1
Performance of European Buyout Investments:
Does Timing Matter?
Maarten van Delft
Name: Maarten van Delft Student number: 1571281 MSc Business Administration Specialization: Finance University of Groningen
2
Table of Contents
1. Introduction ... 4
2. Literature Review ... 6
3. Methodology & Data ... 10
Sample description ... 12
Total sample descriptive statistics ... 12
Statistical insights dataset ... 15
4. Results ... 18
Base Regression ... 19
Timing analysis ... 19
Vintage year analysis ... 20
Exit year analysis ... 22
Economic situation analysis ... 24
Conclusion ... 25
References ... 27
Appendix A: Tables ... 29
3
Performance of European Buyout Investments:
Does Timing Matter?
Maarten van Delft
1Abstract
In this study a sample of 1,814 unique European buyout deals shines new light on performance of European buyout investments. The first contribution of this paper is to provide new descriptive statistics on the performance of European buyout investments. Additionally, a cross-sectional analysis on a new dataset gives new insights on buyout performance. The third aim of this study is to identify the timing aspect of buyout investments versus public markets. An ordinary least squares (OLS) regression is performed were several meaningful variables pop out as drivers for individual buyout investment performance. Market return (+) and duration (-) turn out as significant drivers for investment performance. Furthermore I found a large dispersion of performance between vintage and exit years. Lastly evidence is presented in favor of the money chasing hypothesis; deals initiated in public market boom periods underperform and deals initiated in recession show outperformance.
Keywords: private equity, buyout investment performance JEL-classification: G24, G34
1
Master student at the University of Groningen, Faculty of Economics and Business, Department of Economics, Econometrics and Finance (e-mail for correspondence m.j.b.@student.rug.nl). This paper is written in
fulfillment of the dissertation requirement for the degree of Master of Science in Business Administration,
4
1. Introduction
Private Equity (PE) gained a great deal of interest in recent years. The EVCA2 estimated that since 2007 PE firms have raised EUR 233 billion of funds3. The growing importance of PE as an asset class has drawn the attention of investment professionals worldwide. Besides practitioners, scholars also gained interest in this relatively new asset class. This interest is enforced by the current turmoil in public markets. In the aftermath of the financial crisis (2007-2010) public market securities dropped heavily. Consequently coverage ratios4 of institutional investors (e.g. endowments and pension funds) fell behind. Therefore, from a portfolio construction perspective, institutional investors are in search of alternative asset classes to make up for the losses on the public markets. Although several scholars found that PE funds are pro-cyclical in general (Robinson and Sensoy, 2011; Kaserer and Diller, 2004) there is reason to assume that during some economical periods the performance deviates from the general trend. For
instance Kaplan and Schoar (2005) found that PE funds initiated during economic booms show
underperformance. These findings all rest on fund level studies. Since PE funds invest over several years, research based on fund level vintage years might muddle results. Unfortunately, PE deal level data is not widely available. These data limitations are a consequence of the private nature of the PE asset class which exempts it from disclosure regulations that apply to public market securities. Because of these data availability issues there still is limited understanding of risks, returns and timing of cash flows in PE. This paper aims at filling that gap and investigates the performance of buyout5 investments. Subsequently since the relationship between market returns and buyout performance is not fully understood yet I aim at explaining differences in performance from a timing perspective. Following Kaplan and Strömberg (2009) and Robinson and Sensoy (2011) I investigate buyout performance with respect to public market returns. They have found that periods in which public markets generate high returns PE investments tend perform worse. In spite of these significant results all these authors cover market timing from a fund perspective which may muddle the exact timing. Since individual investments within funds are executed over multiple years there could be a substantial difference in public market based valuation multiples at the time of investment or divestment. The discussed difficulties on data availability are countered with the help of a large global investor in PE (hereafter The Investor). The aggregation level of the dataset in this study however allows for to taking a closer look to these findings. The Investor enables me to counter this dataset issue by providing investment level data. In finding out whether the outcomes on timing regarding
2 European Private Equity and Venture Capital Association 3
EVCA Yearbook 2012
4 In this case, a measure of a pension funds ability to meet its financial obligations.
5 Literature states different labels to buyout transactions (e.g., management buyouts, levered buyouts, management
5 public markets on above mentioned authors also apply to deal specific buyout investments the main question this study is searching to answer is:
1. What is the impact of public market returns on individual European buyouts investments? To operationalize this question I approach timing from three different perspectives. The first is the vintage year in which an investment is done. Besides vintage, exit timing could be of importance when addressing performance. The last timing aspect I am investigating is the public market circumstance during a buyout investment. More specifically I try to find answers to the questions:
- Do vintage years have an impact on individual buyout investment performance? - Do exit years have an impact on buyout investment performance?
- Does public market performance during an investment have impact on individual buyout performance?
The first contribution is to provide new descriptive statistics on European buyout investments. As mentioned before, data availability is one of the main drawbacks in PE research. With the help of The Investor in PE, the data availability issue is countered. The hand-collected dataset existing of 1814 buyout investments in the period 1985-2010 sheds new light on performance of this asset class in Europe. A great dispersion of performance is measured, almost twenty percent of all the investments generate a loss (gross of management fees) and the twenty-five percent of the total investments generate a return over 50% IRR6 gross of management fees. The duration of these investments matches as in the samples of Kaplan and Strömberg (2009) and Lopez-de-Silanes et al. (2010) is relatively long-lived. Median (mean) duration in this sample is 3.75 (4.15) years. It is also found that long –lived investments are certainly not the high performing investments. For this sample a clear negative relation of performance with duration is
measured. Deals with a duration of less than two year (the so-called quick flips), which make up over 16% of the sample, generate a median (mean) return of 74.90% (86.6%) gross of management fees.
The second contribution of this paper is the identification of different performance drivers behind the variation of individual buyout investments. Due to the mentioned scarce data availability, literature until the late 2000s has focused on aggregate performance over time (Kaplan and Strömberg, 2009) or
aggregated in funds (Kaplan and Schoar, 2005; Diller and Kaserer, 2007). The deal-specific nature of the data set used in this paper enables me to go deeper into specific deal performance. I find that small investments perform relatively better than large investments; the relationship between investment size and performance is negative. Furthermore public market performance is a significant driver for individual
6 The Internal Rate of Return (IRR) is an important financial performance measure; see Appendix A Table I for an
6 buyout investments. Building on the latter, a more thorough analysis is performed on the relation of public market returns and PE investments. I find a clear dispersion of performance between different vintage and exit years which indicates that markets are not efficient in a sense of the efficient market hypothesis (Fama 1970). More specifically, vintage year analysis produces results in favor of the money chasing hypothesis of Gompers and Lerner (1999) who found that in cold markets PE investments are undervalued. Exit year analysis on the other hand generates contradicting results; I found that the bad performing public market years 2000 and 2001 generated good buyout exits. Going deeper in the concept of timing I finally investigate buyout investments during different economic cycles. Investments during stable or moderately growing public markets do not under- or outperform. However, investments during recessions are less sensitive to public market returns and show higher performance. Furthermore,
confirming expectations buyout investments during public market booms show underperformance and are more sensitive to changes in public market returns.
In the second section a short background of PE is presented. The third section focuses on explaining the characteristics of the data set and used methodology. The fourth section discusses the results of an Ordinary Least Squares (OLS) regression. In the last section conclusions will be drawn and suggestions for further research will be done.
