Given the measurement difficulties, we shouldn’t be expecting much. The big surprise is that there is a remarkably clear relationship between hours, wages and capital intensity. The capital-intensive sectors have longer annual hours and higher hourly wage rates, exactly what the theoiy would suggest if hours were a perfect indicator of effort. Not only do we find a wage-effort offer curve; we also find it shifting just as the theory suggests. In the 1960s with stable relative prices but improving technologies, the curve shifts “upward” with higher real wages offered at every level of effort.

Starting in the mid-1970s, when relative prices of apparel and footwear and textiles and other labor-intensive goods fell substantially, the offer curve “twists” with wages falling for low-effort contracts but rising for high-effort contracts. In the eighties the curve began shifting “to the right”, with more hours required to attain any given level of earnings. The theory allows this last shift to be due either to the introduction of new machinery (computers) or to a rise in fixed costs other than capital, namely benefits.

The relationship between wages and capital intensity has been discussed in the efficiency wage literature, including papers by Dickens and Katz(1987a), Katz and Summers(I989), and Krueger and Summers(1987). We view the correlation between wages and capital intensity through an entirely different theoretical lens. Empirically, the innovation of this paper is the discovery of a strong correlation between weekly hours and capital intensity.

One surprising feature of the US wage-effort offer curve is that it bends backward in the early years of the sample, with the highest wages offered in sectors with relatively low effort (as measured by hours). The two-digit names of these unusual sectors are transportation, primary metals and printing and publishing. These names cry out to us “union effects.” A union effect on wages can come from market power in the product market since it allows producers to pass on higher costs to consumers who have nowhere else to go.

Beginning in the 1970s consumers did have somewhere else to go: auto imports from Japan and steel imports from a variety of countries. The erosion of union power seems evident in the wage-effort offer curve as the transportation and primary metal sectors increase their hours to conform more closely to the wage-hour contracts offered in other sectors. But our measure of unionization is not able to account completely for this backward bending portion of the wage-effort offer curve. This may be due to the poor quality of the unionization data.