# Research sources and limitations

**Retrieval date for all links unless noted:** 2026-09-05.  
**Method:** literature discovery began with OpenResearch/`orx` searches across OpenAlex and alphaXiv, followed by focused `orx paper` reads for load-bearing studies. Native web retrieval was used for primary datasets, regulator filings, issuer records, and sources not returned cleanly by discovery. The literature retrieval loop was not delegated.

## Preserved project inputs

The supplied article was preserved before analysis as HTML and plain text, with its image files, extracted-link index, figure index, and provenance record. The recorded retrieval time is `2026-09-05T03:28:45.866387+00:00`; the preserved HTML SHA-256 is `f835a287d5e566043649c2ef2d894d992e462fd96132a4e46ab6c7d770d4142a`. Those input snapshots were not edited.

The reproducibility directory separately preserves downloaded ZIP/CSV factor data and PDF reports. `repro/inputs/MANIFEST.md` records direct URLs and SHA-256 hashes. This audit does not redistribute proprietary PitchBook or Cambridge Associates microdata.

## Article and directly cited data

### Damodaran article

**Aswath Damodaran (2026), [“The Scaling and Profitability Trade off: Venture Capital's Weakest Link!”](https://aswathdamodaran.blogspot.com/2026/09/the-scaling-and-profitability-trade-off.html).**

Primary object of the audit. Strength: an explicit operating framework connecting scale to market size, unit economics, capital intensity, competition, and fixed-cost leverage. Limitations: figures often omit transformation code or precise denominator; descriptive evidence is repeatedly extended into causal claims; company, investment, fund, and LP outcomes are sometimes conflated.

### Khosla statement

**Vinod Khosla social-media screenshot embedded in the article.**

The supplied image preserves the exact statement, but not the post to which it replied, the thread, timestamp, edits, or follow-up. A [secondary contemporaneous report](https://economictimes.indiatimes.com/markets/stocks/news/aswath-damodaran-disagrees-with-vinod-khosla-on-the-scaling-vs-profitability-debate-heres-why/articleshow/133730584.cms) helps date the exchange but cannot reconstruct missing context. The audit therefore treats wording as observed and intent/context as unknown.

### NVCA / PitchBook

**National Venture Capital Association, [2026 NVCA Yearbook](https://nvca.org/2026-nvca-yearbook/) and [release](https://nvca.org/press_releases/nvca-releases-2026-yearbook-charts-a-venture-industry-in-transition/), 2026-04-13.**

Primary public source for U.S. VC activity through 2025. It reports $320 billion across 15,352 deals, AI at 65.4% of value, $67 billion raised across 585 funds, and a fall in firm count to 2,984. Limitations: produced by the industry association with PitchBook; some methodological detail and the supplemental XLS are not public. It cannot support an annual 2026 endpoint. The article’s “NCVA” is a typographical error.

### Cambridge Associates

**Cambridge Associates, [“US PE/VC Benchmark Commentary: Calendar Year 2025”](https://www.cambridgeassociates.com/insight/us-pe-vc-benchmark-commentary-calendar-year-2025/), 2026-07-31.**

Primary benchmark page and methodology. U.S. VC returned 21.1% in calendar 2025; the database comprised 2,816 funds formed 1981–2024 with $764 billion in value. Private indexes are capitalization-weighted pooled horizon IRRs net of fees, expenses, and carry; public comparisons use mPME. Limitations: proprietary institutional-quality sample; underlying cash flows and marks are not public; pooled performance is not a median and relies partly on reported NAVs.

### CF Private Equity / PitchBook concentration

**Aaron Miller, [“AI is Steepening the Power Law in Venture Capital”](https://www.commonfund.org/cf-private-equity/ai-is-steepening-the-power-law-in-venture-capital?hs_amp=true), 2026-07-13.**

Exact source for the article’s top-1% chart. CF analyzes U.S. VC-backed exits of at least $10,000 in PitchBook from 2005-01-01 through 2026-06-18. Since 2023 the top 1% represent 80% of total exit value, or 45% excluding the $1.77 trillion SpaceX IPO. Limitation: proprietary microdata prevent replication; exit value is not investor return and the last window is partial and event-dominated.

