PRIVATE EQUITY RETURNS AND DISCLOSURE AROUND THE WORLD*
Douglas Cumming Lally School of Management and Technology Department of Finance and Accounting Rensselaer Polytechnic Institute (RPI) Troy, New York 12180 USA Web: http://ssrn.com/author=75390 http://Douglas.Cumming.com Email:
[email protected]
Uwe Walz J.W. Goethe-Universität Frankfurt/Main Center for Financial Studies Schumannstr. 60 Uni-PF 64 D-60054 Frankfurt/Main Germany Tel.: ++49-69-798-22775 Fax: ++49-798-22753 email:
[email protected] http://www.wiwi.uni-frankfurt.de/profs/walz
12 September 2006
* Acknowledgements. We owe thanks to the workshop and conference participants at the J.W. Goethe-Universität Frankfurt/Main, European Business School, University of Cambridge Judge Institute of Management, Universitità di Bologna Almaweb Graduate School of Business, Universitità di Bologna Forli School of Business, Universitità di Trento Department of Legal Sciences and Faculty of Economics, European Finance Association Annual Conference, European Economic Association Annual Conference, University of New South Wales School of Banking and Finance, the Economic Society of Australia Annual Conference and the European Financial Management Association. We are grateful to CEPRES (Center for Private Equity Research, Frankfurt), especially Daniel Schmidt and the Center for Financial Studies (Frankfurt) for their generous funding and data.
1
PRIVATE EQUITY RETURNS AND DISCLOSURE AROUND THE WORLD
Abstract
This paper studies the returns to venture capital and private equity investment based on a sample of 221 private equity funds that are managed by 72 private equity management firms, in 5040 entrepreneurial investee firms, and spanning 33 years (1971 – 2003) and 39 countries from North and South America to Europe and Asia. We make use of four main categories of proxy variables for value-added activities and risks that explain venture capital and private equity returns: market and legal environment, fund characteristics, entrepreneurial firm characteristics, and the characteristics and structure of the investment.
We further compare actual unrealized returns, as reported to
institutional investors, to the predicted unrealized returns based on the estimates of realized returns. We show that significant systematic biases exist in the reporting of unrealized investments to institutional investors, depending on the level of the earnings aggressiveness and disclosure indices in a country, as well as proxies for the degree of information asymmetry between institutional investment managers and venture capital and private equity fund managers.
Keywords:
Venture Capital, Private Equity, Risk, Return, Law and Finance
JEL Classification: G24, G28, G31, G32, G35
2 1. Introduction
Venture capital (“VC”) and private equity (“PE”) funds are typically considered to be specialized intermediaries which invest in illiquid assets, i.e. their investee firms, and aim to add value over the life of their investment. Hence, there are two obvious issues which relate to the performance of VC and PE funds. First, how pronounced are the value-added contributions of VC and PE funds in their entrepreneurial investee firms relative to other factors which are responsible for the success of the investment? Or to put it differently: what are the determinants of the returns on these investments? Second, due to the illiquidity of the investment, VC and PE funds have to report to their own investors (limited partners of the fund) the valuation of the portfolio of investee firms and hence the implicit return on investment which have yet to be exited and for which a market price therefore typically fails to exist. This latter issue has become particularly important as a matter of public policy in recent ‘transparency lawsuits’ in which public pension funds such as CalPERS, the largest pension fund in the United States (“US”), were forced to disclose the performance results of VC and PE funds that they invested in to the public. Such disclosure has had drastic implications for the VC and PE industry in the US. For example, some VC and PE funds have restricted participation by institutional limited partners who may end up disclosing their performance results,1 and likewise, some pension funds have been forced to rethink their investment strategy in VC and PE funds.2 Furthermore, VC and PE associations3 and investment manager associations4 around the world have been reconsidering the issue of appropriate standards for reporting unrealized returns to institutional investors.
We translate these two broad issues, namely the determinants of returns on realized investments and on unrealized investments in the industry in the following research questions: First, how should the risk and return to PE on realized and unrealized investments be estimated and what determines the performance of PE investors in their entrepreneurial investee firms? Second, do systematic biases exist in the reporting of unrealized investments to institutional investors, and if so, under what conditions are those biases more pronounced?
1
For example, Sequoia Capital has ejected the University of Michigan; http://www.mercurynews.com/mld/mercurynews/business/6390139.htm 2 For example, CalPERS has been forced to reconsider its PE allocations, and in ways that differ relative to what it might otherwise have done but for the public disclosure; see http://www.ventureeconomics.com/vcj/protected/1070549534318.html 3 See, for example, http://www.evca.com/html/PE_industry/IS.asp 4 The Association for Investment Management and Research (AIMR), perhaps the leading international self regulatory organization around the world for investment managers, released new guidelines in September 2003; see http://www.aimr.org/pdf/standards/ipc/sept03/10a.pdf. As well, the National Venture Capital Association recently (as of 3 March 2004) rejected a proposal by the Private Equity Industry Guidelines Group regarding valuation guidelines, creating controversies among the Institutional Limited Partners Association and other industry associations; see http://www.privateequityonline.com/TopStory.asp?ID=4498&strType=1.
3 VC is traditionally defined as seed, start-up, early stage and expansion investing.5
PE, by
contrast, includes VC investments, and also more later stage buyout and turnaround investments, as well as investments in mezzanine firms that might soon be suitable for listing on a stock exchange. In this paper we study both VC and PE investments. However, , we use the term PE as a generic term that encompasses all investments in private firms as mentioned above. Likewise, for ease of exposition, we use the term “PE fund” to include both a VC fund and a PE fund in this paper (and similarly, a “PE manager” refers to the manager of a PE fund that makes and implements investment decisions).
In this paper we address the two above mentioned sets of reasearch questions by using a very detailed data set which gives us cash flow information on the level of the individual investment of 221 PE funds in 5040 investee firms spanning over a time period of 33 years (1971 – 2003). In addition, due to the fact that these investee firms are disbursed over 39 countries, we are able to investigate country-specific effects on the returns of realized as well as unrealized investments. We can calculate precisely the actual Internal Rate of Return (“IRR”) from all cash flows rather than having to rely on a proxy for returns computed from initial and final cash flows.
To study the determinants of realized IRRs, our empirical methods make use of bivariate Heckman sample selection procedures in order to account for selection effects in regards to realized versus unrealized investments (as in Cochrane, 2005), as well as full versus partially realized investments (as an extension to Cochrane, 2005). We make use of four main categories of variables to proxy for value-added activities and risks that explain PE returns: market and legal environment, investor characteristics, entrepreneurial firm characteristics, and the characteristics and structure of the investment. We find that the PE fund, entrepreneur and investment characteristics, as well as the economic environment, all contribute significantly to the success of the investment. We also show that the legal framework in the different countries dealt with in our sample significantly contributes to the performance of the investment: the more sound the legal conditions, the higher the IRRs. A similar result is presented by Lerner and Schoar (2005) based on a PE dataset of 210 investments; however, Lerner and Schoar (2005) study post-money valuations and fund-level returns, and do not have IRR data at the investee firm level. Our dataset, by contrast comprises 5040 investments from both developed and developing countries and precise details on not only realized IRRs but also unrealized IRRs reported by PE managers to their institutional investors. The data also comprise very specific details that significantly extend prior work, and reveal that interests of PE monitoring significantly increases the IRRs of realized investments.
5 Our definition of venture capital and private equity follows that put forth by Venture Economics in the US (www.ventureeconomics.com) and the European Venture Capital Association (EVCA) (www.evca.com).
4 We then extend our analysis to consider the unrealized IRRs, as reported by the PE managers to their institutional investors. No prior paper has considered the issue of reporting of unrealized returns to institutional investors.
In this paper, we compare the reported IRRs on unrealized
investments to what we would predict for such unrealized investments, based on our analysis of realized investments. We confirm the validity of the approach by comparing the realized IRRs to the previously reported unrealized IRRs for a subsample of the data, as shown in the Appendix. We show that there are systematic biases in the reporting of unrealized IRRs relative to what we would expect. These reporting biases are explained in terms of cross-country differences in accounting standards (Bhattacharya et al., 2003), legality (La Porta et al., 1997, 1998; Berkowitz et al., 2003), and proxies for information asymmetry between PE managers and their institutional investors (Gompers and Lerner, 1999).
We thereby provide empirical evidence for our hypothesis that the cost of
overreporting (higher loss of reputation in countries with a more sound legal system) are negatively related to the valuations of unrealized investments. Consistent with our theoretical analysis, we show experienced PE managers (with a reputational capital stock which is jeopardized by valuations that are too high) tend to report significantly lower valuations than their younger counterparts. Furthermore, it turns out that early stage and high-tech unrealized investments are, on average, valued higher than what we would predict based on realized early stage high-tech investments (industries such as computers and biotechnology which typically involve high industry market / book ratios).
Our paper focuses on the returns on investment in each specific entrepreneurial investee firm, as in Moskowitz and Vissing-Jorgensen (2002), Ljungqvist and Richardson (2003a), Das et al. (2003), Hand (2005) and Cochrane (2005). Our data, however, are distinct in that we have very specific details on each transaction and data from 39 countries (discussed in detail below), and we have details on the valuations of unexited investments. Cochrane (2005) and Hand (2005) analyze PE returns at the firm level by using US data sets. Cochrane (2005) investigates the performance of PE investments in individual US investee firms using the VentureOne database. The main focus of his analysis is to eliminate selection biases. We build on his work in this respect. Hand (2005) estimates the return on US VC biotechnology investments with a sample of 194 very detailed transactions. Ljungqvist and Richardson (2003b) use a proprietary data set to analyse the investment behaviour of PE funds in the US. Their main focus is on the investment and exit process (see also Das et al., 2003). Their analysis of the determinants of returns does not, however, take firm characteristics and fund manager details into account and is concentrating on US data only. Cumming and MacIntosh (2003) have data on exits and returns in Canada and US, but are limited in both breadth and depth of transactions. Hege et al. (2003) and Schwienbacher (2003) have similar data comparing Europe and the US, but lack the breadth and depth of data considered in this paper. This paper differs from these other papers by using a more comprehensive and more international data set as well as having a different focus with respect to our research questions. To the best of our knowledge, however, this paper is the first which addresses
5 the analysis of realized returns at the individual firm level with an international data set and which looks into the reporting biases with regard to unrealized investments.
By looking at the performance of PE funds in their individual investee firms we distinguish ourselves from some recent papers which focus on performance of PE investors at the fund level (see Hochberg et al. (2006), Ljungqvist and Richardson (2003b), Kaplan and Schoar (2005) as well as Phalippou and Zollo (2005)). These papers mainly look at PE as a separate asset class and aim to investigate the characteristics and performance of this asset class. In contrast, we go one disaggregation step further by analyzing the main determinants of the PE returns from individual investments and potential reporting biases of unrealized investments.
