Estimating the Elasticity of Intertemporal Substitution with Household-Specific Portfolios

Abstract This paper estimates the elasticity of intertemporal substitution (EIS), allowing for household-specific portfolio. Previous studies that estimated the EIS used financial indexes as a proxy for the risky return on a representative household portfolio. According to the latest data from the 2004 Survey of Consumer Finances, however, the median US stockholders who own stocks directly hold only 3 stock securities. If a large fraction of stockholders do not own a financial index and hold only few individual stocks, then how does that affect inference about household risk aversion? We estimate the EIS using the log-linearized Euler equation derived by Hansen and Singleton (1983) and accounting for household-specific portfolio choice instead of a financial index. Our results show two main findings. First, financial indexes are not a proper substitute for household-specific portfolio. Second, we find support for the standard representative agent assumption that there is a high degree of homogeneity in the EIS across households with different wealth levels (the EIS approximately is 0.22). Our findings have implications for models that assess the comovement between consumption and return on stocks since the value of EIS reflects the comovement level. We argue that a consideration of financial indexes instead of household-specific portfolio explains the small comovement puzzle introduced by Mankiw and Zeldes (1991).