First to simulate PE fee structures on actual net cash flows and translate the fee stream into an annualized return equivalent (about 6 % p.a.). Introduced an Alpha measure (which precedes Direct Alpha by 15 years) and first to compute a Beta-adjusted PME and an effective fund duration (similar to bond duration). Documented a strong negative correlation between fund duration and measured performance.

First to estimate the Beta of different types of private-equity funds. Uses the identifying condition “NPV = 0 across funds” (as in Cochrane) and Monte Carlo experiments to assess estimator accuracy. The method of moments is practical and easy to implement.

First to analyse LBO deal-level cash-flow data and estimate risk exposures. Estimated liquidity-risk premium ≈ 3 % annually and alpha gross of fees ≈ 0. Also first to document a strong correlation between liquidity conditions over an investment’s life and LBO returns.

Replicates results in the often-cited Harris, Jenkinson and Kaplan (2014) and shows high sensitivity to benchmark choice. First to show and argue that PE funds invest in companies much smaller than those in the S&P 500, implying that the S&P 500 is an inappropriate benchmark according to asset-pricing theory. It also introduces the use of small-cap benchmarks that are net of fees and traded, highlights how the Russell 2000 is a poor choice as it significantly underperforms peer indices, and finds that US PE fund returns are close to small/mid-cap US equity benchmarks over the same sample period.

First paper to derive sufficient assumptions under which the condition “NPV = 0 across funds” identifies betas in a model-free framework. Distinctive contribution is to avoid imposing a factor model (as in Cochrane 2005 or Driessen et al. 2012) by using a Bayesian MCMC approach instead. This produces a time-series of true rates of return that generated PE fund returns without relying on self-reported NAVs.

Shows it is possible to infer Carry earned by GPs using only fund TVPI -- which is widely available. Use fund-level cash flows and fee terms for more than 12,000 funds. Estimate total amount earned: exceeds one trillion dollars. Accounts for 18% of investor profits, about equal the contractual value-weighted rate of 19%. The difference reflects the role of hurdle rates and the relatively smooth distribution of fund outcomes. Carry is strongly related to both performance and fund size, and past carry is a stronger predictor of future performance than past returns.

Argues that apparent performance persistence is driven by overlapping fund lives; studied ex-ante persistence and found persistence mainly among losers. Also first to examine the flow-performance relation in PE, documenting strong flows to winners but limited persistence for winners.

Compiled a large cross section of LBO investment returns and documented differences across exit channels, duration and countries. Found lower returns in developing countries, average investment durations around 4 years, and evidence that funds often exit winners earlier while holding losers. Suggested measuring portfolio risk by variance in the loss domain and highlighted that the number of concurrent investments—not fund size—explains key performance differences.

Using a dynamic portfolio model, we show that ex-ante capital commitment materially affects investor allocations and welfare. Investors should under-allocate to PE and use direct investments (e.g., co-investments) to adjust their allocations upward over time. Commitment risk premiums increase with secondary market liquidity and persist even when diversification across many funds is possible.

Documented returns to buyers of secondary buyouts (SBOs). SBOs purchased under pressure to deploy capital underperform; buyers who complement sellers’ skills tend to perform better. Also discussed LPs that participate on both sides of SBO transactions.

First comprehensive survey of PE investors (LPs). Shows that investors with a larger capital allocation to PE are more specialized and have a wider scope of due diligence and investment activities. Other investor characteristics (experience, type, location, compensation structure, number of funds under management) play no role in explaining differences in due-diligence or monitoring. Documents how many days LPs spend on due diligence and their priorities when choosing funds, including reliance on IRR.

First paper to empirically document the existence of ex-post discretionary fees charged by fund managers on the assets they manage on behalf of LPs — the first type of tunnelling documented for a population of institutional investors. These fee payments total about $20 billion, representing over 6% of the equity invested. Fees do not vary with business cycles, company characteristics, or GP performance but differ strongly across GPs and are persistent within GPs. GPs charging the least raised more capital after the financial crisis, when these fees drew public attention. Following publication, many such fees were reportedly refunded, illustrating that even in developed, institutionally dominated markets, large-scale tunnelling can occur without robust regulation.

First paper to describe the existence of ex-post discretionary fees charged by fund managers on assets they manage for LPs, highlighting conflicts of interest and weak alignment of incentives. Explains why fundraising track-record presentations can be misleading, shows how definitions drive large variations in fee bills, and details incentives to exit winners early and hold losers longer, both to inflate IRRs and carried interest.

