How much does responsibility matter in fairness measurement?

Empirical evidence suggests that social acceptance of redistribution depends on whether income differences result from preferences (of which individuals are responsible) or from circumstances. We propose a new empirical method that measures the importance of preferences in the distribution of welfare in the context of tax reforms. We compare two types of Compensating Variation: the standard CV and a new one (CVcirc), which is computed assuming that individuals differ only in circumstances. To obtain these metrics, we first estimate a structural job choice model that allows us to take the preferences/circumstances dyad into account. We then use the estimated parameters to compute our two metrics, leveraging a tax reform and applying a simulation approach `a la McFadden (1999). Implementing our method with Norwegian data, we find that both welfare metrics display a very similar distribution, except at the very top of the households’ income distribution, suggesting this is where responsibility matters.