We quantitatively evaluate general equilibrium effects of input subsidy programs in agriculture. First, we present cross-country evidence that while staple-targeting subsidy programs improve both the productivity of staples and food security, they leave behind producers of valuable cash crops. This trade-off is confirmed by our micro-evidence in Malawi showing that cultivating more staples reduces the overall value of household’s cultivated crops. Then, we build a dynamic general equilibrium model with financial frictions, transaction costs, subsistence consumption constraints and heterogeneous agents making occupational choice between working as labourers in urban areas, and as staple or cash crop farmers in rural areas. Quantifying the model for the case of African input subsidy programs we show the importance of financial development and infrastructure on the optimal policy.