Authors: Godfrey Keller
Publish Date: 2006/07/21
Volume: 33, Issue: 2, Pages: 263-269
Abstract
In models of learning by experimentation that exhibit signal dependence a benchmark using a passive learner has been proposed The use of this benchmark is flawed – first passive learning does not disentangle the effects of knowing that beliefs as well as other state variables might change and we address this issue directly by introducing a naïve learner Secondly and more tellingly passive learning does not do what it is supposed to do namely help measure the gains from active experimentation the naïve learner enables us to illustrate this point in the context of a particular exampleThanks for helpful discussions are owed to Donata Hoesch Thomas Mariotti and especially Margaret Stevens and for constructive comments to an anonymous referee A previous version of this work appeared as Discussion Paper 223 Department of Economics University of Oxford with the title “The inappropriate benchmark when beliefs are not the only state variable”
Keywords: