Diller Kaserer PE Return Drivers PDF

European Financial Management, Vol. 15, No.
3, 2009, 643–675
doi: 10.1111/j.1468-036X.2007.00438.x
What Drives Private Equity Returns? –

Fund Inflows, Skilled GPs, and/or
Risk?∗
Christian Diller
Capital Dynamics, Bahnhofstr. 22, CH-6301 Zugi
E-mail: cdiller.@capdyn.com
Christoph Kaserer
Center for Entrepreneurial and Financial Studies (CEFS) & Department for Financial
Management and Capital Markets, Technische Universität München, Acisstr. 21, D-80290
München
E-mail: christoph.kaserer@wi.tum.de
Abstract
This paper analyzes the determinants of returns generated by mature European
private equity funds. It starts from the presumption that this asset class is
characterized by illiquidity, stickiness, and segmentation. Given this presumption,
Gompers and Lerner (2000) have shown that venture deal valuations are driven
by overall fund inflows into the industry that yield the putative ‘money chasing
deals’ phenomenon. It is the aim of this paper to show that this phenomenon
explains a significant part of the variation in private equity funds’ returns. This
is especially true for venture funds, as they are affected more by illiquidity and
segmentation than buy-out funds. In the context of a WLS-regression approach
the paper reports a highly significant impact of total fund inflows on fund returns.
It can also be shown that private equity funds’ returns are driven by GP’s skills
as well as stand-alone investment risk. In a bootstrapping context we can show
that most of these results are quite stable.
∗
The authors would like to thank the anonymous referee, Giorgio di Giorgio, Didier
Guennoc, Dietmar Harhoff, Ulrich Hege, Donald Hester, Josh Lerner, Bernd Rudolph,
Stefan Ruenzi; the participants of the 2004 EVI conference at Tuck Business School,
Hanover (NH); the Finance and Economics Seminars at LUISS; the RICAFE Conference
2004, Frankfurt; the research workshop on ‘Managing Growth: The Role of Private Equity’
at IESE 2004, Barcelona; the ODEON-CEFS seminars in entrepreneurship and finance,
Munich; the annual meetings of the German Finance Association 2005, Tübingen; the French
Finance Association 2004, Paris; and the German Association of University Professors of
Management 2005, Kiel, for many helpful comments. We are grateful to the European
Venture Capital and Private Equity Association (EVCA) and Thomson Venture Economics
for making the data used in this study available. The normal caveat applies.

C 2008 The Authors
Journal compilation
C 2008 Blackwell Publishing Ltd.
644 Christian Diller and Christoph Kaserer
Keywords: Private equity funds, performance measurement, venture capital,

buyout, IRR, PME, ‘money chasing deals’ phenomenon
JEL classification: G24
1. Introduction
Despite the increasing importance of private equity as an asset class, the economic
characteristics of this industry are not yet fully understood. Among the growing
literature, one strand is focused on the characteristics and determinants of private equity
returns. This paper aims to make a contribution with respect to this topic, the discussion
of which is especially influenced by Gompers and Lerner (2000) and Inderst and Müller
(2004), who emphasize the impact of a specific competitive environment in the private
equity industry.
In frictionless and perfectly competitive capital markets, returns on investments in
private equity funds should be determined by systematic risk only. Neither the personal
skills of the management team, nor the inflow of money into private equity funds, nor
the funds’ characteristics (insofar as they are not related to systematic risk) should have
an impact on the performance of these funds. Due to the specific characteristics of
the private equity asset class, e.g. the illiquidity of the investment, the stickiness of
fund flows, the restricted number of target companies and the segmentation from other
asset classes, the market may be far from being frictionless and perfectly competitive,
at least in the short run. A very important finding in this regard has been presented
by Gompers and Lerner (2000), who show that inflows into venture funds and target
companies valuations correlate positively. Although it is an open question whether
increased valuations are triggered by money pouring into the private equity industry
or whether this money flow is triggered by improved expectations with respect to future
investment opportunities, and hence by increased valuations, Gompers and Lerner (2000)
present evidence that is more consistent with the first explanation. They basically argue
that, at least in the short run, there are a limited number of favourable investments in the
private equity industry that must be matched with a fluctuating capital supply, giving
way to the so-called ‘money chasing deals’ hypothesis.
This hypothesis states that there should be a negative correlation between a fund’s
performance and the amount of money directed towards the private equity industry
during the investment phase of this fund. It is a major challenge of this paper to develop
an approach that is able to test this hypothesis.
More specifically, this paper may extend the existing literature in the following three
ways. First, by using a data set of 200 mature European private equity funds over the
period 1980 to 2003 provided by Thomson Venture Economics (TVE), we are able to
develop a comprehensive regression model explaining a significant part of the variation
in private equity returns. Moreover, due to the small size of the data set, we use a
bootstrap inference in order to check the robustness of the results. Second and more
importantly, we propose a test for the ‘money chasing deals’ hypothesis that basically
relies on the fact that we distinguish between absolute and relative cash inflows into
private equity funds. We show that for a given absolute fund inflow, an increase in the
allocation of money towards a particular fund type has a significant negative impact on
the performance of that fund type. Moreover, this effect is much stronger for venture
funds than for buy-out funds. This seems to suggest that segmentation and stickiness
is present to a greater extent in the venture industry than in the buy-out industry. This

C 2008 The Authors
Journal compilation
C 2008 Blackwell Publishing Ltd
What Drives Private Equity Returns 645
finding strongly supports the ‘money chasing deals’ hypothesis. Third, we find returns
to be associated positively with certain measures that represent the general partners’
skills as well as stand-alone investment risk. However, we find no evidence that private
equity funds’ returns are correlated with stock market returns; moreover, they even seem
to be associated negatively with the development of the economy as a whole.
The paper is organized as follows: in section 2 we review the literature briefly and
present the theoretical background for the statistical tests. Section 3 describes the data
set and some methodological issues. Results are presented in section 4. Section 5
summarizes the results and gives a brief outlook.
2. Related Literature and Theoretical Considerations
2.1. The ‘Money Chasing Deals’ Phenomenon

Among other things, this paper addresses the hypothesis, commonly called the ‘money
chasing deals’ hypothesis, that fund returns are triggered by total fund inflows during
the fund’s inception period. It is interesting in this regard that the venture capital
industry is known to be highly cyclical, with large swings in supply and demand. 1
That being so, we investigate the extent to which such variations in the supply of funds
affect the performance of private equity funds. Our analysis was influenced by three
papers in particular: in Gompers and Lerner (2000) the so-called ‘money chasing deals’
hypothesis was first proposed and analyzed. Ljungqvist, Richardson, and Wolfenzon
(2007) investigate the impact of GPs’ fund-raising and draw-down behaviour on fund
performance. Lastly, Inderst and Müller (2004) developed an equilibrium model that
describes the particular conditions of supply and demand in the venture capital industry.
Gompers and Lerner (2000) argue that the private equity market is far from being
a perfect and frictionless capital market. According to their view this is mainly due to
the fact that this asset class is segmented from other asset classes. Private equity funds,
which account for the largest part of money invested in private equity, are normally
not allowed to invest their committed capital in other asset classes. As a consequence,
even if a GP were aware of an overvaluation in the industry, he would hardly be able
to redirect newly committed funds towards investment projects outside the asset class.
Hence, to the extent that an increase in capital inflow is not matched by an increase
in the number and size of investment projects, the competition to finance companies
increases and, as a consequence, valuations increase. This effect will be amplified if the
asset class is illiquid. 2 In this case, the additional funds pouring into the industry have
to be absorbed mainly by primary markets, i.e. by new investments in target companies.
Private equity fund managers, who believe that current valuations on the market are
too high, find it rather difficult to sell their shares in target companies due to the lack
1
This has been documented, among others, by Lerner (2002).
2
Private equity fund investments can still be considered as being rather illiquid, as stakes of
LPs are traded only infrequently. However, the secondary market for these funds has grown
over the last years. According to statistics presented by the EVCA, secondary transactions
have become the fourth most important exit channel in 2006. About 17% of all exits by
European private equity funds were realized as secondary deals. Hence, the degree of
illiquidity of the private equity asset class may be going to be reduced in the future. Cf.
http://www.evca.com/html/PE industry/facts divestments.asp.

C 2008 The Authors
Journal compilation
of liquidity on secondary markets. Finally, as Ljungqvist, Richardson, and Wolfenzon

(2007) point out, capital flows between GPs and LPs tend to be sticky. It takes longer
than normal to adjust the invested capital to changed expectations or valuations. This
mechanism makes it even more difficult to redirect funds to other asset classes quickly,
which further reinforces the pressure on deal valuations. As a Corollary, it should be
noted that the segmentation and stickiness argument may be more relevant for venture
funds than for buy-out funds. The latter have a much broader set of potential investment
targets, including public equity. Hence, we expect venture funds to be more prone to the
‘money chasing deals’ phenomenon than buy-out funds. 3
Gompers and Lerner (2000) analyze more than 4,000 venture financing rounds of
privately held firms through the period 1987 to 1995. They showed that the firm
valuation in a financing round increases in proportion to the amount of money poured
into the venture industry over the year before the deal was closed. This relationship is
robust and perceivable in economic terms. However, although the authors integrated
a lag structure into their regression model, they were forced to admit that on the
basis of the evidence they provided, it is not obvious whether higher valuations due to
better economic prospects cause higher inflows or whether higher inflows cause higher
valuations. Nevertheless, Gompers and Lerner (2000, p. 316 n.) argue in favour of the
second relationship, i.e. the ‘money chasing deals’ hypothesis. The most important fact
supporting their interpretation was the performance of deals closed in ‘hot’ periods, i.e.
periods with relatively high inflows. The economic performance of these deals was no
better than the performance of deals closed during ‘cold’ periods.
Inderst and Müller (2004) model the relationship between the entrepreneur and venture
capitalist as a double-sided incentive problem. The model consists of two parts: (i) a
model of contracting and bargaining and (ii) a search model that links outside options to
the relative supply of venture capital. The result is an equilibrium in which capital market
characteristics affect the relative supply of, and demand for, capital. These variables
influence the bargaining power and ownership shares, which affect the pricing and
value creation in start-ups. The authors point out that the bargaining power of the venture
capitalist, as well as the venture capitalist’s equity share, decreases if the competition in
the venture capital market increases, i.e. when a lot of money is flowing into the venture
capital market, and vice versa. An important ingredient for this model is the assumption
that in the short run the supply of venture capital adjusts more quickly to changes in
market conditions than the demand, i.e. the supply of new ideas.
In the spirit of these papers, Ljungqvist, Richardson, and Wolfenzon (2007) stress the
importance of the competitive environment faced by the management team of the private
equity fund. First, they argue that due to competition for investment targets, the fewer
the available favourable targets, the more GPs come under pressure. Assuming that the
number of newly founded companies in a particular industry is a good proxy for the total
size of favourable investment projects, they show that the fewer the number of newly
founded companies, the longer it takes to return a given multiple of committed capital
to the LP. In other words, the tougher the competition for favourable investment projects
becomes, the lower the return for the private equity fund. Second, they show that the
greater the inflow of money into private equity funds, the longer it takes to return a given
multiple of committed capital to the LP. Given that the time to return a given fraction of
3
Inderst and Müller (2004) show that this can, in fact, be the equilibrium outcome in a model
where the relative bargaining power of entrepreneurs and venture capitalists depend, among
other things, on the relative scarcity of venture capital.