2. Literature Review
Private Equity (PE) mainly exists out of two types of investments: venture capital (VC) and (levered) buyouts (BO). Although types of target companies in VC and BO may differ both types are often studied together. Reason for this is that organizational structure is similar (fee structure, illiquidity characteristics etc.). Furthermore investors are able to play an important management role in PE, which is uncommon in public investments. The majority of PE investments are structured as limited partnerships which have a finite life (usually ten years, often extended to as long as fourteen years). Institutional investors make up the largest part of PE investors. Investors like pension funds, endowments and banks7 typically have long term visions that match long term illiquid investments. These investors are called Limited Partners (hereafter: LPs) who commit a certain amount of capital to a private equity fund. These funds are
managed by General Partners (hereafter: GPs). Usually GPs identify investment opportunities in portfolio companies. GPs then draw capital from the LPs to acquire a portfolio company (so-called „capital calls‟. Often the investment period is approximately 5 years. When a portfolio company is sold (liquidated) the GPs distribute the invested capital back to the LPs, either in kind or in cash. A GPs‟ compensation is based on a general management fee and a performance fee called „carried interest‟ or „carry‟.
7 In the period 2007-2011, in Europe, 26.6% of the sourcing in PE came from pension funds, 15.5% from
7 Management fees usually are 1.5 to 2.5% per year over the invested capital. The performance fee is typically 20% of the capital gain that exceeds a hurdle rate of 8% (Phalippou, 2008; Metrick and Yasuda, 2010).
The question whether changes in public market returns are influencing buyout performance is inherently linked to a fundamental theory on the market efficiency. The Efficient Market Hypothesis (Fama, 1970) states that prices of securities fully reflect all available information on a security. If demand is perfectly elastic, exogenous changes in supply and demand should not lead to changes in a securities price. More specifically, if the market for buyouts investments is perfectly efficient then changes in vintage, exit year or public market returns should not have an impact on the performance of buyout investments. However, due to the private nature of PE investments and the lack of a free and open market place PE asset class is characterized by a great deal of illiquidity. This raises questions whether markets are efficient at all. Gompers and Lerner (1999) set the first steps in addressing performance from a market return
perspective. They have found that capital inflows to venture markets increase when public markets are going up. Although they leave it open whether this money chasing phenomenon implicates that funds rose in times of high capital in-streams perform worse, Gompers and Lerner (1999) are the first to address timing in a private equity environment. Strongly building upon the findings of Gompers and Lerner (1999) are Diller and Kaserer (2009). They find evidence that indeed funds raised in times of high capital inflows perform worse than funds raised in times of low fund inflows. Additionally they find that PE fund returns are correlated negatively with stock market returns which recognized the earlier finding of high performance when markets are cold and low performance when markets are hot. In contrast, Kaplan and Schoar (2005) found that PE funds move pro-cyclical. Kaplan and Schoar wrote a seminal article on relative performance of the PE funds. They were the first to combine workable datasets with appropriate methodology. Kaplan and Schoar are the first to use the Public Market Equivalent method to compare PE returns to the public market, which nowadays is a common benchmark measure of PE performance. The PME method mimics private equity investments as if they were done in the public market. The actual return multiple then is divided by the mimicked public benchmark multiple to find outperformance (PME>1) or underperformance (PME<1) of a particular fund or individual deal regarding the public benchmark8. This adjustment of methodology is necessary since PE companies are not traded daily as S&P500 companies are. Therefore instead of time-weighted return (TWR) measures private equity uses measures as IRR and money multiple. For purchase and sale of a single stock during a given period of time IRR and TWR are the same. However IRR is very sensitive to both timing and interim cash flows so PME becomes a necessary measure when investments and divestments become irregular. Although PME seems to solve the problem of comparison it does not go completely without weaknesses since it gives a
8 relative performance9. Phalippou and Gottschalg (2009) argue that earlier mentioned authors are
overstating PE performance due to the survivorship’ bias. The survivorship bias puts forward the error that can rise when survivors are composing the largest part of the sample. In this case, GPs that perform well will every year show their good results where bad performing GPs disappear and do not submit their performance figures. Furthermore Phalippou and Gottschalg discuss that it is difficult to value unrealized deals and possibly these deals are overstated. When corrected for these errors Phalippou and Gottschalg find that PE funds are underperforming the public market with three per cent per year net of management fees. When Phalippou and Gottschalg compare returns gross of management fees they find an
outperformance of three per cent per year. The above finding suggests that GPs as intermediaries are able to generate higher returns however due to their high intermediary compensation this outperformance is not paid out to the LP. Finally I will discuss the findings of Robinson and Sensoy (2011). Robinson and Sensoy find evidence in favor of the findings of Kaplan and Schoar (2005); both performance and cash flows are highly correlated over time with public market returns. Consequently buyout funds initiated in hot markets are underperforming in absolute terms (IRR).
Table I
Synopsis on private equity research on fund and deal-level performance
This table provides performance outcomes of recent researches on performance of private equity. The first section describes recent papers on deal specific datasets. The second section presents performance figures of three seminal articles on fund performance.
Authors Region Period # of Deals/Funds Money Multiple average/median Gross IRR average/median Fund Level
Kaplan and Schoar (2005) World 1980-2001 746 1.83* 18%* Diller and Kaserer (2009) Europe 1980-2003 200 1.64* 13.4%* Robinson and Sensoy (2011) World 1984-2010 990 n/a 12% / 10% Phalippou and Gottschalg (2009) World 1980-2003 852 2.11** 15.5%**
Investment level
Achleitner et al. (2010) Europe 1991-2005 206 3.5 / 2.8 43% / 33% Acharaya et al. (2011) Europe 1995-2005 395 4.4 / 3.0 56% / 43% Lopez-de-Silanes et al. (2010) World 1985-2005 7453 n/a / 1.9 n/a / 21% Kaserer (2011) Europe 1997-2010 327 2.93 / 2.36 40% / 29% * Average gross performance figures of only buyout investments.
** Average gross performance figures of buyout and venture investments.
The last decade, when data availability became less of a problem, scholars were able to dive deeper in the concept of PE performance since investment level datasets were provided. Lopez-de-Silanes, Phalippou
9 F.i. if the benchmark loses 50% of its value in a period of time and a private investment only uses 25% the PME
9 and Gottschalg (2010) were the first to collect a dataset of unique private equity deals worldwide. Lopez-de-Silanes et al. have added to existing literature in two ways. First they provide a series of new facts and statistics of private equity worldwide on the biggest existing deal-specific dataset of 7500 deals.
Drawbacks of Lopez-de-Silanes et al. are that they also consider unrealized deals. Furthermore in spite of the richness of control variables Lopez-de-Silanes have gathered, they do not go any further in cross sectional analysis than descriptive statistics. Kaserer (2010) focuses his research on European buyout investments. Kaserer goes further than Lopez-de-Silanes et al. in explaining performance of individual deals. Kaserer has found that two-third of investment IRRs are contributable to earnings enhancement of a GP that buys a company. Almost a third of a company‟s return increase can be contributed to leveraging of the company. However, Kaserer makes use of a rather small sample of only 332 individual deals of only 18 different investors. Achleitner, Braun, Engel, Figge and Tappeiner (2010) improve on existing value creation literature in several ways. First, they do not make use of an Anglo-Saxon focused dataset. Second, Achleitner et al. create a comprehensible framework of value creation for individual buyout deals. The authors analyze three ways in which GPs could create value in a buyout. By using financial engineering GPs can improve a business cash and capital claims and distributions. Operational
engineering is the optimization of real business processes (f.i. supply chain management). And finally a part of the value creation of a deal is created by, as they call it, market effects. These market effects contain for example timing effects that are more difficult for GPs to affect. Acharaya, Gottschalg, Hahn and Cehoe (2011) conduct the fourth important research on investment level. Researching 395 Western European private equity transactions from 1991 to 2007 they find average (median) gross IRRs of 56% (43%) in their dataset. After assessing general return characteristics Acharaya et al. mainly focus on operational engineering. They found that GPs with operational backgrounds (ex-consultants or ex
industry-managers) are able to generate significantly higher outperformance in deals that focus on internal value creation. GPs with financial backgrounds are able to generate higher outperformance in deals with significant M&A events.
