Buzz

The Returns to Private Equity: An Update

Jason Draho, IPOAdvisory


Estimating the returns to private equity is necessary to evaluate the entire asset class as an investment opportunity and the performance of individual fund managers. This subject has been the focus of a growing body of academic research in the past couple of years, some of which was already reviewed in an earlier Buzz article (see The Returns to Private Equity, July 22, 2003).

A few general conclusions can be drawn from those studies. First, private equity funds, both venture capital and buyout, generated returns that were roughly equal to the returns of public equity. Subsequent studies have continued to find the same result. Second, there was persistence in the performance of funds started by the same private equity firm, another fact since corroborated. That is, a follow-on fund was more likely to do well if the preceding fund had generated good returns, and vice versa for poorly performing funds. This pattern implies that experience and demonstrated ability account for quite a bit in predicting future returns. Third, funds started by new firms in vintage years with the heaviest capital inflows subsequently performed the worst. Such inexperienced funds resulted in too much money chasing too few good deals.

Since the first Buzz article, additional studies have examined other aspects of the risk and return properties of private equity. Many of the problems that plague any effort to measure private equity returns - data availability, the flaws with IRR, and valuing unrealized investments - continue to exist. Notwithstanding these general concerns, some of the more interesting findings are presented below.

Risk and Return Characteristics

Assessing private equity as an asset class based solely on the total returns generated provides an incomplete picture of its virtues. Private equity can diversify the portfolios of limited partners if the returns have low correlation with other assets. Thus, even if the returns to private equity are no better than those generated by public stocks, the overall portfolio risk can be lowered without adversely affecting the expected returns.

The most common way to measure an asset's affect on the overall portfolio risk is its beta, which measures how the returns systematically vary with the market return. Estimating a beta for private equity investments is difficult because regularly updated share prices needed to estimate how the returns co-vary with the market are not observed. In spite of this difficulty, some researchers have tried to measure private equity betas.

Two general conclusions emerge from these studies, with the caveat that the findings are only suggestive because of the estimation problems. The first is that private equity betas appear to be fairly close to one and consequently offer relatively limited hedging benefits. Oliver Gottschalg, Ludovic Phalippou and Maurizio Zollo estimated betas for VC and buyout funds to be 1.2 and 1.7, respectively, although the betas had weak explanatory power for the cross section of fund returns. As an alternative, they "marked-to-market" investments; each investment was assumed to return what the market or industry returned over the same period. The beta equivalents against the market IRR using this method were 0.84 and 0.97 for VC and buyout funds, respectively. Frank Kerins, Janet Kilholm Smith and Richard Smith estimated the average betas for recent IPOs in eight technology-oriented industries to be in the 0.7 to 1.25 range, and they were usually less than one.

The second major conclusion is that betas tend to increase as the firm ages and grows, and goes from a private entity to a public company. John Hand estimated that the average beta for biotechnology stocks was only 0.28 in the VC market, but increased to 1.38 in the public market. Kerins, Smith and Smith documented a similar trend for other technology stocks. These patterns suggest that as companies mature they transition from risk that is highly idiosyncratic to risk that is more closely related to the market.

Under the assumption that only systematic risk is relevant for pricing securities, which is the defining property of CAPM, it is not surprising that private equity, with a beta near or just below one, has generated returns roughly equal to the overall market. Yet the preceding paragraph provided evidence that company-specific idiosyncratic risk does affect the returns. Gottschalg, Phalippou and Zollo indeed found that idiosyncratic risk is priced in the returns. The number of investments made by VC and buyout funds, a proxy for the amount of diversification, was negatively related to the returns. In addition, funds that concentrated their investments in a particular industry or investment type produced better returns. Thus, funds that were willing to take on more idiosyncratic risk were rewarded with higher total returns, a result that generally does not hold in the public equity market.

Factors Related to the Returns

Differing types and degrees of risk are not the only factors that affect private equity returns. Specific characteristics of the macroeconomic environment and the contractual arrangement between the VC and the portfolio company have been found to have explanatory power for the cross section of returns.

The macro factors pertain to both the legal environment and the general economic conditions. Douglas Cumming and Uwe Walz examined VC returns in 39 countries and found that the realized IRRs were positively and significantly related to the soundness of the legal conditions in a country, proxied by a legality index. Gottschalg, Phalippou and Zollo demonstrated the effect of economic conditions. VC funds that invested during an economic boom significantly underperformed, possibly because venture deals were too expensive. However, VC funds that invested during periods of unexpectedly high GDP growth outperformed. For buyout funds, they found that the main determinant of performance was the interest rate paid on the very high debt level. Buyout funds that invested during periods of high corporate interest rates, usually occurring in a boom, significantly underperformed. Consequently, buyout funds appear to do better in an economic slump. Stock market fluctuations had relatively little effect on either VC or buyout funds.

The ability and skill of a VC to add value to his or her portfolio company directly translates into higher returns. Gottschalg, Phalippou and Zollo reported that funds that made fewer investments, invested in only a single type of deal (either VC or buyout) or a single industry strongly outperformed. Likewise, Cumming and Walz found that the number of portfolio companies per VC manager was negatively related to the realized IRRs. The ability of the VC to provide valuable advice to the entrepreneurs is constrained when their attention is spread too thin. The effect was not as pronounced at later stage investments, consistent with the claim that most value is created at the early stage. Syndicated investments and the use of convertible securities were associated with significantly higher IRRs. Thus, value can be created through more intensive monitoring and the allocation of contingent control to the VC.

