The disparities in cross-country labor productivity are greater in agriculture than in other industries. I propose that the misallocation of female talent across sectors distorts productivity. I formalize the theory by using a general equilibrium Roy model with gender-specific frictions. If female workers experience higher frictions in nonagricultural sectors, then female workers who are better skilled at non-agricultural jobs may select into agricultural sector. From a sample of 66 countries, I find that low-income countries have higher frictions in non-agricultural industries. By setting frictions to US levels, agricultural labor productivity increases by 4.3-7.6 percent, nonagricultural labor productivity decreases by 0.7-1.4 percent, and GDP per capita increases by 0.8-1.5 percent.