The perils of peer review in economics and other sciences

The quality of researchers' work in economics and other sciences is generally evaluated through a system of peer review. In an experimental test it is shown that the peer review system can be very inefficient by creating a bias towards incremental development of existing methods and against exploration of new methods. Previous studies on this issue have put the blame on biases in individual judgement. Here the inefficiency is shown to occur even when researchers are rational and have perfect information as a result of strategic uncertainty about the extent to which other referees reject new methods. The experiment also shows that the bias generated by peer review can be alleviated by shifting some quality evaluation to non-researchers, even if these are poor at discerning quality.