Errors from the 'Proportionality Assumption' in the Measurement of Offshoring : Application to German Labor Demand

Offshoring – the importing of intermediate materials and services – has expanded rapidly in most industrialized countries and its impact on the labor markets in these countries has been the source of enormous debate in both scholarly and popular circles. Since data on imported inputs at the sectoral level are not available for the US and the UK, empirical research has relied entirely on a proxy-based measure offshoring, using what the OECD refers to as the “proportionality assumption.” That is, every sector is assumed to import inputs of each material and service in the same proportion as its economy-wide use of that input.German input-output data differentiate between domestically purchased inputs and imported inputs, which permits us to calculate a direct measure of sectoral imported input use. In this paper, we compare this measure to the proxy-based measure based on the standard proportionality assumption. We find that the direct measure differs significantly from the proxy-based measure for both services and materials offshoring. To assess the significance of using different measures, we substitute them for each other in standard labor demand equations focusing on German manufacturing between 1995 and 2004. We find that using the direct measure of offshoring gives very different results for labor demand – sometimes of opposite sign – compared to estimates using the proxy-based measure.We perform a simple decomposition of the proxy-based measure and find that it fails to accurately capture the cross-sectoral variation in offshoring intensity because – as a result of the proportionality assumption – it is heavily influenced by the cross-sectoral variation in domestic input demand. The implications of our findings go beyond the case of Germany. They indicate that researchers must be cautious about drawing policy conclusions from estimates using the proxy-based measure of offshoring