the vintage-based integration of the OPEX and CAPEX development in the dynamic-recursive CGE model.
A5: Handling of investments for grids and storage infrastructure
Total system investment costs are composed of three parts: 1) power plants, 2) grids and 3) batteries and P2G facilities. Regarding 1) and 2) we
assume crowding out of other generic investments, meaning that total investment volume does not change between EU-ref and RES-e. On the
contrary, for 3) we account for additional investments for batteries and P2G facilities (only necessary in RES-e), which are modelled explicitly as
additional investments, financed via higher savings and thus a reduction in consumption. The corresponding annuities materialize as additional
annual capital costs of the electricity sector. Figure A 3 shows these additional investments costs by region for the modelled annual time slices
(before 2036 no storage investments are necessary).
20
For convenience, we drop indices for model regions and electricity generation technologies.
21
The region-specific reference unit-costs represent the average LCOE of each technology weighted with the benchmark share of generation.
22
Note that we here drop only region indices.
12
A6: Additional material and explanations for WACC calculation
We use a typical specification of the annuity factor AF of an investment project as the point of departure for assessing the relationship between a
project's lifetime t and its WACC i (Eq. (A1)). The initial settings are i = 10% and t = 40 years. In a first step we reduce the underlying economic lifetime
(e.g. halving it to 20 years) to calculate a new annuity factor. In the second step we then set the lifetime back to its original value (i.e. 40 years) and solve
for the interest rate that reaches the same new annuity factor as determined in the first step. As the AF formula is not invertible, we solve numerically.
For very large reductions in the economic lifetime of a project, the WACC can get prohibitively high. If the project can exploit only 18% of its
usual lifetime (ca. 7 years instead of 40) the WACC doubles for the given numerical example. Figure A 6 shows the relationship between the lifetime
reduction and the corresponding WACC, according to the described method.
A7: Additional results
Fig. A1. Regional economic growth rate, effective labor force growth rate and multi-factor productivity growth per anno for SSP2.1
Fig. A2. Assumptions on annual autonomous energy efficiency improvement per policy region.2
Fig. A3. Annual storage investment costs in billion EUR by region for the modelled time slices (based on [17]).3
13
Fig. A4. Regional return on equity and return on debt rates, based on long-term country data on equity (IMF, 2019) and debt (ECB, 2019; World Bank, 2019) for non-
financial corporations.4
Fig. A5. Type of financing by electricity generation technologies based on Steffen (2018, Fig. 4). We assume an offshore share of 25% for wind power (WI) across
Europe. *Author's assumption for other (conventional) technologies, which are not depicted in Steffen (2018).5
Fig. A6. Relationship between WACC and lifetime of a representative investment project. The marker indicates the initial setting.6
14
Fig. A7. Change in welfare (RES-e versus EU-ref).
15
Fig. A8. Change in unemployment rate of skilled labor (RES-e versus EU-ref).
16
Fig. A9. Change in unemployment rate of unskilled labor (RES-e versus EU-ref).
17
Fig. A10. Change in total investment (RES-e versus EU-ref).
18
Fig. A11. Change in capital rents (RES-e versus EU-ref).
Table A1
Interest rate assumptions of various models used for policy evaluation in the energy-economy-environment context. BU = bottom-up; TD = top-down; (N)LP=(Non)
Linear Programming; PE=Partial Equilibrium; CGE=Computable General Equilibrium; n.s. = not specified; *Evolutionary bottom-up simulation of technology
diffusion & top-down macro-econometrics. #Technology-specific hurdle rates.
Model class Model acronym Method Interest rate assumption Sensitivity analysis References
BU elesplan-m 6%-7% No [17]

BU Calliope LP 10% yes [16]
BU DIETER LP n.s. no [62]
BU EMMA LP 7% no [63]
BU ETSAP/TIAM LP 5% yes [64]
BU ETSAP/TIAM & TIMES-Norway LP 5%# yes [18]
BU REMIND/TIAM NLP 5% yes [65]
BU IMAGE PE 10% no [66,67]
Hybrid FTT-E3ME * 10% yes [30]
Hybrid MESSAGE-MACRO LP/CGE n.s. yes [8,68]
Hybrid PRIMES/GEM-E3 PE/CGE 8% no [69,70]
TD AIM/CGE CGE n.s. no [71]
TD - CGE n.s. no [72]
TD EPPA CGE n.s. no [73–75]
TD GEM-E3 CGE n.s. no [76,77]
19
Table A 2
Aggregate sectors and regions of the WEGDYN CGE model.
Model sector code Aggregated Sectors Model country code Region Name
AGRI Agriculture AUT Austria

