Optimal policy to influence individual choice probabilities, working paper

This paper presents a model in which government may affect outcomes by manipulating individual choice probabilities through the design of the domain of choice or the use of fiscal instruments. Such manipulations are ineffective when individuals are perfectly rational, provided all alternatives are permitted. However, even a small deviation from perfect rationality is shown to call for policy that substantially manipulates choice probabilities. This policy aims to lend weight to alternatives preferred by individuals who are prone, more than others, to make mistakes. At very low levels of rationality, when choices are largely random, it is always socially optimal to entirely eliminate individual choice in order to prevent the errors generated by such choice. It is better to impose one alternative that is not the preferred one for some individuals instead of inducing a completely random draw by everybody