Research

This paper highlights the importance of distinguishing between temporary and permanent migration when assessing the impact of economic shocks and the welfare consequences of public policies. To this end, I develop a spatial general equilibrium trade model that allows for both types of migration between locations. Applying the model to India, I find that temporary migration contributes 1.5 times more to average welfare and three times more to the welfare of the poorest 10% of districts, compared to permanent migration in the baseline economy. Furthermore, temporary migration plays a key role in reallocating labor in response to local economic shocks, generating significantly larger gains in both productivity and welfare relative to permanent migration. Finally, the model accounting for both forms of migration finds that spatial redistributive policies improve average welfare, whereas models limited to permanent migration predict welfare losses under the same policies. These findings underscore the importance of considering temporary migration when studying the transmission of local shocks and the effects of spatial redistribution.

Domestic transport improvements may raise manufacturing productivity by connecting plants to specialized intermediate-input suppliers. I study this channel using India’s Golden Quadrilateral and a plant-specific measure of input market access combining baseline product mixes, national input requirements, supplier geography, and modeled road travel times. Because plants require different inputs, the same road improvement produces different predicted gains within a district. Exploiting this variation with plant, district-by-period, and industry-by-period fixed effects, I find that greater supplier access is associated with higher domestic material-cost shares, broader reported input use, higher value-added labor productivity, and expanded production. A 0.10-log-point larger cumulative access gain corresponds to a 1.16-percentage-point higher domestic material-cost share, 4.8% higher real value added per employee, and 8.4% higher production in 2007–08 relative to 2000–02. The main results persist after accounting for buyer access, with stronger input-breadth and labor-productivity responses among initially smaller plants. The measure identifies economically relevant sourcing opportunities without observed buyer–supplier links.

We study the relationship between informality within the non-agricultural sectors and development. Using cross-country data and district-level evidence from India, we show that the informal employment share within non-agriculture declines non-linearly with development: initially, the non-agricultural informal employment share is stable or even rises with development, but it declines steadily as economies move along the development path. To rationalise this non-linear relationship, we develop a multi-sector general equilibrium model featuring non-homothetic preferences, multiple goods within non-agriculture differentiated by quality, agent heterogeneity by sectoral productivities, occupational choices (self-employment, wage labour or entrepreneurship) and endogenous formalisation choices by entrepreneurs. The model implies slow formalisation early in development and faster formalisation later, as demand shifts towards high-quality goods which tend to be produced formally, raising their relative price, and as equilibrium firm scale rises with development, resulting in a higher informality penalty. Policies seeking to maximise non-agricultural labour productivity contingent on achieving a specific increase in formalisation incentivise formal production in middle income economies, while they deter informality in richer economies.

We document a novel empirical finding: U.S. regions with higher pre-Great Recession exposure to automation experienced significantly faster employment recoveries after the recession. To explain this result, we develop a simple model of firm heterogeneity with capital accumulation. The model predicts that, following a transitory negative TFP shock, economies with higher automation intensity recover more quickly than those with lower automation intensity. In less automation-intensive regions, firms start with relatively lower levels of automation capital, and this gap is exacerbated during the shock. As a result, these regions benefit less from the complementarity between automation capital and labour, leading to a slower employment recovery.