Swamy2009selection

Selection Supersedes Access PAGE 1

Selection Supersedes Access:
1

When does experience pay in Private Equity?
Gitanjali Swamy, Bhavin Shah, Nitin Nohria, Daniel Bergstresser, Irina Zeltser

Introduction
Private equity is one of the fastest growing yet most opaque industries today.
From its inception in the early 50s to today, it’s been at the center of the some of
the most powerful entrepreneurial and wealth creation stories. Not only has the
industry created hundreds of millionaires in both investors and operators, but it
has also launched, nurtured and revived some of today’s most successful
Sections
businesses.
1 Introduction
According to PEI, Thompson and other sources, the PE equity market has been
growing rapidly and reached $2.5 Trillion in 2008. About 60% of the allocation or
2 Common Beliefs & Myths
$1.5 Trillion is allocated to venture and buyout funds and the remainder is
allocated to distressed, mezzanine and other funds. It is not just the sheer size of
3 Analysis, Contention & Results
this market that is compelling but the growth rate is extraordinary as well. In a little
4 Refuting Myths over a decade from 1980 and 1994, the amount of private equity outstanding
rose from less than $5 billion to $100 billion at CAGR of 23%. In the first quarter of
5 Conclusions 2008 alone, U.S. private equity firms raised $58.5 billion up nearly 32% over the
$44.3 billion in the first quarter of 2007.

Buyout funds alone now control over $900 Billion in capital not including leverage,
and venture capital funds control another $350 Billion. Including the impact of
leverage, they have an aggregate buying power of $3 Trillion; that’s enough to
buy more than 40 McDonalds, 10 GEs, and 500 General Motors. What were 296
VC and 40 buyout firms in 1985 has evolved to 741 VC firms and 588 Buyout firms
in 2007. Inspite of the industry growth, LPs are investing less and less in new firms.
From 2002 to 2006, while the total number of funds have grown by a CAGR of
7.85% and the AUM have grown by 43.9%, the number of new funds have only
grown by 1.34% and new fund AUM has only grown by 35%. Most of the new
growth comes from new buyout funds; in the venture industry the total venture
AUM has grown at a CAGR of 30.7% as compared to new venture AUM at half
that rate or 16.2%. In the same period, the number of new venture funds has
shrunk at -2.6% as compared to the total number of venture funds that has grown
at a CAGR of 3.58%.
Note on FOIA
The remaining private equity alternatives such as distressed and lending funds are
Both the US and UK government emerging categories that show the same pattern of fragmentation. This
freedom of information acts require fragmentation, coupled with the lack of transparency, makes it difficult for
disclosure by government agencies investors to assess and compare funds. It has created an environment where
on performance of PE vehicles. The rumors run rife and facts are rarely consistent. While many1 such as Calpers, have
US government FOIA, Title 5, 1966, tried and somewhat succeeded at making private equity fund performance
together with various state more transparent and objective, they have not focused on debunking some of
legislations from 1980 through 2006, the myths surrounding funding behavior and investor selection.
provided us the means to obtain the
source data. Our aim has been to review and challenge the key myths that have historically
guided and continue to guide investor selection behavior by analyzing the
industry-wide and investor-specific performance data that has only recently
become publicly available in the industry.

1 Footnote1: The past works includes experiments by Lerner et al that tried to establish the superior performance of endowments over pension funds
(Smart Institutions, Foolish Choices) or the exploration of persistence by Kaplan, Schoar. However, no work ever examined the causality or reality of
persistence over a long time frame.

This version of the “Selection Supersedes Access” paper was created for the students of Finc-190, Private Equity at Harvard University Extension School spring 2009. This
document contains unpublished research and information which is confidential and proprietary. By accepting a copy of this document, the recipient acknowledges the
confidential and proprietary nature of this information, agrees to hold this information in strict confidence, agrees not to use this information for any purpose without the prior
written consent of theAuthors, and agrees to return this document upon the written request of the Authors. This document may not be reproduced or distributed for any purpose
unless authorized in writing by the Authors.. Contact gms@zucicap.com or gmswamy@fas.harvard.edu with questions.


One dominant myth that has become a crux of driving LP behavior is that
investing in successive funds of same firms or individuals is much safer, more
rewarding than investing in an emerging manager. This is the focus of our
research. According to Northern Trust Global Investments (NTGA) while 40% of
returns come from emerging managers, most LPs such as Calpers, NYCERS
allocate 1% or less to emerging manager programs. A vast majority of pension
funds, endowments, institutions, etc. make it a policy not even to look at or invest
in emerging manager funds. In fact, this situation has been getting progressively
worse; the NVCA reports that the percentage of new funds vs. follow-on funds
decreased from 33% in 2002 to less than 21% in 2007.We have been acting on
one of those urban myths that cannot be proven, but can easily be disproven.

