Tariffs, global value chains, and the incidence of protection: Evidence from US automobiles
This research analyses the impact of tariffs on the US automobile market, highlighting how global value chains mediate economic outcomes. It finds that while vehicle-only tariffs provide modest producer gains, extending them to parts reverses these benefits and doubles consumer losses. Results vary significantly based on firms’ specific input sourcing.
Please login or join for free to read more.
OVERVIEW
Introduction
A central question in the analysis of trade and industrial policy is how protection affects prices, profits, and welfare in markets characterised by market power and globalised production. In many modern industries, firms compete in differentiated-product markets while simultaneously relying on complex global value chains for intermediate inputs. This paper studies these mechanisms in the context of the U.S. automobile market, where almost half of vehicles sold are assembled abroad, and even domestically assembled units rely heavily on imported parts. The percentage of imported parts varies across models and manufacturers, ranging from 25% to 90% in 2024. Consequently, the incidence of protection depends critically on firms’ exposure to foreign parts and how these costs pass through to marginal costs.
Data and descriptive statistics
The researchers constructed a novel model-year level dataset for the U.S. light-duty automobile market covering the period 2015–2024. This dataset combines sales data from S&P Polk, vehicle characteristics from DataOne, and manufacturing data, including assembly location and vehicle foreign content, from the National Highway Traffic Safety Administration (NHTSA). Two notable shifts occurred over the sample period: electric vehicles (EVs) expanded from a negligible share in 2015 to 7.5% by 2024, and SUVs became significantly more prevalent, increasing from 31% of sales to 49% in 2024. By 2024, firms displayed a pronounced bimodal distribution in domestic parts sourcing, reflecting persistent differences in production architectures and supply-chain configurations.
Model
The study develops a structural model of the U.S. automobile market that integrates differentiated-products demand with firm pricing and global input sourcing. On the demand side, a random-coefficients discrete-choice model captures realistic substitution patterns across hundreds of vehicle models and allows for rich heterogeneity in consumer price sensitivity and preferences for vehicle attributes. On the supply side, the framework departs from the common assumption that domestically assembled products are insulated from shocks to foreign input prices. Instead, it allows changes in the cost of imported automotive parts to transmit into manufacturers’ marginal costs, accounting for upstream cost absorption by suppliers and medium-run adjustments in sourcing decisions.
Estimation
Demand parameters were estimated using the generalised method of moments (GMM) by matching simulated outcomes to observed market shares and demographic micro-moments. Identification of the cost channel utilised exchange-rate variation, specifically the lagged real exchange rate between the U.S. and the assembly country. The model was estimated using the PyBLP package, with mean utilities recovered via contraction mapping. The estimation procedure accounts for price endogeneity, as a product’s price is likely correlated with its unobserved quality.
Results
The demand system implies a share-weighted mean own-price elasticity of −6.764, meaning a 1% increase in prices leads to a 6.76% reduction in sales. The supply-side results indicate that for a domestically assembled vehicle that sources 100% of its parts from abroad, a 1% increase in the cost of those parts leads to a 0.715% increase in the vehicle’s marginal cost. This estimate is consistent with external assessments suggesting that approximately 70% to 71% of vehicle value reflects parts costs. The share-weighted average markup across the full sample is 17.2%.
Counterfactuals
The counterfactual analysis yields three core results. First, a 25% tariff on imported vehicles alone reallocates demand toward domestically assembled products, raising U.S. producer surplus by approximately $1 billion but reducing consumer surplus by about $14 billion. Second, extending tariffs to imported parts fundamentally alters these effects: consumer surplus losses roughly double to $32.8 billion, and U.S. producer surplus declines by about $2.58 billion. These aggregate effects mask substantial heterogeneity; firms with lower reliance on foreign inputs, such as Tesla and Jeep, are better able to increase profits, whereas traditional U.S.-headquartered firms like Chevrolet, GMC, and Ford experience losses. Higher-income households account for a substantial share of consumer welfare losses, with the top quintile alone representing about 40%. Third, reinstating EV subsidies generates gains, with consumer surplus increasing by approximately $4.8 billion and U.S. producer surplus rising by $1.8 billion.
Conclusion
The results demonstrate that tariff incidence depends critically on firms’ exposure to global value chains and cannot be inferred from final assembly locations alone. Policies that appear protective when focusing only on final goods trade can impose substantial costs when intermediate inputs are taken into account. This highlights the importance of explicit attention to global value chains and the interaction between cost exposure and market structure when evaluating trade policy in modern industrial sectors.