Do self-funded companies last longer than VC-backed ones? What the public data can and cannot say

A founder with a working business and a little cash in the bank asks a reasonable question: if I raise venture capital, is my company more or less likely to still exist in seven years?

There is a free, public dataset that looks built to answer it, and we ran the comparison on it. The data cannot tell you, and the reasons it cannot are more useful than a number would have been. Somebody also published the same qualitative result in 2021, on the same survey, and we found their paper only after doing the work.

What "survival" means in this kind of study

Three definitions, because everything below depends on them.

Survival is not "the business is doing well". It means the company has not yet been recorded as closed, merged or sold. A company that limps along for seven years survives. A company acquired in year three does not.

Hazard is the risk of that recorded ending in a given year, among the companies still going at the start of it. A hazard ratio compares two groups: 0.71 means about 29% lower per-year risk, 1.00 means the same risk, 1.28 means about 28% higher. The range in brackets after each ratio is a 95% interval: the band of values the data cannot rule out. When that band includes 1.00, the honest reading is "we cannot tell", which is not the same as "there is no difference".

The sample is the Kauffman Firm Survey: 4,928 US businesses that all started in 2004, some of them by buying an existing business or a franchise, surveyed every year through 2011. The public microdata file downloads without registration. It is the best free panel for this question, and it is also 15 years old.

Almost nobody takes venture capital

Classify each firm by the form of financing it took in its first year, assigning each company to exactly one group:

• Venture capital: 40 firms, 0.8%

• Individuals outside the family (angels, friends): 146 firms, 3.0%

• Company, government or other outside equity: 80 firms, 1.6%

• No outside equity recorded, business debt: 1,082 firms, 22.0%

• No outside equity recorded, other debt only: 1,435 firms, 29.1%

• No outside equity and no debt recorded: 2,133 firms, 43.3%

• Unclassifiable (missing debt data): 12 firms, 0.2%

Two published findings put those groups in context. They come from different surveys, so neither describes these firms directly.

Robb and Robinson, looking at first-year capital structure, found that new firms lean on outside debt, not on friends and family and not on outside equity: averaged over all firms including the zeros, outside debt came to $47,847 against $15,935 of outside equity, and seven times as many firms reported outside debt as outside equity. Among the 205 firms that did raise outside equity, the average was over $350,000, roughly twice the total capital of the average firm in the survey. Outside equity is rare and, where it happens, large.

Hurst and Pugsley asked people who were about to start a business what they wanted. Offered the choice between "I want this new business to be as large as possible" and "I want a size I can manage myself or with a few key employees", nearly three quarters picked the second, in every subsample they looked at.

So "bootstrapped versus VC-funded" is not a comparison of two halves of the startup world. It is a comparison between a large, heterogeneous majority and 40 companies. Separate evidence on people preparing to start a business, in a different survey, also suggests that maximising scale is not the typical founder's objective; it does not establish what the firms in these groups wanted.

What the numbers say

Here are the hazard ratios for the outcome fixed before any group results were looked at, which merged closure with sale or merger into one event. (That ordering has to be taken on trust: the plan sits in a private repository. Implementation errors were also corrected after the first results, and the decision to report weighted and unweighted side by side, rather than one as primary, came afterwards.) Survey weights exist to make the sample stand in for the population of 2004 startups, and applying them changes the answer.

• VC (40) vs no outside equity recorded (4,650): weighted 0.71 (0.31 to 1.29); unweighted 1.28 (0.81 to 2.03)

• Any outside equity (266) vs no outside equity recorded: weighted 0.99 (0.78 to 1.23); unweighted 1.04 (0.87 to 1.25)

• No equity and no debt (2,133) vs has debt (2,517): weighted 0.93 (0.84 to 1.02); unweighted 0.97 (0.89 to 1.06)

The VC comparison flips sign depending on whether the weights are applied, and both intervals are wide enough to contain "30% safer" and "twice as risky" at once. That is not a subtle finding, it is a sample size problem: the weighted estimate has an effective sample size of about 21 firms, with roughly three quarters of the weight sitting on 13 of the 40. There are 8 closures and 11 sales or mergers in the group in total.

