Working Paper: NBER ID: w22804
Authors: Matilde Bombardini; Bingjing Li
Abstract: Has the expansion in exports affected pollution and health outcomes across different prefectures in China in the two decades between 1990 and 2010? We exploit variation in the initial industrial composition to gauge the effect of export expansion due to the decline in tariffs faced by Chinese exporters. We construct two export shocks at the prefecture level: (i) PollutionExportShock represents the pollution content of export expansion and is measured in pounds of pollutants per worker; (ii) ExportShock measures export expansion in dollars per worker. The two measures differ because prefectures specialize in different products: while two prefectures may experience the same shock in dollar terms, the one specializing in the dirty sector has a larger PollutionExportShock. We instrument export shocks using the change in tariffs faced by Chinese producers exporting to the rest of the world. We find that the pollution content of export affected pollution and mortality. A one standard deviation increase in PollutionExportShock increases infant mortality by 2.2 deaths per thousand live births, which is about 13% of the standard deviation of infant mortality change during the period. The dollar value of export expansion tends to reduce mortality, but is not always statistically significant. We show that the channel through which exports affect mortality is pollution concentration: a one standard deviation increase in PollutionExportShock increases SO2 concentration by 5.4 micrograms per cubic meter (the average is around 60). We find a negative, but insignificant effect on pollution of the dollar-value export shocks, a potential “technique” effect whereby higher income drives demand for clean environment. We find that only infant mortality related to cardio-respiratory conditions responds to exports shocks, while deaths due to accidents and other causes are not affected.
Keywords: trade; pollution; mortality; China
JEL Codes: F1; I1; Q53
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| pollution-export shock (F64) | infant mortality (J13) |
| pollution-export shock (F64) | SO2 concentration (Q53) |
| SO2 concentration (Q53) | infant mortality (J13) |
| export shock (Y60) | infant mortality (J13) |