اشتغال، ساختار مهارت و تجارت بین المللی : شواهد در سطح بنگاه فرانسه

We use the French Customs files, an exhaustive account of the international trade transactions carried out by firms across the period 1986–1992, to analyze the link between imports, exports, employment, and skill structure of French manufacturing firms. Our data allow us to distinguish between imports of finished goods and imports of intermediate inputs. Our results show that there is a strong correlation between increasing imports, in particular imports of finished goods, and job destruction, most notably destruction of production jobs. Interestingly, the strength of the relation between job destruction and imports is stronger for larger firms. For example, within production jobs, the association between increasing imports of finished goods and destruction of unskilled jobs is only found in large firms. These findings are robust to the introduction of firm-level measures of innovation.

Krugman hypothesized that wage inequality in the United States and unemployment in Western Europe were “the two sides of the same coin”. There is a very large literature that examines the American side of the coin (Bound and Johnson, 1992; Berman, Bound and Griliches, 1994; Revenga, 1992 and Katz and Murphy, 1992). Its conclusion is that the demand for unskilled labor decreased in virtually every industry, interpreting these within-industry evolutions as evidence of skilled-biased technical change. Bernard and Jensen (1997) (BJ, hereafter) using firm-level measures of wages, skills, employment as well as exports, and in contrast to most previous analysts, find that changes in product demand were key in understanding these phenomena. Their analysis confirmed that exporting plants played a key role during this period of rapid change. This paper examines the other side of the coin by investigating the relation between trade and employment in France, taking stock of the decrease in wage inequality in this country up to 1984 and its stability afterwards (see Buchinsky et al., 2003 for a description of these trends from 1967 to 1999).2 We also use firm-level micro-data sources for this purpose. Our analysis distinguishes itself from that of Bernard and Jensen not only because its period is slightly more recent (1986 to 1992) or because we study France. The main difference is our use of French Customs data in which all flows of goods — imports and exports — are recorded. This administrative database also contains the origin or the destination as well as the product that is imported or exported. Of course, a thorough examination of the relationship between trade, most particularly imports, and employment is of potential academic interest given this debate. But, our results should also allow us to understand if the fears regularly expressed about globalization by the popular press, the unions, or some politicians, and not only in Europe, have any empirical support.

We have related total employment growth, changes in the proportion of production workers in total employment, and changes in the proportion of unskilled production workers in total production labor, to changes in trade variables, namely imports and exports of firms. We find evidence that firms importing finished goods (FG) always destroy more employment than firms importing only intermediary inputs (II) conditional on changes in local purchases. The first type of imports may reflect outsourcing strategies. Imports from low-wage countries have a slightly more negative association than average imports, but the difference is minor. Exports are positively associated with employment growth. Turning to skills, we find the usual result that most changes occur within firms, a fact often interpreted as evidence of skill-biased technical change. Using a regression framework, we find that increased FG imports have a negative association with production labor, as well as unskilled labor (at least in the largest firms). Controlling for innovations taking place at the firm does not alter any of our conclusions. Our approach is purely descriptive. To go a step further and identify causal effects, two directions can be taken. In one, valid instruments for firms' imports and exports have to be found. A potential route is the use of variations in exchange rates. In the second direction, the estimation of structural models (as those of Eaton and Kortum, 2002 or Melitz, 2003), should help identify the effects of trade on employment.