Revenga (1992), using a sample of three and four digit manufacturing industries over the 1977 to 1987 period, found that exchange rates have significant implications for employment in the United States, and smaller but still significant effects on wages. The estimates came from a sub-sample of manufacturing industries and focused on the effects of import competition into the United States: the higher the import share of an industry, the more an import price decline or dollar appreciation hurt domestic labor markets.
Instead, using 2-digit industry data for all of manufacturing for a longer sample period -the early 1970s through mid -1990s– we find that industry wages are considerably more responsive than jobs (and hours worked) to exchange rate movements. We also show how this wage responsiveness has been growing over time as U.S. industries become more export oriented. The growth of imported input use by producers provides some offset to the pressures from the export channel partially offsetting its overall effect. More significant wage effects are apparent in industries that have lower price-over-cost markups and with relatively less-skilled workforces further.
These industries are closer to perfect competition and have production biased toward durable goods. Patterns of significant employment responsiveness are different: these employment effects are concentrated in a much smaller group of industries. We also consider labor market adjustment via another margin, through industry use of and payment for overtime activity. We find a qualitatively important role of currency movements. Changes in the U.S dollar significantly affect overtime wages and overtime hours, especially in high markup industries and industries with more skill-intensive production. Industry unionization rates and capital intensity uncorrelated with industry rankings of the elasticity of wage, employment or overtime activity response to the exchange rate.
Our conclusions on wages are quantitatively similar to those reported by Revenga (1992) while our employment estimates differ with both the early Branson and Love study and the more recent findings by Revenga. Differences in the estimation intervals across the studies do not explain these differences.2 Two other differences in approach, one methodological and one related to data, may explain the starkly different findings on employment response across this group of related papers.
On the methodology front, our work focuses on the different channels of exchange rate exposure of an industry while the previous studies focus on one particular channel, import competition. For example, Revenga’s motivation and sample of industries are driven by the goal of estimating the effects that import competition in the United States had on U.S. labor markets. Instead, we stress the importance of decomposing the sources of currency exposure of an industry into its three major components. An industry exposure to exchange rates can arise from its reliance on export revenues, imported inputs into production, and, as highlighted in the previous work, from import competition.