Households hold under-diversified stock portfolios concentrated in firms headquartered near the city where they reside. Theories for this local-bias puzzle assign a causal role for proximity, be it generating an informational advantage or a familiarity bias. These explanations assume that households locate randomly. But in location choice models, a household optimally selects a city depending on many factors. This selection is important since latent location factors might be correlated with latent demand for local stocks. We develop a Heckman (1977)-style model to account for the effect of location choices on portfolio choices. Adjusting for this selection effect drastically reduces the impact of distance on household portfolios and the performance of local stock picks. We point to the importance of accounting for location choices on portfolio choices more generally.