Relative judgments

This paper presents a theory of relative judgments, suggesting that judges evaluate individual cases on the basis of how those cases are ranked in comparison to the other cases in their caseloads. Consequently, judges view a case more severely when their caseloads contain milder cases and more leniently when their caseloads contain graver cases. The paper develops a novel empirical identification strategy that exploits the properties of caseload distribution under random assignment of cases as a source of exogenous variation in judicial exposure to gravity. Using sentencing data, I construct a matched sample of judges randomly located at different ends of the caseloads distribution and demonstrate the existence of relative-judgment bias in their decisions. Judges exposed to lower levels of criminal gravity order longer sentences and are more likely to use the aggravated sentencing guidelines range or depart above the sentencing guidelines recommendations than judges exposed to higher levels of criminal gravity.