SSRN Id3614875
SSRN Id3614875
SSRN Id3614875
Richard M. Ennis
Richardmennis.com
Richardmennis@gmail.com
June 25, 2020
Endowment funds in the U.S., underperform consistently across cohorts of fund size.
Significant underperformance is also observed in cross-sectional analysis of the returns of
the largest individual endowment funds.
Abstract
Endowment funds in the U.S., large and small, significantly underperform passive investment.
Moreover, an analysis of the performance of 43 of the largest individual endowments over the 11
years ended June 30, 2019, reveals that none outperformed with statistical significance, while
one in four underperformed with statistical significance. Alternative asset classes have failed to
deliver putative diversification benefits and have had an adverse effect on endowment
performance. Given prevailing diversification patterns and costs of 1 to 2% of assets, it is likely
that the great majority of endowment funds will continue to underperform in the years ahead.
This paper provides further insight into the performance of educational endowment funds. It
updates returns through June 30, 2019. Additionally, using a novel dataset of returns for
individual educational endowment funds, it analyzes those returns in cross section. The paper
attempts to disentangle the performance effects of (1) equity exposure, (2) portfolio size, (3)
degree of reliance on alternative investments, and (4) style tilts. Like the prior work, this paper
underscores the fact that diversification with high cost is a recipe for failure.
We have become accustomed to hearing and reading of “the endowment model.” There is, in
fact, no canonical model for managing endowment funds. There are, however, certain recurring
investment themes often associated with their management. We refer to these themes a bit more
modestly as elements of endowment style. They include:
Active Management. The ability of endowed institutions to identify and exploit investment
skill is an overarching theme among endowments, which make little use of passive
investments (14% of assets in 2018, according to Greenwich Associates1). Their willingness
to pay for perceived skill is exemplified by their large commitments to hedge funds and
private equity.
Equity Orientation. The belief that equity investments, broadly speaking, should
predominate portfolio holdings is widespread among endowment managers. That said, equity
exposure varies among institutions, largely based on the size of their portfolios. Large funds
maintain an average effective equity exposure of 72%, with figures of 80% or more being not
uncommon. Small funds average 63% (particulars follow).
Private Markets. Private markets are a central focus of many endowment managers and
their advisors. They are viewed as riper for exploitation by skillful investors as well as being
a source of diversification. Along with hedge funds, private investments form the core of
what are commonly known as alternative investments, or “alts.” Larger endowments avail
themselves of private market opportunities and other alts to a greater extent than do the
smaller ones.
1
See “2018 U.S. Institutional Market Trends,” Greenwich Associates, January 2019.
The endowment style of investing produced superior results in the 1990s and early years of this
century, so much so that it has become a subject of enduring interest among investors around the
world.
LONG-TERM PERFORMANCE
Hammond (2020) reviews the long-term performance of educational endowments. Over the last
50 years the average endowment earned 8.5% per year, compared with 9.3% for a 60/40
passive benchmark, for a shortfall of 0.8% per year.2 The full story, though, is richer, as
conveyed by Exhibit 1. The 60/40 portfolio beat the average in four of the five decades, with the
decade ended 2009 being the sole exception. The decade ended in 2009 coincides with the glory
days of the endowment style of investing. The cohort of large endowments, in particular, enjoyed
an advantage of 460 bps in that decade. Two additional observations may help to provide context
for the results summarized in Exhibit 1. First, the modern endowment investing style was not a
factor in the first two decades (70s and 80s) reported on there; in that era stocks and bonds
sufficed for all. Second, equity exposure, broadly defined, is not a constant across cohorts or
over time. From what we can glean from Nacubo archival data, “60/40” was probably a pretty
good benchmark across the board in the earliest decade.3 Equity exposure began to rise,
however, in the decades that followed, especially among the large endowments. We believe that
this fact largely accounts for the return spread among cohorts in the decade ended 1989, when
equities outperformed bonds by a wide margin and before alternative investments had made real
headway with the large endowments. These results leave us with insight into how investment
practice has evolved over half a century. What remains unexplained is what transpired at about
the time of the Global Financial Crisis of 2008 to cause endowments to lose their steam.
2
Nacubo archival records are the source of returns, which are represented as net of fees. “60/40”
incorporates the S&P 500 for stocks and Bloomberg Barclays Aggregate, or its equivalent, for
bonds in the early years.
3
1974 Nacubo Endowment Study
4
Source: Nacubo Endowment Study 2019. While the 2019 Nacubo study reports on seven
cohorts, we combined the Over $100 million - $250 million and Over $250 million - $500 million
cohorts into one (as shown in # 4 in Exhibit 2) for consistency with past Nacubo studies where
only six cohorts are reported. All Nacubo returns are represented as net of fees.