The aim of our exercise is to either prove or disprove the myth of persistence and
to start an exploration of the real causes of better performance in private equity
firms.

Common Beliefs and Myths
In order to research and analyze this myth, we identified and challenged a
number of supporting beliefs that drive investor behavior. Investors have
traditionally believed that

1. Historical returns are a predictor of future performance and Persistence
of performance will continue.
2. Experienced fund managers will be more successful (what about
experienced investors or operators?) (i.e., better deal flow, edge in
deals, better financing, staff selection, etc.)
3. It is less risky to invest in tangential/new businesses founded by
experienced fund managers than to invest in emerging managers with
directly relevant skill sets

Note on Dataset This is based on several fallacious myths that the industry believes. Firstly, the
industry believes that experienced firms will deliver better returns, perhaps due to
Our dataset included all publicly their experience in the industry, history of working together and network of
disclosed pension and endowment data. relationships in the industry. The industry believes that experienced managers in
In particular, our dataset focused on 560 an irrelevant old category are still better bets in a new category than an
funds sponsored by 280 firms. 300 funds emerging manager with differentiable expertise in the same. E.g. a portfolio
had more than 1 successor and 132 manager would rather invest in a new India growth fund offered by an
funds had more than 3 successors. The experienced GP manager specializing in communications early-stage venture
start years ranged from 1978 to 2000. investing in Boston than a new fund composed of a team with individual team
The funds were distributed 36% venture, members have 20 years experience in Indian private equity. The industry believes
14% balanced, 29% buyouts. The funds that emerging manager’s teams will not have the source, operating and exit skill
were predominantly in the US with over set, network, experience to generate superior returns. Finally, LPs believe that
90% centered there but had 4% in investing in emerging managers does not present rewards commensurate with
Europe and 3% in Asia. The funds are the risk being taken.
distributed equally among all industries
with Biotech, Medical, Communications, Thus, the structure of the alternative investment management industry is geared
IT and Retail dominating. The mean (IRR) towards returning fund managers with a lot more obstacles, challenges and
of the data was 11.45%, median 8.45% straight-out refusals for emerging managers. In addition, the structure is inherited
and standard deviation 33%.We made biased towards a specialization model inherited from the public market investing,
almost no subjective assessments to which focuses only on financial specialization (small cap, mid cap, large cap)
segment, classify, manipulate data. The and not optimized for private equity, which is as much about active portfolio
product, industry, geography management with value add from industry, geography and financial expertise.
segmentation was collated from the LP This mentality or approach is true with respect to type of investment fund (i.e.,
sources where disclosed and completed venture, buyout, PE, etc.), industry specialization of fund (i.e., technology,
with data from the company’s own industrials, healthcare, etc.) and size of fund (i.e., early-stage, lower middle-
website. market, middle-market, large-cap, etc.)



Analysis & Contention
Note on Analysis The private equity industry has grown up believing in fallacy of persistence of
2
performance; good funds continue to do well and bad funds continue to badly.
A simple ordinary linear regression showed We have been acting on one of those urban myths that cannot be proven, but
an increasing positive correlation for values can easily be disproven. In fact, the traditional belief is actually completely
of N varying from 1 to 3, going from 0.52 to wrong. Our results show that contrary to common perception, Net IRR (net of fees
0.9. This clearly implied the persistence of and carry) actually shows a pattern of decreasing persistence beyond a small
returns for the first 3 funds. However, after window of 1-3 funds. From an LPs perspective Net IRR is a far more meaningful
fund 3, 4, the correlation changed direction metric since it reflects eventual returns to the LP and not the GP.
and became progressively smaller with N
going to 7. It started at 0.9 and decreased On an anecdotal basis, there are examples of both lack of persistence and
monotonically to 0.22. Eventually it turned persistence in performance. For example, A Partners (a top tier Hamilton lane
negative. rated firm) showed vintage 1998, 2000 returns of -17% and -26.8% respectively.
These were in no way predicted by their 1989, 1993, 1996 funds with IRRs of 14.7%,
A robust regression which down-weighted 77.8% and 189% respectively. On an average, Funds in the top 10% (decile) in
the outliers showed an even more startling performance showed a drop of more than 30% from Fund 1 to Fund 2 and Fund 3.
pattern that showed a monotonically The top decile showed a drop of more than 50% from Fund 4 to Fund 5 and
decreasing and almost immediately Fund6. Finally, among firms having 5 or more funds, 80% showed a decline of
negative correlation at N=2 onwards. The more than 50% from the 5th fund to the 7th fund. But while anecdotal evidence
correlation started at 0.009 and decreased may suggest patterns, the real proof comes from a comprehensive statistical
monotonically through N=7 to -0.39. analysis.