C 2008 The Authors
Journal compilation
money is negatively related with the IRR, and also other return measures, their results
could also be stated as follows: the more money that is poured into the industry in a
given vintage year, the lower is the return of funds closed in that particular vintage year.
The better the prospects of a particular industry, as measured by the number of newly
founded companies in that industry, the higher the fund’s returns will be. These findings
are in perfect accordance with the ‘money chasing deals’ hypothesis. Moreover, they
are further corroborated by recent research presented by Gottschalg and Zipser (2006).
However, Phalippou and Gottschalg (2007) do not find support for this hypothesis.
In section 4 we will present results that corroborate the view that deal valuations are
affected significantly by the amount of funds flowing into the industry. In contrast to
Ljungqvist, Richardson, and Wolfenzon (2007), however, we will set up a direct test of
this hypothesis, i.e. we will present evidence that the private equity funds’ returns are
correlated negatively with excess capital pouring into the industry. The critical part of
this analysis is, of course, the measurement of excess capital. This will be explained in
more detail in section 4.2.
2.2. The Impact of Fund Characteristics on Performance

A substantial part of the literature has focused on the question of whether private
equity funds are able to generate sufficiently high risk-adjusted returns. 4 Ljungqvist,
Richardson, and Wolfenzon (2007) analyze cash flow data provided by one of the largest
institutional investors in private equity in the US between 1981 and 2000. They use the
excess IRR with respect to a S&P 500 investment, to assess a fund’s risk-adjusted
profitability. They document an outperformance of 5 to 8% per year on average.
Phalippou and Gottschalg (2007) analyze the return of a sample of mature private
equity funds provided by TVE on the basis of a profitability index. In this context, they
document a profitability index between 1.01 and 0.92, depending on the treatment of
residual net asset values. Moreover, they report negative average alphas. Kaplan and
Schoar (2005) analyze a data set from TVE that includes 746 funds with vintages years
ranging from 1980 to 2001. By using the PME approach, they show that the average
returns of the funds are quite close to the S&P 500 returns. In fact, they found the PMEs
to be in a range from 0.96 to 1.05 on average. The IRR is around 18%.
A completely different approach is used by Zimmermann, Bilo, Christophers, and
Degosciu (2004). They concentrate on a set of 229 publicy traded private equity vehicles.
Evidently, in such a Case, a straightforward performance calculation applies. They
document substantially larger Sharpe ratios of 1.5 for listed private equity firms than
for traditional asset classes. They calculate a positive correlation between private equity
and the MSCI World of 0.40 and the Global Bond Index of 0.02.
To sum up, the current state of research gives no definite indication as to whether
private equity generates positive or negative risk-adjusted returns. Be that as it may, the
more interesting perspective is related to performance drivers. It has been pointed out
that the putative ‘money chasing deals’ phenomenon is closely related to the illiquidity,
segmentation, and stickiness of private equity markets. From this it follows that the skills
of the management team should have a more significant impact on fund returns than is
4
For a discussion of different methods to measure private equity funds’ returns, see Kaplan
and Schoar (2005) or, more extensively, Phalippou and Gottschalg (2007) and Kaserer and
Diller (2004b).

C 2008 The Authors
Journal compilation
the case for funds that are invested in public market securities. In efficient public markets
a great deal of information, public or private, is incorporated into asset prices. Hence,
the ultimate outcome of an investment strategy should be almost the same, regardless
of whether or not the investor undertakes costly informational activities. In fact, the
literature on the performance of mutual funds provides no clear evidence that fund
returns may be driven by fund managers’ skills, such as selection and timing abilities. 5
In addition, due to the lack of illiqudity and stickiness of public securities markets, there
is as yet no evidence that mutual fund returns are determined by past fund inflows or
by other factors driven by investor sentiment. 6
We expect fund management skills to be much more important in private equity funds
than in public mutual funds. Knowledge about investment opportunities in the private
equity industry may be distributed very unequally and, due to the lack of a continuous
market for this assets, it may take a long time for this information to be disseminated.
In fact, Hege, Palomino, and Schwienbacher (2003) argue that the outperformance of
US venture capital funds relative to their European counterparts is due, at least in
part, to the superior screening abilities of US-based GPs. The first consequence of this
idea is that deal returns should have a much higher volatility than public stock market
prices. 7 If there is a systematic difference in knowledge about private equity investment
opportunities among different management teams, it is to be expected that good deals
will be concentrated in a few fund portfolios, i.e. the portfolios of the skilled management
teams. In fact, it is well known (and this will be corroborated in this paper) that the
distributions of returns on private equity funds are heavily skewed. Finally, if skills are
distributed unequally at a given point in time, it may well be that their distribution is not
independent over time. Hence, we would expect the returns of different funds run by the
same management team to be correlated. This correlation yields the so-called persistence
phenomenon with respect to the returns of private equity funds. The phenomenon has
been documented by Kaplan and Schoar (2005), Ljungqvist, Richardson, and Wolfenzon
(2007) or also Phalippou and Gottschalg (2007). According to Kaplan and Schoar (2005)
it is more pronounced for venture funds. Also in this paper we document persistence in
fund returns, and it seems not to be due to the GP’s market timing abilities. This is in
5
For instance, Henriksson (1984) found only weak evidence to support the idea that mutual
fund managers can time the market effectively, although his methodology has recently been
subjected to criticism; see Goetzmann, Ingersoll, and Ivkovich (2000). More generally, the
evidence on performance persistence in mutual funds indicates that this may be a short-term
phenomenon, at the most. In this regard, see Hendricks, Patel, and Zeckhauser (1993) and,
more recently, Deaves (2004). No evidence in favour of market timing abilities and only
weak evidence in favour of selection abilities has been found by Daniel, Grinblatt, Titman,
and Wermers (1997). Similar results are also documented for closed-end funds; see Madura
and Bers (2002). Short-term persistence seems also to apply to real estate mutual funds; see
Lin and Yung (2004). Moreover, it seems there is no evidence to support the presumption
that fund managers with local information advantages generate superior performance; see
Otten and Bams (2007). In addition, there is no clear evidence in favour of the view that
investor sentiment has an impact on mutual fund returns; see Doukas and Milonas (2004).
6
However, there is evidence that investors chase returns, i.e. that mutual funds that have
been successful in the past attract additional money; see Deaves (2004). Evidence presented
by Madura and Bers (2002) on foreign closed-end funds is, to some extent, compatible with
the view that investor sentiment drives closed-end fund prices.
7
This is confirmed by the findings of Cochrane (2005).

C 2008 The Authors
Journal compilation
contrast to the results presented by Schmidt, Nowak, and Knigge (2006), who find clear
support for market timing abilities during the investment phase of the fund.
As far as fund characteristics are concerned, Ljungqvist and Richardson (2003) found
that a fund’s excess IRR has an inverse U-shaped relationship with fund size, while
Phalippou and Gottschalg (2007) and Kaplan and Schoar (2005) document a significant
positive linear relationship between performance and size. Ljungqvist and Richardson
(2003) did not find a significant relationship between a funds systematic or total risk
and its abnormal return as measured by the excess IRR.
3. Data
We use a dataset of European private equity funds that was provided by Thomson Venture
Economics (TVE). 8 It should be noted that TVE uses the term ‘private equity’ to cover
all venture investing, buyout investing, and mezzanine investing. 9 In fact, TVE provided
various information related to the timing and size of cash flows net of fees, residual
net asset values (RNAV), fund size, vintage year, fund type, fund stage, and liquidation
status for a total of 777 funds for the period 1980-2003. We use the same definitions
as TVE for the different fund types and stages. A synopsis of these definitions can be
found in Table 1.
As one can see from Panel A in Table 2, about 59% of the funds in our sample are
venture capital funds, while the remaining 41% are categorised as buyout funds. The
average fund size, according to the TVE data, is Euro 182.75m. 10 Variation in fund size
is high; the largest fund is 132 times as large as the median fund. Moreover, as one might
expect, buyout funds are on average about 3.7 times as large as venture capital funds.
As far as the stage of the sample funds is concerned, it can be seen that one quarter are
early stage funds, about one seventh are balanced funds, and almost one fifth are late
stage funds. Not surprisingly, the size of the funds differs significantly depending on
their stage focus.
In order to measure fund performance, it is necessary to know the entire cash flow
history. Panel B in Table 2 reveals that there are only 95 liquidated funds in our data
set. If we were to restrict the analysis to these funds only, we would miss out on a great
deal of useful data. It is widely accepted that in order to mitigate this problem it is
permissible to integrate mature, but not yet liquidated, funds into the analysis as well.
While in the literature mature funds are often identified by using rule of thumb methods,
like a minimum fund age or a given number of periods without any distributions, 11 in
this paper a more formal definition of mature funds is introduced. A fund is regarded
as being mature if the residual net asset value is small relative to cumulated cash flows.
In such cases, treating the residual net asset value as a final cash flow is supposed to
have a minor impact on the IRR or any other return measure used. 12 One can think of
8
TVE records private equity data for five different world regions, one of which is Europe.
9
Fund of fund investing and secondaries are also included within the scope of TVE’s usage
of the term ‘private equity’. TVE does not use the term to include angel investors or business
angels, real estate investments, or other investing scenarios outside the public market.
10
It should be noted that TVE calculates the fund size on the basis of committed capital.
11
Such rules are used in Kaplan and Schoar (2005), Ljungqvist, Richardson, and Wolfenzon
(2007), Phalippou and Gottschalg (2007) and others.
12
In fact, it will be shown that the impact of the size of the residual net asset value on

C 2008 The Authors
Journal compilation
Table 1
Private equity funds’ type and stage definitions according to Thomson Venture Economics
(TVE)
Type definitions:
• Venture capital funds (VC): TVE use the term to describe the world of venture investing.
It does not include buyout investing, mezzanine investing, fund of fund investing or
secondaries. Angel investors or business angels are also not included in the definition.
• Buyout funds (BO): TVE use the term to describe the world of buyout investing and
mezzanine investing. It does not include venture investing, fund of fund investing or
secondaries. Angel investors or business angels are also not included in the definition.
Stage definitions:
• Early Stage (ES): A fund investment strategy that includes investment in companies
for product development and initial marketing, manufacturing and sales activities. We
include seed and start-up funds in this definition.
• Balanced/Diversified (B): A venture fund investment strategy that includes investment
in portfolio companies at a variety of stages of development (Seed, Early Stage,
Diversified, Later Stage).
• Late Stage (LS): Development funds provide for the major growth expansion of a
company whose sales volume is increasing. Although the company has clearly made
progress, it may not yet be showing a profit. The money invested is used to finance
the initial development of the young company. Later stage fund investment involves
financing the expansion of a company that is producing, shipping and increasing its sales
volume. In this definition, we include all the funds that are designated as development
(DEV), expansion (EX) and Late Stage (LS).
• Buyout (LBO): The term is used to describe the world of leveraged buy-out investing.
It does not include venture investing, fund of fund investing or secondaries. Angel
investors or business angels are also not included in the definition.
• Buyout (PE): Similar to LBO, with the difference that only nonleveraged buy-out
investing (e.g. MBOs), bridge financing and mezzanine investing is considered here.
two obvious measures for considering a fund to be mature. First, this might be the case
if the residual net asset value RNAVN at the end of a period N is small relative to total
paid-in capital. Defining TD t as the capital paid into the fund at time t, this condition
can be written as follows: 13
RNAVN
N ≤ c1 (1)
t=0 TDt
Second, one might define a fund to be mature, if the residual net asset value is small
relative to total distributions. Defining D t as the distribution paid by the fund at time t,
performance is rather small for these mature funds, especially as far as the PME-measure
is concerned. Nevertheless, it should be noted that net asset values may be biased; see
Dittmann, Maug, and Kemper (2004).
13
In principle, it would be better to use discounted cash flows in the denominator rather
than undiscounted. However, we believe that this difference is not so important here, given
that this effect could be taken into account by adjusting the cutoff points c i . Moreover, the
approach presented here allows the following conditions to be easily transformed into the
very simple condition (4).