Having discussed the most important articles on timing and performance of individual buyout investments we now arrive at hypothesis development. From the literature we have found certain relationships
10 Table II
Variable relationships
In this table a short overview of all variables and their impact on deal performance is presented Variable +/- Why?
Market Return +/-
Kaplan and Schoar (2005) and Lopez-de Silanes et al (2010) found a positive relationship on fund level between market return and fund performance. Kaserer and Diller (2009) found a negative relationship between market return and fund performance.
Investment Size +/-
Acharaya et al (2011), Achleitner et al. (2011) and Ljungqvist and Richardson (2003) found a positive relation between deal size and performance. Lopez-de-Silanes et al. find a negative relation.
DumVintage/DumExit +/-
For firms initiated (exited) in economic booms a negative (positive) sign is expected. Vice versa for deals initiated (exited) in economic downturns a positive (negative) sign is expected (Gompers and Lerner, 2000; Kaserer and Diller 2009 and Kaplan and Strömberg, 2009)
Dum Market Condition +/-
In economic downturns performance is expected to significantly higher. In economic booms the performance is expected to be significantly lower (Robinson and Sensoy, 2011)
In short, I am investigating the relationship between public market returns and buyout investments. I use investment size as a control variable. Since from literature it follows that timing of investments is an important performance driver I make an assessment of the different timing features of an individual investment: the vintage year, the exit year and the public market circumstances during an investment. The three hypotheses are:
H0: The vintage year has an impact on the performance of individual buyout investments
H1: The exit year has an impact on the performance of individual buyout investments
H2: The public market state during an investment has an impact on the performance of individual buyout investments
3. Methodology & Data
Some of the shortcomings of performance measurement already have been mentioned in the previous section. Nevertheless I briefly present the main points were improvement in PE research can be made. First, data on unlisted private equity is not publicly available. Researchers have to find their way in delving into private equity using different databases of different data provides10. Different companies use different assumptions and definitions and so datasets are not always as reliable as researchers‟ state.
10 Companies such as Preqin, Burgiss‟Capital IQ and Thomson Reuters Banker ONE do provide databases. Problem
11 These dataset issues leave room for scholars to create hand-collected controllable datasets and perform the discussed methodology on investment level datasets. Second, investments are done over several years and cannot be liquidated easily. Scholars therefore used intermediary deal valuations to assess the
performance of these unrealized deals. Phalippou and Gottschalg (2009) found that these intermediate valuations could be over or understated and research based on unrealized investments is likely to be biased. Third, especially deal specific data is not widely available. Many scholars have concentrated their efforts on deal specific performance analysis using the best, yet small, data sets available. Besides Lopez-de-Silanes et al (2010) there is no European research that exceeds 400 deals in a multi-year period (In 2010 alone, which is considered a „bad‟ year, European private equity deals exceeded 80011
). Henceforth there is room for new research based on larger datasets.
In order to find the effect of public market returns on private equity deals this study uses an Ordinary Least Squares (OLS) regression to measure the impact. Using different dummy variables for, timing cross-sectional analysis will be executed to get to a better understanding of non-operational return drivers of individual buyouts. The first variable that is tested is market return12. Earlier research produced ambiguous results. Kaplan and Schoar (2005) found a positive relationship between market return and PE fund performance. Diller and Kaserer however found a negative relationship between fund performance and a public benchmark. Since Lopez-de-Silanes et al (2010) have found a positive impact of public market returns and buyout investments, I expect the same positive relationship for my dataset. In line with Kaserer (2011) the logarithm of the investment value is added as control variable to the equation.
Although Lopez-de-Silanes et al. (2010) have found a negative relationship between investment size and performance my expectation is (in line with Kaserer (2010)) that the relationship is positive. The reason for this assumption is the significantly smaller investment sizes in my dataset13. The last part of the OLS regression will be taken out to test the timing aspect of individual buyout investments. I use different timing dummies to clarify different aspects and variations of timing of investments. The first dummy variable that is tested is vintage year. The vintage year is the year in which the investment is initiated. The second dummy variable I introduce is the exit year, this is the year in which the investment is sold or written off by the GP. With these two variables I am able to point out whether buyout investments in certain years are performing better or worse regarding the public markets. The last variable is a dummy variable for the market circumstance during the investments. Although PE investments are perceived to correlate positively with public markets in general, Robinson and Sensoy (2011) point out that there is
11 Thomson Reuters Bankers ONE 12
12 variation in performance during different economic cycles. Taken together the OLS regression estimate will be:
In this equation α is the constant variable. Performance is the performance variable. Although I cross check different gross and net performance measures as PME, excess IRR and money multiples this paper will focus on the gross IRR as performance measure mainly due to its importance in the industry and the high correlation between different measures14. MARKETRETURN is the return a similar investment in the MSCI Europe would have returned. logINVESTMENTSIZE is a control variable of the logarithm of the investment size in dollars15. Duration is a control variable for the duration in years. dumVINTAGE is a binary variable for the entry year of a deal, dumEXIT is a binary variable for the exit year of a deal. The dummy variables dumMARKETCIRCUMSTANCES represents a binary variable of different market circumstances during the entire life of an investment. is the error term. IRR, multiple and PME are
used as performance variables, these performance measures will be further explained in the following section.
In short, the impact of public market returns upon buyout performance is measured. From literature control variable investment size is added. With dummy variables an assessment is made regarding timing of investments in different vintage years, exit years and different public market circumstances.
Sample description
Total sample descriptive statistics
This research investigates the performance of 1814 individual European buyout investments provided The Investor. Since this dataset is accumulated by a single, albeit large, investor the representativeness of the dataset is a natural concern. In spite of this possible drawback this dataset improves on current academic literature in several ways. First, the used dataset of individual buyout investments enables this study to go beyond fund level performance. The lack of individual deal data is a drawback in the commonly used datasets of commercial data providers such as Preqin, Burgiss‟ Capital IQ or Thomson Banker ONE (formerly known as Thomson Venture Economics or VE). Especially, the latter has been used by many
14 In Appendix A, Table IV the correlation matrix is presented
15 Because the deals go back to as far as 1985 deals in PPMs are presented in different pre-Euro currencies. I have
13 seminal articles (f.i. Kaplan & Schoar, 2005; Phalippou & Gottschalg, 2009) on performance of private equity. Second, the dataset is accumulated with the help of a large global private equity investor who collects data directly from the source, the General Partners‟ Private Placement Memorandums. This method of data collection avoids reporting and survivorship biases in this dataset which are common in commercial databases. (Stücke, 2011; Harris, Jenkinson and Stücke, 2010). Third, large data providers depend in their data collection on the willingness of GPs to send in their performance data. Stücke (2011) provides evidence for the case that a serious downward bias exists in these large datasets due to deficient cash flow updates and the use of faulty carried-forward interim Net Asset Values. This study evades these pitfalls by only using fully realized investments in the analysis. Finally, The Investor also keeps track of performance of GPs in which they do not invest in. It therefore goes beyond the studies of Kaserer (2011) and Acharaya et al. (2011) and follows the data gathering method of Lopez-de-Silanes (2010) which seems to measure the European buyout universe to a more realistic extend. One major disadvantage of the dataset used in this research is that only total value of paid-in money and paid-out money is available, intermediate cash flows are not provided in the larger part of the PPMs. This prevents me to cross-check the IRRs which are presented in the PPMs; I am restricted to GP provided IRRs. A second shortcoming of the provided data is that intermediate cash flows are not available. Therefore the assumption has to be made that all the invested capital is invested at the entry date and the total capital is paid out at the exit date. This prevents this research to exactly analyze the timing effect of cash flows. Additionally the data collection method used offers several advantages in investigating buyout investments. It, however, should be mentioned that this collection method is not free of selection biases. The Investor focuses on small and mid-market investments and aims for top tier PE funds, therefore the performance of investments could be biased. Moreover, since the total population of European buyout investments is not available, it is impossible to perform robustness checks regarding the selection bias.
