The descriptive and predictive adequacy of theories of decision making under uncertainty/ambiguity

Abstract

In this paper we examine the performance of theories of decision making under uncertainty/ambiguity from the perspective of their descriptive and predictive power, taking into account the relative parsimony of the various theories. To this end, we employ an innovative experimental design which enables us to reproduce ambiguity in the laboratory in a transparent and non-probabilistic way. We find that judging theories on the basis of their theoretical appeal, or on their ability to do well in testing contexts, is not the same as judging them on the basis of their explanatory and predictive power. We also find that the more elegant theoretical models do not perform as well as simple rules of thumb.

Impact of Outcome Ambiguity on Self‐insurance and Self‐protection: Experimental Evidence

“…In aggregate, consistent with intuition, most empirical research show that individuals behave to avoid ambiguity (Hey et al, 2009;Hey et al, 2010;Pace and Hey, 2011;Morone and Ozdemir, 2012). Thus, for example, in the case of tornado that household certainly knows that it will hit the house according to the warning (with known probability), they will try to find ways to mitigate the loss size.…”

Controlling for presentation effects in choice

“…Obviously, estimates could be inconsistent simply due to differences in the subject pools, but in this case, all models will detect inconsistency. Note that consistency has been argued to be negatively correlated with in‐sample accuracy, since a model that does not fit well in‐sample may still be particularly robust due to being “simpler” (Hey, Lotito, and Maffioletti (2010)). This may put Focal at a disadvantage.…”

“…Do other models fail in this respect, and is the relevance of presentation effects economically substantial? The general idea, to test behavioral models by evaluating validity in‐sample and out‐of‐sample using typical data sets, follows a literature comprising analyses of decision under risk (Harless and Camerer (1994), Wilcox (2008), Hey, Lotito, and Maffioletti (2010)), learning (Camerer and Ho (1999)), strategic choice in normal‐form games (Camerer, Ho, and Chong (2004)), and social preferences (De Bruyn and Bolton (2008)).…”

Risk-sensitive inverse reinforcement learning via semi- and non-parametric methods

“…Polytopic risk measures constitute a rich class of risk measures, encompassing a spectrum ranging from risk neutrality ( scriptP = { p } ) to worst-case assessments ( scriptP = normalΔ | normalΩ | ); see also Chow and Pavone (2014) and Shapiro et al (2014). We further note that the ambiguity interpretation of CRMs is reminiscent of Gilboa and Schmeidler’s (1989) minmax EU model for ambiguity-aversion, which was shown to outperform various competing models in Hey et al (2010) for single-stage decision problems, albeit with more restrictions on the set scriptB .…”