Working Paper

"Attribute-Based Subsidies and Market Power: An Application to Electric Vehicles" Revise & Resubmit, Journal of Political Economy, covered by VoxChina 
with Panle Barwick and Shanjun Li

• • Attribute-based subsidies are commonly used to promote the diffusion of energy-efficient products which are often manufactured in industries with significant market power. To understand the impacts of these subsidies, we first develop a theoretical framework of attribute-based subsidies in the presence of market power. The model illustrates that the welfare impact of subsidies critically hinges on firms' product attribute choices and their implications for environmental externalities and market power. We then develop and estimate an equilibrium model with endogenous product attributes using comprehensive data on China's vehicle market. Based on the model estimates, we conduct counterfactual simulations to examine the impacts of different subsidy designs. Relative to attribute-based subsidies, uniform subsidies favor small and environmental-friendly vehicles but exacerbate the quantity distortion from market power for high-quality products. In contrast, subsidies based on the driving range and battery capacity or vehicle weight generate a large consumer surplus by improving product quality and mitigating market power. Capacity-based subsidies induce attributes valued by consumers, mitigate market power, and lead to the largest welfare gain at a moderate loss of environmental benefit. The findings highlight the importance of incorporating attribute choice and market power considerations in designing attributes-based policies.

"Learning by Doing in the Global Electric Vehicle Battery Industry and Implications for Government Policies" Revise & Resubmit, American Economic
        Review, covered by BFI Briefing, VoxChina, and EPIC Insight
        with Panle Barwick, Shanjun Li, and Nahim Zahur

• The global battery industry has achieved significant cost savings: electric vehicle (EV) battery costs have dropped by more than 90% over the past decade. This study assesses the extent to which this sharp decline in lithium-ion battery prices is attributable to learning-by-doing, and quantifies the impact of government policies (subsidies and local protection policies) on learning. Our analysis is based on a rich dataset consisting of 1) model-level sales, prices, and attributes of EVs and 2) battery suppliers and characteristics, for 13 countries from 2013 to 2020, that together account for over 95% of global EV sales. We estimate a structural model of the global EV industry, accounting for consumer vehicle choices, EV makers' pricing decisions, and bilateral bargaining between EV manufacturers and battery suppliers over battery prices. We recover the model-implied battery costs and evaluate how battery costs change with the accumulated production experience of battery suppliers, finding that learning-by-doing explains a substantial portion of the observed battery cost reductions. Lastly, we use the estimated model to examine the strategies of the US and China to protect their domestic battery industries: both countries require automakers to use domestically produced batteries to receive EV subsidies. We simulate equilibrium outcomes without such restrictions to evaluate the impact of local protectionism on cost-saving innovations in the global battery industry.

"Subsidies versus Tradable Credits for Electric Vehicles: The Role of Market Power in the Credit Market"  ( Job Market Paper )

• • With the wider adoption of electric vehicles (EVs), policymakers across countries are considering switching from subsidies to tradable emission credit systems. This policy change would eliminate subsidy expenditures while introducing a credit market that could exhibit two-sided market power. Traditional automakers in need of credits form an oligopsony on the credit demand side, whereas EV makers create an oligopoly on the credit supply side. This paper first provides a theoretical framework to demonstrate how imperfect competition in the credit market affects equilibrium outcomes. According to the model, firms leverage their market power in the output market to manipulate the credit price. This incentive encourages them to raise the prices of their primary fuel-type vehicles, exacerbating the market power problem in the output market. To empirically validate these theoretical predictions, this study then estimates a structural model that simultaneously accounts for equilibrium product portfolios, vehicle prices, and the credit price, with data from forty major cities in China. Counterfactual simulations separately evaluate the welfare impact of 1) removing government spending and 2) introducing a credit market. Withdrawal of subsidy expenditures raises the effective marginal costs of vehicle production, lowering total welfare by 7.4 billion RMB per quarter, amounting to 13.9% of the total subsidy expenditure. In addition, firms' strategic behaviors in the credit market reduce consumer surplus by 2.2 billion RMB per quarter, and this reduction dominates the increase in profits and environmental cost savings, generating a further welfare loss.

"Equitable Energy Transitions? The Efficiency and Distributional Effects of Subsidies for Used Electric Vehicles" 
        with Hunt Allcott, Tess Snyder, and Levi Kiefer

• • We study the efficiency and distributional effects of the Inflation Reduction Act (IRA) tax credits for purchasing used electric vehicles (EVs), which aimed to address concerns that new EV tax credits primarily benefit higher-income buyers. We show theoretically that under certain conditions, tax credits for new versus used EVs have the \textit{same} economic incidence because they interact through used EV resale values. However, using confidential dealership transaction data, we find that used EV prices increased by only a limited amount after the IRA was enacted and the tax credits became available, suggesting that the initial economic incidence fell primarily on EV buyers who were eligible for the credit. Bunching of transaction prices below the credit's \$25,000 price threshold increased markedly in 2024 when buyers could immediately receive the credit amount as a cash rebate. We then assess the long-run welfare effects of EV tax credits using a novel non-stationary dynamic structural model of new and used vehicle markets.