The Unequal Effects of Trade Liberalization: Theory and Evidence from India

This paper focuses on the effects of a recent episode of trade liberalization in India. India underwent a massive reform in 1991 which involved slashing tariffs and opening up different industrial sectors to foreign direct investment. This episode, which is described in detail in a section of the paper, represents, for its size and impact, an attractive experiment to assess the validity of the theory. More precisely, we construct a three-dimensional panel for the period 1970 to 1996 using "Annual Survey of Industries" (ASI) data with variations over 3-digit industry, state and time. The available data include value added, capital, labor and profits for each industry-state-time observation. We use a measure of output per worker in the period just before liberalization (relative to the most productive state-industry observation in the same year) as a proxy of the distance to frontier for a particular 3-digit industry. We interact this variable with a reform measure which is zero before 1991 and takes on a value of one thereafter to test whether the distance to frontier prior to reform influences the post-reform performance. We first document the effects of the reform on labor productivity, and then consider separately the effects on total factor productivity, profitability employment and investments. This provides a test of the first prediction of the theory, i.e., that firms closer to the frontier respond more to the threat of entry introduced by liberalization. Second, we consider state-specific labor market regulations. To this aim, we use a measure of the direction of labor regulation constructed by Besley and Burgess (2002), which coded state amendments to the Industrial Dispute Act as pro-labor, neutral or pro-capital. The level of this variable in 1991, which captures the relative bargaining powers of workers and employers at the moment of the reform, is then interacted with the dummy for the reform and used as an explanatory variable for separate regressions for labor productivity, TFP etc.. The estimated coefficient on this interaction term provides an inference on whether labor regulations mattered in determining the effect of the reform on state industrial performances. Finally, we consider the three way interaction between distance to frontier, labor regulation and reform to examine whether the impact of regulation on industrial performance is larger for industries that are closer to the world frontier. The regression analysis show that (i) state-industries that are closer to the technological frontier experienced larger increases in real manufacturing output, labor productivity, TFP and rents (profits) in the post-reform period. (ii) pro-worker labor regulations have a negative effect on the growth of the same variables, and this effect is strengthened by liberalization. Both results hold true after controlling for state, industry and state-industry fixed effects plus time dummies. Robustness checks include adding controls for state dummies interacted with the dummy for reform to show that the results are due, as the theory suggests, to variations within industries. Also, adding industry-specific time trends to account for different convergence patterns across industries does not change the results.