Matt Yglesias recently tweeted a link to this data from the Metropolitan Police Department of the District of Columbia, showing the city's number of homicides and the clearance rate of those homicides for every year from 2002 through 2011. The really striking part, of course, is the 95.4% closure rate in 2011. That's insane! At the beginning of the period the rate was just 55%. The percent of murders that go unsolved has, in other words, been divided by nine over the last decade in D.C. How did that happen? Did the police get nine times as effective? Or twice as effective, if we look at the yes-solved percentage? Well, that's a somewhat complicated question, and it's tough to know which of the effects at play here are to the credit of the police. But it's clear what is going on here: the MPDC has a certain murder-investigating capacity. Based on these numbers it's somewhere between 110 and 140 cases per year. But the murder rate has been falling like a stone, from 262 homicides in 2002 to just 108 in 2011. The result is that the same total amount of murder-clearing represented a much lower percentage in 2002 than it did in 2011.
Or to put it another way, I had Microsoft Excel calculate trend lines for the plot of both solved and unsolved homicides against total homicides. There's a positive correlation in both cases, of course, but the comparison between the two is an interesting one. Every extra homicide, according to these data, will on average correspond to an extra 0.325 cleared homicides. And the total homicide rate explains 79% of the variance in number of cleared homicides. But every extra homicide, of course, leads to .674 uncleared homicides (after all, it must be one or the other, so the two slopes add up to 1). In this case, though, the R-squared value is 94%. That is to say, based on the data the total murder rate provides an explanation for the changes in the number of unsolved murders which is very nearly complete. We need look basically no further to find out what determines the number of murders which go unsolved in D.C. every year. Oh, and these simple linear trend lines suggest that, if no murders were committed, 63 of them would be solved anyway. That's sixty-three out of zero murders.
Now, obviously that's not right. Both figures are bounded above by the total number of murders and below by zero. And that means that extra reductions in total murders would be unlikely to result in extra reductions to the number of unsolved murders. After all, that number in 2011 was... five, as best I can tell. In a city of over two-thirds of a million people, only five murders went unsolved that year. That's the point at which returns will be diminishing, though of course the main marginal gain of one less murder is one more person not being murdered. But we know that one of the biggest factors in actually deterring crime through a penal code is the certainty of punishment, not the severity. If you know that there's a 95% chance that you'll get caught if you commit a murder, you'll be a lot less likely to commit it than if you think it's almost a coin-flip whether you'll get away, especially because nearly everyone will see themselves as being in the more competent half of the population and will therefore believe that they will do the murder well enough to be one of the ones getting away. Five percent sounds a lot more like zero, though. So what we should see, I think, would be a pretty strong feedback loop, where once the clearance rate gets close enough to 100% it will help push the overall murder rate down, even though that won't make clearing the last handful of murders any easier.
And that's the link to what I mentioned in my title, what you should do about an impending crime wave. Let's say we accept the Kevin Drum hypothesis that the mid-20th century crime wave was largely the consequence of environmental lead poisoning, which I think there's good reason to do. And let's also say we have a time machine and can go back to the 1960s, or even the 1950s, and warn them about the impending lead-induced crime wave. What policies would it be appropriate for the country to implement, in anticipation of a surge in violence and criminality that there is good reason to believe is the fault of an environmental neurotoxin? There are, I think, serious justice problems with the approach we actually took. If a crime wave is the result of, I dunno, people just sort of deciding to become more evil, you can see the case for massively ramping up punishments. But it seems decidedly unfair to impose upon a population a regime of mass incarceration for multiple decades if the population is basically the victim of industrialization.
What makes more sense, however, is to do everything possible to keep the certainty of punishment as high as possible. Because while there are justice concerns about punishing one lead-induced criminal more severely because he and everyone else has been lead-poisoned, their crimes are still crimes. And a big part of the problem with a crime wave is that it can have the exact opposite feedback loop that I was saying might start happening in D.C.: as crime goes up, it will become harder and harder for the police to keep up and solve as high a proportion of their cases, which will substantially reduce the marginal incentive people have not to commit crimes because they might get caught. So if you knew that the lead crime wave was coming, the appropriate response would be, I think, to just shovel a lot more resources toward the criminal justice system. More police. Not necessarily more aggressive police, which I think we got as somewhat of a substitute for the "more" part, and certainly not racist police whose attempts to enforce the laws will inflame racial tensions and make the whole situation worse. But also more courts. More judges. More prosecutors. Oh, and more public defenders, because again we are trying to keep certainty of punishment up without reducing fairness, and also because the more likely you are to be punished for crimes you didn't commit, the lower your incentive not to commit crimes, ceteris paribus. If you knew the crime rate was about to double, you might want to consider doubling the resources of your criminal justice system, from top to bottom. You want, in other words, to boost your investigative capacity, the thing that seems currently to sit a little over 100 murders per year for the D.C. police. And if you can keep up with the crime wave, and keep the clearance rate from plummeting precisely as the overall crime rate rises, you should be able to cut off the feedback loop and keep the deterrent effects in place.
What we did, instead, was to simultaneously jack up punishments and reduce the procedural fairness safeguards designed to keep the chance of being wrongly convicted as low as possible, without particularly boosting the system's resources. That's wrong in every conceivable way. It permitted the odds of getting away with any given crime to go way up. It simultaneously weakened, especially for black people, the causal link between not committing crimes and not being punished for them; in the black community in particular, this may have seriously undermined the incentivizing effects of having a criminal justice system at all. (The incentive effects of mass wrongful conviction is, I think, an under-appreciated point in public discussion of crime issues.) And the increased sentences had an assortment of unfortunate side-effects, but most notably, I think, are the effect of increasing the sense of unfairness in the system (especially when combined with the first two problems of the overall approach) and giving prisons a massively stronger ability to act as engines of criminality.
Basically, we did it exactly wrong, in ways that were probably mostly about racial hatred. In an awful lot of ways that's the story of our nation's entire history, isn't it?
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment