Incentives Don't Matter For Criminals?
Now, since I don't have the full study available and is only able to rely to Slate's account of it and the brief NBER description, I am unable to respond as effectively as possible.
However, even with this limited information, the study seems flawed for several reasons. The study bases its analysis on behavioral changes from the weeks before to the weeks after the 18th birth day of criminals. Yet first of all, punishment aren't really increased that much on account of that 18th birthday. The risk of prison rises from 3% to 17%, meaning that it is still a very low risk, especially since this is only the risk of getting imprisoned if you are caught. If you add in the fact that in most crimes, the criminal won't get caught, the risk of getting caught remains very small. And so since the increase in punishment is fairly small, the effect could hardly be expected to be anything but small.
Secondly, they did in fact find that arrest rates fell from the age of 17 to 19, only they didn't find any acceleration in the rate of decline the first weeks after the 18th birthday compared to the weeks before. Yet looking at only a few weeks is potentially very misleading as mere coincidences could have skewed the arrest rates over such short time periods. Had they tried to compare arrest rates the hours before and after, then the potentially misleading nature of the data would have been even more evident, but this exist here too (albeit to a lesser extent).
Finally, I find the conclusion of the authors that their results show criminals to be "impatient" because they keep committing crimes after they turn 18 to be curious, to say the least. To say they are impatient implies that they rationally should wait, but wait for what? For a time machine that will make them 17 again?