Income Uncertainty and Heterogeneous Labour Supply Responses: The Role of Risk Preference Among Gig Workers in India

Abstract

This paper examines how income uncertainty shapes labour supply decisions among platform-based gig workers in India, with explicit attention to the moderating role of individual risk preferences. Using primary survey data collected from 200 gig economy workers across Hyderabad, Bengaluru, and Mumbai — spanning ride-hailing, delivery, and freelancing platforms — we construct individual-level income volatility measures and elicit risk preferences through a combination of the Domain-Specific Risk-Taking (DOSPERT) scale and a behavioural lottery task. Our core finding is that responses to income uncertainty are strongly heterogeneous and systematically mediated by risk attitudes. Risk-averse workers exhibit a pronounced buffer-stock labour supply response: a one standard deviation increase in income volatility raises weekly hours by approximately 12.7 hours (OLS) and 16.2 hours (IV), consistent with precautionary labour supply theory. Risk-tolerant workers, by contrast, show a modest negative or null response, suggesting they substitute uncertainty with optimised job selection rather than hours expansion. Platform diversification — measured by the number of platforms a worker simultaneously participates in — follows a similar heterogeneous pattern. Two-stage least squares estimates using platform rating variance as an instrument for income uncertainty confirm causal identification and yield estimates broadly consistent with OLS. Subgroup analyses reveal that income-education and platform-type heterogeneities further amplify these patterns, with low-education and ride-hailing workers displaying the largest precautionary responses. Robustness checks using quantile regression, Tobit, and Heckman selection models corroborate our main findings. These results carry important implications for gig-platform design, social insurance architecture, and the welfare of informal workers in emerging economies.