The standard quantile regression showed The objective of our exercise was to successfully prove or disprove what makes
exactly the same pattern with a minor better or worse private equity firms. The first step was to examine whether any
flattening of the curve at N=1, 2. The persistence existed; i.e. did good firms continue to do well and bad firms continue
correlation started at 0.2 and decreased to do badly. In order to see if past performance was in any way a determinant of
monotonically to -0.3 at N=7. It turned future, we conducted our experiments on nearly 600 closed funds from the public
negative at N=5. A winsorized regression disclosures of the top pension funds. In particular the classifications we compare
that eliminated the outliers all together and correlate the (net of fees) performance of N successive funds of a GP with
showed a similar pattern with a slower rate the net performance of the N+1th fund. N was varied from 1 to 10. If indeed
of increase through 1, 2 from 0.22 to 0.39 persistence held, the correlation coefficient would go up or stay constant but
and flattening of the correlation through would never go down. A negative correlation coefficient would indicate the
3,4, followed by a rapid monotonic decrease reverse of persistence.
going from 0.39 down to -0.1. To summarize, our experiments show that in fact the performance of N+1th fund
does NOT directly track2 the performance of the prior N (N=1,2,3,4,….10) funds. In
short, good funds do not continue to do well and bad funds do not continue to
Figure 1. Ordinary Linear Regression do badly. In fact, our comprehensive analysis shows that persistence is limited
Correlation Fund Performance vs. Prior Fund and at most restricted to a window of 3 funds. In general, there is decreasing
Performance. correlation between successive funds performances.

In conclusion, the industry belief in the persistence of returns is a fallacy. The
analytical selection of new good managers overwhelmingly trumps the
advantages of relationship access to existing managers in good private equity
fund management.

2
Footnote2: The correlation between the performance of N (N=1, 2, 3, 4….10) past funds and the next N+1th fund shows decreasing correlation as N is increased. In short, there
is limited persistence of performance. Contrarily, there is decreasing correlation and in some cases even an inverse (negative) correlation with increasing funds. A robust
regression which down-weighted the outliers showed an even more startling pattern that showed a monotonically decreasing and almost immediately negative correlation at
N=2 onwards. The correlation started at 0.009 and decreased monotonically through N=7 to -0.39. We used the Stata environment to perform multivariate regression analysis.
Our analysis adjusted for vintage variation in performance by creating an independent variable for each vintage year and using it as part of the regression analysis.



Refuting the Myths
Figure 2. Robust Regression Correlation
Fund Performance vs. Prior Fund
In fact, our experiments and analysis of underlying data show that none of the Performance.
common myths in private equity investing hold in reality.

Historical returns alone are NOT a predictor of future performance in private
equity firms and Persistence of performance will NOT continue. In fact, our
comprehensive analysis shows that persistence is limited and at most restricted
to a window of 3 funds. In general, there is decreasing correlation between
successive funds performances. A simple ordinary linear regression showed an
increasing positive correlation for values of N varying from 1 to 3, going from
0.52 to 0.9. This clearly implied the persistence of returns for the first 3 funds.
However, after fund 3, 4, the correlation changed direction and became
progressively smaller with N going to 7. It started at 0.9 and decreased
monotonically to 0.22. Eventually it turned negative. A robust regression which
down-weighted the outliers showed an even more startling pattern that
showed a monotonically decreasing and almost immediately negative
correlation at N=2 onwards. The correlation started at 0.009 and decreased
monotonically through N=7 to -0.39. This result clearly implied that in general
persistence does not apply except for a small window and if the outliers were
less pivotal, persistence did not hold at all.

