Ambiguity and Probabilistic Information

Experiments detecting ambiguity aversion often rely on the assumption that probabilities are exogenously given for some uncertain events. However, the canonical models that accommodate ambiguity into economic theory, such as the maxmin expected utility (MEU) and Choquet expected utility (CEU) models, are purely subjective. These models do not specify how subjects could incorporate exogenous probabilities into decisions. We study two approaches for embedding exogenous probabilities in the context of the thought experiments suggested by Mark Machina. We show that Machina’s choice behavior entails fundamentally different consequences for the ambiguity models mentioned; although it violates the CEU model, it is consistent with the MEU model. For the latter model, Machina’s experiments can test whether individuals adhere to expected utility for prospects whose consequences occur with the exogenously given probabilities. This paper was accepted by Manel Baucells, decision analysis.

On the falsifiability and learnability of decision theories

We study the degree of falsifiability of theories of choice. A theory is easy to falsify if relatively small data sets are enough to guarantee that the theory can be falsified: the Vapnik–Chervonenkis (VC) dimension of a theory is the largest sample size for which the theory is “never falsifiable.” VC dimension is motivated strategically. We consider a model with a strategic proponent of a theory and a skeptical consumer, or user, of theories. The former presents experimental evidence in favor of the theory; the latter may doubt whether the experiment could ever have falsified the theory. We focus on decision‐making under uncertainty, considering the central models of expected utility, Choquet expected utility, and max–min expected utility models. We show that expected utility has VC dimension that grows linearly with the number of states, while that of Choquet expected utility grows exponentially. The max–min expected utility model has infinite VC dimension when there are at least three states of the world. In consequence, expected utility is easily falsified, while the more flexible Choquet and max–min expected utility are hard to falsify. Finally, as VC dimension and statistical estimation are related, we study the implications of our results for machine learning approaches to preference recovery.

Testing Ambiguity Models through the Measurement of Probabilities for Gains and Losses

This paper reports on two experiments that test the descriptive validity of ambiguity models using a natural source of uncertainty (the evolution of stock indices) and both gains and losses. We observed violations of probabilistic sophistication, violations that imply a fourfold pattern of ambiguity attitudes: ambiguity aversion for likely gains and unlikely losses and ambiguity seeking for unlikely gains and likely losses. Our data are most consistent with prospect theory and, to a lesser extent, α-maxmin expected utility and Choquet expected utility. Models with uniform ambiguity attitudes are inconsistent with most of the observed behavioral patterns. (JEL D81, D83, G11, G12, G14)

THE RE-OPENING OF DUBINS AND SAVAGE CASINO IN THE ERA OF DIVERSIFICATION

In Dynamic Programing, mixed strategies consist of randomizing the choice of actions. In some problems, such as portfolio management, it makes sense to diversify actions rather than choosing among them purely or randomly. Optimal betting in casinos and roulette by a gambler with fixed goal was studied by Dubins and Savage [9] and their school without the element of diversification (betting simultaneously on different holes of the roulette), once it was proved (Smith's theorem - Smith [16], Dubins [8] and Gilat and Weiss [10]) that diversification does not increase the probability of reaching the goal. We question the scope of this finding, that was based on the assumption that the holes on which gamblers can bet are disjoint, such as 1 and BLACK in regular roulette. A counter example is provided in which holes are nested, such as 1 and RED. Thus, it may be rational for gamblers with a fixed goal to place chips on more than one hole at the table.This note is related to a joint work with Michèle Cohen on the preference for safety in the Choquet Expected Utility model.

Ambiguity Models and the Machina Paradoxes

Machina (2009) introduced two examples that falsify Choquet expected utility, presently one of the most popular models of ambiguity. This article shows that Machina's examples falsify not only the model mentioned, but also four other popular models for ambiguity of the literature, namely maxmin expected utility, variational preferences, α-maxmin, and the smooth model of ambiguity aversion. Thus, Machina's examples pose a challenge to most of the present field of ambiguity. Finally, the paper discusses how an alternative representation of ambiguity-averse preferences works to accommodate the Machina paradoxes and what drives the results. (JEL D81)