In this section we examine returns because an ex-post analysis of returns provides guidance as to how discount factors in valuations might vary depending on the characteristics of the investment. To begin, it is important to point out that performance statistics vary widely in terms of data quality. Performance indicators are generally not made available to the public, and therefore large representative samples are hard to obtain. Details vary depending on the number of transactions, the years of the transactions, and the different variables in the dataset examined.
Cochrane (2005) examines the VentureOne database from its beginning in 1987 to June 2000, which consists of 16,613 financing rounds, with 7,765 investee firms. That data enable one to compare market returns to venture capital fund returns while controlling for selection biases in terms of which realized returns are observed. Cochrane estimates that the average log return is 15% per year and finds a beta (the covariance between market returns and venture capital fund returns divided by the variable of market returns) to be 1.7 and the alpha (the performance of venture capital fund returns above that which would otherwise be predicted by the Capital Asset Pricing Model [CAPM], which accounts for systematic risk) to be 32%.
The statistics in reference to venture capital fund returns presented by Cochrane are interesting, but the dataset used cannot account for features of venture capital that are central to what it is that makes venture capital fund investment distinct from other types of investment. That is, the dataset used provides scant details in terms of venture capital fund characteristics, entrepreneurial firm characteristics, and investment structure characteristics. To be able to value a venture capital fund investment, ideally we would like to analyze more than alphas and betas in a CAPM framework. Venture capital funds hold a small nondiversified portfolio of entrepreneurial firms and pay great attention to the structure of their investment by writing detailed contracts to
mitigate agency costs and idiosyncratic risks. It is worthwhile to assess how returns vary pursuant to evaluating these details to come up with a more accurate portrayal of value drivers.
Arguably the most comprehensive and detailed venture capital and private equity dataset collected to date is the CEPRES (Center for Private Equity Research) dataset, which is based out of Goethe University of Frankfurt, Germany. The dataset has been described in prior work, including Cumming et al. (2004), Cumming and Walz (2004), and Nowak et al. (2004). The sample contains cash flow information at the level of the individual investment for 5,038 portfolio firms in 221 private equity funds spanning a time period of 33 years (1971 to 2003). The data indicate very detailed information including all cash flow between venture capital funds and entrepreneurial firms, 4 fund characteristics (such as age, fund number, portfolio size per fund manager), entrepreneurial firm characteristics (such as stage of development at the time of first investment and industry), and investment characteristics, such as whether the fund returns are observed by the lead venture capital fund investor in a syndicate, the presence of syndicated investors, co-investment from the same fund managers operating a different fund, board seats, the use of convertible securities, amounts invested, and the standard deviation of cash flows
Based on the CEPRES data, Cumming and Walz (2004) identified a number of factors that potentially influence the performance of venture capital and private equity investments. The data are presented for internal rates of return (IRRs) that are fully realized versus IRRs that are yet to be realized but reported to institutional investors, are summarized in Table 22.1 .
5 Table 22.1 further indicates differences in mean and median IRRs for realized and unrealized IRRs for different funds, entrepreneurial firm and investment characteristics, and differences across countries and in different economic conditions. Part A summarizes all of the investments in the dataset. Part B considers differences for across different market conditions and legal standards in different countries. Part C considers fund-specific characteristics, Part D considers entrepreneurial firm characteristics, Part E considers investment characteristics, and Part F considers country differences.
Cochrane (2005) examines the VentureOne database from its beginning in 1987 to June 2000, which consists of 16,613 financing rounds, with 7,765 investee firms. That data enable one to compare market returns to venture capital fund returns while controlling for selection biases in terms of which realized returns are observed. Cochrane estimates that the average log return is 15% per year and finds a beta (the covariance between market returns and venture capital fund returns divided by the variable of market returns) to be 1.7 and the alpha (the performance of venture capital fund returns above that which would otherwise be predicted by the Capital Asset Pricing Model [CAPM], which accounts for systematic risk) to be 32%.
The statistics in reference to venture capital fund returns presented by Cochrane are interesting, but the dataset used cannot account for features of venture capital that are central to what it is that makes venture capital fund investment distinct from other types of investment. That is, the dataset used provides scant details in terms of venture capital fund characteristics, entrepreneurial firm characteristics, and investment structure characteristics. To be able to value a venture capital fund investment, ideally we would like to analyze more than alphas and betas in a CAPM framework. Venture capital funds hold a small nondiversified portfolio of entrepreneurial firms and pay great attention to the structure of their investment by writing detailed contracts to
mitigate agency costs and idiosyncratic risks. It is worthwhile to assess how returns vary pursuant to evaluating these details to come up with a more accurate portrayal of value drivers.
