The elasticity of intertemporal substitution is an important determinant of the response of saving and consumption to the real interest rate. Summers (1984) argued that the intertemporal substitution effect was strong, whereas Hall (1988), drawing evidence from U.S. data, concluded that its value is close to zero. Hall maintained that the previous higher estimates obtained by Summers (1982) and others are due to inappropriate treatment of the time aggregation bias and can therefore be dismissed.
This paper, which addresses the specification issues raised by Hall, extends the earlier research to an international context by examining data from twenty OECD countries. The Kreps-Porteus nonexpected utility preference is adopted, and distributional restrictions are imposed to derive a simple relation that governs the covariation of consumption growth and asset returns, which allows unambiguous identification of the intertemporal substitution parameter.
The single-equation generalized method of moments estimates for each of the seven major industrial countries are typically small and imprecise, corroborating Hall’s earlier finding from the U.S. data. The full information maximum likelihood estimation, however, gives larger and more precise values for the parameter, possibly because of the efficiency gain of system estimation. The panel procedure also yields relatively large estimates. Overall, the multicountry evidence seems to contradict the hypothesis of zero intertemporal substitution.
The results presented in this paper imply, among other things, that a shift toward expenditure taxation would probably lead to increases in private savings.