Evidence Atlas · Private Equity Research

Private Equity Evidence Atlas

Read the findings on value creation, performance persistence, benchmarks, leverage, covenants, US tax and regulation. Check each claim’s dated sources and limits before using it.

Review results

The useful answer is often a boundary, not a universal number.

The reviews separate empirical evidence, vendor methodology, market context and controlling authority. That produces usable conclusions without inventing a current market range from an agency report or treating an older sample as live evidence.

EA-01: value creation.Use a deal-specific bridge with the attribution order disclosed. The reviewed studies do not support one portable ranking or fixed IRR contribution range.
EA-02: persistence.Past performance contains information, but US buyout top-quartile persistence weakens when measured with information available at successor fundraising. Venture capital and different performance metrics must remain separate.
EA-03: benchmarks.MSCI, Preqin, PitchBook and Cambridge Associates use different populations and methods. A current gap or winner cannot be quantified without normalized same-date, same-fund extracts.
EA-04A: leverage.Current agency policy and broad credit reports do not establish a representative lower-middle-market sponsor entry leverage range. None is published.
EA-04B: covenants.The current Federal Reserve bank survey excludes M&A-purpose loans from its covenant question. It cannot establish sponsor-loan covenant-lite prevalence or a universal package.
EA-04C: US tax.Section 163(j) and section 1061 statements are limited to general US federal rules, dated by taxable year and subject to entity, partnership, exception and transaction facts.
EA-04D: US regulation.The vacated SEC adviser rules, adopted Form PF timeline, pending non-final April 2026 proposal and annual HSR threshold are recorded with distinct statuses and dates.

Claims and sources

Seven contested claims, with the method left visible.

About the review record

EA-01 through EA-03 have completed primary-source method reviews. The former EA-04 currency warning is split into leverage, covenants, US tax and US regulation so one dated source cannot make unrelated claims look current. Every card joins the claim to direct sources, stance, dependencies, scope, currentness rules and the canonical page that owns the practical explanation.

How to read the review labels

Claim register

7 method-reviewed public claims

Working finding

No stable universal return split was found. A deal-specific bridge should separate operating performance, multiple movement and net-debt change, while disclosing its order, return basis, sample and exclusions.

The primary-source review confirms that the studies ask different questions with different samples and methods. Operations and earnings growth matter in several settings, while multiple changes, leverage and tax benefits, entry price, selection and financing conditions also affect observed outcomes. The linked guide now presents the bridge as a scoped, synthetic teaching method rather than a universal ranking.

Unit: Deal-level value-creation attribution
Segment: Buyouts; not validated specifically for the lower middle market
Geography: Mixed international and US/European evidence
Evidence horizon: Publications reviewed through 2025; observed deal or entry vintages extend through 2018, while a Swedish firm-outcomes study extends through 2022
Excluded: Not a fund-level net-return decomposition and not a forecast of current deal returns

Currentness rulePrimary-source review recorded 2026-08-01

Do not publish a universal percentage or basis-point contribution. Treat publication date and sample end separately, and state the bridge order, deal/fund grain, gross/net basis, benchmark, leverage treatment and holding-period method.

Next trigger: Reopen when a genuinely independent primary buyout-attribution study adds a later, fully disclosed and comparable realized sample.

Primary-source register5 reviewed sources

Qualifies

Private Equity: Value Creation and Performance

Christian Graf, Christoph Kaserer and Daniel M. Schmidt, Private Equity: Value Creation and Performance (2012)

Sample boundary: 2008-06

Open primary source

Publisher metadata and chapter methodology · retrieved 2026-08-01

Supports an operating-performance lane in a narrow sample, but its residual construction and fixed-leverage assumption do not establish a universal buyout split.

CEPRES provider-linked evidence

CEPRES supplied voluntary proprietary transaction data, and the public author record lists one coauthor as CEPRES-affiliated. The directly attributed result is limited to regulated financial institutions.

The 87-90% residual result applies to regulated financial-institution subsets under a fixed-leverage assumption. The authors call extension to other industries indirect evidence.

Recorded method fields

Data provider: CEPRES voluntary member-contributed proprietary transaction cash flows; coauthor Daniel M. Schmidt was affiliated with CEPRES
Unit: Fully realized transaction-level cash flows; attribution subsets include 280 financial-institution PE deals, 134 buyouts and 67 bank holdings
Return basis: Gross deal returns before fund management fees and carried interest
Value-created definition: Linearized IRR split between public-industry P/B market timing and a residual labeled operational improvement; interaction omitted
Market / beta: No asset-beta or factor-risk model. P/B changes proxy market timing in the attribution subset; nine Datastream sector total-return indices are used only in broader performance comparisons
Leverage: Deal leverage is not observed in the attribution; regulated-sector leverage is assumed constrained or fixed
Holding period: Exact entry-to-exit dates; holding periods shorter than one quarter excluded

Contradicts

Do Buyouts (Still) Create Value?

Shourun Guo, Edith S. Hotchkiss and Weihong Song, Do Buyouts (Still) Create Value? (2011)

Sample boundary: 2006

Open primary source

NBER working-paper record and published-study methodology · retrieved 2026-08-01

Finds operating gains, industry multiple changes and tax benefits all economically important, contradicting a universal operations-first ranking.

Independent public US take-private data

Public-to-private US sample assembled from public deal, filing, accounting and financing sources; independent of the CEPRES stream but old and large-deal-specific.

Large US public-to-private deals ending in 2006 are not a current or lower-middle-market sample. Terminal-value estimation and post-buyout-data availability introduce model and selection risk.

