Events

Associate Professor
Department of Economics
University of Toronto

We propose a sharp test to assess the validity of the exclusion and monotonicity assumption in judge leniency designs. Our sharp test exploits all the relevant information in the observed data distribution to refute the model and will not make discordant recommendations. When the validity of the design is rejected, we show that a variant of the marginal treatment effect (MTE) can be identified under weaker partial monotonicity and partial exclusion assumptions. We apply our test to the Philadelphia court data studied by Stevenson (2018) and demonstrate that it outperforms existing non-sharp tests by significant margins in simulation studies.