One structural detail matters for the first two comparisons. The questionnaire never asks sole proprietors about outside equity at all, and treating that gap as "no outside equity recorded" puts 1,634 sole proprietors into the comparison group. So "outside equity versus none" is also, in large part, "incorporated versus not".

We also split the merged outcome, separating closure from sale or merger. This was decided after seeing the results, so treat it as a lead, not a finding. The split suggests the two movements run in opposite directions and cancel: outside-equity firms were sold or merged more often (unadjusted, 33 such events in 266 firms), while firms with no equity and no debt were sold or merged less often than firms with debt (hazard ratio 0.60, bootstrap interval 0.44 to 0.80, 87 events). The VC-only split rests on 11 sale or merger events, roughly five effective after weighting, and is too unstable to report as a result, so the numbers are not given here. And the survey cannot tell a triumphant exit from a rescue sale, so even a solid estimate here would not mean what a founder would want it to mean.

Someone already published this

Keogh and Johnson (2021), in the Journal of Entrepreneurship, Management and Innovation, open access under CC BY, ran a Cox survival model on the same survey. Their financing variables include a venture capital indicator, and they state the baseline explicitly: "We treat the excluded category as implicitly self-funded." Their venture capital hazard ratio is 1.0343, standard error 0.4776, z of 0.07. In their words, after listing the financing sources that did move survival, "Other financing choices show no statistically significant effects." They also flag the same wall we hit: "there were relatively few equity-financed or venture capital-financed firms", so their results "may be unintentionally biased".

That is the same qualitative answer, published first, and it must be credited. It is not, however, the same study, and it would be wrong to present the two as a replication:

• Groups. Keogh and Johnson (2021): overlapping binary indicators, with a residual omitted category. This check: six mutually exclusive groups by first-year form.

• Mergers and acquisitions. Keogh and Johnson: firms that merged or were acquired removed from the sample entirely. This check: kept, and separated from closure.

• Not returning the survey. Keogh and Johnson: counted as the end of survival. This check: observation stopped, not assumed to have ended (censored).

• Model. Keogh and Johnson: Cox proportional hazard. This check: discrete-time complementary log-log.

• Survey weights. Keogh and Johnson: not mentioned in the paper. This check: reported weighted and unweighted.

• Firms. Keogh and Johnson: 3,768. This check: 4,916 classified of 4,928.

• VC firms in the first year. Keogh and Johnson: 20 (2004). This check: 40.

Different group construction, different handling of acquisition and attrition, different model, and twice the VC count. A hazard ratio of 1.03 and one of 1.28 are not the same estimate either. What they share is the conclusion, and the conclusion is theirs first.

"Survives longer" would still not mean "self-funding causes survival"

Suppose the data had come back clean and bootstrapped firms had lasted longer. That still would not be the sentence a founder wants, for a reason that has nothing to do with sample size: investors choose.

Puri and Zarutskie, tracking US firms from birth over two decades in Census data, found that venture capitalists invest in young firms where "potential for large scale" is an important criterion, and that VC-financed firms are at remarkably larger scale than matched non-VC firms at every point in the life cycle, successes and failures alike. Any raw survival gap mixes the effect of the money with the effect of being the kind of company that gets offered money, and the second is large.

Where researchers have engineered that selection away, they end up estimating something narrower than "should I raise". Kerr, Lerner and Schoar used a discontinuity in two US angel groups, where a small change in how much collective interest a pitch attracted produced a discrete jump in the chance of funding. Around that threshold, funding still raised survival and growth, though not access to follow-on money. That is a local effect of angel groups on companies that chose to pitch angel groups: a real finding, and not an answer about venture capital in general.

One further piece of evidence belongs here, and it is the one claim in this article we could not verify at source. The Japanese draft this article comes from reports, from the published abstract, that the meta-analysis by Rosenbusch, Brinckmann and Mueller (2013) found the small positive association between VC funding and firm performance disappears once industry selection is controlled. The article is paywalled and its abstract is not available from Crossref, Semantic Scholar or any open repository we could reach, so that sentence rests on a secondary source: the draft's reading of the abstract. Do not lean on it.

