YOUTH LABOR MARKETS IN THE U.S: Results 9

The final analysis addresses the issue of measurement of local unemployment rates. As explained in Section III, the unemployment rates for individuals residing outside of metropolitan areas (as well as those residing in metropolitan areas for which unemployment rates are not reported in Employment and Earnings) are rates for the entire non-metropolitan area of their state of residence. Consequently, unemployment rates may measure local labor market conditions much more accurately for those residing in SMSAs for which separate unemployment rates are reported in Employment and Earnings. The estimates of the key specifications, both with and without adult tenure included, were therefore recomputed using only observations on individuals residing in this subset of SMSAs in each of the early labor market years as well as 1992 (since the unemployment rate for each of these years is required).

Men Women Men Women
(1) (2) (3) (4)
OLS loe wage eauation estimates:
Longest tenure attained, -.006 .015
five-year post-schooling period (.01) (.012)
Number of jobs held, -.008 .004
five-year post-schooling period (.007) (.007)
IV log waee eauation estimates:
Longest tenure attained, .05 .14
five-year post-schooling period (.04) (-04)
Number of jobs held, -.06 .04
five-year post-schooling period (.05) (-04)
Tenure .04 .08 .03 .07
(01) (.02) (.02) (.01)
Tenure squared * 10′2 -.19 -.55 -.22 -.35
(.12) (.13) (.15) (.11)
Minimum unemployment rate -.05 -.08 -.06 -.08
on current job (.02) (.01) (.02) (.01)
Current unemployment rate .02 .03 .02 .02
(.01) (.01) (01) (.01)
First stage estimates:
Uj. .29 .36 -.33 -.19
(.05) (.11) (.09) (.13)
Up .02 .03 -.12 -.16
(04) (.04) (08) (.08)
-.13 -.00
(04) (.08)
(1VUj,) -.03 -.07
(.03) (.05)
<vlj2-uj2) .02 .02
(-03) (05)
(Uv-Up) .01 -.00
(.03) (.05)
(Ulj4-u.j4) -.02 -.13
(.03) (06)
(Ujj5-U.j5) .01 .00
(03) (.07)
F-statistic for instruments in first stage 13.4 26.0 6.6 2.6
P-value for Hausman exogeneity test .12 .00 .30 .43
P-value for test of overidentifying .44 .98 .85 .11
restrictions
P-value for test of overidentifying restrictions when minimum unemploymen rate on current job is excluded .07it .99 .84 .02

The results are reported in Table 7.

Excluding tenure Including tenure
Men Women Men Women Men Women Men Womer
OLS log waee eauation estimates: (1) (2) (3) (4) (5) (6) (7) (8)
Longest tenure attained, five-year post-schooling period .023(.016) .017(.014) .015(.016) .023(015)
Number of jobs held, five-year post-schooling period -.022(.009) -.010(.009) -.015(.009) -.007(.009)
IV log wage eauation estimates:
Longest tenure attained, five-year post-schooling period .11(.05) .18(06) .073(045) .14(05)
Number of jobs held, five-year post-schooling period -.14(07) .015(.05) -.121(074) .03(.05)
Tenure .06(02) .09(.02) .05(02) .05(02)
Tenure squared * 10′2 -.31(.16) -.62(.18) -.41(.22) -.25(.15)
Minimum unemployment rate -.08 -.08 -.11 -.08 -.08 -.08 -.10 -.08
on current job (.02) (-02) (.04) (.02) (.02) (.02) (.04) (.02)
Current unemployment rate .05 .04 .08 .04 .05 .04 .07 .03
(-02) (.01) (.02) (.01) (.01) (01) (.02) (.01)
First stage estimates:
Uii .26 .48 -.45 -.33 .28 .61 -.39 -.26
(.10) (.29) (.15) (21) (09) (.27) (14) (.21)
U* -.08 .02 -.15 -.24 -.10 .05 -.10 -.24
(.07) (-10) (.13) (.12) (.07) (.09) (.12) (12)
Up -.20(.10) .09(20) -.21(.09) .17(.19)
(Uyi-U,,) -.02(-05) -.03(.07) -.01(.05) -.03(.07)
(Uj2-U.j2) .09(.05) -.04(.07) .06(.05) -.04(.07)
avUj,) -.07(.05) .07(.08) -.05(.05) .08(.08)
(IV u*) .03(.05) -.20(.10) .04(.05) -.21(.09)
flVUjj) .03(.05) .07(.10) -.03(.05) .08(.09)
F-statistic for instruments in first stage 7.2 10.7 4.7 1.5 8.6 12.8 3.7 1.7
P-value for Hausman exogeneity test .05 .01 .07 .63 .16 .02 .15 .48
P-value for test of overidentifying restrictions .32 .97 .91 .11 .32 .85 .77 .37
P-value for test of overidentifying restrictions when minimum unemployment rate on current job is excluded .02 .98 .90 .02 .02 .86 .76 .11
N 474 434 384 363 474 434 384 363

For the most part, the magnitudes of the estimated coefficients of local unemployment rates in the first-stage regressions and in the IV estimates of the wage equations are larger in absolute value, consistent with a reduction in measurement error in unemployment rates. More importantly, for almost all of the estimates the evidence of beneficial effects of early job stability is stronger than in Table 6 and the corresponding estimates in Tables 2-5. The most notable difference, perhaps, is that even in the specifications including adult tenure, the estimated effects of early job stability for men are positive, and significant or nearly so at the ten-percent level (or better). Thus, for this subsample for which the instruments are measured more accurately, there is relatively strong evidence of positive returns to early job stability for men and women, even conditional on adult tenure.-.22/span-.41