Transcript of "Less Prison, More Police, Less Crime: How Criminology Can Save the States from Bankruptcy"

Opinions or points of view expressed are those of the author(s)/presenter(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice.

An NIJ Research for the Real World Seminar
Lawrence Sherman, Ph.D.
Wolfson Professor of Criminology, University of Cambridge; Director of the Jerry Lee Center of Criminology at the University of Pennsylvania

Introduction by NIJ Acting Director Kristina Rose.

April 21, 2010

Transcript

Kristina Rose: We're here at the National Institute of Justice, and it's wonderful to see such a large crowd. And it's wonderful to see so many folks that I don't recognize, which means people found it important enough to come from agencies outside of OJP, outside of DOJ, to come listen to this very special speaker that we have here with us today.

And before I get started I'd like to recognize just a few people in the audience. We have Mr. Jerry Lee. Jerry Lee is of the Jerry Lee Center of Criminology at the University of Pennsylvania. Thank you for coming here today. We also have Peter Neyroud. Where is Peter? Peter here is the head of the National Policing Improvement Agency in the United Kingdom. He happens to be stuck here in the United States right now, so we were fortunate to have him come to our lecture today. So thanks to both of you for being here, and all the rest of our distinguished guests.

Today's lecture is entitled “Less Prison, More Police, Less Crime: How Criminology Can Save the States from Bankruptcy.” And we have Dr. Larry Sherman here from the University of Pennsylvania … he is a Wolfson Professor of Criminology at the University of Cambridge in the U.K.

So in order to get us started today, I want to introduce Laurie Robinson, who'll be introducing Dr. Sherman. And for those of you who are outside of the Department of Justice and may not know Laurie, she's our Assistant Attorney General and was sworn in in November of 2009, having served in that acting capacity since January of 2009. Now Laurie was OJP's Assistant Attorney General throughout the Clinton administration, and we thank our lucky stars every day that she decided to come back and lead our agency, because she is really second to none in terms of her passion and her commitment to the work of the Office of Justice Programs.

And it just so happens that Laurie and Larry were colleagues at the University of Pennsylvania together, and so we feel very good about the fact we could reunite the two of you again today. So Laurie has a very important hearing tomorrow before the House Appropriations Committee, so I want to thank you, Laurie, for taking the time to be with us today. That's especially meaningful in light of your very tight schedule and important hearing tomorrow. So please join me in welcoming Assistant Attorney General Laurie Robinson.

[Applause.]

Laurie Robinson: Well, good afternoon to all of you, and I can assure you this is much more fun than prepping for the appropriations hearing. Well, I am actually very delighted, first of all, to welcome all of you here today, and secondly to welcome my very good friend and former colleague Larry Sherman to the Office of Justice Programs. Now I have to tell you that Larry wins the award for the person who made the greatest effort to get to Washington, D.C. He was in England two days ago and he did not swim here, but he close to … did that to get to OJP today. He took the ferry to Calais, driving to the port and taking his car across. Then he drove to Barcelona and waited breathlessly to see whether the plane was going to take off. The plane did take off; he said it was about a 50/50 chance. He took that plane to New York and then took the Amtrak down. As he was coming down he was revising the slides and e-mailing some of us the revised slides, so I don't think he's had sleep in two days. But despite that, he managed to get here. So Larry, we're awfully glad you did.

Now for those of you who may not recall, Larry, back in 1996, was the co-author, along with colleagues at the University of Maryland, of what I think is a landmark publication put out by OJP, entitled “Preventing Crime: What Works, What Doesn't, What's Promising,” which I think is really a bible in our field in kind of defining evidence-based approaches. Then in 1998 in a lecture at the Police Foundation, he talked about the notion of evidence-based policing … and that has gone on to spur a lot of discussion and debate also in the field, really drawing from ideas from medicine. And in 2000, went on to become one of the co-founders of the Campbell Collaboration. Larry is a past president of more associations than most of us could name if we were hard-pressed: The American Society of Criminology, the International Society of Criminology, American Academy of Political and Social Science, the Academy of Experimental Criminology, and the list could go on.

So with that, let me ask Larry Sherman to come and join us here at the podium. Thank you, Larry.

[Applause.]