The main sources of data are PPMs of fund raising GPs. Furthermore, presentations and digital track records are also taken into consideration. This method of obtaining data directly from the GP has the major advantage that all the data can be checked from the source. After gathering all data sources 2651 deals remain in the initial sample. Subsequently non-buyout deals, performance outliers and deals with incomplete data points are filtered out, which leaves 1814 unique buyout investments16.
Since the dataset is constructed mainly from PPMs it differs from the largest part of academic papers on performance. This is due to the fact that it contains information on individual investments. Granting that not all PPMs come in the same format, the PPMs usually contain historical per investment track records which usually at least present information on: entry date, exit date, investment cost, investment proceeds,
14 company industry and company location, Gross IRR and Gross Multiple. Detailed information on all the variables used in this paper is presented in Table I of Appendix A. Table III provides the descriptive statistics of the most important variables. This table shows that the average (median) duration of European buyout deals is 4.15 (3.75) years. This is largely in line with results that Kaserer (2011, WP) has found in a comparable sample of 327 identical buyouts. The median of investment cost in this study is $ 7.97 million. This is substantially smaller than in Lopez-de-Silanes et al. (2010) who found a median of $ 13.0 million for non-US developed country buyouts. The reason for this large difference with the results of the study of Lopez-de Silanes et al (2010) is the focus on small and midmarket buyouts and the
underrepresentation of large and mega buyouts in this sample. The variable MSCI Market Return
describes the annualized return of a passive investment identical in timing and size of a particular deal in the sample which has been done in the MSCI Europe index.
Table III
Descriptive Statistics
This table shows the descriptive statistics of the total sample of realized European buyout investments from 1985 to 2010. The statistics shown in this table are representing the total sample of 1820 observations. In Appendix A, Table II the descriptive statistics per cross-section can be found.
Panel A. Descriptive Statistics of total sample
Mean Median Maximum Minimum Std. Dev.
Vintage 1998 1998 2010 1985 4.63 Exit Year 2002 2002 2011 1987 4.7 Duration 4.15 3.75 14.10 0.09 2.25 Investment Size 21.96 7.97 440.46 0.04 40.11 Investment Proceeds 63.98 17.10 1875.34 0.00 148.82 Market Return 11.37% 12.13% 48.78% -47.36% 11.15% Gross IRR 36.10% 28.67% 498.89% -100.00% 69.33% Gross Multiple 2.95 2.27 53.00 0.00 3.44 Gross PME 2.07 1.57 32.93 0.00 2.44 Gross Alpha 24.73% 18.20% 497.58% -127.77% 67.78% Net IRR 21.98% 17.13% 490.55% -100.00% 58.44% Net Multiple 2.25 1.79 40.89 0.00 2.46 Net PME 1.59 1.25 25.41 0.00 1.79 NET Alpha 10.49% 6.40% 453.20% -127.72% 56.49%
15 2005; Robinson and Sensoy, 2011). This performance however, comes at the cost of a substantially higher risk shown in the difference of standard deviations (11.15% for public investments and 69.33% for private investments). The multiple represents the time which is needed to earn back the investment costs after full realization of a deal (the times the investment cost are earned back after full realization of a deal). In this sample the mean (median) multiple of a European buyout is 2.95 (2.27) which is largely the same result as Kaserer (2011) who found a mean (median) of 2.93 (2.36). The PME (Public Market Equivalent) is a measure perfected by Kaplan and Schoar (2005). This measure is used to benchmark private deals to comparable investments in the public market. It simply mimics the capital calls and pay-outs of a private investment as if they took place in the public market place. The multiple of the private investment is then divided by the mimicked public deal. A PME lower than 1
implicates that the private investment has underperformed the public market. A PME higher than 1 indicates the outperformance of public to private investments17. The mean (median) deal PME gross of fees in this sample is 2.07 (1.57). This implicates that an average European buyout deal gross of fees would have outperformed an investment in the MSCI Europe Index more than 2 times. The three (gross of management fees) performance measures (IRR, Multiple and PME) are also presented net of management fees. Since many GPs are reticent in providing fee structures it is important to mention that these net performance measures are based on an estimate of a common fee structure18. As Table 2 shows average (mean) yearly return on European buyouts is 21.98% (17.13%). Buyout investments will on average (median) return 2.25x (1.79x) the invested capital. Benchmarked to comparable public investments on average (median) an investment in a buyout will outperform the market 1.59 (1.25) times. The annual concrete outperformance net of fees or “alpha” is 10.49% (6.40%).
Statistical insights dataset
This particular data collection provides a wealth of cross-sectional variables. Therefore, it is meaningful to take a closer look to cross-sectional statistics. This following section will discuss important statistics per cross section to provide a better understanding of how performance of buyouts evolves through different dimensions.
An important cross section for private equity investors is the entry year or investment of vintage year. Although time frame of investments is 1985-2010, the largest part (93.2%) of the investments has an
17
Kaplan and Schoar (2005) where the first to benchmark private to public deals using PME-method in a large Thomson Reuters dataset. Although the setbacks of this dataset are discussed in Stücke (2011) the PME method now is the best practice in used in academic papers to compare private to public investments.
18
16 initial investment date in the years between 1990 and 2005. In the first place it is interesting to take a look at the entry and exit dates of the investments. Figure 1 presents the distribution of entries and exits. Interestingly, the number of exits peaked in 2000 during the internet bubble and in the period 2005-2007 just before the financial crisis. Exits are somewhat more dispersed over the dataset; 82.49% of the exits took place between 1996 and 2007.
Figure 1: Distribution of number of entries and exits
In order to corroborate the presumptions that the database studied is indeed existing of mainly midmarket buyout deals the distribution of investment cost is presented in figure 2. As can be derived from this figure the 56% of the investments are not exceeding 10 million USD19. The latter finding confirms the postulation that the dataset is comprised of mainly small and midmarket investments. It furthermore is important to emphasize that this study takes a closer look investment size; this graph does not describe company size.
19
Although this research describes performance of European buyouts the chosen currency is US dollars. This is due to the fact that for deals before 4 January 1999 could have been calculated to a synthetic Euro exchange rate for the years 1985-1998. I have chosen to calculate the investment cost on the base of a real currency available throughout the entire timespan; the US dollar. Cost is the only variable based on USD, all performance figures are calculated on the base of the real currency preventing the inclusion of a dollar exchange rate risk.
