I consider the aggregate impact of low intermediate input intensity in the agricultural sector of developing countries. In a dynamic general equilibrium model with idiosyncratic shocks, incomplete markets, and subsistence requirements, farmers in developing countries use fewer intermediate inputs because it limits their exposure to uninsurable shocks. The calibrated model implies that Indian agricultural productivity would increase by 16 percent if markets were complete, driven by quantitatively important increases in both the average real intermediate share and measured TFP through lower misallocation. I then extend the results to consider the importance of risk in other contexts. First, the introduction of insurance decreases cross-country differences in agricultural labor productivity by 14 percent. Second, scaling the introduction of improved seeds to decrease downside risk reduces inequality by reallocating resources from rich to poor farmers via equilibrium effects. This reallocation substantially increases aggregate productivity relative to what would be expected from extrapolating the partial equilibrium impact.