C 2008 The Authors
Journal compilation
this yields the following condition:

RNAVN
N ≤ c2 (2)
t=0 Dt
Although both measures are rather similar, they are not equivalent. In fact, some
simple transformations show that implicitly there is a different weighting of draw-
downs and distributions in both approaches. 14 For any c 1 = c 2 , condition (2) is biased
towards funds with high distributions, i.e. funds with high gains, while condition (1) is
biased towards funds with low distributions, i.e. funds with high losses. In fact, it can
be shown that condition (2) tends to include more funds with a DPI 15 larger than one,
while condition (1) tends to include more funds with a DPI smaller than one. 16 Hence,
as both approaches might introduce a bias, although in an opposite direction, we chose
to define the cutoff point for mature funds as a combination of these two measures. By
inverting both sides of the inequalities (1) and (2) and adding the right-hand as well as
left-hand sides of these expressions, i.e. assigning equal weight to both measures, we
get the following condition:
N N
t=0 TDt + t=0 Dt c1 + c2
≥ (3)
RNAVN c1 c2
Finally, inverting this inequality once again and setting q = c1 c2

c1 +c2
yields
RNAV N
N N ≤q (4)
t=0 TDt + t=0 Dt
Of course, q must be chosen arbitrarily. 17 In this paper we will set q equal to 0.1 and
0.2, respectively. Hence, we add a nonliquidated fund to our sample if its residual net
asset value does not exceed 10% resp. 20% of the undiscounted sum of all previously
accrued cash flows, disregarding their sign. For these funds, it is assumed that the
residual net asset value is paid out to the LPs by the end of the observation period.
In what follows we will distinguish three different subsamples. First, the sample of all
liquidated funds. Second, sample I, which consists of all liquidated funds plus all funds
14
The reader should note that the residual net asset value of a fund can be defined as the sum
of all takedowns plus the net book value gains(BVG) minus the sum of all paid out dividends,
i.e. the following equation holds: RNAVN = t=0 TDt + BVGN − t=0 Dt . Now, substituting
N N
N
this equation into equation (1) yields the following result: t=0 TDt (1 − c1 ) + BVGN ≤
N N
Dt . The same transformation with respect to equation (2) gives: t=0 T Dt + BVGN ≤
N t=0
t=0 Dt (1 + c2 ).
As one can see, the two conditions will only lead to the same cutoff point,
N
if the condition Nt=0TDt = cc12 holds.
D
t
t=0 N
The reader should note that the following definition holds: DPI N =
15 Dt
Nt=0
TDt
.
t=0
16
To see this,
note that according to footnote 14 condition (1) can be rewritten as(i) BVGN ≤
N
t −
N
TDt (1 − c1 ), while condition (2) can be written as (ii) BVGN ≤ t=0 Dt (1 +
N
t=0 D
N t=0
c2 ) − t=0 TDt . Now, for any c 1 = c 2 the right-hand side of (ii) is larger (smaller) than the
right-hand side of (i), for all funds with DPI N > 1 (DPI N < 1).
17
Note that by using the definition RVPI N = RNAV N
N
TD
condition (4) can also be written as:
t=0 t
1+DPI N
RVPI N
≥ q1 .

C 2008 The Authors
Journal compilation
that had not been liquidated by 30, June, 2003 and that satisfy condition (4) for q =
0.1. Third, sample II, which has the same definition as sample I with the exception that
q = 0.2 holds. A short description of these three samples is given in Panel C of Table 2.
Sample I consists of 200 funds, while sample II has 262 funds. This is a preceptible
increase, given that we have only 95 liquidated funds. For a large part of our analysis
we will concentrate on the intermediate sample I; hence, a more detailed description of
this sample can be found in Table 3.
4. Empirical Results
4.1. Return Distribution of European Private Equity Funds

We start by briefly presenting some basic results with respect to the return distribution
of European private equity funds. First, it should be noted that there is an ongoing
debate on how to measure the return distribution of an illiquid investment. Given that
this paper is focused on the determinants of private equity returns, we do not emphasize
this issue. 18 However, because the shortcomings of the IRR are well-known we use
three alternative performance measures in our study: the PME, the excess-IRR, and the
undiscounted payback period. The PME is defined as the ratio of the present value of
all cash distributions over the present value of all take-downs. The year-by-year realized
return on the MSCI Europe equity index is used as the discount rate. 19 While we have
private equity cash flows net of fees in our data set, we can only observe the returns on a
market index gross of management fees; hence, we reduce market returns by an assumed
management fee of 50bp per year. 20 As a consequence, the net yearly return is equal to
the gross yearly return, as indicated by the index performance, times 0.995. Finally, we
also report the excess-IRR, defined as a fund’s IRR minus the return on a simultaneous
stock market index investment, as well as the payback period, defined as the number of
months it takes before cumulated distributions equal cumulated take-downs. It should
18
A more detailed discussion of this issue in the context of the data set used here can be
found in Kaserer and Diller (2004a). For an extensive discussion see also Phalippou and
Gottschalg (2007).
19
The PME is also used by Kaplan and Schoar (2005) and Phalippou and Gottschalg (2007),
although the latter call it profitability index (PI). Of course, alternative performance measures
could be used as well. For instance, depending on the reinvestment hypothesis, one could
argue that cash flows have to be discounted with a risk-free rate of return or that this
hypothesis should apply, at least, to take-downs, because investors may hold their committed
capital in liquid accounts. Kaserer and Diller (2004b) propose the bond market equivalent
(BME) as an alternative performance measure, where cash flows are discounted with the
realized return on a European government bond index. The reader should note that the rank
correlation between the PME, as defined here, and the BME according to Kaserer and Diller
(2004b), is 0.96. Hence, for the purposes of this paper, both performance measures would
lead to similar results. For the sake of conciseness we do not report the results related to the
BME here.
20
According to Miller (2005) passively managed stock funds have total expense ratios (TERs)
of about 50bp. According to anecdotic evidence collected by the authors the TER of large
European index funds should be about 30 to 50bp Moreover, Miller (2005) reports that the
average TER of actively managed institutional large-cap funds is 77bp.

C 2008 The Authors
Journal compilation
Table 2
Sample characteristics
The complete data set provided by Thomson Venture Economics (TVE) includes 777 European private
equity funds raised over the period 1980 to 2003. TVE use the term ‘private equity’ to describe the
world of all venture investing, buyout investing and mezzanine investing. Following TVE, we use the
following type definitions. Venture capital funds (VC) represent the world of venture investing. It
does not include buyout investing, mezzanine investing, fund of fund investing or secondaries. Angel
investors or business angels are also not included in the definition. Buyout funds (BO) represent the
world of buyout investing and mezzanine investing. Moreover, the following stage definitions are used.
Early Stage (ES) is a strategy for fund investment that involves investment in companies for product
development and initial marketing, manufacturing and sales activities. We included seed and start-up
funds in this definition. Balanced/Diversified (B) is a strategy for venture fund investment that includes
investment in portfolio companies at a variety of stages of development (Seed, Early Stage, Diversified,
Later Stage). Late Stage funds include development funds that provide for the major growth expansion
of a company whose sales volume is increasing. Later stage fund investment also involves financing the
expansion of a company that is producing, shipping and increasing its sales volume. Size is measured
as total capital committed to a fund.
Panel A: Characteristics according to fund-stages and -types
Type of Funds All Venture Capital Funds VC BO

Total Total
Stage of Funds Early Stage Balanced Late Stage
# of funds 777 197 116 143 456 321
in % 100.0% 25.4% 14.9% 18.4% 58.7% 41.3%
Size (m€)
Average 182.75 70.89 144.13 60.50 86.26 319.81
Median 47.80 28.20 40.35 30.00 31.20 85.20
Stdev 513.04 122.55 435.79 109.38 243.66 722.35
Panel B: Characteristics according to liquidation status
All Liquidated funds Nonliquidated funds

# of funds 777 95 682
in % 100.0% 12.2% 87.8%
Size (m€)
Average 182.75 52.14 202.87
Median 47.80 26.20 53.10
Stdev 513.04 103.62 546.30
Panel C: Characteristics according to subsamples
Liquidated Funds Sample I Sample II

# of funds VC 47 99 131
BO 48 101 131
Total 95 200 262
Size in m€ Mean 24.61 52.10 54.32
VC Median 18.90 26.20 28.50
Stdev 24.37 96.90 91.15
Size in m€ Mean 79.09 104.53 188.16
BO Median 34.60 46.90 50.90
Stdev 139.26 150.83 600.34
Total Mean 52.14 78.05 121.24
Median 26.20 33.10 39.10
Stdev 103.62 128.89 433.76

C 2008 The Authors
Journal compilation

654
Table 3
Characteristics of funds in sample I
C 2008 The Authors

For not yet liquidated funds the residual net asset value was considered as a final distribution. The multiple is the ratio of the sum over all distributions to the sum
Journal compilation
over all take downs. Maturity is measured in years.
Vintage Number of funds Total take downs (m€) Mean multiple Mean maturity
Year
VC BO Total VC BO Total VC BO Total VC BO Total
1980 1 0 1 5.18 n/a 5.18 4.02 n/a 4.02 21.00 n/a 21.00
1981 3 1 4 37.21 42.13 79.34 1.65 2.57 1.88 14.33 17.00 15.00

1982 1 0 1 17.38 n/a 17.38 1.14 n/a 1.14 10.00 n/a 10.00
1983 4 0 4 66.78 n/a 66.78 1.64 n/a 1.64 16.75 n/a 16.75
1984 3 5 8 70.88 108.42 179.30 1.29 1.74 1.57 17.33 17.00 17.13
1985 14 3 17 284.21 130.45 414.66 1.30 1.98 1.42 13.79 14.00 13.82
1986 9 6 15 178.27 119.10 297.37 1.38 1.17 1.29 12.22 14.33 13.07
1987 11 8 19 964.58 541.93 1506.52 1.65 1.72 1.68 14.45 12.75 13.74
1988 13 18 31 506.28 715.10 1221.38 1.33 1.33 1.33 12.69 12.33 12.48
1989 14 11 25 1057.69 1104.14 2161.84 3.65 1.20 2.57 13.86 11.09 12.64
1990 8 10 18 381.60 1205.67 1587.27 2.22 1.65 1.90 12.00 11.20 11.56
1991 6 9 15 100.95 451.89 552.84 1.44 1.84 1.68 11.67 10.78 11.13
1992 5 4 9 111.57 186.25 297.82 2.22 2.62 2.40 11.00 11.00 11.00
Christian Diller and Christoph Kaserer
1993 2 6 8 79.52 590.48 670.00 3.70 1.82 2.29 10.00 7.50 8.13
1994 0 9 9 n/a 1445.49 1445.49 n/a 2.31 2.31 n/a 8.67 8.67
1995 0 3 3 n/a 606.96 606.96 n/a 1.49 1.49 n/a 8.00 8.00
1996 3 0 3 30.07 n/a 30.07 5.95 n/a 5.95 5.33 n/a 5.33
1997 2 4 6 274.47 402.76 677.22 1.98 1.45 1.63 6.00 5.25 5.50
1998 0 2 2 n/a 203.59 203.59 n/a 1.07 1.07 n/a 5.00 5.00
1999 1 0 1 37.26 n/a 37.26 2.04 n/a 2.04 4.00 n/a 4.00
2000 1 0 1 8.16 n/a 8.16 5.29 n/a 5.29 3.00 n/a 3.00
be noted that all four return measures display a statistically highly significant degree of
rank correlation. 21
The private equity fund performance on the basis of the four different performance
measures is resumed in Tables 4 and 5. The average IRR of samples I and II is
considerably higher than the IRR of the subsample consisting of liquidated funds only.
In fact, starting with an average IRR of 10% for the liquidated sample, we reach an
IRR of about 13% for sample I and 14% for sample II. These figures are slightly lower
than the results reported by Kaplan and Schoar (2005) and Phalippou and Gottschalg
(2007). Simultaneously, the standard deviation of the IRRs increases significantly when
the dataset is expanded.
From Table 4 it should be noted that the IRRs for samples I and II are calculated by
assuming that the residual net asset value could be realized immediately as a final cash
flow. Evidently, this is a quite strong assumption. However, given the selection criteria
for samples I and II that were presented in Section 3, we expect a bias in the net asset
value not to have a too strong impact on the results. In fact, as shown in Table 4 the mean
IRR will only decrease from 12.69% to 12.41%, if it is assumed that only 50% of the
residual net asset value can be realized as a final cash flow. However, by assuming a total
write-off of the residual net asset value, the mean IRR goes down to 11.64% for sample
I, which equals a reduction of almost 2 percentage points or about 8% percent. For
sample II the impact of writing-off the residual net asset values is substantially stronger;
the mean IRR would decrease from 14.07% to 13.47%, for a 50% consideration of the
residual net asset value, down to 10.97% for total write-off. As an additional point, one
should note that the impact of writing-off the residual net asset value on the IRR is
stronger for VC-funds than for BO-funds.
Moreover, we would like to stress the highly skewed distribution of all the return mea-
sures except the payback. This is in line with our presumption that unequally distributed
skills and industry knowledge among GPs generate a skewed return distribution. It also
should be noted that the average payback in all three different subsamples is about
90 months, or 7.5 to 7.8 years. This figure is very close to the result of Ljungqvist and
Richardson (2003), who document a payback period of slightly less than seven years.
In Table 5 the results for the excess-IRR as well as the PME are presented. The
excess-IRR is in the range of 0.58% to 6.68%. For sample I it is 4.45%, which is in line
with the results presented by Ljungqvist and Richardson (2003). The impact of a total
write-off of the residual net asset value is rather small, as in this case the excess-IRR
goes down to 3.97%. However, for sample II the impact is quite substantial. Depending
on the sample definition the PME is between 0.86 and 1.03. For sample I it is equal to
0.96, which is very close to the results presented by Phalippou and Gottschalg (2007)
and Kaplan and Schoar (2005). For a total write-off of residual net asset values this
average goes down to 0.89, which is equal to a 7% reduction. This is in line with results
presented by Phalippou and Gottschalg (2007), who document a 9% reduction in the
mean performance index by going from a 100% consideration of residual net asset values
to a 0% consideration. However, for sample II the reduction is, once again, much more
pronounced, showing that the parameters used for defining mature funds together with
the assumptions with respect to the residual net asset value actually can have a strong
impact on the performance results. Altogether these results are an important rationale
21
For sample I, the rank correlation coefficient of the IRR with the PME is 0.921, with the
excess-IRR 0.969 and with the payback -0.742. The degree of correlation is almost the same
for the different fund types. Due to limited space we do not report detailed results here.

