Cold Turkey vs. gradualism : evidence on disinflation strategies from a laboratory experiment

Disinflation can be implemented gradually or via Cold Turkey - an immediate change of policy - with the latter being mainly recommended by theory and empirical literature. But Cold Turkey may only be superior because it is endogenously selected for favorable environments. To eliminate this endogeneity and to disentangle the credible push through of a disinflation policy from ex-ante credibility, I run an experiment where a central banker has to decide for a disinflationary strategy and four forecasters try to coordinate on it. The design abstracts from any rigidities and provides full information so that Cold Turkey is the Nash equilibrium. But Cold Turkey fails to be the most successful strategy because forecasters react sluggishly due to limited reasoning. Cold Turkey does not speed up learning or increase reasoning, is less successful and is reversed more often.