We model a multi-stage supply chain for EVs from battery production to vehicle distribution. Given industrial policies, firms select where to open facilities at each stage. This is a difficult combinatorial choice problem that we solve with a fast mixed integer linear programming formulation. We estimate the variable and fixed costs parameters using SMM. Counterfactual simulations reveal a tension between boosting EV adoption and promoting domestic supply chains. Due to increasing returns, even unconditional subsidies raise the number of factories in the subsidizing region—by about 16% for EVs and 7% for cells in North America, and even more in Europe. Theoretically, local assembly requirements can push down delivered marginal costs relative to unconditional subsidies. Empirically, local content requirements quadruple the expansion of cell factories in America, but they drive up costs and reduce subsidy uptake, undoing more than half of the EV adoption stimulus coming from pure buyer subsidies.

This paper studies spatial interdependencies in multi-plant production. I develop a model in which each firm decides locations of plant set and variable markups where its plants sell. Such location decision is guided by two competing forces: amplification of a firm’s competitive advantage through plant expansion and diminishing marginal benefits due to cannibalization. Despite a combinatorial discrete choice problem, the model is estimated efficiently provided the location game is submodular and aggregative. With this framework, I investigate the spatial organization of cement firms responding to environmental policy changes, and show that neglecting the interdependencies biases the estimate of carbon leakage.

Recent decades have witnessed the growing importance of trade in intermediate goods and pursuit of free trade agreements (FTAs). They distort firms’ sourcing decisions internationally through preferential tariffs and rules of origin (RoOs), a set of criteria that define the origin of a product to qualify for preferential access. The paper distinguishes trade diversion through RoOs from tariff reduction on intermediate goods, focusing on the automotive industry. Car assemblers’ decisions of how much to acquire from which supplier are modeled for every auto part. With the derived gravity trade equation, the estimation identifies significant diversion in intermediate sourcing and the effect is nonlinear with respect to the restrictiveness of RoOs. The shift from foreign to regional inputs exhibits a humped shape and it peaks when the required minimum FTA content is between 50% and 60%. Impacts of RoOs are further decomposed to four channels: export destinations of final goods, magnitude of preferential treatments, price and cost penalty of intermediate goods. Results show that the RoO effects are stronger when car exports are mainly intra-FTA. However, there are mixed forces driving the effects of preferential tariff margins of cars and values of parts.