C 2008 The Authors
Journal compilation
Table 4

656
Distribution of IRR(δ) and Payback Period for Different Fund Types

IRR(δ) is the fund’s internal rate of return calculated on basis of its cash flows. For those funds that were not yet liquidated by the end of the observation period,
the fraction δ of the residual net asset value is considered as a final distribution to limited partners. The payback gives the number of months it takes until for the
first time the sum of distributions is at least as high as the sum of take downs.
C 2008 The Authors

Journal compilation
IRR(δ) IRR(δ) IRR(δ) Payback in Month
δ=1 δ = 0.5 δ=0 δ=1

Liquidated Funds
Average 7.32% 12.64% 10.01% 108.53 83.63 94.62
Median 4.77% 9.79% 7.28% 110.00 84.50 95.50

75th Percentile 12.98% 18.67% 14.24% 143.50 113.75 120.50
25th Percentile −4.00% 8.23% 0.00% 87.50 62.25 66.50
Min −12.12% −13.66% −13.66% 32.00 21.00 21.00
Max 103.73% 88.05% 103.73% 215.00 139.00 215.00
Stdev 17.82% 17.67% 17.85% 41.38 33.61 38.99
Sample I
Average 12.00% 13.39% 12.69% 11.83% 12.87% 12.41% 10.77% 12.54% 11.64% 102.79 78.38 90.35
Median 8.05% 10.80% 9.14% 7.53% 10.25% 8.93% 7.40% 9.59% 8.77% 104.00 70.00 90.00
75th Percentile 15.65% 18.76% 17.13% 14.83% 18.36% 17.02% 14.28% 17.36% 16.27% 127.50 106.75 118.50
25th Percentile 1.90% 9.00% 4.45% 0.38% 6.37% 3.85% 0.00% 1.37% 0.39% 74.50 54.25 61.50
Min −13.56% −13.66% −13.66% −14.53% −17.78% −17.78% −55.76% −17.79% −55.76% 16.00 20.00 16.00
Max 153.91% 88.05% 153.91% 142.27% 88.05% 142.27% 141.49% 88.05% 141.49% 215.00 169.00 215.00
Stdev 22.06% 16.18% 19.34% 22.02% 17.02% 18.82% 23.37% 17.20% 20.51% 41.90 33.94 39.86
Sample II
Average 12.50% 15.63% 14.07% 11.08% 14.39% 13.47% 8.68% 13.26% 10.97% 99.58 81.48 90.09
Median 7.40% 11.00% 9.56% 7.40% 10.80% 10.76% 7.23% 8.89% 8.07% 96.50 71.00 8400
75th Percentile 16.31% 19.95% 18.17% 15.24% 19.52% 17.21% 14.06% 19.36% 16.23% 127.00 108.75 118.00
25th Percentile 0.00% 1.69% 0.50% 0.00% 0.97% 0.19% 0.00% 0.69% 0.07% 69.50 54.25 60.75
Min −13.56% −13.66% −13.66% −33.56% −14.58% −33.56% −64.08% −33.25% −64.08% 16.00 18.00 16.00
Max 181.90% 133.25% 181.90% 181.90% 132.42% 181.90% 180.49% 131.72% 180.49% 215.00 200.00 215.00
Stdev 24.95% 20.59% 22.89% 25.90% 21.52% 22.47% 28.89% 22.10% 25.50% 42.84 37.35 40.97

Table 5
Excess-IRR and PME of Private Equity Funds
The excess IRR is defined as a fund’s IRR minus the return on the MSCI Europe equity index that can be achieved by investing at fund closing and selling at
the end of a fund’s lifetime. The PME is the ratio of the present value of all cash distributions over the present value of all take-downs. For non liquidated funds
C 2008 The Authors

Journal compilation
the fraction δ for the residual net asset value is considered as a final cash flow in the IRR- and PME-calculations. The year-by-year realized return on the MSCI
Europe is used as the discount rate. In order to take into account the management fees of a public equity investment, we multiplied the yearly realized index return
by 0.995; i.e. we assumed the management fees of a public equity investment to be equal to 50bp per year.
Excess-IRR PME
δ=1 δ=0 δ=1 δ=0

Liquidated Funds
Average −2.27% 3.37% 0.58% 0.82 0.90 0.86
Median −4.17% −0.77% −2.70% 0.68 0.89 0.80
75th Percentile 1.76% 9.47% 5.21% 0.97 1.24 1.10
25th Percentile −10.84% −7.08% −9.21% 0.33 0.51 0.42
Min −22.24% −24.00% −24.00% 0.07 0.06 0.06
Max 90.99% 84.13% 90.99% 6.97 2.79 6.97
Stdev 17.41% 19.14% 18.42% 1.01 0.53 0.81
Value-weighted 0.94
What Drives Private Equity Returns
Sample I
Average 3.62% 5.29% 4.45% 2.83% 5.13% 3.97% 0.98 0.94 0.96 0.91 0.87 0.89
Median −1.37% 1.57% 0.61% −1.33% 0.51% 0.31% 0.75 0.86 0.82 0.68 0.83 0.75
75th Percentile 5.94% 12.56% 10.24% 5.84% 12.85% 9.18% 1.17 1.24 1.23 1.10 1.17 1.12
25th Percentile −8.02% −6.17% −7.32% −8.61% −6.91% −7.32% 0.40 0.59 0.51 0.39 0.52 0.44
Min −22.24% −24.00% −24.00% −63.48% −27.03% −63.48% 0.01 0.06 0.01 0.00 0.02 0.00
Max 169.35% 84.13% 169.35% 166.93% 78.67% 166.93% 6.97 2.79 6.97 6.97 2.79 6.97
Stdev 24.27% 17.16% 21.01% 25.45% 17.39% 21.82% 1.15 0.51 0.89 1.12 0.49 0.87
Value-weighted 1.04
657

658
C 2008 The Authors

Journal compilation
Table 5
Continued.
Excess-IRR PME
δ=1 δ=0 δ=1 δ=0

Sample II
Average 5.10% 8.25% 6.68% 1.61% 6.34% 3.98% 1.01 1.06 1.03 0.83 0.84 0.83
Median 0.64% 3.53% 1.71% −1.21% 2.51% 0.61% 0.76 0.92 0.85 0.58 0.76 0.68
75th Percentile 8.22% 12.87% 11.23% 6.23% 11.35% 9.04% 1.22 1.35 1.27 0.95 1.12 1.06
25th Percentile −6.99% −5.04% −5.92% −7.91% −6.91% −7.56% 0.44 0.61 0.55 0.32 0.44 0.39
Min −22.24% −20.00% −22.24% −70.91% −39.13% −70.91% 0.01 0.06 0.01 0.00 0.00 0.00
Max 176,00% 127.00% 176.00% 176.00% 127.00% 176.00% 7.27 4.61 7.27 6.97 3.30 6.97
Stdev 25.07% 20.63% 22.96% 29.92% 22.49% 26.52% 1.15 0.70 0.95 1.10 0.63 0.90
Value-weighted 1.16
for why we concentrate on sample I for the following analysis. An additional argument
is the fact that all the performance measures presented here, are higher for sample I,
even with total write-off of the residual net asset value, than for the sample of liquidated
funds.
4.2 Performance, Fund Inflows and Market Sentiment

As explained in Section 2.1, the basic idea of the putative ‘money chasing deals’
phenomenon is a mismatch in the supply of, and demand for, capital in the private
equity industry. From an empirical perspective, the basic problem is that this mismatch
cannot be detected simply by looking at the supply side, i.e. at capital inflows in
the private equity industry. One would have to know the extent to which these
inflows are due to improved economic prospects in the industry, i.e. to an increased
demand for capital, and the extent to which supply exceeds this demand. Ljungqvist,
Richardson, and Wolfenzon (2007) adopt an approach in which they rely on the
assumption that investment opportunities in the private equity industry are captured
by market-to-book-ratios. However, to the extent that valuations on the public stock
market might be correlated with valuations in the private equity segment, it is unclear
whether the market-to-book-ratio is a proxy for investment opportunities or for deal
competition.
Given the difficulty just outlined, we follow a different approach in this paper. In
our view, total fund inflow is, basically, triggered by the economic prospects of the
private equity industry. However, in the short-run there might be a mismatch between
the amount of capital that can be invested in new favourable investment projects and
the amount of money pouring into the industry. It could be argued that the impact
of this mismatch would be greater the more the fund flows are directed towards a
particular part of the private equity industry. For instance, if capital inflows increase
by 10% and, simultaneously, the share of early stage venture funds, later stage venture
funds, buy-out funds, etc., is the same as the year before, this supply shock has to be
borne by the whole private equity industry. In fact, the shock would be more harmful,
if this 10% increase had to be absorbed by a particular fund type, e.g. early stage
funds, alone. Therefore we make a distinction between the total fund inflow into
the private equity industry, as a proxy for future investment opportunities, and the
relative fund inflow, i.e. the allocation between different fund types, as a proxy for deal
competition. 22
In this approach, of course, it is assumed that the private equity industry is not
only segmented from other asset classes, but it is also segmented within itself, at least
to a certain extent. The most important reason for this segmentation is the fact the
LPs normally expect a well-defined investment focus, as defined in the partnership
agreement. 23 . Hence, a buy-out fund manager is not allowed to divert investment funds
22
It could be argued that it might be better to measure demand-supply conditions in
every segment directly by looking, for instance, at the actual investment activities of
each fund in every segment. Unfortunately, company specific investment data is not
available.
23
According to the EVCA reporting guidelines, the fund should clearly disclose
its investment focus; for more details, see http://www.evca.com/images/attachments/
tmpl 9 art 19 att 702.pdf