Recorded method fields

Data provider: SDC, Dealogic, SEC filings, Compustat, Dealscan, news archives and Fama-French portfolios
Unit: Company/deal returns to total pre-buyout and post-buyout capital
Return basis: Includes transaction fees in invested capital and company-paid monitoring fees in expenses; does not net LP management fees or carry
Value-created definition: Nominal and market/industry-adjusted total-capital returns, including interim financing cash flows and terminal value
Market / beta: S&P 500 adjustment assumes asset beta of one; industry adjustment assumes the deal asset beta equals the levered beta of its Fama-French industry
Leverage: Measures pre- and post-buyout leverage, tax benefits and financing; later regressions instrument leverage with loan rates
Holding period: Mean observed duration 3.86 fiscal years and median 3.32; missing terminal values use matched-industry EBITDA multiples with sensitivities

Qualifies

Value Creation in Private Equity

Markus Biesinger, Cagatay Bircan and Alexander Ljungqvist, Value Creation in Private Equity (2020; revised 2023)

Sample boundary: 2017

Open primary source

EBRD Working Paper 242 methodology and sample · retrieved 2026-08-01

Links successful execution of several value-creation actions to gross outcomes, but does not estimate one additive buyout return split.

EBRD data-owner and author overlap

The EBRD is the data-owning LP and employer of two authors; the emerging-market sample is dominated by growth and venture investments rather than buyouts.

A mixed emerging-market sample dominated by growth and venture deals cannot establish a universal developed-market buyout ranking. EBRD is both the data-owning LP and employer of two authors.

Recorded method fields

Data provider: EBRD LP archive, GP investment memos and quarterly reports, EBRD monitoring reports, and Orbis company accounts
Unit: Deal/company; 1,136 coded value-creation plans and 946 realized-return observations
Return basis: Gross deal MOIC and Kaplan-Schoar-style PME versus MSCI Emerging Markets; not net to LPs and not risk, liquidity or leverage adjusted
Value-created definition: Execution of 23 coded actions across five strategy groups linked to realized deal returns and company outcomes
Market / beta: No beta model; PME benchmark does not hold risk, liquidity or leverage constant
Leverage: Leverage is measured as a company outcome rather than isolated as an additive return contribution
Holding period: Controls for duration and entry/exit years; company outcomes extend up to five years after exit

Qualifies

Pricing, Returns, and Sources of Value Creation in Buyouts

Roland Fuss, Stefan Morkoetter, Dominic Rainsborough and Tereza Tykvova, Pricing, Returns, and Sources of Value Creation in Buyouts (2025)

Sample boundary: Entry vintages through 2018; exit-observation cutoff unresolved

Open primary source

SSRN abstract, revision metadata and disclosed sample · retrieved 2026-08-01

Shows that entry pricing changes performance and the apparent EBITDA-versus-multiple mix, reinforcing the method and sample boundary without clearing a numeric share.

Unresolved proprietary-data source

The accessible primary record does not identify the proprietary provider or disclose enough return-basis, beta, leverage or holding-period detail to normalize the estimates.

This is a 2025 working paper with transactions initiated through 2018. Public primary metadata does not disclose the exit-observation cutoff or enough method detail to support a numeric attribution share.

Recorded method fields

Data provider: Proprietary and unidentified in accessible primary metadata; overlap with commercial datasets remains unresolved
Unit: Deal-level buyout transactions
Return basis: Unresolved in accessible primary metadata
Value-created definition: Relates entry and exit pricing, competition, fund-cycle stage, EBITDA growth and deal performance; no universal contribution share is disclosed
Market / beta: Unresolved in accessible primary metadata
Leverage: Unresolved in accessible primary metadata
Holding period: Unresolved in accessible primary metadata

Qualifies

Plundered or profitably pumped-up? The effects of private equity takeover

Anders Kärnä and Samantha Myers, Plundered or profitably pumped-up? The effects of private equity takeover (2025)

Sample boundary: 2022

Open primary source

Sveriges Riksbank Working Paper 444 methodology and results · retrieved 2026-08-01

Finds higher debt without an average productivity gain in its setting. This later-period firm-outcomes evidence qualifies universal operating-improvement language, but it is not sponsor-return attribution.

Independent Swedish registry data

Independent Swedish registry and PitchBook firm-outcomes stream; it does not measure sponsor or LP returns and therefore cannot supply return attribution percentages.

This is a firm-outcomes study, not a sponsor-return attribution. Sweden, matching design and the low-rate-heavy sample limit generalization, but it is a current-sample counterexample to universal productivity claims.

Recorded method fields

Data provider: Serrano Swedish company accounts and PitchBook deal identification
Unit: Firm-year operating and balance-sheet outcomes, not deal, fund or LP returns
Return basis: Not applicable; the study does not measure realized sponsor or LP returns
Value-created definition: Matched event-study changes in operating, employment, productivity, profitability, tax and debt outcomes after PE takeover
Market / beta: No return beta; models include firm, industry and calendar-year fixed effects
Leverage: Debt is a directly observed company outcome and rises about 20% after LBO, largely through short-term debt
Holding period: Event window runs from five years before to six years after takeover; no sponsor-return holding-period attribution

Primary-source refresh2026-08-01

Primary metadata and available methods were reviewed, and the canonical guide was reconciled to the method boundary. The studies differ in sample, unit, return basis and attribution method, so no universal driver ranking or fixed IRR contribution range is presented.

Primary sources checked: yes

Working finding

Past performance contains information, but prior fund quartile is not a universal investable rule. In US buyouts, top-quartile persistence largely disappears when performance is measured as investors could observe it at successor-fund fundraising; venture-capital persistence is stronger.

The studies answer different questions. Harris and coauthors separate final ranks from information available when the next fund raised and find little US buyout top-quartile persistence on the latter basis, especially after 2000, with weaker outcomes carrying more signal. Korteweg and Sorensen find persistent expected-return heterogeneity but little investable persistence because realized outcomes are noisy. Braun and coauthors find that relative PME and dollar-value persistence can diverge at deal, fund and manager grain. None turns the last fund's quartile into a standalone selection rule.

