Kai Wu

Abstract: This paper studies how regulatory shocks propagate through production networks and who bears the burden. I exploit a natural experiment—the 2017 Xiong'an pollution shutdown campaign in China, which imposed mandatory closures on polluting industrial operations across a newly designated development zone—combined with administrative VAT data covering the universe of firm-to-firm transactions. Targeted firms' purchases fell by 55% and sales by 53%. The shock propagated in both directions: upstream suppliers at mean exposure lost 11.8% of sales; downstream customers saw purchases drop by 9.5%. Effects extend to second-degree trade partners and persist for at least three years. Small firms bear the entire propagation loss while large firms are completely buffered. The mechanism is two-sided complementarity: for small firms, trading partners are gross complements on both the input and output sides, so that losing one partner forces contraction with all others; for large firms, partners are gross substitutes. A general equilibrium model of shock transmission in production networks with scale economies and size-varying elasticities formalizes these patterns, showing that ignoring supply chain spillovers leads to overregulation and that network externalities create novel channels through which emissions taxes dominate quantity controls.

Crowding is a common disamenity that exists in many settings, but little is known about its monetary cost. This paper estimates the willingness-to-pay (WTP) to avoid crowding in public transportation with a revealed preference framework. We leverage an off-peak pricing discontinuity in the Beijing Subway that generates exogenous temporal variation in price. Passengers traveling between a pair of stations choose the optimal departure time, trading off between the price, the expected level of crowding, and the deviation from the ideal time of travel. We develop a novel approach to allocate passengers to trains and calculate real-time crowding. To address the endogeneity in crowding, we construct an instrumental variable based on the number of overlapping trips that start from and end in different stations, and are thus driven by plausibly unrelated demand shocks. We estimate the marginal WTP to reduce in-train crowding by one passenger per square meter to be about 40% of the average fare for a typical 40-minute ride. With average crowding of 3.4 persons per square meter, a trip generates a crowding externality of roughly 140% of the fare. An optimal crowding tax to address the externality raises welfare but disproportionately harms low-income passengers; in contrast, a two-class configuration improves welfare for both income groups by soliciting self-selection.

I investigate the impact of political consolidation on the corporate tax burden, leveraging the variation from a nationwide reform in China that shifted governance autonomy from annexed to annexing political units. Using administrative firm-level tax payment data from 2008 to 2015 and a heterogeneous difference-in-differences approach, I find divergent shifts in the effective corporate income tax rate among affected areas following the annexation. On average, annexing areas experienced a 1.8% decrease in the effective tax rate, contrasted with a 1.4% increase in annexed areas. These changes worked to make corporate tax rates more uniform across annexed and annexing areas. Also, I find the observed effects are largely attributed to the adjustment in less transparent tax break programs, over which local governments possess greater discretion, underscoring the pivotal role of local taxation rearrangement along with the annexation. Heterogeneity analysis suggests political favoritism and tax competition are additional mechanisms contributing to these findings.