Persistence in venture capital returns

You've backed a firm whose debut venture fund ranks as a top performer. Should you also invest in its next fund? Andrew Conner analysed the evidence.

Investors in private equity often value gaining access to the best venture capital firms. Building a portfolio of “top-quartile” funds is a common goal. Both of these ideas implicitly assume that new funds raised by the managers of the previous best funds will themselves be superior.

In this article, we test the validity of the concept of persistence in venture capital returns. We compare the performance of one-time funds to that of fund families. We answer the question of whether venture capitalists improve their performance in subsequent funds or if they are, in fact, at their best when they make up a new team that is hungry to prove their abilities. Finally, we determine whether successor funds to top-quartile funds are more likely than average to be top-quartile funds themselves.

THE DATASET
We analysed the funds in the Venture Economics database. Our dataset included mature venture capital funds in the database from vintage years 1980-1995. As we have shown elsewhere, funds aged at least eight years are sufficiently mature to enable us to draw general conclusions regarding performance. Although from the Venture Economics database it is impossible to discern the identity of individual venture capital funds and/or firms, it is possible to link individual funds to the firms that managed them via a unique identifier. Our dataset included a total of 485 venture capital firms and 645 funds.

EXHIBIT 2: AGE WHEN RAISING A FOL LOW-ON FUND

1 Year 2 Years 3 Years 4 Years 5 Year 6 Years 7 Year 8 Year
Proportion of funds in sample 18% 25% 24% 16% 11% 4% 1% 1%

EXHIBIT 3: SUMMARY STATISTICS OF SINGLE AND MULTI-FUND GROUPS

Standard
Standard Deviation Deviation of
Mean IRR Median IRR of IRR Mean TME Median TME TME
All Funds In Sample (n=645) 13.4% 8.5% 25.6% 2.19 1.68 2.20
Single-Fund Group (n=367) 16.4% 9.3% 31.7% 2.37 1.69 2.72
Multi-Fund Group (n=278) 9.3% 7.8% 13.1% 1.95 1.66 1.19
Predecessor Funds (n=160) 7.4% 7.1% 10.1% 1.82 1.65 1.00
Successor Funds (n=160) 10.5% 7.9% 14.9% 2.03 1.63 1.35

EXHIBIT 4: VENTURE CAPITAL FUND PERFORMANCE BY ROMAN NUMERAL

1 II III IV V VI
Average Fund IRR 7.7% 8.4% 15.0% 16.3% 16.3% 0.9%
Average Fund TME 1.83 1.89 2.32 2.54 2.30 1.10
n 118 110 39 8 2 1

CONCLUSION
In this research brief we study persistence in venture capital returns. Our results show that stand-alone funds perform fairly well and that first-time funds that raise follow-on funds tend to improve their performance in successive funds. Additionally, exceptional performance in a predecessor fund tends to be a useful indicator of similar performance in a successor fund. Follow-on funds raised by fund managers whose prior fund had a top-quartile IRR had a 44 percent chance of achieving a top-quartile IRR and those with a bottom quartile successor fund had a 48 percent chance of repeating their bottom quartile IRR, two statistical results that support the existence of persistence in the returns of venture capital funds with greater than 99 percent confidence.

NOTES
1. From Wolfram research: The binomial distribution gives the discrete probability distribution of obtaining more than n successes out of N Bernoulli trials (where the result of each Bernoulli trail is true with probability p) equals: which follows the incomplete Beta distribution: In this case, random variability (no persistence) would dictate that a top-quartile fund be followed by a top-quartile fund with 25% probability. Given that 21 successors of our 48 topquartile predecessors were also top quartile, we have the probability of greater than 20 successes equals: Alignment Capital Group (www.alignmentcapital. com) is a full-service private equity consulting firm based in Austin, Texas. Andrew Conner is an Associate with Alignment Capital Group. His responsibilities include performing due diligence on investment managers, providing strategic portfolio management advice and conducting original research.