Unit: Fund sequences, private-equity firms and deal/fund manager-value measures
Segment: US buyout and venture capital must remain separated
Geography: US persistence evidence plus a multi-market proprietary buyout study
Evidence horizon: Publications reviewed through 2026; the newest underlying cash-flow boundary is 2020 and the studied fund vintages are older
Excluded: Not an endorsement, manager ranking, access model or claim that a prior quartile predicts an investable successor-fund outcome

Currentness rulePrimary-source review recorded 2026-08-02

State the strategy, performance measure, observation date, successor-fund timing, sample vintages and gross/net basis. Do not translate ex-post final ranks, manager NPV persistence or venture results into a live US buyout selection rule.

Next trigger: A later independent study adds post-2020 cash flows or tests observable-at-fundraising buyout persistence in newer vintages.

Primary-source register3 reviewed sources

Supports boundary

Has persistence persisted in private equity?

Robert S. Harris, Tim Jenkinson, Steven N. Kaplan and Rudiger Stucke, Has persistence persisted in private equity? (2023)

Sample boundary: Cash flows through 2020-12-31; fund vintages 1984-2015

Open primary source

Journal article, sample construction and fundraising-time persistence tests · retrieved 2026-08-02

Supports the central boundary: final ex-post ranks show more persistence than investors could observe at successor fundraising, and the US buyout result differs from venture capital.

Burgiss institutional cash-flow data

Institutional-investor cash-flow data provide a strong LP perspective but remain one proprietary provider stream and end in 2020.

The study separates final, ex-post performance from information observable when a successor fund was raised. Its vintage and cash-flow boundaries must remain visible, and buyout results must not be generalized to venture capital.

Recorded method fields

Data provider: Burgiss institutional-investor cash flows, now part of MSCI
Unit: Fund sequences within private-equity firms, separated into US buyout and venture-capital samples
Return basis: Net-to-LP fund performance, including PME and absolute-return measures; observable-at-fundraising and final ranks are analyzed separately
Value-created definition: Not a value-creation attribution study; the outcome is successor-fund performance persistence
Market / beta: PME comparisons use public-market opportunity cost; absolute-return measures are also reported
Leverage: Fund and deal leverage are not normalized into the persistence result
Holding period: Fundraising-time information is separated from later final performance; cash flows run through 2020

Qualifies

Skill and Luck in Private Equity Performance

Arthur Korteweg and Morten Sørensen, Skill and Luck in Private Equity Performance (2017)

Sample boundary: 1,924 fully liquidated funds from 831 firms; vintages 1969-2001

Open primary source

Peer-reviewed article record, accepted manuscript and variance-decomposition method · retrieved 2026-08-02

Finds persistent expected-return heterogeneity after controlling for overlapping funds, while also finding little investable persistence because realized outcomes are noisy.

Preqin fund-level return data

Commercial fund-level records and a model-based variance decomposition answer a different question from observable-at-fundraising cash-flow persistence.

The variance decomposition estimates persistent expected-return heterogeneity, but noisy realized fund outcomes leave little investable persistence. It does not make a prior quartile rank a standalone selection rule.

Recorded method fields

Data provider: Preqin fund-level performance records
Unit: Fund returns and private-equity firms across 831 managers
Return basis: Net-of-fees fund returns modeled through a variance decomposition; underlying cash-flow PME is not the primary input
Value-created definition: Not a deal-attribution study; persistence is decomposed into long-term firm, overlapping-fund and idiosyncratic components
Market / beta: No deal-level beta adjustment; the model separates persistent expected-return heterogeneity from noise and overlapping funds
Leverage: Fund and deal leverage are not isolated
Holding period: Overlapping contemporaneous funds are modeled explicitly to avoid mechanical persistence

Qualifies

Size, Returns, and Value: Do Private Equity Firms Allocate Capital According to Manager Skill?

Reiner Braun, Nils Dorau, Tim Jenkinson and Daniel Urban, Size, Returns, and Value: Do Private Equity Firms Allocate Capital According to Manager Skill? (2026)

Sample boundary: Fund vintages through 2011; deals through 2015; cash flows and fair values through 2018

Open primary source

Published article, proprietary sample and relative-PME-versus-NPV analysis · retrieved 2026-08-02

Shows why the metric matters: relative PME persistence can fade while dollar NPV persistence behaves differently. It cannot validate a prior-fund quartile selection rule.

Proprietary gross-deal data

The gross deal and firm dataset is distinct from an LP-net successor-fund rank test, and its cash-flow boundary ends in 2018.

This study measures gross deal and manager value under multiple metrics. Relative PME and dollar NPV can show different persistence patterns, so its result cannot be substituted for net-to-LP successor-fund quartile persistence.

Recorded method fields

Data provider: Proprietary deal, fund cash-flow and fair-value dataset
Unit: Buyout deals, funds and private-equity firms
Return basis: Gross deal-level relative PME and dollar NPV, with fund and firm aggregation
Value-created definition: Manager value is evaluated in both relative-return and dollar-value terms
Market / beta: Relative PME benchmarks public-market performance; NPV answers a different capital-allocation question
Leverage: Leverage is not converted into an LP-net persistence measure
Holding period: Deals run through 2015 and cash flows or fair values through 2018; this is not a live 2026 portfolio sample

Primary-source refresh2026-08-02

Primary methods were reconciled across final versus fundraising-time ranks, buyout versus venture, net fund returns versus gross manager value, and older sample boundaries. Past performance contains information, but no standalone prior-quartile buyout selection rule is presented.

Primary sources checked: yes

Working finding

Private-markets benchmarks are not interchangeable because source populations, inclusion rules, update lags, classifications and aggregation methods differ. Those differences do not, by themselves, prove a large current performance gap.

