Social Diversity and Spread of Pandemic: Evidence from India

TL;DR

This study investigates how caste-group diversity in Indian districts influenced COVID-19 spread during lockdown, revealing that caste-homogeneous areas experienced slower infection growth and highlighting the role of community health workers in managing public health crises.

Contribution

It provides empirical evidence linking caste-group homogeneity to slower COVID-19 spread and emphasizes the importance of community health workers and decentralized responses.

Findings

Caste-homogeneous districts took more days to reach infection thresholds.

Slower infection growth observed in caste-homogeneous districts during lockdown.

Community health workers' engagement increased in caste-homogeneous localities.

Abstract

Compliance with the public health guidelines during a pandemic requires coordinated community actions which might be undermined in socially diverse areas. In this paper, we assess the relationship between caste-group diversity and the spread of COVID-19 infection during the nationwide lockdown and unlocking period in India. On the extensive margin, we find that caste-homogeneous districts systematically took more days to cross the concentration thresholds of 50 to 500 cases. Estimates on the intensive margin, using daily cases, further show that caste-homogeneous districts experienced slower growth in infection. Overall, the effects of caste-group homogeneity remained positive and statistically significant for 2.5 months (about 76 days) after the beginning of the lockdown and weakened with subsequent phases of the lockdown. The results hold even after accounting for the emergence of initial hotspots before lockdown, broader diffusion patterns through daily fixed effects, region fixed effects, and dynamic administrative response through time-variant lagged COVID-19 fatalities at the district level. These effects are not found to be confounded by differential levels of testing and underreporting of cases in some states. Consistent estimates from bias-adjusted treatment effects also ensure that our findings remain robust even after accounting for other unobservables. We find suggestive evidence of higher engagement of community health workers in caste-homogenous localities, which further increased after the lockdown. We posit this as one potential channel that can explain our results. Our findings reveal how caste-group diversity can be used to identify potential hotspots during public health emergencies and emphasize the importance of community health workers and decentralized policy response.