5
The indexes used are Russell 3000, MSCI ACWI ex-U.S. and Bloomberg Barclays Aggregate.
6
We use the term “risk-adjusted return” to describe value added as the intercept resulting from
regression of a composite’s or fund’s return on a benchmark index. The terms “risk-adjusted
return,” “intercept” and “alpha” are used interchangeably throughout the paper. We use “excess
return” to describe the simple difference between the returns of a composite and its benchmark.
Exhibit 3
Benchmark Weights and Performance Statistics of
Endowments by Size Cohort for the Eleven Years Ended June 30, 2019
7
Equal-weighted average of underlying institutions’ allocations to alt assets within the cohort.
8
The value factor is defined as the difference between the returns of Russell 3000 Value and
Russell 3000 Growth indexes.
Exhibit 4
Breakdown of Excess Return Between Value Tilt and Other Sources
for the Eleven Years Ended June 30, 2019
Excess Return
Due to
Other Bets
Composite Benchmark Due to &
Cohort Return Return Total Value Bet Costs
1. Smallest 5.58% 6.85% -1.26% -0.47% -0.79%
2. 5.32 6.90 -1.58 -0.17 -1.41
3. 5.21 6.79 -1.58 -0.23 -1.35
4. 5.25 7.03 -1.78 -0.15 -1.63
5. 5.47 7.42 -1.94 -0.13 -1.82
6. Largest 5.94 7.52 -1.58 0.00 -1.58
9
The Nacubo Study reports on seven size cohorts, one of which comprises 107 institutions with
endowments greater than $1 billion in value. However, Nacubo does not identify individual
schools. To create the individual fund dataset, we began reviewing the annual reports of schools
with the largest endowments, starting with Harvard University. We soon discovered that many
either do not report endowment fund returns, do not report the 11 consecutive returns required
for the present study or have fiscal-year-ends other than June 30. We wound up acquiring
complete return series for 35 of the 50 largest endowments. From there, we added data for other
schools with assets greater than $1 billion, as best as we could locate them. An indication that the
final sample of 43 funds may not be representative of all 107 funds (with assets greater than $1
billion) is that the average 11-year annual return of the sample exceeds that of Nacubo’s $1-
billion-plus fund cohort by approximately 40 bps. While the 43 individual fund returns we
obtained are not suspect in our minds, they may not be truly representative of the large fund
cohort. Accordingly, we use the individual fund returns only in cross-sectional analysis and not
as indicative of the large fund cohort itself. We assume, without confirmation, that returns are net
of fees.
Exhibit 5
Diversification and Performance of Large Endowment Funds
for the Eleven Years Ended June 30, 2019
Effective
Rank School Equity R2 Alpha t-stat
Exposure
1 Massachusetts Institute of Technology 64% 0.835 2.07% 1.25
2 Bowdoin College 71% 0.870 2.03% 1.26
3 Michigan State University 69% 0.969 1.39% 1.94
4 Williams College 71% 0.931 0.99% 0.86
5 University of Pennsylvania 67% 0.941 0.73% 0.73
6 Columbia University 68% 0.953 0.56% 0.60
7 University of California 69% 0.957 0.32% 0.36
8 University of Richmond 63% 0.937 0.29% 0.30
9 Dartmouth College 70% 0.917 0.25% 0.20
10 Princeton University 77% 0.910 0.23% 0.16
11 Rice University 71% 0.978 0.18% 0.29
12 University of Missouri 68% 0.995 0.01% 0.05
13 University of Virginia 76% 0.966 -0.01% -0.02
14 University of Notre Dame 73% 0.955 -0.12% -0.13
15 Rutgers University 62% 0.975 -0.18% -0.31
16 Yale University 77% 0.902 -0.21% -0.14
17 Wellesley College 69% 0.967 -0.22% -0.29
18 Northwestern University 62% 0.940 -0.42% -0.44
19 Brown University 74% 0.919 -0.87% -0.66
20 Pennsylvania State University 72% 0.987 -0.92% -1.75
21 University of Rochester 70% 0.956 -0.97% -1.05
22 Amherst College 74% 0.952 -1.06% -1.07
23 North Carolina State University 71% 0.888 -1.18% -0.79
24 Stanford University 78% 0.969 -1.19% -1.40
25 Vanderbilt University 60% 0.875 -1.24% -0.93
26 Purdue University 74% 0.990 -1.25% -2.75
Conclusion
The overarching conclusion of this section is that endowment funds have underperformed
passive investment by a significant margin during the study period, no matter how one
slices the data.
