YOUTH LABOR MARKETS IN THE U.S: Discussion and Conclusions 4

The fact that the IV estimates indicate net positive returns to early job stability in the earliest years in the labor market is not inconsistent with an explanation based on variation in returns to search. Although the explanation does require some positive returns to job shopping, it does not require that there be no positive returns to job stability in the earliest years in the labor market.

Rather, there can be competing beneficial effects of job stability and job shopping. However, according to this explanation, selection of those with low returns to search-and hence worse adult matches-into careers marked by early job stability, leads OLS estimates to understate the relative return to job stability, while IV estimates suggest that, on balance, the returns to job stability in the earliest years in the labor market are greater.
Finally, the earlier discussion argued against the possibility that a positive correlation between unemployment rates and match quality leads the IV estimate of the effect of early job stability to be more upward biased than the OLS estimate. The argument was that the positive relationship between and Sit.-reflecting those with good matches experiencing early job stability-would outweigh this source of bias. However, if, as suggested here, jiit and Sit. are negatively correlated, then upward bias in the IV estimates relative to the OLS estimates is a more plausible possibility, and suggests that the IV estimates of the positive effects of early job stability could be overstated.