### Kenneth French factor data

**Kenneth R. French, [Data Library](https://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_Library.html) and [short-term reversal construction](https://mba.tuck.dartmouth.edu/pages/faculty/ken.french/Data_Library/det_st_rev_factor.html).**

Primary public factor/portfolio data. Short-term reversal is the average of low prior-month return portfolios minus high prior-month return portfolios, using six value-weighted size/return intersections. Momentum uses prior months 2–12; long-term reversal and HML are distinct. The downloaded releases were created with the 2026-07 CRSP database and are preserved. Limitation: the article does not disclose its aggregation code, and its “bottom 2 sextiles” chart wording does not map transparently to the canonical factor file; exact plotted values were not reproducible under two reasonable aggregation conventions.

### Jay Ritter IPO data

**Jay R. Ritter, [IPO Data](https://site.warrington.ufl.edu/ritter/ipo-data/), updated 2026-03-09; [age table](https://site.warrington.ufl.edu/ritter/files/IPOs-Age-of-Companies-Going-Public.pdf); [sales and earnings table](https://site.warrington.ufl.edu/ritter/files/IPOs-Sales.pdf).**

Primary public summary tables through 2025. The U.S. operating-company sample excludes small best-efforts deals, unit offers, ADRs, closed-end funds, natural-resource limited partnerships, SPACs, REITs, banks/S&Ls, and firms without timely CRSP coverage. Age conventions account for predecessors/rollups; sales are trailing and inflation adjusted; profitability uses trailing or, when necessary, latest-fiscal-year earnings. Limitations: public PDFs are summary tables, not all firm-level inputs; article-specific averages are not accompanied by code; annual medians/percentages should not be confused with pooled firm statistics. The chart labeled tech appears to mix or use all-IPO counts and extends through 2026, beyond the public series.

**Jay R. Ritter (2008), [“A Review of IPO Activity, Pricing, and Allocations”](https://site.warrington.ufl.edu/ritter/files/2016/01/A-Review-of-IPO-Activity-Pricing-and-Allocations-2002-08.pdf).**

Supports the historical statement that 19% of 1980s IPOs had negative trailing earnings, implying roughly 81% profitable under that sample. Limitation: vintage paper and conventions may differ from later updated tables.

## Empirical literature

### VC performance and persistence

**Steven N. Kaplan and Antoinette Schoar (2005), [“Private Equity Performance: Returns, Persistence, and Capital Flows”](https://doi.org/10.1111/j.1540-6261.2005.00780.x), *Journal of Finance*.**

Finds average net fund returns approximately equal to the S&P 500 in the Venture Economics sample, substantial heterogeneity, persistence across successive funds, larger follow-on funds after good performance, and worse follow-on prospects for boom-time entrants. Limitation: old sample and known commercial-database coverage issues; combines buyout and VC analyses and predates modern mega-funds.

**Robert S. Harris, Tim Jenkinson, Steven N. Kaplan, and Rüdiger Stucke (2023), [“Has Persistence Persisted in Private Equity?”](https://doi.org/10.1016/j.jcorpfin.2022.102326), *Journal of Corporate Finance*.**

Uses Burgiss institutional-investor cash flows and carefully distinguishes final prior-fund performance from information actually available at the next fundraising. Little top-quartile persistence remains for post-2000 buyouts, but VC persistence survives. Limitation: access and manager continuation still shape the observed sample; persistence is not proof of a single mechanism or guaranteed future alpha.

