I mentioned yesterday that we do not know whether capital punishment is a deterrent or not. What about imprisonment sentences? That is much harder to establish without very precise coding of convictions in multiple jurisdictions with different sentences for the same crime. Finding or establishing such a dataset should be very difficult. But maybe we can find some partial answers in indirect ways.
Nadia Campaniello, Theodoros Diasakos and Giovanni Mastrobuoni use an interesting natural experiment in Italy. There, the parliament occasionally decides on mass pardons to reduce jail crowding. When such proposals are being discussed, suicide rates in Italian prisons drop. That means clearly that prisoners do not like ex-post being in prison, and sufficiently to make life depend on it. If this matters also ex-ante (before they head to jail and in particular before the decide to commit a crime), we should see a deterrence effect. But this may be a big if.