Economists perceive crime - at least, crimes such as theft - as the result of a simple cost/benefit analysis by the perpetrators. A potential criminal - and we're all potential criminals - simply weighs the expected benefit of a particular crime against its expected cost, and if the former is greater than the latter the crime will be committed. The expected benefits are of course the expected payoff from doing the crime - the value of the stuff you steal given the probability that you succeed. The expected costs are the costs of arrest, trial, and incarceration given the probability of all of those events.
So, when one goes about thinking how to reduce crime one can find ways to reduce the expected benefits or the increasing he expected costs. For the latter, this is done by increasing the probability of arrest (more cops or more power to those cops), and by increasing the probability of conviction and/or the length of incarceration. In other words, if you're more likely to get arrested, more likely to get convicted, and more likely to get a longer prison sentence then you're more likely to be deterred from committing the crime.
However, the cost of imprisonment isn't uniform across individuals. While there might be some rough baseline psychological cost of being locked up, approximately equivalent across the board, the cost is largely a function of our alternative opportunities. That is, if Ken Lay is locked up for 10 years he presumably gives up 10 years of travel, big homes, nice restaurants, and generally all the finer things in life. If Random Poor Criminal gets locked up for 10 years, he/she is sacrificing far less.
Now, none of this necessarily ignores ethics - the psychological costs of violating your own ethical code can be part of the costs of committing a crime - but the point is that even with identical ethics, the cost/benefit tradeoff can differ substantially across individuals. In particular, poorer individuals with less to lose will be more likely to commit crimes.
So, the bleeding heart liberal approach to all of this is to observe that one way to increase the expected cost of committing a crime - and to therefore increase deterrence - is to increase the legitimate economic opportunities and more generally improve the lifestyle of the poor. From society's point of view, this isn't simply an issue of niceness - incarcerating people is expensive, as is maintaining an expensive police force. Whether backed up by evidence or not (and I'll admit I'm not too up on this research generally ), the liberal perception is that a way to reduce crime is to improve the legal opportunities for advancement of poorer individuals by improving education, etc..
Talk Left alerts us to a study claiming that Chicago's crime problem is related to its lack of affordable housing. I can't comment on the quality of this study, but it obviously fits nicely into my bleeding heart liberal biases. Adequate and stable housing is a precondition for access to the economic system in this country (as is, increasingly, things like an email address). To the extent this kind of research holds up, we may find that using public money to improve the lot of the poor is a more efficient way to reduce crime than the standard deterrence - the potential for long prison sentences.
