Credibility of propensity score matching estimates. An example from Fair Trade certification of coffee producers

Propensity score matching (PSM) is an increasingly popular method for evaluation studies in agricultural and development economics. However, statisticians and econometricians have stressed that results rely on untestable assumptions, and therefore, guidelines for researchers on how to improve credibility have been developed. We follow one of these guidelines with a data set analysed by other authors to evaluate the impact of Fair Trade certification on the income of coffee producers. We provide thereby a best practice example of how to evaluate the credibility of PSM estimates. We find that a thorough assessment of the assumptions made renders the data we use not suitable for a credible PSM estimation of the effects of treatment. We conclude that the debate about the impact of Fair Trade certification would greatly benefit from a detailed reporting of credibility checking.