Inter-firm Networks across Intra-national Borders: Theory and Evidence from India

The organisation of production in modern economies features large-scale complex networks of firms spanning across industries and locations. As a result, how firms choose to source their intermediate inputs is an important determinant of aggregate welfare. Sustained growth of the private sector in a lower-income country requires that idiosyncratic distortions at the firm level and firm-to-firm level be low to mitigate losses in aggregate production efficiency. These distortions, be they due to lack of infrastructure, institutional reasons or administrative inefficiencies lower aggregate welfare. The study of the effects of these idiosyncratic distortions to date has been limited by (a) a lack of detailed micro-data on firm-to-firm linkages extending across a wide geography and (b) the use of continuum economy models that assume away granularities to deliver deterministic aggregate variables. This project develops a theory and estimation framework for inter-firm production networks in the presence of these distortions and utilises a new rich administrative dataset of Indian firm-to-firm linkages to address these limitations and shed light on the channels through which the input-output multiplier operates.

First, Panigrahi will build a granular model of multi-region inter-firm production networks featuring idiosyncratic distortions both at the firm-level and firm-to-firm level where both the intensive and extensive margins of inter-firm trade are determined within the model. Second, a novel, extremely rich and unusually large dataset is collected on roughly 30 million firm-to-firm linkages between around 2 million firms located across 142 districts within 5 Indian states spanning five years from administrative tax records. Third, taking advantage of the full volume of data, Panigrahi develops an econometric framework to estimate this model of network formation. Finally, he uses the estimated model to quantify the macroeconomic effects of a recent policy reform in India where harmonisation of commercial tax structures reduced distortions in inter-firm transactions across state borders.

Over 166 countries in the world, including many LICs, employ VAT as the primary means of commercial taxation. As a result, these governments collect similar administrative data to that used here. The generalised nature of this project's model makes it convenient for policymakers to adapt the parameters to any country of interest by utilising similar micro-data. The project will also directly inform the Government of India as it tries to formulate and make optimal changes to tax structure and revenue sharing arrangements while considering the general equilibrium effects of such a change.


Piyush Panigrahi

University of California, Berkeley