C 2008 The Authors
Journal compilation
into early stage companies, even if he is aware of significant overpricing in the buy-out
segment. 24
Consequently, in order to test the ‘money chasing deals’ hypothesis, the above-
mentioned variables are defined as follows: The variable Fund type inflow absolute
is the total amount of capital allocated to VC (BO) funds during the vintage year of the
fund. This time series is reported by EVCA for all European private equity funds. The
variable Fund type inflow relative is the ratio of the investment funds allocated to all
VC (BO) funds in a particular vintage year to the total investment funds allocated to
private equity in that year.
A first hint with respect to the soundness of this reasoning may be found by simply
looking at the correlation coefficients. 25 As expected, the Fund type inflow relative-
variable has a significantly positive correlation with the payback period and a negative
correlation with the IRR of the fund. In other words, the higher the ratio of funds
allocated to a particular fund type, the longer it takes to pay back capital to LPs and
the smaller are the returns. Moreover, in such years of overshooting investments capital
commitments to a fund are significantly lower, which is in accordance with the view
that GPs in such periods are worried whether they will be able to dispose of sufficient
valuable investment projects. Further supporting evidence results from looking at the
Fund type inflow absolute-variable. Funds that are closed in years with a high inflow
into that particular fund type have a short payback period and a high return in terms
of IRR; moreover, their committed capital is typically higher as well as their speed of
take-downs.
For a more in-depth analysis of the putative ‘money chasing deals’ phenomenon, a
multivariate regression approach has to be set up. Due to the presence of heteroscedastic-
ity in our data set, a WLS-regression approach was chosen. Data inspection showed that
IRR variance is substantially higher for small funds than for large ones. 26
Regression specifications (1) to (3) in Table 6 reveal that, as predicted, funds raised
in years with high absolute fund inflow generate high returns. This result is almost
unaffected, if the change in the absolute fund inflow is used instead of the absolute
capital inflow, as can be seen from regression specifications (4) to (6) in the same
table. The relative inflow, i.e. the share of new funds that is invested either in VC-
or in BO-funds, has a highly significant negative impact on the fund’s returns, as
predicted. Both effects are not only statistically highly significant, but seem also to
be very robust. Basically, size and significance remains unaffected throughout all the
different regressions specifications that are tested in this paper. Moreover, in regression
equations (2) to (5) of Table 6 we use, as an alternative, the share of newly invested funds
that are allocated to venture funds only. In this case also, the regression parameter is
negative and highly significant. If, instead the share of funds allocated to buy-out funds
only was used as an independent variable, there was no significant effect. 27 Hence, the
change in the fraction of funds allocated to a particular fund type has a clear impact on
24
It should be noted, however, that Cumming, Fleming, and Schwienbacher (2005) present
evidence for a perceivable style drift in private equity funds. Evidently, there is a limited
shift in the investment focus of a private equity fund over its lifetime. This is not harmful to
our approach because we are focusing on the year of fund inception.
25
Because of limited space, we do not report detailed results here.
26
In fact, by applying a Levene-test, the null-hypothesis that residuals have equal variance
has to be rejected on a 5% significance level.
27
For simplicity, these regression results are not reported here.

C 2008 The Authors
Journal compilation

Table 6
WLS Estimation Results on Fund Returns: IRR as the Dependent Variable (Sample I)
C 2008 The Authors

Journal compilation
As dependent variables we use the IRR, as defined in the text. Stage is an ordinal variable where we assigned the following numbers to funds with different stage
focus: Early Stage = 0, Balanced/Diversified = 1, Developed/Late Stage = 2, LBO = 3 and Private Equity = 4. We assume that the risk of an investment policy
of a fund is greater, the lower the value of this variable is. MSCI return p.y. is the annualized return over the lifetime of the fund. MSCI return in VY is the return
in the vintage year. GDP growth rate p.y. is the annualized growth over the lifetime of the fund. IRR of preceding fund, else 0 is the IRR of the preceding fund,
if one exists; otherwise, it is set to zero. Fund type inflow relative is the ratio of the! total funds allocated to VC (BO) funds during the vintage year of the fund,
if it is a VC (BO) fund. VC fund inflow relative is the ratio of the total funds allocated to funds during the vintage year of the fund, if it is a VC fund; else, it is
zero. Fund type inflow absolute are the total funds allocated to VC (BO) funds during the vintage year of the fund, if it is a VC (BO) fund. Absolute difference
of fund type inflow is the difference between the fund type inflow of the Vintage Year and the year before. Timing Take Downs 36 mths is defined according to

the definition proposed by Schmidt, Nowak, and Knigge (2006, p. 12 n.). The variable is supposed to measure the timing ability of the fund manager. The weight
applied to each observation is SizePowerValue . We use ∗∗∗ , ∗∗ , and ∗ to denote significance at the 1%, 5%, and 10% level (two-sided).
Dependent Variable: IRR
(1) VIF (2) VIF (3) VIF (4) VIF (5) VIF (6) VIF
Constant 0.875 0.722 0.697 1.327 1.121 0.948
(0.000) ∗∗∗ (0.000) ∗∗∗ (0.000) ∗∗∗ (0.000) ∗∗∗ (0.000) ∗∗∗ (0.000) ∗∗∗
Stage −0.032 1.930 −0.034 3.409 −0.024 2.001 −0.046 1.858 −0.051 3.280 −0.032 2.091
(0.011) ∗∗ (0.040) ∗∗ (0.039) ∗∗ (0.000) ∗∗∗ (0.004) ∗∗∗ (0.014) ∗∗
MSCI Return p.y. 0.110 1.221 0.113 1.234 −0.057 1.274 0.009 1.156 0.016 1.179 −0.101 1.175
(0.142) (0.143) (0.472) (0.909) (0.846) (0.211)

MSCI Return in VY −0.104 1.151 −0.100 1.152 −0.081 1.159 −0.139 1.167 −0.131 1.166 −0.087 1.209
(0.044) ∗∗ (0.059) ∗ (0.111) (0.009) ∗∗∗ (0016) ∗∗ (0.101)
GDP Growth Rate p.y. −0.231 1.024 −0.227 1.024 −0.154 1.032 −0.319 1.096 −0.312 1.096 −0.208 1.207
(0.000) ∗∗∗ (0.000) ∗∗∗ (0.007) ∗∗∗ (0.000) ∗∗∗ (0.000) ∗∗∗ (0.002) ∗∗∗
IRR of Preceding 0.459 1.027 0.385 1.037 0.511 1.036 0.425 1.032 0.372 1.046 0.505 1.042
Fund, Else 0 (0.028) ∗∗ (0.071) ∗ (0.009) ∗∗∗ (0.044) ∗∗ (0.087) ∗ (0.012) ∗∗
661

662
C 2008 The Authors

Journal compilation
Table 6
Continued.
Dependent Variable: IRR
VC Fund Inflow −0.159 3.263 −0190 3.175

Relative (0.017) ∗∗ (0.007) ∗∗∗
Fund Type Inflow −0.408 1.841 −0.318 2.007 −0.519 1.776 −0.352 2.006
Relative (0.000) ∗∗∗ (0.005) ∗∗∗ (0.000) ∗∗∗ (0.009) ∗∗∗
Fund Type Inflow 0.066 1.129 0.063 1.138 0.037 1.221

Absolute (0.000) ∗∗∗ (0.000) ∗∗∗ (0.003) ∗∗∗
Absolute Difference 0.127 1.108 0.119 1.100 0.058 1.325

of Fund Type Inflow (0.000) ∗∗∗ (0.000) ∗∗∗ (0.052) ∗
Timing Take Downs −0.029 1.068 −0.028 1.071

36mths (0.190) (0.230)
N 190 190 158 182 182 152

Adjusted R2 0.391 0.364 0.166 0.344 0.310 0.143
Significance 0.000 0.000 0.000 0.000 0.000 0.000
Power Value −0.100 −0.100 0.100 −0.200 −0200 0.000

Log-likelihood 95.776 91.775 105.242 86.465 81.940 96.761
venture fund returns, while no impact can be detected for buy-out funds. In our view,
these findings strongly corroborate the ‘money chasing deals’ hypothesis.
It should be noted that we also find the stock market return in the vintage year of
the fund to have a negative impact on its final return. Although no straightforward
explanation can be presented for this result, it could be argued that the variation in
public equity valuations captures an additional factor that drives the overshoot of capital
supply responsible for the ‘money chasing deals’ hypothesis.
In order to test robustness, the estimations presented in Table 6 were redone under four
modifications. First, we used White heteroscedasticity consistent estimators instead of
a WLS-estimation. Second, we used the logarithm of IRR as the dependent variable and
made a logarithmic transformation of those independent variables where it made sense
to do so. Results are reported in Table 7. As one can see, there is neither a change in the
significance, nor a substantial change in the size, of the coefficients. Hence, the results
may be considered to be robust with respect to potential outliers as well as distributional
assumptions.
Third, it may be argued that by using the absolute fund inflow (expressed in terms of
Euros) as the only non normalized variable in the regression, there is potential for a bias
to arise. For this reason, we ran regressions with alternative specifications. For instance,
in both Tables 6 and 8, the alternative Absolute difference of fund type inflow-variable
was introduced, which is the difference between the fund type inflow of the vintage year
and the year before. Moreover, in Table 7, the logarithm of the absolute fund inflow was
used. As one can see, the highly significant impact of the variable measuring the absolute
money inflow is unchanged. We also used the percentage change in the absolute inflow;
the results are unchanged. 28 From these results it can be seen that the positive impact of
the money inflow into the private equity industry is extremely robust.
Fourth, the reader may wonder whether we would still find support for the ‘money
chasing deals’ hypothesis if we were to use a finer grid of fund types. Therefore,
we split up the set of venture capital funds into funds focused on seed, start-up, or
expansion investments. The set of buy-out funds could not be refined further, because no
additional information on investment styles within this fund type had become available.
Recalculating the Fund type inflow relative-variable under this new grid and rerunning
the regression equations presented in Table 6 once again generated results that were
similar in significance and size. 29
As has already been pointed out there is no straightforward performance measure
for illiquid private equity funds. Hence, the possibility that results are driven by the
performance measure used should be ruled out. In fact, Table 8 gives the results for
the same regression specifications used in Table 6, with the important difference that
the excess-IRR is used as the dependent variable. As one can see, most of the results are
unchanged, at least as far as the sign and the significance of the regression parameters
are concerned. 30
28
Results for this last regression are not reported here.
29
Due to limited space, these results are not reported here. It should be noted, however, that
the Fund type inflow relative-variable turned out to be significant at the 5.7%-level. This
loss in significance may be due to the fact that data records on fund inflow for this finer
grid were not directly available, but had to be reconstructed from additional sources. Also
for that reason, the number of observations was reduced perceptibly.
30
The same is true for the regression results using the PME as the dependent variable. For
reasons of space we do not report these results here.

