Does the In-Group Bias in Preferences Get Crowded-Out/In?

This paper reports on an online experiment that tests whether the in-group bias in preferences is crowded-out/in when a policy is introduced that materially encourages in-group biased behaviour. The bias in preferences is crowded-out in the aggregate and the policy is not successful: it does not change the extent of the bias in the aggregate. However, we also find that there is individual heterogeneity in our subject pool and while a subset of subjects clearly reveal crowding-out of the in-group bias preference, there is another subset where this preference is crowded-in by the policy intervention. This suggests that whether there is crowding-out or crowding-in, in general, in the aggregate depends on the relative size of these two subsets of the population