Housing returns and intertemporal substitution in consumption: estimates for industrial economies

This paper uses housing returns to estimate the elasticity of intertemporal substitution (EIS) in consumption for fifteen advanced economies over the postwar period 1950-2015. As housing is the main asset for the majority of households, returns on housing are better suited to estimate the EIS than the asset returns typically considered in the literature, i.e., equity and bill returns. An estimable regression equation for aggregate consumption growth and returns is obtained from the aggregation of the consumption Euler equations of heterogeneous agents. As the regression equation includes unobserved omitted variables, we use instrumental variables estimation. We exploit both the temporal and spatial dimensions of the panel by instrumenting the domestic return using its own lag and a cross-country average of foreign returns. Both instruments are strong and allow to test the overidentifying restriction. The restriction holds once we control for common international growth and financial factors in the regression equation. We report a baseline elasticity estimate of about 0.21. This is substantially larger than the elasticities estimated from equity and bill returns which, in line with the extant literature, are found not to be significantly different from zero.