COA Coal GRC Greece
CRP Chemical, rubber, plastic products EEU# Eastern Europe
ELY* Electricity NEU# Northern Europe
EXT Extraction SEU# Southern Europe
FTI Food and textile industries WEU# Western Europe
GAS Gas AFR Africa
I_S Iron & Steel: basic production and casting CAN Canada
NMM Mineral products CHN China
OIL Oil ECO Emerging economies
P_C Petroleum, coke products IND India
PPP Paper, pulp and paper products LAM Latin America
SERV Other services and utilities OIGA Oil and gas exporting countries
TEC Tech industries RASI Rest of South & East Asia
TRN Transport REU Rest of Europe
CGDS Capital goods USA United States
Notes: *Represented by eight generation technologies (Solid fuels, Petroleum, Gas, Nuclear, Hydropower, Biomass, PV, Wind) and one subsector for Collection and
Distribution. #EEU includes: Bulgaria, Czech Republic, Hungary, Poland, Romania, Slovakia, and Slovenia. NEU includes: Estonia, Lithuania, Latvia, Denmark,
Finland, United Kingdom, Ireland, Norway, and Sweden. SEU includes: Croatia, Cyprus, Spain, Italy, Malta, and Portugal. WEU includes: Belgium, Germany, France,
Liechtenstein, Iceland, Luxembourg, and the Netherlands.
Table A3
Regional aggregates of the WEGDYN model.
Model Aggregate name Aggregated countries

code
AUT Austria Austria

GRC Greece Greece
EEU Eastern Europe Bulgaria, Czech Republic, Hungary, Poland, Romania, Slovakia, Slovenia
NEU Northern Europe Estonia, Lithuania, Latvia, Denmark, Finland, United Kingdom, Ireland, Norway, Sweden
SEU Southern Europe Croatia, Cyprus, Spain, Italy, Malta, Portugal
WEU Western Europe Belgium, Germany, France, Liechtenstein, Iceland, Luxembourg, Netherlands
CHN China China
IND India India
CAN Canada Canada
USA USA USA
REU Rest of Europe Albania, Switzerland, Bosnia-Herzegovina, Makedonia, Serbia, Moldavia
ROI Rest of industrialised Australia, New Zealand, Japan
countries
ECO Emerging economies South Africa, Hong Kong, Armenia, Azerbaijan, Georgia, Kazakhstan, Kyrgyzstan, Russian Federation, Tajikistan, Turkmenistan, Ukraine,
Uzbekistan, Brazil, Mexico, Indonesia, Republic of Korea, Pakistan, Belgium, Turkey
LAM Latin America Argentina, Belize, Bolivia, Chile, Costa Rica, Dominican Republic, Guatemala, Honduras, Jamaica, Nicaragua, Panama, Peru, Paraguay, El
Salvador, Trinidad and Tobago, Uruguay, Puerto Rico, Bahamas, Barbados, Cuba, Guyana, Haiti, Suriname
OIGA Oil and gas exporting Angola, Democratic Republic of the Congo, Nigeria, Ecuador, Venezuela, United Arab Emirates, Bahrain, Algeria, Egypt, Iran, Iraq, Israel,
countries Jordan, Kuwait, Lebanon, Libya, Morocco, Oman, Occupied Palestinian Territory, Qatar, Saudi Arabia, Syrian Arab Republic, Tunisia, Yemen
RASI Rest of South & South Cambodia, People's Democratic Republic Lao, Macao Special Administrative Region China, Vietnam, Brunei Darussalam, Malaysia,
East Asia Philippines, Singapore, Thailand, Bangladesh, Sri Lanka, Nepal, Fiji, New Caledonia, Papua New Guinea, French Polynesia, Solomon Islands,
Vanuatu, Samoa, Afghanistan, Bhutan, Maldives, Myanmar, Timor-Leste
AFR Africa Benin, Benin, Burkina Faso, Botswana, Côte d'Ivoire, Cameroon, Ethiopia, Ghana, Guinea, Kenya, Madagascar, Mozambique, Mauritius,
Malawi, Namibia, Rwanda, Senegal, Togo, United Republic of Tanzania, Uganda, Zambia, Zimbabwe, Mongolia, Burundi, Central African
Republic, Congo, Comoros, Cape Verde, Djibouti, Eritrea, Gabon, Gambia, Guinea-Bissau, Equatorial Guinea, Liberia, Lesotho, Mali,
Mauritania, Niger, Sierra Leone, Somalia, Swaziland, Chad
Table A4
Installed base year capacity and benchmark generation 2011 based on [48].
Solid Fuels Petroleum Gas Nuclear Hydro Biomass PV Wind
AUT GW 1,262 1,105 4,425 - 12,795 648 298 1,267

GWh 5,431 1,014 14,343 - 37,782 4,517 174 1,936
GRC GW 4,235 2,499 4,363 - 3,250 55 683 1,469
GWh 31,063 5,915 13,938 - 4,275 207 610 3,315
EEU GW 53,300 2,548 13,488 11,776 13,455 1,133 2,640 3,874
GWh 254,540 4,204 37,130 93,655 32,136 13,492 2,747 6,482
(continued on next page)
20
Table A4 (continued)
Solid Fuels Petroleum Gas Nuclear Hydro Biomass PV Wind
NEU GW 37,000 15,672 46,669 22,155 23,353 10,915 2,241 16,241

