Is three strikes working?

I really don't know. But this isn't the test:

News that there are now 132 violent and sexual offenders who have been convicted of a 'Strike' offence and given their first 'Strike' warning comes as no surprise and simply serves to highlight that the 'Three Strikes' policy is working exactly as ACT intended, ACT New Zealand Justice Spokesman Hilary Calvert said today.

"These offenders now know their behaviour will not be tolerated, and will be well aware of the consequences of further similar offending," Ms Calvert said.

Here's the test. Take the set of all offences going back a few years. If there's a decrease post Three-Strikes coming into effect in those offences on the strike list relative to non-strikeable offences, Three-Strikes is effective. If not, it isn't, or at least isn't so far.

If New Zealand follows American experience, we ought to see a decrease even in first-strike offences, but a small increase in non-strikeable offences as offenders substitute into categories that have lower long-run cost. We might also see a severity shift within the list of strikeable offences: fewer of the lowest severity crimes (substitution down into non-strikeable offences) and a higher proportion of the more severe strikeable offences (as the granting of a first strike acts as a level shift in the severity of future sentences, though marginal deterrence is maintained within the second strike).

Give it another year or so, then run the test. If nobody else has done anything on it by early next year, I'll assign it as an honours project.

I prefer staking out what an appropriate test would be before taking it to the data.