Figure 3.3a: Smoothed Variation in 2-Digit Industry Hours from Manufacturing Mean
Figure 3.3b: Smoothed Variation in 2-Digit Industry Hours from Manufacturing Mean

Notable exceptions to the stability of these results are highlighted in black. They include especially transport and primary metals, each which experienced an increase in average hours of about 3 hours per week. These two industries were the same two that formed the backward bending portion of the wage-effort offer curve in the seventies. The increase in the number of hours in Figure 3.4: Capital Intensity, Wages and Hours these two industries is clearly tied to the same reasons the backward bending portion of the wage-effort curve disappeared from the data, i.e. most likely the decline of unionization. Those changing in the opposite direction, albeit on a smaller scale, include furniture and food which both experienced a 2 hour decline. The food sector and the furniture sector experienced a small decline in their relative capital intensity compared to other manufacturing industries, perhaps explaining the decline in the average number of weekly hours worked in this industry in detail.

Displays of the 1990 4-digit industry data across 4 digit Manufacturing Industries, 1990

The 4-digit industry data in the NBER productivity database are considerably more noisy than the 2-digit data, but the same pattern emerges. Capital-intensive sectors pay high wages and have long hours.

Figure 3.4 shows the distinctly positive relationship between average weekly hours and average real weekly wages for production workers across the industries included in the NBER data in 1990.11 Also striking is the strong positive relationship between the capital intensity of the sector with both average hours worked per week by production workers as well as weekly wages. Of these two, the relationship between capital and hours is noisier. This is likely due to short-term fluctuations in industry demand that are absorbed more by hours than by wages as well as due to the existence of greater noise in reported hours than in reported weekly earnings.

We expect a positive relationship to exist between weekly hours and weekly wages in the short run due to the inelasticity of the industry specific labor supply and normal fluctuations in relative industry demand. This could explain the upward sloping relationship seen in any one year. Yet as seen in the 2-digit data, the pattern of wages and hours across sectors is very stable over time. For example the correlation of weekly wages across sectors between 1990 and 1960 is .82 as can be seen in Table 3.1 which reports the cross-industry correlations of hours and wages at the various sample years, remarkably stable even over a thirty-year time period. Hours across sectors are also very stable. The correlation of weekly hours across sectors in 1960 and 1990 is .56.