Welfare economics and bounded rationality: the case for model-based approaches

What is it about?

This paper looks at the issues for a policy maker who notices inconsistent choices made by agents. If said agent has imperfect attention which stops it from acting rationally, the relation R that connects x with y, and shows that x is preferred if x is chosen over y, no longer holds. The agent may have chosen x because they truly prefer x to y, but also because they might not have paid sufficient attention to y. Thus, the policy maker must first check that the data fits a specific model of imperfect attention and secondly attempt to infer from data whether it was true preference or merely inattention that led to the agent’s choice. While the paper does briefly discuss the model-less approach, it dismisses it in favour of the model-based approach, which it explores much more in depth. The paper states that a model-based approach must use models that are testable and that equate a postulated, unobservable choice procedure to the agent’s observable. It denotes the advantages of a model-based approach as information efficiency, a higher degree of precision in understanding the true connection between choices and welfare, and a greater scope than that of a model-less approach which relies on a greater number of assumptions. Having said this, the paper also reveals three types of failures and suggests ways to address each one. In type I failure, two or more are undistinguishable in terms of choice. In type II failure, only one model is compatible with the data, but there is more than one possible interpretation of the model’s outcome. And type III failure is when there are multiple preferences that fit an identified model structure.

Featured Image