C 2008 The Authors
Journal compilation
Table 7

White Heteroskedasiticity-Consistent Estimation Results: IRR and Ln(1 + IRR) as Dependent Variables (Sample I)
664
Stage is an ordinal variable where we assigned the following numbers to funds with different stage focus: Early Stage = 0, Balanced/Diversified = 1, Developed/Late
Stage = 2, LBO = 3 and Private Equity = 4. We assume that the risk of the investment policy of a fund is greater, the lower the value of this variable is. Here, we
use the absolute value of the variable in all three equations. MSCI return p.y. is the annualized return over the lifetime of the fund. MSCI return in VY is the return
C 2008 The Authors

in the vintage year. GDP growth rate p.y. is the annualized growth over the lifetime of the fund. IRR of preceding fund, else 0 is the IRR of the preceding fund, if
Journal compilation
one exists; otherwise, it is set to zero. Fund type inflow relative is the ratio of the total funds allocated to VC (BO) funds during the vintage year of the fund, if it
is a VC (BO) fund. Fund type inflow absolute are the total funds allocated to VC (BO) funds during the vintage year of the fund, if it is a VC (BO) fund. Absolute
difference of fund type inflow is the difference between the fund type inflow of the Vintage Year and the year before. We use ∗∗∗ , ∗∗ , and ∗ to denote significance
at the 1%, 5%, and 10% levels (two-sided).
(1) (2) (3)

Dependent variable IRR ln(1 + IRR) ln(1 + IRR)

Constant 0,848 0,717 Constant 0,304
∗∗∗ ∗∗∗
0,002 0,000 0,121
Stages −0,031 −0,025 Stages −0,366
∗∗ ∗∗ ∗∗∗
0,017 0,012 0,001
MSCI return p.y. 0,123 0,056 ln(1 + MSCI return p.y.) 0,114
0,121 0,351 0,274
MSCI return in VY −0,976 −0,077 ln(1 + MSCI return in VY) −0,091
∗∗ ∗∗ ∗∗
0,035 0,044 0,047
GDP growth rate p.y. −0,227 −0,180 ln(1 + GDP growth rate p.y.) −0,377
∗∗ ∗∗ ∗∗
0,025 0,012 0,043
IRR of preceding fund, else 0 0,445 0,395 ln(1 + IRR of preceding fund, else 1) 0,371
∗∗ ∗∗ ∗∗
0,021 0,012 0,030
Fund Type inflow relative −0,405 −0,329 ln(Fund Type inflow relative) −0,025
∗∗∗ ∗∗∗ ∗∗∗
0,000 0,000 0,000
Fund Type inflow absolute 0,069 0,044 ln(Fund Type inflow absolute) 0,111
∗∗∗ ∗∗∗ ∗∗∗
0,000 0,000 0,000
N 190 190 190
Adjusted R2 0,418 0,348 0,262
∗∗∗ ∗∗∗ ∗∗∗
Significance 0,000 0,000 0,000
log-likelihood 94,387 137,112 125,266

Table 8
WLS Estimation Results on Fund Returns: Excess-IRR as Dependent Variable (Sample I)
As dependent variable we use the Excess-IRR, as defined in the text. Stage is an ordinal variable where we assigned the following numbers to funds with different
C 2008 The Authors

Journal compilation
stage focus: Early Stage = 0, Balanced/Diversified = 1, Developed/Late Stage = 2, LBO = 3 and Private Equity = 4. We assume that risk of the investment
policy of a fund is greater, the lower the value of this variable is. MSCI return p.y. is the annualized return over the lifetime of the fund. MSCI return in VY is the
return in the vintage year. GDP growth rate p.y. is the annualized growth over the lifetime of the fund. IRR of preceding fund, else 0 is the IRR of the preceding
fund, if one exists; otherwise, it is set to zero. Fund type inflow relative is the ratio of the total funds allocated to VC (BO) funds during the vintage year of the
fund, if it is a VC (BO) fund. VC fund inflow relative is the ratio of the total funds allocated to funds during the vintage year of the fund, if it is a VC fund; else, it
is zero. Fund type inflow absolute are the total funds allocated to VC (BO) funds during the vintage year of the fund, if it is a VC (BO) fund. Absolute difference
of fund type inflow is the difference between the fund type inflow of the Vintage Year and the year before. Timing Take Downs 36 mths is defined according to
the definition proposed by Schmidt, Nowak, and Knigge (2006, p. 12 n.). The variable is supposed to measure the timing ability of the fund manager. The weight

applied to each observation is SizePowerValue . We use ∗∗∗ , ∗∗ , and ∗ to denote significance at the 1%, 5%, and 10% levels (two-sided).
Dependent Variable: Excess-IRR

Constant 0.758 0.612 0.552 1.306 1.102 0.969
(0.000) ∗∗∗ (0.001) ∗∗∗ (0.002) ∗∗∗ (0.000) ∗∗∗ (0.000) ∗∗∗ (0.000) ∗∗∗
Stage −0.027 1.894 −0.031 3.311 −0.023 1.966 −0.048 1.788 −0.057 3.094 −0.038 2.016
(0.027) ∗∗ (0.064) ∗ (0.053) ∗ (0.000) ∗∗∗ (0.001) ∗∗∗ (0.004) ∗∗∗
MSCI Return in VY −0.086 1.051 −0.081 1.050 −0.098 1.069 −0.153 1.053 −0.144 1.049 −0.122 1.113
(0.084) ∗ (0.109) (0.050) ∗∗ (0.003) ∗∗∗ (0006) ∗∗∗ (0.022) ∗∗

(0.000) ∗∗∗ (0.000) ∗∗∗ (0.010) ∗∗∗ (0.000) ∗∗∗ (0.000) ∗∗∗ (0.000) ∗∗∗
IRR of Preceding Fund, 0.432 1.027 0.359 1.037 0.496 1.035 0.383 1.032 0.321 1.045 0.456 1.040
Else 0 (0.040) ∗∗ (0.095) ∗ (0.012) ∗∗ (0.074) ∗ (0.144) (0.027) ∗∗
VC Fund Inflow −0.158 3.211 −0216 3.082
Relative (0.017) ∗∗ (0.002) ∗∗∗
665

666
C 2008 The Authors

Journal compilation
Table 8
Continued.


Fund Type Inflow −0.397 1.831 −0.325 1.986 −0.542 1.759 −0.383 1.997
Relative (0.000) ∗∗∗ (0.005) ∗∗∗ (0.000) ∗∗∗ (0.006) ∗∗∗
Fund Type Inflow 0.079 1.035 0.076 1.055 0.056 1.064
Absolute (0.000) ∗∗∗ (0.000) ∗∗∗ (0.000) ∗∗∗
Absolute Difference 0.159 1.081 0.150 1.079 0.103 1.272
of Fund Type Inflow (0.000) ∗∗∗ (0.000) ∗∗∗ (0.001) ∗∗∗
Timing Take Downs −0.030 1.062 −0.025 1.060
36mths (0.185) (0.298)
N 190 190 158 182 182 152
Adjusted R2 0.470 0.450 0.209 0.416 0.390 0.159
Significance 0.000 0.000 0.000 0.000 0.000 0.000
Power Value −0.100 −0.100 0.100 −0.200 −0200 0.000

Log-likelihood 93.607 90.066 101.789 82.896 78.955 91.383
Finally, a bootstrap regression approach further corroborates the robustness of

the results. 31 For that purpose we resampled the data set randomly by making 200
independent draws with replacement. Then the WLS-regression was recalculated for this
new data set. These steps were repeated 1.000 times. In this way we got a distribution for
all the regression parameters, which allowed us to calculate different confidence levels.
The results for the regression specification with the excess-IRR as independent variable
are reported in Table 9. 32 As one can see, the difference in the regression parameter
estimations is small in size and almost negligible with respect to the significance
levels.
4.3. The Impact of GPs’ Skills and Risk on Fund Performance

As expected, we also found statistically significant evidence for persistence in fund
returns. The results suggest that an increase of 1 percentage point in the IRR of the
preceding fund leads to an increase in the range of 0.4 to 0.7 percentage points in
the IRR of the follow-on fund, as can be seen from Tables 6, 7 and 10. 33 The results
also hold by dropping single-time funds, i.e. funds with no predecessor, from the data
set. The hypothesis that funds’ returns are persistent was tested by using the variable
IRR of preceding fund, else 0, which is the IRR of the preceding fund, if one exists;
otherwise, it is set to zero. These results are robust and statistically significant in our
different regression equations for the IRR or excess-IRR as dependent variable. 34 The
bootstrap approach also provides evidence for the persistence phenomenon, although
the significance of the result is lower there. Finally, it should be noted that persistence
could also be found by using contingency tables. For that purpose, we define every
fund as being a winner if its IRR was above the median, and a loser otherwise. Now,
the empirical probability of a follow-on fund to be classified as a winner or a loser,
contingent on the fact that the preceding fund was a winner or a loser, too, is 63%.
Similarly, the contingent probability that the follow-on fund of a losing fund will realize
an above-median performance is only 18%. Applying an odds-ratio-test inspired by
Brown and Goetzmann (1995), a Z-test inspired by Malkiel (1995) and a χ 2 -test inspired
by Kahn and Ruud (1995), the null hypothesis that the classification of the follow-on
fund does not depend on the classification of the preceding fund can be rejected at a
95% significance level by all three tests. 35
It could be argued that these results may be biased, because it cannot be ruled out
that the vintage year of the follow-on fund is close to the vintage year of the preceding
fund. In that case, the positive return correlation might be caused by similar market
conditions. For that reason, an additional contingency table was constructed, in which
any fund-pair was dropped if the inception of a follow-on fund was within four years of
31
For an introduction to the bootstrap method, see MacKinnon (2002).
32
For reasons of space the results with respect to the IRR and PME-variable are not reported
here. However, for those variables also, the WLS regression results are clearly corroborated
by the bootstrap analysis.
33
Almost the same result holds also for the excess-IRR, as can be seen from Tables 8 and
10.
34
If the PME is used as the dependent variable, however, the persistency variable is no longer
significant.
35
In this regard, see also Kaserer and Diller (2005).

C 2008 The Authors
Journal compilation
Table 9

Bootstrap WLS Estimation Results on Fund Returns: Excess-IRR as Dependent Variable (Sample I)
668
Here we use a bootstrap WLS regression approach with 1’000 random resamplings of the data set. For an introduction to this method, see MacKinnon (2002). The
equations are numbered as in Table 8. As dependent variables we use the Excess-IRR, as defined in the text. Stage is an ordinal variable where we assigned the
following numbers to funds with different stage focus: Early Stage = 0, Balanced/Diversified = 1, Developed/Late Stage = 2, LBO = 3 and Private Equity =
C 2008 The Authors

Journal compilation
4. We assume that the risk of the investment policy of a fund is greater, the lower the value of this variable is. MSCI return p.y. is the annualized return over the
lifetime of the fund. MSCI return in VY is the return in the vintage year. GDP growth rate p.y. is the annualized growth over the lifetime of the fund. IRR of
preceding fund, else 0 is the IRR of the preceding fund, if one exists; otherwise, it is set to zero. Fund type inflow relative is the ratio of the total funds allocated
to VC (BO) funds during the vintage year of the fund, if it is a VC (BO) fund. VC fund inflow relative is the ratio of the total funds allocated to funds during the
vintage year of the fund, if it is a VC fund; else, it is zero. Fund type inflow absolute are the total funds allocated to VC (BO) funds during the vintage year of the
fund, if it is a VC (BO) fund. Absolute difference of fund type inflow is the difference between the fund type inflow of the Vintage Year and the year before. Timing
Take Downs 36 mths is defined according to the definition proposed by Schmidt, Nowak, and Knigge (2006, p. 12 n.). The variable is supposed to measure the
timing ability of the fund manager. The weight applied to each observation is SizePowerValue . We use ∗∗∗ , ∗∗ , and ∗ to denote significance at the 1%, 5%, and 10%

levels (two-sided).

