The estimates presented in Table 2 and Figure 5 ignore the potentially
offsetting increase in women’s labor force participation illustrated in Figure 4. We
did not include women’s labor force participation rates in our initial model since
they may reflect choices (and so be a function of the poverty rate), rather than
reflecting primarily prices or constraints like our measures of unemployment and
wages. To examine the importance of the trend toward increasing female employment,
however, we add the fraction of women between the ages of 25 and 64 who
are employed to our regression model.8 Columns 4–6 of Table 2 show how this
addition changes the estimated effects of our labor market variables. The inclusion
of the female employment variable has virtually no effect on the other estimated
labor market coefficients prior to 1980 and very little effect on the estimates in the
post-1980 period. At the same time, the female employment variable itself is
strongly negatively correlated with the poverty rate in the later period (with no
significant impacts in the earlier period). Using the coefficients in column 6, we
again create counter factual poverty rates for each year, this time using the female
employment rate along with the labor market variables. This predicted poverty rate