Current first-party methodology records from MSCI, Preqin, PitchBook and Cambridge Associates show materially different sourcing, population, lag, estimation, peer-group and maintenance rules. A historical four-database study nevertheless found qualitatively similar aggregate performance conclusions. Without licensed, same-date and same-fund extracts, the Atlas does not quantify a 2026 vendor gap or rank a winner.

Unit: Database populations, methodology rules and fund benchmark cohorts
Segment: Closed-end private-capital funds; eligibility and strategy definitions differ by provider
Geography: Global coverage as defined by each provider
Evidence horizon: First-party methodology pages reviewed through 2026-08-02; no normalized current performance extracts were compared
Excluded: Not a vendor ranking, quantified current performance gap, market-size estimate or claim that any database is complete

Currentness rulePrimary-source review recorded 2026-08-02

Treat methodological non-equivalence as verified and the size or direction of current performance disagreement as unmeasured. Quantification requires same-date, same-fund extracts normalized for strategy, vintage, geography, cash-flow basis and benchmark construction.

Next trigger: A provider materially revises its methodology or comparable licensed extracts become available for a reproducible overlap test.

Primary-source register5 reviewed sources

Qualifies

What Do Different Commercial Data Sets Tell Us About Private Equity Performance?

Gregory W. Brown, Robert S. Harris, Tim Jenkinson, Steven N. Kaplan and David T. Robinson, What Do Different Commercial Data Sets Tell Us About Private Equity Performance? (2015)

Sample boundary: Historical Burgiss, Cambridge Associates, PitchBook and Preqin datasets available to the authors

Open primary source

Working paper abstract, matched-dataset comparison and reported aggregate results · retrieved 2026-08-02

Qualifies the method-disagreement finding: the study found qualitatively similar historical aggregate conclusions, so methodology differences alone do not establish a large current performance gap.

Historical four-database comparison

The comparison directly uses Burgiss, Cambridge Associates, PitchBook and Preqin, but it is a historical snapshot rather than a current 2026 overlap test.

The historical comparison found qualitatively similar aggregate performance across four datasets, subject to sample and matching limitations. It does not quantify a current 2026 vendor gap.

Recorded method fields

Data provider: Burgiss, Cambridge Associates, PitchBook and Preqin
Unit: Commercial database fund samples and aggregate performance results
Return basis: Net fund cash-flow performance where available, compared across database samples
Value-created definition: Not a value-creation study; it tests whether database choice changes aggregate performance conclusions
Market / beta: Public-market-equivalent and absolute-return comparisons depend on the underlying database records
Leverage: Not separately measured
Holding period: Historical database snapshots; not a same-date 2026 extract

Supports boundary

MSCI Private Capital Benchmarks Summary Q1 2026

MSCI, Private Capital Benchmarks Summary Q1 2026 (2026)

Sample boundary: 2026-Q1

Open primary source

Official benchmark report, data population and methodology disclosures · retrieved 2026-08-02

Supports methodological non-equivalence through disclosed population, classification and lag rules; it does not quantify how far MSCI performance differs from another vendor.

MSCI LP cash-flow data

LP-contributed cash flows, rules-based classification, quarterly reconstruction and no subscription-line adjustment are provider-specific design choices.

This first-party methodology record explains MSCI's population, classification and reconstruction choices. It does not independently establish the magnitude of a vendor performance gap.

Recorded method fields

Data provider: Institutional-LP contributed fund cash flows
Unit: Since-inception quarterly fund and benchmark cohorts classified under MSCI PACS
Return basis: Net of fees and carried interest; subscription-line effects are not adjusted out in the report methodology
Value-created definition: Not applicable; the source defines benchmark construction and coverage
Market / beta: Public-market comparisons are separate from the private-capital peer benchmark construction
Leverage: Subscription-line use is not normalized out of reported fund cash flows
Holding period: Quarterly since-inception records; reconstruction may introduce an approximately 90-day reporting lag

Supports boundary

Private Capital Performance Data Guide

Preqin, Private Capital Performance Data Guide (2025)

Open primary source

Official collection, estimation, current-range and peer-group methodology · retrieved 2026-08-02

Supports non-interchangeability through its source-selection, estimation, recency and peer-group rules, without establishing a current direction or size of cross-vendor error.

Preqin multi-source data

Preqin combines GP, LP, freedom-of-information and public records under a provider source hierarchy and may estimate selected return fields.

This first-party guide discloses Preqin's source hierarchy, estimation and peer-group rules. It cannot prove a quantitative cross-vendor gap without normalized same-date extracts.

Recorded method fields

Data provider: GP submissions, LP and freedom-of-information disclosures, listed-manager reports and other public records, with one selected source per fund and as-of date
Unit: Fund observations and vintage/strategy peer groups
Return basis: Reported or, in limited cases, estimated IRR and multiples; quartiles are equal-weighted within eligible peer groups
Value-created definition: Not applicable; the source defines data collection and benchmark construction
Market / beta: Private peer-group statistics are distinct from public-market-equivalent analysis
Leverage: Fund-level financing effects are not standardized into a cross-vendor adjustment
Holding period: Current-range observations can span up to five quarters; peer groups may expand when minimum counts are not met

Supports boundary

PitchBook Benchmarks

PitchBook, PitchBook Benchmarks methodology (retrieved 2026)

Open primary source

Official benchmark help page and disclosed peer-group rules · retrieved 2026-08-02

Supports the method boundary by documenting peer-group construction and timing rules; it does not establish a current vendor ranking.

PitchBook proprietary data

The proprietary database, eight-fund peer-group minimum and conditional prior-quarter extension are provider-specific design choices.

The first-party page documents peer-group and quarter-extension rules. It supports methodological non-equivalence, not a claim that PitchBook is higher, lower or more complete than another vendor.