Effective equity exposure is the overwhelming determinant of risk and return, with 99%
of return variance explained by stock and bond indexes alone across size-cohorts of
funds.
A value bias is evident. It has had a negative effect on risk-adjusted performance during
the period of this study, mainly among smaller endowments. Even there, it accounts for a
minor part of observed underperformance.
Advocates of alternative investments claim that they are a source of diversification and
incremental risk-adjusted return. The sections that follow examine these propositions.
Exhibit 6
Diversification Patterns of Large Endowment Funds Pre- and Post-GFC:
R2 and Tracking Error
1. 2. 3. 4. 5.
U.S. Stocks Admit Admit Admit Admit
and Bonds Non-U.S. Real Private Hedge
Only Stocks Estate Equity Funds
Exhibit 7 graphically illustrates that the benchmark comprising U.S. stocks and bonds plus non-
U.S. stocks is the near-exclusive driver of return for the large fund composite. It shows the
regression of the large fund composite on its benchmark consisting solely of stocks and bonds in
the proportions shown for Cohort 6 in Exhibit 3. The slope (beta) is 0.99. R2 is .993 and the
standard error of the regression is a minimal 1.4%. The intercept, or alpha, is -1.46% (t-statistic
of -2.4). We note in passing that the annual standard deviation of return for composite and
benchmark are nearly identical, at 11.10% and 11.16%, which is to say there is no evidence of
“volatility dampening” in the return series of the alts-heavy composite. Moreover, in this
analysis we make no attempt to adjust for the return-smoothing characteristic of alternative
investments, which account for 51% of the assets of the composite.
25.0%
Endowment Composite Return
20.0%
15.0%
10.0%
5.0%
0.0%
-5.0%
-10.0%
-15.0%
-20.0%
-25.0%
-30.0% -20.0% -10.0% 0.0% 10.0% 20.0% 30.0%
Benchmark Return
These are remarkable results: Stock and bond indexes capture the return-variability
characteristics of alternative investments in the composite of large endowment funds for all
intents and purposes. Alternative investments do not have a meaningful impact. The finding
that the correlation between a composite of funds averaging 51% alts exposure and a marketable
securities benchmark is near-perfect runs counter to the popular notion that the return properties
of alts differ materially from those of stocks and bonds. That, after all, is an oft-cited reason for
incorporating alternative investments in institutional portfolios. But as we see here, alt returns
simply blend in with broad market returns in the context of standard portfolio analysis.
Furthermore, evidence of volatility-dampening is absent.
10
Exhibit 9 is similar to Exhibit 8. It shows total fund alpha versus alts allocation for the six
endowment fund size-cohorts (described in Exhibit 2) through June 30, 2019. The regression line
pertains only to Cohorts 1-5. The R2 is .89. The slope coefficient is statistically significant (t-stat
of -4.9), and the standard error of regression is a mere 0.10%. A very strong relationship exists,
in other words. For funds with alts allocations of up to about 40% of total assets, the story is
the same as for the public funds, namely: (1) alts detract from performance and (2) the
more you have, the worse you do. For Cohorts 1-5, a reduction in total fund alpha of ~20 bps is
associated with every 10% of assets allocated to alts. A 40% alts allocation results in a penalty of
79 bps.
11
-0.90%
-1.10% #1
Intercept (Alpha)
-1.30%
#6
#3
-1.50%
#2
-1.70% 64 bps
#4 #5
-1.90%
-2.10%
0% 10% 20% 30% 40% 50% 60%
Exposure to Alternatives
In terms perhaps better suited to trustees of endowment funds with less than a billion in assets
and others who may not be conversant with statistical jargon, the message here is this: Liquidate
your alternative investments and put the proceeds into index funds. Do it now.
