Factors That Affect Venture Capital Realized Returns and Reported Unrealized Returns 2

  Part D of Table 22.1 indicates differences in IRRs depending on stage of development and industry. Recall the development stages that were defined in Chapter 1.  Earlier-development stages are associated with lower median realized IRRs but higher average IRRs, which means that there is much greater variance in IRRs and higher potential upside with earlier stages of investment.  Similarly, for high-tech investments (in industries with higher market/book values), average IRRs are higher but median IRRs are lower, which likewise reflects greater variance in IRRs and greater upside potential in high-tech industries. In the dataset there are 14 realized investments from firms that are publicly listed, and their IRRs are very high. The investments appear to be IPO
allocations to venture capital funds for which the fund managers were able to flip the investment shortly after the IPO for a substantial capital gain (see Chapter 19 on IPO underpricing). Notice that start-up stage investments tend to have the greatest unrealized valuations. Valuation standards (Chapter 7) typically suggest that valuations not deviate from zero for recent investments, particularly at the seed stage.  But where it is possible to indicate an appreciation in investment values, this is often done at the start-up stage. Both average
and median valuations of start-up and early-stage investments are significantly larger than the realized IRRs for those investment stages.
  Part E of Table 22.1 indicates differences in IRRs depending on specific investment characteristics. IRRs of realized investments tend to be higher where unaffiliated funds syndicate, while co-investments by affiliated funds tend to show lower IRRs. Recall from Chapter 5 that funds often prohibit co-investment of fund capital in firms which have obtained prior funding from affiliated funds managed by the fund manager because fund managers may have an incentive to use new capital from recently raised funds to bail out the bad investments of prior affiliated funds (e.g., a low IRR buyback exit would look better than a write-off). Regression analyses in Cumming and Walz (2004) show that syndicated investments tend to yield 73% higher IRRs than on syndicated investments, while co-investments tend to yield 33% lower IRRs than non-co-investments. The relation between IRRs and board seats and convertible securities that enable periodic cash flow back to the venture capital fund prior to exit tends to be sensitive to the econometric specification in Cumming and Walz (2004). One explanation is that the use of the convertibles and board seats is endogenous to expected exit outcomes.  Convertible securities with periodic cash flow tend to be positively associated with IRRs, but only for positive IRRs; in other words, cash flows back to the venture capital fund periodically where the entrepreneur is expected to be able to pay back the investor. Board seats tend to be associated with IRRs that are less than zero; in other words, venture capital fund managers are more likely to sit on boards of firms that are performing poorly. Finally, smaller initial investment amounts tend to be associated with lower IRRs. The econometric evidence in Cumming and Walz (2004) shows that an increase in the initial investment from US $ 1 million to US $2 million tends to be associated with an increase in IRRs by 3%, while an increase in the initial investment from US $19 million to US $ 20 million tends to be associated with an increase in IRRs by 0.2%.
  Part F of Table 22.1 reports the data by country and legal origin. IRRs may also vary depending on legal standards in different countries, although it is difficult to predict the relationship between IRRs and legal conditions    On one hand, given risks are more pronounced in countries with inferior legal standards, IRRs may be greater in countries with inferior legal standards in order to compensate for such risks. On the other hand, higher legal standards are associated with lower information asymmetry and lower agency costs that
enhance the efficiency of advice provided by venture capital fund managers to entrepreneurs and mitigate venture capital fund -entrepreneur conflicts (see also Chapter 16). As such, we might expect a positive relation between legal standards and IRRs. The data indicate that English legal origin countries have experienced the highest median realized IRRs (17.49%), while German legal origin countries experienced the lowest median realized IRRs (10.95%).
Tests for differences in medians (rows 96 –101), however, are statistically significant for differences in medians between English and French legal origin.
There are no statistically significant differences in means across legal origins, and this finding is explained by the high variability in returns, consistent with Cochrane’s (2005) evidence for the United States. There are no statistically significant differences in medians across legal origins for unrealized returns.
Mean unrealized returns are highest in German legal origin countries (89.97%), but differences in mean unrealized returns are not significant for German legal origin relative to other legal origin countries (and again, this is due to the high variance). Mean English legal origin unrealized returns are 54.25%, and significantly higher than mean French legal origin unrealized returns (19.10%) and mean Scandinavian legal origin unrealized returns (14.10%). Further, it is noteworthy that for all legal origins, median unrealized returns are lower than median realized returns. Overall, therefore, Part F of Table 22.1 indicates legal origins and country-specific factors do not appear to play as great a role in driving differences in means and medians, at least relative to market and legal factors (Part B of Table 22.1 ), fund characteristics (Part C), portfolio firm characteristics (Part D), and investment characteristics (Part E). In particular, it is noteworthy that the differences in Legality and accounting standards (Part B, rows 6 –11) appear to be stronger drivers of differences in realized and unrealized returns than the legal origins variables in Part F of Table 22.1 .
  The available evidence to date in fact indicates that IRRs are positively correlated with legal conditions. Cumming and Walz (2004) show that median realized IRRs are higher in countries with Legality indices above 20, but mean IRRs are higher in countries with Legality indices below 20. In other words, there is greater variability in IRRs in countries with lower legal standards. Cumming and Walz (2004) show a positive relation between IRRs and legal standards with a dataset from 39 countries and over 5,000 transactions, even after controlling for selection effects and other factors in multistep multivariate regression analyses. Similarly, Lerner and Schoar (2005) analyze 210 investments in 25 developing countries and show a positive relation between post money valuation and legal conditions. Cumming et al. (2006) show a positive relation between legal conditions and the probability that a venture capital–backed firm will exit via an IPO rather than a private sale or write-off, and this effect is robust to statistical selection effects.
  It is also noteworthy in Table 22.1 that the mean unrealized IRRs are higher in countries with lower legal standards (while the median unrealized IRR is slightly higher in countries with Legality indices above 20). Cumming and Walz (2004) show that unrealized IRRs tend to be higher than what we would otherwise expect based on factors that drive realized IRRs. As well, for a subsample of their data, Cumming and Walz (2004) can compare actual realized IRRs with previously reported unrealized IRRs and find legal conditions are a
key factor in explaining venture capital fund managers tendency to overvalue unrealized IRRs when reporting to their institutional investors.
  It is also noteworthy that Cumming and Walz (2004) find evidence that the structure of the investment tends to influence the propensity to overreport valuations on unexited investments. Valuations tend to be overreported for nonsyndicated investments and for co-investments. The intuition is that fund managers are less prone to exaggerate unexited investments when there is a syndicated investor that is potentially independently valuing the investment. By contrast, co-investments involve the same venture capital fund manager and, as discussed, there are agency problems with co-investment that would be consistent with a tendency to overvalue the investment. Convertible securities with periodic cash flows are less likely to be overvalued because the stream of cash flow does in fact make it easier to value the investment (since the valuation does not strictly rest on the estimate of the exit value). Finally, larger investments tend to have a shorter investment horizon until exit and valuations are less opaque. InTable 22.1 we note that the valuation average IRRs of unexited investments of less than US $2,500,000 are 91.8%, while average realized IRRs of exited investments of less than US $2,500,000 are 63.5%, and average IRRs of unexited investments of more than US $2,500,000 are only 34.6%, while average realized IRRs of exited investments of more than US $ 2,500,000 are 75.6%.