Recorded method fields

Data provider: PitchBook private-capital database
Unit: Preformed peer groups by vintage, location, fund size, access point and strategy
Return basis: Fund performance benchmarks subject to product methodology and an eight-fund minimum peer group
Value-created definition: Not applicable; the source defines benchmark construction
Market / beta: Private peer groups are not a public-market risk adjustment
Leverage: No cross-vendor leverage normalization is described
Holding period: Prior-quarter returns may be extended under disclosed conditions

Supports boundary

Private Investment Benchmarks FAQ

Cambridge Associates, Private Investment Benchmarks FAQ (undated; retrieved 2026)

Open primary source

Official benchmark population, source, exclusion and one-quarter-lag disclosures · retrieved 2026-08-02

Supports methodological non-equivalence through the disclosed source, universe and reporting lag; it cannot quantify a current same-fund performance gap by itself.

Cambridge Associates GP statements

Quarterly GP statements, an institutional-quality eligibility screen and a one-quarter lag define a distinct provider population.

The first-party FAQ documents Cambridge Associates' institutional-quality universe, exclusions, source records and one-quarter lag. It does not supply a normalized current comparison to other vendors.

Recorded method fields

Data provider: Quarterly GP financial statements collected by Cambridge Associates
Unit: Eligible institutional private-investment funds and benchmark cohorts
Return basis: Fund-level performance from GP financial statements, subject to Cambridge Associates inclusion rules
Value-created definition: Not applicable; the source defines benchmark population, sourcing and release timing
Market / beta: Private peer benchmarks are separate from public-market comparisons
Leverage: No cross-vendor financing normalization is disclosed on the public page
Holding period: One-quarter reporting lag due to private-fund manager reporting time frames

Primary-source refresh2026-08-02

Current first-party methodology records verify that provider populations, source hierarchies, lags, estimation, peer groups and maintenance differ. A historical cross-dataset study found qualitatively similar aggregate results, so no current gap or vendor winner is inferred without normalized extracts.

Primary sources checked: yes

Working finding

No universal current LBO leverage range is supported by the reviewed official sources. The former 6x supervisory concern is not a current market-clearing norm, and the available current agency reports cover different lending populations.

OCC and FDIC withdrew their participation in the 2013 leveraged-lending guidance in December 2025, so its 6x concern cannot be presented as a universal current agency rule or market norm. The 2025 Shared National Credit review covers large facilities shared by supervised institutions, while the May 2026 Financial Stability Report provides broad credit conditions. Neither identifies lower-middle-market sponsor entry leverage, and neither covers the entire private-credit market.

Unit: Supervisory policy, large shared credits and broad business-credit conditions
Segment: US leveraged lending; not a representative lower-middle-market sponsor-loan dataset
Geography: United States, with agency scope kept explicit
Evidence horizon: Official status and market-context sources reviewed through 2026-08-02; current deal-level entry multiples were not observed
Excluded: No market leverage range, underwriting recommendation or claim about all banks, private lenders or sponsor-backed deals

Currentness rulePrimary-source review recorded 2026-08-02

Do not publish a current leverage range from supervisory guidance or broad credit reports. A numeric market statement requires a dated, representative loan-level source with lender channel, company size, calculation definition and period disclosed.

Next trigger: A representative current lower-middle-market sponsor-loan dataset becomes publicly reviewable or a named agency changes its leveraged-lending policy.

Primary-source register3 reviewed sources

Supports boundary

Interagency Statement on OCC and FDIC Withdrawal of the Interagency Leveraged Lending Guidance

FDIC and OCC, Interagency Statement on Withdrawal of the Interagency Leveraged Lending Guidance (2025)

Open primary source

Official interagency withdrawal statement · retrieved 2026-08-02

Supports the published boundary that the withdrawn 2013 guidance's 6x concern is not a universal current supervisory rule or market-clearing leverage norm.

Official agency policy

The action speaks only for the named agencies and supervised institutions; it is not a loan-level market dataset.

The statement withdraws OCC and FDIC participation in the 2013 guidance. It is a supervisory-policy source, not evidence for a market-clearing debt/EBITDA range or a universal rule across lenders.

Recorded method fields

Data provider: Official agency action
Unit: Supervisory guidance applicable to institutions under the named agencies
Return basis: Not applicable; this is not a return or pricing dataset
Value-created definition: Not applicable; this source establishes the status and scope of prior guidance
Market / beta: Not applicable
Leverage: Explains withdrawal of agency participation; no current market leverage band is established
Holding period: Effective status checked as of 2026-08-02

Qualifies

Shared National Credit Program Review: 2025

Federal Reserve, FDIC and OCC, Shared National Credit Program Review: 2025 (released 2026)

Sample boundary: Credits originated by 2025-06-30

Open primary source

Official 2025 program review, population and origination cutoff · retrieved 2026-08-02

Supplies current large syndicated-credit risk context while demonstrating why the population cannot establish an LMM entry leverage range.

Shared National Credit large-loan data

The program covers commitments of at least $100 million shared by three or more supervised institutions, not a representative lower-middle-market sponsor-loan population.

The review provides current large syndicated-credit risk context. Its population does not represent lower-middle-market sponsor entry leverage or the full private-credit market.

Recorded method fields

Data provider: Federal banking agencies' Shared National Credit review
Unit: Large shared credit facilities
Return basis: Not applicable; supervisory credit-quality classifications and commitments
Value-created definition: Not applicable; this is credit-risk context
Market / beta: Not applicable
Leverage: Leveraged status and credit quality are supervisory classifications, not a current LBO leverage range
Holding period: Annual review with an origination cutoff of 2025-06-30

Context

Financial Stability Report - May 2026

Board of Governors of the Federal Reserve System, Financial Stability Report (May 2026)

Sample boundary: Market conditions generally through 2026-04-23

Open primary source

Official May 2026 report and broad business-credit indicators · retrieved 2026-08-02

Provides current financing-condition context but does not clear a numeric sponsor-backed LBO leverage claim.

