Impact of crop diversification on farm income in arid and semi-arid regions of India

This study examines how crop diversification into high-value crops affects net farm income among agricultural households in India’s arid and semi-arid regions, using two rounds of unit-level of the NSSO Situation Assessment Surveys (70th (2012–13) and 77th (2018–19)). To address selection bias from non-random diversification, the study applies Propensity Score Matching (PSM) based on logit estimates of households’ propensity to diversify. Impacts are estimated within the region of common support using multiple matching algorithms (nearest-neighbor, kernel, and radius), complemented by post-matching balance diagnostics, and Rosenbaum bounds sensitivity analysis. Descriptive results show that around one-third of households are diversified, with diversification declining over time. Diversified households are systematically better positioned in terms of institutional and informational access, including a higher likelihood of receiving formal credit, crop insurance, and technical advice, and they have a higher share of working-age members. Furthermore, the logit estimates indicate that crop insurance, institutional credit, and media-/extension-based advice are key correlates of diversification, while the role of irrigation weakens in the later period. Across matching specifications, the Average Treatment Effect on the Treated is positive in both rounds but statistically insignificant, implying modest income gains amid substantial income variability. These estimates represent potential income gains of 3–20% relative to mean real farm income, though the wide range reflects substantial heterogeneity in household income across the dryland belt. Balance tests indicate improved comparability after matching, and sensitivity results suggest robustness to sizeable hidden bias. Overall, diversification as a climate-smart agricultural practice appears more relevant for resilience and risk management than for large mean income growth.