Consumption adjustment to real interest rates: Intertemporal substitution revisited

This paper investigates the degree of elasticity of intertemporal substitution in consumption using post-war US aggregate data. Previous findings suggest that the elasticity of substitution is unlikely to be much above 0.1 and may well be zero. In contrast, I find strong evidence that there is a statistically significant positive response of consumption growth to changes in expected real interest rates. The elasticity estimates cluster around 0.3. Previous weak results are attributed to either the inappropriate choice of instruments or the use of an inadequate measure of consumption. This finding is robust to considerations of the time aggregation bias, different sample periods, and alternative formulations of the permanent income consumption model which include 'rule-of-sum' consumers and borrowing constraints. For example, if the Campbell and Mankiw model is adopted as an approximate description of US aggregate time series in consumption, income and real interest rates, the implied elasticity of intertemporal substitution for permanent income consumers could be as high as 0.8, which implies that the coefficient of relative risk aversion is around 1.25.