Public employment programs may affect poverty both directly through the income they provide and indirectly through general equilibrium effects. We estimate both effects, exploiting a reform that improved the implementation of India's National Rural Employment Guarantee Scheme (NREGS) and whose rollout was randomized at a large (sub‐district) scale. The reform raised beneficiary households' earnings by 14%, and reduced poverty by 26%. Importantly, 86% of income gains came from non‐program earnings, driven by higher private‐sector (real) wages and employment. This pattern appears to reflect imperfectly competitive labor markets more than productivity gains: worker's reservation wages increased, land returns fell, and employment gains were higher in villages with more concentrated landholdings. Non‐agricultural enterprise counts and employment grew rapidly despite higher wages, consistent with a role for local demand in structural transformation. These results suggest that public employment programs can effectively reduce poverty in developing countries, and may also improve economic efficiency.