Intertemporal Correlation Aversion—A Model-Free Measurement

Decisions with risky consequences at multiple points in time are driven not only by risk attitudes and time preferences but also by attitudes toward intertemporal correlation (i.e., the correlation between outcomes at different points in time). This paper proposes a model-free method to measure degrees of intertemporal correlation aversion. We disentangle attitudes toward positive and negative intertemporal correlation, which can differ if expected intertemporal utility is violated. In an experiment, subjects on average exhibited correlation aversion both for lotteries with positive correlation and for lotteries with negative correlation. That is, they disliked positive correlations and liked negative correlations. At the individual level, we found heterogeneity and remarkably, many subjects being insensitive to intertemporal correlations. Moreover, for most subjects, expected intertemporal utility was violated because attitudes toward positive and negative correlation differed.