In real-world settings, a firm's application for industrial policy incentives must be approved by multiple actors, including politicians, bureaucrats, and firms themselves (through self-selection). Who ultimately picks winners, whom they favor, and how favoritism interacts with incentive design remain open questions. Using confidential data on industrial subsidy applications and decisions in a large Indian state, we document several stylized facts. First, most variation in winning subsidies is conditional on final bureaucrat approval. While bureaucrats approve over 90% of applications, less than 30% of approved subsidies are paid, with an average delay of 3.5 years among winners. Second, winners are actively chosen at the payment stage: each subsidy release order covers either a single high-profile plant or an industrial cluster. Firm bargaining power (proxied by size and in-state headquarters) predicts earlier payments. Third, payments support struggling firms: in a shift-share design based on firms' pre-pandemic product mixes, those facing larger negative demand shocks in 2020 are more likely to receive payments in 2023 for investments made years earlier. Favoritism is more pronounced for subsidies on variable inputs (e.g., sales tax and electricity), which can be filed years after the eligible investment. These results underscore the challenge of insulating industrial policy from political influence, as constrained funds controlled by politicians lead to favoritism in a black box, years after investments.
We study the benefits of economic integration from reducing policy-induced barriers to trade. A landmark 2017 fiscal reform in India substantially reduced barriers to crossing internal state borders. Using the reform as a natural experiment and aggregate data on trade flows, we estimate gravity regressions and find that each additional border in a shipping route reduces trade by 15%. Calibrating a quantitative trade model to this elasticity, we find that reducing all such border frictions would increase GDP by 3%. To examine how supply chains may have reorganized, and the implications this has for gains from trade, we intend to exploit detailed micro-level data which we constructed from the universe of VAT records in India.
We examine the impact of trade liberalization on structural change patterns in India. Leveraging district-level variations in sectoral composition, we find that districts with greater tariff reductions experienced larger declines in manufacturing employment shares. By extending Matsuyama’s 1992 model of deindustrialization to include a non-tradable service sector, we demonstrate analytically and through simulations that India's observed deindustrialization and service-led growth can be qualitatively attributed to trade liberalization. We aim to structurally estimate the model parameters to quantify the role of trade liberalization in driving these structural changes.