Working Paper: NBER ID: w28073
Authors: Matilde Bombardini; Bingjing Li; Francesco Trebbi
Abstract: In the two decades straddling China's WTO accession, the China Shock, i.e. the rapid trade integration of China in the early 2000's, has had a profound economic impact across U.S. regions. It is now both an internationally litigated issue and the casus belli for a global trade war. Were its consequences unexpected? Did U.S. politicians have imperfect information about the extent of China Shock's repercussions in their district at the time when they voted on China's Normal Trade Relations status? Or did they have accurate expectations, yet placed a relatively low weight on the subconstituencies that ended up being adversely affected? Information sets, expectations, and preferences of politicians are fundamental, but unobserved determinants of their policy choices. We apply a moment inequality approach designed to deliver unbiased estimates under weak informational assumptions on the information sets of members of Congress. This methodology offers a robust way to test hypotheses about the expectations of politicians at the time of their vote. Employing repeated roll call votes in the U.S. House of Representatives on China's Normal Trade Relations status, we formally test what information politicians had at the time of their decision and consistently estimate the weights that constituent interests, ideology, and other factors had in congressional votes. We show how assuming perfect foresight of the shocks biases the role of constituent interests and how standard proxies to modeling politician's expectations bias the estimation. We cannot reject that politicians could predict the initial China Shock in the early 1990's, but not around 2000, when China started entering new sectors, and find a moderate role of constituent interests, compared to ideology. Overall, U.S. legislators appear to have had accurate information on the China Shock, but did not place substantial weight on its adverse consequences.
Keywords: China shock; trade policy; US Congress; NTR status
JEL Codes: F13; P16
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| US legislators had accurate information on the China shock (F69) | US legislators did not give substantial weight to its adverse consequences (K37) |
| A one interquartile difference in the value of the shock (C22) | decreases the probability of voting in favor of NTR for China (D72) |
| A one interquartile difference in ideology (D79) | increases the probability of supporting NTR (C34) |
| The information available to politicians (D72) | allowed them to forecast about 55% of the variation in the China shock from 1990 to 2001 (F17) |
| Democrats being more informed than Republicans (D72) | affects their voting behavior regarding NTR (D72) |
| Perfect information about upcoming shocks (D84) | would not have changed overall support for China’s NTR status substantially (F69) |