If early job stability enhances adult labor market outcomes conditional on the adult characteristics in X, then oc in equation (1) should be positive when S is the longest tenure attained, and negative when S is the number of jobs held. In GN, OLS estimates generally indicated that the coefficients on longest tenure attained and number of jobs were small and insignificant, and often did not have the expected sign.
The component riit is the usual idiosyncratic error, assumed to be uncorrelated with S and X, and independently and identically distributed across observations. The component fiit is an unobserved measure of the quality of the adult job match, and hence is assumed to be positively correlated with ln(w). In turn, |ijt is presumed to be positively related to the quality of the job match attained during the five-year post-schooling period, jiit., for two reasons. First, some individuals remain on the same job and hence retain their earlier match. Second, all else the same individuals with good matches in the early years may be more likely to have good matches as adults even if they changed jobs, because voluntary job changes, at least, are likely to be in the direction of even better matches. Thus, because jiit. is likely to be positively correlated with early job stability, (ijt is also likely to be positively correlated with early job stability, suggesting that OLS estimation of equation (1) will result in upward biased estimates of the effects of early job stability. no teletrack payday loans
The concern with omitted job match quality suggests that the error term e consists of two components, The strategy for estimating equation (1) in the presence of an unobserved variable related to match quality is to use an instrumental variable for S, the measure of early job stability. What is required is a variable or set of variables that explains variation in S, but does not directly affect the current wage-specifically, it must be uncorrelated with ц1Г Various measures of the unemployment rate faced by the individual in the immediate post-schooling period are used. Variation in unemployment rates faced by young people should be related to early job stability. For example, if a recession occurs soon after a person enters the labor market, the likelihood of holding a job for a long period would be reduced. Similarly, the number of jobs the person held in this period might increase, if the principal effect of the recession is to cause jobs to end. In this case, a weak youth labor market reduces early job stability. On the other hand, the relationship between early unemployment and early job stability could go in the opposite direction, as, for example, slack labor markets reduce wage offers from firms to employed workers and hence deter mobility. The direction of this relationship is an empirical question; what is required, however, is that youth labor market conditions help to predict early job stability.