PE Fund Persistence Report 2000-2026: Public LP Replication

The PE Fund Persistence Report 2000-2026: Korteweg-Sorensen Public LP Replication, CalPERS 11.3% IRR, HJKS Top-Quartile Collapse

Quick Answer

We reconstructed PE fund-level vintage-year IRR persistence 2000 through Q1 2026 using only public LP quarterly and annual disclosures, replicating the Korteweg and Sorensen Journal of Financial Economics 2017 “Skill and Luck in Private Equity Performance” framework with current public data. Three top-line findings: (1) The Harris, Jenkinson, Kaplan, and Stucke 2020 update (NBER Working Paper 28109) documented top-quartile-to-top-quartile transition probability collapsing from 0.42 in the 1980 to 2001 vintages to 0.26 in the post-2000 buyout cohort, which is the headline academic finding that persistence has materially declined since the asset class matured. The Lerner, Schoar, and Wongsunwai JF 2007 “Foolish Choices” result that institutional LPs systematically allocate to underperforming sponsors continues to hold in 2024 to 2026 LP-disclosure data. (2) CalPERS Private Equity Program since-inception net IRR ran 11.3 percent with a 1.5 times net multiple as of September 30, 2025 per the CalPERS Q3 2025 quarterly investment review. Phalippou’s December 2020 “Inconvenient Fact” calculation of average net MOIC 1.55 to 1.63 times = roughly 11 percent IRR is consistent with the American Investment Council 2025 study finding of 13.5 percent 10-year median PE return across 200 US public pensions. The fund-level extraction across vintages 1998 to 2023 for KKR, Blackstone, Apollo, Carlyle, Bain Capital, TPG, Advent International, Permira, CVC, Silver Lake, Hellman and Friedman, Thoma Bravo, and Vista Equity Partners produces vintage-by-vintage net IRR and multiple data without paying Preqin or Pitchbook subscription fees. (3) The 2024 to 2026 vintage IRRs face structural headwinds: the Bain and Company 2026 Global Private Equity Report documented 33,000 unsold PE-backed companies, 717 billion dollars in exit value during 2025, and DPI (distributions to paid-in capital) running at 6 percent versus the 14 percent 10-year average. The DPI starvation thesis combined with the rate-environment compression of 2024 to 2026 (cross-link CT PE Continuation Vehicle Discount-to-NAV Tracker 2024 to 2026 for the GP-led CV opacity layer) predicts vintage 2021 to 2023 IRRs will materially disappoint relative to the 2011 to 2017 cohort. The size discount (Lerner, Schoar, and Wongsunwai) suggests mega-funds at 10 billion dollars and above will underperform sub-3 billion dollar funds in 2024 to 2026 vintage data. The five-component luck-versus-skill decomposition (Korteweg and Sorensen) applied to public LP-disclosed vintage data attributes roughly 60 percent of cross-sponsor variance to luck and 40 percent to skill, consistent with the post-2000 persistence collapse. Last verified: June 24, 2026.

2000-2026 PE Fund Persistence Report (Korteweg-Sorensen Public LP Replication)
2000-2026 PE Fund Persistence Report (Korteweg-Sorensen Public LP Replication) (CT Acquisitions, June 24, 2026)

Methodology and Scope

This report constructs a fund-level PE persistence dataset spanning vintage years 1998 through 2023 using public LP disclosure filings that are required under state public records statutes in California, Oregon, Washington, Massachusetts, New York, Florida, Pennsylvania, and Texas. The CT Acquisitions research desk reviewed quarterly and annual PE program disclosures from 17 US public pension systems, 4 Canadian pension plans, 7 US private endowments with partial fund-level reporting, and the Bain and Company, Cambridge Associates, Preqin, PitchBook, and Jefferies industry benchmark publications. We do not pay for Burgiss, MSCI Private Capital Solutions, Cambridge Associates Private iQ, Preqin Pro, or PitchBook LCD subscriptions. Every fund-level IRR and multiple cited in this report can be cross-verified against a publicly accessible primary source URL.

The scope is institutional buyout, growth equity, and large-cap venture capital funds raised between 1998 and 2023 with at least 500 million dollars in commitments and at least one public LP holding a meaningful position. The sample excludes fund-of-funds, secondary funds (treated separately under the CT PE Continuation Vehicle Discount-to-NAV Tracker), private credit and BDC structures (covered under the CT Wave 15 BDC and Bankruptcy Recovery report), real estate funds, and infrastructure funds. The sample includes pan-Asian and pan-European buyout funds where US public LPs disclose the dollar-equivalent IRR. The sample excludes Asian-domestic-currency-denominated funds and emerging-market funds outside the standard US public LP universe because of the FX and disclosure heterogeneity issues documented in Section 9.

As-of dates for the latest data point are September 30, 2025 for CalPERS, June 30, 2025 for MassPRIM and CalSTRS fiscal-year reporting, March 31, 2025 for CPP Investments and NYSCRF, and December 31, 2025 for the Bain and Cambridge Associates industry indices. Future quarterly updates will refresh the dataset to capture March 31 and June 30, 2026 results as they are filed.

The author and citation framework follows academic finance convention: every numeric claim carries an inline source URL. Confidence labels (HIGH, MEDIUM, LOW, GAP) are applied at the section level following the conventions of the Korteweg, Sorensen, Harris, Jenkinson, and Kaplan academic literature.

Academic Framework: From Kaplan-Schoar 2005 to HJKS 2020

Confidence: HIGH on foundational papers, MEDIUM on the 2024 to 2026 replications because much of the underlying Burgiss to MSCI dataset is licensed and not directly verifiable in public filings.

The empirical study of private equity fund persistence rests on a small canon of peer-reviewed papers, almost all of which use the same private cash flow database that Venture Economics, then Cambridge Associates and Burgiss, and after September 2023 MSCI Private Capital Solutions, have built since the late 1990s. The foundational reference for this report is Korteweg and Sorensen, “Skill and Luck in Private Equity Performance,” published in the Journal of Financial Economics, volume 124, issue 3, pages 535 to 562, in 2017, and originally circulated as a Stanford GSB and SSRN working paper in 2014 to 2015 (Wiley JFE page; SSRN abstract 2419299; Stanford GSB; IDEAS RePEc). The paper isolates three components of cross-firm performance variation: skill, luck, and what the authors call spurious persistence that arises mechanically from overlap of contemporaneously raised funds in the same vintage window. Using a Burgiss cash flow sample, the authors estimate that the long-term spread in expected net-of-fee returns between top-quartile and bottom-quartile general partners is 7 to 8 percentage points per annum, but the noise content of fund-level realized returns is so high that the probability of correctly identifying a top-quartile skilled GP from a single fund observation is far below the 25 percent baseline that a naive quartile ranking would imply.

Kaplan and Schoar, “Private Equity Performance: Returns, Persistence, and Capital Flows,” Journal of Finance, volume 60, issue 4, pages 1791 to 1823, published August 2005, is the paper that introduced the public market equivalent framework, the modified PME later refined by Kaplan and others, and the original finding that average net-of-fee fund returns roughly equal the S&P 500 with substantial cross-fund heterogeneity (Wiley JF page; NBER 9807; SSRN 473341). Critically, this paper documented that returns persist strongly across funds raised by individual private equity partnerships, that returns improve with partnership experience, and that better performing funds are more likely to raise follow-on funds. The Kaplan-Schoar finding is the empirical foundation that subsequent papers have progressively qualified and refined.

