This is the ninth in this year’s series of posts by PhD students on the job market
Over the past two decades, trade in ‘dirty goods’, goods that harm the environment, such as used vehicles, and waste and scrap, has been rising dramatically. Most of them flow to developing economies for the purpose of consumption or recycling, but at the same time creating substantial environmental consequences. To address environmental concerns, some countries and regions impose trade restrictions on such goods. However, the effectiveness and efficiency of those policies remain unclear. Do trade restrictions protect the local environment? Do they improve welfare? Are decentralized policies socially efficient? My job market paper answers these questions in the context of intercity import restrictions on used vehicles in China.
The empirical setting
Vehicle tailpipe emissions are regulated under emission standards which set the limit of pollutants allowed in exhaust gases released from an internal combustion engine. China introduced China 1, the first set of standards, in 2001, and tightened the standards over time from China 1 to China 6. Auto makers need to produce new vehicles that comply with the effective standard. Therefore, older vehicles are dirtier because they subject to less stringent standards, and also because tailpipe emissions increase as vehicles age.
In 2008, Beijing started to use the emission standard to restrict imports of used vehicles from other cities. Residents in Beijing were prohibited from buying used vehicles below China 3 from other cities but free to purchase used vehicles regardless of emission standards within the local market. This policy, an emission-based import restriction in essence, was quickly followed by many other cities. By the end of 2015, almost 95% of cities in China had adopted the import restriction on used vehicles, most of which used China 4 as the minimum standard for imports.
How does the import restriction affect trade flows of used vehicles and air quality?
Based on the universe of used vehicle registration data from 2013 to 2015, I compiled bilateral trade flows of used vehicles across 312 cities (prefecture-level divisions) in China at the quarterly level. I collected data on the timing and stringency of import restrictions on used vehicles by city from government websites, news media, etc. Out of the 312 cities, 95 had adopted the restriction before 2013, 207 adopted the restriction during 2013-2015, and 10 remained unrestricted. I also obtained ambient air quality measures of the aerosol optical depth (AOD) which strongly correlate with PM2.5, and aggregated the data to city-quarter level from 2013 to 2015.
I use the doubly-robust DID estimator, which is robust to the ‘negative-weighting issue’ in the standard two-way fixed effects estimator, to estimate the effect of adopting import restrictions on net imports of used vehicles. The estimates of dynamic effects do not show much of pre-trends (Figure 1), alleviating the concern that cities might adopt the restrictions because of fears that imports are increasing suddenly. On average, one city’s net imports of restricted used vehicles decreased by 409 units in the next year, which is about 0.1% of the mean motor vehicle stock by city. The decrease in net imports does not lead to a significant decrease in ambient air quality in the next year.
Figure 1: Treatment effects of adopting import restrictions on net imports of used vehicles
Do cities strategically interact in adopting the import restriction?
Locally enforced policies could be strategically interdependent when local policies generate spillovers to other jurisdictions. To test whether one city’s restriction adoption depends on other cities’ restrictions, I regress one city’s restriction dummy on lagged weighted average of other cities’ restriction dummies. The weights are constructed as the share of used vehicle trade between the adopting city and every other city, over the total trade of used vehicles of the adopting city in the first quarter of 2013. Higher weights are given to more important partners in the trade of used vehicles. The results show that cities’ import restrictions are strategic complements --- one city is 96 percentage points more likely to restrict if all the other cities have adopted such restrictions one quarter before.
A structural model that features trade flows of vehicles
To quantify the welfare effect of import restrictions, I develop a multi-sector multi-region general equilibrium trade model à la Armington (1969) featuring the bilateral trade of vehicles across locations. Demands for vehicles are characterized by a nested utility function that aggregates vehicle consumptions from different locations and different sectors. Supply of vehicles are characterized by a Cobb-Douglas production function that combines labor and a composite intermediate input. Through market clearing conditions, general equilibrium prices and trade flows of vehicles could be solved.
I use the model to simulate four policy counterfactuals: (1) unilaterally optimal restrictions in which each city chooses an emission-based import restriction to maximize its own welfare (gains from trade net of environmental costs from vehicle lifetime emissions); (2) non-cooperative Nash equilibrium in which all cities simultaneously choose their unilaterally optimal restriction, taking other cities' restrictions as given; (3) nationally optimal import restrictions in which a social planner decides which cities should restrict imports of used vehicles at what stringency to maximize the total social welfare; and (4) emission taxes levied on both imported and locally traded used vehicles based on lifetime emission damages. I use ‘exact hat algebra’ to solve for changes in equilibrium outcomes under different counterfactuals and quantify welfare effects.
I find that unilaterally restricting imports of used vehicles have two effects on welfare. It reduces gains from trade but can increase environmental benefits through two channels: reducing net imports and increasing scrappage. Restricting high-polluting vehicles makes some cities better off. However, these unilateral restrictions generate significant spillovers onto other cities and incentivize cities to adopt or tighten restrictions in response. The decentralized equilibrium increases social welfare by 3.2 billion RMB relative to the status quo as of 2013, but it is socially inefficient. A centralized policy to restrict used vehicle imports in fewer cities with less stringent standards could increase national welfare gains by 26% relative to decentralized equilibrium. To put the centralized optimal import restrictions into perspective, they can achieve 14% of emission reductions and 12% of welfare gains attainable under emission taxes.
Who gains and who loses?
Under import restrictions, the welfare effects are heterogenous across cities. The poorest third cities (by GDP per capita) gain the most, the middle third gain modest, and the richest third lose. This is mainly driven by the pattern of used vehicle trade --- most used vehicles flow from higher income cities to lower income cities. Import restrictions reduce net inflows of used vehicles, thus generating environmental benefits to lower income cities but environmental costs to higher income cities. These effects could be reinforced or (partially) offset by the cross-boundary pollution transmission, which depends on wind directions and other meteorological conditions.
Implications for policy
Could countries or regions use trade restrictions on dirty goods to protect local environment? My paper uses the example of import restrictions on used vehicles to show that trade restrictions can be used as a second-best environmental policy that brings in environmental benefits and increases welfare, especially for lower income economies. In addition, interjurisdictional spillovers incentive cities to adopt more restrictions. Although it could overshoot the socially optimal restriction, it helps to achieve environmental goals when a centralized policy is not feasible due to political economy reasons. However, the effectiveness of trade restrictions can be limited, as it focuses on trade flows but does not directly target the (much larger) local existing vehicle fleet.
Hui Zhou is a PhD student at Cornell University.