GWh 156,059 5,140 187,271 152,642 92,283 41,504 276 37,279
SEU GW 21,925 28,403 87,690 7,756 41,597 4,641 11,676 31,884
GWh 101,136 38,013 254,152 57,718 97,945 18,422 20,476 62,136
WEU GW 61,678 15,825 59,679 88,558 29,383 6,359 25,769 42,143
GWh 302,436 12,282 228,677 602,729 75,986 54,456 22,976 68,411
Table A5
OPEX, investment costs (2011 and 2050) and economic lifetime of electricity generation technologies [80,83].
Investment costs [EUR/kW] Economic lifetime [years] Operating expenditures [EUR/kWh]
2011 2050 AUT GRC EEU NEU SEU WEU
Solid Fuels 1,523 1,523 40 118 78 95 79 73 94

Petroleum 400 400 30 309 168 329 397 312 197
Gas 653 653 30 87 95 126 75 86 82
Nuclear 6,528 6,528 40 - - 89 84 81 102
Hydro 3,263 3,263 100 2 3 3 2 2 3
Biomass 2,485 1,951 30 26 11 82 38 16 28
PV 3,800 445 25 4 2 3 4 2 4
Wind 2,563 1,330 25 35 24 29 36 29 38
Table A6
Technology-specific LCOE 2011 in EUR/kWh [48,84].
LCOE 2011 [EUR/kWh] AUT GRC NEU WEU EEU SEU
Solid Fuels 0.11 0.07 0.08 0.08 0.10 0.06

