Our regressions examine the effect of exchange rates entered alone in the regressions, which has been the standard in this literature, and also examine the effects of exchange rates interacted with the industry-specific and time-varying channels of trade. While our theory spelled out three distinct channels (exports, import competition and imported inputs) ultimately we use only two interacted channels in the regressions: (i) the export to production share in the industry, and (ii) the share of imported inputs into production costs (Campa and Goldberg 1997). We limited ourselves to these two channels because of the high within-industry correlations between import penetration and imported input use.
The regressions also include the prices of two other inputs: capital and energy. We use the long-term interest rate, measured by the yield in long-term U.S. Government bonds, as a measure of the cost of capital. Real oil prices, measured by the average annual dollar price per barrel of crude petroleum reported by the International Monetary Fund, are used as our measure of energy prices. The net effect for labor demand of an increase in either of these two prices depends on the substitutability versus complementarities of each of these factors with labor. To the extent that technological substitution between that input and labor is possible, an increase in the input price leads to an increase in the demand for labor. To the extent that the factors are complements, the sign of this correlation will be negative.
Our measures of labor market activity are central to the empirical work. The model of labor market equilibrium in Section II did not specify the unit of observation for labor. Labor was a continuous input that could be adjusted by infinitesimal amounts as needed. Empirically, labor activity during a certain period can be measured by several proxies, such as the average number of employees or as the total number of hours worked by employees (including overtime effort). As surveyed by Hammermesh (1993), a number of factors determine whether a producer’s response to stimuli is through hiring (firing) new workers or through an increase (decrease) in the number of hours that the existing staff work.
These factors include: the nature of the shock (transitory versus permanent)15; the industry costs of hiring/ firing versus expanding/ contracting work effort; and the types of contracts signed with workers. The tendency toward the use of overtime employment instead of changes in the number of employees is expected to be higher when the shocks are temporary, hiring and firing costs are high, and the labor pool is more skilled or requires more job-specific skills.
Our analysis looks at the endogeneity of both employment and total hours worked in each 2-digit manufacturing industry, as well as at industry wages and measures of industry overtime activity. The employment series is the total number of non-farm employees in the industry, as reported by the Bureau of Labor Statistics.