Lawrence Sherman: Thank you, Laurie. That's wonderful, absolutely. Thank you. Well, thank you all for coming today, and special thanks to NIJ and to OJP for giving me the chance to talk with you about some of the most important issues facing our field. And increasingly it's apparent that what the cost of criminal justice may become is fewer school teachers, fewer nurses, fewer public services that many of us would deem essential, but which are in some ways more controllable than the costs of criminal justice. I just finished my slides this morning when I picked up the Washington Post and read that the forecast is between 100,000 and 300,000 teachers may be laid off over the next year to deal with the reduction in revenues that governments are experiencing. But with the state's heavy role in education, it's clear that the state cost of imprisonment has become a major part of decisions like that.

I learned today that the Federal Bureau of Prisons, operating the largest prison population in the country, is going to add another 7,000 inmates each year for the foreseeable future, way over capacity. I think that if we were to take a look at the entire portfolio of criminal justice and ask whether we've got our investments in the right place, we might rapidly conclude what a recent review by Dan Nagin and Frank Cohn has shown, which is that essentially, we own stock in a company called prison, where we don't know what the stock price is, we never get to see a dividend statement, but every year they send us a really big bill and it's bigger than the bill from the year before.

So here's my argument: that if criminal justice policy these days were based on the current evidence of cost-effectiveness, we would have less money spent on prison; we would have more money spent on police; and we would have fewer serious crimes with less total harm and fewer states threatened with bankruptcy.

We essentially don't know whether prison has a general deterrent effect, which is perhaps its best justification. We know more, but not enough, about the specific effect it has on the people who go to prison. But this is where this incredibly thorough review that Cohn and Nagin — comes in, because the weak evidence we have suggests that there is, in fact, no effect of imprisonment on the imprisoned. Or if there is, it's a net increase in offending. In general, we can point to other evidence that suggests that the incapacitation effect of prison is wasted on many, if not most of the people who are behind bars under the logic that they could be Willie Hortons. But in fact, most of the people behind bars are not Willie Hortons. They might commit more crimes if they were outside, but they wouldn't commit absolutely horrible crimes. Few people in prison are Willie Hortons, but many people who are Willie Hortons are getting probation or parole from the evidence we have, and really only Virginia has developed a risk-based sentencing guidelines framework. And that one could be criticized arguably because it focuses more on the fact of reconviction, or even the frequency, but not on the seriousness, which is really what the Willie Horton point is all about.

So the cost-effectiveness of the police is much better understood because of research funded by NIJ over the years, where the general deterrent effect of imprisonment, I'm sorry, of policing is clearly established by having randomized control trials, rigorous tests of the effects of police patrol when they're concentrated in hot spots. Increasing less rigorous but consistent evidence, the traffic and minor crime enforcement by police reduces crimes like robbery. And some evidence, perhaps the least compelling, that the more police we have, the less crime we have overall, although many of us would argue that that depends entirely on how the police spend their time, including whether they spend their time actually going after some of the most serious offenders for the purposes of getting them behind bars and keeping them there as long as possible because they are very dangerous people. On a much less severe level, the situational crime prevention strategies for problem-oriented policing, as recently reviewed by David Weisburd and others, have proven successful in at least getting clear reductions in those problems for communities' central issues for community policing.

Perhaps less encouraging, but actually very important for us to know is the recent review on the prosecution of juveniles where police play a critical role, and in fact other evidence that shows that actually police might be preventing more crime if they arrested fewer people, or prosecuted fewer people, and actually looked at other kinds of resolutions which are historically quite common in policing but which were abandoned under the legalistic framework of recent years.

So the question of whether having less prison and more police would lead to less crime really depends on this notion that what we ought to be doing with our entire portfolio is trying to lock up the worst and manage the rest. Locking up the worst in order to have less serious crime, and managing the rest in order to have less cost for people who remain in the community, but who can be given more crime-preventive attention by police, probation and parole than would be the case under the current system in which there is an increasing … caseload for probation and parole people, and in which the police probably don't do nearly as much to pull the levers in relation to probationers and parolees as some of the demonstration projects suggest they could do.