Petroleum 0.27 0.15 0.36 0.17 0.30 0.29
Gas 0.08 0.09 0.07 0.07 0.12 0.08
Nuclear - - 0.10 0.10 0.10 0.09
Hydro 0.03 0.04 0.04 0.05 0.04 0.03
Wind 0.09 0.09 0.10 0.09 0.09 0.09
Biomass 0.10 0.04 0.07 0.09 0.08 0.03
Solar 0.13 0.09 0.15 0.12 0.13 0.09
Table A7
Detailed overview of WACC settings
Solid fuels Petroleum Gas Nuclear Hydro Biomass PV Wind
Uniform AUT 8.0% 8.0% 8.0% 8.0% 8.0% 8.0% 8.0% 8.0%
GRC 8.0% 8.0% 8.0% 8.0% 8.0% 8.0% 8.0% 8.0%
EEU 8.0% 8.0% 8.0% 8.0% 8.0% 8.0% 8.0% 8.0%
NEU 8.0% 8.0% 8.0% 8.0% 8.0% 8.0% 8.0% 8.0%
SEU 8.0% 8.0% 8.0% 8.0% 8.0% 8.0% 8.0% 8.0%
WEU 8.0% 8.0% 8.0% 8.0% 8.0% 8.0% 8.0% 8.0%
MAIN AUT 12.9% 13.0% 13.5% “n.a.” 13.0% 8.3% 2.9% 5.0%
GRC 4.5% 4.5% 4.5% “n.a.” 4.5% 4.9% 5.3% 5.2%
EEU 11.3% 11.4% 11.5% 11.4% 11.4% 9.7% 7.7% 8.5%
NEU 9.8% 9.9% 10.2% 9.9% 9.9% 6.6% 2.9% 4.3%
SEU 8.6% 8.6% 8.9% 8.6% 8.6% 6.5% 4.1% 5.0%
WEU 14.7% 14.8% 15.4% 14.8% 14.8% 9.6% 3.5% 5.8%
DRR AUT 12.9% 13.0% 13.5% “n.a.” 13.0% 8.3% 1.3% 1.6%
GRC 4.5% 4.5% 4.5% “n.a.” 4.5% 4.9% 2.4% 1.7%
EEU 11.3% 11.4% 11.5% 11.4% 11.4% 9.7% 3.5% 2.8%
NEU 9.8% 9.9% 10.2% 9.9% 9.9% 6.6% 1.3% 1.4%
SEU 8.6% 8.6% 8.9% 8.6% 8.6% 6.5% 1.9% 1.6%
WEU 14.7% 14.8% 15.4% 14.8% 14.8% 9.6% 1.6% 1.9%
FFR AUT 28.3% 35.3% 35.7% “n.a.” 13.0% 8.3% 2.9% 5.0%
GRC 22.8% 29.6% 29.6% “n.a.” 4.5% 4.9% 5.3% 5.2%
EEU 27.3% 34.2% 34.3% 11.4% 11.4% 9.7% 7.7% 8.5%
NEU 26.2% 33.2% 33.4% 9.9% 9.9% 6.6% 2.9% 4.3%
SEU 25.4% 32.3% 32.5% 8.6% 8.6% 6.5% 4.1% 5.0%
WEU 29.6% 36.7% 37.0% 14.8% 14.8% 9.6% 3.5% 5.8%
(continued on next page)
21
Table A7 (continued)
Solid fuels Petroleum Gas Nuclear Hydro Biomass PV Wind
COMB AUT 28.3% 35.3% 35.7% “n.a.” 13.0% 8.3% 1.3% 1.6%
GRC 22.8% 29.6% 29.6% “n.a.” 4.5% 4.9% 2.4% 1.7%
EEU 27.3% 34.2% 34.3% 11.4% 11.4% 9.7% 3.5% 2.8%
NEU 26.2% 33.2% 33.4% 9.9% 9.9% 6.6% 1.3% 1.4%
SEU 25.4% 32.3% 32.5% 8.6% 8.6% 6.5% 1.9% 1.6%
WEU 29.6% 36.7% 37.0% 14.8% 14.8% 9.6% 1.6% 1.9%
FFR_low 1/2 lifetime AUT 14.1% 15.2% 15.6% “n.a.” 13.0% 8.3% 2.9% 5.0%
GRC 7.2% 8.5% 8.5% “n.a.” 4.5% 4.9% 5.3% 5.2%
EEU 12.7% 13.9% 14.0% 11.4% 11.4% 9.7% 7.7% 8.5%
NEU 11.4% 12.7% 12.9% 9.9% 9.9% 6.6% 2.9% 4.3%
SEU 10.4% 11.7% 11.9% 8.6% 8.6% 6.5% 4.1% 5.0%
WEU 15.7% 16.8% 17.3% 14.8% 14.8% 9.6% 3.5% 5.8%
FFR_med AUT 18.3% 21.6% 21.9% “n.a.” 13.0% 8.3% 2.9% 5.0%
1/4 lifetime GRC 12.5% 15.8% 15.8% “n.a.” 4.5% 4.9% 5.3% 5.2%
EEU 17.2% 20.4% 20.6% 11.4% 11.4% 9.7% 7.7% 8.5%
NEU 16.1% 19.4% 19.6% 9.9% 9.9% 6.6% 2.9% 4.3%
SEU 15.2% 18.6% 18.7% 8.6% 8.6% 6.5% 4.1% 5.0%
WEU 19.7% 22.9% 23.3% 14.8% 14.8% 9.6% 3.5% 5.8%
References [18] D. García-Gusano, K. Espegren, A. Lind, M. Kirkengen, The role of the discount rates
in energy systems optimisation models, Renew. Sustain. Energy Rev. 59 (2016)
56–72, https://doi.org/10.1016/j.rser.2015.12.359.
[1] UNFCCC, Adoption of the Paris Agreement, United Nations, Paris, France, 2015. [19] B. Steffen, The importance of project finance for renewable energy projects, Energy
[2] J. Rockström, O. Gaffney, J. Rogelj, M. Meinshausen, N. Nakicenovic, Econ. 69 (2018) 280–294, https://doi.org/10.1016/j.eneco.2017.11.006.
H.J. Schellnhuber, A roadmap for rapid decarbonization, Science 355 (2017) [20] S. van Renssen, Investors take charge of climate policy, Nat. Clim. Change 4 (2014)
1269–1271, https://doi.org/10.1126/science.aah3443. 241–242, https://doi.org/10.1038/nclimate2175.
[3] J.-C. Hourcade, M. Jaccard, C. Bataille, F. Ghersi, Hybrid modeling: new answers to [21] N. Komendantova, T. Schinko, A. Patt, De-risking policies as a substantial de-
old challenges. Introduction to the special issue of “The Energy Journal”, Energy J. terminant of climate change mitigation costs in developing countries: case study of
27 (2006) 1–11. the Middle East and North African region, Energy Policy 127 (2019) 404–411,
[4] R. Loulou, M. Labriet, ETSAP-TIAM, The TIMES integrated assessment model Part I: https://doi.org/10.1016/j.enpol.2018.12.023.
model structure, Comput. Manag. Sci. 5 (2008) 7–40, https://doi.org/10.1007/ [22] T. Schinko, N. Komendantova, De-risking investment into concentrated solar power
s10287-007-0046-z. in North Africa: impacts on the costs of electricity generation, Renew. Energy 92
[5] S. Paltsev, J.M. Reilly, H.D. Jacoby, R.S. Eckaus, J. McFarland, M. Sarofim, (2016) 262–272, https://doi.org/10.1016/j.renene.2016.02.009.
M. Asadoorian, M. Babiker, The MIT Emissions Prediction and Policy Analysis [23] G. Frisari, M. Stadelmann, De-risking concentrated solar power in emerging mar-
(EPPA) Model: Version 4, (2005) Cambridge, MA. kets: the role of policies and international finance institutions, Energy Policy 82
[6] F. Bosello, The role of economic modelling for climate change mitigation and (2015) 12–22, https://doi.org/10.1016/j.enpol.2015.02.011.
adapation stratgies, in: A. Markandya, I. Galarraga, E. Sainz de Murieta (Eds.), [24] B. Steffen, T.S. Schmidt, A quantitative analysis of 10 multilateral development
Routledge Handbook of the Economics of Climate Change Adaptation, Routledge, banks' investment in conventional and renewable power-generation technologies
