An Intolerance of Failure? Evidence from U.S. Private Equity
In our 2018 Diverse Asset Management Study (found here), we show that the asset management industry is characterized by extremely low levels of diversity in ownership. Overall, diverse-owned managers - defined as asset managers with significant female or ethnic minority ownership - control just 1.3% of all US-based managers’ assets . Our results also indicate that the lack of diversity is unlikely to be explained by differences in fund performance. In particular, the data show no statistically meaningful difference between the performance of diverse- and non-diverse-owned funds. Motivated by these findings, the Bella Private Markets team, together with the John S. and James L. Knight Foundation, began researching a new explanation: that investors are more likely to curtail capital commitments to underperforming diverse-owned funds than to non-diverse funds.
According to this “intolerance of failure” theory, low levels of diversity might be explained by a double-standard in the way investors react to fund manager performance. Specifically, investors may be more likely to reduce their investments with poorly performing diverse-owned managers than with non-diverse owned managers with the same performance. To the extent that diverse-owned managers face higher “penalties for failure” compared with their non-diverse peers, such firms may struggle to survive during times of low returns.
To test this theory, the Bella team gathered data from Preqin , a leading provider of information on alternative investments, to see how past performance affects future fundraising outcomes for US-based diverse-owned and non-diverse-owned private equity (PE) fund managers. We measure an asset manager’s performance as the average (excess) multiple of its prior funds . A “multiple” reflects how many times an investor receives her initial investment (after fees) in cash distributions from a fund. If an intolerance of failure exists, any dip in performance should reduce a diverse-owned manager’s ability to fundraise by more than would the same performance decline for a non-diverse-owned manager. In other words, we would find that the fundraising activities of non-diverse managers are less sensitive to dips in performance.
First, we need to isolate how past performance affects a manager’s ability to raise a new PE fund. Using regression analysis, we can “hold constant” other factors that may influence the relationship between performance and fundraising - such as fund size, fund type, and general economic conditions. For example, larger funds might, on average, have lower returns and be raised in periods with better fundraising outcomes. In this case, failure to account for fund size would cause us to underestimate the true penalty for failure.
To quantify fundraising ability, we look at how a decrease in past performance impacts a manager’s likelihood of raising a new PE fund in a given year. Figure 1 (below) shows the results of our analysis. We find that, on average, a 1-unit decrease in performance is associated with a 24.5 and 14.3 percentage point reduction in the probability of raising a new fund for women- and minority-owned managers. To get a sense of the magnitude, such a decrease in performance may be interpreted as the difference between a manager raising average funds and a manager raising funds returning one multiple below the average. In contrast, the same dip in performance only leads to a 3.1 percentage point decrease in the change of raising a new fund for non-diverse-owned managers. The evidence suggests that the fundraising activities of women-and minority-owned managers are more sensitive to past performance.
Second, we estimate a more general version of our model that simply asks whether underperforming PE firms are less likely to raise a follow-on fund in a given year. An “underperforming” manager has started PE funds that perform below the average for funds of the same vintage year. On average, we find that an underperforming non-diverse-owned manager is 9.6 percentage points less likely to raise a new fund compared to an overperforming non-diverse-owned manager. The estimated “penalties for underperformance” for women-and minority-owned managers were larger - at 27.9 and 17.3 percentage points, respectively. We should note, however, that the difference in the underperformance penalty between diverse-owned and non-diverse-owned managers is only statistically meaningful for women-owned managers. That is, the 27.9 percentage point penalty for underperformance by women-owned funds is highly unlikely to have occurred randomly, while there is a possibility that the 17.3 percentage point penalty for minority-owned managers may be due to chance.
Finally, we ask whether a manager’s past performance affects the amount of capital raised in the next fund. Specifically, we looked at the relationship between our firm-level performance metric and fund size growth, which measures the percentage change in the amount of capital raised between a manager’s prior fund and their current fund. Because of the limited number of women-and minority-owned firms, we combine these two groups in a single diverse ownership category . We find that, on average, reducing a diverse-owned manager’s performance by 1-unit is associated with a 53-percentage point decrease in fund size growth relative to diverse owned fund managers. The results suggest that even if poorly performing diverse-owned managers succeed in raising a new fund, they may struggle to raise as much capital as their non-diverse peers.
As we compare these results with those of our earlier study of diverse representation, we see an intriguing conundrum. Not only are there fewer diverse-owned firms that manage proportionately less capital than their representation in the population would suggest, but investors may be less likely to give them “the benefit of the doubt.” Poor performance by diverse-owned managers appears to be penalized more severely relative to similar non-diverse peers not only in terms of the likelihood of raising a subsequent fund but also with respect to the amount of capital that the manager can raise.
It is important to note that our research does not suggest that an intolerance of failure is the only explanation for the lack of diverse ownership in asset management. Rather, we view it as one of many potential causes. For example, some researchers have suggested gender differences in the willingness to compete, in opportunities for mentorship, and in occupational choices as potential explanations for underrepresentation . While such explanations are potentially important drivers of the lack of diverse ownership, we leave their exploration, and the formulation of potential policy solutions for underrepresentation, for future research.
Click here to view the full report, containing all of our findings. To learn about all of our industry insights and updates, subscribe below to receive our free monthly email newsletter.
 Private equity managers are considered women-owned (minority-owned) if at least 50% of the firms’ equity is held by women (minorities).
 Source: www.Preqin.com
 We measure an asset manager’s performance by looking at how its results across all its previous funds compare to similar funds. To do this, we use the “excess multiple,” or the difference of a given fund’s performance from the average. We calculate an “excess multiple” by computing the average multiple of funds in the same vintage year and subtracting that from a given fund’s actual multiple. A positive excess multiple indicates “above average” performance, while a negative excess multiple indicates “below average” performance.
 A PE manager is considered diverse-owned if it is women-owned, minority-owned, or both.
 Sutter, Matthias. Glätzle-Rützler, Daniela. “Gender Differences in the Willingness to Compete Emerge Early in Life and Persist. Management Science. Volume 61, Issue 10. October 2015. Pages iv-vii, 2281-2547. See also, Polachek, Solomon William. “Occupational Self-Selection: A Human Capital Approach to Sex Differences in Occupational Structure.” Review of Economics and Statistics 63. 1981. Pages 60-69.