Venture Capital Returns in Europe

Most of the established estimates for the returns to private equity are based on the performance of US funds. This largely reflects the fact that the US private equity market was far more developed than its European counterparts throughout the 1980s and 1990s. Two recent studies have examined venture capital returns in Europe, although the findings are only suggestive because of the limited available data.

A safe assessment is that VC funds either based in Europe or investing exclusively in European countries have performed far worse than US-oriented funds. Gottschalg, Phalippou and Zollo found that exclusive European funds returned about half as much as US-invested funds. Ulrich Hege, Frederic Palomino and Armin Schweinbacher examined funds in Belgium, Germany, Sweden, France, the Netherlands and the UK. They too found a significant gap in performance between US portfolio firms and their European counterparts, both in terms of type of exit and rates of return.

The latter study examined some of the specific differences in the return patterns to US and European funds and proposed explanations for their divergence. The total length of an investment was strongly negatively related to the corresponding return in the US, but the relationship was almost as strongly positively in Europe. The same pattern held for the average duration between financing rounds. Furthermore, portfolio firms that received a larger percentage of their total VC financing in the first round produced better returns. These patterns are consistent with US VCs doing a better job of screening out the good projects initially and pushing them to a successful exit event sooner.

The gap in performance could be attributed to differences in the contractual relationships between the VCs and entrepreneurs. VCs in the US were much more assertive in demanding financial contracts - usually convertible securities - that transfer residual control rights to them in case of poor performance and they activated contingent control more frequently, measured by the replacement of entrepreneurs and the termination of projects. There was a positive correlation between the control exercised by the VCs and the success of the investment. Overall, it would appear that European VCs are more deal-makers than they are active monitors. They still seem to be lagging in their ability to select projects and provide value-added services to their portfolio companies.

Valuing Unrealized Investments

A major challenge to estimating fund IRRs is how to value investments that have not yet being exited. The IRRs that fund managers report to Venture Economics and their own LPs are highly contingent on the method used to value unrealized investments. Three common methods are used to value unrealized investments, which, in order, result in increasingly larger reported IRRs. The first option is to assign active investments a value of zero, which does downwardly bias the reported IRRs. The second method is to use the book value for unexited investments. Finally, unrealized investments can be marked-to-market, which involves the greatest degree of subjectivity. Since fund managers are more inclined to mark-to-market when doing so increases the asset value, this can result in an overestimation of the true IRR.

Without a uniform standard for reporting the IRRs of unrealized investments, especially across countries, the incentives of fund managers can lead to a mis-reporting of active investments. Experienced and prestigious GPs have more to lose if they over-report the unrealized IRRs, measured in terms of damage to their reputation. Such a concern is not as strong a deterrent for younger VCs, who actually have an incentive to overstate the IRRs if they are trying to raise a second fund.

Cumming and Walz tested these hypotheses by comparing the reported IRRs on unexited investments in 39 countries to the predicted return, based on analysis of realized investments. They found systematic positive biases in the reporting of returns relative to what was expected. Experienced VCs tended to report significantly lower valuations than their younger counterparts. Overreporting by VCs was worse in countries that have the lowest legality indices, lowest disclosure indices, and the most aggressive earnings indices. The negative correlation between overreporting and the soundness of the legal system thus implicitly captures the deterrent effect that reputation concerns have on the incentive to overreport.

Other patterns were also documented for overreporting. Early stage and high-tech unexited investments were valued higher than what was predicted. Funds with larger portfolios per general partner were more inclined to overreport. These funds were less profitable and may have felt the need to overstate the unrealized returns. Overreporting was less extensive in syndicated investments, supporting the notion that different VCs provide a check on each other's actions. Overreporting was more extensive when the public equity markets performed poorly, suggesting that VCs are very reluctant to mark down investments. The takeaway is that LPs should remain suspicious when the GPs report the IRRs of their current funds, using some of these findings to pinpoint the most likely culprits of overreporting.


References

Douglas Cumming and Uwe Walz (2004), "Private Equity Returns and Disclosure around the World," University of Alberta working paper.

Oliver Gottschalg, Ludovic Phalippou and Maurizio Zollo (2003), "Performance of Private Equity Funds: Another Puzzle?" Insead working paper.

John Hand (2004), "Determinants of the Returns to Venture Capital Investments," University of North Carolina working paper.

Ulrich Hege, Frederic Palomino and Armin Schweinbacher (2003), "Determinants of Venture Capital Performance: Europe and the United States," LSE working paper No. 001.

Frank Kerins, Janet Kilholm Smith and Richard Smith (2003), "Opportunity Cost of Capital for Venture Capital Investors and Entrepreneurs," Claremont University working paper.

Appendix

The sources and details of the data used in each study are presented below.

Douglas Cumming and Uwe Walz examined the returns from 221 VC and private equity funds, spanning 32 years (1971-2003) and 39 countries. The data is from the Center for Private Equity Research (CEPRES) in Frankfurt, Germany.

Oliver Gottschalg, Ludovic Phalippou and Maurizio Zollo used PE fund data from Venture Economics and VentureXpert. It covered funds raised from 1980 to 2002.

John Hand used data from 154 US biotechnology companies that went public over the period 1992-2003.

Ulrich Hege, Frederic Palomino and Armin Schweinbacher used data constructed from the VentureXpert database provided by Venture Economics. They had a total of 147 firms in the European sample and 234 firms in the US sample.

Frank Kerins, Smith and Richard Smith used firms from eight technology-oriented industries that went public during 1995 to 2000, resulting in 2,623 firm-year observations.