London, 2014. from 2006 to 2015, Nature Energy 4 (2019) 75–82, https://doi.org/10.1038/
[7] J.R. McFarland, J.M. Reilly, H.J. Herzog, Representing energy technologies in top- s41560-018-0280-3.
down economic models using bottom-up information, Energy Econ. 26 (2004) [25] T.S. Schmidt, Low-carbon investment risks and de-risking, Nat. Clim. Change 4
685–707, https://doi.org/10.1016/j.eneco.2004.04.026. (2014) 237–239, https://doi.org/10.1038/nclimate2112.
[8] S. Messner, L. Schrattenholzer, MESSAGE–MACRO: linking an energy supply model [26] O. Tietjen, M. Pahle, S. Fuss, Investment risks in power generation: a comparison of
with a macroeconomic module and solving it iteratively, Energy 25 (2000) fossil fuel and renewable energy dominated markets, Energy Econ. 58 (2016)
267–282, https://doi.org/10.1016/S0360-5442(99)00063-8. 174–185, https://doi.org/10.1016/j.eneco.2016.07.005.
[9] N. Bauer, I. Mouratiadou, G. Luderer, L. Baumstark, R.J. Brecha, O. Edenhofer, [27] F. Egli, B. Steffen, T.S. Schmidt, A dynamic analysis of financing conditions for
E. Kriegler, Global fossil energy markets and climate change mitigation – an ana- renewable energy technologies, Nature Energy 3 (2018) 1084–1092, https://doi.
lysis with REMIND, Clim. Change 136 (2016) 69–82, https://doi.org/10.1007/ org/10.1038/s41560-018-0277-y.
s10584-013-0901-6. [28] G. Bachner, K.W. Steininger, K. Williges, A. Tuerk, The economy-wide effects of
[10] C. McGlade, P. Ekins, The geographical distribution of fossil fuels unused when large-scale renewable electricity expansion in Europe: the role of integration costs,
limiting global warming to 2 °C, Nature 517 (2015) 187–190, https://doi.org/10. Renew. Energy 134 (2019) 1369–1380, https://doi.org/10.1016/j.renene.2018.09.
1038/nature14016. 052.
[11] N. May, K. Neuhoff, Financing Power: Impacts of Energy Policies in Changing [29] J.C. Steckel, M. Jakob, The role of financing cost and de-risking strategies for clean
Regulatory Environments, DIW Berlin, German Institute for Economic Research,, energy investment, International Economics 155 (2018) 19–28, https://doi.org/10.
2017, https://ideas.repec.org/p/diw/diwwpp/dp1684.html , Accessed date: 19 1016/j.inteco.2018.02.003.
March 2018. [30] J.-F. Mercure, H. Pollitt, J.E. Viñuales, N.R. Edwards, P.B. Holden, U. Chewpreecha,
[12] L. Hirth, J.C. Steckel, The role of capital costs in decarbonizing the electricity P. Salas, I. Sognnaes, A. Lam, F. Knobloch, Macroeconomic impact of stranded fossil
sector, Environ. Res. Lett. 11 (2016) 114010, https://doi.org/10.1088/1748-9326/ fuel assets, Nat. Clim. Change 8 (2018) 588–593, https://doi.org/10.1038/s41558-
11/11/114010. 018-0182-1.
[13] D. Angelopoulos, R. Brückmann, F. Jirouš, I. Konstantinavičiūtė, P. Noothout, [31] P. Noothout, D. de Jager, L. Tesniére, S. van Rooijen, N. Karypidis, R. Brückmann,
J. Psarras, L. Tesnière, B. Breitschopf, Risks and Cost of Capital for Onshore Wind F. Jirouš, B. Breitschopf, D. Angelopoulos, H. Doukas, I. Konstantinavičiūtė,
Energy Investments in EU Countries, Energy Environ. 27 (2016) 82–104, https:// G. Resch, The Impact of Risks in Renewable Energy Investments and the Role of
doi.org/10.1177/0958305X16638573. Smart Policies - Final Report, ECOFYS, 2016.
[14] B. Sweerts, F.D. Longa, B. van der Zwaan, Financial de-risking to unlock Africa's [32] R. Dimitrov, J. Hovi, D.F. Sprinz, H. Saelen, A. Underdal, Institutional and en-
renewable energy potential, Renew. Sustain. Energy Rev. 102 (2019) 75–82, vironmental effectiveness: will the paris agreement work? Wiley Interdiscip. Rev.
https://doi.org/10.1016/j.rser.2018.11.039. Clim. Change 10 (2019) 1–12, https://doi.org/10.1002/wcc.583.
[15] Oxera, Discount Rates for Low-Carbon and Renewable Generation Technologies, [33] CTI, Unburnable Carbon 2013: Wasted Capital and Stranded Assets, Carbon Tracker
Oxera Consulting Ltd, Oxford, UK, 2011https://www.oxera.com/wp-content/ Initiative & The Grantham Research Institute, 2013.
uploads/2018/03/Oxera-report-on-low-carbon-discount-rates.pdf , Accessed date: [34] M. Campanale, J. Leggett, J. Leaton, Unburnable Carbon: Are the World's Financial
22 February 2019. Markets Carrying a Carbon Bubble? (2011) https://www.carbontracker.org/
[16] S. Pfenninger, J. Keirstead, Comparing concentrating solar and nuclear power as reports/carbon-bubble/ , Accessed date: 18 February 2019.
baseload providers using the example of South Africa, Energy 87 (2015) 303–314, [35] R. Heede, N. Oreskes, Potential emissions of CO2 and methane from proved reserves
https://doi.org/10.1016/j.energy.2015.04.077. of fossil fuels: an alternative analysis, Glob. Environ. Chang. 36 (2016) 12–20,
[17] G. Pleßmann, P. Blechinger, How to meet EU GHG emission reduction targets? A https://doi.org/10.1016/j.gloenvcha.2015.10.005.
model based decarbonization pathway for Europe's electricity supply system until [36] P.A. Griffin, A.M. Jaffe, D.H. Lont, R. Dominguez-Faus, Science and the stock
