The most likely alternative explanation for the correlation between wages and capital intensity is complementarity of human and physical capital. We include measures of human capital as well as rate of unionization in our equations that explain wages and hours. Unionization and education both have positive simple correlations with weekly wages and weekly hours. Controlling for capital intensity of the sector, both education and union membership have a positive and statistically significant effect on wages but do not have a measurable effect on hours. Even after controlling for education and unionization, there remains strong evidence of the positive relationship between wages and effort that we are looking for.
Another possible explanation for the apparent wage-effort offer curve is rent-sharing with rents especially high in capital intensive sectors. We explore this possibility by using an imperfect measure of industry rents and do not find that rents can explain away our findings.
Our analysis could be contaminated by business cycle effects. The first response to a slowdown in sales is a reduction in hours and only when the slowdown is judged to be long-lived is there a reduction in employment. Then when sales begin to grow again the first response is to increase hours, followed later by an increase in employment. The variation in employment causes opposite variation in capital per worker because capital is a very slow-moving series. This cycle in hours and capital per worker causes us concern about our choice of dates for estimating the wage-effort offer curve since the movements in the curve over time may be mostly due to the cycle. We control for the cycle by estimating the wage-effort offer curve at business cycle peaks and business cycle trough, and then comparing peak with peak and trough with trough.
The theory that drives the data analysis concerns the demand for labor, but the market of course has to have a supply side as well. We do briefly explore one labor-supply variable: gender. We find what is already rather well known: females are more likely to be employed in low effort-low hours sectors.
In the next section of this paper we review pertinent aspects of the theory offered in Learner( 1997b). Also in Section 2 we summarize related literature. We argue that we offer a unique theoretical viewpoint that is distinct from efficiency wage literature and that leads us to estimating an entirely new equation that explains hours as a function of the capital intensity of the sector. In Section 3 we discuss graphical displays of the 2-digit SIC data on hours and wages. These displays conform remarkably well with the theory. The capital-intensive sectors have long hours and high hourly wage rates. We also present a formal analysis at the 4-digit level that backs up what is evident in the 2-digit displays. Finally, we offer some summary and concluding remarks in Section 4.