Correlation Concern

TL;DR

This paper introduces a model where decision makers account for uncertainty in variable correlations by evaluating options under multiple correlation scenarios and choosing based on the worst-case expected outcome.

Contribution

It axiomatizes a novel decision model that incorporates correlation perception uncertainty, extending traditional expected utility frameworks.

Findings

The model captures decision-making under correlation ambiguity.

It provides a formal basis for worst-case evaluation in correlated variables.

The approach offers new insights into choices under uncertainty.

Abstract

In many choice problems, the interaction between several distinct variables determines the payoff of each alternative. I propose and axiomatize a model of a decision maker who recognizes that she may not accurately perceive the correlation between these variables, and who takes this into account when making her decision. She chooses as if she calculates each alternative's expected outcome under multiple possible correlation structures, and then evaluates it according to the worst expected outcome.