First paper to show why and how IRR is deeply flawed in the PE context. Key reason is endogenous cash-flow timing. Early exits of winning deals create unrealistic reinvestment assumptions that inflate IRRs. Proposes MIRR and NPV as more reliable alternatives.

Shows that Yale’s endowment is not the model it is often claimed to be. The frequently cited “30% return” was never an annualized rate of return but rather an IRR — something Yale itself never described as such, nor had anyone else. Also show that their reported returns changed very little from one year to the next, and show how this was due to large venture capital distributions of the 1990s. Following publication, Yale acknowledged that their numbers were based on IRR. As suggested in this paper, they subsequently separated venture capital and buyout returns — and, as predicted, the venture capital figures appeared even more unrealistic (93% per annum), while buyout returns were around 12% per annum. Then, in 2020, again as recommended here, Yale began showing its 20-year IRR, thus excluding those early distributions. Both VC and BO returns then converged to around 11%. In short, the so-called “Yale model” never existed.

Reviews industrial-organization and finance research on price shrouding and applies it to private equity. First to argue that most PE contracts function as take-it-or-leave-it products rather than arm’s-length negotiations, with headline fees disconnected from total costs. Highlights how complexity and lack of comparability impede competition, providing a clear rationale for regulatory intervention.

Shows that, despite similar language, carried-interest waterfalls often provide weak alignment between GPs and LPs. Proposes contract modifications to strengthen alignment and ensure carried interest qualifies as a capital gain rather than ordinary income.

Highlights limitations of using accounting data to infer value-addition in PE. Differences in payout policies, acquisition frequency, and goodwill amortization make ratios such as ROA or growth comparisons misleading. Concludes that the magnitude of true post-LBO earnings improvement remains uncertain.

This paper measures the aggregate performance of the US private equity industry as of 2019 and estimates the total carried interest accrued to general partners. It provides a detailed explanation of how to approximate the total carry earned and offers a comprehensive review of the empirical evidence on private equity performance, fee structures, and reporting conventions. US private equity returns have been broadly similar to those of US public equity markets. Yet, because of the vast sums committed and the asymmetric fee structure, general partners collectively earned several hundred billion dollars in carried interest. Findings challenged one of the industry’s most entrenched narratives and received extensive media coverage. The paper remains one of the most downloaded finance articles on SSRN.

Uses natural language processing to create a listed Private Equity exposure index highly correlated with LBO fund indices, demonstrating how thematic investing can be automated with big data techniques.

Decomposes value creation into leverage, multiple expansion, operational improvement, and timing, showing that Hilton’s adjusted value-add was similar to Marriott’s. Highlights how difficult — and essential — it is to measure true performance drivers in private equity.

Analyzes the “democratization” of private equity, arguing that retail products risk exploitation without transparency, governance, and fair pricing safeguards.

Explains why the since-inception IRR misrepresents investor wealth growth and how its misuse fosters the illusion of PE outperformance. Shows that returns reported by well-known PE firms such as KKR and Apollo hardly change from one year to the next over decades. Advocates replacing si-IRR with horizon IRRs and NPV-based metrics.

A review of the latest data (June 2025) suggests that prior findings (Phalippou 2014, 2020) still hold. The performance of private equity is broadly in line with public markets in the US and probably in Europe. There is little controversy about fund performance per se; disagreements stem from how benchmarks are chosen and how data is filtered.

Informal abstract: A theoretical analysis showing that when sellers have private information and buyers are optimistic about adding value, seller-provided loans is better than equity retention. This structure mitigates information asymmetry and improves efficiency, except when excessive buyer optimism eliminates demand for such loans.

Reveals a strong link between the 2016 Brexit vote and COVID-19 outcomes: districts with the most Remainers had roughly 33 % lower death rates, 25 % lower infection rates, and higher vaccination uptake than those with the fewest. The gap widened after the first wave, highlighting how cultural and belief differences shape policy effectiveness.

Documents that about half of the value destruction in M&A stems from a small subset of deals (≈7 %) in which the target is itself a serial acquirer. Such “defensive” acquisitions are highly value-destructive. Target acquisitiveness strongly predicts both deal completion and acquirer announcement returns.

Contemporaneous and independent of Nagel (2005), this paper shows that the value premium exists only among stocks with low institutional ownership, disappearing when arbitrage limits are high. Finds the effect holds for both long and short positions, casting doubt on risk-based explanations.