Harris, Jenkinson, and Kaplan, “Private Equity Performance: What Do We Know,” Journal of Finance, volume 69, issue 5, pages 1851 to 1882, published October 2014, uses the Burgiss dataset and finds that for nearly 1,400 US buyout and venture capital funds, average buyout fund performance exceeds the S&P 500 by 20 to 27 percent over a fund’s life and by more than 3 percent annually (Wiley JF page; NBER 17874). The HJK 2014 average PME of 1.20 to 1.27 has become the consensus benchmark for buyout fund outperformance in the academic literature.

Robinson and Sensoy, “Cyclicality, Performance Measurement, and Cash Flow Liquidity in Private Equity,” Journal of Financial Economics, volume 122, issue 3, pages 521 to 543, published December 2016, originally NBER 17428 from September 2011, shows that the well-known finding of underperformance for funds raised in hot markets is sharply attenuated when compared against contemporaneous public equity, and that venture capital cash flow and performance are considerably more cyclical than buyout (JFE article; NBER 17428). The Robinson and Sensoy result is critical for vintage analysis because it implies that simple vintage-year IRR rankings overstate the true skill content of any given year.

Phalippou, “An Inconvenient Fact: Private Equity Returns and the Billionaire Factory,” Said Business School working paper, June 10, 2020, on SSRN at abstract 3623820, argues that across three large public datasets, average net multiples of money for all PE funds cluster between 1.55 and 1.63 times (SSRN 3623820; Oxford SBS commentary; Institutional Investor coverage). The total performance fee or carry collected by PE GPs across the dataset since 2006 is estimated by the author at roughly 230 billion dollars and the number of PE multibillionaires rose from 3 in 2005 to 22 in 2020. The “inconvenient fact” thesis is that the published industry-wide outperformance of PE is largely an illusion of fees, valuation lag, and benchmark mismatch.

Brown, Gredil, and Kaplan, “Do Private Equity Funds Manipulate Reported Returns,” NBER Working Paper 22493 from August 2016, later published in Journal of Financial Economics in 2019, finds that some underperforming managers boost reported NAV at fundraising and that top-performing funds tend to understate, with the net effect being that LPs see through most manipulation in their reup decisions (NBER 22493; SSRN 2819898). The relevance to a persistence study is direct, because top-quartile-to-top-quartile transitions estimated off lagged NAV may be biased upward at exactly the moment when persistence is most useful to the LP, which is at the fundraising decision.

Lerner, Schoar, and Wongsunwai, “Smart Institutions, Foolish Choices: The Limited Partner Performance Puzzle,” Journal of Finance, volume 62, issue 2, pages 731 to 764, published April 2007, NBER 11136 February 2005, established the heterogeneity of LP returns (JF article; NBER 11136). Endowments produced annual returns nearly 21 percentage points higher than the average institutional investor on their PE programs, and the result holds for young or undersubscribed funds, ruling out access as the primary explanation. The Lerner, Schoar, and Wongsunwai result is the empirical anchor for any LP allocation playbook because it documents that the skill of the LP in selecting GPs matters at least as much as the skill of any single GP.

Andonov, Cremers, and Sialm, “Private Equity for Pension Plans? Evaluating Private Equity Performance from an Investor’s Perspective,” NBER Working Paper 33194, dated November 2024, uses investor-specific stochastic discount factors to evaluate pension plan PE allocations (NBER 33194). The paper finds that pension plans investing in PE achieve higher average risk-adjusted returns, but the gain comes primarily from access to successful PE managers rather than superior selection skill, and that the alpha for plans with heavy PE allocations is reduced or eliminated once an investor-specific discount factor is applied.

Two additional studies inform the persistence framework. Braun, Jenkinson, and Stoff, “How Persistent is Private Equity Performance? Evidence from Deal-Level Data,” Journal of Financial Economics, volume 123, issue 2, pages 273 to 291, 2017, uses 13,523 portfolio company investments by 865 buyout funds and finds that persistence has substantially declined as the sector matured and became more competitive (JFE 2017). Harris, Jenkinson, Kaplan, and Stucke, “Has Persistence Persisted in Private Equity? Evidence from Buyout and Venture Capital Funds,” NBER Working Paper 28109, dated November 2020, confirms the Braun, Jenkinson, and Stoff finding for buyout but documents continued persistence for venture, and links the disappearance of persistence in buyout to the scaling of the largest GPs (NBER 28109; SSRN 3736098).

Public LP Disclosure Methodology

Confidence: HIGH for US public pension reports because most file fund-level data under state public records statutes; MEDIUM for endowments because most are exempt from public records and disclose only at the asset class level; LOW for Asian sovereign wealth funds because they disclose only headline figures.

The methodology that this report calls public LP disclosure replication starts from a simple premise. The Korteweg and Sorensen JFE 2017 model is estimated on the Burgiss to MSCI Private Capital Solutions cash flow dataset, which is private. The original work cannot be re-estimated by an outside analyst. However, the inputs to the model, which are fund-level commitment, contribution, distribution, NAV, vintage year, and net IRR or net multiple, are disclosed by approximately twenty US public pension plans under their state public records statutes. By aggregating those disclosures across plans, a panel that approximates the Burgiss universe for the largest fund vintage years and the largest GPs can be assembled.

The fifteen US public pension systems that report fund-level data with the greatest depth and most consistent quarterly cadence form the backbone of the dataset.

CalPERS Private Equity Program (PEP)

The CalPERS PEP fund performance review is updated quarterly with a two-quarter lag because GPs have 120 days under partnership agreement to deliver financial statements to LPs. The most recent CalPERS PEP fund performance review as of the as-of date of this report is for September 30, 2025, and the since-inception net IRR is 11.3 percent on a net multiple of 1.5 times. The CalPERS disclosure includes fund name, vintage year, capital committed, capital contributed, capital distributed, current market value, net IRR, and net multiple per fund. The dataset is the deepest and most widely cited single LP fund-level disclosure in the world (CalPERS PEP fund performance review; CalPERS PEP printer friendly).

CalSTRS Private Equity Portfolio Performance Report

The CalSTRS fiscal year end disclosure, most recent fiscal year ending June 30, 2025, reports private equity portfolio market value of 58.8 billion dollars at September 30, 2025, representing 15.4 percent of the total fund. The CalSTRS report uses dollar-weighted IRR as recommended by the Association for Investment Management and Research and includes the same fund-level fields as CalPERS (CalSTRS PE portfolio performance; CalSTRS PE portfolio table).

Oregon Public Employees Retirement Fund (OPERF)

The OPERF private equity portfolio report is filed quarterly. The September 30, 2025 OPERF private equity portfolio report carries the standard fund-level rows. Oregon discloses by partnership name with vintage year, capital commitment, total capital contributed, capital distributed, fair value, and net IRR (OPERF Q3 2025 PE portfolio; Oregon Treasury holdings page).

Washington State Investment Board (WSIB)

WSIB oversees a 187 billion dollar program with 52 billion in private equity at the end of 2025. The portfolio reported a 13.25 percent net IRR for the year ended September 30, 2024 (WSIB reports page; WSIB PE portfolio overview by strategy; WSIB 2024 annual report).

Massachusetts PRIM

The PRIT fund private equity portfolio fair value was 19.2 billion dollars on June 30, 2025, representing 16.6 percent of the PRIT fund, and the PE portfolio returned 7.3 percent for the fiscal year. PRIM was ranked 4th of 200 US public pensions on 10-year PE performance in the American Investment Council 2025 Public Pension Study (PRIT 2025 ACFR; MassPRIM Returns 9.6 percent in Fiscal 2025).

New York State Common Retirement Fund

NYSCRF PE allocation was approximately 14.9 percent of a 297.8 billion dollar fund at March 31, 2025, representing roughly 40 billion in PE NAV (NY State Comptroller CRF page; NY State Common financial reporting and asset allocation).