17
Figure 2: Distribution of investment value (in million $) and median gross IRRs per subset at entry
The duration of deals that this study finds is largely in line with former authors who have investigated European buyout performance. The average (median) holding period for this dataset is 4.12 (3.72) which is in line with Lopes-de Silanes et al. (2010) who have found a median of 3.75 for non-US developed country buyouts. Kaserer (2011), finds a slightly higher average of 4.7 the median of 3.8 is closer to this database outcome. Figure 3 provides a closer look on duration distribution. Besides the fact that there is an interesting decrease in the number of deals while duration increases, it is also interesting to see a clear decrease in performance when deals take longer to exit.
As far as industry distribution concerns I find a clear focus in Industrial investments as can be seen in figure 420. The sectors Financials and Information Technology appear to outperform comparing to the other sectors. A reason for the outperformance of the tech sector could be the ability of GPs to adequately time the technology wave of the late 90s and early 2000s.
20 When PPMs did not provide the company industry, company/sector matches were made using Thomson Reuters
Banker ONE database which includes the largest part of European buyout deals. To classify companies consistently the Standard and Poor‟s Global Industry Classification Standard (GICS) is used in this study.
1.79 2.53 2.75 2.54 2.10 2.20 1.96 1.39 1.67 1.04 1.94 0.00 0.50 1.00 1.50 2.00 2.50 3.00 0 50 100 150 200 250 300 350 400 <1 year 1-2 2-3 3-4 4-5 5-6 6-7 7-8 8-9 9-10 10-15 # of investments IRR
18 Lastly, I present a geographical cross-section in figure 5. The graph immediately shows immediately follows from the graph that the UK deals are representing almost half (47%) of the deals. This is largely in line with Lopez-de-Silanes et al. (2010) who found 40% of the non-US deals to be originated in the UK. Interestingly, in this sample the Nordic countries (Sweden, Finland, Denmark and Norway) show a remarkable high performance comparing to the other countries.
4. Results
This section describes the results found in the OLS regression. First, the impact of the public market return, investment size and duration on performance are described using OLS regression technique. Secondly, the variables investment vintage year and exit year and market circumstances are discussed with the use of dummy variables.
17% 54% 38% 17% 45% 26% 28% 40% 29% 27% 0% 10% 20% 30% 40% 50% 60% 0 100 200 300 400 500 600 700 800 900 1000 # of investments Median 24% 40% 43% 21% 33% 29% 42% 39% 39% 28% 29% 27% 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% 0 100 200 300 400 500 600 700 800 900 1000 # of investments Gross IRR Figure 4 Distribution of investments and gross median IRRs per GICS sector
19
Base Regression
Table IV the results of the base regression are presented. In the first column the positive and significant relation between the market return and buyout performance is presented. The second column shows the expected positive relation between investment size and buyout performance. In the last column the estimates of the linear multivariate regression are presented. The intuition of the last column is that, in general, when public markets go up with 100% returns on IRRs of buyout investments go up 132.75%. Secondly, when investment size increases with one million dollar the IRR is expected to increase with 7.00%. This finding contradicts with the results of Lopez-de-Silanez et al (2010) who found a negative coefficient for the logarithm of size. One explanation for this different finding is that average investment size in that study is almost twice as large as the investment size in this study. In general these findings are in line with earlier studies of Kaserer (2010), Achleitner et al. (2011) and Acharaya et al. (2011).
Duration is used in Kaserer (2011) as a control variable. Although it could be important control variable it is excluded in the IRR regression due to the fact that it is not an endogenous variable. The relationship of duration and performance is therefore tested using another performance measure which does not rely on deals‟ duration, the multiple. In line with expectations the OLS produces the same outcome for the multiple as for the IRR; there is a strong negative relationship between duration of an investment and its performance. OLS estimates with multiple as dependent variable are showed in Table IV panel B.
Table IV
Base Regression
This table shows the results of Ordinary Least Squares analysis. The first table shows the results for the gross
IRR and the second table provides regression results for net performance figures. In the columns 1-3 the estimates
are presented for the simple linear regression. In the last column the estimates of the base multivariate analysis is presented. Coefficients are presented in the first line. Standard errors are presented in parentheses.
Panel A 1 2 Both
Market Return 1.3158 *** 1.3275 ***
(0.1428) (9.3085)
Log Investment Size 0.0629 ** 0.070029 ***
(0.0262) (0.0256) Constant 0.2114 *** 0.3074 *** .15044 *** (9.3013) (0.0276) (0.0318) Adj. R² 0.044 0.012 0.048 Number of Investments 1841 1841 1841
*** significant at 1%, ** significant at 5%, * significant at 10%
Timing analysis
20 for buyout performance during different market circumstances will be discussed. Furthermore, all
following regression estimates use the Gross IRR as dependent variable because it is the most commonly used performance figure in private equity and it correlates highly with the other performance figures.21
Vintage year analysis
In Table V the regression estimates for the vintage year dummies are presented. This is the first step in assessing market timing of GPs with respect to the public market. Although the total dataset consists of vintages from 1985 to 2010 only vintages in the sample with more than 75 observations are analyzed. With these results I can observe the differences in performance throughout different vintage years of
21
1992 1993 1994 1995 1996 1997 1998
Ind. Variables Coeff. P-Value Coeff. P-Value Coeff. P-Value Coeff. P-Value Coeff. P-Value Coeff. P-Value Coeff. P-Value Market Return 1.6848 0.0000 1.6789 0.0000 1.7311 0.0000 1.7382 0.0000 1.6709 0.0000 1.5903 0.0000 1.6709 0.0000
0.1238 *** 0.1238 *** 0.1273 *** 0.1254 *** 0.1291 *** 0.1247 *** 0.1237 ***
Log investment size 0.1527 0.0000 0.1510 0.0000 0.1544 0.0000 0.1566 0.0000 0.1582 0.0000 0.1560 0.0000 0.1605 0.0000
0.0184 *** 0.0183 *** 0.0184 *** 0.0184 *** 0.0185 *** 0.0186 *** 0.0191 *** Intersect Dummy 0.7698 0.0033 1.4190 0.0000 -0.1610 0.4321 -0.2552 0.1181 -0.1821 0.1467 -0.1058 0.2449 -0.0628 0.2884 0.2617 *** 0.2877 *** 0.2049 0.1632 0.1254 0.0909 0.0592 Slope Dummy -4.6731 0.0067 -7.7803 0.0000 0.3721 0.7311 0.5539 0.5431 0.7549 0.2006 1.6994 0.0051 -0.1107 0.8901 1.7227 *** 1.7311 *** 1.0825 0.9106 0.5897 0.6056 *** 0.8015 Adjusted R² 0.0400 0.0486 0.0366 0.0388 0.0365 0.0411 0.0361 Observations 1814 1814 1814 1814 1814 1814 1814 1999 2000 2001 2002 2003 2004 2005
Ind. Variables Coeff. P-Value Coeff. P-Value Coeff. P-Value Coeff. P-Value Coeff. P-Value Coeff. P-Value Coeff. P-Value Market Return 1.7876 0.0000 1.8630 0.0000 1.7645 0.0000 1.7451 0.0000 1.6873 0.0000 1.6238 0.0000 1.5718 0.0000
0.1260 *** 0.1258 *** 0.1221 *** 0.1242 *** 0.1258 *** 0.1232 *** 0.1231 ***
Log investment size 0.1383 0.0000 0.1338 0.0000 0.1455 0.0000 0.1566 0.0000 0.1544 0.0000 0.1517 0.0000 0.1470 0.0000
0.0196 *** 0.0194 *** 0.0187 *** 0.0186 *** 0.0185 *** 0.0186 *** 0.0185 *** Intersect Dummy 0.0815 0.1676 0.0262 0.6945 0.3691 0.0001 0.2104 0.3369 0.0811 0.7461 -0.2483 0.2570 0.0435 0.7623 0.0591 0.0667 0.0966 *** 0.2190 0.2504 0.2190 0.1437 Slope Dummy -2.1668 0.0053 -3.9495 0.0000 -4.4558 0.0000 -1.7410 0.0979 -0.3361 0.7536 2.2246 0.0422 1.9282 0.0072 0.7759 *** 0.6647 *** 0.9342 *** 1.0513 * 1.0705 1.0945 ** 0.7172 *** Adjusted R² 0.0409 0.0552 0.0475 0.0384 0.0354 0.0398 0.0501 Observations 1814 1814 1814 1814 1814 1814 1814 Table V
Vintage Year Dummies
This table presents the results for the regression analysis where all vintage intersect dummies for the years 1992-2005 are regressed singular with the base IRR regression without an constant.
