House Prices, Consumption, and Their Common Causes

Abstract

This paper provides evidence that common drivers of consumption and house prices, as opposed to wealth or collateral channels, are responsible for most of the positive correlation between the two series. Using micro data from the PSID and the CEX, I illustrate the relative importance of the common causality channel using three different methods. First, I show that consumption elasticities for owners with respect to statewide housing appreciation can be reduced by about half with the inclusion of controls for the per capita income and the unemployment rate of the household's state of residence. In the second approach, I use renters as a control group and show that owners do not display any significant difference in their consumption response to house price changes. Lastly, the paper presents evidence showing that consumption elasticities are statistically and economically significant with respect to broad changes in house prices, at the state or MSA levels but are insignificant with respect to self-reported owner estimates. I argue that the differential effects found using these alternative measures is the result of a trade-off between two different types of noise. While owners' estimates contain a great deal of sampling noise, the change in HPI is a noisy measure for individual housing appreciation because the substantial heterogeneity in individual appreciation rates within a locality. If owners make consumption and savings decision based on perceived wealth, owners' estimates are the appropriate measure to estimate wealth effects. If heterogeneity in individual home appreciation rates is significant, change in HPI is a poor measure for changes in household housing wealth, though it could still be spuriously correlated with consumption due to common drivers.

Keywords: wealth effects, house prices, marginal propensity to consume, consumption

JEL Classification: D12, E21, H31