Theoretical models of entrepreneurship posit that attitudes toward risk, entrepreneurial ability, and preferences for autonomy are central to the individual's decision between self-employment and wage/salary work.  None of the studies in the rapidly growing empirical literature on entrepreneurship, however, have been able to test whether these factors are important determinants of self-employment.  I provide indirect evidence on this hypothesis by examining the relationship between drug dealing and legitimate self-employment.  A review of ethnographic studies in the criminology literature indicates that drug dealing may represent a useful proxy for low risk aversion, entrepreneurial ability, and a preference for autonomy.

            The 1980 wave of the National Longitudinal Survey of Youth (NLSY) contained a special section on participation in illegal activities, including questions on selling marijuana and other "hard" drugs.  I use the answers to these questions and data from subsequent years of the NLSY to examine the relationship between drug dealing as a youth and legitimate self-employment in later years.  Using various definitions of drug dealing and specifications of the econometric model, I find that drug dealers are 11 to 21 percent more likely to choose self-employment than non drug dealers, all else equal.  I also find that drug dealers who sold more frequently, used drugs less frequently, or reported receiving income from drug dealing are more likely to choose self-employment than other drug dealers.  After ruling out a few alternative explanations, I interpret these results as providing indirect evidence that risk aversion, entrepreneurial ability, and preferences for autonomy are important determinants of self-employment.

Theoretical models of entrepreneurship posit that attitudes toward risk, entrepreneurial ability, and preferences for autonomy are central to the individual's decision between self-employment and wage/salary work.  None of the studies in the rapidly growing empirical literature on entrepreneurship, however, have been able to test whether these factors are important determinants of self-employment.  I provide indirect evidence on this hypothesis by examining the relationship between drug dealing and legitimate self-employment.  A review of ethnographic studies in the criminology literature indicates that drug dealing may represent a useful proxy for low risk aversion, entrepreneurial ability, and a preference for autonomy.

            The 1980 wave of the National Longitudinal Survey of Youth (NLSY) contained a special section on participation in illegal activities, including questions on selling marijuana and other "hard" drugs.  I use the answers to these questions and data from subsequent years of the NLSY to examine the relationship between drug dealing as a youth and legitimate self-employment in later years.  Using various definitions of drug dealing and specifications of the econometric model, I find that drug dealers are 11 to 21 percent more likely to choose self-employment than non drug dealers, all else equal.  I also find that drug dealers who sold more frequently, used drugs less frequently, or reported receiving income from drug dealing are more likely to choose self-employment than other drug dealers.  After ruling out a few alternative explanations, I interpret these results as providing indirect evidence that risk aversion, entrepreneurial ability, and preferences for autonomy are important determinants of self-employment.

Theories of market failures and targeting motivate the promotion of entrepreneurship training programs and generate testable predictions regarding heterogeneous treatment effects from such programs. Using a large randomized evaluation in the United States, we find no strong or lasting effects on those most likely to face credit or human capital constraints, or labor market discrimination. We do find a short-run effect on business ownership for those unemployed at baseline, but this dissipates at longer horizons. Treatment effects on the full sample are also short-term and limited in scope: we do not find effects on business sales, earnings, or employees.

Approximately 9 out of 10 high school students who have access to a home computer use that computer to complete school assignments.  Do these home computers, however, improve educational outcomes?  Using the Computer and Internet Use Supplement to the 2001 Current Population Survey, I explore whether access to home computers increases the likelihood of school enrollment among teenagers who have not graduated from high school.  A comparison of school enrollment rates reveals that 95.2 percent of children who have home computers are enrolled in school, whereas only 85.4 percent of children who do not have home computers are enrolled in school. Controlling for family income, parental education, parental occupation and other observable characteristics in probit regressions for the probability of school enrollment, I find a difference of 1.4 percentage points.  Although the evidence is mixed on whether the errors are correlated, I also estimate bivariate probit models for the joint probability of school enrollment and owning a home computer and find larger effects (7.7 percentage points).  Use of computers and the Internet by the child's mother and father are used as exclusion restrictions.  The estimates are not sensitive to alternative combinations of exclusion restrictions and alternative samples.

Approximately 9 out of 10 high school students who have access to a home computer use that computer to complete school assignments.  Do these home computers, however, improve educational outcomes?  Using the Computer and Internet Use Supplement to the 2001 Current Population Survey, I explore whether access to home computers increases the likelihood of school enrollment among teenagers who have not graduated from high school.  A comparison of school enrollment rates reveals that 95.2 percent of children who have home computers are enrolled in school, whereas only 85.4 percent of children who do not have home computers are enrolled in school. Controlling for family income, parental education, parental occupation and other observable characteristics in probit regressions for the probability of school enrollment, I find a difference of 1.4 percentage points.  Although the evidence is mixed on whether the errors are correlated, I also estimate bivariate probit models for the joint probability of school enrollment and owning a home computer and find larger effects (7.7 percentage points).  Use of computers and the Internet by the child's mother and father are used as exclusion restrictions.  The estimates are not sensitive to alternative combinations of exclusion restrictions and alternative samples.