If we have people who are active offenders but only offend in hot spots, then more policing in hot spots could actually reduce their rate or frequency of offending. If we have problem-oriented policing, altering the factors that cause crime in those places, then they would also be less likely to offend. So this perspective brings together the causes of criminal events with this idea of very active offenders, which is still separate from the third and critical question of the most serious offenses and who's going to commit those offenses, which is arguably the role — the best role that prison can play within the total portfolio. And then if we add crime victims into the equation — Why would we want to do that? They're only the people suffering all this harm — the evidence is pretty clear that they would much rather have diversionary opportunities like restorative justice conferences that are also effective in reducing crime as opposed to things that simply impose retribution for the sake of retribution.

Well clearly there's a lot of people involved in all of this. The criminologists at work, who I will be talking about and attempting to integrate today, are coming at this from many different angles, but I think there really is something emerging that comes close to being a policy prescription.

And so the outline is, I'd like to go into the premises behind that policy prescription, then introduce some key concepts that I think are essential for understanding the total portfolio of criminal justice that's actually split across three levels of government in the United States, in which the police are paid for at the local level, prisons are paid for at the state level, and that the federal level is the one that has the capacity to generate the information, and potentially the ideas and leadership, that could integrate these portfolios around these concepts of a policy system where you begin to see which buttons to push: a new idea for thinking about crime beyond specific offenses, but rather a bottom line for crime which is an index that looks at the harm that crime causes. Then focusing everything we do in relation to the cost of criminal justice around the risk of serious crime that constitutes that harm index, as well as all of that coming together in what might be the beginnings of a general theory of crime, prisons and police. I'll briefly touch, time permitting, on evidence on prisons and policing that lies behind all this, but then try to come very clearly on the question of who can push the buttons at the end of this analysis.

So let's start with the premises. And that would mean that what criminal justice needs is for people wanting criminal justice to protect them from their loss of liberty. They also want offenders held accountable for crimes, and they need to have criminal justice cost taxpayers as little as possible. Arguably that's the one where we're really falling down. And in order to do better at those things, we need a system of criminal justice that produces consequences for every decision that are predictable, based on good evidence, and chosen democratically by managing the system in ways that we intend it to perform.

So the key concepts here are of a policy system where we know what buttons to push, the Crime Harm Index, the risk-based policies and a general theory.

Now what's this idea about push-button systems?

Recently Paul Krugman suggested that before John Maynard Keynes, economics really didn't have a way of creating policy, of identifying the buttons that government could use to get under control things like mass unemployment during the depression. And that in the period before Keynes, there were case studies in economics, there were trends, histories, lots of micro-level things, theories, explanations, but there really weren't any policy prescriptions. There was no identification of what government could do to get the economy started again in the 1930s.

But after Keynes presented a very different way of looking at this, a kind of switch in perspective, what happened to economic policy is that government accepted the idea that it had some key buttons it could push and that pushing could help a lot, which led to lots of work about which ones to push, how to push them or when to push them, with what effect, and how those buttons affect each other.

So if we think about this in terms of criminology, at the moment, our case-by-case crime policy, by which we're adding 7,000 people to the federal prison system every year, is a kind of laissez-faire economics in which the assumption is nobody in the executive branch can do anything about this, these are judicial decisions, these are decisions that have their own logic, and it would be wrong for us to interfere. Well, it might be wrong to interfere at the case-by-case level, but when do you stop? When 50 percent of the GDP is going into criminal justice just because that's the way the independent judicial process would push us? I don't think anybody would want to see that eventuality.

And yet, unless we accept the idea that we can't leave the system at a laissez-faire, case-by-case basis to do whatever it wants, but we have to be interventionist, we will not be able to figure out this relationship among the different parts of the criminal justice portfolio, and the question of whether we're better off investing as much as we have in prisons or whether moving some of that to policing or probation and parole would make more sense.

And in fact, you could say the evidence suggests that we are pushing the wrong buttons on a case-by-case basis with too much prison for too little benefit, not enough optimizing of what the police do, and not enough system policy for pushing the right buttons that could be guided by a general theory. Now that was, of course, very important for Keynes.

Central to this is an idea that I think economics has done very well with, both the gross domestic product and the consumer price index. It's a way of taking a large amount of complex indicator and information data that comes together in a single bottom line. It gets widely reported. It has big effects on how people pursue business and do lots of things. We could do that in crime rather than getting people confused about the homicide rates going down but the robbery rates going up; but the National Crime Victimization Survey is showing one trend, and the police data is showing another.