*** significant at 1%, ** significant at 5%, * significant at 10%
*** significant at 1%, ** significant at 5%, * significant at 10%
22 individual deals. I perform three vintage year dummy analyses: only using intersect dummies, only using slope dummies and adding both intersect as slope dummy to the regression. Intersect dummy analysis will point out whether there is reason to assume under or outperformance of buyout investments in certain vintage years. The slope dummy measures the sensitivity to changes in market return in a vintage year. Constants are left out of the equation in all presented regressions to be able to compare performance between vintages without the impact of a constant. The estimates for only intersect and slope dummy regressions are presented in Appendix A Table III-IX. Here I will pay special attention to the regressions in which both intersect as well as slope dummies are included since these regressions seem to model the performance of buyout investments best due to the interaction effect22. From Table IV it can be read that for the only four years in the sample where MSCI Europe returns were negative (1992, 2000, 2001 and 2002)23 I found positive intersect dummies and negative slope dummies. In 1992 and 2001 both dummy variables were significant enforcing expectations that in bad public market years buyout investments are performing better and are less likely to have a positive relation to public market returns. In spite of these findings there still is too little evidence to accept the hypothesis that in years with badly performing public markets buyout investments show more outperformance.
Exit year analysis
In this section identical dummy regression analysis as for vintage years is performed for exit years. The results for exit year OLS estimations are presented in Table VI. As with vintage year dummies I will here focus on the OLS regressions where both intersect as well as slope dummies are included. Estimates for only intersect and only slope dummy regressions are added in Appendix A, Table IV. When both dummy variables are added I found contradicting results to the assumption that in good public market years exits are performing well. For exit years 2000, 2001 and 2002 (in which public markets generated negative results). I found positive intersect dummies and negative slope dummies. This suggests that buyout exits in negative MSCI Europe return years are showing outperformance to buyouts made in years with non-negative MSCI Europe years. The contradicting results for this dummy analysis in 2006 and 2007 (good MSCI Europe years) point out that this method might not be optimal to address exit year performance in different exit years in comparison to public market returns. A second explanation for this result that exits in downward economic cycles are profiting from the gains made in the years before the downfall. One argument for this latter explanation is that it takes to 2003 for exit intersect dummies to turn negative which could imply a lagged performance downfall for exits. In short, there is evidence for significant out or underperformance as well as difference in sensitivity of buyout investments to public market
fluctuations.
23
1996 1997 1998 1999 2000 2001
Ind. Variables Coeff. P-Value Coeff. P-Value Coeff. P-Value Coeff. P-Value Coeff. P-Value Coeff. P-Value Market Return 1.6742 0.0000 1.6454 0.0000 1.7833 0.0000 1.6845 0.0000 1.7310 0.0000 1.8284 0.0000
0.1234 *** 0.1261 *** 0.1353 *** 0.1256 ** 0.1244 *** 0.1243 ***
Log investment size 0.1552 0.0000 0.1562 0.0000 0.1505 0.0000 0.1543 0.0000 0.1517 0.0000 0.1353 0.0000 0.0184 *** 0.0184 *** 0.0186 *** 0.0184 *** 0.0185 *** 0.0189 *** Intersect Dummy -0.1857 0.6547 -0.5495 0.0619 -0.0570 0.7897 0.4296 0.1343 0.3397 0.0210 0.1562 0.0211 0.4151 0.2942 * 0.2136 0.2868 0.1471 ** 0.0677 ** Slope Dummy 1.5098 0.5913 2.9632 0.0367 -0.2195 0.7927 -2.2356 0.1516 -2.4550 0.0099 -2.9497 0.0000 2.8114 1.4173 * 0.8350 1.5584 0.9504 *** 0.6614 *** Adjusted R² 0.0356 0.0379 0.0369 0.0365 0.0389 0.0483 Observations 1814 1814 1814 1814 1814 1814
*** significant at 1%, ** significant at 5%, * significant at 10%
2002 2003 2004 2005 2006 2007
Ind. Variables Coeff. P-Value Coeff. P-Value Coeff. P-Value Coeff. P-Value Coeff. P-Value Coeff. P-Value Market Return 1.8432 0.0000 1.7284 0.0000 1.6779 0.0000 1.6620 0.0000 1.6115 0.0000 1.6789 0.0000
0.1250 *** 0.1248 *** 0.1239 *** 0.1242 *** 0.1248 *** 0.1238 *** Log investment size 0.1322 0.0000 0.1498 0.0000 0.1549 0.0000 0.1628 0.0000 0.1581 0.0000 0.1510 0.0000
0.0190 *** 0.0192 *** 0.0191 *** 0.0190 *** 0.0187 *** 0.0183 *** Intersect Dummy 0.0510 0.5065 -0.0283 0.6635 -0.0060 0.9361 -0.1472 0.0852 -0.3361 0.0042 1.4190 0.0000 0.0767 0.0651 0.0751 0.0855 * 0.1174 *** 0.2877 *** Slope Dummy -3.5582 0.0000 -2.3152 0.0192 0.2410 0.7273 0.7073 0.2286 2.1721 0.0008 -7.7803 0.0000 0.8033 *** 0.9881 *** 0.6909 0.5872 0.6499 *** 1.7311 *** Adjusted R² 0.0491 0.0383 0.0354 0.0369 0.0413 0.0486 Observations 1814 1814 1814 1814 1814 1814
*** significant at 1%, ** significant at 5%, * significant at 10%
Table VI Continued Exit year dummies
Table VI Exit year dummies
24
Economic situation analysis
Since exit dummy analysis in the previous section showed results contradicting with theory, I assess timing from another perspective. In this section dummy variables are applied for different economic circumstances. In spite the fact that public market returns in general are positively related to buyout investment performance following Gompers and Lerner (1999) and Kaplan and Schoar (2005) there is reason to assume that in different market circumstances this relation is enforced or attenuated. Here I apply different dummy variables for the public market performance during a given buyout investment. When markets go up with more than 20% during a buyout investment an economic boom dummy is appointed. When markets rise between 10% and 20% a moderate growth dummy is designated. In the case markets rise 0% to 10% a stable market dummy is appointed and if markets generate a negative return a recession dummy is appointed. This method enables me to show different relationships between buyout investment performance and different economic circumstances. Again, here I only present the outcomes for the intersect and slope dummy regression estimates. As can be read from Table VII, the results for a stable and moderate growth market are not significant. However, for investments during economic boom periods the results are significant. In column 1 the results for economic boom periods are shown. I find a negative and significant intersect dummy and a positive and significant slope dummy. This finding has the implication that, in general, boom period initiated buyout investments are
underperforming. Furthermore the sensitivity to public market returns is stronger pointing out a stronger rise in performance of buyout investments when the MSCI Europe increases in boom periods. For downward public markets, in line with expectations, I find the contrary. Buyout deals initiated in
economic recession perform better in comparison with the total sample. More specifically the positive and significant intersect dummy shows us that in the base case buyout investments during economic down cycles perform better. The negative and significant slope dummy point out that during recession periods buyout investment performance is less sensitive to changes in public markets.