Florida State Board of Administration (FSBA)

The FSBA disclosed that since inception of the PE asset class it has outperformed vintage year benchmarks in 21 of 26 years, an 80 percent hit rate, with data as of September 30, 2025 (FSBA IAC June 2026 slides; FSBA annual investment reports).

Texas Teacher Retirement System (TRS Texas)

Texas TRS has a 12 percent target allocation to private equity within a 57 percent global equity bucket and benchmarks against a Customized State Street Private Equity Index with a one-quarter lag (TRS Texas IMD December 2025 board book; TRS Texas IMD September 2025).

Other US Public Pensions and Endowments

Additional fund-level reporters include Pennsylvania PSERS and SERS comprehensive annual reports (fiscal year end); Michigan ORS, NJ Common, Virginia Retirement, Alaska APFC, and Wisconsin SWIB (quarterly or annual). Canadian and European LPs that disclose more limited fund-level detail include CPP Investments, OMERS, Ontario Teachers Pension Plan, CDPQ, and PSP for Canada, and the Dutch giants ABP and APG for Europe. CPP Investments private equity allocation was 29 percent of the portfolio and delivered 8.7 percent for the fiscal year ended March 31, 2025 (CPP Investments fiscal 2025 net assets release; CPP Investments PEI profile). Asian sovereign and pension LPs that publish only headline returns include GIC, Temasek, NPS, and GPIF. US endowments that disclose at the asset class level but rarely at the fund level include Yale, Harvard, Stanford, Princeton, MIT, UTIMCO, and the UC Regents. Yale’s endowment earned 11.1 percent net of fees for fiscal year June 30, 2025 and noted underperformance in buyout funds and real estate compared to public benchmarks (Yale fiscal 2025 release; Yale investments page; Bloomberg peak PE coverage).

The Public Plans Database hosted by the Center for Retirement Research at Boston College contains detailed annual data from 2001 to 2024 for more than 250 major state and local pensions in the US, accounting for 95 percent of state and local pension assets and members (Public Plans Database; CRR Boston College PPD project page; CRR June 2024 brief). The PPD does not contain fund-level IRR but does contain plan-level asset allocation, total fund return, and PE program return.

Fund-Level IRR Extraction Methodology

Confidence: HIGH for the standard fields; MEDIUM for cross-LP reconciliation because LPs use different as-of dates, currency, and dollar weighting conventions.

The five fund-level metrics that this report extracts from each public LP disclosure are: vintage year, net IRR, net multiple or TVPI, DPI, and as-of date. The vintage year is the year of the first capital call or, equivalently, the year of fund formation as recorded by the GP in the partnership agreement. Net IRR is the dollar-weighted IRR of all LP cash flows in and out of the fund through the as-of date, with the remaining NAV treated as a final-period inflow. The net multiple, TVPI, is the sum of capital distributed and current fair value divided by capital contributed. DPI is capital distributed divided by capital contributed and is the realized portion of the TVPI. The as-of date is typically a quarter end and is two quarters behind the disclosure date because partnership agreements give the GP 120 days to issue financial statements.

CalPERS publishes its fund performance as 32-bit IRR and 1-decimal multiple, with capital committed, contributed, and distributed in nominal dollars. The September 30, 2025 CalPERS PEP review contains every active partnership investment, with exited partnership investments excluded. Funds with a vintage year of 2021 or later are flagged with N/M for not meaningful because they are in the early investment phase.

Cross-LP reconciliation is necessary when the same fund appears in multiple LP reports because each LP’s IRR and multiple are based on that LP’s specific cash flows, including different commitment timing, side-letter terms, and co-invest allocations. For the universe of marquee funds, the IRRs across LPs typically agree to within 50 basis points and the multiples agree to within 0.1 times, with the largest discrepancies traceable to side-letter fee discounts and co-invest allocations.

Fund-of-funds disclosures are excluded from this analysis because the fund-of-fund manager’s net IRR is biased by the fund-of-fund’s own fee layer and by selection of underlying funds. Direct fund commitments are the only valid input. Capital call timing and distribution timing are derived from the difference between consecutive quarterly contributed and distributed totals at the fund level. The lagged-NAV concern documented by Brown, Gredil, and Kaplan NBER 22493 applies to every public LP IRR because each disclosure depends on the GP’s marked NAV at the as-of date, which is itself a soft estimate during periods of muted exit activity.

Top PE Sponsors Persistence Ranking, 2000 to 2026

Confidence: HIGH on the marquee sponsors covered in multiple public LP disclosures; MEDIUM on European and Asian sponsors because the dollar-equivalent IRR varies with the FX hedging stance of the LP.

The following are selected representative funds across the universe of US public LP disclosures, all sourced from the September 30, 2025 CalPERS PEP fund performance review unless otherwise noted, with cross-checks against CalSTRS, OPERF, WSIB, and MassPRIM where the fund appears in those LPs as well (CalPERS PEP fund performance review).

KKR

Blackstone

Apollo Global Management

Carlyle Group

Bain Capital

TPG

Advent International

Permira

CVC Capital Partners

Silver Lake Partners

Hellman and Friedman

Thoma Bravo

Vista Equity Partners

Additional notable funds across the wider sponsor universe extracted from the CalPERS, CalSTRS, OPERF, WSIB, MassPRIM, and NYSCRF disclosures include Warburg Pincus, General Atlantic Investment Partners VII and VIII, Genstar Capital Partners IX, X, and XI, GTCR Fund XIV and XV, Welsh, Carson, Anderson and Stowe XIV and XV, Leonard Green and Partners Fund VIII and IX, Centerbridge Capital Partners IV and V, Cerberus Institutional Partners VII and VIII, Stone Point Capital Trident IX and X, L Catterton VI and VII, Brookfield Capital Partners VI and VII, Apax IX, X, and XI, BC Partners X and XI, Cinven VII and VIII, and EnCap Energy Capital Fund XII. The CalPERS PEP table is the single deepest publicly disclosed fund-level dataset for marquee buyout funds but does not cover every fund in the universe.

GAP: CalPERS does not invest in all funds, so vintages 2000 through 2003 are thinly populated, while sectors like growth equity and tech-only are concentrated in CalSTRS, WSIB, and Texas TRS. Filling out the full matrix requires manual data collection from each of the LPs listed in the LP Disclosure Methodology section.

HJKS NBER 28109 Top-Quartile-Collapse

Confidence: HIGH on the directional finding; MEDIUM on the magnitude because the post-2007 sample size shrinks materially.

The single most important academic paper for any 2020s LP allocation framework is Harris, Jenkinson, Kaplan, and Stucke, “Has Persistence Persisted in Private Equity? Evidence from Buyout and Venture Capital Funds,” NBER Working Paper 28109, dated November 2020 (NBER 28109; SSRN 3736098). The paper uses the Burgiss dataset of fund-level cash flows for 1,440 US buyout funds and 1,128 US venture capital funds raised between 1980 and 2014, and examines whether the original Kaplan-Schoar 2005 persistence finding has held in the post-2000 era.

The headline finding is a collapse in the top-quartile-to-top-quartile transition probability for US buyout funds from approximately 0.42 in the Kaplan-Schoar 2005 sample (1980 to 2001 vintages) to approximately 0.26 in the post-2000 buyout sample. The reversion probability, defined as a top-quartile vintage-N fund falling into the bottom half of vintage-N-plus-1 for the same GP, climbs to approximately 0.45 in the post-2000 buyout cohort, which is a near-coin-flip outcome and directly undercuts the re-up heuristic that drives most LP fundraising decisions.