2050, Energy Strategy Reviews 15 (2017) 19–32, https://doi.org/10.1016/j.esr. market: investors' recognition of unburnable carbon, Energy Econ. 52 (2015) 1–12,
2016.11.003. https://doi.org/10.1016/j.eneco.2015.08.028.
22
[39] J.P. Painuly, Barriers to renewable energy penetration; a framework for analysis, system perspective, Economics of Energy & Environmental Policy 6 (2017) 7–31,
Renew. Energy 24 (2001) 73–89, https://doi.org/10.1016/S0960-1481(00) https://doi.org/10.5547/2160-5890.6.1.wsch.
00186-5. [63] L. Hirth, The optimal share of variable renewables: how the variability of wind and
[40] I. Diaz-Rainey, B. Robertson, C. Wilson, Stranded research? Leading finance jour- solar power affects their welfare-optimal deployment, Energy J. 36 (2015)
nals are silent on climate change, Clim. Change 143 (2017) 243–260, https://doi. 149–184, https://doi.org/10.5547/01956574.36.1.6.
org/10.1007/s10584-017-1985-1. [64] T.H.Y. Føyn, K. Karlsson, O. Balyk, P.E. Grohnheit, A global renewable energy
[41] N. Bauer, C. McGlade, J. Hilaire, P. Ekins, Divestment prevails over the green system: a modelling exercise in ETSAP/TIAM, Appl. Energy 88 (2011) 526–534,
paradox when anticipating strong future climate policies, Nat. Clim. Change 8 https://doi.org/10.1016/j.apenergy.2010.05.003.
(2018) 130–134, https://doi.org/10.1038/s41558-017-0053-1. [65] N. Bauer, C. McGlade, J. Hilaire, P. Ekins, Divestment prevails over the green
[42] J. Rozenberg, A. Vogt-Schilb, S. Hallegatte, Instrument choice and stranded assets paradox when anticipating strong future climate policies, Nat. Clim. Change 8
in the transition to clean capital, J. Environ. Econ. Manag. (2018), https://doi.org/ (2018) 130–134, https://doi.org/10.1038/s41558-017-0053-1.
10.1016/j.jeem.2018.10.005 in press. [66] M. Hoogwijk, D. van Vuuren, B. de Vries, W. Turkenburg, Exploring the impact on
[43] G. Kieffer, T.D. Couture, Renewable Energy Target Setting, IRENA, Abu Dhabi, cost and electricity production of high penetration levels of intermittent electricity
United Arab Emirates, 2015. in OECD Europe and the USA, results for wind energy, Energy 32 (2007)
[44] A. Mahone, C.K. Woo, J. Williams, I. Horowitz, Renewable portfolio standards and 1381–1402, https://doi.org/10.1016/j.energy.2006.09.004.
cost-effective energy-efficiency investment, Energy Policy 37 (2009) 774–777, [67] D.E.H.J. Gernaat, D.P. Van Vuuren, J. Van Vliet, P. Sullivan, D.J. Arent, Global
https://doi.org/10.1016/j.enpol.2008.11.033. long-term cost dynamics of offshore wind electricity generation, Energy 76 (2014)
[45] J. Mayer, G. Bachner, K.W. Steininger, Macroeconomic implications of switching to 663–672, https://doi.org/10.1016/j.energy.2014.08.062.
process-emission-free iron and steel production in Europe, J. Clean. Prod. 210 [68] P. Sullivan, V. Krey, K. Riahi, Impacts of considering electric sector variability and
(2019) 1517–1533, https://doi.org/10.1016/j.jclepro.2018.11.118. reliability in the MESSAGE model, Energy Strategy Reviews 1 (2013) 157–163,
[46] G. Bachner, B. Wolkinger, J. Mayer, A. Tuerk, K.W. Steininger, Risk assessment of https://doi.org/10.1016/j.esr.2013.01.001.
the low-carbon transition of Austria's steel and electricity sectors, Environ. Innov. [69] P. Fragkos, N. Tasios, L. Paroussos, P. Capros, S. Tsani, Energy system impacts and
Soc. Trans. (2018), https://doi.org/10.1016/j.eist.2018.12.005 in press. policy implications of the European Intended Nationally Determined Contribution
[47] A. Aguiar, B. Narayanan, R. McDougall, An overview of the GTAP 9 data base, and low-carbon pathway to 2050, Energy Policy 100 (2017) 216–226, https://doi.
Journal of Global Economic Analysis 1 (2016) 181–208, https://doi.org/10.21642/ org/10.1016/j.enpol.2016.10.023.
JGEA.010103AF. [70] E3MLab, The PRIMES Model 2016-7, (2017) http://www.e3mlab.eu/e3mlab/
[48] E.C., Country Datasheets - June 2016 Update, EU Commission, DG ENER, Unit A4, PRIMES%20Manual/The%20PRIMES%20MODEL%202016-7.pdf.
2016, https://ec.europa.eu/energy/en/data-analysis/country. [71] H. Dai, S. Fujimori, D. Silva Herran, H. Shiraki, T. Masui, Y. Matsuoka, The impacts
[49] E.C., Country Datasheets - March 2017 Update, EU Commission, DG ENER, Unit A4, on climate mitigation costs of considering curtailment and storage of variable re-
2017, https://ec.europa.eu/energy/en/data-analysis/country. newable energy in a general equilibrium model, Energy Econ. 64 (2017) 627–637,
[50] B. Bednar-Friedl, T. Schinko, K.W. Steininger, The relevance of process emissions https://doi.org/10.1016/j.eneco.2016.03.002.
for carbon leakage: a comparison of unilateral climate policy options with and [72] H. Dai, X. Xie, Y. Xie, J. Liu, T. Masui, Green growth: the economic impacts of large-
without border carbon adjustment, Energy Econ. 34 (Supplement 2) (2012) scale renewable energy development in China, Appl. Energy 162 (2016) 435–449,
168–180, https://doi.org/10.1016/j.eneco.2012.08.038. https://doi.org/10.1016/j.apenergy.2015.10.049.
