Summary
Aim
To estimate the effect of bail decisions on the likelihood of receiving a prison sentence, failure to appear and offending on bail.
Method
A dataset of 42,362 first bail hearings taking place after the ‘show cause’ amendments to the Bail Act (2013) was constructed and linked to final case outcomes and offending data. Quasi-random assignment of bail magistrates with differing propensities to grant bail was used to address problems of selection bias and partial observability. Further analyses were undertaken to determine the proportion and characteristics of defendants who were sensitive to magistrate leniency. Robustness checks were conducted to determine the sensitivity of estimates to different specifications.
Results
The marginal effect of additional releases is an increase in the rate of offending from 2.3 per cent to 13.3 per cent, a decrease in the rate of imprisonment from 59.0 to 49.0 per cent and an increase in the rate of failure to appear from 2.1 per cent to 11.1 per cent for those defendants. Thus, remanding ten additional defendants increases the number imprisoned by one, and reduces the number of offending and failing to appear by 1.1 and 0.9 on average. These estimates are causal and net of differences in observed characteristics and selection bias, but applicable only to a subset of defendants whose bail status is sensitive to magistrate leniency. The likelihood of failing to appear and of offending on bail for these defendants does not exceed the general rate among those released on bail.
Conclusion
Taken together, the results show that bail refusal has a significant incapacitation effect on crime and failure to appear. These benefits should, however, be considered alongside the considerable cost to the correctional system and the individual arising from increased imprisonment rates. There is limited evidence for the influence of selection bias in regards to imprisonment but not in relation to crime or failure to appear.