Finally, survival is not the outcome most founders are actually asking about. Hall and Woodward, using data covering close to the universe of US venture-funded companies from 1987 to 2008, estimate that the typical venture-backed entrepreneur received an average of $5.8 million in exit cash, and that "almost three-quarters of entrepreneurs receive nothing at exit". Company growth, investor return, founder payout and founder satisfaction are four different measures, and a result about one of them says nothing about the others.

What would change the picture, and what is still unknown

• A newer representative cohort. The survey follows companies founded in 2004 and stops in 2011. Nothing here speaks to 2026 fundraising norms.

• An outside-equity question asked of everyone. Until the sole-proprietor skip is fixed or modelled, the "no outside equity" group is partly a legal-form group, and quantifying how much that contaminates published estimates would be a genuine contribution.

• Exits labelled by type. Without telling a rescue sale from a successful acquisition, splitting closure from sale buys less than it looks like.

• Founder wellbeing by funding type. We found no peer-reviewed study comparing self-funded against VC-backed founders on satisfaction or wellbeing. That search was not exhaustive, so read it as an unfilled cell rather than a proven absence.

• A sample with enough VC firms. 40 in a 4,928-firm cohort is what a representative survey of all new businesses gives you. Answering the question properly needs a sample drawn where VC is common, which stops it being a statement about startups in general.

How this was checked

The Japanese draft behind this article cited 16 references and had read none of them in full, and that was not a paywall problem: six were cited through their NBER abstract pages, and all six of those PDFs download freely from the same site.

For this article we opened the source for every claim but one. Keogh and Johnson, Robb and Robinson, Hurst and Pugsley, Puri and Zarutskie, Kerr, Lerner and Schoar, and Hall and Woodward were read as full PDFs, and the quoted phrases and numbers above come from their text, not from abstracts. The Rosenbusch meta-analysis is the exception, flagged above where it is used.

What this piece can claim for itself is the shape of the question. "Do bootstrapped companies last longer" cannot be answered as posed: fewer than 1% of new US companies take venture capital, the groups differ in size, industry and legal form from day one, and how you define the ending almost determines what you can claim. The published answer, Keogh and Johnson's, is that there is no detectable survival difference. Our reading of the same data agrees, and agrees too weakly to count as support.

References

Hall, R. E., and Woodward, S. E. (2010). The burden of the nondiversifiable risk of entrepreneurship. American Economic Review, 100(3), 1163-1194. https://doi.org/10.1257/aer.100.3.1163

Hurst, E., and Pugsley, B. W. (2011). What do small businesses do? NBER Working Paper 17041. https://www.nber.org/papers/w17041

Keogh, D., and Johnson, D. K. N. (2021). Survival of the funded: Econometric analysis of startup longevity and success. Journal of Entrepreneurship, Management and Innovation, 17(4), 29-49. CC BY. https://doi.org/10.7341/20211742

Kerr, W. R., Lerner, J., and Schoar, A. (2010). The consequences of entrepreneurial finance: A regression discontinuity analysis. NBER Working Paper 15831. Published as Kerr, Lerner and Schoar (2014), Review of Financial Studies, 27(1), 20-55. https://www.nber.org/papers/w15831

Puri, M., and Zarutskie, R. (2008). On the life cycle dynamics of venture-capital- and non-venture-capital-financed firms. NBER Working Paper 14250. Published as Puri and Zarutskie (2012), Journal of Finance, 67(6), 2247-2293. https://www.nber.org/papers/w14250

Robb, A. M., and Robinson, D. T. (2010). The capital structure decisions of new firms. NBER Working Paper 16272. Published as Robb and Robinson (2014), Review of Financial Studies, 27(1), 153-179. https://www.nber.org/papers/w16272

Rosenbusch, N., Brinckmann, J., and Mueller, V. (2013). Does acquiring venture capital pay off for the funded firms? A meta-analysis on the relationship between venture capital investment and funded firm financial performance. Journal of Business Venturing, 28(3), 335-353. Not read; see the note above. https://doi.org/10.1016/j.jbusvent.2012.04.002

Data: Ewing Marion Kauffman Foundation, Kauffman Firm Survey public-use microdata, 2004 to 2011, logically imputed version. https://www.kauffman.org/entrepreneurship/research/kauffman-firm-survey/