Ind. Variables Coeff. P-Value Coeff. P-Value Coeff. P-Value Coeff. P-Value
Market Return 1.3177 0.0000 1.6179 0.0000 1.6588 0.0000 2.2933 0.0000
0.1718 *** 0.1358 *** 0.1232 *** 0.1391 ***
Log investment size 0.1585 0.0000 0.1536 0.0000 0.1742 0.0000 0.0539 0.0137
0.0184 *** 0.0185 *** 0.0203 *** 0.0218 ** Intersect Dummy -0.3254 0.0776 -0.0882 0.5620 -0.0853 0.2200 0.1708 0.0091 0.1843 * 0.1520 0.0695 0.0654 *** Slope Dummy 1.8536 0.0122 0.7907 0.4233 0.2144 0.8417 -2.9964 0.0000 0.7389 ** 0.9873 1.0738 0.6612 *** Adjusted R² 0.0416 0.0361 0.0380 0.0731 Observations 1814 1814 1814 1814
*** significant at 1%, ** significant at 5%, * significant at 10%
Boom Mod. Growth Stable Recession
Table VII
Market Circumstance Dummies
25
Conclusion
In this paper I investigate the performance of European buyout investments. Private Equity has
experienced an enormous growth over the recent decade. In spite of growing commitments to this asset class academic work is scarce. Some research on PE performance is available, especially on fund level. Due to the private nature of this asset class, however, zooming in at an investment level is more difficult. PE investors generally are reticent in providing performance data and especially in the case of deal specific investments. With the help of a large investor in European PE I am able tackle at least some of the data availability issues that are inextricably linked to PE research.
With a hand-collected database of 1814 unique pan European buyout investments this paper tries to address buyout investment performance from a timing perspective. The first contribution of this paper is that it provides descriptive statistics of an ill-researched asset class. I find that European buyouts gross of fees produce a median IRR, gross of management fees, of 28.67%. Furthermore I found an estimated excess IRR net of management fees (alpha) of 6.40%. Average duration of investments is 4.15 years. The median investment size for this sample is $7.97 million. In addition I provide insights on the buyout industry with cross-sectional statistics. The largest part of the sample investments are in industrial, IT and consumer goods companies. The UK has by far the largest buyout industry in this sample; almost 45% of the included deals in the sample are located in the UK.
In the second part of this paper I build on earlier efforts of Kaplan and Schoar (2005) and Robinson and Sensoy (2011) who researched PE from a timing perspective. I found a great dispersion of sensitivity and performance of buyout deals regarding public market during different vintage and exit years. I therefore conclude that the first two hypotheses are accepted; both vintage and exit year have an impact on buyout investment performance. For vintage years the assumption seems to hold that bad public market years produce good PE vintages. The sample contains four years in which public markets generate negative returns, in those vintage years buyout investments showed absolute higher IRRs. For high return public market years however no significant result was measured.
The same analysis is performed for exit years. I do not find evidence for the assumption that in well performing public market years exits are higher as well. In contrary I found significant results higher performance in the dotcom bubble years 2000 and 2001. An explanation for this result might be that M&A exit multiples are lagged. In other words it takes some time after an economic downfall for multiples to adjust to the economic situation.
26 outperformance. In line with expectations I however do find reason to assume that deals during economic recessions show outperformance regarding non-recession deals. Correspondingly, recession deals are less sensitive to changes in public market returns. For economic boom period buyout investments the opposite results are found. Buyout investments in thriving markets are underperforming non-boom investments and they are more sensitive to changes in public market returns. Therefore I accept also the last
hypothesis; public market circumstances have impact on the performance of buyout investments. These results are in line with the money chasing hypothesis which states that hot markets generate bad PE vintages. This paper also found evidence against the efficient market hypothesis, the variation in performance between different vintage and exit years and market circumstances show us that buyout investment markets are not efficient.
To conclude this study some limitations are discussed. Due to the private nature of Private Equity data availability is limited. Although this paper is able to describe PE performance on an investment level some drawbacks have to be taken into account. First, a selection bias which cannot be controlled for is quite possibly at stake since only investments available to The Investor are included in the sample. The Investor is a large investor in European PE and does also collect data on funds they do not participate in. In spite of this the investor also focuses on small and midmarket funds and aims to invest in top tier PE funds. It therefore is possible that well performing investments are over represented in this sample. A second limitation related to data availability is the lack of intermediate cash flows. Although buyout investments are characterized by large initial capital calls and large eventual distributions I am not able to take specific timing into account. Before drawing conclusions on the findings of any paper on PE, these limitations should always be held into account.
27
References
Acharaya, V. V., O. Gottschalg, M.Hahn, and C.Kehoe, 2011. “Corporate Governance and Value Creation: Evidence from Private Equity”. Working Paper Available at SSRN: http://
ssrn.com/abstract=1324016.
Diller, C, and C. Kaserer, 2009, What Drives Private Equity Returns?? Fund Inflows, Skilled GPs, and/or Risk?, European Financial Management 15, 643-675.
Fama EF. Efficient capital markets: a review of theory and empirical work. Journal of Finance 1970; 25; 383-417.
Gompers, P., and J. Lerner, 2000, Money Chasing Deals? The Impact of Fund Inflows on Private Equity Valuations, Journal of Financial Economics 55, 281–325.
Kaplan, S.N., and A. Schoar, 2005, Private Equity Performance: Returns, Persistence, and Capital Flows, Journal of Finance 60, 1791–1823.
Kaplan, S. N.,B.A. Sensoy, and P.J.Strömberg, 2002, „How well do venture capital databases reflect actual investments?‟, Working Paper (University of Chicago)
Kaplan, Steven, and Per Strömberg, 2009, Leveraged buyouts and private equity, Journal of Economic Perspectives 23, 121-146.
Kaserer, C. 2011, “Return attribution in Mid-Market Buy-Out Transactions: New Evidence from Europe”. Working Paper available at SSRN:http://ssrn.com/abstract=1946110
Lopez –de-Silanes, F., L. Phalippou and O. Gottschalg, 2010, ”Giants at the Gate: Diseconomies of Scale in Private Equity”. Working Paper.
Kaserer, C., and C. Diller, 2009, What Drives Private Equity Returns? – Fund Inflows, Skilled GPs, and/or Risk?, European Financial Management 15, 643-675.
Ljungqvist, A., and M. Richardson, 2003, The Cash Flow, Return, and Risk Characteristics of Private Equity, NBER Working Paper 9494.
28 Phalippou L., 2010, Risk and Return of Private Equity: An Overview of Data, Methods and Results, In D.J. Cumming, ed.: Private Equity: Fund Types, Risks and Returns, and Regulation, chapter 12, 257-282, John Wiley & Sons, New Jersey, USA.
