Drought Shocks and Firm Performance: A Study of Causal Effects Using Machine Learning

This paper explores the effects of drought shocks on firms’ performance based on the generalized random forest by integrating farmers and firms into the same analytical framework. The results indicate that drought shocks have a negative impact on firms' revenues, employment size, and profit. The negative impact of drought shocks on firms has significant heterogeneous effects across firms with different assets, industries, and ages. Robustness tests reveal that the data used in the study contains sufficient information so that the results are not affected by omitted variables and other confounding factors. In addition, the results remain robust when non-randomization, repeated shocks, and spillover effects are considered. Mechanistic analyses reveal that drought shocks affect firms through demand and cost effects. On the one hand, drought shocks affect the consumption level of regional farmers, which reduces the demand for firms’ products, thus affecting firms’ revenues and employment. On the other hand, drought shocks lead to a decline in demand for firms' products that is greater than the decline in firms' costs, which in turn has an impact on firms' profits. The results of the adaptive strategies analysis show that promoting agricultural technology, improving agricultural facilities, and promoting financial inclusion in the region can effectively mitigate the negative impacts of drought shocks on firms’ performance.