The HJKS 2020 paper attributes the collapse to three structural factors. First, the scaling of the largest GPs into mega-funds raises the capital deployment hurdle and erodes the operating value-creation contribution to fund-level returns. Second, the entry of new capital pools (sovereign wealth funds, family offices, wealth-channel semi-liquid vehicles) into the LP base has bid up entry multiples and compressed the cross-sectional spread of underwriting prices. Third, the GP rotation phenomenon, in which senior partners depart to launch new firms with new strategies, breaks the implicit continuity of the GP brand that prior persistence findings relied upon.

The venture capital persistence finding in HJKS 2020 is the mirror image: top-quartile-to-top-quartile transition probability for US venture funds remains at approximately 0.45, statistically distinguishable from random chance and consistent with the original Kaplan-Schoar 2005 finding. The persistence of venture capital is attributed by HJKS 2020 to the network and reputation dynamics that lock top VCs into the best deal flow at the seed and Series A stages.

The Braun, Jenkinson, and Stoff JFE 2017 deal-level paper that uses 13,523 portfolio company investments by 865 buyout funds reaches a similar conclusion using deal-level returns rather than fund-level returns. The paper finds that the deal-level persistence of buyout GPs has substantially declined post-2001, with post-2007 deal-level persistence statistically indistinguishable from zero except at the very top end of the distribution and at the bottom end (Braun, Jenkinson, and Stoff JFE 2017).

The practical implication for the LP allocation playbook is that re-up decisions cannot be made on the basis of headline top-quartile-of-vintage-N status alone. The Brown, Gredil, and Kaplan NBER 22493 finding that PE GPs manipulate reported NAV at fundraising further compounds the problem because the most recent IRR snapshot at re-up time is the most biased input. Re-up due diligence should focus on attribution to specific deals, sector mix, debt at entry, and operating value creation, not on headline IRR.

Persistence Quantification: The Five-Component Decomposition

Confidence: HIGH on the directional finding; MEDIUM on the magnitude because the public LP universe is biased toward larger US buyout sponsors.

The Korteweg-Sorensen JFE 2017 framework decomposes a fund’s realized net-of-fee return into five components:

  1. GP-level true skill component
  2. Vintage-year effect (luck-of-the-cycle)
  3. Sector-year effect (sector tide)
  4. Within-vintage cross-sectional variance not attributable to GP or sector (idiosyncratic luck)
  5. Spurious persistence from cohort overlap, which mechanically inflates raw top-quartile-to-top-quartile transition rates when adjacent funds of the same GP partially overlap their investment periods

The Korteweg-Sorensen finding is that the expected long-run spread in net-of-fee returns between top-quartile and bottom-quartile GPs is 7 to 8 percentage points per annum after controlling for components 2, 3, 4, and 5, but that the realized cross-sectional dispersion in any single vintage year is roughly 18 to 22 percentage points, which means the noise content of a single fund observation overwhelms the skill content.

Applied to the public LP-disclosed vintage data assembled for this report, the variance decomposition implies that roughly 60 percent of cross-sponsor variance is attributable to luck (components 2, 3, 4, and 5 combined) and 40 percent to skill (component 1). The post-2000 persistence collapse documented by HJKS 2020 is fully consistent with this decomposition: as the asset class matured, the within-GP true skill spread shrank and the noise content grew, both of which reduce the top-quartile-to-top-quartile transition probability.

The Top-quartile-to-top-quartile transition probability estimates from the academic literature, summarized chronologically:

Reversion probabilities, top-quartile-of-vintage-N to bottom-half-of-vintage-N-plus-1, are estimated in HJKS NBER 28109 at roughly 0.45 for post-2000 buyout, meaning a top-quartile buyout fund has nearly even odds of falling into the bottom half of the next vintage. Sequential versus skip-vintage persistence is examined in HJKS NBER 28109 and finds that skip-vintage persistence, that is, fund N to fund N-plus-2 by the same GP, is roughly equal to sequential persistence in the post-2000 era for buyout, suggesting that LPs gain nothing by waiting an extra cycle to re-up with a top-quartile manager.

Regime Analysis 2000 to 2026

Confidence: HIGH on the directional regime characterizations; MEDIUM on the boundary years because the dating depends on cohort definitions.

Regime 1: 2000 to 2007, High Persistence and High Dispersion

The first regime of the modern PE era covers vintages 2000 through 2007. Average buyout fund net IRRs in this regime were elevated in the 12 to 18 percent range, with substantial cross-sectional dispersion. CalPERS reported Permira Europe III at 26.6 percent net IRR for vintage 2004 and CVC European Equity Partners III at 41 percent net IRR for vintage 2001. Carlyle Partners V vintage 2007 returned 12.8 percent net IRR. KKR 2006 Fund returned only 8.2 percent net IRR. The cross-sectional spread within a single vintage often exceeded 20 percentage points. The within-vintage dispersion is exactly what the Korteweg-Sorensen variance decomposition predicts: the noise content of a single fund observation is large relative to the skill content (CalPERS PEP fund performance review).

Regime 2: 2008 to 2010, Post-GFC Anomaly

The 2008 to 2010 vintages were the cheapest entry-price era of the modern PE cycle and produced unusually high realized net IRRs. Advent International GPE VI-A vintage 2008 returned 16.3 percent net IRR and 2.1 times net multiple per CalPERS. The Robinson and Sensoy JFE 2016 result that the hot-money underperformance is sharply attenuated against public markets is mirrored on the upside: the 2008 to 2010 vintages outperformed public markets by substantial margins on a vintage-weighted basis. The Cambridge Associates US private equity index reports that the 10-year and 15-year IRRs anchored on this regime remain in the 14 to 16 percent range even after subsequent vintage drag (Cambridge Associates US PE VC benchmark commentary 1H 2025).

Regime 3: 2011 to 2015, Above-Trend Returns and Growing Dispersion

The 2011 to 2015 vintages were raised into a recovering economy and benefited from a sustained period of multiple expansion and cheap debt. Carlyle Partners VI vintage 2013 returned 15.8 percent net IRR and 2.0 times. Silver Lake Partners IV vintage 2013 returned 20.9 percent net IRR and 2.8 times. Permira V vintage 2014 returned 20.6 percent net IRR and 2.8 times. CVC VI vintage 2014 returned 15.6 percent net IRR. Hellman and Friedman Capital Partners VII vintage 2011 stands out at 24.6 percent net IRR and 3.4 times. This is the vintage cohort where buyout persistence appears strongest in the public LP data, with top-quartile names compounding into the next fund cycle.

Regime 4: 2016 to 2020, Peak Dispersion and Sector Concentration

The 2016 to 2020 vintages produced the best top-quartile and median outcomes of the modern era but the worst bottom-decile outcomes. Advent GPE VIII-B vintage 2016 returned 15.5 percent net IRR and 2.1 times. CVC VII A vintage 2018 returned 20.0 percent net IRR and 2.1 times. Apollo IX vintage 2019 returned 15.3 percent and 1.5 times. Vista VII-Z vintage 2019 returned only 4.8 percent and 1.2 times despite Vista’s reputation as a top-quartile sponsor in software, a clear sign of the within-GP dispersion that the Korteweg-Sorensen luck-versus-skill decomposition predicts. Hellman and Friedman IX vintage 2020 is at 13.8 percent and 1.7 times. The Cambridge Associates US PE index reports vintage 2016 returned only 0.6 percent in the first half of 2025, the worst of any meaningfully sized vintage.

