Structural models of location choice use observed demand to estimate household preferences. However, household demand may be partly determined by borrowing constraints, limiting households’ choice set. Credit availability differs across locations, households, and years. We put forward a model of neighborhood choice where mortgage approval rates determine households’ choice set. Using household-level data, geocoded transactions, and mortgage applications for the San Francisco Bay area, we find that including borrowing constraints leads to higher estimated preferences for better performing schools and majority-white neighborhoods. General equilibrium estimates of the relaxation of lending standards provide two out-of-sample predictions: between 2000 and 2006, (i) a compression of the price distribution and (ii) a decline in black exposure to Whites. Both predictions are supported by empirical observation. Joint with Romain Ranciere.

Click here to read the paper.