Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 116 of the 122 references it cites were found in a registry or an index; none of the 19 sources its claims cite is checked for support yet.

Smart and Illicit: Who Becomes an Entrepreneur and Do They Earn More?*

Open access
Levine, Ross, Rubinstein, Yona · 2017 · The Quarterly Journal of Economics

Abstract

We disaggregate the self-employed into incorporated and unincorporated to distinguish between “entrepreneurs” and other business owners. We show that the incorporated self-employed and their businesses engage in activities that demand comparatively strong nonroutine cognitive abilities, while the unincorporated and their firms perform tasks demanding relatively strong manual skills. People who become incorporated business owners tend to be more educated and—as teenagers—score higher on learning aptitude tests, exhibit greater self-esteem, and engage in more illicit activities than others. The combination of “smart” and “illicit” tendencies as youths accounts for both entry into entrepreneurship and the comparative earnings of entrepreneurs. Individuals tend to experience a material increase in earnings when becoming entrepreneurs, and this increase occurs at each decile of the distribution.

Overview

CoChat Summary

TL;DR

Disaggregating the self-employed into incorporated and unincorporated resolves the puzzle of whether entrepreneurship pays: incorporated self-employed—a better proxy for entrepreneurs—earn more per hour and work more hours than salaried workers, while unincorporated self-employed earn less. Entrepreneurs are distinguished by a pre-labor-market mixture of high cognitive ability and aggressive/illicit risk-taking, and they experience an earnings boost when they incorporate.

Key findings

  • Incorporated self-employed earn more and work more: CPS median hourly earnings are about 80% greater than the unincorporated and 33% more than salaried; NLSY median incorporated earns 41% more per hour and works 27% more hours than the median salaried, while median unincorporated earns about 21% less per hour.
  • After controls, incorporated average residual hourly earnings are 48% greater than salaried counterparts (abstract/intro; median residual reported as 28% greater there, though later CPS/NLSY tables show smaller median gaps, e.g., 3.8% and 6.4%). Switching from salaried to incorporated is associated with an 18% average residual hourly earnings boost (6.4% at median in NLSY Table 8); unincorporated average residual earnings are about 3% lower than salaried.
  • Sorting traits: incorporated are more educated, from higher-income two-parent families with better-educated mothers; as teens they score higher on AFQT and self-esteem, more often aspire to be managers, and engage in more illicit/aggressive/risk-taking activities. High AFQT plus high Illicit is associated with almost 60% greater probability of incorporation (Table 5).
  • Fat tails/option value: incorporated earnings distribution has much fatter tails; 90th-percentile incorporated enjoy a 30% boost relative to their own successful salaried earnings, while 10th-percentile suffer a 10% drop (Figure 2a; introduction cites 70% at the 90th). 65% of incorporated call themselves entrepreneurs vs 44% unincorporated and 17% salaried; incorporated are more than twice as likely to have contributed to a patent application.

Methods

  • CPS March Annual Demographic Survey files, work years 1994–2010; prime-age 25–55; excludes group quarters, missing demographics, agriculture/military, and allocated earnings; earnings regressions use white non-Hispanic males. Used for summary statistics, transition matrices, and OLS/quantile earnings regressions.
  • NLSY79, 12,686 individuals aged 15–22 in 1979, surveyed 1979–2010; uses cross-sectional, supplemental, and military samples. Pre-labor-market measures include AFQT, Rosenberg Self-Esteem, Rotter Locus of Control, managerial/professional aspirations, Illicit Activity Index, parental education, family income, and two-parent family at 14; 2010 entrepreneur and patent questions.
  • Analysis: multinomial logit for selection into salaried/unincorporated/incorporated; OLS and quantile regressions with individual fixed effects and change-in-earnings specifications; IV using lagged employment type; three-sector Roy model to frame sorting and comparative earnings.

Limitations

  • Paper states it does not evaluate the average impact of incorporation on earnings and does not assess risk-adjusted investment returns to starting a business, as it does not account for the full array of costs and risks.
  • Main analyses focus on white, full-time, full-year males (though paper says results are robust to broader samples); NLSY79 is younger and smaller than CPS, and incorporated self-employed are older so a smaller percentage of the NLSY sample is incorporated.
  • Evident from text: incorporation is an imperfect proxy for entrepreneurship; employment type and earnings are self-reported; the Illicit Activity Index may proxy risk aversion, though the paper reports no strong association with a risk-aversion indicator after controls. The observational design means estimates are sorting/associations, not causal effects of incorporation.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1093/qje/qjw044

Cites

116 of 116 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Verification

Claims the paper makes

0 Valid · 5 Not valid · 13 Not checked · 210 Own results and interpretations

151 more claims

Create a free account to see all 228 claims and to verify the 10 references still open.

See all claims — free account
  • The quote only partly states this claim. 4
  • It may state two separate claims as one. 43
  • Probably not a substantial claim. 28
  • Borderline: one of the tests for a substantial claim is unclear. 1

Create a free account to see these claims.

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

4 of 4

Citations

Cites

Research works printed but not linked: 1

  1. Kushner, 2012, “Machine Politics: The Man Who Started the Hacker Wars,”, New Yorker

Cited by

See all 793

Metadata

Authors
Levine, R., & Rubinstein, Y.
Affiliations
University of California, Berkeley (US)
Venue
The Quarterly Journal of Economics
Published
2016-11-15
Volume
132
Issue
2
Pages
963-1018
ISSN
  • 0033-5533 (ISSN-L)
  • 1531-4650
OpenAlex
W2980504233
Cited by
792
References
122
Citations by year
2026: 73 · 2025: 94 · 2024: 83 · 2023: 82 · 2022: 75 · 2021: 99 · 2020: 94 · 2019: 62 · 2018: 43 · 2017: 33 · 2016: 20
FWCI
37.63
Type
journal article
Language
en
Open access
Locations
  • The Quarterly Journal of Economics (publishedVersion)
  • SSRN Electronic Journal (publishedVersion)
  • London School of Economics and Political Science Research Online (London School of Economics and Political Science) (acceptedVersion)
  • RePEc: Research Papers in Economics (submittedVersion)
  • London School of Economics and Political Science Research Online (London School of Economics and Political Science) (submittedVersion)
Funders
Economic and Social Research Council (ES/M010341/1)
Topics
  • Social Sciences › Business, Management and Accounting › Management of Technology and Innovation › Entrepreneurship Studies and Influences
  • Social Sciences › Economics, Econometrics and Finance › Economics and Econometrics › Firm Innovation and Growth
  • Social Sciences › Sociology and Political Science › Migration, Ethnicity, and Economy
Keywords
entrepreneurship, self-employment, incorporated businesses, unincorporated businesses, manual skills, learning ability, self-esteem, illicit activities, youth characteristics, educational attainment

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

A language model drew the claims from the paper on 2026-11-17, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-11-17 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.