(1) (2) (3) (4) (5) (6)
∗∗∗ ∗∗ ∗∗∗ ∗∗∗ ∗∗∗ ∗∗∗
Constant 0.5367 0.4614 0.6199 0.8923 0.7369 1.0576
∗ ∗ ∗∗
Stage −0.0124 −0.0201 −0.0211 −0.0256 −0.0350 −0.0378
MSCI Return p.y.
∗ ∗∗ ∗ ∗∗∗ ∗∗∗ ∗∗
MSCI Return in VY −0.1044 −0.1072 −0.1179 −0.1724 −0.1610 −0.1617
∗∗∗ ∗∗ ∗∗∗ ∗∗∗ ∗∗∗ ∗∗∗
GDP Growth Rate p.y. −0.1632 −0.1569 −0.1749 −0.2302 −0.2090 −0.2754
∗ ∗ ∗
IRR of Preceding Fund, Else 0 0.4216 0.3691 0.4260 0.3543 0.3366 0.3429
∗∗ ∗∗∗
VC Fund Inflow Relative −0.1416 −0.1747
∗∗∗ ∗∗∗ ∗∗∗ ∗∗∗
Fund Type Inflow Relative −0.2707 −0.3270 −0.3529 −0.4013
∗∗∗ ∗∗∗ ∗∗∗
Fund Type Inflow Absolute 0.0638 0.0557 0.0614
∗ ∗∗
Absolute Difference of Fund Type Inflow 0.0832 0.0622 0.1143
Timing Take Downs 36mths −0.0254 −0.0216
N 190 190 158 182 182 152
# of Bootstrap Samples 1000 1000 1000 1000 1000 1000
Power Value −0.10 −0.10 0.10 −0.20 −0.20 0.00
the vintage year of the preceding fund. Once again, it could be shown that the probability
of a follow-on fund to be classified as a winner or a loser, contingent on the fact that
the preceding fund was a winner or a loser, too, is 67%. Again, the no-persistence null
hypothesis could be rejected at usual significance levels. To sum up, these results fit
nicely into the picture of a sticky and segmented asset class. Furthermore, the results are
close to those provided by Kaplan and Schoar (2005), who find the regression coefficient
of the variable ‘IRR of the proceeding fund’ to be 0.47. Also Phalippou and Gottschalg
(2007) document a significant persistence effect.
As an additional point it should be noted that we do not find market timing abilities
to be responsible for persistent returns. In fact, this is exactly what we would expect for
a fund industry with sticky capital flows. It should be noted here that we use the market
timing ability variable Timing Take Downs 36 mths proposed by Schmidt, Nowak, and
Knigge (2006, p. 12 n.). This variable basically expresses whether or not a GP tends to
call a take-down during a phase of low market valuation. More specifically, we defined
the timing ability variable to be the correlation between the relative market valuation
and the investment activity of the fund. 36 If the correlation is negative, the timing ability
of the fund management is good because the management was able to invest in periods
of low valuations. It should be noted, however, that there is a potential overlap between
this measure of timing ability and the impact of fund inflows, as discussed in the
preceding section. In fact, if a fund were to take-down all the committed capital already
in the inception year, the timing variable would, basically, reflect the correlation of the
fund inflow and the stock market return in the three years around the inception. To some
extent this information is already integrated in the analysis presented in the preceding
section However, in reality take-down is evenly distributed over the first years. For our
sample, on average 23% percent of committed capital is called in the first year, while
60% is called over the three years after inception. 37 Hence, even if fund inflow in the
year of inception and stock market returns in that year were be correlated, the timing
variable might be uncorrelated with both variables. This is exactly what we find in our
data set.
Schmidt, Nowak, and Knigge (2006) found investment timing abilities for 63% of
private equity funds. In our data sample I, 51.9% of funds show a negative correlation,
but the results of our regression equations with the IRR, Excess-IRR and PME as the
dependent variable show no significant timing ability for private equity funds.
Moreover, from Tables 6 and 8, one can see that if the timing variable is included as an
additional variable, the adjusted R2 decreases substantially. The simple explanation may
be that due to missing values for the timing variable, the size of the data set is decreased
36
The approach is similar to that proposed by Schmidt, Nowak, and Knigge (2006). To
identify the relative monthly valuation levels of the MSCI Europe equity index I t at month
t, we calculate a moving average that depends on the absolute index level 18 months (= k)
before and after the behold month. The following
equationdescribes the
moving average
for the 36-month period: M A(2k) = 2k1 · 12 It−k + 12 It+k + τ =t−(k−1) Iτ . To compute the
t+(k−1)
relative market valuation κ of the MSCI index of the current month, we divide the absolute
index level by the MA(36): κ = M A(2k)It
. To calculate the investment activity level η of a
private equity fund, we divide the negative cash flows of each month t by the total sum
of all take-downs (TD) of this fund: η = TT DTt D . Then, the timing ability is defined as
j=0 j
corr(κ, η).
37
Cf. Kaserer and Diller (2004b); a similar result is reported by Ljungqvist and Richardson
(2003).

C 2008 The Authors
Journal compilation
by 32 observations. However, this may raise concerns as to whether results are driven
by a small data subset. In order to tackle this concern, the regressions presented in
Tables 6 and 8 were repeated for the subset of those funds for which values for the
timing variable are available. It turns out that although the adjusted R2 goes down to
roughly 20% in these cases, the size and significance of the coefficients are almost
unaffected.
Finally, we would like to address one important question that, to the best of our
knowledge, has yet to receive an answer in the literature: to what degree can private
equity returns be explained by the stand-alone risk and/or systematic risk of the fund’s
investment policy? As was shown in Section 2.2, results reported in the literature are
quite inconclusive in this regard. To a certain extent, this may be due to the fact that it
is not clear how risk should be measured for an illiquid asset class. Here we propose
two measures of risk. First, there is the investment policy, as defined by the investment
focus in terms of the stages of portfolio companies. This can be regarded as a measure
of the stand-alone risk of a fund. 38 This idea is in line with Cochrane (2005) reporting
that early round financing in venture capital is more risky than later round financing. 39
As can be seen from Tables 6 and 8, we find clear support for the idea that a riskier
investment policy, characterized by the fund stage, leads to higher returns. For that
purpose we defined five different stage categories, three for the venture funds and two
for the buy-out funds. 40 These stages were assigned numbers from 0 to 4, where the
stage with the highest risk, i.e. early stage, was assigned the number 0 and the stage
with the lowest risk, i.e. private equity, was assigned the number 4. 41 As we can see
from Tables 6 and 8, there is a significant difference in returns for the different stages. 42
The expectation of this difference is between 2 and 3 percentage points for every stage
tier. Hence, the expected difference in the IRR between an early stage fund and a private
equity fund is in the range of 8 to 12 percentage points. This result is in accordance
with results presented by Cochrane (2005), who showed that the higher risk of early
financing rounds in venture capital is associated with a higher return. Moreover, this is
in line with the model proposed by Jones and Rhodes-Kropf (2003), according to which
38
Evidently, one can think of more appropriate measures for stand-alone risk of a fund. For
instance, the focus of the fund on a particular country or industry, or the number of portfolio
companies, might be better measures of stand-alone risk. Unfortunately, this information
was not available in the data set.
39
A related theoretical argument can be found in Berk, Green, and Naik (2004), who show
that projects with high technical uncertainty also bear high systematic risk in the context of a
multistage investment decision process. The impact of the systematic risk over the lifetime of
such a multistage project is largest at the beginning. From this one could conclude that early
stage investment do not only bear a higher idiosyncratic risk, but also a higher systematic
risk.
40
The stages for the venture funds are early stage, balanced/diversified, later stage. For the
buy-out funds we have the stages leveraged buy-out and private equity. Stage information
was provided by TVE.
41
Stage is an ordinal variable where we assigned the following numbers to funds with
different stage focus: Early Stage = 0, Balanced/Diversified = 1, Developed/Late Stage =
2, LBO = 3 and Private Equity = 4. We assumed that the risk of the investment policy of a
fund increased according to how low the value of this variable is.
42
It should be noted, however, that the results are only weakly significant in the context of
the bootstrap analysis; see Table 9.

C 2008 The Authors
Journal compilation
venture capital returns should be affected by the amount of idiosyncratic risk, due to
principal-agent problems.
In order to check for the robustness of this result, we used an alternative specification
to measure the impact of the investment focus on the return. The regression results
reported in Table 10 were derived by using a dummy variable for four out of the five
identified investment stages and setting the stage ‘private equity’ as the base case.
According to the definition of the different stages, this last stage can be regarded as the
least risky; hence, we would expect the regression coefficients of the dummy variables
to have a positive sign. In fact, as one can see from Table 10, the coefficients for all stage
dummies are significantly positive. Moreover, the coefficient for the riskiest stage, as
defined by the dummy ES, is higher in all cases than the coefficient on the stage with the
second lowest risk, as defined by the dummy LBO. Hence, the results are in accordance
with the presumption that private equity returns depend positively on the risk exposure
of the investment policy. As a Corollary, it should be noted that all the other coefficients
in Table 10 correspond in size and significance to the formerly reported results.
As a final remark it should be noted that according to the results presented in Tables 6
and 8 private equity returns seem to be negatively correlated with the overall economic
development as measured by the annualized GDP growth rate p.y. over the lifetime of
a fund. 43 Although we cannot go into the details here, it should be mentioned that this
effect might be a simple interest effect, taking into account the fact that interest rates
are correlated positively with GDP growth, but have a negative impact on fund returns.
However, additional tests undertaken with our data do not corroborate this view. Hence,
this result certainly deserves additional attention in future research.
5. Conclusion
We analyzed a comprehensive data set of mature European private equity funds. Our
main focus was to offer new insights into the determinants of funds’ returns. For that pur-
pose, we started from the assumption that the private equity asset class is characterized by
illiquidity, stickiness, and segmentation. It has been argued in theoretical and empirical
papers that these characteristics can cause an over- or undershooting of private equity
asset prices, at least in the short run. Most importantly, Gompers and Lerner (2000) have
shown that venture deal valuations are driven by overall fund inflows into the industry,
which yield the putative ‘money chasing deals’ phenomenon. In addition, Ljungqvist,
Richardson, and Wolfenzon (2007) have shown that the investment behaviour of a GP
depends on fund inflows into the industry.
It was the aim of this paper to document that the ‘money chasing deals’ hypothesis also
explains a significant part of variation in private equity funds’ returns. This is especially
true for venture funds, because they are more affected by illiquidity and segmentation
than buy-out funds. In the context of a WLS-regression model we corroborated the
importance of fund flows for explaining the funds’ returns. Moreover, we were also able
to show that GPs’ skills, as well as the stand-alone investment risk of a fund, have a
significant impact on its returns. Additionally, they seem to be unrelated to stock market
returns and to be correlated negatively with the growth rates of the economy as a whole.
According to a bootstrapping inference the results seem to be quite stable.
43
The result also holds if the PME is used as the dependent variable. Moreover, the result is
also robust to all the different regression specifications that have already been described.

C 2008 The Authors
Journal compilation

672
Table 10
WLS Results with Stage Dummies (Sample I)
C 2008 The Authors

As dependent variable we use the IRR and Excess-IRR, as defined in the text. Stage Dummy ES (B; LS; LBO) is a dummy variable that is 1 if the fund is categorized
Journal compilation
as early stage (balanced; late stage; leveraged buy out) fund. Otherwise it is zero. MSCI return p.y. is the annualized return over the lifetime of the fund. MSCI
return in VY is the return in the vintage year. GDP growth rate p.y. is the annualized growth over the lifetime of the fund. PME of preceding fund, else 0 is the
PME of the preceding fund, if one exists; otherwise, it is set to zero. Fund type inflow relative is the ratio of the total funds allocated to VC (BO) funds during
the vintage year of the fund, if it is a VC (BO) fund. VC fund inflow relative is the ratio of the total funds allocated to funds during the vintage year of the fund,
if it is a VC fund; else, it is zero. Fund type inflow absolute are total funds allocated to VC (BO) funds during the vintage year of the fund, if it is a VC (BO)
fund. Timing Take Downs 36 mths is defined according to the definition proposed by Schmidt, Nowak, and Knigge (2006, p. 12 n.). The variable is supposed to
measure the timing ability of the fund manager. The weight applied to each observation is SizePowerValue . We use ∗∗∗ , ∗∗ , and ∗ to denote significance at the 1%, 5%,

and 10% levels (two-sided).
Dependant Variable IRR Excess-IRR PME
Constant 0.706 0.524 0.601 0.383 2.738 2.652
∗∗∗ ∗∗∗ ∗∗∗ ∗∗ ∗∗∗ ∗∗∗
(0.000) (0.002) (0.000) (0.022) (0.001) (0.000)
Stage Dummy ES 0.147 3.392 0.109 3.469 0.129 3.263 0112 3.342 0.663 3.344 0.501 3.357
∗∗∗ ∗∗ ∗∗ ∗∗ ∗∗ ∗∗
(0.010) (0.043) (0.022) (0.038) (0.013) (0.011)
Stage Dummy B 0.189 2.637 0.116 2.564 0.163 2.604 0.099 2.543 0.817 3.053 0.463 2.630
∗∗∗ ∗∗ ∗∗∗ ∗∗ ∗∗∗ ∗∗
(0.001) (0.019) (0.003) (0.047) (0.001) (0.011)
Stage Dummy LS 0.129 4.104 0.064 4.017 0.111 3.987 0061 3.903 0.767 4.069 0.343 3.891
∗∗ ∗∗ ∗∗∗ ∗
(0.017) (0.201) (0.041) (0.224) (0.003) (0.063)
Stage Dummy LBO 0.095 3.013 0.090 2.924 0.089 2.949 0.091 2.889 0.432 3.717 0.368 3.018
∗∗ ∗∗ ∗∗ ∗∗ ∗∗ ∗∗∗
(0.013) (0.011) (0.019) (0.012) (0.018) (0.005)
MSCI Return p.y. 0.116 1.251 −0.058 1.309
(0.124) (0.467)
MSCI Return in VY −0.095 1.157 −0.072 1.169 −0.077 1.060 −0.090 1.079 −0.264 1.051 −0.211 1.073
∗ ∗
(0.064) (0.155) (0.122) (0.071) (0.202) (0.230)