Federal Reserve broad-credit context

Broad debt, spread and debt-service indicators do not reveal sponsor-specific entry debt/EBITDA or cover all private lenders.

This report supplies broad financing-condition context. It does not disclose sponsor-specific lower-middle-market entry debt/EBITDA or covenant prevalence.

Recorded method fields

Data provider: Federal Reserve and referenced market/credit data
Unit: Financial-system and broad business-credit indicators
Return basis: Not applicable; market and balance-sheet condition measures
Value-created definition: Not applicable; this is macro-financial context
Market / beta: Broad market spreads and conditions, not deal-specific risk adjustment
Leverage: Broad business leverage and debt-service measures; no sponsor entry multiple
Holding period: Conditions generally observed through 2026-04-23

Primary-source refresh2026-08-02

Official sources establish current agency-policy status and broad credit context, but their populations do not support a representative lower-middle-market sponsor entry leverage range. No numeric range is published.

Primary sources checked: yes

Working finding

No universal current covenant package or covenant-lite prevalence is supported for sponsor-backed lower-middle-market loans. The current bank survey is directional and excludes M&A-purpose loans from the relevant covenant question.

The Federal Reserve's April 2026 Senior Loan Officer Opinion Survey measures whether banks changed C&I covenant terms, but the relevant question excludes merger-and-acquisition-purpose loans. That makes it useful context for bank lending conditions, not evidence of sponsor-debt covenant prevalence. Actual protections remain contract-specific and must be read from the governing credit agreement within a defined loan market.

Unit: Bank survey responses and named loan-document provisions
Segment: US C&I lending context; sponsor-backed M&A loans are not measured by the relevant survey question
Geography: United States
Evidence horizon: Official survey and supervisory-policy sources reviewed through 2026-08-02
Excluded: No covenant-lite prevalence estimate, standard covenant package, legal interpretation or assumption about a specific credit agreement

Currentness rulePrimary-source review recorded 2026-08-02

Do not convert directional bank-survey responses or an undated market primer into sponsor-debt prevalence. State the loan market, period and document population, then verify each protection in the executed agreement.

Next trigger: A dated, representative sponsor-backed lower-middle-market loan-document study becomes publicly reviewable or the Federal Reserve changes the survey population.

Primary-source register2 reviewed sources

Qualifies

April 2026 Senior Loan Officer Opinion Survey - Table 1

Board of Governors of the Federal Reserve System, April 2026 Senior Loan Officer Opinion Survey, Table 1

Sample boundary: Responses due 2026-04-03; released 2026-05-04

Open primary source

Official April 2026 survey Table 1, covenant question and exclusions · retrieved 2026-08-02

Qualifies broad covenant narratives: the survey can describe direction in bank C&I terms, but not sponsor-loan covenant-lite prevalence or a standard package.

Federal Reserve bank survey (non-M&A C&I)

The relevant directional covenant question excludes M&A-purpose loans and surveys banks rather than executed sponsor credit agreements.

The survey measures directional changes in bank C&I terms and expressly excludes M&A-purpose loans from the relevant covenant question. It cannot establish sponsor-backed loan covenant-lite prevalence.

Recorded method fields

Data provider: Federal Reserve survey of senior loan officers
Unit: Respondent banks and directional changes in C&I lending standards and terms
Return basis: Not applicable; survey responses rather than loan-level pricing or performance
Value-created definition: Not applicable; this source measures lending-condition changes
Market / beta: Not applicable
Leverage: No borrower debt/EBITDA range is reported for sponsor-backed transactions
Holding period: Directional change over the survey period, not a stock measure of outstanding loan documents

Context

Interagency Statement on OCC and FDIC Withdrawal of the Interagency Leveraged Lending Guidance

FDIC and OCC, Interagency Statement on Withdrawal of the Interagency Leveraged Lending Guidance (2025)

Open primary source

Official withdrawal statement and continuing safety-and-soundness boundary · retrieved 2026-08-02

Provides policy context but cannot substitute for reading the governing agreement or sampling executed sponsor-loan documents.

Official agency policy

Supervisory principles and agency scope do not measure the prevalence of contractual protections in a defined loan market.

The statement withdraws OCC and FDIC participation in the 2013 guidance. It is a supervisory-policy source, not evidence for a market-clearing debt/EBITDA range or a universal rule across lenders.

Recorded method fields

Data provider: Official agency action
Unit: Supervisory guidance applicable to institutions under the named agencies
Return basis: Not applicable; this is not a return or pricing dataset
Value-created definition: Not applicable; this source establishes the status and scope of prior guidance
Market / beta: Not applicable
Leverage: Explains withdrawal of agency participation; no current market leverage band is established
Holding period: Effective status checked as of 2026-08-02

Primary-source refresh2026-08-02

The current bank survey is directional and excludes M&A-purpose loans from its covenant question; agency policy does not measure executed contract prevalence. No covenant-lite percentage or universal package is published.

Primary sources checked: yes

Working finding

US federal tax assumptions must be refreshed by taxable year and taxpayer. Section 163(j) generally limits deductible business interest through a statutory formula, while section 1061 generally requires more than three years for specified gains allocated through an applicable partnership interest to remain long-term.

The IRS's December 2025 section 163(j) overview and Public Law 119-21 show that adjusted-taxable-income computation and coordination rules changed on different effective dates. IRS section 1061 guidance preserves the general more-than-three-year rule but also points to reporting mechanics, exceptions and fact-specific application. Entity type, elections, partnership allocations, asset character and taxable year can change the answer.

Unit: US taxpayer, taxable year, partnership interest and statutory provision
Segment: Federal business-interest limitation and applicable-partnership-interest holding-period rules
Geography: United States federal tax only
Evidence horizon: Statute and IRS guidance reviewed through 2026-08-02
Excluded: No state, local or non-US tax conclusion; no tax advice or determination for a specific entity, partner, asset or transaction

Currentness rulePrimary-source review recorded 2026-08-02

Every model or article must identify the taxable year, taxpayer and relevant exception or election. Use the enacted statute as controlling authority and treat IRS FAQs as general explanation, not a substitute for fact-specific analysis.