Experienced fund managers will NOT always be more successful l. It is not just
the years of experience but the nature of experience. Moreover, experienced
fund managers tend to raise their fee/carry compensation. Thus, the net IRR or
returns to the investor actually does not show a superior return even when the
performance goes up. To illustrate this point, we looked at the top 10% (decile)
Note on Experiment
in our data set. We aggregated and averaged the performance data by fund
numbers. On an average, net IRR performance dropped 30% from Fund 2 to
We adjusted for the impact of vintage in
Fund 3 and over 50% from Fund 5 to Fund 6 or Fund 7.
all experiments by creating an extra set
of independent variables that reflected
It is NOT less risky to invest in tangential/new businesses founded by
the vintage. The correlation was run
experienced fund managers than to invest in emerging managers with directly
using these independent variables and in
relevant skill set. Once again our correlation results disprove this myth. As
effect adjusting for the impact of
explained earlier, our comprehensive analysis shows that persistence is limited
vintage. Some variables that we wish we
and at most restricted to a window of 3 funds. In general, there is decreasing
had access to but did not for this set of
correlation between successive funds performances. We compared the
experiment were, the Write-down,
correlation of ex
Organizational Turnover and Fee/Carry
terms. These variables would allow us to
see the impact of and adjust for changes
in the fund team, the pinpoint the cause
of return as an overall portfolio vs. single
company impact. For instance Google
was singlehandedly responsible for the
Figure 2. Data Extraction, Clean-up and Analysis Process returns of Kleiner vintage 1996 fund.
Fee/Carry terms would allow us to
determine whether the degradation in
persistence were due to a decrease in
the quality of the investment choices or
due to increase in compensation
demanded by a successful firm. We did
not adjust for management changes in
fund or subsequent significant fund size
changes by firm managements



3

Note on Results
Conclusions
Most of the results are statistically
In conclusion, we found that persistence of returns is a fallacious argument to
significant. But some data points should
justify LP investment choices in particular GPs. In reality, selection significantly
be ignored. In the OLP, The significance
supersedes access and therefore investors must spend more time in finding new
(P<|t|) varied from 4.3 exp-13 down to
GPs with the right strategic, operational and financial attributes, rather than
0.1 at N=5. Thus, N=6, 7 could be ignored
assume that proprietary access to past top-performers will guarantee future
in the OLP regression.
returns.
However, the robust and quantile
We believe that this result may be attributable to several factors.
regressions were more statistically
significant in the larger N’s. The
• Impact of successful firms increasing fees/carry. The increases in
significance (P<|t|) varied from 0.08 at
performance don’t translate to increases in the returns to the LP
N=3 and decreased down to 0.003 at
investors because successful funds increase their share (fees/carry) of
N=7. N=1, 2 were not statistically
the returns. This phenomenon has been observed in many industries
significant. The quantile regression
including the mutual fund industry. This mis-alignment of manager
showed a variegated pattern once again
incentives and investor incentives results in poor returns to the investor.
N=5 thru 7 were usually statistically
A good illustration of this came from the 2005 Mayfield XII, which
significant for most quantiles. At quantile
raised its fee to 2.5% and carry to 30% as compared to the industry
30, the significance was 0.01, implying at
standard of 2%/20, on the grounds that the1995 Mayfield IX, their most
1% error.
recent closed fund, returned a 49% IRR. The prior fund $1Bil, 2000
Mayfield XI, wasn’t closed and was only X% invested. LPs such as the
Kirsch foundation exited the fund with an IRR of -1.3%. Both Harvard
and MIT declined to participate because of extortionate fees but
Mayfield closed $325Mil irrespective through investors such as the
State of Alaska.

• Impact of industry and macro economic cycles/changes. Thus
expertise, relationships in any particular industry or product are not Figure 3. Percentage Num of First Funds vs.
perpetually valuable. A good firm must be willing to continually re- Follow on Funds.
invent itself and experienced fund managers with one specialization
do not guarantee future success either in the same industry or a
tangential one. Matrix Partners, one of the best east coast firms
returned a 213% and 516% IRR for 1995 Matrix IV and 1998 Matrix V
funds based on its investments and expertise in the communications
sector. In particular, the returns were very sensitive to the Cascade
and Sycamore investments both of which were communications
companies founded by a particular entrepreneur. By 2000, the
entrepreneur started funding his own companies and no longer
worked with Matrix the communications sector collapsed and 2000
Matrix VI returned a dismal -8%.

• Impact of GP staff turnover. The private equity organizations
themselves change and with a new set of partners/investing
professionals; the firm is not the same. It cannot be reasonably
expected that LPs will get the same returns. With the departure of
Luminaries such as Vinod Khosla, Kevin Compton, John Doerr was the
sole remaining star at Kleiner Perkins. None of the almost brand new
stable of partners/principles all of whom were luminaries but not
venture capitalists were able to provide returns. Since Doer’s 1999
Google IPO, the firm has yet to demonstrate a killer IPO. The partner
and investment staff turnover didn’t impact Kleiners ability to raise
funds, just its ability to deliver returns.