Robinson, David T., and Berk A. Sensoy, 2011, Private Equity in the 21st Century: Liquidity, Cash Flows and Performance from 1984-2010, Working Paper available at SSRN:http://ssrn.com/abstract= 1890777
29
Appendix A: Tables
Table I
Variable descriptions
This table contains the descriptions of all important variables used in this research.
Variable Variable Description
PE Firm A private equity (PE) firm is an organization that undertakes investments in private markets. This research focuses on buyout deals which are only a part of the total world of private equity. Henceforth other deals which are considered private equity investments like venture capital, distressed financing, land, real estate, mezzanine infrastructure or timber are omitted.
PE Fund A private equity (PE) fund is (in this research) a buyout investment fund which is managed by the PE firm (or General Partner, GP). Firms can manage several funds over time. The existence of PE funds is finite, usually funds have a duration of ten years which can be extended by one year periods after ten to fourteen years.
Investment
In this research an investment can be seen as a buyout investment by a PE firm in a certain company. All add-on investments and divestments are included. Debt en public equity investments are excluded
Vintage year The year of initiation of an private equity investment.
Exit year The year a private equity deal is exited.
Duration The duration is the length of the investment given in years. It is calculated by subtracting the entry date from the exit date. The entry and exit dates are provided by fund PPMs and PE firm due diligence files and presentations. Investments in the data sources which did not contain exit dates are omitted from the dataset. Doing so this research only contains of investments which are fully realized.
IRR The internal rate of return, gross of fees, of the investment. In 81.1% of the cases, the IRR which is provided by the PE firm is used. This PE firm IRR is based on cash flow data which is not always provided in PPMs and due diligence material. In 18.9% of the cases the IRR was not given by the PE firm. There are some reasons to omit the IRRs from the PPMs. One reason is that it is not meaningful (f.i. very short positive investments tend to have excessive IRRs which give a biased view on the investment performance). Another reason of this omission can be the weak performance of an investment and so forth a negative IRR which in some PPMs is not shown. In those cases I have calculated the IRRs using the duration and multiple data. Since duration of invested capital is usually shorter than duration of the entire investment the IRRs I have calculated will in most cases be more conservative than the real cash flow based IRRs. IRRs can be presented gross and net of fees.
30 Table I (continued)
Variable Variable Description
PME The public market equivalent (PME) is a easy to interpret figure which shows the performance of a private investment to a public investment. A PME > 1 means outperformance of the public markets and a PME < 1 means underperformance of the private investment to the public market. To calculate this measure it is assumed that all investments are made at t=0 and all distributions are made on the exit date. This multiple than is divided by the multiple of an comparable investment which is made in the public market with exact the same timing. PMEs can be presented both gross and net of fees.
Bankrupt A classification of all the investments that have reported PPM returns of 0.
Home Run A classification of the investments which have a reported PPM IRR > 50%.
Quick Flip A classification of deals which have a a duration < two years.
31 Table II
Cross-Sectional Variable Descriptive Statistics
This table shows the statistics of performance for different cross-sections. Panel A presents the statistics for vintage year. Panel B shows the statistics for exit
year. In panel C the descriptive statistics for market circumstance are shown.
Finally in Panel D the statistics for different countries are presented.
Panel A
Mean Median Max Min.
32 Table II (continued)
Panel B
Mean Median Max Min. Std. Dev. Obs.
1987 1.92 1.92 1.92 1.92 NA 1 1988 1.08 0.92 2.72 -0.31 1.06 8 1989 0.46 0.50 1.13 -0.13 0.37 8 1990 1.06 0.49 4.84 0.00 1.39 12 1991 0.37 0.28 1.44 -0.66 0.76 7 1992 0.59 0.46 2.03 -0.74 0.69 19 1993 0.58 0.54 3.02 -1.00 0.83 33 1994 0.62 0.48 4.92 -0.61 0.93 39 1995 0.50 0.48 4.05 -1.00 0.90 55 1996 0.35 0.32 3.51 -1.00 0.63 73 1997 0.48 0.39 4.99 -1.00 0.85 92 1998 0.41 0.29 3.43 -1.00 0.66 131 1999 0.43 0.33 4.27 -1.00 0.74 110 2000 0.36 0.26 3.87 -1.00 0.72 150 2001 0.26 0.27 2.78 -1.00 0.68 105 2002 0.22 0.17 2.25 -1.00 0.57 97 2003 0.10 0.16 4.02 -1.00 0.59 119 2004 0.27 0.29 1.58 -1.00 0.49 136 2005 0.30 0.22 2.69 -1.00 0.64 169 2006 0.45 0.37 3.15 -1.00 0.60 163 2007 0.46 0.35 3.61 -1.00 0.67 151 2008 0.34 0.33 2.36 -1.00 0.65 47 2009 0.06 0.07 1.82 -1.00 0.71 24 2010 0.19 0.24 1.73 -1.00 0.63 46 2011 0.10 0.21 1.10 -1.00 0.61 19 All 0.36 0.29 4.99 -1.00 0.69 1814 Table II (continued) Panel C
Mean Median Max Min. Std. Dev. Obs.
Boom 0.62 0.51 4.99 -1.00 0.77 416
Moderate Growth 0.39 0.32 4.84 -1.00 0.65 597
Stable 0.17 0.15 3.02 -1.00 0.54 535
Recession 0.27 0.25 4.92 -1.00 0.80 266
33 Table II (continued)
Panel D
Mean Median Max Min. Std. Dev. Obs.
UK 0.29 0.27 4.99 -1.00 0.64 861 France 0.34 0.29 3.43 -1.00 0.55 313 Germany 0.52 0.28 4.02 -1.00 0.93 155 Italy 0.48 0.36 4.84 -1.00 0.85 109 Netherlands 0.45 0.39 2.59 -1.00 0.65 80 Sweden 0.62 0.41 4.05 -1.00 0.78 65 Spain 0.31 0.29 2.25 -1.00 0.45 64 Finland 0.38 0.33 2.81 -1.00 0.73 38 Poland 0.29 0.21 3.87 -1.00 0.80 36 Switzerland 0.51 0.43 2.36 -1.00 0.59 24 Denmark 1.06 0.40 4.27 -0.68 1.56 15 Belgium 0.19 0.14 0.90 -0.65 0.38 12 Norway 0.88 0.72 2.15 -0.58 1.02 7 Austria 0.09 0.23 0.47 -0.79 0.45 6 Czech Republic 0.44 0.28 1.21 0.11 0.44 5 Ireland 0.09 0.13 0.28 -0.12 0.15 5 Portugal 0.07 0.08 0.13 -0.02 0.07 4 Romania 0.49 0.44 0.76 0.30 0.20 4 Hungary 0.03 0.19 0.43 -0.54 0.50 3 Bulgaria 0.92 0.92 0.92 0.91 0.01 2 Luxembourg 0.94 0.94 1.22 0.65 0.40 2 Slovenia 0.22 0.22 0.30 0.13 0.12 2 Serbia 0.31 0.31 0.31 0.31 NA 1 Turkey 0.33 0.33 0.33 0.33 NA 1 All 0.36 0.29 4.99 -1.00 0.69 1814
IRR PME Multiple
Initial sample 2651 - -
-No GP IRRs or exit dates 2415 - -
-Deletion of identical deals 2392 22.31% 1.32 2.12
Deletion of non-buyout deals 1857 30.01% 1.78 2.39
Deletion of IRR>500% and multiples>50 1814 28.67% 1.57 2.27 Table III
Sample Construction
This table describes the filters applied to create the eventual sample. The first column describes the filter applied. The second column presents the residual deals after the filter. The third column provides performance statistics.