What we need, I think, is a consensus of experts and practitioners to do something, not what has been done since the Wickersham Commission in 1929, which invented the Uniform Crime Reporting crime index, but rather something much more like what economists have. What would that be? It would be a challenge to the Bureau of Justice Statistics to develop it and start publishing it on a regular basis.

So I hereby suggest to those from BJS here today to accept that challenge. I also suggest to the United States Senate that they might want to confirm the director who's been nominated by our President to lead the Bureau of Justice Statistics. Oh and by the way, maybe the same for NIJ, but I digress.

[Laughter.]

Let us think what a director of NIJ and BJS could do jointly in developing a formula that would be weighted according to the seriousness of each crime so that when we look at all of the offenses that somebody let out of prison might commit, we're not just talking about, did that person get reconvicted once or more. We're not just talking about whether that person committed a property offense or a violent offense, or both. But we're actually adding up a cost or a harm level for each offense, which we know is going to be much higher for murder than it's going to be for burglary, and the burglary is going to be higher than it is for breaking into the window of a car and taking a package worth $10 out of the back seat of the car; of course it costs $500 for the window.

The British have actually estimated the cost of these different kinds of offenses and they're now publishing some cost indexes of crime in their annual British Crime Survey, which have actually shown reductions in the costs of crime in recent years. But we don't even have to do it with money. I know that money as a cost of crime is a very offensive thing to some people. They don't want to hear about the idea that a murder is worth $5 million. They'll say, “How can you ever calculate the loss of a life?” So it could just be done with public opinion data, which is quite extensive now in terms of the relative seriousness of crime. And to develop this reliable set of crime categories into a single index would help a lot. To do what? Well, among other things, to figure out how to invest our portfolio. Because we could compare the impact on the Crime Harm Index from investing in prisons versus investing in police.

Among other things, a crime harm index would get us beyond these assumptions of neo-classical deterrence, and indeed economics, that all crimes can be counted in terms of equal harm: offenders committing crimes of equal harm, prisons preventing equal harm per inmate, and so on.

What we would get is something better than the very bad push-button criminology that you can find on the Internet. Things like this graph purporting to show that the drop in the incarceration rate for violent crime in the '60s was followed by this huge run-up in violent crime that was only gotten under control by substantially increasing the incarceration rate.

If you look at it a little differently, that is in the total incarceration rate, we see that we've had this massive one-time and never-ceasing run-up from the mid-'70s in the incarceration rate across the U.S.

At the same time that for much of that period, from '75 is when the increase began. By the mid-'80s you see potentially no influence of that growth and incarceration on the violent crime rate.

But what you do see in the '90s when those rates do start to drop from the British, sorry, from the National Crime Victimization Survey, is an alternative interpretation, which is that increasing numbers of police, or perhaps better software, better research from NIJ, different kinds of policing that were more sensitive to the risk of crime by time of day and location, that all of that was making the kind of difference that we've seen in terms of the steady drop in violent crime with, of course, many social factors including, surprisingly, immigration which, contrary to politics, seems to be lowering the crime rate in many big cities around the country.

So, in other words, we need something a lot more complicated than looking at prison and police, and prison and crime in totality, because prisoners and potential prisoners vary highly in terms of how dangerous they are. And the incapacitation effects on the Crime Harm Index would vary enormously depending on which people we're putting behind bars. The imprisonment effects would also vary by their age, their first offense and so on with huge impacts for policing in terms of how to redirect police resources in a way that would have maximum impact on the Crime Harm Index.

So by drawing us to the risk of serious harm, we would see the capacity to develop risk-based policies and not just kind of one-size-fits-all policies for every crime that comes to our attention, but a seriousness grading, a kind of red, yellow and green approach to expending criminal justice resources in the kind of triage that's obviously necessary in our current cost environment for every part of the system: for police, for prosecutors, for pre-trial people, for sentencing, for prisons, parole and probation.

And if we think about hot spots of policing, which the Japanese National Police Research Agency has so brilliantly posted here in three dimensions, where crime is heavily concentrated in a small portion of Tokyo, and incidentally which the Japanese National Police Agency has shown absolutely no interest in using in their allocation of patrol resources in Tokyo, at least not the last time I heard.