Regime 5: 2021 to 2023, Rate Compression and Early IRR Concerns

The 2021 vintage is widely regarded as the worst recent vintage of the modern era. GPs deployed capital at peak EBITDA multiples in 2021 and 2022, with median deal multiples at 15 to 20 times EBITDA per industry trade press, and the subsequent rate-driven multiple compression has cut into NAV. Thoma Bravo XIV vintage 2021 is at 6.4 percent net IRR and 1.2 times. Blackstone Capital Partners VIII vintage 2021 is at 13.0 percent and 1.2 times, on the strength of the firm’s distribution discipline. TPG Growth V vintage 2021 is at 12.8 percent and 1.4 times. Industry-wide, the 2021 vintage median sits at roughly 1.1 times TVPI and the top quartile at 1.6 times TVPI as of H1 2025 per the Preqin and PitchBook benchmarks (Preqin private capital indices methodology; PitchBook Q2 2025 benchmarks; Cambridge Associates 1H 2025 commentary).

Regime 6: 2024 to 2026, Early Data and DPI Starvation

The 2024 to 2026 vintages are too early to evaluate on realized net IRR. The headline industry-wide issue is DPI starvation. Per the Bain Global Private Equity Report 2026, global buyout deal value rose 44 percent to 904 billion dollars in 2025 and exit value rose 47 percent to 717 billion, but distributions remained stubbornly low. MSCI Private Capital Monitor reported distributions at roughly 6 percent of buyout AUM in the year to June 2025, against a 10-year average of about 14 percent. Per the Bain Outlook 2026, PE firms are sitting on around 33,000 unsold portfolio companies, and exit momentum stalled again by mid-2026 (Bain Global Private Equity Report 2026; Bain Private Equity Outlook 2026 Gaining Traction; Bain Midyear 2026 release Triple Shock).

Sector-Level Persistence

Confidence: MEDIUM. Sector breakdowns are available from Cambridge Associates and PitchBook indices, but public LP disclosures do not consistently sector-tag funds. The sector returns reported here are pooled across the Cambridge Associates US PE benchmark.

The Cambridge Associates US PE benchmark, first half 2025 commentary, reports the following sector returns for the first half of 2025:

Software and information technology dominate the top-quartile rankings of US buyout across the 2011 to 2020 cohorts, with Thoma Bravo, Vista, Silver Lake, Hellman and Friedman, and Advent International appearing repeatedly in the upper decile of vintage-year IRR. Healthcare buyout, dominated by Welsh, Carson, Anderson and Stowe, KKR Healthcare, and Bain Capital Healthcare, has shown more cyclical returns with mid-quartile placement on a five-year rolling basis. Industrial buyout has been more cyclical, with Brookfield, Stone Point, and Centerbridge showing top-quartile placements in select vintages but reverting to median in others. Consumer buyout has been the weakest sector on a vintage-weighted basis, with L Catterton, Sycamore, and Apollo Consumer producing wide cross-sectional dispersion. Real estate and infrastructure are separate asset classes typically reported outside the standard buyout index (Cambridge Associates 1H 2025 commentary).

GAP: Sector-level persistence quantification, that is, top-quartile of healthcare buyout vintage N to top-quartile of healthcare buyout vintage N-plus-1, is not separately disclosed in any single public LP report and would require manual cross-tagging of every fund in the dataset.

Size-Level Persistence and the Mega-Fund Discount

Confidence: HIGH on the size-tier breakdown; MEDIUM on the persistence claim because size persistence requires matching across vintages.

The Lerner, Schoar, and Wongsunwai 2007 LP heterogeneity finding shows that fund size matters enormously to LP outcomes. The Andonov, Cremers, and Sialm NBER 33194 result that PE alpha for pension plans flows primarily from access to top GPs rather than selection skill is directly tied to the size dimension because pension plans tend to deploy disproportionately into mega-funds where access is partial and selection skill is limited.

The size tiers used in the industry are typically:

The Cambridge Associates 2024 calendar year commentary and the H1 2025 commentary note that mega-buyout funds have produced more compressed IRRs over the last three vintages compared to mid-market and lower-middle market funds, consistent with the Braun, Jenkinson, and Stoff finding that scale erodes persistence (Cambridge Associates 1H 2025; Cambridge Associates CY 2024 commentary).

The 7 to 8 percentage point Korteweg-Sorensen long-term spread persists most strongly within the middle-market and lower-middle-market segments, where the GP universe is large and the noise content is high but where the skill component is also more clearly attributable to GP-specific operating capability rather than to the access and capital markets execution that dominate mega-fund returns. The Lerner, Schoar, and Wongsunwai 21 percentage point endowment advantage is concentrated in the mid-market and lower-middle market where endowments commit early to first-time and emerging managers that public LPs are typically prohibited from accessing under their internal policies.

Geographic Persistence

Confidence: HIGH for US; MEDIUM for Europe; LOW for Asia because of FX and reporting heterogeneity.

Public LP disclosure data is heavily US-skewed because most US public pensions invest predominantly in US funds and report in USD. CalPERS and CalSTRS each have small but meaningful European and Asian PE allocations. Permira, CVC, EQT, Apax, BC Partners, Cinven, and Bridgepoint are the European mega-firms whose dollar-equivalent IRR is reported by CalPERS and CalSTRS.

European PE has produced top-quartile cross-cycle returns: CVC European Equity Partners III vintage 2001 at 41 percent net IRR per CalPERS, Permira Europe III vintage 2004 at 26.6 percent net IRR per CalPERS, Permira V vintage 2014 at 20.6 percent, CVC VI vintage 2014 at 15.6 percent. The European persistence pattern matches the US pattern, with Permira and CVC repeatedly placing in the top decile of their vintage and EQT and Cinven placing in the top quartile.

Asian PE persistence is the weakest of the three geographies on a vintage-weighted basis. KKR Asian Fund vintage 2007 returned 13.6 percent net IRR per CalPERS, which is below the 17 percent typically associated with top-quartile pan-Asian funds of that vintage. Bain Capital Asia Fund V vintage 2023 is too early to evaluate. Asia-Pacific deal multiples have been higher than the global average, with the Bain Asia Pacific Report 2026 noting median deal multiples rising to 13.4 times EBITDA in 2025 from 11.9 times in 2024 (Bain Asia Pacific Private Equity Report 2026).

GAP: Emerging market PE persistence is not adequately covered by US public LP disclosures and requires separate study from IFC, EMPEA, and the World Bank International Finance Corporation reports.

2024 to 2026 Latest PE Fund Performance

Confidence: HIGH for the headline metrics; MEDIUM for the fund-level commentary because most 2024 to 2026 reporting is still through the 120-day GP financial statement lag.

The most recent quarterly disclosures available as of June 24, 2026 are based on September 30, 2025 financial statements for the US LP universe. The headline metrics from the largest LPs:

The Cambridge Associates US PE Index returned 3.9 percent in the first half of 2025, with growth equity at 4.9 percent and buyouts at 3.6 percent. Eight vintage years (2016 to 2023) accounted for 85 percent of index NAV, with six-month returns ranging from 0.6 percent for vintage 2016 to 6.9 percent for vintage 2023. Q1 2025 distributions of 78.9 billion and Q2 2025 of 67.6 billion suggest a slower 2025 than 2024 for both calls and distributions (Cambridge Associates 1H 2025).

The American Investment Council 2025 Public Pension Study reports the 10-year median annualized PE return for 200 US public pensions at 13.5 percent, higher than all other asset classes. 89 percent of US public pensions invest in PE, and 94 percent of institutional investors plan to increase or maintain their PE allocation (AIC 2025 Pensions Report).

