Fairness Preferences and Default Effects

Abstract

An influential subset of the literature on distributional preferences studies how preferences condition on characteristics such as workers' relative productivity. In this study we establish that there are default effects when such conditional fairness preferences are measured using the "inequality acceptance" method. Depending on the default, implemented inequality decreases by over 65% and cross-country differences are not observed. To organize the data, we develop a simple framework in which agents form a reference point based on a combination of the distribution suggested by their fairness ideal and the default. We use this framework to illustrate that choice data from different defaults is needed to separately identify the fairness ideal and effect of the default, and discuss best practices for measuring fairness preferences.