This paper studies neighborhood effects using a dynamic general equilibrium model. Households choose where to live and how much to invest in their child's human capital. The return on parents' investment is determined in part by their child's ability and in part by a neighborhood externality. We calibrate the model using data from Chicago in 1960, assuming that in previous decades households were randomly allocated to, and then could not move from, neighborhoods with different total factor productivity (TFP). This restriction on neighborhood choice allows us to overcome the fundamental problem of endogenous neighborhood selection. We use the calibrated model to study Wilson's (1987) hypothesis that racial equality under the law need not ensure equality of opportunity due to neighborhood dynamics. We examine the consequences of allowing for mobility, equalizing TFP, or both. In line with Wilson, 1987, sorting can lead to persistent inequality of opportunity across locations if initial conditions are unequal. Our results highlight the importance of forward‐looking agents.