This is actually a metaphor for the risk of crime at every stage of the system. This is in relation to patrol, but if we think about that as a population of offenders that's out there and that a prosecutor is going to get an arrest population where some of the people are very dangerous, sticking up out of that landscape, and most of them are low-risk. And right now the policy, as I understand it at the U.S. Department of Justice, is maximum charge on everybody. One size fits all. But this metaphor suggests that these are people of very different sizes, in terms of how much harm they can cause to our society.

So, if we want to then be responsive to this kind of heterogeneity in the population, we would accept one of the main insights of economics from a century ago, Vilfredo Pareto, identifying not a normal curve but a hockey stick by which, and those of us who remember the RAND Corporation report on selective incapacitation, we see that maybe 10 or 20 percent of any population is producing way over half of all of the bad things or good things if you're a business, this is what the frequent flyer distribution looks like, or it did before last week. And it's a well-known distribution that helps businesses maximize the money they make. It helps allocate health care in emergency rooms where they use a triage system, where a huge percentage of the people who die are going to be in a tiny percentage of the people who walk in the door. But they have systems for assessing that risk almost immediately and doing everything possible, and when they fail they get into a big amount of trouble.

But our criminal justice system, by and large, is not using the Pareto curve; it's using the good old normal distribution which was, after all, discovered by a criminologist in 1815, but I think we've done something more important, lately, than that.

We've done something that will help us deal with the Willie Horton problem, which is a problem that I define not in terms of Michael Dukakis losing the 1988 election, but rather in terms of these rare, extremely serious, highly damaging crimes that can just blow away all sorts of judgment at the routine case level. And what I think may have happened in the years since Willie Horton is that every time a case like that comes down the pike, such as the parolee who murdered a Philadelphia police officer in late 2008, the system just tries to defend itself from the public blame and in the case of Pennsylvania, the parole board put a freeze on all parole releases. There are now Pennsylvania inmates in other states who have to be housed there because the population of inmates rose so rapidly when Philadelphia had that kind of a Willie Horton experience.

The question is not how do we react to it after the fact, but how do we anticipate it and indeed try to prevent it by introducing forecasting that is focused not on recidivism per se, but on the risk of somebody being a very serious offender who would commit an outrageous crime.

And that's a slightly different take on where we were 30 years ago when the RAND Corporation produced a report on selective incapacitation. You wouldn't know that that report had been rejected by the National Academy of Sciences in 1986 in a report on career criminals, supported by NIJ, and I think an excellent piece of scholarship in terms of assessing the accuracy of the tools that the RAND Corporation was using at that time to identify who was a high-frequency offender and who wasn't.

The Pew Trust report a year ago said we really need risk-based policies, but they point to qualitative systems for assessing risk, which are not reliable, which have been outperformed by quantitative systems ever since 1954 when the first comparison was done by [unintelligible] and his colleagues at Minnesota. And the argument that was made in 1986 I think made sense when the incarceration rate was so much lower. And that is that there's too much overprediction. There's too many people who would be identified as high-volume offenders that you'd put behind bars who wouldn't be high-volume offenders, so we shouldn't do it. But I think that what happened instead was that, in the absence of the most reliable guidance possible for criminology, which kept its hands clean, prosecutors and judges had to go ahead and make the best forecast they could make from the seat of their pants, from their own experience, from looking at a rap sheet. And as a prosecutor recently said to me, “What can I get out of one of these forecasts that you guys are doing that I can't tell from looking at the rap sheet myself?” And what I didn't say to him was, “Well, you might find out, for example, all the times you were wrong,” which you're not normally in a position to do in the course of the work of a prosecutor.

So what I think we have to recognize is that actuarial risk assessments could actually lower the incarceration rate. And that was the ethical problem that the National Academy of Sciences report had with it. They were afraid it was going to raise the incarceration rate. I think the evidence now is — we've had kind of a control group of what happens if we don't use actuarial risk assessments. Well, whether it caused it we don't know, but we know that the prison rate has skyrocketed. So, if we can't do an experiment of a randomized kind, at least we could, at the national level, we could try this locally and see whether using actuarial risk assessments would actually lead to fewer people being put in prison.

