Working Papers
Do Firms Substitute Across Domestic and Foreign Sources? Micro Evidence and Macro Implications (Job Market Paper )
Midwest Macro (scheduled), Southern Economic Association (scheduled), Midwest International Trade 2026, World Bank Brown Bag 2025
How easily firms substitute between domestic and foreign suppliers within an input is central to several questions in international economics, but credible empirical estimates remain limited. I provide one of the first direct plant-level estimates of the elasticity of substitution between domestic and foreign suppliers within narrowly defined inputs. Using rich Indian plant-level data, I estimate an elasticity of 0.50, implying complementarity between foreign and domestic sourcing and substantially below recent estimates based on aggregate data. Aggregating the same data across plants to mimic a national-level input-sourcing dataset yields estimates three to four times larger. I derive and empirically validate conditions under which ignoring firm-level heterogeneity can generate a large upward bias in aggregate data estimates. To understand its macro importance, I embed my estimates in a general equilibrium model with product level input-output linkages. A quantitative application to the dollar shock induced by the 2013 Taper Tantrum shows that a standard calibration of foreign and domestic sourcing as substitutes, rather than complements, understates the real GDP loss by 13.3% and shapes which firms are more exposed to the shock.
We study how poor domestic transport infrastructure shapes the distributional effects of integration with world markets. Exploiting India’s large-scale trade liberalization in the 1990s, together with a digitized road network, we find that reductions in input tariffs generate substantially larger manufacturing gains in districts with better access to international ports, accompanied by faster reallocation of workers out of agriculture and into manufacturing and producer services, and greater migration from more remote regions. We interpret these findings through a quantitative spatial model. A key mechanism is that manufacturing intermediates are more time-sensitive in transit than manufacturing output, leading to larger increases in the relative returns to manufacturing in districts with better port access. The model replicates our main empirical findings and reveals stark distributional inequalities from liberalization, with gains in real income, real wages, and real rents heavily skewed toward regions with better port access. Finally, a simultaneous highway expansion targeted at improving port connectivity eliminates these spatial inequalities in welfare gains while generating net aggregate benefits.
Recent shifts in global trade patterns have received a lot of attention, but the implications of these shifts for domestic value added in exports (DVAR) have so far been overlooked. This paper documents substantial increases in Viet Nam’s aggregate DVAR since 2018, reversing a long declining trend. Analysis using matched firm-level and customs data reveals that idiosyncratic US demand shocks had large positive effects on firm-level DVAR and output. Firms respond to the shocks by exporting more to both US and non-US markets, and their labor productivity increases, both that are consistent with the economies of scale. Firms also increase their use of domestic materials and reduce reliance on imported materials, suggesting sensitivity to implicit rules of origin. Finally, the paper highlights the role of firm ownership, with non-Chinese owned firms driving the increase in domestic value added.
We examine the role of transportation and communication infrastructure in shaping structural transformation by embedding it in a spatial model of non-homothetic demand and multi-sector production. We show that service-led structural change can both be a cause of service productivity growth, or a consequence of improved connectivity that allows districts to source tradable goods from others; rising real incomes increase demand for (non-tradable) services, which have to be supplied locally. We quantify the importance of these channels using detailed data on road and cellular network expansion, an instrumental variables strategy, and representative household and establishment data in India. Infrastructure — particularly cellular networks — and sectoral productivity each account for about half the aggregate real-income gains between 1999–2009. While infrastructure-driven gains are lower in initially more urban areas, spatial complementarities through a national network expansion are larger in them. Put together, we conclude that infrastructure shapes both the sectoral and spatial incience of structural transformation.