YOUTH LABOR MARKETS IN THE U.S: Results 8

Columns (6)-(9) report the estimates first using the cohort average unemployment rates and individual-level deviations from these, defined over each of the five post-schooling years, and then using the five-year averages. In columns (7) and (9) the IV estimates of the effect of number of jobs held are smaller than in column (5), and insignificant. However, in these estimations the p-values for the overidentifying restrictions are quite low (.18 and .08), suggesting that the estimates in column (5) are preferred, presumably because the estimates in columns (6)-(9) use the individual-level unemployment rates that may be partly endogenous. The estimate of -.08 in column (5) implies that a one standard deviation (2.63) increase in the number of jobs held in the immediate post-schooling period lowers adult wages by 21 percent.

OLS First staee IV First staee IV First staee IV First stage IV
(1) (2) (3) (4) (5) (6) (7) (8) (9)
Number of jobs held, five-year post-schooling period -.00(-01) .03(.05) -.14(.10) .00(.04) .07(05)
Schooling .06 -.04 .06 -.04 .05 -.05 .06 -.04 .06
(.01) (07) (.01) (.07) (.02) (-07) (01) (.07) (.01)
Experience .01 .90 -.01 .93 .14 .93 .01 .85 -.05
(.02) (.14) (.05) (.14) (.10) (.14) (.04) (14) (.05)
Experience squared * 1 O’2 .25 -6.1 .40 -6.3 -.61 -6.3 .27 -5.8 .65
(.15) (.87) (.33) (.88) (.65) (.88) (.28) (.84) (.35)
Currently married .02 -.36 .03 -.31 -.02 -.37 .02 -.26 .03
(.03) (.18) (.03) (.18) (05) (.18) (.03) (.17) (.03)
Non-white .03 -.98 .06 -.91 -.10 -.93 .03 -.86 .10
(04) (.25) (.06) (.25) (.11) (.25) (.06) (.24) (.06)
Minimum unemployment rate -.07 .10 -.07 -.05 -.08 .12 -.07 .14 -.07
on current job (01) (.10) (.01) (.08) (.02) (.10) (01) (09) (.01)
Current unemployment rate .02 .08 .02 .01 .03 .07 .02 .05 .03
(.01) (.06) (-01) (.06) (.01) (.06) (.01) (.05) (.01)
ц. -.08(06)
U.J2 -.01(.05)
-.06(.04)
Uj4 -.02(.05)
ц* -.02(.07)
u. -.26(.14) -.26(.14)
и,, -.17(.08) -.18(.08)
Ч.. -.31(.40)
(U.rU,,) -.07(.05)
(iVUj.) .01(.05)
(uij3-u,o -.01(.06)
(Uij4-u.j4) -.13(06)
flVUjs) .02(07)
(Uii-Uj.) -.17(.05)
F-statistic for instruments in first stage 2.5 2.4 2.9 6.2
P-value for Hausman exogeneity test .60 .16 .93 .18
P-value for test of overidentifying restrictions .00 .05 .01 .07
P-value for test of overidentifying restrictions when minimum unemploymen rate on current job is excluded it .00 .03 .00 .44

Finally, column (3) of Table 6 reports estimates with controls for contemporaneous tenure in the wage equation, for the preferred specification. The small negative estimated effect of number of jobs held in the OLS estimates is no longer present after controlling for adult tenure. The IV results indicate a negative coefficient on number of jobs held in the immediate post-schooling period, although as for longest tenure attained the estimated effect conditional on adult tenure is smaller (falling from -.08 to -.06) and insignificant, implying that some of the effect of early job stability arises through higher adult tenure.

Paralleling Table 4, Table 5 reports results for women for the number of jobs in the immediate post-schooling period. The OLS estimates in column (1) indicate no effect of number of jobs held. In the remaining columns of the table, the IV estimates of the effect of number of jobs held range from positive to negative, although they are never statistically significant. However, at least in columns (2)-(7), the F-statistics in the first-stage regressions are rather low (2.4-2.9), and in all columns the overidentifying restrictions are rejected at the five- or ten-percent level even when the minimum unemployment rate on the current job is included. Thus, for women, the data (using these instruments, at least) do not appear to permit us to identify the exogenous effects of number of jobs held in the post-schooling period.