Is one judging head the same as three : a natural experiment on individuals vs teams

This paper utilises a natural experiment to investigate the differences between decisions taken by individuals and teams. In 2012 a judicial reform in Greece replaced three-member judicial panels with single-member ones. The assignment of judges to panels is exogenous, via a lottery. The assignment of defendants to panels is based on a date cut-off. Crucially, cases are tried by both types of panels in parallel. Our dataset contains 1723 observations, the universe of decisions regarding drug trafficking offences over 18 months from the court of Athens. We find that single-member panels are significantly more likely to convict a defendant. The length of sentences does not differ by panel-type but the variance of sentences imposed by single-member panels is lower than that of three-member panels, in line with a model of group polarisation due to information aggregation. We complement our analysis with an original survey of 142 judges that sheds more light on the mechanisms driving our results