**Ramana Nanda, Sampsa Samila, and Olav Sorenson (2020), [“The Persistent Effect of Initial Success: Evidence from Venture Capital”](https://doi.org/10.1016/j.jfineco.2020.01.004), *Journal of Financial Economics*.**

Investment-level evidence: an additional IPO among a VC’s first ten investments predicts up to an 8% higher IPO rate later, but the timing/location component does not persist; early success improves access to later rounds and larger syndicates. This provides an access/reputation alternative to pure selection skill. Limitation: IPO occurrence is not a complete cash return measure.

**Arthur Korteweg and Stefan Nagel (2016), [“Risk-Adjusting the Returns to Venture Capital”](https://doi.org/10.1111/jofi.12390), *Journal of Finance*.**

Develops a generalized PME/SDF approach for infrequent, skewed, cross-dependent cash flows. Start-up investments retain positive abnormal returns, while fund abnormal returns are near zero and can become negative under an index-put-based SDF. Limitation: conclusions depend on the available payoff sample and admissible risk model; it does not describe current 2026 performance.

**Robert S. Harris, Tim Jenkinson, and Steven N. Kaplan (2014), [“Private Equity Performance: What Do We Know?”](https://doi.org/10.1111/jofi.12154), *Journal of Finance*.**

Uses cash flows from nearly 1,400 funds sourced through more than 200 institutional investors and compares Burgiss with Venture Economics, Preqin, and Cambridge. It documents material data-source differences and vintage sensitivity; VC outperformed in the 1990s and underperformed in the 2000s. Limitation: historical cohort and institutional-investor selection.

### VC decisions, contracts, and governance

**Paul Gompers, Will Gornall, Steven N. Kaplan, and Ilya A. Strebulaev (2020), [“How Do Venture Capitalists Make Decisions?”](https://doi.org/10.1016/j.jfineco.2019.06.011), *Journal of Financial Economics*.**

Survey of 885 institutional VCs at 681 firms. Respondents rate management most important in selection, deal selection most important to value creation, and report substantial post-investment guidance, recruiting, and connections. Limitations: self-report, respondent selection, 2015–2016 practices, and no causal estimate of value added.

**Steven N. Kaplan and Per Strömberg (2004), [“Characteristics, Contracts, and Actions: Evidence from Venture Capitalist Analyses”](https://www.nber.org/papers/w8764), *Journal of Finance*.**

Studies 67 investments by 11 VC firms. Business, management, and external risks are linked to contractual allocation of cash-flow and control rights and to expected monitoring. Limitation: small, selected, historical sample; contract responses do not guarantee effective governance.

**Josh Lerner and Ramana Nanda (2020), [“Venture Capital’s Role in Financing Innovation”](https://doi.org/10.1257/jep.34.3.237), *Journal of Economic Perspectives*.**

Balanced review of VC strengths and limitations. Highlights the narrow class of innovations fitting VC, concentration among capital allocators, and relaxation of historically intensive governance. Limitation: synthesis rather than a single causal design; social-welfare effects remain uncertain.

### Private capital, crossover investors, and IPO timing

**Sungjoung Kwon, Michelle Lowry, and Yiming Qian (2020), [“Mutual Fund Investments in Private Firms”](https://doi.org/10.1016/j.jfineco.2019.10.003), *Journal of Financial Economics*.**

Documents a large rise in private-company investment by 16 mutual-fund families and tests demand- and supply-side explanations. Mutual-fund capital is primarily incremental primary financing and IV estimates imply firms remain private about 2.8 years longer. Limitations: only 16 families, data largely through 2016, instrument validity is an assumption, and investment returns face private-value measurement problems. An [author summary](https://corpgov.law.harvard.edu/2017/04/18/mutual-fund-investments-in-private-firms/) provides sample counts and estimates.

**Sergey Chernenko, Josh Lerner, Yao Zeng, and Francesca Cornelli (2021), [“Mutual Funds as Venture Capitalists? Evidence from Unicorns”](https://doi.org/10.1093/rfs/hhaa100), *Review of Financial Studies*.**

Contract-level evidence shows larger and more stable mutual funds are more likely to invest; compared with VCs, they have weaker cash-flow rights, less board involvement, and stronger redemption rights consistent with liquidity needs. Limitation: unicorn-focused contractual sample; does not establish whether crossover capital improves or harms company performance.