Arguably the most comprehensive and detailed venture capital and private equity dataset collected to date is the CEPRES (Center for Private Equity Research) dataset, which is based out of Goethe University of Frankfurt, Germany. The dataset has been described in prior work, including Cumming et al. (2004), Cumming and Walz (2004), and Nowak et al. (2004). The sample contains cash flow information at the level of the individual investment for 5,038 portfolio firms in 221 private equity funds spanning a time period of 33 years (1971 to 2003). The data indicate very detailed information including all cash flow between venture capital funds and entrepreneurial firms, 4 fund characteristics (such as age, fund number, portfolio size per fund manager), entrepreneurial firm characteristics (such as stage of development at the time of first investment and industry), and investment characteristics, such as whether the fund returns are observed by the lead venture capital fund investor in a syndicate, the presence of syndicated investors, co-investment from the same fund managers operating a different fund, board seats, the use of convertible securities, amounts invested, and the standard deviation of cash flows
Based on the CEPRES data, Cumming and Walz (2004) identified a number of factors that potentially influence the performance of venture capital and private equity investments. The data are presented for internal rates of return (IRRs) that are fully realized versus IRRs that are yet to be realized but reported to institutional investors, are summarized in Table 22.1 .
5 Table 22.1 further indicates differences in mean and median IRRs for realized and unrealized IRRs for different funds, entrepreneurial firm and investment characteristics, and differences across countries and in different economic conditions. Part A summarizes all of the investments in the dataset. Part B considers differences for across different market conditions and legal standards in different countries. Part C considers fund-specific characteristics, Part D considers entrepreneurial firm characteristics, Part E considers investment characteristics, and Part F considers country differences.
Although venture capital fund managers are obliged to periodically provide reports of the value of the unexited investments to their institutional investors, it has to be noted that by the very fact that the investments that are to be valued as“ unrealized, ” their valuation by the venture capital fund managers are by all accounts subjective and extremely difficult to corroborate. Venture capital fund managers therefore may have an incentive to misreport valuations to their institutional investors for various obvious reasons, most notable of which is to attract more capital for investment.
Hege et al. (2003), Cumming and Walz (2004), and others focus on IRRs in performance measurement of venture capital investments, despite the fact that IRRs are subject to manipulation (see, e.g., Damodaran, 2006), 6 for a number of reasons. Cumming and Walz explain that perhaps the most important reason is that the venture capital and private equity funds in the CEPRES sample do report IRR for realized and unrealized investments. As Cumming and Walz show, the fund managers do at times manipulate unrealized IRRs in their reports to institutional investors. Hence, it is appropriate to consider IRR because this is what is reported to the institutional investors. It is likewise particularly important to assess robust in regress estimates of factors that influence IRRs, by considering alternative adjusted metrics of performance and various causal mechanisms that
influence performance, as well as multistep selection effects (Cumming and Walz, 2004). Herein we do not go into details beyond summarizing the CEPRES data, since we believe this is an area of evolving research. We do, however, summarize factors that plausibly influence performance that are present in the CEPRES data, as well as discuss in the next subsection other factors that could be considered in valuation and performance that are not present in any current dataset.
The data reported in Table 22.1 indicate that realized IRRs are significantly higher when the country-specific Morgan Stanley Capital International ( “ MSCI ” ) index returns over the contemporaneous investment period have been higher. Accounting for selection biases as to which firms are fully exited, Cumming and Walz estimate the beta to be approximately 1.45 for all venture capital and private equity investments, which is slightly less than Cochrane’s (2005) beta estimate of 1.7 for venture capital fund investments in the United States. The data in Table 22.1 indicate that the average (median) realized IRR is 58.07% (20.21%) when MSCI returns over the contemporaneous period have been greater than 3.5%. When MSCI returns are less than 3.5%, the average (median) realized IRR is 108.24% ( –10.99%). For unrealized investments, the average (median) IRR is 76.88% (9.32%) when MSCI returns over the contemporaneous period have been above 3.5%. When MSCI returns are less than 3.5%, the average (median) realized IRR is 59.07% (0.00%). It is noteworthy that with the very high-standard deviations in IRRs, differences in average values are statistically insignificant. However, differences in medians are statistically significant. Median realized IRRs are higher in times of better market conditions, but median unrealized IRRs are higher in times of worse market conditions. In other words, venture capital fund managers tend to overreport the value of their unexited investments in times of worse market conditions relative to what one might otherwise expect given existing market conditions (Cumming and Walz, 2004).
It is interesting to note that a variety of fund characteristics are related to fund IRRs. Based on the Venture Economics data from 1980 to 2001, Kaplan and Schoar (2005) estimate that U.S. fund managers with better performance by 1% on a prior fund achieve a 77-basis-point better performance on the subsequent fund, demonstrating persistence in private equity performance over time. Given this persistence in performance, entrepreneurs tend to prefer financing from more reputable venture capital fund managers. Based on a hand-collected sample (from a dataset collected by MIT) of 148 entrepreneurs, Hsu (2004) estimates that entrepreneurs are three times more likely to accept a financing offer from a high-reputation venture capital fund manager, and these high-reputation venture capital fund managers acquire equity from the investee entrepreneurial firms at a 10 to 14% discount. As well, Lerner et al.(2007) find that different institutional investors tend to be systematically better at selecting fund managers that achieve superior performance results (or have better access to such fund managers).