3 The same pattern was show by other statistical metrics such as mean+standard deviation, max etc. This data suggests that in fact, experienced
fund managers were NOT more successful.



Figure 4. First Time vs. Follow-on Venture
• Impact of capital oversupply. The increasing amount of capital, increasing fund Funds
size, and increasing number of funds are not balanced by a corresponding
increase in investment talent. Thus, firms can neither find high quality talent to
support increased fund deployment nor do they have the time to develop talent
in-house. In 2007, Blackstone raised $21.7 billion for the world's biggest private-
equity fund, just as a global credit crunch slowed leveraged buyouts. The fund was
triple the $6.45 billion pool that Blackstone raised five years ago. These funds were
raised over the year inspite of the catastrophic earnings, revenue and share price
collapse. Over the same year, Blackstone’s revenue fell 94 per cent from $1.23
billion to $68.5 million, earnings collapsed from more than $1Bil to a net loss of
$255Mil and the stock fell from its post IPO price of $35/share to $12/share.

• Impact of getting rich fat and happy when rewards independent of performance.
Principals in a successful firm lose their incentive to perform after reaping inordinate
rewards that are not tied to performance. There is significant incentive
misalignment between fund managers that reap heavy rewards from
management fees alone and investors whose returns are determined by
performance. Stephen Schwarzman, chairman and cofounder of the
aforementioned Blackstone Group, owns 39% of Blackstone, which operates
leveraged buyout and real estate funds, and funds of funds. Last year Blackstone
earned $2.3 billion on an average of $62 billion in assets. Schwarzman took home:
$940 million. The real kicker is that he only paid 15% federal income tax on that
income. That's the same tax bracket that a single worker is in if she earns between
$16,575 and $40,600 (assuming she claims the personal exemption and standard
deduction). A single worker earning between $40,600 and $85,850 is in a 25%
bracket. Net net, there is a limited motivation for an investment professional
rewarded in the aforementioned manner to perform.

The excessive focus on returning funds and lack of focus on emerging funds is neither
justified nor a “safe” strategy. It is not geared to generate the best returns for the corpus,
investors, clients, etc. The best investment strategy is to re-invest in the same firms at most
within a small window of opportunity of about 3 funds/terms or about 7 years. Instead an LP
must continually seed new funds in order to continue to reap returns over the long run.

We recommend a 30-50% allocation for emerging managers as opposed to the miniscule
percentage that LPs traditionally allocate. Coincidentally, that is strategy that “Smart Lps ”
like Yale Endowment etc. also apply in practice.

Sadly, the industry is doing just the opposite. From 2002 to 2006, the percentage of new firms
vs. old firm funds raised has dropped from almost 30% down to less than 22%. During that
time, while the total number of funds has grown by a CAGR of 7.85% and the AUM have
grown by 43.9%, the number of new firm funds have only grown by 1.34% and new fund
AUM has only grown by 35%. Most of the new growth comes from new buyout funds; in the
venture industry the total venture AUM has grown at a CAGR of 30.7% as compared to new
firm venture AUM at half that rate or 16.2%. In the same period, the number of new venture
firm funds has shrunk at -2.6% as compared to the total number of venture funds that has
grown at a CAGR of 3.58%.

In effect, the industry is creating an unstable situation that is progressively biasing the
investment selection towards poorer returns. It is a myth that investing in experienced
managers guarantees predictable or above average returns. However, having disproven
this myth does not mean we have completely identified what makes a better manager.

We are still conducting our experiment and future publications will include.
- Market trends/shifts
- Manager/partner departures
- Spinoffs vs. truly new starts
- Geography
- Larger fund sizes vs. same fund sizes
- Fee vs. carry trade-offs

Our next step is to expand the analysis to account for variations in industry, geography and
product focus. Following that, we would be able to do a detailed causal analysis that
deduces which factor – strategy, management, culture, diversity, vintage years, incentives
and needs (net worth), fund size, fee vs. carry trade-offs, motivations etc. actually determine
the performance. the “Selection Supersedes Access” paper was created for the students of Finc-190, Private Equity at Harvard University Extension
This version of School spring 2009. This

Swamy2009selection

More Related Content

What's hot (17)

Viewers also liked (6)

Similar to Swamy2009selection (20)

Recently uploaded (20)

Swamy2009selection