First to highlight the existence of a “factor zoo,” showing that several rational-risk models proposed to explain asset-pricing puzzles were not robust to changes in test assets—evidence of model snooping.

Early empirical test for a liquidity-risk premium among mutual funds. Finds no robust premium once persistence and cross-sectional correlation are handled correctly via clustered errors and Fama–MacBeth regressions. Although unpublished, its liquidity metric has been widely used in later literature, often without attribution.

As part of my PhD (2000–2004) I prepared several versions of this paper, presented it in 2005, posted it on SSRN in 2005 and it was eventually published in 2009. The paper contains several contributions that are often overlooked: simulation of fee structures on net cash flows (translating fee streams into an annualized equivalent ≈6% p.a.), an alpha measure obtained by adding a constant to benchmark returns, and a Beta-adjusted PME together with an effective fund duration concept. The paper also documents a negative correlation between fund duration and measured performance — an important stylized fact that affects averaged IRR comparisons.

A recurring critique is that results are driven by TVE dataset anomalies. The TVE dataset used here is the same source as Kaplan and Schoar (2005), and for LBO funds our figures match theirs. The discrepancy on VC PMEs is due to an inactivity filter: some VC funds showed flat NAVs for prolonged periods (no reported cash flows); treating NAVs unchanged after three years with no cash flows leads to different results for VCs. For LBO funds this filter makes negligible difference.

Methodologically, we were the first to simulate fee structures using actual net-of-fees cash flows, demonstrating fees are dominated by the fixed component (not carry) and converting them into an annualized cost equivalent. We also proposed an 'alpha' obtained by shifting benchmark returns until NPV is zero — a construction with drawbacks when alpha is large due to reinvestment assumptions, which we discussed in the working-paper footnote.

Key takeaways: (1) an early risk-correction attempt via Beta-adjusted PME; (2) a fee-simulation exposing material fixed-fee costs; (3) clarity on data treatment choices (inactivity filter) and their implications for VC vs LBO results. These methodological points are why I include this longer note here: many critiques focus on headline PMEs without engaging the underlying data and filters.

Explains the method-of-moments approach searching for Betas that best fit fund cash flows under the identifying condition “NPV = 0 across funds” (as in Cochrane). We used fund-level data and Monte Carlo simulations to assess estimator performance. The method is easily implemented (even via simple grid search in Excel for single-factor models) though point estimates are sensitive to idiosyncratic volatility; the Monte Carlo checks provide useful guidance on estimator accuracy.

This paper assembled a large cross-section of LBO investment returns (sourced from fundraising prospectuses). It was initially rejected at multiple journals due to concerns about selection bias in prospectus-based data. We apply tests and corrections and argue that main qualitative results—investment duration, exit-channel patterns and the role of portfolio concentration—are robust despite data-source limitations. The paper explores diseconomies of scale and offers portfolio-level risk metrics (variance in loss domain) that better reflect concentrated private-investment risks.

Using a dataset of 4,403 LBO deal-level cash flows (gross of fees) we estimate betas in a four-factor model including liquidity risk and find an annual liquidity premium of roughly 3%. Working at the deal level reduces complications from intermediate cash flows (they are relatively rare), letting us apply standard inference techniques. We also study cross-sectional determinants of returns and find that liquidity conditions during the life of an investment are a significant predictor of performance.

Demonstrates that apparent outperformance of private equity versus public benchmarks is highly sensitive to benchmark choice. Using Capital IQ to show PE investments are typically in smaller companies, this paper argues that the S&P 500 is often an inappropriate benchmark; small and mid-cap, traded, net-of-fees benchmarks provide a more comparable yardstick. We also show Beta-adjustments can materially change conclusions depending on benchmark selection.

This paper derives sufficient conditions under which searching for Betas that make NPV=0 identifies the true Betas in a linear multi-factor model and introduces a Bayesian MCMC approach to recover a time series of returns without relying on self-reported NAVs. The main novelty is generating a time-series of true rates of return that produced observed fund cash flows — a useful object for many further analyses even if point estimates of Alpha/Beta are not the primary focus. Aggregation at the fund level is necessary because idiosyncratic volatility at the investment level biases naive estimators.

Notes: this survey-based work documents how larger allocators to PE behave—more specialization, deeper due diligence, and broader activity. It was initially desk-rejected by some journals that are averse to survey methods, but subsequent research has validated the importance of survey evidence for understanding LP practices.