House Prices and Consumption : A New Instrumental Variables Approach

While the elasticity of housing supply has served as a popular instrument for inferring the causal effect of housing price growth on economic outcomes, concerns about the exclusion restriction have emerged. We introduce a methodology for creating instruments using the pre-existing supply of housing structures of different characteristics changes in the price of these characteristics. Unlike other common strategies for inferring the effects of housing price shocks, our Bartik-like instrument is a strong predictor of housing price growth in both the cross-section and over time even after controlling for county and time fixed effects, as well as a wide array of local demographic, business cycle, and industry characteristics. Using longitudinal household data, our instrument produces an elasticity of consumption with respect to local housing price growth of 0.12, corresponding to a marginal propensity to consume out of housing wealth equal to 0.8-1.03 cents on the dollar. These effects are concentrated among the young and those likely to be facing tight borrowing constraints