C 2008 The Authors
Journal compilation
∗∗∗ ∗∗∗ ∗∗∗ ∗∗∗ ∗∗ ∗∗∗
(0.000) (0.008) (0.000) (0.010) (0.029) (0.001)
IRR of Preceding 0.435 1.064 0.446 1.073 0.400 1.064 0.426 1.072 0.086 1.060 0.069 1.056
∗∗ ∗∗ ∗ ∗∗
Fund, Else 0 (0.039) (0.023) (0.060) (0.033) (0.466) (0.483)

Fund Type Inflow −0.435 2.746 −0.224 3.018 −0.406 2.717 −0.229 2.959 −1.981 2.658 −0.823 2.919
∗∗∗ ∗ ∗∗∗ ∗ ∗∗∗ ∗
Relative (0.001) (0.099) (0.003) (0.096) (0.002) (0.100)
Fund Type Inflow 0.064 1.179 0.031 1.306 0.076 1.075 0.051 1.133 0.176 1.070 0.094 1.128
∗∗∗ ∗∗ ∗∗∗ ∗∗∗ ∗∗∗ ∗∗
Absolute (0.000) (0.014) (0.000) (0.000) (0.000) (0.025)
Timing Take Downs −0.028 1.076 −0.030 1.071 −0.079 1.078
36mths (0.196) (0.177) (0.331)
N 190 158 190 158 190 158
Adjusted R2 0.401 0.178 0.476 0.220 0.167 0.124
Significance 0.000 0.000 0.000 0.000 0.000 0.001
Power Value −0.100 0.100 −0.100 0.100 −0.500 0.000
Log-Likelihood 98.893 107.835 96.080 104.316 −199.700 −97.086

673
Of course, some interesting questions had to be left open for future research. Most
importantly, the issue of the degree of idiosyncratic risk as well as market risk inherent
in the private equity asset class needs to be addressed in more detail. Related to this,
the dependency of fund returns on systematic risk also needs to be analyzed further.
Both issues are especially important for making decisions regarding asset allocation for
private equity. The same is true with respect to the issue of the dependence of private
equity returns on the economic development. Our very puzzling result indicating that
there is a negative correlation between these two variables also deserves additional
attention. Our understanding of the economic role of private equity may be influenced
significantly by these results.
References
Berk, J. B., Green, R. C. and Naik V., ‘Valuation and return dynamics of new venture’, Review of
Financial Studies, Vol. 17, 2004, pp. 1–35.
Brown, S. J. and Goetzmann, W. N., ‘Performance persistence’, Journal of Finance, Vol. 50, 1995,
pp. 679–98.
Cochrane, J. H., ‘The risk and return of venture capital’, Journal of Financial Economics, Vol. 75,
2005, pp. 3–52.
Cumming, D. J., Fleming, G. A. and Schwienbacher, A., ‘Style Drift in Private Equity’, SSRN eLibrary,
2005.
Daniel, K., Grinblatt, M., Titman, S. and Wermers, R., ‘Measuring mutual fund performance with
characteristic-based benchmarks’, Journal of Financial and Quantitative Analysis, Vol. 52, 1997,
pp. 1035–58.
Deaves, R., ‘Data-conditioning biases, performance, persistence and ows: The case of canadian equity
funds’, Journal of Banking and Finance, Vol. 28, 2004, pp. 673–94.
Dittmann, I., Maug, E. and Kemper, J. ‘How fundamental are fundamental values? valuation methods
and their impact on the performance of german venture capitalists’, European Financial Manage-
ment, Vol. 10, 2004, pp. 609–38.
Doukas, J. A. and Milonas, N. T., ‘Investor sentiment and the closedend fund puzzle: Out-of-sample
evidence’, European Financial Management, Vol. 10, 2004, pp. 235–66.
Goetzmann, W., Ingersoll, J. and Ivkovich, Z., ‘Monthly measurement of daily timers’, Journal of
Financial and Quantitative Analysis, Vol. 35, 2000, pp. 257–90.
Gompers, P. A. and Lerner, J., ‘Money chasing deals? The impact of fund inows on private equity
valuations’, Journal of Financial Economics, Vol. 55, 2000, pp. 281–25.
Gottschalg, O. and Zipser, D., ‘Money Chasing Deals and Deals Chasing Money - The Impact of
Supply and Demand on Buyout Performance’, SSRN eLibrary, 2006.
Hege, U., Palomino, F. and Schwienbacher, A., ‘Determinants of venture capital performance: Europe
and the united states’, RICAFE Working Paper, 2003.
Hendricks, D., Patel, J. and Zeckhauser, R., ‘Hot hands in mutual funds: Short-run persistence of
relative performance, 1974–1988’, Journal of Finance, Vol. 48, 1993, pp. 93–130.
Henriksson, R. D., ‘Market timing and mutual fund performance: An empirical analysis’, Journal of
Business, Vol. 57, 1984, pp. 73–96.
Inderst, R. and Müller, H. M., ‘The effect of capital market characteristics on the value of start-up
firms’, Journal of Financial Economics, Vol. 72, 2004, pp. 319–56.
Jones, C. M. and Rhodes-Kropf, M., ‘The Price of Diversifiable Risk in Venture Capital and Private
Equity’, SSRN eLibrary, 2003.
Kahn, R. N. and Ruud, A., ‘Does historical performance predict future performance’, Financial
Anlaysts Journal, Vol. 51(November–December), 1995, pp. 43–52.
Kaplan, S. N. and Schoar, A., ‘Private equity performance: Returns, persistence and capital ows’,
Journal of Finance, Vol. 60, 2005, pp. 1791–823.

C 2008 The Authors
Journal compilation
Kaserer, C. and Diller, C., ‘Beyond IRR once more’, Private Equity International (July/August),
2004a.
Kaserer, C. and Diller, C., ‘European Private Equity Funds - A Cash Flow Based Performance Analysis’,
SSRN eLibrary, 2004b.
Kaserer, C. and Diller, C., ‘Private equity funds: Return characteristics, return drivers and consequences
for investors’, Journal of Financial Transformation, Vol. 15(December), 2005, pp. 107–17.
Lerner, J., ‘Boom and Bust in the Venture Capital Industry and the Impact on Innovation’, Federal
Reserve Bank of Atlanta Economic Review, Vol. 87(4), 2002, pp. 25–40.
Lin, C. Y. and Yung, K., ‘Real estate mutual funds: Performance and persistence’, Journal of Real
Estate Research, Vol. 26, 2004, pp. 70–93.
Ljungqvist, A. and Richardson, M. P., ‘The Cash Flow, Return and Risk Characteristics of Private
Equity’, SSRN eLibrary, 2003.
Ljungqvist, A., Richardson, M. P. and Wolfenzon, D., ‘The Investment Behavior of Buyout Funds:
Theory and Evidence’, SSRN eLibrary, 2007.
MacKinnon, J. G., ‘Bootstrap inference in econometrics’, Canadian Journal of Economics, Vol. 35,
2002, pp. 615–45.
Madura, J. and Bers, M. K., ‘The performance persistence of foreign closed-end funds’, Review of
Financial Economics, Vol. 11, 2002, pp. 263–85.
Malkiel, B. G., ‘Returns from investing in equity mutual funds 1971 to 1991’, Journal of Finance,
Vol. 50, 1995, pp. 549–72.
Miller, R. M., ‘Measuring the True Cost of Active Management by Mutual Funds’, SSRN eLibrary,
2005.
Otten, R. and Bams, D., ‘The performance of local versus foreign mutual fund managers’, European
Financial Management, Vol. 13, 2007, forthcoming.
Phalippou, L. and Gottschalg, O., ‘Performance of Private Equity Funds’, Review of Financial Studies,
2007, forthcoming.
Schmidt, D., Nowak, E. and Knigge, A., ‘Private equity funds: Return characteristics, return drivers
and consequences for investors’, Journal of Financial Transformation, Vol. 16(April), 2006,
pp. 123–34.
Zimmermann, H., Bilo, S., Christophers, H. and Degosciu, M., ‘The risk and return of publicly traded
private equity’, WWZ/Department of Finance, Working Paper No. 6/04, 2004.

C 2008 The Authors
Journal compilation

Diller Kaserer PE Return Drivers PDF

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Diller Kaserer PE Return Drivers PDF

Hochgeladen von

Copyright:

Verfügbare Formate

European Financial Management, Vol. 15, No.

What Drives Private Equity Returns? –

Keywords: Private equity funds, performance measurement, venture capital,

2. Related Literature and Theoretical Considerations

2.1. The ‘Money Chasing Deals’ Phenomenon

of liquidity on secondary markets. Finally, as Ljungqvist, Richardson, and Wolfenzon

2.2. The Impact of Fund Characteristics on Performance

this yields the following condition:

Finally, inverting this inequality once again and setting q = c1 c2

4.1. Return Distribution of European Private Equity Funds

Type of Funds All Venture Capital Funds VC BO

Panel B: Characteristics according to liquidation status

All Liquidated funds Nonliquidated funds

Panel C: Characteristics according to subsamples

Liquidated Funds Sample I Sample II

C 2008 The Authors

C 2008 Blackwell Publishing Ltd

Distribution of IRR(δ) and Payback Period for Different Fund Types

C 2008 The Authors

VC BO Total VC BO Total VC BO Total VC BO Total

C 2008 Blackwell Publishing Ltd

C 2008 The Authors

C 2008 Blackwell Publishing Ltd

C 2008 The Authors

C 2008 Blackwell Publishing Ltd

4.2 Performance, Fund Inflows and Market Sentiment

C 2008 The Authors

C 2008 Blackwell Publishing Ltd

(0.142) (0.143) (0.472) (0.909) (0.846) (0.211)

C 2008 The Authors

C 2008 Blackwell Publishing Ltd

Fund Type Inflow 0.066 1.129 0.063 1.138 0.037 1.221

Absolute Difference 0.127 1.108 0.119 1.100 0.058 1.325

Timing Take Downs −0.029 1.068 −0.028 1.071

N 190 190 158 182 182 152

Power Value −0.100 −0.100 0.100 −0.200 −0200 0.000

C 2008 The Authors

(1) (2) (3)

C 2008 Blackwell Publishing Ltd

C 2008 The Authors

C 2008 Blackwell Publishing Ltd

Dependent Variable: Excess-IRR

(0.084) ∗ (0.109) (0.050) ∗∗ (0.003) ∗∗∗ (0006) ∗∗∗ (0.022) ∗∗

C 2008 The Authors

Dependent Variable: Excess-IRR

C 2008 Blackwell Publishing Ltd

Power Value −0.100 −0.100 0.100 −0.200 −0200 0.000

Finally, a bootstrap regression approach further corroborates the robustness of

4.3. The Impact of GPs’ Skills and Risk on Fund Performance

C 2008 The Authors

C 2008 Blackwell Publishing Ltd

Dependent Variable: Excess-IRR

C 2008 The Authors

C 2008 Blackwell Publishing Ltd

Dependant Variable IRR Excess-IRR PME

C 2008 Blackwell Publishing Ltd

Log-Likelihood 98.893 107.835 96.080 104.316 −199.700 −97.086

Das könnte Ihnen auch gefallen