Research

Working Papers

Household labor reallocation provides a potentially important channel for rural households to adapt to changing weather patterns. Exploiting the temporal and spatial variation of drought occurrence in India, I find that drought reduces the share of agriculture labor hours by 110 hours (3% at the mean). This reduction is driven by households that do not own land. Motivated by these facts, I develop a model of labor allocation across the agriculture and non-agriculture sectors to analyze how droughts may affect structural transformation. My results imply that projected increase in the spatial extent of droughts over the coming years will induce landowning households to allocate 2% more labor to agriculture and induce landless households to reduce their agricultural labor. The net effect is a 1% reduction in agricultural labor. While small in percentage terms, this implies that 2.5 million individuals would leave agriculture. I use the model to analyze how projected climate change would affect the cost to the government of achieving its stated target of increasing the manufacturing share of GDP to 25% by 2035. Under climate change, I find that the government would need to subsidize non-agriculture wages by a higher margin due to the presence of land market frictions.

Droughts are becoming increasingly common in India, where 50\% of the labor force works in agriculture, and most agricultural production is rainfall dependent.  This paper investigates the extent to which rural households adapt to drought by reallocating labor from agriculture to other sectors of the economy. I combine high-resolution data on drought with panel data on about 8,000 Indian households collected over a 20-year period in rural areas in 184 districts. I use household-level fixed effects regressions to estimate how drought affects household consumption and diversification from agriculture, and to investigate the mechanisms underlying these effects.  I find that household consumption declines by 5.9% in response to drought and agricultural jobs decline by 2.9%  in the year following a drought. Further, I find that these effects are mediated by job skills and land ownership. Specifically, I find that households with working members who have completed primary education account for most of the workers who exit the agricultural sector. In contrast, I find that households that own land increase their agricultural labor share after experiencing a drought. Thus, while I find that drought causes households to diversify away from agriculture on aggregate, the extent of this structural change is mitigated by the behavior of landowners. Cultural norms, relative prices, and land market transaction costs provide potential explanations for this behavior.


Environmental Justice for Seniors? Evidence from the Superfund Program with Jonathan Ketcham and Nicolai Kuminoff (manuscript available upon request)

We study how the effects of “Superfund” hazardous waste sites’ listing and deletion during the 2000’s differed across race, income, and health among US seniors. Using a random 20% sample of the US Medicare population from 1999 through 2013 and a spatial difference–in–difference regression method, we find the probability to move out is 5-7% higher for seniors who live within 3.5km of a site in response to the designation of a Superfund site compared to seniors living within 3.5-7km. On average, seniors who move away from sites reduce their exposure to PM2.5 and move further from other sites, but the reductions in pollution are smaller for poorer, sicker, and non-white movers. In addition, we find that Black, Hispanic, and poorer people are 5-6% more likely to move within 3.5km of a site that is not yet proposed to be cleaned up compared to 3.5-7km of a site. We find nearly symmetric results upon the completion of cleanup of a Superfund site. These findings add to the environmental justice literature by providing new evidence on pollution- exposure and sorting for the Medicare population who are defined by the EPA as a vulnerable group to pollution based on age, income, and health.

Work in Progress

Resource traps and child well-being in Zambia with Valerie Mueller 

Theoretical models predict quite diverse effects of resource boom on investments in children. On the one hand, the household income shock may redirect patterns of basic consumption towards human and physical capital investments (Glewwe and Jacoby, 2004). Furthermore, recent literature suggests that income shocks can have quite gendered consequences on patterns of investment due to its consequences on marriage market conditions and bride price (Corno et al., 2020), as well as the increased demand for unpaid labor as wages attract more household members to engage in paid work. On the other hand, studies have suggested that positive income shocks raise the opportunity cost of education, encouraging adolescents to seek employment rather than continue their education (de Brauw and Giles, 2017). This paper estimates the effects of copper mining on human capital accumulation in the Copperbelt region of Zambia. We exploit the exogeneity of international prices and distance from mines to identify causal effects. In particular, we explore how the copper boom in the 2000s have influenced school attendance among children between 12-18 years of age, whether these patterns are different across boys and girls and what mechanism explain this effect. Using a distance threshold regression, we define our treatment areas to be within 30km of a mine. 30-100km of a mine is the control group. We find that school attendance has fallen by 7.4 pp more within 30 km of a mine after the copper boom, compared to 30-100km of a mine. This effect is higher among boys compared to girls. We explore the potential mechanisms underlying the motivation for declines in educational investment: 1) credit-constrained households diversifying income practices to save during positive income shocks (permanent income hypothesis); 2) increased demand for unpaid work in the household and shifts in the intra-household allocation of labor; 3) contribution of positive income shocks to marriage market dynamics.

Segregation and Connectivity: Evidence from India with Manaswini Bhalla, Manisha Goel and Gaurav Khanna

It is well-documented that Indian cities are segregated along caste and religious lines (Adukia et al, 2019; Bharathi et al, 2021). In the recent past, major Indian cities have witnessed an expansion of metro rail which is intended to reduce commuting times in congested cities. Neighborhoods with improved public goods such as new metro rail terminals may influence housing market dynamics that could affect residential sorting. This could lead to segregation which affects economic outcomes (Chetty et al, 2020). This is the first paper to investigate how urban commuting infrastructure affects segregation in India. In this paper, we document how the expansion of metro rail in Bangalore impacts residential segregation along caste and religious lines. We use a simple model of residential sorting incorporating preference for commuting facilities and demographic composition of neighborhoods to motivate our empirical specifications. We use a novel dataset of the universe of property sales in Bangalore between 2014-2018. Our empirical strategy relies on the distance between new metro rail terminals and property locations. We assign treatment and control neighborhoods based on their distance to the nearest metro rail terminal. Exploiting the caste identity of buyers and sellers, we calculate inflows and outflows of various caste and religious groups in different neighborhoods. We identify the effect of metro rail expansion on the composition of neighborhoods using the temporal and spatial variation in the opening of metro links. This paper would add to the urban economics literature, particularly the branch studying how place-based investment policies affect residential segregation.

 Airport Expansion, Air Pollution, and Death Count in Mexico City with Luis A. Fernández Intriago