In Table 22.1 , Part C, we present the data from Cumming and Walz (2004) by fund characteristics. The data do not indicate specific trends in performance based on the age of the fund manager or the number of funds operated by the fund manager. The data do, however, indicate that the fund manager is much more likely to perform better where portfolio size in terms of the number of investee entrepreneurial firms per fund manager is smaller. The regression estimates in Cumming and Walz (2004) are consistent in showing portfolio size per manager as a very important factor in explaining realized IRRs. A change in portfolio size per manager from 10 to 20 investee entrepreneurial firms, for example, is associated with an expected reduction in realized IRRs by 10%. This evidence is consistent with the findings presented in Chapters 16 and 18 (as well as the theoretical work of Kanniainen and Keuschnigg, 2003, 2004; Keuschnigg, 2004b; and Bernile et al., 2007) that venture capital fund managers with larger portfolios provide less advice to the investee entrepreneurial firms.
Hege et al. (2003), Cumming and Walz (2004), and others focus on IRRs in performance measurement of venture capital investments, despite the fact that IRRs are subject to manipulation (see, e.g., Damodaran, 2006), 6 for a number of reasons. Cumming and Walz explain that perhaps the most important reason is that the venture capital and private equity funds in the CEPRES sample do report IRR for realized and unrealized investments. As Cumming and Walz show, the fund managers do at times manipulate unrealized IRRs in their reports to institutional investors. Hence, it is appropriate to consider IRR because this is what is reported to the institutional investors. It is likewise particularly important to assess robust in regress estimates of factors that influence IRRs, by considering alternative adjusted metrics of performance and various causal mechanisms that
influence performance, as well as multistep selection effects (Cumming and Walz, 2004). Herein we do not go into details beyond summarizing the CEPRES data, since we believe this is an area of evolving research. We do, however, summarize factors that plausibly influence performance that are present in the CEPRES data, as well as discuss in the next subsection other factors that could be considered in valuation and performance that are not present in any current dataset.
The data reported in Table 22.1 indicate that realized IRRs are significantly higher when the country-specific Morgan Stanley Capital International ( “ MSCI ” ) index returns over the contemporaneous investment period have been higher. Accounting for selection biases as to which firms are fully exited, Cumming and Walz estimate the beta to be approximately 1.45 for all venture capital and private equity investments, which is slightly less than Cochrane’s (2005) beta estimate of 1.7 for venture capital fund investments in the United States. The data in Table 22.1 indicate that the average (median) realized IRR is 58.07% (20.21%) when MSCI returns over the contemporaneous period have been greater than 3.5%. When MSCI returns are less than 3.5%, the average (median) realized IRR is 108.24% ( –10.99%). For unrealized investments, the average (median) IRR is 76.88% (9.32%) when MSCI returns over the contemporaneous period have been above 3.5%. When MSCI returns are less than 3.5%, the average (median) realized IRR is 59.07% (0.00%). It is noteworthy that with the very high-standard deviations in IRRs, differences in average values are statistically insignificant. However, differences in medians are statistically significant. Median realized IRRs are higher in times of better market conditions, but median unrealized IRRs are higher in times of worse market conditions. In other words, venture capital fund managers tend to overreport the value of their unexited investments in times of worse market conditions relative to what one might otherwise expect given existing market conditions (Cumming and Walz, 2004).
It is interesting to note that a variety of fund characteristics are related to fund IRRs. Based on the Venture Economics data from 1980 to 2001, Kaplan and Schoar (2005) estimate that U.S. fund managers with better performance by 1% on a prior fund achieve a 77-basis-point better performance on the subsequent fund, demonstrating persistence in private equity performance over time. Given this persistence in performance, entrepreneurs tend to prefer financing from more reputable venture capital fund managers. Based on a hand-collected sample (from a dataset collected by MIT) of 148 entrepreneurs, Hsu (2004) estimates that entrepreneurs are three times more likely to accept a financing offer from a high-reputation venture capital fund manager, and these high-reputation venture capital fund managers acquire equity from the investee entrepreneurial firms at a 10 to 14% discount. As well, Lerner et al.(2007) find that different institutional investors tend to be systematically better at selecting fund managers that achieve superior performance results (or have better access to such fund managers).
In Table 22.1 , Part C, we present the data from Cumming and Walz (2004) by fund characteristics. The data do not indicate specific trends in performance based on the age of the fund manager or the number of funds operated by the fund manager. The data do, however, indicate that the fund manager is much more likely to perform better where portfolio size in terms of the number of investee entrepreneurial firms per fund manager is smaller. The regression estimates in Cumming and Walz (2004) are consistent in showing portfolio size per manager as a very important factor in explaining realized IRRs. A change in portfolio size per manager from 10 to 20 investee entrepreneurial firms, for example, is associated with an expected reduction in realized IRRs by 10%. This evidence is consistent with the findings presented in Chapters 16 and 18 (as well as the theoretical work of Kanniainen and Keuschnigg, 2003, 2004; Keuschnigg, 2004b; and Bernile et al., 2007) that venture capital fund managers with larger portfolios provide less advice to the investee entrepreneurial firms.
Posts