In addition, there are other VC and PE cross-country data sets which are to a certain extent related to ours. The scope of our data is similar to Black and Gilson (1998) and Jeng and Wells (2000), whose data are based on aggregate industry figures and comprise no transaction-specific information. Lerner and Schoar (2005) present cross-country data on 210 transaction structures from developing countries, and focus the analysis on the determinants of transaction structure. Lerner and Schoar (2005) also present data showing a negative relation between law quality and fund-level returns as well as post-money valuations (among other things such as deal terms; see also Bottazzi et al., 2006); however, as most of their investments have not been realized, they do not analyze investment-level returns. Lerner et al. (2006) present US returns data indicating sources of funds’ impact returns, consistent with the results of Mayer et al. (2005) based on an international data set. Cumming and MacIntosh (2003) and Cumming et al. (2006) have data on exits and returns in Canada and the US and Australasia, respectively, but are limited in both breadth and depth of transactions. Gompers et al. (2003) present a large data set on entrepreneurial firms across different European countries, but do not consider information pertaining to PE finance. Our paper is distinct in that we consider extensive details relating to returns in an international context, and for the first time ever we consider reports of returns on unrealized investments.
This paper is organized as follows. A discussion of the institutional setting of PE funds’ disbursements of cash flows as well as their reporting to institutional investors is provided in section 2. In section 3 we derive the theoretical hypotheses which form the basis for our empirical analysis. The data are described in section 4.
Section 5 provides an analysis of realized IRRs with
consideration of sample selection issues. Section 6 compares unrealized IRRs to predicted IRRs, and companion analyses of comparing realized IRRs to prior reported unrealized IRRs for a subset of the data is presented in the Appendix. A discussion of limitations and extensions is provided in section 7. Concluding remarks follow in the last section.
6 2. The Institutional Setting of PE Fund Disclosures to Institutional Investors
In order to set the stage for our analysis of PE funds’ return on investments which have been exited and those in which the PE fund is still invested we discuss in this section briefly the institutional background of PE financing. PE funds are financial intermediaries between entrepreneurial firms and institutional investors (Sahlman, 1990). It is in fact widely recognized that PE funds exist because of the pronounced information asymmetries and principal-agent problems in financing start-up firms, and the inability of institutional investors in terms of time and skill to select suitable entrepreneurial firms in which to invest. PE funds are therefore typically set up as limited partnerships that exist for a 10 year period, with an option to continue for a further 3 years, so that investments may be selected and brought to fruition in an exit. The PE manager, which selects, monitors and adds value to the investment, is the general partner and the institutional investors are the limited partners (Gompers and Lerner, 1999). Each year, PE managers report valuations on unrealized portfolio investments to their institutional investors. There is however a principal-agent problem between PE managers and their institutional investors.
While the measurement of returns on
investments which have been exited is straightforward since the resulting cash flows have to be distributed to the limited partners, matters are more difficult with respect to the reporting of returns on unrealized investments since they hinge on the valuations determined by the PE managers.
PE funds are subject to an audit review each year. This means that usually the auditors come in and review the appropriateness of the carrying value of the overall portfolio of investments of the fund in the fund's accounts. Typically, the auditors do not have authority/ability6 to certify the value of any individual specific entrepreneurial firm that has not yet been realized, but in order to ascertain the accuracy of the overall portfolio valuation and sign it, they do need to "look over the shoulder" of the PE managers to see how they have been valuing the entrepreneurial firms. These valuations are generally based on the financial statements and any other management reports of the entrepreneurial firms provided to the PE managers.
The valuations of unrealized investments in the PE industry have not been formally regulated around the world (Lerner et al., 2004). In many PE markets (such as Australia, Canada, the US, the United Kingdom (“UK”), and Continental Europe), only guidelines for valuations have been issued by the national VC and PE associations. The guidelines typically recommend that valuations should not deviate from “Cost” within the first 12 months from the time of first investment.7 “Cost” is based on values established through an initial outside-led investment round with a sophisticated unrelated party 6 Many of the PE managers that the authors have met in Australia, North America and Europe have indicated “ability” would be a more appropriate word than “authority” in this sentence. 7 See, for example, http://www.evca.com/pdf/EVCA%20Guidelines/Guidelines_valuation.pdf for the EVCA guidelines.
7 that was not a strategic investor. “Market” value, by contrast, is based on the values of the most recent independent round (Lerner et al., 2004, p. 4). However, PE funds are not obliged to follow these guidelines closely.
It is important to note that the lack of formal rules for valuation of unrealized investments does not at all imply that there is no systematic direct link between valuations and accounting standards. Regardless of whether valuation is at Cost or Market, the price paid by any investor will be based on future cash flows which are expectations based on the entrepreneurial firm’s current earnings. Therefore, financial reporting in PE should have an effect on valuations (Armstrong et al., 2005; Hand, 2005). The financial statements or financial reporting of the entrepreneurial firms, in turn, are affected by the prevailing accounting standards in the respective country (and their enforcement), since these accounting standards apply to private firms as well as to public firms. In short, despite the fact that the PE industry valuations are not directly regulated, there is a link between accounting standards and valuations because financial reporting by, and the preparation of, financial statements for the entrepreneurial firms are regulated,8 and because current earning figures form the basis for the valuations of PE backed firms.
We note that our data does not comprise direct information on the financial statements of the individual entrepreneurial firms. The hypothesis we test is therefore that accounting standards do have an impact on earnings management and financial reporting, and that reporting of earnings has an effect on valuations of unrealized PE investments. Accounting standards (and their enforcement) have an effect on valuation directly and indirectly. If valuation broadly follows the guidelines of the different industry standards, the prices paid in previous rounds are affected by the earnings announced by the firm. We further note that even if valuations are not based on prices paid in former financing rounds, PE managers are indirectly influenced by the accounting figures, as deviating substantially from them needs to be justified to the limited partners (institutional investors) and auditors. The notion of indirect effects of accounting standards on the quality of disclosure reflects the idea that accounting standards reflect the overall legal framework and the entire “culture” of proper reporting. We investigate this hypothesis of indirect effect by looking at the impact of the legal and accounting framework on the degree of over-reporting.
Several papers in recent research literature are consistent with our core hypothesis that a country’s accounting standards affect the quality of valuations by PE managers to their institutional investors directly as well as indirectly. Verecchia (1983) and Gigler (1994) provide theoretical models of voluntary disclosure which result in full and partial disclosure equilibria. Healy and Palepu
8 The same accounting standards apply to publicly listed and private firms; see Hand (2005) for the US, Ball and Shivakumar (2005) for the UK, and Burgstahler et al. (2004) for Europe.
8 (2001) show that voluntary disclosure is related to both the economic and institutional environment. In a number of studies in analogous work in different contexts outside of the realm of PE finance, the quality of the governance environment (both within the PE fund as well as the external auditing standards of a particular country) turns out to be positively correlated with disclosure quality (Ho and Wong, 2001). PE managers likewise view accounting information as being extremely important and tied to contracts with entrepreneurs (Mitchell et al., 1995). The performance of realized VC-backed firms is significantly linked to accounting information (Armstrong et al., 2005; Hand, 2005), and round-to-round VC returns prior to exit events are also closely linked to financial statement information (Hand, 2005). Finally, a contemporaneous paper shows that the legal and institutional environment is important for the quality of VC disclosures in Belgium (Beuselinck et al., 2005), which is consistent with our cross-country empirical analysis.
In the empirical analyses in this paper we make use of different accounting indices that allow a broud comparison of private versus public reporting standards across countries. In our analyses we compare public valuations (of realized investments which are at least partially traded publicly (after IPOs)) with private valuations (before exits occurred). Our approach is motivated by accounting research on earnings management and valuation reports for private and public firms (Burgstahler et al., 2006; Ball and Shivakumar, 2005). By providing a first look at valuations of unrealized PE investments in an international setting we provide a relevant contribution to this related literature.
3. Reporting Biases with Unrealizd Investments: Hypotheses.
In this section we summarize hypotheses pertaining to potential factors that affect the valuations of unrealized investments. There is a principal-agent relationship between institutional investors (principals) and PE managers (agents). Institutional investors face information asymmetries in that the PE fund holds illiquid assets in investee firms which do not have a Market value until divestment or a realization event. PE managers have an incentive to overvalue unrealized investments in order to attract other institutional investor capital to raise follow-up funds. This incentive to overvalue is significant and especially pronounced for first-time PE managers without a track record of successful divestments. For example, young PE managers grandstand to institutional investors by taking investee firms public sooner than that which would otherwise be appropriate for the investee firms (Gompers and Lerner, 1999).
The expected cost of overvaluation is the loss of reputation. Reputational concerns have been shown to be extremely important for VCs in the context of IPOs (Megginson and Weiss, 1991). Note that reputation costs are not directly associated with reporting overvaluations of unrealized investments. When the investment comes to fruiting and the PE fund exits, if the exited value is
9 notably less than the reported value then it may be the case that the decline in value is attributable to adverse changes in market conditions between the time of reporting and the time of exit. Changes in value may also be attributable to idiosyncratic factors associated with the investee firm for which the PE manager was not aware at the time of reporting of unrealized value. Consistent with Hand (2005, 2006), we may expect that it is more difficult for institutional investors to compare reports of valuations on unrealized investments to the subsequent actual exit outcome in countries with less stringent accounting standards; that is, there is a comparative dearth of reliable reported information that enables the institutional investor to ascertain whether the change in value is due to actual changes in value from time of report to exit, or simply due to overvaluation.
Reporting of overvaluations will occur where the marginal benefits in the current period outweigh the marginal costs of overvaluation in future periods. Trade-offs between overvaluations and reputation costs have been addressed in prior theoretical work (see, e.g., Benabou and Laroque, 1992; Stocken, 2000; Neus and Walz, 2005). This work shows insiders (management) disclose their private information truthfully provided that the investors are sufficiently patient, the accounting system is sufficiently useful for assessing the credibility of the insider’s disclosure, and the insider’s disclosure is evaluated over a sufficiently long period. The primary hypotheses tested in this paper likewise compare the costs of reporting overvaluations (the expected discounted reputational losses arising in future periods) against the benefits of reporting overvaluations (the expected increase in probability of raising a new fund in the present period). Our first hypothesis focuses on international differences in accounting standards. The other hypotheses focus on factors that proxy for information asymmetries faced by institutional investors: ¾ Less stringent accounting standards lower the expected reputation costs to PE managers associated with reporting overvaluations of unrealized investments to institutional investors. Less stringent accounting standards also make reporting overvaluations more difficult to distinguish from actual changes in firm value from the time of reporting to the time of exit. We further expect less reporting of overvaluations after the introduction of the SarbanesOxley Legislation on 30 July 2002. ¾ Inexperienced PE managers with a comparative dearth of reputational capital face lower expected costs of reporting overvaluations, and hence a stronger incentive to overvalue. ¾ Overvaluations are more likely for earlier stage investments since time to exit is longer for investee firms in the earlier stages of investment, and reporting of overvaluations is more likely to be blurred from changes in market conditions and other idiosyncratic factors that affect firm value.