**Michael Ewens and Joan Farre-Mensa (2020), [“The Deregulation of the Private Equity Markets and the Decline in IPOs”](https://doi.org/10.1093/rfs/hhaa053), *Review of Financial Studies*.**

Uses the 1996 NSMIA deregulation to identify a rise in private capital supply, larger late-stage firms, later IPOs, and stronger founder bargaining power. Explicitly argues the new equilibrium is not simply an IPO-market failure and cautions against treating NSMIA as the only cause. Limitation: does not estimate whether delayed listing causes weaker economics, governance failures, or lower welfare.

### Passive ownership and price formation

**Ian Appel, Todd Gormley, and Donald Keim (2016), [“Passive Investors, Not Passive Owners”](https://doi.org/10.1016/j.jfineco.2016.03.003), *Journal of Financial Economics*.**

Russell 1000/2000 assignment variation links passive mutual-fund ownership to more independent directors, fewer takeover defenses, more equal voting rights, and improved longer-term performance. Limitation: local index-assignment design and governance outcomes; not a test of VC behavior or every passive-fund effect.

**Jeffrey Coles, Davidson Heath, and Matthew Ringgenberg (2020 working paper), [“On Index Investing” / index-inclusion evidence](https://www.nber.org/papers/w27593).**

Reports lower stock-price informativeness around index inclusion and no detected competition effect, offering evidence for a price-demand/information channel. Limitation: working-paper status, contested measurement, and no direct link to the French short-term reversal series or private-company scaling.

This pair is intentionally included to show the live disagreement: passive ownership can affect both price formation and governance, and one effect should not be used as a universal characterization.

## Issuer and regulatory primary sources

### SpaceX

**Space Exploration Technologies Corp., [Form 10-Q for 2026-06-30](https://www.sec.gov/Archives/edgar/data/1181412/000162828026052535/spcx-20260630.htm).**

Confirms a June 2026 IPO of 638.9 million Class A shares at $135, including the overallotment, and net proceeds of $85.675 billion. Limitation: a post-offering company filing, not an independent valuation assessment.

**SpaceX, [SEC-filed U.K. offering disclosure](https://www.sec.gov/Archives/edgar/data/1181412/000162828026040874/spacexukfwp.htm), June 2026.**

Gives pro forma 7,380,196,910 Class A and 5,695,668,265 Class B shares and the $135 offer price. Their product is approximately $1.765 trillion. Limitation: offer-price implied equity value differs from proceeds, enterprise value, and trading market capitalization.

**Nasdaq, [IPO opening report](https://www.nasdaq.com/newsroom/spacex-ipo-rocket-company-launches-historic-ipo), 2026-06-12; [overallotment/first-day report](https://www.nasdaq.com/newsroom/spacex-makes-history-raising-85-billion-through-nasdaq-listing), 2026-06-16.**

Exchange confirmation of $135 offer, $150 opening trade, approximately $1.77 trillion implied offer valuation, $85.7 billion total raise after overallotment, and approximately $2.1 trillion first-day market cap. Limitation: exchange promotional newsroom, though the core event data agree with SEC filings.

### Amazon

**Amazon, [1997 shareholder letter reproduced in SEC filing](https://www.sec.gov/Archives/edgar/data/1018724/000119312517120198/d373368dex991.htm).**

Contemporaneous articulation of scale-first but disciplined reinvestment: large market, repeat orders, analytical return tests, program termination, and long-run cash-flow focus. Limitation: management narrative and a selected success; later outcomes cannot prove the strategy was ex ante optimal for other firms.

### Google and Facebook

**Google, [2004 S-1](https://www.sec.gov/Archives/edgar/data/1288776/000119312504073639/ds1.htm).**

Reports 2003 net revenue of $961.874 million and net income of $105.648 million, confirming profitability before IPO. Limitation: one exceptional advertising model; historical accounting and stock-compensation treatment require care.

