The “China Shock” Revisited

Abstract

We exploit a decomposition of gross trade flows into their value added components to reassess the relationship between increased imports from China and manufacturing jobs in US local labour markets following the seminal paper of Autor, Dorn, and Hanson (2013, ADH). Decomposed trade flows enable us to address identification and measurement issues inherent to gross trade data. In particular, it allows us to remove US value added in Chinese exports from the exposure measure which is mechanically correlated with the dependent variable and overstates the volume of the trade shock. In addition, the decomposition permits to correct for double counting, to remove primary and services inputs in manufacturing exports, and to assign competition to the upstream industry that supplied the value added rather than the final exporting industry. This further reduces the volume of the shock and improves the accuracy of the import exposure measure. Consequently, we find considerable differences in the pattern of regions that are most affected by the trade shock and show that imports from China can explain less of the decline in US manufacturing than what gross trade data would suggest. We then separate the shock into a China-driven domestic reform and a thirdcountry-driven value chain component, and find in line with ADH that the smaller, but still negative labour market effects are indeed China driven. Finally, we observe that the negative effects identified in ADH are not present in the 2008-2014 period, as labour market adjustment has largely concluded. The long time needed for adjustment may have been prolonged by the evolution of China's comparative advantage.