The second and third columns show how the impact of labor market opportunities
has changed over time. Initially, we looked at three periods: 1967–1979,
1980–1989 and 1990–2003, which roughly coincide with the calendar decades and
include (in each period) a combination of boom and bust years. The results for the
1980s and 1990s are very similar, however, so we have combined them for ease of
presentation. The difference in the estimates across the second and third columns
shows quite strikingly that the impact of the labor market on poverty has weakened
over time. In the later period, the estimated coefficients on the labor market variables are roughly half of their estimated values in 1967–1979. Blank (1993) also
notes that the effect of economic growth (measured by growth in real GNP) fell
substantially during the 1980s, because growth in the 1980s consisted of stagnant
median wages and growing wage inequality. Our results take this finding a step
further: even after controlling for both median wage growth and inequality at the
bottom of the distribution, we see a dramatic reduction in the relationship between
labor market variables and the poverty rate. An interesting question is why the
predictive power of these different labor market variables seems to be changing
over time. One possibility stems from the rise in female employment—as more
women work, the shock to total household income associated with events like a
husband’s job loss may decline.