Socioeconomic status and judicial disparities in England and Wales, 1870-1910

In a criminal sentencing system based on optimal deterrence, groups will receive different average sentences based on disparities in conviction probabilities, with longer prison sentences balancing more frequent acquittals. A taste-based model makes no such prediction. I compare these models using trial records from England and Wales in 1870, 1883, and 1910, years with extensive jury trials and broad latitude in sentencing for judges. In the earlier years, higher status defendants accused of property crimes receive substantially longer sentences, with no such difference for violent crimes. In contrast, higher status defendants are less likely to be convicted at trial, both when accused of property crimes and of violent crimes. In a Heckman selection empirical model, selection into sentencing has a large effect on the magnitude of these sentencing patterns. For property crimes, there is negative selection on unobservable traits, possibly explained by using imperfect proxies for social class. I conclude that an optimal sentencing and deterrence model is more supported for property crimes than for violent ones, with judges balancing disparate rates of conviction from juries.