Joint elicitation of elasticity of intertemporal substitution, risk and time preferences

Abstract

The elicitation of the elasticity of intertemporal substitution (EIS), discount factor and risk attitude parameters in dynamic models is of central importance to economics, finance and public policy. This paper suggests an alternative method to jointly elicit and estimate these three parameters using experimental data. We employ a new model based on dynamic quantile preferences, where individuals maximize the stream of future $\tau$-quantile utilities, for $\tau \in \left(0,1\right)$. These preferences are simple, dynamically consistent and monotonic. In the quantile model, the risk attitude is captured by the quantile $\tau$ of the payoff distribution, while the EIS and the discount factor are related to the utility function describing individual's intertemporal behaviour, hence allowing for complete separability between risk, EIS and discount factor. The estimation of the parameters of interest uses a structural maximum likelihood method. Individual's risk aversion is estimated below the median. The discount factor is marginally smaller than estimates reported in the literature, and the EIS is slightly larger than one, which suggests that utility over time is concave. The estimates for the elasticity contrast with those reported by the existing studies using observational disaggregated data, which in general find an elasticity smaller than one.