Within a life cycle labor supply model, my estimated elasticity of daily hours supplied will be determined by wage changes arising from shifts in the wage profile, changes in the profile slope, and movements along a given lifetime wage profile. Thus not only could the wage profile have changed, but whereas in the 1890s workers may have preferred to take their recreation through reductions in the length of the work day (perhaps because the work day was so long), later generations of workers may have preferred reductions in the length of the work week, work year, and work life. Increased bunching in annual hours worked over the life cycle, among men both in and out of the labor force, is striking. Circa 1900 men’s annual hours were 2641 at ages 45 to 54, 2465 at ages 55 to 64, and 2118 at ages 65 to 74. Between 1940 and 1990, annual hours of men aged 45 to 54 increased from 1732 in 1940 to 1874 in 1960 and then to 1896 in 1990, largely because of the decline in part-year work; among men aged 55 to 64 annual hours fell 1732 in 1940 to 1577 in 1960 and 1291 in 1990. Among men aged 65 to 74 the decline has been particularly pronounced – from 940 hours in 1940 to 590 hours in 1960 and then to 313 hours in 1990.
I can use my three datasets to determine whether men in the labor force are now more likely to bunch work time during prime ages than they were in the 1890s and whether this is attributable to changes in the wage profile. Although the wage profile was flatter in the 1890s than in either 1973 or 1991, the hourly wage peaked at around age 40 to 45 in all three years. However, in the 1890s daily hours worked peaked around age 27 and then declined slightly, bearing very little relation to the wage, whereas in 1973 and in 1991 hours worked have the familiar hump shape.
The intertemporal substitution of daily hours worked with respect to the wage rose from-0.072 in the 1890s to 0.125 in 1973 and0.147in 1991 (seeTable9). If the retired had been included in the samples, the increase in the intertemporal elasticity of substitution would probably have been even greater. The changing slope of the labor supply curve may therefore reflect individuals’ increasing willingness to bunch their hours of work over the life cycle.