10
¾ Investments for which information asymmetries are more pronounced (such as high-tech and early stage firms) are more likely to be overvalued since the expected costs associated with reporting overvaluations are lower; that is, it more difficult for the institutional investor to disentangle negative shocks associated with changes in market conditions and other idiosyncratic factors that affect value from overvaluation. ¾ Syndicated investments are less likely to be overvalued since such overvaluations might be revealed to, or checked by, the other syndicated investors.
In our empirical analyses of unrealized returns in the subsequent sections of this paper, we initially consider factors that impact realized returns in order to compare unrealized returns with expected returns on unrealized investments. Realized returns are expected to be greater in times of better market conditions (Cochrane, 2005), legal conditions (Lerner and Schoar, 2005) and improved governance (Hochberg et al., 2006; Ljungqvist and Richardson, 2003a,b). These issues are discussed further below in conjunction with the empirical analyses.
The next section introduces the data to enable tests of these hypotheses. Multivariate tests are presented in section 5. A discussion of extensions and concluding remarks follow in the last sections.
4. Data
4.1. Data Description
Our data set was collected by the Center of Private Equity Research (CEPRES) in Frankfurt, Germany. The data comprise 221 PE funds managed by 72 PE managers, 5040 observations of entrepreneurial investee firms (3826 VC and 1214 late stage mezzanine and buyout) spanning 33 years (1971 – 2003), and 39 countries from North and South America to Europe and Asia.9 For confidentiality reasons, names of funds, managers and firms etc. are not disclosed. The data comprises 2498 fully realized investments, 954 partially realized investments,10 and 1665 unrealized investments. The data are depicted in Figures 1 and 2. The volume of data (Figure 9
Specifically, the countries include Argentina, Austria, Belgium, Brazil, Canada, China, Czech, Denmark, Finland, France, Germany, Greece, Guatemala, Hong Kong, Iceland, India, Indonesia, Ireland, Israel, Italy, Japan, Korea, Luxembourg, Malaysia, Netherlands, Norway, Philippines, Poland, Portugal, Puerto Rico, Romania, Russia, Singapore, Spain, Sweden, Switzerland, Taiwan, the UK, and the US. 10 A partially realized investment involves a disposition whereby less than 100% of the investor’s ownership interest has been sold (for empirical work on topic, see Gompers and Lerner, 1999; Cumming and MacIntosh, 2003; for theoretical work on topic, see Neus and Walz, 2005). The complete return on the investment cannot be perfectly measured for partial exits in our dataset.
11 2) is consistent with that reported elsewhere (the volume of transactions in the US, see Gompers and Lerner, 1999; Lerner, 2002a). The presence of realized, partially realized and unrealized investments (Figure 2) is a useful attribute of the data as it enables us to consider selection effects (Heckman, 1976, 1979) in two dimensions: realized versus unrealized investments, and full versus partial exits.11 In estimating the determinants of returns, we consider selection effects on both dimensions (consistent with Cochrane, 2005, but Cochrane does not have details on partial exits).
[Figures 1 and 2 About Here]
It is important to note that our data comprises details on the actual IRRs for realized investments (accounting for all cash flows between the investor and the entrepreneurial firm). This is distinct from other papers (Cochrane, 2005), who appears to proxy returns based on initial cash flows and final cash flows. Our actual IRRs are not approximations; the data are extremely precise.
Further, our data comprises details on the IRRs of unrealized investments; that is, the IRRs reported to institutional investors by the PE managers on unrealized transactions. These unrealized IRRs in the data were reported in the period from June 2000 – Sept. 2003 (and the investment dates are indicated in Figure 2). The presence of reported unrealized IRRs is unique and significant, as it enables us to compare these reported unrealized returns with the predicted returns for unrealized investments (based on our analysis of returns derived from realized investments). No previous paper has provided data on this topic. We believe that this dimension of data and analysis is an important new contribution to the literature, particularly in light of the recent US CalPERS case.12
The data comprise detailed information on a number of different transaction-specific variables, as summarized and defined in Table I.13 The types of variables are broken down into 4 primary categories: market and legal factors, PE fund characteristics, entrepreneurial firm characteristics, and investment characteristics. These variables are used in the ensuing empirical analyses.
[Table I About Here]
4.2. Summary Statistics 11
It would be possible to present selection effects for a third dimension if we had a complete set of details for each transaction. That is, we exclude some of the observations in the empirics where we do not have a complete set of details on each variable of interest. This approach is consistent with that advocated in Greene (1997). Cochrane (2005) adopts a similar approach in excluding observations with incomplete data. 12 The CalPERS case and related issues were discussed in the Introduction. 13 There are a few additional details in the data that are not reported in this paper. The main reason for this is that we believe our theoretical model has captured the important aspects that pertain to the research questions at hand. Excessive reporting of other variables would detract from the central focus.
12
Summary statistics are presented in Table II and are separated into five categories: (A) all funds, (B) market and legal factors, (C) PE fund characteristics, (D) entrepreneurial firm characteristics, (E) transaction specific characteristics. Comparison tests are explicitly provided for average and median returns across fully realized versus unrealized or only partially realized investments14.
[Table II About Here]
The data indicate that the median unrealized IRR is 0.00 for all transactions (Table I, Part A, row 1), but the average unrealized IRR is 63.23%. Realized IRRs, by contrast, have a median of 16.99% and an average of 68.67%. Median realized IRRs are significantly greater than median unrealized IRRs, but average realized IRRs are not statistically different from one another. The insignificance of the differences in average values is attributable to the very large standard deviations of the returns. The fact that PE returns have a massive variance has previously been reported with US data (Cochrane, 2005). The dispersion of the returns in our data is graphically depicted in Figure 1.
Some very interesting differences are observed in the data with regard to breakdowns of market and legal factors. When public equity markets experience high returns (Table II, Part B, row 2), realized returns are greater than unrealized returns; however, when public equity markets experience low returns (Part B, row 3), unrealized returns are greater than realized returns. The same applies to risk-free returns (Part 3, rows 4 and 5) in hot versus cold market periods. Also, note that unrealized investment returns are sticky downwards at 0.00% (see Figure 1). The data clearly indicate that PE managers do not tend to write off the value of an investment below its book value, until such losses are actually realized.
With regard to the legal and accounting indices (Table II, Part B, rows 6-11), the average realized IRRs are insignificantly different from the average unrealized IRRs, but the median IRRs are higher for realized investments. It is significant that for unrealized IRRs, the average IRRs are higher among countries with lower legality and accounting indices. Due to the massive variance in IRRs, these differences of means tests are not statistically significant, but are nevertheless indicative of trends in the data that warrant further investigation in the multivariate analyses below.
Part C of Table II reports the data according to various PE fund characteristics. While average and median realized returns are higher among funds within PE managers of different vintages
14 Comparison tests across different characteristics within the grouping of realized or unrealized investments are apparent, but not explicitly provided for reasons of conciseness.
13 (rows 12 and 13), note that funds managed by PE managers of younger vintage (3 or fewer funds per manager) have much higher median unrealized IRRs. Regarding the age of a particular fund, younger funds are less inclined to report losses on unrealized investments (row 15). It is also worthy of attention that funds with large portfolios have statistically significantly higher average unreported IRRs relative to reported IRRs. This latter result suggests that funds which add less value to their entrepreneurial firms15 are more inclined to exaggerate their IRR performance on unrealized investments. Part D of Table II reports the data according to entrepreneurial firm characteristics.16 The data indicate that for the start-up and early stages of investment (rows 19 and 20) for which informational opaqueness is very pronounced, unrealized IRRs are greater than realized IRRs. Unrealized IRRs are less than realized IRRs at the latter development stages. It is also noteworthy that the average unrealized IRRs of firms in industries with high market/book values are quite high (more than 100%; but insignificantly different from realized IRRs due to the high variance).
Part E of Table II reports the data according to investment characteristics. Lead investors (row 30) report very high average IRRs on unrealized investments. Average unrealized returns among firms for which convertible securities were used, and among firms with high standard deviations of cash flows are high as well. But again these differences are not significantly different from average realized IRRs due to the high variance.
Table III provides a correlation matrix for the subsample of realized investments. The correlations provide some insight into the univariate relations between the variables. For instance, realized returns are positively associated with public market returns, convertible securities and syndication are negatively related to co-investment. The correlations also indicate potential collinearity problems across the variables considered in the multivariate analyses reported in the subsequent tables.17
[Table III About Here]
The univariate statistics provide only initial, albeit important, insights. Therefore, we provide in the subsequent section a multivariate analysis of the determinants of realized returns across countries. Thereafter section 6 considers the difference between unrealized returns as reported to 15 Portfolio size (in terms of the number of investees) per investment professional within a PE manager and valueadded are inversely related (Kanniainen and Keuschnigg, 2003a, b). 16 Note that across countries the definition of a seed, start-up, early stage and even expansion stage firm is a little difficult due to differences in conventions across countries. For many of the firms included in the data we were unable to obtain a reliable definition, and therefore use an ‘unknown’ category (row 22). 17 The results reported below are very robust, mainly due to the large number of observations in the sample. Alternative specifications not explicitly provided are available upon request.
14 institutional investors and predicted realized returns for the unrealized investments. As mentioned earlier, these predicted realized returns emerge by applying our estimations for the realized investments to the unrealized investments.
5. Multivariate Analysis of IRR Performance
5.1. Empirical Methods
Our aim in this section is to study the determinants of fully realized returns in our crosscountry sample described above. There are two approaches that appear to have been employed in previous work. On the one hand, one could use OLS on a subsample of the fully realized returns (our impression is that this is the approach used by Ljungqvist and Richardson, 2003b, Table VI, for a US sample). On the other hand, one could account for sample selection issues with regard to realized and unrealized investments (as considered by Cochrane, 2005, for a US sample). In this paper, we use OLS on the subsample of realized returns, and then analyze the robustness of those results concerning sample selection corrections.
