The truth would then lie somewhere in between the OLS estimates (because the OLS estimates would be prone only to the downward bias from this negative correlation), and the IV estimates of quite large positive effects. On net, though, because the OLS estimates of the effects of early job stability are essentially zero, the results would still imply beneficial effects of early job stability.
Overall, then, once account is taken of the endogenous determination of early job stability, the evidence indicates that there are positive effects of this stability on adult wages, suggesting that policies that exogenously increase early job stability might have net beneficial effects. This conclusion should, however, be treated cautiously for four reasons.
First, there may be other possible explanations of the differences between the OLS results indicating no effects of early job stability, and the IV results indicating beneficial effects, which merit further investigation. Particularly given that the results from correcting for the endogeneity of early job stability are at odds with the initial hypothesis, and that the explanation of these results based on heterogeneity in the returns to search was developed ex post, there is good reason to consider other explanations.
Second, the estimates rely on the identifying assumption that the unemployment rate during the early years in the labor market is a valid instrument for early job stability.