Can we really trust offshoring indices?

This paper argues that indices of (business) service and material offshoring built on sectoral input-output data may actually measure something different than what we think they should. Applying shift-share analysis we decompose the variation over time of a commonly used class of such indices into two components: one related to the intensity in the import of intermediate inputs, and the other associated with the use of such inputs in the production of manufacturing goods. Using data from input-output tables of 21 European countries from 1995 to 2006, we show that in the case of service offshoring, in most countries a larger part of the variance is driven by the raising share of (domestically produced) services used in manufacturing production, while the share of imported services contributes to a much smaller extent. When we focus on the subset of business services, evidence shows a relatively larger tendency towards relying on imported rather domestically produced inputs. Instead, in the case of material offshoring there is evidence that foreign suppliers have substituted domestic ones. However, this pattern is strongest in countries, such as Estonia, Hungary and Slovenia, where incoming multinationals, rather than domestic firms offshoring production may be the driving force.