We also have a whole new generation of tools pioneered by people like Richard Berk, who are using quantitative and not qualitative efforts to do risk assessments, looking primarily at prior charges, not convictions, where offenders live, how old they are, what their gender is, not a presentence report, but something that's based on tens of thousands of previous criminal histories with these patterns that make it much more like a short-term weather report. And like it or not, short-term weather forecasting has become much more accurate in this country. It sure beats volcano forecasting. And what we can get out of the short-term weather forecast is a pretty good basis, a lot better, we know from many tests, than subjective or qualitative forecasting. You can read more about this in the Journal of the Royal Statistical Society, Series A, in the article by Berk and his colleagues last year.

But it's part of this continuation by which the statistical forecasting is doing better, especially with the cheap supercomputers, the vast numbers of cases and data mining that we have at the moment.

And let me just illustrate this notion that risk is a combination of the frequency of offending and seriousness of offending when you apply it to this kind of population.

Now Geoffrey Barnes, who's here today from the Jerry Lee Center of Criminology at Penn, several years ago did a two-year follow-up on some of Richard Berk's forecasts in relation to the forecast of homicide, what was the analysis in terms of identifying the — well, let's just take it from there. The high risk population is only two percent of the probation population in Philadelphia. It's about 45,000 people. The 38 percent medium group, the yellow here, is neither high nor low in the classification, with the low risk being about 60 percent.

Now, what happens for the overall high-risk group, which I can't show you because I'm technologically inept, but what I can tell you is that the top two percent is eight times more likely to be charged with any offense than the bottom group. That is to say, the rate for every hundred of them is eight times higher, and it's 75 times higher for murder. So this classification is accurate at identifying a group of people whose frequency of murder and attempted murder charges is 75 times higher than the lowest-risk group, which is, after all, 60 percent of the population.

Thirty-seven charges of murder or attempted murder for every 100 offenders in the high risk group. Three hundred and eighteen charges of serious crime: murder, rape, robbery, aggravated assault. And 1,800 charges for all crimes in the top two percent. These are people on probation. And many people look at these numbers and say, “Why are they on probation? Why aren't they in prison?” And the answer is, because nobody knew. The judges didn't know. The defense attorneys didn't know. The prosecutors didn't know until they went on probation and then the risk assessment was done, which is why the judges in Philadelphia have asked to have the risk assessment being done before sentence is passed.

So this brings us to a general theory, which is where I'm going to have to conclude, I'm afraid — a general theory of crime, prisons and police with three basic propositions. The first one is that the higher the proportion of inmates who are high-frequency and high-harm, the more cost-effective prisons will be in lowering the Crime Harm Index. The second is that the higher the proportion of police time spent on high-risk places, victims and offenders, the more cost-effective police will be in lowering the Crime Harm Index. And thirdly, the more low-risk offenders that police hold accountable for their crimes without using prison, starting alternatives to prison right at the police interaction in the immediate wake of an offense, with a victim still present in many cases, the more cost-effective the police may be at reducing the Crime Harm Index.

So let's just make some observations about these. First for the prisons, the idea that you have to have a higher proportion of inmates who are high-frequency and high-harm, the more cost-effective prisons will be. That could mean that prisons would have to have many fewer inmates in order to be cost-effective, or more cost-effective than they are. And why is that? Because there is diminishing returns to locking up more and more people if the basis for locking them up is their risk — the contribution they make to the Crime Harm Index.

So if you're creaming off the most dangerous people and concentrating them on prison, that is what will make prison cost-effective. Anything else will be wasting your money for no benefit. For example, lots of low-level drug transportation people who are arguably well-represented in the federal prison population but who are not posing any risk to public safety. And the question could be for the tax payers, do we want to lock them up when there's really no benefit that's gained for it, other than perhaps minimizing our successful efforts to win the war on drugs.

I think in terms of police time, we have to consider that just adding police is not predicted to cause less crime or harm. Rather, reductions in crime and the harm of crime could come even with fewer police, depending on how the police are used. So that the critical point here in evidence-based policing and risk-focused policing is not just having more police; it's doing the most evidence-based things for the greatest benefit and the greatest cost-effectiveness which, incidentally, requires more costing data than we have right now in a lot of the police research. And it could be something for NIJ to pursue more relentlessly in terms of any program evaluation, to be much clearer about what the cost benefit is in relation to not just crimes prevented, but also the total harm from crime. These are not commonly used outcomes in criminology journals or anywhere else.

