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

Citation

Levine, R., & Rubinstein, Y. (2017). Smart and illicit: Who becomes an entrepreneur and do they earn more? The Quarterly Journal of Economics, 132(2), 963–1018. https://doi.org/10.1093/qje/qjw044

Chapter Summary

This is a journal article of 56 pages with an introduction, four further sections and a conclusion, and it can be read as one extended empirical argument. It asks two questions that the title states: who becomes an entrepreneur, and whether entrepreneurs earn more than other workers. The answer rests on a single measurement decision, which is to split the self-employed into the incorporated and the unincorporated.

I. Introduction

Opens with a puzzle. Economists since Adam Smith and Schumpeter have treated entrepreneurs as drivers of growth, yet studies that use self-employment as a stand-in for entrepreneurship find that the self-employed earn no more than salaried workers and look much like them in education, aptitude and family background. The authors argue that self-employment lumps together very different people. They propose the incorporated self-employed as a closer proxy for Schumpeterian entrepreneurs, and preview their findings on traits, earnings and the combination of "smart" and "illicit" tendencies.

II. Data and Summary Statistics Across Employment Types

Introduces three data sources. The March supplements of the Current Population Survey provide a large cross-section and a two-year matched panel, with workers classified as salaried, unincorporated self-employed or incorporated self-employed. The National Longitudinal Survey of Youth 1979 (NLSY79) follows young people from their teens into mid-career and records cognitive and noncognitive traits before they enter the labour market. The US Department of Labor's Dictionary of Occupational Titles measures the skills that jobs require. Summary statistics already separate the incorporated from the rest in earnings, hours, gender, race and education.

III. The Schumpeterian Entrepreneur and Other Self-employed Business Owners

Links the Schumpeterian view, in which entrepreneurs break from routine and introduce novelty, to testable predictions about job skills. Using the Dictionary of Occupational Titles, the authors ask whether the incorporated and unincorporated perform different activities, whether they open different kinds of businesses, and whether incorporation reflects success after the fact. They find that the incorporated self-employed and their businesses demand more nonroutine cognitive skills, while the unincorporated and their firms demand more manual skills.

IV. Who Becomes an Entrepreneur? Evidence from the NLSY79

The traits section. It describes the cognitive and noncognitive measures (an aptitude test score, self-esteem, locus of control and an index of illicit activity in youth built from questions on behaviours such as skipping school, vandalism or shoplifting), summarises them by employment type, and then analyses selection on traits, on labour market productivity, and the link between traits, employment types and job task requirements. High aptitude, high self-esteem and a record of illicit activity in youth are associated with later incorporated self-employment, and the combination of high aptitude with illicit activity is what predicts it most strongly.

V. Entrepreneurs: Do They Earn More?

The longest section, on earnings. It sets up a baseline specification, estimates it separately on the CPS and the NLSY79, and then deals in turn with the selection problems that could distort the comparison: selection into incorporated self-employment, selection out of it, and whether the smart-and-illicit combination matters for earnings in entrepreneurial businesses. It closes with the distribution of earnings and a measure of risk, the coefficient of variation. The central results are that the incorporated earn more than comparable salaried workers and the unincorporated earn less, and that much of the incorporated earnings advantage reflects who selects into incorporation, with an additional gain on switching.

VI. Conclusions

Summarises the case for disaggregating the self-employed and restates the findings: entrepreneurs, measured by incorporation, have a distinctive mixture of cognitive and noncognitive traits and earn more, and it is the high-ability person who tends to break the rules as a youth who is especially likely to become a successful entrepreneur.

Key Concepts

Measurement

• Incorporated versus unincorporated self-employment: the paper's proxy for entrepreneurs and for other business owners, based on the legal form of the business.
• Schumpeterian entrepreneurship: the view of entrepreneurs as creative leaders who disrupt existing markets and take risks.
• Nonroutine cognitive versus manual skills: the job characteristics used to show that the two groups do different work.

Traits and Selection

• Cognitive and noncognitive traits: aptitude on one side, and self-esteem, locus of control and illicit behaviour on the other.
• Smart and illicit: the combination of high aptitude scores and rule-breaking behaviour as a teenager.
• Selection into and out of self-employment: the concern that earnings gaps reflect who becomes an entrepreneur, not what entrepreneurship does.

Earnings

• Individual (person) effects: controlling for fixed characteristics of workers using panel data.
• Coefficient of variation of earnings: the measure of earnings risk used to compare entrepreneurs with salaried workers.

Critical Analysis

Strengths

• A measurement idea with consequences. Splitting the self-employed by incorporation reconciles the puzzle that motivates the paper, since the incorporated and unincorporated differ in skills, traits and earnings in different directions.
• Several data sets used together. The paper combines a large survey (CPS) with a longitudinal survey that records traits before entry (NLSY79) and an occupational database on task requirements, so each claim is checked in more than one way.
• Selection is taken seriously. The earnings section lists four specific selection challenges, such as incorporating after a successful start or leaving after failure, and addresses each with panel data and the analysis of switching.
• Traits studied before careers begin. Measuring aptitude, self-esteem and illicit activity in the teenage years avoids reading traits back from outcomes.

Limitations

• Not experimental. The authors state that they do not evaluate the effect of randomly making someone a business owner but of what happens when a person chooses to become self-employed, so the results are patterns in observational data.
• A coarse proxy. Incorporation is self-reported and the authors acknowledge that it covers many legal forms; the proxy is validated through task requirements and not observed directly.
• United States and a particular cohort. The evidence uses US surveys, and the trait data come from people who were teenagers or young adults in 1979, so it is open to question how far the results extend to other countries and later cohorts.
• A constructed illicit measure. The index is built from self-reported youth behaviours, and "smart and illicit" is defined by thresholds on the aptitude score and the index.

Where It Fits in the Course

Among the public and labour economics readings, this is the only single empirical study. Cahuc, Carcillo and Zylberberg's Labor Economics is a graduate textbook that surveys the whole field, and Ehrenberg, Smith and Hallock's Modern Labor Economics is an undergraduate text that builds the analysis from supply and demand and applies it to policy. The paper is best read after one of them, as a worked example of how economists use survey data, panel methods and selection arguments on a concrete labour market question.

Gruber's Public Finance and Public Policy, Stephen Smith's Taxation: A Very Short Introduction and Piketty's The Economics of Inequality cover the economics of government, taxation and inequality, and the paper's findings on earnings dispersion connect to the inequality theme.

How to Use This Book

The paper suits advanced undergraduates and graduate students who can read regression tables: an introductory econometrics course, with some familiarity with fixed effects and panel data, is enough for the main argument.

A practical reading order:

• First pass: Sections I and VI (the introduction and the conclusion) give the argument in a few pages.
• Second pass: Section III, to see why incorporation is a proxy for entrepreneurship, then Section IV on traits.
• For method: Section V, especially its list of selection challenges and how each is handled.

As an exercise, restate the claim that incorporated business owners earn more as three separate statements about levels, about selection and about the gain on switching, and look for the evidence for each.