Next trigger: A new statute, Treasury regulation, IRS release, court decision or modeled taxable year changes the applicable rule.

Primary-source register3 reviewed sources

Supports boundary

Questions and answers about the limitation on the deduction for business interest expense

Internal Revenue Service, Questions and answers about the limitation on the deduction for business interest expense (2025)

Open primary source

Official IRS section 163(j) overview and Public Law 119-21 update notes · retrieved 2026-08-02

Supports the current general formula and the need to date model assumptions after statutory changes, without determining a taxpayer's deduction.

IRS general guidance

The IRS page is official but expressly general; the statute, regulations and taxpayer facts control.

The IRS page is a general overview and states that the law controls. Entity type, taxable year, elections, exceptions, partnership allocations and other facts can change the result.

Recorded method fields

Data provider: Official IRS explanatory guidance
Unit: Taxpayer and taxable-year application of the business-interest limitation
Return basis: Not applicable; federal income-tax deduction rules
Value-created definition: Not applicable; this source defines the deductible-interest framework
Market / beta: Not applicable
Leverage: Interest deductibility is not equivalent to a permitted leverage level
Holding period: Tax-year specific; refreshed after Public Law 119-21 changes

Supports boundary

Public Law 119-21

Public Law 119-21 (2025)

Open primary source

Enrolled statutory text and provision-specific effective dates · retrieved 2026-08-02

Supports the boundary that section 163(j) assumptions must be tied to the modeled taxable year rather than copied from an older corpus source.

Controlling US statute

The enacted text is controlling primary authority, but separate amendments carry separate effective dates and interact with existing law.

The statute is controlling primary authority, but individual provisions have different effective dates and require fact-specific application. The Atlas supplies a currency boundary, not tax advice.

Recorded method fields

Data provider: Enacted statutory text
Unit: Statutory provision and effective date
Return basis: Not applicable
Value-created definition: Not applicable; this source changes federal tax rules
Market / beta: Not applicable
Leverage: Amends interest-deduction computation and coordination, not market leverage
Holding period: Effective dates differ by provision and taxable year

Supports boundary

Section 1061 Reporting Guidance FAQs

Internal Revenue Service, Section 1061 Reporting Guidance FAQs (undated; retrieved 2026)

Open primary source

Official IRS section 1061 reporting guidance and general holding-period rule · retrieved 2026-08-02

Supports the narrow more-than-three-year general rule while preserving exceptions, look-through questions and fact-specific application.

IRS general guidance

The guidance summarizes a general rule and reporting mechanics; the code, regulations, exceptions and transaction facts determine treatment.

The general more-than-three-year rule has exceptions, look-through and reporting mechanics. The page cannot determine treatment without the partnership, asset and taxpayer facts.

Recorded method fields

Data provider: Official IRS explanatory and reporting guidance
Unit: Partner, partnership interest, asset disposition and reporting item
Return basis: Not applicable; character and holding-period rules for certain gains
Value-created definition: Not applicable; this source defines section 1061 reporting treatment
Market / beta: Not applicable
Leverage: Not applicable
Holding period: Generally requires more than three years for affected gain to remain long-term, subject to statutory and regulatory detail

Primary-source refresh2026-08-02

The US federal section 163(j) and section 1061 boundaries were checked against enacted law and current IRS guidance. The record remains general, taxable-year-specific and expressly non-advisory.

Primary sources checked: yes

Working finding

US regulatory claims must identify the agency, rule or action, jurisdiction and effective date. The SEC's 2023 private-fund adviser rules were vacated effective June 5, 2024. Form PF has a separate adopted compliance timeline and a pending April 2026 SEC/CFTC proposal. The 2026 HSR size-of-transaction threshold is $133.9 million for relevant closings on or after February 17, 2026.

The SEC's technical amendments confirm that the specified 2023 private-fund adviser rules are not in force after the court vacatur. A separate joint SEC/CFTC action extended the compliance date for specified Form PF amendments to October 1, 2026, subject to later action; that date has now passed and a further extension has been reported but not yet verified against the primary SEC and CFTC records, so this record is marked refresh-required until that check is done. An April 2026 SEC/CFTC proposal would change separate Form PF requirements but remains pending and non-final. The FTC's annual HSR update sets the current size-of-transaction threshold, while the full statutory tests, aggregation rules and exemptions still determine filing obligations.

Unit: Named US agency action, rule, reporting deadline or transaction threshold
Segment: Private-fund advisers, Form PF filers, pending Form PF rulemaking and transactions potentially subject to HSR
Geography: United States federal regulation
Evidence horizon: Official SEC, CFTC and FTC records reviewed through 2026-08-02
Excluded: No legal advice, filing determination or claim that unrelated adviser, securities, antitrust or state obligations changed

Currentness rulePrimary-source review recorded 2026-08-02

Name the exact rule or threshold, agency, action type, affected population and effective or compliance date. Keep proposed, adopted, vacated and effective rules separate, and rerun transaction-specific legal analysis.

Next trigger: The SEC, CFTC, FTC, a court or Congress changes the named action, Form PF timeline or annual HSR threshold.

Primary-source register4 reviewed sources

Supports boundary

Technical Amendments Reflecting Vacatur of the Private Fund Advisers Rules

Securities and Exchange Commission, Technical amendments reflecting vacatur of the Private Fund Advisers rules (2024)

Open primary source

Official SEC technical amendments and vacatur status · retrieved 2026-08-02

Supports the current statement that the named 2023 rules were vacated effective June 5, 2024 and are not in force.

SEC official rule status

The record applies to the specified 2023 private-fund adviser rules and amendments, not every adviser obligation.

