This twisting of the curve of observed wage-effort offers is not compatible with a representative worker model with a stable utility function since the contracts in the capital-intensive sectors have unambiguously improved while those in the labor-intensive sectors have unambiguously deteriorated. This twisting of the curve could be an equilibrium if workers have heterogeneous preferences. Differences in adversity to effort may also explain a portion of the apparent rigidity of the labor market to changes in relative wages across sectors: despite the shift in the wage-effort offer curve, the distribution of production workers across sectors has remained fairly stable. This is illustrated in Table 3.6 which reports employment by quintile of capital intensity. The sectors with the lowest capital/labor ratios experienced an 8.5% reduction in employment between 1970 and 1980. The third and fourth quintiles experienced the greatest gains while there was actually a decline in employment in the most capital-intensive sectors.

In the 1980s the wage-effort offer curve began to shift to the right, a movement that is depicted in Figure 3.7. In words, for the same wage levels, workers were required to work more hours. The increase in weekly hours was between 2 and 3 hours during this 13-year period, more at the upper end of the wage-effort curve than at the lower end. One possible explanation for this shift is increasing real capital rental costs coming from the rise in the demand for capital induced by the economic liberalizations in Asian and Latin America. Another possible explanation is the introduction of new more expensive equipment (computers and robots) which shifts the curve as illustrated in the middle panel of Figure 2.5. A third possibility is the business cycle. There was a 4-point decline in unemployment after the peak during the recession in 1983. The strong correlation between employment and weekly hours was highlighted in Figure 3.2. Yet there are a number of reasons to doubt the business cycle role in the sharp increase in weekly hours. One reason is the magnitude of the change. The larger change in unemployment between 1969 and 1983 (a reduction of 5.5 points) only lead to a 1.25 hour decrease in average weekly hours, much smaller than the 2 hour increase seen after 1983.

Figure 3.7: Derived Wage-Effort Offer Curves, 1980-1993
We think that a likely cause for part of the shift in the wage- Figure 3.8: Benefit to Wage Costs in Manufacturing, 1989=100 effort offer curve since 1980 was an increase in the quasi-fixed labor costs emphasized by Oi: training, payroll taxes and worker benefits. Whether it is capital or benefits, firms can save costs paid on a per-worker basis by getting more work out of each worker; thus more benefits for more hours. Figure 3.8 plots the ratio of a BLS employee benefits index relative to wages per worker in manufacturing in the 1980’s. Between 1980 and 1996 this index increased about 25%. Note that the majority of this increase occurred after 1986. Unfortunately the data does not go back beyond 1979, so there is no way to compare this trend to earlier periods.

Controlling for Business Cycles
According to our estimates, the wage-effort offer curve has varied systematically over time, shifting up, then twisting and finally shifting right. The 1970 and 1980 dates at which we estimate the wage-effort offer curve were selected to conform with Leamer’s(1997) claim that the 1970s were the Stolper-Samuelson decade in which there was a significant decline in the relative prices of labor-intensive tradables. But these years and the other five-year intervals at which we estimate the offer curve select different points in the business cycle, and some of the observed shifting may be due to the cycle not the fundamentals. In this section we show that in fact the timing is not essential. Comparisons, peak-to-peak and trough-to-trough, show the same shifting of the wage-effort offer curve. payday loans online direct lenders only