Earlier research considered evidence on the relationship between early job market stability and adult labor market outcomes by exploring the correlations between a wide range of individuals’ youth labor market experiences and their labor market outcomes as more mature adults, in a multivariate framework that controlled for other adult characteristics (Gardecki and Neumark, 1998, hereafter GN). GN reported estimates of wage regressions for individuals in their late-20s to mid-30s, controlling for the usual ingredients of wage regressions-schooling, experience, etc-at the time the wage was measured, but adding in measures of youth labor market experiences over the first five years in the labor market, including number of jobs, longest job held, labor market experience, industry and occupation changes, etc. The results suggested that adult labor market outcomes are for the most part unrelated to the stability of early labor market experiences, especially for men, although as many studies have found, training bestowed longer-term benefits. This evidence was interpreted as undermining the case for policy initiatives to create more early job market stability in U.S. labor markets.

However, that evidence may be misleading if, for example, good job matches are likely to have resulted in early job stability. For the most part, GN documented that the estimated partial correlations between early job market stability and adult earnings were essentially zero, and argued that the positive bias in the estimates suggests that exogenous increases in early job market stability have if anything adverse effects on adult labor market outcomes.4 This is consistent with the view that job shopping is a crucial source of earnings growth for young workers, as the process of searching for better matches generates low early job market stability but ultimately higher adult wages. Although policy innovations that lead to better job matches would, undeniably, be helpful, this evidence suggests that policies targeting early job stability per se may be harmful. The present paper moves beyond this earlier work by studying explicitly the bias in the estimated relationship between early job market stability and adult labor market outcomes that stems from omitted job match quality or other sources, leading-it is hoped-to estimates of the causal effects of changes in early job market stability, which are a necessary input for policy evaluation. credit