Our sample selection corrections involve multiple steps. The first step involves determining the probability of an exit (either full or partial). The second step involves determining the probability of a full exit (versus a partial exit, as defined above), taking into account the first step consideration of an actual exit (this Heckman-like methodology is easily adopted with Limdep Econometric Software; see Greene, 2002). The third step is the linear regression, explaining returns with the sample selection correction based on steps one and two (Heckman, 1976, 1979). It is noteworthy that our results are quite robust in relation to alternative specifications of the sample selection corrections (alternative specifications not specifically reported are available upon request), but not as robust in relation to the standard OLS estimates in the subsample of fully realized exits. Our approach is based on previous work and considers a multi-step Heckman-like sample selection correction on realized/unrealized exits and full/partial exits. This approach is intuitively sensible, and out-performs other single-step sample selection corrections (again, available on request but not explicitly reported), as well as standard OLS methods on the subsample of realized exits (which is explicitly indicated).
Our econometric specifications are the function of the following variables: (1) Probability of observing an actual exit = f {age of investment} (2) Probability of a full exit = f {age of investment, legality, stage of investment, country dummy variables, industry dummy variables, exit year dummy variables, syndication | Actual Exit in regression (1)}
15 (3) Realized returns = f {market and legal conditions [MSCI returns, risk-free returns, legality, committed capital in market at investment date], PE fund characteristics [number of funds and portfolio size per investment professional within the PE manager], entrepreneurial firm characteristics [stage of development, industry market/book, country of residence dummy variables, industry dummy variables, exit year dummy variables], investment characteristics [lead investment, syndicated investment, co-investment, board seats, convertible security, standard deviation of amount invested, and initial amount invested] | Actual Exit [regression (1)] and Full Exit [regression (2)]}
The particular variables were defined above and summarized in Table I. The age of the investment seemed to be a natural explanatory variable for the probability of an actual exit. The longer a particular firm is in the portfolio of the PE fund, the more likely it will be divested. This refers to bad as well as good ventures.
In view of the large number of regressions reported below, we do not present a large variety of robustness checks. considered.
Alternative specifications with different right-hand-side variables were
The results are extremely robust, which is attributable to the large number of
observations in the data. For instance, in a prior version of the paper we presented twice as many regressions with exclusions for industry, country and exit year dummy variables, and the results were extremely robust. We have not presented such robustness checks in order to provide a more succinct and clear presentation. These alternative specifications are available upon request.
Other variables present in the data (such as other measures of market returns, etc.) were considered but deemed less relevant. We present regressions in which the left- and right-hand-side variables, besides dummy variables,
are in logs.
As such, the coefficients are interpreted as
elasticities. We also considered the regressions in levels; those results were very similar, and are available upon request. Furthermore, we considered the use of different explanatory variables in the earlier steps (1) and (2) regressions for the selection effects (some of which are reported in the tables), but this did not affect the core results discussed below.
There could be concern about the fact that some of the right-hand-side variables are endogenous. For example, syndication might be endogenous if project quality affects the probability of syndication. We did consider this issue, but were limited by the absence of ideal instruments (as in Brander et al., 2002). Some instruments considered included fund characteristics (such as fund location, if different from the entrepreneurial firm, and fund size), which could be more closely connected to syndication than to returns themselves. As we did not find material differences in the eventual results, we have only reported the straight estimates without the use of instrumental variable
16 methods. As other papers in this field do not use instrumental variables when explaining returns on investment as a function of syndication and other investment characteristics (see, for example, Brander et al., 2002, on the effect of syndication on returns in Canada) we report specifications without the use of instruments as well. Nevertheless, additional specifications are available upon request. Other papers (Cochrane, 2005; Ljungqvist and Richardson, 2003a,b) do not consider the effect of any investment characteristics on returns (unlike, for example Hand, 2005, who does consider more detailed firm-specific information). We feel less comfortable with dropping these variables, as they have been used, for example, to explain the performance of VC-backed IPOs and the ensuing returns to VCs (Barry et al., 1990; Megginson and Weiss, 1991; Gompers and Lerner, 1999; Hand, 2005).
Failure to consider these variables may result in a more serious problem
regarding omitted variables in relation to endogeneity. In any event, the main results are quite robust and alternative specifications, not explicitly reported, are available upon request. 18
5.2. Empirical Results
We present the results for the subsample of investments in which the first round was VC in nature (in the seed, start-up, early or expansion stages) (Table IV Panel A), as well as the results of the full sample that also comprises the different types of later stage investments (Table IV Panel B). We point out certain differences in the estimates in the different samples. We explicitly present two models in each Panel. Model (1) is the standard OLS concerning the subsample of fully realized exits. Model (2) is the three-step bivariate Heckman-corrected estimates based on actual exits versus no exit and full exits versus partial exits.
[Table IV Panels A and B About Here]
It should be noted that the data indicate a superior fit in regard to the Heckman corrected Model (2) relative to the simple OLS Model (1) concerning the subsample of realized returns. Adjusted R2 and likelihood ratio, Akaike and other model selection criteria all point to the appropriateness of Models (2) (this is the case more so for Panel A on the subsample of earlier stage investments than for Panel B on the full sample including the later stage investments).
18 Ideally, our specifications in each step would involve different explanatory variables (Puhani, 2000). To some extent we are able to achieve this, as the right-hand-side variables do not completely overlap. For instance, the age of the investment is in steps (1) and (2) but not step (3). Our reported results are robust with regard to alternative specifications. A limitation in our dataset is that in many cases the precise exit vehicle (IPO, acquisition, buyback) is unknown, and hence that dimension cannot be explored with the data. Nevertheless, this is not a significant limitation for our research question as there is no causal relation between exit vehicle choice and returns; exit choice would be endogenous to a good project with high returns. At a general level of comparison, steps (1) and (3) are consistent with Cochrane (2005), steps (1) and (2) are consistent with Cumming et al. (2006) and Step (3) is consistent with Ljungqvist and Richardson (2003b) and Brander et al. (2002).
17 Selection Effects
Step 1 selection regressions in Model (2) indicate that the exit itself is more likely to be observed in the data the longer the duration of the investment. This is an obvious point, and this variable has been used in previous work on the topic with a sample of US data (Cochrane, 2005).
Step 2 selection regressions in Model (2) consider the determinants of full versus partial exits. A partial exit facilitates ownership transfer when it is relatively difficult for the new owner to value and monitor the firm. Consistent with previous work (Gompers and Lerner, 1999), PE funds will choose a partial exit when informational problems faced by the new owners are more pronounced. The funds will then complete the exit and fully divest once the new owners feel sufficiently confident to undertake total ownership of the firm. Our specifications control for industry factors, year effects, stage of development at first investment, investment duration, and investment syndication (consistent with Gompers and Lerner, 1999; see also Cumming and MacIntosh, 2003). The data indicated that controls for other factors were not warranted. Furthermore, we did not want to over-specify the full/partial exits regressions as it is undesirable for the different Heckman regressions to have righthand-side variables that are overly correlated between equations (Puhani, 2000). The data indicates that full exits are more likely the longer the investment duration and less likely for syndicated investments. The duration evidence is intuitive, in that a longer investment duration until the first exit date (albeit not necessarily the full exit date) facilitates certification of quality vis-à-vis the entrepreneurial firm and its new owners (Megginson and Weiss, 1991). The syndication evidence is less intuitive, as syndication itself could certify quality. Syndication, however, may be a sign of informational problems, and there could be some underlying variable which is unobserved driving both the syndication decision and the partial exit. These details are not apparent in our data. The evidence in regard to stage of development is sensitive to the inclusion or exclusion of the dummy variables for industries, countries and exit years in Model (2). Finally, Table IV Panel B indicates that full exits are more likely for countries with higher legality indices, which is only to be expected, as there is less of a need to certify quality via a partial exit when investors have better legal protection and more certainty.
Even though alternative specifications for the first-step selection regressions were considered including alternative right-hand-side variables, single-step mechanisms versus multiple step, etc., they did not materially impact the results presented, nor those discussed below pertaining to IRRs. Different specifications are available upon request. Given these preliminary selection regressions, we now turn to an analysis of the returns based on the Heckman corrections, with comparison to standard OLS regressions based on the subsample of fully realized IRRs.
18 The Impact of Market and Legal Factors on Returns
The importance of considering selection effects in the data is perhaps best illustrated by the first variable, the log of the MSCI return. In the OLS specifications on the restricted sample (Model (1)), public market returns are statistically unrelated to exit outcomes. Only in Model (2) do we find a statistically significant and positive coefficient which is of course expected as per the Capital Asset Pricing Model. The data indicate that the beta coefficient on the log MSCI index is slightly greater than one.19
The coefficient on the legality index is positive and significant in all of the specifications. This indicates that legal protection facilitates PE returns, consistent with the role of legality in public markets (see La Porta et al., 1997, 1998).
The committed capital variable is negative and significant in Model (1), and positive and significant in Model (2). We would expect an a priori negative coefficient if excess capital pushes up deal prices and lowers returns (Gompers and Lerner, 2000).
The reversal of the sign of this
coefficient is likely attributable to the fact that the presence of a realized versus unrealized investment is closely connected to the market conditions at the time of investment. Our unrealized investments are those that had not been realized in the period from June 2000 to September 2003. The sample selection corrections on this dataset indicate that the greater capital inflows were associated with higher returns.20
The Impact of Fund Characteristics on Returns
We report results for two primary fund characteristics: the numerical order of PE funds managed by the PE manager (latter funds may perform better as their managers are more experienced), and portfolio size (number of investees) per investment professional within a PE manager. On the one hand, we do not find a significant effect of the numerical order of funds on returns. On the other hand, the effect of portfolio size per investment professional within a PE manager is highly significant and economically large (consistent with Kanniainen and Keuschnigg,
19 By contrast, Cochrane (2005) finds the log CAPM market coefficient to be around 90%. Cochrane’s data are only from the US, which might suggest that VC returns across countries are slightly riskier than in the US. However, Cochrane accounts for at most 1% of the variation of the returns in his sample, while our consideration of other variables offers an explanation for more than 35% of the variation in returns. Furthermore, our sample selection mechanisms are different (for example, we include a step for full and partial exits, described above). Given the risks and illiquidity in the VC and PE market, it is natural to expect a beta on the market return variable to be greater than one. 20 This result is not directly comparable to Gompers and Lerner (2000), since we have a specific dataset where nonrealizations are concentrated on a period after the Internet bubble market crash. As well, we have a multitude of countries in the data. Further, our focus is not on initial valuations of investments, as considered by Gompers and Lerner (2000) but rather on returns. As indicated in note 16, note that the coefficient on committed capital is not affected by collinearity with the MSCI variable.