Cohort 6 (funds with assets greater than $1 billion) stands apart from Cohorts 1-5 in Exhibit 9,
statistically as well visually. Total fund alpha of Cohort 6, although still negative by nearly 1.5%,
is 64 bps better than the regression equation would indicate for an average allocation to alts of
51% (a 6.5-sigma outlier relative to the regression line). It is reasonable to conjecture that the
more skilled practitioners of alternative investing are to be found among those with the heaviest
allocations there. Which is to say, there is arguably an indication of skill among at least some of
the practitioners of alternative investing. Anecdotally, three (and only three) stand out with
positive total fund alphas of greater than a percentage point. They are MIT, Bowdoin College
and Michigan State University. All maintain alts allocations of 55% or more and, yet, achieved
alphas at the total fund level of +2.07%, +2.03% and +1.39%%, respectively. Alas, even these
fetching figures do not rise to the level of statistical significance.10
10
There is a parallel here with the experience of professional gamblers. We know that the
majority of amateurs entering a casino will depart losers. Studies have shown that a tiny
percentage (about 1%) of professional gamblers exit winners with some consistency. (These
12
Exhibit 10
Average Annual Excess Return of Alternative Investments
Before and After the GFC
Pre-GFC Post-GFC
(1994-2008) (2009-2019)
3.3% 3.4%
1.3%
-1.0%
-2.9%
-6.6%
Real Estate LBOs Hedge Funds Real Estate LBOs* Hedge Funds
* Post-GFC LBO return of -2.9% is for the period 2009-2014
Sources: Cambridge Associates, FTSE NAREIT, L’Her er al. (2016), Sullivan (2020)
Exhibit 11 illustrates the excess return of the large endowment composite over 21 years. It is the
picture of a paradigm shift that dates to fiscal year 2009.
happen to be the whales.) Another group (about 5%) usually exit losers, albeit with smaller
losses than the Average Joe. The three large endowments cited above are analogous to the group
of about 1% of gamblers cited in a study of fantasy-sports betting by McKinsey & Co. In both
cases, there are rare winners in a game marked by a preponderance of losers of varying degree.
11
For real estate, we subtract the returns of the FTSE NAREIT All-Equity REIT Index from
those of the Cambridge Associates Real Estate Index, using quarterly IRRs to estimate TWRs for
the Cambridge series. For buyout funds these are the average excess returns reported by L’Her et
al. (2016) in Tables 3 and 4 for size-, leverage- and sector-adjusted returns. The hedge fund
excess returns are as reported by Sullivan (2020).
13
20.0% 18.5%
15.0%
Annual Excess Returns
10.0%
6.4%
5.0% 3.6%
3.1% 2.9% 2.9%
2.3%
1.5%
0.5% 0.5% 0.4%
0.0%
0.0%
-0.9% -0.6%
-1.1% -1.5%
-2.4% -2.3%-2.3%
-3.2%
-5.0% -3.8%
2009
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
COST AND DIVERSIFICATION
We posit that portfolios of marketable securities cost the investor 0.5% to 0.7% of asset value,
the percentage varying with the mix of stocks and bonds, the use of passive versus active
management and turnover rates. The cost of alternative investments begins at about 1% of asset
value for open-end diversified core equity (ODCE) real estate funds. Estimates of the cost of
private equity investing approximate 6.0% of invested capital.12 The cost of hedge funds, non-
core real estate and private debt fall between those extremes. We put the cost of a typical
portfolio of diverse alternative investments in the range of 2 to 4% of asset value annually. We
estimate that across the spectrum of endowment funds, their cost of investing ranges from 1 to
2% of assets.13 Assuming a cost of 0.6% for marketable securities and 2.5% for a diverse
portfolio alternative investments, a portfolio with an allocation of 50% to alts would incur an
annual cost of 1.55% of assets. The median R2 of the individual funds reported on in Exhibit 5 is
.96 with an associated standard error of return relative to benchmark of 2.7%. If an endowment
fund with that degree of diversification incurs costs of 1.5% annually, the likelihood of it
12
See McKinsey & Co. (2017) and Phalippou and Gottschalg (2009).
13
See our previous work for a detailed discussion of the cost-estimation procedure.
14
CONCLUSION
Notwithstanding the existence of a handful of arguably skillful endowment fund managers in the
realm of alternative investments, the vast majority of endowment funds incur costs that
overwhelm the limited opportunity to exploit mispricing. The alt-heavy approach to investing
has failed to provide a diversification benefit and has significantly underperformed simpler
approaches employing stocks and bonds alone. Absent a change in strategy, the great majority of
endowment funds, large and small, are likely to underperform by a significant margin in the
years ahead. Those confident of their ability to identify truly profitable alternative investments
consistently should concentrate those investments to a greater extent, just as they should do with
traditional active portfolios. Absent such confidence, they should shift assets to passive
investments.
REFERENCES
Ennis, Richard M. 2020. “Institutional Investment Strategy and Manager Choice: A Critique.”
Journal of Portfolio Management (Fund Manager Selection Issue): 104-117.
McKinsey & Co. “Equity Investments in Unlisted Companies: Report for the Norwegian
Ministry of Finance.” November 2017.
Phalippou, L., and O. Gottschalg. 2009. “The Performance of Private Equity Funds.” The Review
of Financial Studies 22 (4): 1747–1776.
14
The probability reflects the area under a normal distribution curve with a tracking error of
2.7% and incorporating a cost of 1.5% per year.
15