**Facebook, [2012 Form 10-K](https://www.sec.gov/Archives/edgar/data/1326801/000132680113000003/fb-12312012x10k.htm).**

Reports 2011 revenue of $3.711 billion and operating income of $1.756 billion, confirming profitability before the 2012 IPO. Limitation: selected winner and platform-specific economics.

### WeWork

**The We Company, [2019 draft-registration filing](https://www.sec.gov/Archives/edgar/data/1533523/000162827919000272/filename1.htm).**

Defines committed revenue backlog and contribution margin, and describes the long-lease/flexible-membership model. The SEC’s [comment correspondence](https://www.sec.gov/files/foia-requests-for-the-we-company.pdf) questioned cancellation, term, and revenue-backlog disclosure. Limitation: draft filing was withdrawn and metrics were management-defined.

**WeWork, [2022 Form 10-K](https://www.sec.gov/Archives/edgar/data/1813756/000181375623000016/we-20221231.htm).**

Provides a clean duration comparison: about 15-year initial lease terms, 19-month average membership agreements, $27.9 billion future undiscounted lease costs, and $2.5 billion committed sales. Limitation: later snapshot after restructuring; it illustrates model exposure but cannot isolate the causal effect of earlier growth.

### Ferrari

**Ferrari N.V., [FY2025 results filed with the SEC](https://www.sec.gov/Archives/edgar/data/1648416/000164841626000018/fnvfy2025results.htm), 2026-02-10.**

Reports 13,640 shipments, €7.146 billion net revenue, 29.5% EBIT margin, and €1.538 billion industrial free cash flow, with explicit “carefully-managed volume” language. Limitation: industrial FCF is a company-defined non-GAAP measure; a same-date mass-automaker market-cap comparison would still be descriptive, not causal.

## Measurement references

**U.S. Securities and Exchange Commission, [Beginner’s Guide to Financial Statements](https://www.sec.gov/about/reports-publications/beginners-guide-financial-statements).**

Authoritative basic distinction among operating, investing, and financing cash flows and explanation of noncash/working-capital reconciliation. Limitation: educational overview, not a comprehensive accounting standard or a definition of free cash flow.

**Financial Accounting Standards Board, [Summary of Statement No. 2: Accounting for Research and Development Costs](https://fasb.org/page/PageContent?bcpath+=+tff&pageId=%2Freference-library%2Fsuperseded-standards%2Fsummary-of-statement-no-2.html).**

States the U.S. accounting rule that R&D costs are charged to expense when incurred. Limitation: superseded-statement summary; current Codification contains exceptions and detailed scope. Economic capitalization is an analytical adjustment, not GAAP.

## Reproducibility and known gaps

The saved scripts use Python’s standard library only:

- `repro/reconstruct_french_factors.py` parses the preserved French annual and monthly tables and writes both mean annual compounded-return and annualized mean-monthly summaries by decade.
- `repro/verify_claim_math.py` computes SpaceX’s offer-price implied equity value and gives counterexamples showing that skewness does not fix the fraction above the mean.
- `repro/french-factor-decades.csv` is generated output.

Known limitations:

1. PitchBook exit microdata and the NVCA supplemental XLS were not available publicly; no exact recreation of the CF concentration chart or article’s NVCA chart was attempted.
2. Cambridge fund cash flows/NAVs are proprietary; the audit relies on published methodology and tables.
3. Ritter’s public materials are summary PDFs. The article’s custom period transformations, tech/all-IPO classification, and 2026 additions are undisclosed.
4. The original Khosla thread was not recovered; only the embedded statement is treated as observed.
5. The French factor inputs are current releases and historical values may be revised with CRSP updates. The article’s exact aggregation code is absent.
6. Company examples illustrate mechanisms but are selected and cannot estimate average treatment effects.
7. No paid databases, compute, transactions, external messages, or publication were used.