19 2003a,b; Keuschnigg, 2004; Cumming, 2006): the estimated elasticity ranges from 0.31 to 0.51 in Table IV Panel A and from 0.25 to 0.33 in Panel B. The smaller economic effect in Panel B is consistent with the widely held view that PE managers add less value to later stage investments than to earlier stage investments (see Gompers and Lerner, 1999, for the majority of the seminal work on this point). Overall, although prior seminal work specifically examining VC returns (Cochrane, 2005; Ljungqvist and Richardson, 2003b) has not considered the effect of portfolio size, the evidence from this dataset suggests this is quite an important variable. In fact, constructive advice and monitoring provided by the PE manager to the firm is the primary element that distinguishes PE from other forms of more passive financial intermediation (Gompers and Lerner, 1999).
The Impact of Entrepreneurial Firm Characteristics on Returns
A somewhat surprising result in the data is that the earlier stage investments did not outperform the later stage investments (see also the summary statistics in Table II). The earlier stage investment variables (for stage of investment at first investment) are generally insignificant, or negative and significant. The only exception is in Table IV, Panel A, Model 1 for start-ups, but that coefficient estimate is not robust. By contrast, the late-stage investment variable (pre-IPO; see Panel B in Table IV) yielded higher returns. Similarly, there were a total of 14 realized returns for investments in publicly listed firms in the data, and these performed well (see Table II, and Table IV Panel B).21 For the most part, the regressions are robust for controls for different country dummy variables, industry dummy variables, and year of exit dummy variables.
The Impact of Investment Characteristics on Returns
Investment structures appear to have a significant effect on returns in a number of different dimensions.
First, syndication significantly enhances returns, consistent with the view that
syndication facilitates value-added investments (Gompers and Lerner, 1999; Brander et al., 2002). Second, co-investment and allocation of board seats are associated with lower returns.
One
explanation for this result is that co-investment and allocation of board seats are more likely for poorly performing investments, as indicated by Gompers and Lerner (1999).22 Third, the use of convertible securities with periodic cash flows (Table I) enhance returns, consistent with the view that the use of convertible securities gives rise to incentives for the PE manager to provide value-added advice and to efficiently monitor the firm, and incentives for the investee to work (for recent 21 These investments actually appear to be IPO allocations in which investment banks enabled the PE fund to buy in to a publicly listed firm early on in order to take advantage of IPO underpricing. 22 As indicated above, this may indicate that these variables are endogenous. Alternative specifications with instruments did not materially affect the results, as discussed above. As well, specifications in which this (and related investment-specific) variables were dropped altogether did not materially affect the results pertaining to the other variables reported in the tables.
20 theoretical work, see, Casamatta, 2003; Schmidt, 2003; Repullo and Suarez, 2004). Convertible securities also facilitate efficient exits (Berglöf, 1994; Bascha and Walz, 2001), which is in line with our finding that higher returns are associated with the use of convertible securities.
Consistent with the evidence discussed above pertaining to the stage of investment, the evidence on the size of the initial investment indicates lower returns on smaller investments (but this effect is statistically significant in Panel A only). As regards the economic significance, the estimated elasticity is approximately 0.10 in Panel A, which is statistically significant only when selection effects are considered.
The standard deviation of the cash flows is positively associated with returns. Variability in the size of cash flows is a direct proxy for the risk of the investment, so we would expect a positive coefficient in that riskier investments are associated with greater returns. This positive association is stronger and more robust for the subsample of earlier stage investments (Panel A) compared to the full sample of all investments (Panel B).
The Impact of Selection Effects on Realized Returns
It is important to point out that both selection effects have a statistically and economically important impact on the measurement of returns. The Lambda A and B coefficients are both negative and statistically significant in Models 3 and 4 in Table IV Panels A and B, with the exception of Model 3 in Panel A, where Lambda B is marginally insignificant. The negative sign of these coefficients indicates that realized returns are systematically lower than unrealized returns.
In
particular, unrealized returns are roughly 7-9% higher than realized returns, and unrealized returns from partial exits are roughly 1-2% higher than realized returns. This implies that the degree to which unrealized returns are overstated is greater for completely unrealized exits compared to partially realized exits. This is in line with our findings that exits lead to more realistic valuations. Our examination of completely unrealized exits is therefore the focus of our analysis in the next section.
In order to verify the correctness and importance of selection effects, one particularly revealing OLS regression (not reported in the Tables) considered the IRRs of our entire sample as the LHS variable. Using our variables from above and adding a dummy for the unrealized investments reveals a highly significant (t-statistic of 14.979, significant at the one percent level) positive coefficient (equal to 2.049) for the unrealized dummy variable. We interpret this as a strong indication of overvaluation. In the next section we investigate the determinants of the differences between reported and predicted valuation of unrealized investments.
21 Overall, the models in Table IV Panels A and B correspond closely to the data. Adjusted R2 values are quite high (ranging from 28% to 35%). By contrast, other approaches to measuring returns with US VC data which do not consider selection effects account for up to 13% of the variation in returns (Ljungqvist and Richardson, 2003b), and papers which do not account for PE value-added activities and transaction specific details account for only 1% of the variation in returns (Cochrane, 2005). The high R2 values in our specifications are useful for reliably predicting returns of unrealized investments, as considered in the next section.
6. Analysis of Reporting Biases in Unrealized Investments
6.1. Empirical Methods
In our data we observe one investment valuation for each unexited investee firm in the period from June 2000 – Sept. 2003. There is not more than one valuation per investee firm observed in this period. Our analysis of unrealized returns compares the reported valuation of unrealized investments with the predicted performance of unrealized investments. The validity of this approach is confirmed with complementary analyses in the Appendix.23
There are 3 steps to comparing predictred returns for unrealized investments with reported returns on unrealized investments. First, we estimate the realized IRRs based on a set of explanatory variables that proxy for market and legal conditions, PE fund and entrepreneurial firm characteristics, and the characteristics of the transaction for each of the observations for the realized returns (collectively represented by XRealized in equation (1)) as discussed in the previous section.
(1) IRRRe alized = α 1 + β 1 X Re alized + residuals Second, we generate a vector of predicted returns for the full sample of all investments based on the estimated coefficients in equation (1), as follows: ∩
∩
(2) IRRPr edicted = α 1 + β 1 X Unrealized Finally, we compare the difference between the reported unrealized returns by PE managers to the institutional investors and predicted returns, and regress this difference on a set of explanatory variables.
(3) IRRUnrealized − IRRPr edicted = α 3 + β 3 X Unrealized + residuals The estimates of equation (3) are presented below in subsection 6.2 in Tables VI and VII. These differentiated amounts are expressed as log(1+Reported IRR) – log(1+Predicted IRR), which could alternatively be interpreted as log((1+Reported IRR)/(1+Predicted IRR)), so that the economic 23
There, we perform an out-of-sample test by comparing the predicted returns of investments for which unrealized returns are reported in our data with the later actual realized returns. We show that our model is fairly robust in explaining this difference in this “out-of-sample” test as well.
22 significance of the coefficients directly indicate the relative degree of overstatement of unrealized returns. Our explanatory variables comprise four categories that proxy for information asymmetry between PE managers and their institutional investors: (1) market and legal conditions, alongside different accounting indices used in recent accounting research on international differences in reporting standards (Bhattacharya et al., 2003, and Burgstahler et al., 2006), (2) PE fund characteristics, (3) entrepreneurial firm characteristics, and (4) transaction specific characteristics.
Table V provides a correlation matrix for the subsample of unrealized investments. The lefthand-side variable in equation (3) is also reported in Table V for Model (1) and in Table VI Panel B for the full sample of all PE investments. For the most representative sample of all data, see Model (1) in Table VI Panel B, where Log(1+IRR Reported)-Log(1+IRR Expected) = Log((1+Reported IRR)/(1+Predicted IRR) has an average value of 4.96 (median of 4.87). In other words, (1+Reported IRR)/(1+Predicted IRR) = exp(4.96) = 143%. This means that from June 2000 – Sept. 2003, PE managers on average reported unrealized IRRs that were 143% of the magnitude predicted by our econometric model (taking into account market conditions at the time of reporting, legal conditions and the characteristics of the investment, as presented in section 5). This value of 143% indicates significant reporting of overvaluations on average, since the value should be 100% if actual and expected returns were equal. See also subsection 4.2 above for summary statistics on realized versus unrealized IRRs.
[Table V About Here]
The accounting indices developed by Bhattacharya et al. (2003) are used in Table V, as well as in Table VI Panels A – C. The first index measures the extent of earnings aggressiveness, which is based on a time series measure of accruals scaled by lagged total assets for each firm (see Bhattacharya et al., 2003, Table I, column 2). Higher earnings aggressiveness index values imply lower quality accounting standards. We also use the disclosure level in a country, based on Bhattacharya et al. (2003, Table I, column 6), Saudagaran and Diga (1997), Table II, and Center for International Financial Analysis and Research (1995); higher figures imply better disclosure. The advantage of these two indices is that they cover almost all of the countries in our sample, and are economically relevant for average audit quality in a given country (which is relevant to PE reporting; see sections 2 and 3).
Table V indicates that the extent to which PE funds overvalue their unrealized IRRs is greater among countries with poor accounting standards. Reporting overvaluations is less pronounced in the
23 period after the introduction of Sarbanes Oxley legislation.24
Table V further indicates that
overvaluation is more pronounced when there is more committed capital on the market and among PE funds that invest in more entrepreneurial firms per investment professional within a PE manager, and less pronounced among PE funds that syndicate their investments and use convertible securities.
One potential concern with the data and empirical strategy is that the IRRs of many unrealized investments are reported at “0.00%” (i.e., reported at cost) to institutional investors. This creates two concerns in our regression analyses in the subsequent section. The first concern is that the difference between actual reported IRR and the predicted IRRs on the unrealized investment would simply be the negative of the predicted IRR, indicating a complete dearth of information of reported valuations. The second concern is that the relationship between accounting standards and (over-) valuations stem from the fact that badly-performing firms are held at their cost valuations (i.e. with an IRR of 0.00%) while the well-performing firm experience a further investment stage leading to higher (market) valuations.25 We therefore consider the robustness of the results by excluding all transactions from the data and regressions where the reported unrealized IRR was “0” (this applied to approximately 25% of our sample). The results (discussed immediately below) are extremely robust with this alternative method,26 as explicitly shown in Table IV Panel C presented below.
