Various factors might account for the change in the distribution of daily hours worked by different wage deciles. The distribution of daily hours may be a poor indicator of total or yearly hours, particularly in the 1890s when workers who experienced large amounts of seasonal unemployment may have traded off a longer day for a shorter year. The length of the work day may be a poor indicator of the intensity of work, especially if some jobs require large amounts of time spent on the job but not working. Hours legislation may have lowered the hours worked by men and women in the lowest wage deciles. The number of daily hours supplied by workers in the lowest wage decile may have fallen relative to the number of hours supplied by workers in the top decile. Technological change such as electrification that allows firms to use different shifts of workers may have decreased firms’ demand for daily hours from each individual worker, but disportionately so for hours of work of lower skilled and hence lower paid workers.

Weekly and Yearly Hours

The highly inegalitarian distribution of daily work hours in the past translated into an unequal distribution of weekly and yearly hours. In the 1890s workers who reported that Sunday work was required were more likely to work a longer day, as were those who reported either no reduction or an increase in Saturday hours. Similarly in 1991 men who worked a longer usual day reported working longer usual weekly hours and more days per week.
I am also able to rule out the possibility that the longer hours of the lower paid in the 1890s were making up for their greater seasonal employment. Workers in the lowest deciles of the wage distribution did not report a disproportionate number of days lost due to sickness or unemployment. When I re-estimate the wage as yearly earnings divided by the yearly number of hours worked (the product of 307 days and days lost times daily hours worked), the results remain virtually unchanged. Furthermore, the number of days lost by the individual worker in the past year has a negligible, but negative, effect on his usual hours of work. The mean number of days lost by workers in the same 3 digit census occupation is also negatively related to the length of the working day.