Key Takeaways
- • Kaplan and Schoar's 2003 working paper (published in the Journal of Finance in 2005) found the average private equity fund's net return roughly matched the S&P 500, with an equal-weighted PME of 0.96 and a capital-weighted PME of 1.05, while the median fund trailed at 0.74 and 0.82.
- • The paper's headline finding was persistence: a manager whose prior fund landed in the top third had a 55% chance of repeating there, versus roughly one in three by chance.
- • A 2020 follow-up by Harris, Jenkinson, Kaplan and Stucke found little or no evidence of persistence for buyout funds when measured with information available at fundraising; only 57% of buyout funds reporting top-quartile results at fundraising ended up top-quartile (63% for post-2000 funds).
- • Venture capital persistence remains in the 2020 study, though it weakened after 2000.
- • Fees are the central drag: Phalippou and Gottschalg estimated fees at about 6% a year, turning a roughly 3% gross annual outperformance into a roughly 3% underperformance net of fees.
Summary
A 2003 working paper by Steven Kaplan of the University of Chicago and Antoinette Schoar of MIT Sloan remains widely cited in debates over whether private equity deserves a place in institutional portfolios. The authors analyzed 746 mostly liquidated funds from Venture Economics data, largely started before 1995, of which about 78% were venture capital and 22% buyout. Using the public market equivalent (PME), which compares fund cash flows against the S&P 500, they found the average fund roughly matched the index. The equal-weighted average PME was 0.96 and the capital-weighted figure was 1.05. Venture capital funds averaged 1.21 on a capital-weighted basis and buyout funds 0.93. The median fund trailed, at 0.74 equal-weighted and 0.82 capital-weighted, and results varied widely, with the 25th percentile at 0.45 and the 75th at 1.14.
The paper's most influential result concerned repeatability. A 1% better result in a manager's prior fund was associated with a 54 basis point better result in the next fund. The authors also noted that better-performing funds did not grow as fast as average ones, and that most limited partners describe top funds as oversubscribed. The authors acknowledged limits, including voluntary reporting, a sample tilted toward larger funds, and the absence of risk adjustment. They also lacked fund-level fee data and relied on typical terms of 1.5% to 2.5% annually plus 20% of profits.
A 2020 NBER working paper by Harris, Jenkinson, Kaplan and Stucke, which Kaplan co-authored, complicates that picture. The authors noted that the Venture Economics data used earlier had since been shown to be flawed. Using cash-flow records for 2,222 U.S. funds with vintages from 1984 through 2014, they confirmed persistence when measuring final results. However, committees choosing managers see only interim results, and on that basis only 57% of buyout funds reporting top-quartile results at fundraising ended up top-quartile. Using fundraising-date information, the study found little or no evidence of buyout persistence, including after 2000. Venture capital persistence remained, though it declined after 2000.
Fees feature prominently in the debate. Phalippou and Gottschalg, in the Review of Financial Studies (2009), estimated that after corrections funds beat the S&P 500 by about 3% a year before fees but trailed it by about 3% after fees, with fees estimated at about 6% a year. Metrick and Yasuda found roughly two-thirds of managers' expected revenue came from fixed components. Phalippou's 2020 paper estimated about $230 billion in performance fees collected by 2006–2015 vintage funds as of end-2019.
The counter-evidence is significant. In a 2014 study using Burgiss data on nearly 1,400 funds, Harris, Jenkinson and Kaplan found the average buyout fund outperformed the S&P 500 net of fees by more than 3% a year. The 2020 persistence study also reported average net PME of 1.20 for buyouts and 1.22 for venture capital across 1984–2014 vintages. Researchers continue to disagree on benchmarks, treatment of unsold investments, and risk adjustment. The consensus is narrower: results vary widely, fees are large, and the evidence for picking tomorrow's winners from yesterday's record is weak for buyouts, though venture capital persistence endures.
The paper's most influential result concerned repeatability. A 1% better result in a manager's prior fund was associated with a 54 basis point better result in the next fund. The authors also noted that better-performing funds did not grow as fast as average ones, and that most limited partners describe top funds as oversubscribed. The authors acknowledged limits, including voluntary reporting, a sample tilted toward larger funds, and the absence of risk adjustment. They also lacked fund-level fee data and relied on typical terms of 1.5% to 2.5% annually plus 20% of profits.
A 2020 NBER working paper by Harris, Jenkinson, Kaplan and Stucke, which Kaplan co-authored, complicates that picture. The authors noted that the Venture Economics data used earlier had since been shown to be flawed. Using cash-flow records for 2,222 U.S. funds with vintages from 1984 through 2014, they confirmed persistence when measuring final results. However, committees choosing managers see only interim results, and on that basis only 57% of buyout funds reporting top-quartile results at fundraising ended up top-quartile. Using fundraising-date information, the study found little or no evidence of buyout persistence, including after 2000. Venture capital persistence remained, though it declined after 2000.
Fees feature prominently in the debate. Phalippou and Gottschalg, in the Review of Financial Studies (2009), estimated that after corrections funds beat the S&P 500 by about 3% a year before fees but trailed it by about 3% after fees, with fees estimated at about 6% a year. Metrick and Yasuda found roughly two-thirds of managers' expected revenue came from fixed components. Phalippou's 2020 paper estimated about $230 billion in performance fees collected by 2006–2015 vintage funds as of end-2019.
The counter-evidence is significant. In a 2014 study using Burgiss data on nearly 1,400 funds, Harris, Jenkinson and Kaplan found the average buyout fund outperformed the S&P 500 net of fees by more than 3% a year. The 2020 persistence study also reported average net PME of 1.20 for buyouts and 1.22 for venture capital across 1984–2014 vintages. Researchers continue to disagree on benchmarks, treatment of unsold investments, and risk adjustment. The consensus is narrower: results vary widely, fees are large, and the evidence for picking tomorrow's winners from yesterday's record is weak for buyouts, though venture capital persistence endures.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.