To check the robustness of our results we compare those in Table VII and Table VI with accounting indices developed by Burgstahler et al. (2006). We use the same notation and variables as in Burgstahler et al. (2006), which include the following. EM1 – Avoidance of Small Losses – measures the degree to which firms use accounting discretion to avoid reporting losses. The use of this index is consistent with the view that PE funds are often loathe to write down the book value of their investments for a variety of reasons, especially to avoid the appearance of carrying “living dead” investments that will be difficult to sell. EM2 – Magnitude of Total Accruals in relation to Cash Flow from Operations – measures the extent to which firms use reporting discretion to, for example, to boost earnings in years of poor performance. This is particularly relevant for our sample in which PE fund reportings to their institutional investors took place over the period 2000 – 2004. EM3 – Smoothing of Operating Earnings vis-à-vis Cash Flow – refers to the extent to which firms reduce the variability of reported earnings using accruals. This is a relevant measure for PE fund reportings because actual exit transactions might be facilitated by reducing the appearance of earnings variability
24 Note that Sarbanes Oxley is relevant for PE disclosures beyond the US borders since US institutional investors are major players within the international private equity scene. (Our data, however, do not enable an analysis of the extent to which institutional investors are from different countries.) Many managers seeking to raise funds would do their best to meet US regulatory standards. 25 Another counterargument against this concern, besides our findings in Table IVC is that there is no need to keep these firms at cost valuations and that more stringent accounting rules would force PE funds to write these firms (partially) off, while with less stringent accounting standards it is feasible for PE funds to keep the valuations of these firms at their cost-level. 26 We also considered ribbons around “0”, and this likewise did not materially affect the results reported herein.
24 of the investee firm before the actual exit. For instance, in the case of an IPO exit, the apparent riskiness of the firm will be lower where the firm’s prior financial statements (as reported in the firm’s prospectus) exhibit smoother earnings streams. EM4 – Correlation between Accounting Accruals and Cash Flow from Operations – is an alternative measure of earnings smoothing (EM3), as larger magnitudes of this negative measure indicate smoothing of reported earnings unrelated to the firm’s economic performance. Finally, EM (Aggregate) – Aggregate Measures of Earnings Management – is the percentage sum of EM1 to EM4. Note that these accounting indices developed by Burgstahler et al. (2006) are defined for a subset of countries in our sample: Austria, Belgium, Denmark, Finland, France, Germany, Greece, Italy, the Netherlands, Portugal, Spain, Sweden and the UK. As such, the use of these accounting indices facilitates a dual robustness check for (1) a subsample of countries, and (2) different measures of accounting standards throughout various countries. The Burgstahler et al. (2006) accounting standards are also direct pertinent measures for private firms.
[Table VI Panels A – C and Table VII Here]
6.2. Multivariate Econometric Results
The dependent variables for each model in Tables VI and VII correspond to the model figures presented in Table IV. We report two different specifications for each of the Table IV models (the results are very robust for concerns with collinearity, among other things; additional specifications not explicitly presented are available upon request). The variables are as defined in Table I.
Overall, the multivariate econometric analyses indicate that the data are strongly consistent with the hypothesis that valuations of unrealized investments are higher when information asymmetries faced by institutional investors are more pronounced. These information asymmetries are more pronounced in countries with lower accounting standards, and for investors, investees and transactions that are more opaque.
The results in both Tables VI and VII indicate that valuations are overstated by PE funds in countries with less strict disclosure indices, and in countries with higher earnings aggressiveness indices (Bhattacharya et al., 2003)27 (see Table VI). The results are quite robust concerning countries with lower quality accounting standards as developed by Burgstahler et al. (2006) (see Table VII). While we note some limitations of our data in section 7 below, overall there is very strong evidence of a need for a stricter legal and accounting environment in curbing overvaluation of unrealized PE fund 27 Note that the legal indices are all expressed in logs, with the exception of the earnings aggressiveness index. The reason for this is as follows. The values in this index are small, typically negative, fractions (see Bhattacharya et al., 2003, Table I). We considered conversion into logs (with arbitrary rescaling to make a log transformation possible). The estimates in logs (available upon request) yielded very similar results without qualitative differences in interpretation of any of the results.
25 returns reported by PE managers to their institutional investors. More stringent accounting rules reduce significantly the incentives (and possibilities) to exaggerate unrealized IRRs. This, in turn, makes the valuations more informative, thereby benefiting the industry as a whole.
We had expected a direct and indirect link between accounting standards and overvaluations. The direct empirical link between accounting standards and reporting biases may exist where accounting standards affect the quality of financial statements and hence firm valuation (as in Armstrong et al., 2005, and Hand, 2005). The indirect empirical link between accounting standards and reporting biases exists when the opaqueness of earnings is negatively correlated with cultural and legal forces that together create an environment in which PE funds face higher incentives to overvalue their unrealized investments and a smaller probability that such overvaluations will be noticed by institutional investors and auditors. The data clearly indicate a negative relation between accounting standards and firm valuation reported by insiders in the absence of market prices.
Although
accountants have for centuries reported the value of net assets rather than firm value, recently the US, the UK, Australia, and the European Union, for example, have issued standards requiring recognition of balance sheet amounts at fair value, and changes in their fair values in income (Barth and Landsmann, 1995; SFAS 141 and 142).
Our evidence of PE fund disclosures of unrealized
investments indicates that insiders should have minimal discretion in valuing assets and/or liabilities for which there is a limited market, particularly in countries with weak accounting standards.
It should be noted, too, that reporting of overvaluations was much more pronounced before the introduction of the Sarbanes Oxley legislation in July 2002 (Table IV tests this effect explicitly; as well, Table VII, too, presents consistent evidence with yearly fixed-effect regressions). This is further evidence of a need for a strong regulatory environment in mitigating PE fund over-statement of unrealized IRRs to their institutional investors.
Aside from the accounting standards which give rise to external information opaqueness, many of the other categories of variables are significant in ways that are similarly related to (internal) information opaqueness proxied by the PE fund and entrepreneurial firm characteristics. Valuations reported by less experienced PE managers are significantly higher. This can be attributed to the fact that older more experienced PE managers have more reputational capital at stake whereas younger PE managers have an incentive to signal with higher valuations thereby increasing the probability of closing the next fund.28 Overvaluation of unrealized IRRs is more pronounced among PE funds with larger portfolios per investment professional within the PE manager, among early-stage and expansion-stage investee firms, firms in high market/book industries and certain industries in
28 This is in line with the findings of Gompers and Lerner (1999) who provide empirical evidence that reputation drives the use of covenants between VC funds and investors.
26 particular (such as the Internet), and among smaller firms. The results pertaining to portfolio size are consistent with theoretical work of Kanniainen and Keuschnigg (2003b) in that funds with larger portfolios add less value to their investees, tend to be less profitable (see sections 4 and 5), and therefore have a greater incentive to overstate valuations.
The findings pertaining to stage of
development and industry factors are consistent with Gompers and Lerner (1999); informational asymmetries are more pronounced with earlier stage high tech firms, thereby providing greater scope for exaggerating unrealized returns.
The data also indicates less overvaluation among syndicated investments and when the PE fund holds a convertible security that provides periodic cash flows. PE managers are less inclined to report overly high valuations when their actions might be monitored (and therefore potentially revealed) by other syndicated investors. The use of convertible securities facilitates more robust firm valuations (Brennan and Kraus, 1987).
Overvaluation is more pronounced in weak public markets. PE funds are loathe to devalue the book value of their investee firms, as a write off would signal negative information to the new owner(s) of the investee firm at the time of actual exit. As indicated in Table II and Figure 1, unrealized returns are rarely less than 0.00%.
Overall, the data support the notion that overvaluation of unrealized IRRs takes place where information asymmetries faced by the institutional investor are more pronounced. In the context of our international sample, it is perhaps most important that the data indicate that this reporting of overvaluations is more pronounced in countries with lower quality accounting standards. We showed the robustness of this latter result in relation to different accounting indices recently developed in the literature, and to different countries. We nevertheless acknowledge that there are limitations in our analyses and that there is ample scope for further research, as outlined in the next section.
7. Limitations, Alternative Explanations and Future Research
We consider our analysis to be a sound step towards understanding the driving forces behind the returns of PE investments and regard it as a first contribution to the analysis of reporting of unrealized investments. Due to the structure of our data, there are some limitations to our analysis as well as scope for future research along these lines. First, some of unexplained variance of our IRR measure is without doubt due to deal-specific characteristics in the relationship between the PE fund and their entrepreneurial firms. Since this would, however, require much more detailed data than that we have here, this is beyond the scope of the present analysis.
27 Second, one might argue that some of our RHS variables (such as syndication) might be endogenous. A potential cure for this, such as an instrumental variable approach, is however hampered by the fact that we do not have any exogenous instruments which are uncorrelated with the IRRs but correlated with the potentially endogenous variables. Since we are, however, also interested in the effect of the investment characteristics on a PE fund’s returns per se, we do not consider this to be such a problematic issue. We do not drop these variables, as they have been used, for example, to explain the performance of VC-backed IPOs and the ensuing returns to VCs (Barry et al., 1990; Megginson and Weiss, 1991; Gompers and Lerner, 1999). Failure to consider these variables may result in a more serious problem of omitted variables with respect to endogeneity.
Third, we did not differentiate among the various exit channels chosen. This is due to the fact that our information on exit channels chosen is not very complete. With the partial set of information available for those variables over some of the observations, we did not find selection effects that materially give rise to biases in our reported results.
Furthermore, there are at least three potential routes of future research which are rather straightforward. First, it would be of interest to investigate the impact of overvaluation of unrealized IRRs on subsequent fundraising. With our data we do know that there is a positive correlation between fund size and overvaluation, but this does not imply causality, as the fundraising is related to the funds raised subsequent to the reporting of the overvaluation. This dimension could be considered alongside other factors that affect fundraising in future work. It would also be interesting to link reporting biases with subsequent investment activities of the fund. These issues, however, are beyond the scope of our paper.
Second, it would be very interesting to track actual ultimate performance to reported performance. This would allow us to look into the overvaluation issue more deeply by not only comparing predicted with reported valuation, but by looking at the actual developments of the reported valuation. Given the structure of our dataset and the confidential data, this is not possible for the full sample of our data. Nevertheless, this is possible for a subsample of 80 observations (investee firms) in our sample. We report this information and carry out backtesting on this subsample in the Appendix. The results are extremely similar to that which we have reported in section 6.
Finally, it could always be interesting to follow up the sample with additional data from other time periods. Our data focuses on disclosure in the aftermath of the Internet bubble; other time periods, as the data become available, could be of interest. Our data are the first ever to consider the reporting of IRRs on unrealized investments by PE managers to their institutional investors. This disclosure issue is of fundamental practical importance in the VC and PE industries around the world
28 (see notes 2 – 6 for references and a discussion), and whether or not the degree of over-exaggeration reported herein applies in other periods could be a useful avenue for further research.
