Analysing global value chains using input-output economics : proceed with care

Input-output economics has become a popular tool to analyse the international fragmentation of value chains, especially now that several multi-regional tables that cover large parts of the global economy have become available. It has been argued that these tables, when analysed with the help of the input-output economics toolbox, can provide better insights about global value chains than can be obtained by case studies of individual value chains. We argue that there are several problems related to the aggregated nature of the input-output table that may lead to large distortions and biases in the aggregate picture about global value chains that is obtained by input-output analysis. There are three main sources behind the distortion obtained in static decompositions of value chains: the average nature of value added to output ratios in the tables, the emergence of production cycles in the process of aggregating several value chains into a single table, and the characteristic of the so-called inverse Leontief matrix to even out the value added distribution. We provide an overview of how these distortions work, and argue that under a wide range of circumstances, input-output methods tend to overstate the contribution of the final sector to the value chain. We also show that this bias does not vanish when we compare input-output decompositions at two different points in time.