Bias in the ‘Proportionality Assumption’ Used in the Measurement of Offshoring
& William Milberg
Published: December 2012
Most studies of offshoring rely on a ‘proportionality assumption’ where every sector is assumed to import each material and service input in the same proportion as its economy-wide use. We assess the bias resulting from this assumption. Since Germany collects imported inputs directly, we are able to compare the direct and proxy measures, where the proxy is constructed with the proportionality assumption. The proxy fails to accurately capture the variation in services offshoring intensity because – as a result of the proportionality assumption – it is strongly influenced by the variation in demand for domestic inputs. Estimation of the effect of offshoring on labour demand for 35 manufacturing sectors in Germany over 1995–2006 shows that the direct and proxy-based measures of services offshoring give very different results. The implication goes beyond the case of Germany: researchers must be cautious about drawing policy conclusions from estimates using the proxy of offshoring.