8. Concluding Remarks
By using a unique dataset to analyze the performance of PE investments around the world, we investigated the determinants of IRRs of realized investment as well as the relative valuations of unrealized investments as reported to institutional investors. Due to the availability of the entire cash flow series for each investment of the respective PE fund we were able to calculate the precise profitability of the investment for realized investments. We did not have to rely on a return proxy based on the first inflow and last cash outflow as in prior research. The cash flow data further indicates functional characteristics of securities that provide periodic cash flows and upside potential, and this makes a significant difference to measuring returns as well as to the degree of overvaluation of unrealized returns. In short, the richness of our data enables us to account for different entrepreneurial firm characteristics, as well as differences in transaction structures. The global nature of the data set makes it possible to investigate potentially important aspects of economic and financial set-ups, rules and institutions and their impact on PE returns and reporting behaviour.
Since our data sample comprises a rather long period of time (1971-2003) we are able to avoid too strong a focus of the recent boom and bust years but are able to derive a broader picture. The volume of data at different points in time is consistent with the overall market and its development over time and throughout the different countries.
Our empirical approach makes use of Heckman selection procedures to correct for potential selection biases in our IRR data of realized investments. We are able to explain a high degree (up to 36%) of the total variation in the IRR among the different observations. Previous papers on this topic have accounted for between 1% and 13% of the variation in returns on US VC investments. Our data indicate the importance of selection effects in realized versus unrealized returns, as well as for full versus partial exits.
Our main empirical findings mostly support our theoretical considerations. Most importantly we find with respect to the realized IRRs that more advice (and monitoring) provided by the PE manager and the use of incentive-compatible financial instruments (convertible securities) contribute to a significant increase of the IRRs of realized returns. These findings, as well as the positive impact of the legality index are robust throughout our different modelling specifications.
29 To the best of our knowledge we are the first to look at the valuation behaviour of PE funds with respect to their unrealized investments. Especially in the light of recent industry developments, this is an important issue with immediate policy implications. Our empirical analysis supports our previous theoretical results. The main findings show that especially less experienced PE managers as well as those involved in early-stage investments are more eager to overvalue. In contrast, syndication proves to lower the incentives of PE funds to overstate the value of their unrealized investments. Most importantly, from our point of view, is the very robust and significant impact of accounting standards and legal framework on the reporting behaviour of PE managers. Less stringent accounting rules and weak legal systems clearly seem to facilitate overvaluation thereby decreasing the information content of these valuations. There are two implications. The first one concerns policy making and the behaviour of the PE industry as a whole. More stringent accounting standards are in the interest of institutional investors, PE funds and the economy as a whole (inducing the provision of more risk capital if the communication between institutional investors and PE funds is less distorted). The second issue is to what extent the overvaluation strategy is successful with respect to fund raising, and therefore distorts the allocation of capital throughout PE funds and throughout countries. The relationship between the reporting of valuations of unrealized investments and future fund raising is a very important question. Future research on this topic is warranted.
Appendix: Backtesting Reported IRRs versus Subsequent Realized IRRs
In this appendix we compare the actual realized IRRs to the unrealized reported IRRs for a subsample of 80 observations (investee firms) from 11 countries for which both the realized and unrealized reported IRR are known.29 The reported IRRs were between 2000 and 2003, while the realized IRRs were between 2001 and 2005. The average [median] duration between the reported IRR and the realized IRR was 2.6 years. The average [median] unrealized reported IRR was 219.71% [2.56%], while the average [median] subsequently realized IRR was 98.46% [8.70%], and for this subsample our predicted average [median] IRR (based on Model (1) of Table IV Panel B) was 15.22% [7.75%]. The correlation between out-of-sample average realized IRRs and our predicted IRRs is 0.45, which is significant at the 1% level. It is important to note that the average reported unrealized IRR was more than 100% higher than the average realized IRR. It is further important to note that while the average level predicted by our model is less than the out-of-sample realized IRR, there is a high positive correlation between our predicted IRR and the out-of-sample realized IRR.
In Table VIII we present regressions that analyze the differences between the reported unrealized IRRs and predicted IRRs (as per the methodology reported in section 6 of this paper) and
29 The 80 investee firms are from Canada, Finland, France, Germany, Israel, Norway, Spain, Sweden, the Netherlands, the UK, and the US.
30 the reported unrealized IRRs and the subsequently realized IRRs. We use the subsample of the 80 investments for which this information is known.
Model (A1) presents OLS estimates of the
determinants of the difference between the unrealized reported IRRs and the predicted IRRs, which is analogous to the regressions Model (1b) in Table VI Panel B (although there are fewer included righthand-side variables in Model (A1) due to the comparative dearth of observations). In Models (A2), (A3) and (A4) in Table VIII, the dependent variable is the difference between unrealized reported IRRs and subsequently realized IRRs for the same investments. There are 80 observations; that is, 80 investee firms which have subsequently been realized.
[Table VIII About Here]
Table VIII indicates very similar findings to those reported in section 6. In Model (A1), which is presented to provide a basis of comparison to the section 6 results for the subsample of 80 observations, notice that the results are analogous to that reported in section 6. Unrealized reported IRRs are significantly higher than that which we would expect for countries with weaker accounting standards (for a higher disclosure index), for high-tech industries with high industry market/book values, and for non-syndicated investments and investments for which convertible securities were not used (Model (A1)). Further, notice in Models (A2), (A3) and (A4) that unrealized reported IRRs are similarly higher than the subsequently realized IRRs for investments with weaker accounting standards (for a higher earnings aggressiveness index) and investments for which convertible securities were not used.
The econometric results in Table VIII do not perfectly overlap for the prediction model and the actual realizations for the subsample of 80 firms for which we can make this comparison. Nevertheless, the main qualitative results are analogous and robust to alternative specifications. Overvaluation is more pronounced among countries with weak accounting standards, and certain characteristics of the investment such as the use of convertible securities which mitigates the tendency for PE managers to report overvalued unrealized investments to their institutional investors.
References
Armstrong, C., Davila, A., and Foster, G. (2005) Venture-backed private equity valuation and financial statement information, Review of Accounting Studies 11, 119-135. Ball, R., and Shivakumar, L. (2005) Earnings quality in U.K. private firms comparative loss recognition timeliness, Journal of Accounting & Economics 39, 83-128. Barry, C.B., Muscarella, C.J., Peavy III, J.W., and Vetsuypens, M.R. (1990) The role of venture capitalists in the creation of public companies: evidence from the going public process, Journal of Financial Economics 27, 447-471.
31
Barth, M and Landsmann, W (1995), Fundamental issues related to using fair value accounting for financial reporting, Accounting Horizons 9, 97-107. Bascha, A., and Walz, U. (2001) Convertible securities and optimal exit decisions in venture capital finance, Journal of Corporate Finance 7, 285-306. Berglöf, E. (1994) A control theory of venture capital finance, Journal of Law, Economics and Organization 10, 247-267. Berkowitz, D., Pistor, K., and Richard, J.F. (2003) Economic development, legality, and the transplant effect, European Economic Review 47, 165-195. Beuselinck, C., Deloof M., and Manigart, S. (2005) Private equity investments and disclosure policy, Mimeo. Bhattacharya, U., Daouk H., and Welker, M. (2003) The world price of earnings opacity, The Accounting Review 78, 641-678. Black, B.S., and Gilson, R.J. (1998) Venture capital and the structure of capital markets: banks versus stock markets, Journal of Financial Economics 47, 243-277. Bottazzi, L., M. Da Rin and Hellmann, T. (2006). What is the role of legal systems in financial intermediation? Theory and evidence, Mimeo. Brander, J.A., Amit, R. and Antweiler, W. (2002) Venture capital syndication: improved venture selection versus the value-added hypothesis, Journal of Economics and Management Strategy 11, 423-452. Brennan, M.J., and Kraus, A. (1987) Efficient financing under asymmetric information, Journal of Finance 42, 1225-1243. Burgstahler, D., L. Hail and Leuz, C. (2006) The importance of reporting incentives: earnings management in European private and public firms. The Accounting Review, forthcoming. Center for International Financial Analysis & Research (1995) International Accounting and Auditing Trends, 4th Edition. Edited by V. B. Bavishi. Princeton, NJ: CIFAR. Casamatta, C., (2003) Financing and advising: optimal financial contracts with venture capitalists, Journal of Finance 58, 2059 - 2086. Cochrane, J., (2005) The risk and return to venture capital, Journal of Financial Economics 75, 3-52. Cumming, D.J., (2006) The determinants of venture capital portfolio size: empirical evidence, Journal of Business 79, 1083-1126. Cumming, D.J., and MacIntosh, J.G. (2003) A cross-country comparison of full and partial venture capital exits, Journal of Banking and Finance 27, 511-548. Cumming, D.J., Fleming, G., and Schwienbacher, A. (2006). Legality and venture capital exits, Journal of Corporate Finance 12, 214-245 Das, S., Jagannathan, M., and Sarin, A. (2003) Private equity returns: an empirical examination of the exit of venture-backed companies, Journal of Investment Management 1, 152-177.
32
Fischer, P., and Verrecchia , R.E., (2004) Disclosure Bias, Journal of Accounting and Economics , 223-250 Gigler, F. (1994) Self-enforcing voluntary disclosures, Journal of Accounting Research 32, 224-240. Gompers, P.A. and Lerner, J. (1997) Risk and reward in private equity investments: the challenge of performance assessment, Journal of Private Equity (Winter 1997): 5-12. Gompers, P.A. and Lerner, J. (1999) The Venture Capital Cycle, Cambridge: MIT Press. Gompers, P.A., and Lerner, J. (2000) Money chasing deals?: The impact of fund inflows on the valuation of private equity investments, Journal of Financial Economics 55, 281-325. Gompers, P., Lerner, J., and Desai, M. (2003) Institutions, capital constraints, and entrepreneurial firm dynamics: evidence from Europe, Mimeo. Greene, W. (2002) Limdep Econometric Software, Econometric Software, Inc. Hand, J.R.M., (2005) The value of relevance of financial statements within and across private and public equity markets, The Accounting Review 80, 613-648. Healy, P., and Palepu, K. (2001) Information asymmetry, corporate disclosure, and the capital markets: a review of the empirical disclosure literature, Journal of Accounting and Economics 31, 405-440. Heckman, J. (1976) The common structure of statistical models of truncation, sample selection, and limited dependent variables and a simple estimator for such models, Annals of Economic and Social Measurement 5, 475-492. Heckman, J. (1979) Sample selection bias as a specification error, Econometrica 47, 153-161. Hege, U., Palomino, F., and Schwienbacher, A. (2003) Determinants of venture capital performance: Europe and the United States, Mimeo. Ho, S.S.M., and Wong, K.S. (2001) A study of the relationship between corporate governance structures and the extent of voluntary disclosure, Journal of International Accounting Auditing and Taxation 10, 139-156. Hochberg Y., Ljungqvist, A.P., and Lu, Y. (2006) Whom you know matters: Venture capital networks and investment performance, Journal of Finance, forthcoming. Hsu, D. (2004) What do entrepreneurs pay for venture capital affiliation? Journal of Finance 59, 1805-1844. Inderst, R., and Mueller, H. (2004) The effect of capital market characteristics on the value of start-up firms, Journal of Financial Economics 72, 319-356. Jeng, L.A., and Wells, P.C. (2000) The determinants of venture capital funding: evidence across countries, Journal of Corporate Finance 6, 241-289. Kanniainen, V., and Keuschnigg, C. (2003a) The optimal portfolio of start-up firms in venture capital finance, Journal of Corporate Finance 9, 521-534.
