Who owns the firm? Ownership, structure, and productivity disparities in India’s unincorporated enterprises

Purpose This study aims to examine how caste shapes enterprise ownership, structural conditions, and productivity in India’s unincorporated sector. It asks whether the rapid expansion of informal enterprises over time has reduced or reproduced caste-based inequality. Design/methodology/approach The study draws on unit-level data from the National Sample Survey (67th round, 2010–11; 73rd round, 2015–16) and the Annual Survey of Unincorporated Sector Enterprises (ASUSE) 2022–23, 2023–24 and 2025. The methodology combines descriptive profiling of ownership and enterprise characteristics; a pooled multinomial logit to assess caste-linked sorting, pooled ordinary least squares (OLS) and Poisson pseudo-maximum likelihood (PPML) regression to estimate the productivity gap; and Blinder-Oaxaca decomposition to separate disparities. Findings The findings reveal a clear paradox: the informal sector enterprise expanded rapidly, but caste inequality in enterprise ownership and returns persisted. Establishments increased from 57.7 million in 2010–11 to 83.5 million in 2025, yet Scheduled Caste (SC) and Scheduled Tribe (ST) ownership remained below their population shares. SC and ST entrepreneurs remained more concentrated in precarious locations and low-capital activities, while upper-caste enterprises retained advantages in permanent premises and productivity. Across the pooled productivity models, the most robust disadvantage is borne by SC-owned enterprises. The decomposition shows that 69.8% of SC-“Others” productivity gap is explained by location, registration and digital adoption, while the remaining portion remains unexplained, pointing to discrimination and unequal social capital. Originality/value The study shifts attention from wages and employment to caste differences in enterprise ownership, structure, and productivity, showing that firm growth did not reduce caste gaps in ownership, business structure, or returns.