Trade integration, economic geography and productivity: The Indo-Sri Lanka FTA

Although there is much empirical evidence to show that good firms become exporters, the literature is less lucid regarding the benefits of exporting. This paper disentangles the direction of the causality to show that exporting improves firm performance. It uses Indian plant-level data (over 1995-2008) for 330 firms across six product categories, which experienced sharp increases in exports to Sri Lanka, which then became an important destination market for these products. I generate measures of total factor productivity by estimating production functions using plant-level physical output data. To deal with the problem of self-selection bias, I use instrumental variables that predict export status but are uncorrelated with unobserved productivity. As a robustness check, I model the exporting decision explicitly and jointly estimate it with the production function. I follow Levinsohn and Petrin (2003) and use intermediate inputs to deal with the simultaneity problem. I also conduct panel-data regressions at the industry (4-digit NIC) level to estimate the relationship between productivity and measures of international exposure, such as export shares. I also study how firm performance differs with regards to firm location, and model the effects of economic geography variables such as market access and agglomeration. This paper contributes to the empirical literature by measuring the effects of learning-by-exporting, and makes the case that these effects are more significant for firms that enjoy the advantages of geography.