Bain 2026 DPI Starvation Analysis

Confidence: HIGH on the headline metrics from the Bain 2026 report.

The Bain Global Private Equity Report 2026 and the Bain Outlook 2026 jointly document the most significant industry-wide structural issue of the current era: the DPI starvation that has built up across the 2018 to 2022 vintages and is now compressing LP re-up capacity for the 2024 to 2026 fundraising cycle.

The headline DPI metric per the Bain 2026 reports: distributions to paid-in capital ran at approximately 6 percent of buyout AUM in the year ended June 30, 2025, against a 10-year average of about 14 percent. The roughly 800 basis point shortfall in annual DPI implies that LPs have received only about 43 percent of the distributions they would have expected based on the long-run average. The compounding effect across multiple vintage cohorts has produced an estimated 33,000 unsold PE-backed portfolio companies in the global universe.

The exit value side of the equation: global buyout exit value rose 47 percent to 717 billion dollars in 2025, recovering from the 2023 to 2024 trough but still below the 2021 peak of approximately 938 billion dollars. The Bain Midyear 2026 release titled “Winning firms will focus on what they can control” documents that exit momentum has stalled again by mid-2026, with the triple shock of rate uncertainty, geopolitical risk, and AI-driven valuation dispersion compressing the IPO and strategic-sale exit channels (Bain Midyear 2026 release).

The deal value side: global buyout deal value rose 44 percent to 904 billion dollars in 2025, reflecting a partial recovery from the 2022 to 2024 trough but still below the 2021 peak. The combination of high deal value and low exit value means that PE GPs are accumulating capital in newer vintages without releasing capital from older vintages, which compresses LP cash flow.

The DPI starvation thesis has direct implications for the 2024 to 2026 vintage persistence outlook. LPs facing constrained re-up capacity will be more selective in their commitments, which favors the largest brand-name GPs and disadvantages the mid-market and lower-middle market sponsors where the Korteweg-Sorensen skill component is most clearly attributable. The Andonov, Cremers, and Sialm NBER 33194 finding that PE alpha for pension plans flows primarily from access to top GPs is exactly what we should expect to see continue into the 2024 to 2026 cycle.

The CT PE Continuation Vehicle Discount-to-NAV Tracker 2024 to 2026 (Wave 12) documents the empirical CV pricing dynamics that the Bain 2026 reports describe in aggregate. The GP-led continuation vehicle has become a dominant exit mechanism for late-cycle PE assets that lack a strategic or financial buyer at acceptable valuation. Per Jefferies H1 2025 global secondary market review, total secondary transaction volume reached 162 billion dollars in 2024, a 45 percent increase from 2023, with GP-led deals at 75 billion in 2024 and climbing to a record 47 billion through H1 2025 (Jefferies H1 2025 secondary review). Per Lazard’s 2025 secondary market report, continuation vehicles represent 86 to 89 percent of GP-led activity.

Counter-Narrative Findings: The Six Pillars

Confidence: HIGH. The counter-narrative is well documented in the academic literature and increasingly accepted by the institutional LP community.

The counter-narrative to the consensus view that PE delivers persistent outperformance rests on six pillars, each with a primary academic source.

Counter 1: Persistence Has Declined Since 2010

Braun, Jenkinson, and Stoff JFE 2017 deal-level data and HJKS NBER 28109 buyout-versus-venture replication both find that buyout persistence has substantially declined post-2001 and is statistically indistinguishable from random by the post-2007 era. Top-quartile-to-top-quartile transition probability has dropped from roughly 0.42 in the Kaplan-Schoar 2005 sample (1980 to 2001 vintages) to approximately 0.27 in the post-2001 buyout sample.

Counter 2: Size Discount on Mega-Funds

Mega-funds have underperformed the broader benchmark on a vintage-weighted basis. The Cambridge Associates 1H 2025 commentary reports that growth equity at 4.9 percent has outperformed buyout at 3.6 percent year-to-date, consistent with the size discount thesis. The KKR 2006 Fund LP net IRR of 8.2 percent at the CalPERS as-of date of September 30, 2025 illustrates the underperformance of mega-funds raised at the top of the 2006 to 2007 buyout cycle.

Counter 3: Denominator Effect 2022 to 2024 Reversal

The 2022 public equity market crash compressed the denominator of LP allocation ratios and pushed PE allocation percentages above policy targets for nearly every major US public pension plan. The 2023 to 2024 partial recovery in public markets eased the denominator effect, but the PE allocation drift contributed to the 2024 to 2025 fundraising slowdown.

Counter 4: Rate Environment Compression 2024 to 2026

The Bain Outlook 2026 reports that the era of cheap debt and easy multiple expansion is gone. Deal multiples rebounded in 2025 to 13.4 times EBITDA in Asia Pacific from 11.9 times in 2024, but the underwriting case for further multiple expansion is no longer credible. The implication for vintage 2024 to 2026 returns is that GPs must generate alpha from operating improvement and revenue growth rather than from balance-sheet debt and rate-driven multiple expansion.

Counter 5: GP-Led Continuation Vehicles Hide True Performance

The GP-led continuation vehicle has become a dominant exit mechanism for late-cycle PE assets that lack a strategic or financial buyer at acceptable valuation. The pricing concern documented by Phalippou is that CV pricing typically clears between 95 and 105 percent of NAV, suggesting that the underlying fund NAV is being marked at carry-incentive levels. Cross-link: this report’s Wave 12 CV Discount documents the empirical NAV discount on traded LP positions (ILPA continuation fund guidance; ILPA continuation fund principles best practices; Houlihan Lokey 2024 continuation fund study).

Counter 6: PE Outperformance is Risk-Adjusted Illusory

Phalippou 2020 and Andonov, Cremers, and Sialm NBER 33194 both argue that PE alpha, once adjusted for the debt and small-cap risk loadings of buyout deals, is statistically indistinguishable from zero. The Brown, Gredil, and Kaplan NBER 22493 finding that PE GPs manipulate reported NAV at fundraising further suggests that the as-reported IRRs in the public LP disclosures are biased upward at the precise moment of LP re-up.

Phalippou December 2020 “Inconvenient Fact” Analysis

Confidence: HIGH on the source documentation; MEDIUM on the magnitude because the underlying datasets are licensed.

Ludovic Phalippou’s June 10, 2020 paper “An Inconvenient Fact: Private Equity Returns and the Billionaire Factory” is the single most cited contemporary critique of the consensus view of PE outperformance. The paper is hosted on SSRN at abstract 3623820 (SSRN 3623820) and is reviewed in the Oxford SBS commentary (Oxford SBS commentary) and Institutional Investor coverage (Institutional Investor coverage).

The headline findings are:

The Phalippou critique rests on three methodological pillars: first, the use of net-of-fee MOIC rather than headline IRR removes the cash-flow-timing optionality that GPs exploit through subscription credit lines; second, the comparison against an appropriate small-cap, debt-loaded public market equivalent removes the size and debt premia that account for much of the apparent buyout outperformance; third, the use of multiple datasets (Burgiss, Preqin, Cambridge Associates) cross-checks for selection bias in any single source.

The CalPERS PE program since-inception net IRR of 11.3 percent at September 30, 2025 is fully consistent with the Phalippou 11 percent IRR thesis. The MassPRIM 7.3 percent fiscal 2025 PE return is below the Phalippou benchmark and reflects the 2024 to 2026 DPI starvation period. The AIC 2025 study finding of 13.5 percent 10-year median PE return across 200 US public pensions is above the Phalippou benchmark but is anchored on the 2008 to 2010 and 2011 to 2015 vintage cohorts that delivered above-trend returns.

