How do the health benefits of reducing air pollution translate into aggregate income gains? Income gains can differ from health benefits because labor productivity may vary in the places where people experience lower pollution, and because labor reallocation to those places may reinforce agglomeration economies. This paper shows that incorporating two physical features of pollution sources - their location and long-distance dispersion tendency - can alter the importance of labor productivity and reallocation, and lead to very different income gains from pollution control policies that otherwise produce similar population health benefits. To understand these interactions, I develop a spatial equilibrium model that accounts for the movement of both pollution and people across space. I apply this model to study income gains from two archetypal pollution control policies that target non-industrial sources in India andproduce similar population exposure reductions, but in very different places. One policy controls agricultural fires in northwestern India that spread pollution across much of north India, and another policy reduces localized emissions from sourcessuch as vehicles within India’s 10 largest cities. Accounting only for differential labor productivity in the places experiencing lower pollution, I find that the latter policy leads to a 3 times larger GDP gain relative to the former. Further accounting for labor reallocation and agglomeration economies leads to a 6 times larger GDP gain. These results have implications for spatial targeting of pollution, especially in polluted-yet-poor LMICs.