U.S. Job Flows and the China Shock Appendix B — For Online Publication

This section looks at manufacturing employment responses to Chinese import exposure. Table B.1 replicates Table 2 of AADHP and presents a stacked first-difference model for the two time periods 1992-99 and 1999-2011 with the change in import penetration and a dummy for each time period as the regressors, along with results for sub-periods 1992-99, 1999-2011 and 1999-2007. We find that rising import penetration lowers domestic employment under all specifications. OLS and IV estimation give similar sign and statistical significance but the IV coefficient is about 1.7 times larger than the OLS coefficient in column 3 with CBP data and 1.8 times larger with NETS data. The estimate in column 3, where import penetration is instrumented with Chinese import penetration into other advanced economies, implies that a 1 percentage point rise in industry import penetration reduces domestic industry employment by 1.33 percentage points. This decrease is 0.47 percentage points when using NETS data and identical for the period 1992-2007, showing that results are very similar if we restrict attention to the years preceding the Great Recession. Deaths of firms in these periods seem to be driving this result. The coefficients on births, expansions and contractions are all very small and statistically insignificant. For column 3, the estimated share of deaths in total Chinese-induced job reallocation is 0.76. As with AADHP, the largest effect in abso-