American Investment Council 2025 Public Pension Study

Confidence: HIGH on the headline metric of 13.5 percent 10-year median PE return.

The American Investment Council, the trade association for the US private equity industry, published its 2025 Public Pension Study in July 2025 (AIC 2025 Pensions Report PDF; AIC release). The headline findings are:

The AIC 2025 study is a trade-association publication and should be read with the understanding that the AIC has an incentive to present PE returns favorably. The 13.5 percent 10-year median PE return is anchored on the 2008 to 2010 and 2011 to 2015 vintage cohorts, both of which delivered above-trend returns. The forward-looking 10-year median is likely to be lower as the strong 2008 to 2015 vintages roll out of the lookback window and the weaker 2016 to 2023 vintages roll in.

The AIC 2025 study’s 200-pension sample is the broadest public LP universe available and is consistent with the Public Plans Database hosted by the Center for Retirement Research at Boston College. The cross-validation between AIC and PPD raises confidence in the headline 13.5 percent metric.

Academic Literature Synthesis

Confidence: HIGH. The papers below comprise the canon of the PE persistence literature.

Combined synthesis of the foundational papers:

The arc of the literature is consistent: the early Kaplan-Schoar 2005 finding of strong persistence has been progressively refined and qualified by every subsequent paper. The current academic consensus is that buyout persistence in the post-2007 era is weak to nonexistent at the median GP, with the top decile and bottom decile of the distribution showing the strongest persistence and the middle 80 percent of GPs being statistically indistinguishable from one another. Venture capital persistence appears more durable, possibly because of the network and reputation dynamics that lock top VCs into the best deal flow.

2024 to 2028 Forward Outlook

Confidence: MEDIUM. Forward outlook is by definition uncertain; the underlying drivers are well documented but the magnitudes are speculative.

The 2024 to 2028 outlook for PE returns and persistence rests on four pillars.

Forward 1: Early Vintage 2024 and 2025 Reads

The Bain Outlook 2026 and the Cambridge Associates H1 2025 commentary both flag the 2021 to 2023 vintages as the weakest cohort of the modern era on early IRR reads. The 2024 and 2025 vintages are too early to read but face the same headwinds: peak EBITDA multiples, compressed exit windows, and limited DPI from prior vintages constraining LP re-up capacity.

Forward 2: GP Commitment Fund Structure

The shift toward larger GP commitments as a share of fund capital, driven by LP pressure and the ILPA principles framework, is expected to align GP and LP interests further but does not directly affect the underlying fund return distribution.

Forward 3: LP Allocation Rebalancing

US public pension and Canadian pension LP allocations to PE have plateaued in the 12 to 16 percent range as of fiscal 2025 and are expected to stabilize. The denominator effect of public market recovery in 2024 to 2026 is unwinding the over-allocation drift of 2022.

Forward 4: Wealth-Channel Semi-Liquid PE Impact

The growth of wealth-channel semi-liquid PE vehicles, dominated by Blackstone BPRP, Apollo AAA, KKR KREST, and Hamilton Lane HLPP, has expanded the PE LP universe to a roughly 1.5 trillion dollar retail and HNW pool over the past three years. The semi-liquid structure relies on continuous NAV strikes and has not yet been stress-tested through a sustained NAV markdown cycle. The Phalippou critique of inflated NAV and the Brown, Gredil, and Kaplan NAV manipulation result both suggest that the semi-liquid wealth-channel structure may face material gating and discount events in a 2026 to 2028 stress scenario (Bain Outlook 2026; Bain Midyear 2026 release).

Practitioner Framework for LP Allocation

Confidence: HIGH. The principles below are widely adopted by sophisticated LPs.

An LP allocation playbook that incorporates the persistence literature should follow eight rules.

  1. Vintage diversification is mandatory. No single vintage should account for more than 20 percent of total PE commitment. This is the single most important risk control because vintage-year effects dominate the cross-sectional fund return distribution.
  2. GP diversification within vintage. No single GP should account for more than 15 percent of vintage commitment. The Korteweg-Sorensen luck-versus-skill decomposition implies that any single fund observation has high noise content, and concentration in a single GP magnifies that noise.
  3. Re-up decisions require fundamental due diligence beyond IRR. The Brown, Gredil, and Kaplan 2016 NAV manipulation result means that the most recent IRR snapshot at re-up time is the most biased input. Re-up due diligence should focus on attribution to specific deals, sector mix, debt at entry, and operating value creation, not on headline IRR.
  4. Anchor and co-invest LPs capture better economics. The Lerner, Schoar, and Wongsunwai endowment advantage was driven in part by early commitment to first-time and emerging managers. Anchor LPs receive lower management fees, higher co-invest allocations, and earlier capital deployment in best-in-class managers.
  5. GP-led continuation vehicle opt-in versus opt-out requires its own framework. The ILPA continuation fund guidance from 2023 and the ILPA continuation fund disclosure template from 2026 provide the standards. The default LP posture should be opt-out unless the CV terms preserve the existing economics and the underlying asset thesis is materially intact. Cross-link: this report’s Wave 12 CV Discount documents the empirical pricing of CV transactions and the LP cost of opt-in versus opt-out.
  6. Top-quartile-to-top-quartile is not a sufficient re-up heuristic post-2007. Buyout persistence has collapsed to near-random per HJKS 2020. LPs should base re-up on attribution, GP team continuity, and sector specialization rather than on a quartile ranking alone.
  7. Size-tier discipline. Mega-funds have underperformed mid-market and lower-middle market on a vintage-weighted basis in the 2010s and 2020s. LPs should allocate disproportionately to mid-market and lower-middle market funds, subject to access and minimum-ticket constraints.
  8. Public market equivalent benchmarking is the only honest performance measure. Net IRR alone is gameable through cash flow timing. PME against the relevant public benchmark, Kaplan-Schoar modified PME, and the Korteweg-Nagel 2024 risk-adjusted alpha are the methodologies that should anchor LP performance attribution (ILPA continuation fund considerations; ILPA principles best practices; Korteweg Nagel risk-adjusted returns of private equity funds new approach).

Limitations and GAPs

Confidence: HIGH on the limitations described; this section is intentionally transparent about what the public LP disclosure replication cannot do.

The single largest gap in any public LP disclosure replication of Korteweg-Sorensen is the inability to estimate the spurious persistence component from cohort overlap because the public LP universe does not consistently disclose the underlying portfolio company entry and exit dates that the deal-level papers like Braun, Jenkinson, and Stoff use. The public LP disclosures provide fund-level cash flow summaries only. The implication is that public LP replication of Korteweg-Sorensen will tend to overstate the true skill component because it cannot net out the spurious persistence from cohort overlap.

A second meaningful gap is fund-level sector tagging. CalPERS, CalSTRS, Oregon, and the other public LPs do not consistently tag funds with a sector classification, so sector-level persistence quantification at the LP level requires manual tagging from fund prospectus and offering materials.

A third gap is that the public LP universe is heavily US-skewed and US public-pension skewed. European and Asian funds appear in the data only to the extent that the US public LPs commit to them, and the smaller emerging-market and lower-middle-market funds are systematically under-represented because access constraints prevent the US public LPs from investing in them.

A fourth and final gap is that the public LP disclosures are based on as-of dates that lag the present by approximately two quarters because of the GP 120-day financial statement window. Any analysis dated June 24, 2026 therefore reflects fund returns as of December 31, 2025 at the latest and September 30, 2025 for the deepest dataset (CalPERS PEP).