[51] P. Capros, A. De Vita, N. Tasios, P. Siskos, M. Kannavou, A. Petropoulos, [73] J.R. McFarland, J.M. Reilly, H.J. Herzog, Representing energy technologies in top-
S. Evangelopoulou, M. Zampara, D. Papadopoulos, C. Nakos, L. Paroussos, down economic models using bottom-up information, Energy Econ. 26 (2004)
K. Fragiadakis, S. Tsani, P. Karkatsoulis, P. Fragkos, N. Kouvaritakis, L. Höglund 685–707, https://doi.org/10.1016/j.eneco.2004.04.026.
Isaksson, W. Winiwarter, P. Purohit, A. Gomez Sanabria, S. Frank, N. Forsell, [74] H.D. Jacoby, J.M. Reilly, J.R. McFarland, S. Paltsev, Technology and technical
M. Gusti, P. Havlik, M. Obersteiner, H.P. Witzke, M. Kesting, EU Reference Scenario change in the MIT EPPA model, Energy Econ. 28 (2006) 610–631, https://doi.org/
2016 - Energy, Transport and GHG Emissions Trends to 2050, (2016) https://ec. 10.1016/j.eneco.2006.05.014.
europa.eu/energy/sites/ener/files/documents/REF2016_report_FINAL-web.pdf , [75] M.H.M. Babiker, J.M. Reilly, M. Mayer, R.S. Eckaus, I. Sue Wing, R.C. Hyman, The
Accessed date: 21 March 2018. MIT Emissions Prediction and Policy Analysis (EPPA) Model : Revisions,
[52] M. Fürsch, S. Hagspiel, C. Jägemann, S. Nagl, D. Lindenberger, E. Tröster, The role Sensitivities, and Comparisons of Results, MIT EPPA model, 2001, http://dspace.
of grid extensions in a cost-efficient transformation of the European electricity mit.edu/handle/1721.1/3574 , Accessed date: 31 January 2019.
system until 2050, Appl. Energy 104 (2013) 642–652, https://doi.org/10.1016/j. [76] E3MLab, The GEM-E3 Model Manual, (2017) http://www.e3mlab.eu/e3mlab/GEM
apenergy.2012.11.050. %20-%20E3%20Manual/GEM-E3_manual_2017.pdf.
[53] R. Kumar, Estimation of cost of capital, Valuation - Theories and Concepts, Elsevier, [77] P. Capros, D. Van Regemorter, L. Paroussos, P. Karkatsoulis, M. Perry, J. Abrell,
2016, pp. 93–118, , https://doi.org/10.1016/B978-0-12-802303-7.00004-8. J. Ciscar, J. Pycroft, B. Saveyn, Institute for Prospective Technological Studies,
[54] R.J. Clews, Political and environmental risks, tax and insurance, Project Finance for GEM-E3 Model Documentation, Publications Office, Luxembourg, 2013http://dx.
the International Petroleum Industry, Elsevier, 2016, pp. 225–247, , https://doi. publications.europa.eu/10.2788/47872 , Accessed date: 8 April 2015.
org/10.1016/B978-0-12-800158-5.00013-X. [78] B.C. O'Neill, E. Kriegler, K. Riahi, K.L. Ebi, S. Hallegatte, T.R. Carter, R. Mathur,
[55] F. Polzin, F. Egli, B. Steffen, T.S. Schmidt, How do policies mobilize private finance D.P. van Vuuren, A new scenario framework for climate change research: the
for renewable energy?—a systematic review with an investor perspective, Appl. concept of shared socioeconomic pathways, Clim. Change 122 (2014) 387–400,
Energy 236 (2019) 1249–1268, https://doi.org/10.1016/j.apenergy.2018.11.098. https://doi.org/10.1007/s10584-013-0905-2.
[56] J. Ondraczek, N. Komendantova, A. Patt, WACC the dog: the effect of financing [79] IEA, World Energy Outlook 2016, Inernational Energy Agency, 2016.
costs on the levelized cost of solar PV power, Renew. Energy 75 (2015) 888–898, [80] I.S.E. Fraunhofer, Current and Future Cost of Photovoltaics. Long-Term Scenarios
https://doi.org/10.1016/j.renene.2014.10.053. for Market Development, System Prices and LCOE of Utility-Scale PV Systems,
[57] IMF, Return on Equity - Non-financial Corporations Sector (Financial Soundness 2015, https://www.ise.fraunhofer.de/content/dam/ise/de/documents/
Indicators - FSIs), (2019) http://data.imf.org/. publications/studies/AgoraEnergiewende_Current_and_Future_Cost_of_PV_
[58] World Bank, Lending interest rate - FR.INR.LEND, https://databank.worldbank. Feb2015_web.pdf.
org/, (2019). [81] NREL, U.S, Solar Photovoltaic System Cost Benchmark: Q1 2017, (2017) https://
[59] ECB, MFI Interest Rate Statistics - Balance Sheet Item A2I, (2019) https://sdw.ecb. www.nrel.gov/docs/fy17osti/68925.pdf.
europa.eu/. [82] IRENA, Renewable Power Generation Costs in 2017, (2018) https://www.irena.
[60] B. Meyer, G. Ahlert, Imperfect markets and the properties of macro-economic-en- org/-/media/Files/IRENA/Agency/Publication/2018/Jan/IRENA_2017_Power_
vironmental models as tools for policy evaluation, Ecol. Econ. 155 (2019) 80–87, Costs_2018.pdf.
https://doi.org/10.1016/j.ecolecon.2017.06.017. [83] A. Schröder, F. Kunz, J. Meiss, R. Mendelevitch, Current and Prospective Costs of
[61] E. Campiglio, Beyond carbon pricing: the role of banking and monetary policy in Electricity Generation until 2050, DIW Data Documentation, No. 68, Deutsches
financing the transition to a low-carbon economy, Ecol. Econ. 121 (2016) 220–230, Institut Für Wirtschaftsforschung (DIW), Berlin, 2013, p. 105.
https://doi.org/10.1016/j.ecolecon.2015.03.020. [84] ECOFYS, Subsidies and Costs of EU Energy - Annex 4-5, ECOFYS, Berlin, Germany,
[62] W.-P. Schill, A. Zerrahn, F. Kunz, Prosumage of solar electricity: pros, cons, and the 2014.
23

