نگرشها نسبت به خطر و خود اشتغالی کارگران جوان

A high degree of risk tolerance is often regarded as one of the fundamental characteristics of entrepreneurs. Using multiple responses on risky income gambles in the 1979 National Longitudinal Survey of Youth (NLSY79), I investigate the effect of individual risk tolerance on the probability of entry into self-employment. I construct a measure of individual level of risk tolerance that is corrected for reporting error and that varies with age and other covariates that potentially affect self-employment decision. I find that risk tolerance is an important determinant of the decision to enter self-employment. However, I find that the estimated effect of risk tolerance on the probability of entering self-employment is dramatically understated if measurement error is not taken into account. In addition, I find that that accounting for the correlation between risk tolerance and other covariates is important to correctly assess the effects of the other determinants of self-employment while it has a trivial effect on the estimated marginal effect of risk tolerance.

The number of self-employed workers in the U.S. has grown dramatically in recent decades. Estimates show that over ten percent of the labor force is self-employed when incorporated businesses are counted (Hipple, 2004) and a quarter of young men are self-employed at some time in their early careers (Ferber and Waldfogel, 1998). Because self-employment is often considered a way for disadvantaged workers to achieve economic prosperity and upward mobility (Fairlie, 2004), these rising rates have sparked policy makers' and researchers' interest in the determinants of self-employment. In this paper, I investigate the role of risk preference on self-employment entry decision using repeated measures of individual risk preference from the 1979 National Longitudinal Survey of Youth (NLSY79). While individual risk tolerance has long been considered a key characteristic of entrepreneurs (Knight, 1921), empirical research has been limited by a lack of data on individual risk preference—and even when data are available, empirical researchers have typically assumed that self-reported individual risk attitudes are measured without errors (Cramer et al., 2002, Guiso and Paiella, 2005 and Dohmen et al., 2005). The important feature of this paper is that I address the issues of measurement error and time variation in self-reported risk tolerance by using multiple responses to identical “income gamble” questions asked in the NLSY79. Following Barsky et al. (1997) and Kimball et al. (2005), I correct my measure of risk tolerance for measurement error. In light of arguments that individual risk tolerance is likely to decrease with age (Morin and Suarez, 1983, Bakshi and Chen, 1994 and Sahm, 2006) and in order to account for the correlation between risk tolerance and other determents of self-employment, I also allow the measure of risk tolerance to vary with age and other covariates that potentially affect self-employment decisions. I find that the estimated effect of risk tolerance on the probability of entering self-employment is dramatically understated if measurement error is not taken into account. While the latter feature proves to have a small effect on the estimated effects of risk tolerance, it affects the estimated coefficients for other determinants of self-employment. I find that individual risk tolerance plays an important role in the self-employment entry decision. An increase in relative risk tolerance from the 10th percentile to the 90th percentile increases the predicted probability of entry into self-employment by 35%. However, using a one-time measure of risk tolerance attenuates its estimated effect by a staggering 90% because measurement error cannot be netted out. This result underscores the need to use multiple responses in order to correct self-reported risk tolerance for reporting error. In addition, I find that that accounting for the correlation between risk tolerance and other covariates included in self-employment decision model is important to correctly assess the effects of the other determinants of self-employment.

In this paper, I identify the effects of individual risk tolerance on the probability that a young man enters self-employment. To construct a measure of relative risk tolerance, I use responses to “income gamble” questions asked in two years of the NLSY79, and I use this within-person variation to account for both survey response error and variations due to aging and other covariates that potentially affect self-employment decision. I then construct a measure of relative risk tolerance by assuming constant relative risk aversion. My estimates reveal that relative risk tolerance has a large, positive, and statistically significant effect on the probability of entering self-employment. An individual whose level of risk tolerance is one standard deviation above the mean is 13% more likely to enter self-employment than is an otherwise identical man. However, the estimated effect of risk preference on self-employment entry is attenuated by 90% when I fail to account for measurement error in self-reported risk tolerance. This evidence implies that the “income gamble” questions used in the NLSY79 and other surveys elicit valuable information on individual risk tolerance, but that it is crucially important to collect multiple responses and to correct them for reporting error. In addition, it turns out that accounting for the correlation between risk tolerance and other covariates in self-employment decision model is important in correctly assessing the effects of the other covariates. The fact that risk tolerance proves to be empirically important is consistent with an absence of markets to insure against losses from business failures. Because individuals cannot insure against this risk, they must be risk tolerant in order to enter self-employment. If self-employment and business formation are considered desirable from an individual or social standpoint, there may be legitimate policy concerns regarding insurance market against the income risk born by the self-employed.