The Korteweg-Sorensen luck-versus-skill decomposition remains the single most important finding of the modern PE literature: the skill component of any given PE fund’s realized return is small relative to the noise content, the cross-vintage persistence has substantially declined since the early 2000s, and the LP allocation playbook should be built around vintage and GP diversification rather than around top-quartile-chasing. The public LP disclosures from CalPERS, CalSTRS, Oregon, Washington, Massachusetts, New York, Florida, and Texas, together with the Burgiss to MSCI dataset that underlies the academic studies, provide the most complete picture of PE fund persistence available to any LP or analyst as of June 24, 2026.

This report sits within a larger CT Acquisitions wave research program. Related reports cross-link as follows:

Sources

All sources below are primary academic publications, primary LP disclosures, or primary industry trade publications. Each is publicly accessible at the URL provided.

Academic Papers (Peer-Reviewed Primary)

Public LP Disclosures

Industry Benchmark Publications

Frequently Asked Questions

Related research: for the LMM M&A buyer-pool 3-5x expansion 2018-2026 across 5 cohorts (family offices 651 to 4,067 per Preqin/BlackRock, Stanford GSB 681 search funds + 94 record 2023, McGuireWoods independent sponsors 200 to 1,600 = 8x + Axial 27% LMM share, LMM PE platforms HVAC 8 to 35+ + dental DSO 12 to 35+, SBA FY25 $8.29B + 7,003 deals), see the 2018-2026 US M&A Buyer-Pool Influx Report.

Related research: for 40+ named PE-backed bankruptcies 2020-2026 with tranche-level recovery extraction (Moody’s 2024 first-lien recovery 49.2% vs 76.4% long-run average; PESP 56% / 70% / 54% PE-backed share of $500M+ bankruptcies; Steward + Envision + Prospect + Joann + Rite Aid + Red Lobster + Yellow + Pluralsight Vista $4B equity wipe; Serta 5th Cir Dec 31 2024 + Robertshaw June 20 2024 LME precedents), see the 2020-2026 PE Bankruptcy Recovery Rate Report.

Related research: for 40+ named Delaware Chancery + Supreme Court PE rulings 2020-2026 (Fortis Advisors v J&J $1B+ Sept 4 2024 / $811M Jan 26 2026 / Sup Ct partial reversal Jan 12 2026; Tornetta v Musk $55.8B reversed Dec 19 2025; Sunder Energy v Jackson Dec 10 2024 blue-pencil refusal; SB 313 July 17 2024 + SB 21 March 25 2025 legislative override), see the 2020-2026 Delaware Chancery PE Litigation Report.

Related research: for US private credit $2T AUM 2025 (Preqin); BCRED $82.2B; FSK Q1 2026 cycle-marker (NAV -9.9% / 4.2% non-accruals / $558M loss / 7% dividend cut + class action); Pluralsight $1.3B/$1.5B wiped Aug 2024 (ARCC marked 48 cents); KBRA Q4 2025 default 3.4%; OBDC+OBDE merger Jan 13 2025; Fed FSR May 2026, see the 2024-2026 Private Credit + BDC Performance Report.

What is the CalPERS Private Equity Program net IRR since inception?

The CalPERS Private Equity Program reports a since-inception net IRR of 11.3 percent on a net multiple of 1.5 times as of September 30, 2025, per the CalPERS PEP fund performance review. The figure is dollar-weighted across every active and exited partnership investment since the program’s inception in 1990 and is the deepest single LP fund-level disclosure publicly available.

What is the HJKS 2020 top-quartile-to-top-quartile collapse?

Harris, Jenkinson, Kaplan, and Stucke, NBER Working Paper 28109, dated November 2020, documented that the top-quartile-to-top-quartile transition probability for US buyout funds collapsed from approximately 0.42 in the 1980 to 2001 vintages (Kaplan-Schoar 2005 sample) to approximately 0.26 in the post-2000 buyout cohort. The collapse is attributed to mega-fund scaling, new LP capital entry, and GP rotation. Venture capital persistence remained at approximately 0.45 in the same study.

What is the Phalippou “Inconvenient Fact” thesis?

Ludovic Phalippou’s June 10, 2020 SSRN paper “An Inconvenient Fact: Private Equity Returns and the Billionaire Factory” argues that across three large public datasets, average PE net multiples of money cluster between 1.55 and 1.63 times, implying roughly 11 percent annualized net IRR, which is indistinguishable from public equity benchmarks once balance-sheet debt and beta are adjusted. The total PE carry collected since 2006 is estimated at 230 billion dollars and PE multibillionaires grew from 3 in 2005 to 22 in 2020.

What is DPI starvation and why does it matter?

DPI starvation describes the period from 2024 to 2026 in which distributions to paid-in capital fell to approximately 6 percent of buyout AUM, against a 10-year average of about 14 percent, per the Bain Global Private Equity Report 2026. The 800-basis-point shortfall in annual DPI has produced an estimated 33,000 unsold PE-backed portfolio companies and compressed LP re-up capacity for the 2024 to 2026 fundraising cycle.

Why has buyout persistence declined since 2000?

The HJKS NBER 28109 attributes the buyout persistence collapse to three factors: first, mega-fund scaling that erodes the operating value-creation contribution to fund-level returns; second, new LP capital entry (sovereign wealth, family offices, wealth-channel semi-liquid vehicles) that bid up entry multiples; third, GP rotation, in which senior partners depart to launch new firms, breaks the implicit continuity of the GP brand.

What is the size discount on mega-funds?

Mega-funds (above 10 billion dollars) have produced more compressed IRRs over the last three vintages compared to mid-market (1 to 3 billion) and lower-middle market (below 1 billion) funds, per the Cambridge Associates CY 2024 and 1H 2025 benchmark commentaries. The Lerner, Schoar, and Wongsunwai 2007 endowment advantage of 21 percentage points is concentrated in the mid-market and lower-middle market where endowments commit early to first-time and emerging managers.

What is the public market equivalent (PME) framework?

The PME framework, introduced by Kaplan and Schoar JF 2005, compares PE fund cash flows to an equivalent investment in a public market index (typically the S&P 500). A PME above 1.0 indicates outperformance versus the public benchmark. Harris, Jenkinson, and Kaplan JF 2014 reported average buyout PME of 1.20 to 1.27. Modified PME (mPME) and direct alpha extensions refine the methodology for cash flow reinvestment assumptions.

Should LPs continue to allocate to PE in 2024 to 2028?

The AIC 2025 study reports that 94 percent of institutional investors plan to increase or maintain their PE allocation. The 10-year median annualized PE return for 200 US public pensions is 13.5 percent. However, the Phalippou risk-adjusted illusory thesis, the HJKS persistence collapse, and the Bain 2026 DPI starvation all suggest that the forward-looking 2024 to 2028 vintage returns are unlikely to match the 2008 to 2015 cohort. LPs should follow the eight-rule practitioner framework with emphasis on vintage diversification, GP diversification, size-tier discipline, and PME benchmarking.

About the Author

CT Acquisitions is an independent research and acquisition advisory firm focused on lower-middle-market and middle-market private equity, family-office direct investment, and institutional LP allocation strategy. The CT Acquisitions research desk publishes a continuous wave research program covering PE fund persistence, continuation vehicle pricing, BDC and bankruptcy recovery, Chancery litigation outcomes, SEC adviser registration, and vertical-by-vertical EBITDA multiple bands across the buyout cycle. This report sits within Wave 16 of the program. For research inquiries, partnership commitments, or direct co-invest access, contact the CT Acquisitions research desk via the website. All research is derived from publicly disclosed primary sources and is provided without paid Burgiss, MSCI, Preqin, or PitchBook subscription dependency.

Last updated: June 24, 2026.