Most economists have a glib answer: The worse your legal options, the better crime looks. But the more I think about this response, the weaker it seems. Here's a striking fact about crime: A lot of it is almost never lucrative. There's little money in assault, drug possession, or drunk driving, to take some of the main offenses that land people in jail. The same goes for rape, and probably most murder too.
These observations bring to mind the famous Levitt-Dubner observation that a lot of drug dealers earn minimum wage. They have a "tournament" story - the superstar payoffs for the drug lords at the top of the pyramid attract massive entry at the bottom. But there's a much simpler theory: Dealing drugs - like most illegal behavior - is an inane strategy for escaping poverty.
What's my alternative? Crime is just one of many, many "social pathologies" that are over-represented among the poor: alcoholism, drug abuse, smoking, obesity, illegitimacy, etc. None of these are good escape routes from poverty. So instead of trying to explain why "poverty causes crime" or "poverty causes obesity," it makes sense to look for common causes of poverty and social pathologies.
Like what? In a paper just accepted by Kyklos, Scott Beaulier and I point to a simple candidate: irrationality. People who have biased beliefs about practical matters, and/or exercise poor impulse control, are likely to screw up their lives across the board. So it's hardly surprising that poverty and self-destructive behavior go hand in hand. Rather than being a natural response to poverty, a lot of crime can be seen as objectively self-destructive behavior that happens to have an unusually large amount of collateral damage.
Update: "Drug driving" was a typo; I meant to write "drunk driving." Sorry for any confusion.