33 Kanniainen, V., and Keuschnigg, C. (2003b) Start-up investment with scarce venture capital support, Journal of Banking and Finance 28, 1935-1959. Kaplan, S.N., and Schoar, A. (2005) Private equity returns: persistence and capital flows, Journal of Finance 60, 1791-1823 Keuschnigg, C., (2004) Taxation of a venture capitalist with a portfolio of firms, Oxford Economic Papers 56, 285-306 Keuschnigg, C., (2003) Optimal public policy for venture capital backed innovation, Mimeo. Keuschnigg, C. and Nielsen, S.B. (2001) Public policy for venture capital, International Tax and Public Finance 8, 557-572. Keuschnigg, C. and Nielsen, S.B. (2003a) Tax policy, venture capital and entrepreneurship, Journal of Public Economics 87, 175-203. Keuschnigg, C. and Nielsen, S.B. (2003b) Taxes and venture capital support, Review of Finance 7, 515-538. La Porta, R., Lopes-de-Silanes, F. Shleifer, A. and Vishny, R. (1997) Legal determinants of external finance, Journal of Finance 52, 1131-1150. La Porta, R., F. Lopez-De-Silanes, F. Shleifer, A. and Vishny, R. (1998) Law and finance, Journal of Political Economy 106, 1113-1155. Lerner, J., Hardymon, F. and Leamon, A. (2004) Between a rock and a hard place: valuation and distribution in private equity, Harvard Business School Case 9-803-161. Lerner, J., and Schoar, A. (2005) Does legal enforcement affect financial transactions? The contractual channel in private equity, Quarterly Journal of Economics 120, 223-246. Lerner, J., Schoar, A. and Wong, W. (2006) Smart institutions, foolish choices?: The limited partner performance puzzle, Journal of Finance, forthcoming. Ljungqvist, A.P., and Richardson, M.P. (2003a) The investment behavior of private equity fund managers. Mimeo. Ljungqvist, A.P., and Richardson, M.P. (2003b) The cash flow, return and risk characteristics of private equity. Mimeo. Mayer, C., Schoors, K. and Yafeh, Y. (2005) Sources of funds and investment activities of venture capital funds: evidence from Germany, Israel, Japan and the UK, Journal of Corporate Finance 11, 586-608. Megginson, W.L., and Weiss, K.A. (1991) Venture capitalist certification in initial public offerings, Journal of Finance 46, 879-903. Mitchell, F., Reid, G.C. and Terry, N. (1995), Post investment demand for accounting information by venture capitalists, Accounting and Business Research 25, 186-196. Moskowitz, T., and Vissing-Jorgensen, A. (2002) The returns to entrepreneurial investment: a private equity premium puzzle? American Economic Review 92, 745-778.
34 Neus, W., and Walz, U. (2005) Exit timing of venture capitalists in the course of an initial public offering, Journal of Financial Intermediation 14, 253-277. Phalippou, L., and Zullo, M. (2005) Performance of private equity funds: another puzzle? Mimeo. Puhani, P.A., (2000) The Heckman correction for sample selection and its critique, Journal of Economic Surveys 14, 53-68. Repullo, R., and Suarez, J. (2004) Venture capital finance: a security design approach, Review of Finance 8, 75-108. Sahlman, W.A. (1990) The structure and governance of venture capital organizations, Journal of Financial Economics 27, 473-521. Saudagaran, S., and Diga, J. (1997) Financial reporting in emerging capital markets: characteristics and policy issues, Accounting Horizons 11, 41-64. Schmidt, K.M. (2003) Convertible securities and venture capital finance, Journal of Finance 58, 1139-1166. Schwienbacher, A. (2003) An empirical analysis of venture capital exits in Europe and the United States, Mimeo. Verecchia, R.E. (1983) Discretionary disclosure, Journal of Accounting and Economics 5, 179-194.
35
Figure 1. Histograms of Fully Realized, Partially Realized and Unrealized IRRs
1200
1000
800
# Entrepreneurial Firms
600 400 Fully Realized IRRs
200
Partially Realized IRRs 0 -100 -50
0
50
Unrealized IRRs 100
IRR (%)
200
300
400
500
1000 5000
Figure 2. Number of Investments by Year
900
800
700
600
500 Number of Investments 400
300
200
100
0 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 Year
36 Table I. Definition of Variables Variable Internal Rate of Return (IRR)
Description The exact IRR based on all current discounted cash flows (for unrealized investments the reported valuation is used as the last cash flow).
Market and Legal Factors MSCI Return
The country-specific MSCI return over the contemporaneous investment period.
Risk Free Return
The US risk-free return over the contemporaneous investment period. (Euro and other risk-free returns were also considered but did not materially change the results.)
Committed Capital Overall Market at Date of Investment
The industry total committed venture capital in the overall US market (as reported by Venture Economics) in the year of investment. This variable is a proxy for deal flow competition. The measure is from the US and not specific countries in the data to avoid correlation with the Legality index.
Legality Index
Weighted average of following factors (based on Berkowitz et al., 2003): civil versus common law systems, efficiency of judicial system, rule of law, corruption, risk of expropriation, risk of contract repudiation, shareholder rights (as per La Porta et al., 1998). Higher numbers indicate 'better' legal systems.
Country Earnings Aggressiveness Index
Bhattacharya et al.'s (2003, Table I) measure of firms and accountants tendency in a nation to incorporate economic gains in a more timely fashion than economic losses (i.e., the opposite of accounting conservatism). Higher numbers indicate more aggressive earnings reporting (i.e., less conservative accounting practices).
Country Disclosure Level Index Private Firm Accounting Indices Sarbanes Oxley
Bhattacharya et al.'s (2003, Table I) accounting disclosure measures per country. The higher the number, the better the disclosure. Burgstahler et al.’s (2006, Table I) accounting standards for earnings management in private and public firms in the European Union. The higher the number, the worse the accounting disclosure. EM1=Avoidance of Small Losses. EM2=Magnitude of Total Accruals relative to Cash Flow from Operations. EM3=Smoothing of Operating Earnings vis-à-vis Cash Flow; EM4=Correlation between Accounting Accruals and Cash Flows from Operations; EM Aggregate = Average of the Percentage Ranks from EM1 – EM4. A dummy variable equal to one for PE fund disclosures of unrealized IRRs to institutional investors after July 30, 2002 (and in year 2003 in the data), and zero otherwise (the disclosures first started in the 2000 data).
Fund Characteristics Fund Number in the PE Firm Age of Specific PE Fund Portfolio Size (# Investees) / # General Partners Entrepreneurial Firm Characteristics Seed Stage
The number of PE funds the PE firm had operated prior to this current fund. The age (in days) of the PE fund. The number of entrepreneurial firms in the PE fund / the number of investment professionals of the general partners in the fund.
A dummy variable equal to 1 for financing provided to research, assess and develop an initial concept.
Start-up Stage
A dummy variable equal to 1 for financing provided to firms for initial product development and marketing. Firms may be in the process of being set up or may have been in business for a short time, but have not sold their product commercially.
Early Stage
A dummy variable equal to 1 for financing provided to firms with product in testing and/or pilot production. The firm may or may not be generating revenue, and has usually been in business less than 30 months.
Expansion Stage
A dummy variable equal to 1 for financing provided to firms in need of development capital. The financing is provided for the growth and expansion of a firm, which may or may not break even or trade profitably. Capital may be used to: finance increased production capacity; market or product development; provide additional working capital.
Late Stage
A dummy variable equal to 1 if the firm has reached profitable operating levels.
MBO/MBI
A dummy variable equal to 1 for buyout financing either of MBO or MBI form. MBO: A buyout in which external managers take over the firm. Financing is provided to enable a manager or group of managers from outside the target firm to buy into the firm with the support of private equity investors. MBI: A buyout in which the target’s management team acquires an existing product line or business from the vendor with the support of private equity investors.
LBO
A dummy variable equal to 1 for a buyout in which the new firm's capital structure incorporates a particularly high level of debt, much of which is normally secured against the firm’s assets.
Publicly Listed Firm
A dummy variable equal to 1 for firms with a listing on a stock exchange.
Industry Market / Book
The industry market/book ratio for the firm's primary industry.
Industry Dummy Variables
Dummy variables equal to 1 for the firm's primary industry.
Country Dummy Variables
Dummy variables equal to 1 for the firm's country of primary residence.
Year of Exit Dummy Variables
Dummy variables equal to 1 for the year of exit.
Investment Characteristics Lead Investment
A dummy variable equal to 1 if the investor was the lead investor, 0 if not the lead investor, and 0.5 if unknown.
Syndicated Investment
A dummy variable equal to 1 if the investment was syndicated, 0 if not syndicated, and 0.5 if unknown.
Co-Investment
A dummy variable equal to 1 if the investment was co-invested (2 or more PE funds in the same PE firm investing in the same entrepreneurial firm), 0 if not a co-investment, and 0.5 if unknown.
PE Board Seat(s)
A dummy variable equal to 1 if the investor had board seat(s), 0 if no board seats, and 0.5 if unknown.
Convertible Security with Actual Periodic Cash Flows
A dummy variable equal to 1 if the investor held a convertible security which functionally provided for periodic cash flows back to the investor prior to exit. The standard deviation of the cash flows provided to the entrepreneur from the investor, scaled by (divided by) the initial investment amount. The standard deviation is calculated in the usual way based on all flows of funds between the PE fund and entrepreneurial firm; for example, a little amount of capital in one round combined with a massive amount in the next round increases the risk measure. Smaller initial investments are also deemed riskier with the scaling (but the results herein are robust to different scaling measures). The initial investment value (in real 2003 US dollars).
Standard Deviation of Cash Flows to Entrepreneur / Initial $ Invested Initial Amount Invested
37
Table II. Summary Statistics
PE Fund Characteristics Part A
All Funds
1
All Funds in the Data
Part B
Market and Legal Factors
Unrealized / Partially realized Ent Firm Investments # Ent Average Median Firms IRR IRR
Fully Realized Ent Firm Investments # Ent Average Median Firms IRR IRR
2619
63.23
0.00
2419
68.67
Difference Tests Means
Medians
16.99
0.22
p 3.5%
611
76.88
9.32
1908
58.07
20.21
-1.14
p