The SEC record states that the 2023 private-fund adviser rules were vacated effective June 5, 2024. It does not imply that all adviser obligations were vacated or that other rules are unchanged.

Recorded method fields

Data provider: Official SEC rulemaking and court-vacatur record
Unit: Named SEC rules and amendments
Return basis: Not applicable; regulatory status
Value-created definition: Not applicable; this source establishes whether specified rules remain in force
Market / beta: Not applicable
Leverage: Not applicable
Holding period: Status checked as of 2026-08-02; vacatur effective 2024-06-05

Qualifies

Form PF Compliance Date Extension

Securities and Exchange Commission and Commodity Futures Trading Commission, Form PF compliance-date extension (2025)

Open primary source

Official joint SEC/CFTC Form PF compliance-date action · retrieved 2026-08-02

Qualifies any broad 'SEC private-fund rules' statement by keeping the Form PF compliance date separate and subject to later agency action.

SEC official rule timeline

Form PF amendments have a separate rulemaking and compliance timeline from the vacated private-fund adviser rules.

The compliance-date action concerns specified Form PF amendments. Proposals, adopted rules, compliance dates and vacated rules must not be merged into one regulatory-status claim.

Recorded method fields

Data provider: Official joint SEC/CFTC rulemaking record
Unit: Form PF amendment and compliance date
Return basis: Not applicable; regulatory reporting
Value-created definition: Not applicable; this source establishes a reporting timeline
Market / beta: Not applicable
Leverage: Form PF may collect leverage information but this record does not establish a market leverage norm
Holding period: Compliance date checked as 2026-10-01, subject to later SEC/CFTC action

Qualifies

Form PF; Reporting Requirements for All Filers

Securities and Exchange Commission and Commodity Futures Trading Commission, Form PF; Reporting Requirements for All Filers (proposed 2026)

Open primary source

Official SEC/CFTC proposed-rule record and proposal status · retrieved 2026-08-02

Qualifies the Form PF timeline by identifying a pending April 2026 proposal without treating proposed relief as adopted or effective.

Pending SEC/CFTC Form PF proposal

The proposal is a separate, non-final rulemaking and cannot be merged with adopted Form PF requirements or compliance dates.

This is a pending proposal, not an adopted or effective rule. It cannot be treated as current reporting relief or as changing the separate October 1, 2026 compliance date unless the agencies take final action.

Recorded method fields

Data provider: Official SEC/CFTC proposed-rule record
Unit: Proposed amendments to Form PF filing and reporting obligations
Return basis: Not applicable; proposed regulatory reporting changes
Value-created definition: Not applicable; this source identifies pending rulemaking
Market / beta: Not applicable
Leverage: The proposal concerns Form PF reporting; it does not establish a market leverage norm
Holding period: Proposed 2026-04-20; public comments were due 2026-06-23; no final rule recorded as of 2026-08-02

Supports boundary

Current HSR Thresholds

Federal Trade Commission, Current HSR Thresholds (2026)

Open primary source

Official FTC current-threshold table and effective date · retrieved 2026-08-02

Supports the dated $133.9 million threshold for relevant closings on or after February 17, 2026, without making a transaction-specific filing determination.

FTC annual HSR threshold

The annually adjusted size-of-transaction threshold is only one part of HSR analysis; aggregation, other tests and exemptions still control.

The $133.9 million size-of-transaction threshold applies to relevant closings on or after February 17, 2026, but the full HSR tests, aggregation rules and exemptions still control. This is not legal advice.

Recorded method fields

Data provider: Official FTC annual threshold notice
Unit: Transaction and annual statutory threshold
Return basis: Not applicable; premerger-notification threshold
Value-created definition: Not applicable; this source states a filing threshold
Market / beta: Not applicable
Leverage: Not applicable
Holding period: Annual threshold effective 2026-02-17 and subject to the next annual adjustment

Primary-source refresh2026-08-02

Official SEC, CFTC and FTC records were checked separately by action and effective date. The vacated adviser rules, adopted Form PF compliance timeline, pending non-final proposal and annual HSR threshold are not collapsed into one broad regulatory claim.

Primary sources checked: yes

How the research Atlas differs from a deal Evidence Ledger

01 · System Boundary

Atlas and Ledger answer different evidence questions.

The Atlas does not replace the deal-specific Evidence Ledger, and it does not duplicate the canonical articles linked from each card.

Evidence Ledger

One deal claim, source to decision

Preserves the reported value, transformation, reviewer override and accepted value for a specific underwrite.

Question: What can this deal team rely on?

Evidence Atlas

One public claim across research sources

Preserves support, contradiction, qualification, shared data, scope, evidence horizon and the next currentness check.

Question: What can UpLevered responsibly teach?

Source methods, review dates and reuse limits

How the research Atlas differs from a deal Evidence Ledger.

See the conflict before repeating the claim. The Evidence Atlas organizes private-equity research by source stance, evidence stream, scope and currentness instead of converting a source count into false consensus.

Current read

Seven scoped records now separate what the evidence supports from what it cannot support. Persistence is conditional, benchmark methods are not interchangeable, no universal current leverage range or covenant-lite share is claimed, and US tax and regulatory statements carry explicit dates and jurisdiction limits.

Public claims: 7
Primary sources: 24
Source relationships: 25
Dated reviews: 7

04 · Method, Rights and Review Rule

Corpus provenance is a routing layer, not current authority.

Provenance

Normalized, claim-level joins

Claims, sources, claim-source stance and dependency, and verification events remain separate records. Public cards show direct source links and the method fields that determine comparability.

Currentness

No badge without a check

A dated review records what was checked. It does not make an old empirical sample current, turn first-party methodology into an independent performance test, or create a universal numeric answer.

Rights

Metadata and paraphrase only

This page republishes no source text, figures, tables or vendor data. It uses bibliographic metadata, direct public links and UpLevered-authored method summaries.