Bachner 2019

Uploaded by

Copyright:

Available Formats

Bachner 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

Bachner 2019

Uploaded by

Copyright:

Available Formats

Energy Strategy Reviews 26 (2019) 100373

Contents lists available at ScienceDirect

Energy Strategy Reviews

Costs or benefits? Assessing the economy-wide effects of the electricity T

ARTICLE INFO ABSTRACT

1. Introduction Irrespective of how BU and TD models are combined, the results of

3.1. WACC revisited

Fig. 4. Change in bottom-up derived unit-costs (RES-e versus EU-ref).

Fig. 5. Change in electricity supply price (RES-e versus EU-ref).

Fig. 6. Change in gross domestic product (RES-e versus EU-ref).

A2: Underlying model and shared socio-economic pathway assumptions

A3: Electricity generation technology data

A5: Handling of investments for grids and storage infrastructure

A6: Additional material and explanations for WACC calculation

Fig. A7. Change in welfare (RES-e versus EU-ref).

Fig. A10. Change in total investment (RES-e versus EU-ref).

Fig. A11. Change in capital rents (RES-e versus EU-ref).

BU elesplan-m 6%-7% No [17]

AGRI Agriculture AUT Austria

Model Aggregate name Aggregated countries

AUT Austria Austria

Solid Fuels Petroleum Gas Nuclear Hydro Biomass PV Wind

AUT GW 1,262 1,105 4,425 - 12,795 648 298 1,267

Solid Fuels Petroleum Gas Nuclear Hydro Biomass PV Wind

NEU GW 37,000 15,672 46,669 22,155 23,353 10,915 2,241 16,241

Investment costs [EUR/kW] Economic lifetime [years] Operating expenditures [EUR/kWh]

2011 2050 AUT GRC EEU NEU SEU WEU

Solid Fuels 1,523 1,523 40 118 78 95 79 73 94

LCOE 2011 [EUR/kWh] AUT GRC NEU WEU EEU SEU

Solid Fuels 0.11 0.07 0.08 0.08 0.10 0.06

Solid fuels Petroleum Gas Nuclear Hydro Biomass PV Wind

Solid fuels Petroleum Gas Nuclear Hydro Biomass PV Wind

You might also like