i (2)Table 12 shows that between 1973 and 1991 inequality in the sum of husbands’ and wives’ weekly earnings widened. Because the dispersion in husbands’ hours of work widened as husbands in the highest combined household earnings deciles increased their hours while those in the lowest income deciles decreased their hours, 28 percent of the increase in earnings inequality could be accounted for by increases in the hours of husbands and wives holding wives’ labor force participation constant. Counteracting the increasing dispersion of husbands’ hours worked was the narrowing of dispersion in hours of wives in the labor force and the increase in wives’ labor force participation in low earnings decile households. Once increases in wives labor force participation rates are accounted for then only 17 percent of the increase in earnings inequality among married couples could be accounted for by changes in hours worked.

The distribution of work hours was very inegalitarian in the 1890s when the most highly paid worked 2 hours less per day than the lowest paid. By 1973 differences in hours worked between the top and bottom deciles were small and by 1991 workers in the top wage decile worked the longest day. In the 1890s inequality of daily work hours equalized income, but between 1973 and 1991 it magnified weekly earnings inequality, accounting for 26 percent of earnings inequality between the top and bottom deciles among men, more than all of the earnings inequality among women, and 17 percent of the increase in total household earnings inequality among husband and wife households.

I argued that much of the change in the inequality of the length of the work day could be explained by declines in the relative number of daily hours workers were willing to supply. I showed that in the past, in contrast to recent times, the labor supply curve was strongly backwards bending. It was so backwards bending that wage increases more than explain the two hour decline in the length of the work day from 1890 to 1919.
I also showed that a rising intertemporal elasticity of substitution may account for the change in the elasticity of supply of daily hours of work estimated from a regression of daily hours of work on the hourly wage. In the 1890s men were more likely to smooth hours over their work lives than they were in 1973 and in 1973 they were more likely to smooth hours over the life cycle than they were in 1991. Several factors could explain the increased bunching of hours over the life cycle. The work day is no longer so long that it leaves individuals with little free time in the day. Better savings and borrowing vehicles may have weakened the link between current consumption and current income. Returns to experience that are not reflected in the aggregate wage profile may have increased. Private pensions and Social Security provide strong financial incentives to take leisure at older ages. The rise of mass tourism and increases in fixed costs of work such as those incurred in commuting are additional explanations. Explaining the increased bunching of work hours over the life cycle remains a topic for future research. Regardless of what the explanation is, the results of this paper imply that although the rich and the poor will always differ in terms of income, income differences no longer mean that the poor have less time for fun.