Abstract:

This paper studies how people aggregate over risk and time when evaluating intertemporal lotteries. In an experiment, subjects are asked to report their present certainty equivalents of receiving a lottery at two different points in time. These two lotteries are perfectly positively correlated, perfectly negatively correlated, or independent. People who first aggregate over risk, are insensitive to intertemporal correlations. We measure non-parametric degrees of correlation aversion and disentangle them into positive correlation aversion and negative correlation seeking. Subjects were both positive correlation averse and negative correlation seeking. They were found to be intertemporal correlation averse across lotteries and time frames. The results show that subjects aggregate first over time, or recursively over risk and time. Two framings were considered between subjects: one that encouraged subjects to first aggregate over risk, and another one that encouraged them to first aggregate over time.While these framings resulted in lower present certainty equivalents for the framing that encouraged to aggregate first over risk, they did not affect degrees of intertemporal correlation aversion. Thus, subjects consistently aggregated first over time or recursively, even in the framing that encouragedto aggregate first over risk.