The only difference relative to men comes in columns (8) and (9), when the unemployment rate variables averaged over the five post-schooling years are used. In this case, the sign of the first-stage estimate of the cohort average unemployment rate is negative and insignificant, whereas it was positive and significant for men. In addition, the F-statistic for the instruments in the first-stage regression is only 2.4, compared with much larger values in the other columns. This suggests that small sample bias may be non-negligible, and that, more generally, the instruments in this specification do not provide much identifying information. This is also reflected in the much larger standard error of the estimated coefficient of longest tenure attained in column (9). Thus, the estimates in columns (8) and (9) are not very informative. Electronic Payday Loans Online
Finally, the estimated effect of longest tenure attained falls, but remains positive (.14) and statistically significant, once controls for contemporaneous tenure are included in the preferred specification, as reported in column (2) of Table 6. Note also that the small positive effect of early tenure in the OLS estimates goes away after conditioning on current tenure. Overall, then, the beneficial effects of early job stability-as measured by longest tenure attained-are stronger for women, appearing even conditional on adult tenure.