But I think if we're going to get serious about using criminology to manage the criminal justice portfolio, that's the kind of direction we have to go in getting a better evidence base for making the police more cost-effective, so that if we do reduce prison population to free up more money to invest in the police that that investment can have the greatest yield possible, and we can be more confident that investing more money in the police will have a greater impact on the Crime Harm Index than taking money away from prisons would scare people and make them worry about.

Especially in light of this third point which is perhaps the most difficult for some people to accept if they want to throw the book at people. And that is the very clear evidence in at least three different domains — youth violence, domestic violence, and restorative justice — that diversions from prosecution using less legalistic and more emotionally-centered ways of creating reconciliation in communities between people who have harmed others and the people they have harmed, that these efforts which reflect traditional justice — they're actually quite widely practiced in Europe off the books, especially in rural areas — that these practices are more cost-effective for police. In doing what? Not in punishing bad guys, but in preventing future crime.

And if you read the principles on which Robert Peale founded the British police in 1829, he talks about prevention as the number one goal and not to be seen to be doing justice just because being seen to be tough or retributive is popular, but rather that we judge the police by their effectiveness in preventing crime. You could say that would apply to the entire criminal justice system.

So my title, I think it's now clear to you, is just a little bit off. Because I'm not really saying, “Less prison, less police will give us,” I'm sorry, “Less prison, more police will give us less crime.” It's not necessarily so. But it is easier to say it that way. And a risk-based policy is crucial to add to this notion with a theory of cost-effective effects on the Crime Harm Index which depends on it because more prison could in fact lead to less crime, but only if it's optimized, and therefore if we look at the risk distributions, it seems quite likely that more prison would not increase cost-effectiveness under almost any scenario.

So taking cost into account is critical. But how do we do this? And let me just close with the question of who is the — I'm skipping the evidence here which will be on the slides that are posted and that have … there we go. OK. There. Right.

Who can push the buttons? In the first instance, the governors can. The governors are the ones who have the greatest incentive; it's their states going bankrupt. No evidence of this, perhaps, is better that Arnie Schwarzenegger seeking a constitutional amendment to reduce the amount of money spent on prisons to be equal to the amount spent on the University of California. He doesn't say how that's to be accomplished except perhaps through privatization. Arguably it could be happening through sentencing guidelines, parole violation alternatives, bigger [inaudible] with state funding for more local police if the localities themselves, especially through their prosecutors, identify ways to reduce the use of prison at the same time that they reduce crime by being more selective and especially focused on the most serious offenders who are forecast through risk assessment.

So the county prosecutors and judges can make a huge difference and then so can the local police chiefs where we have good reason to believe that there's a lot more potential, a lot more room for reallocation of resources to the high-risk places and times, much more investment in problem-oriented policing, although again, the cost of problem-oriented policing in relation to the gain still needs to be developed through research as we have done with restorative justice where police-led conferences between victims and offenders in the U.K. resulted in an $8 of crime prevented for every $1 spent on investing in the police time in that kind of activity. That's the kind of cost-effectiveness information that OMB or the local budget officers can work with.

But ultimately I think I have the most hope in federal leadership showing the way beyond ideas like a constitutional amendment.

I think that the political evidence is that the way the states have been doing it so far is a really bad idea, that reducing prison by early release is likely to backfire if money appears to be the only motive. And if there's no analysis of public safety as the benefit to be gained by reducing prison, there's no claim that less prison could cause less crime.

I would argue that the front end beats the back end with arrest and prosecution, sentencing guidelines, risk analysis, every step of the criminal justice process so that we can say that the reason the prison population is going down is because we don't want to send those people to prison on purpose. Why? Because it would waste your money. Because they're not dangerous enough. They're not the kind of people that you want to lock up.

And that has been very clear in the U.K. where for three years, right up until the current election, the government released 80,000 people 14 days prior to the end of their sentence. And of course many crimes were committed in the two weeks period after their release. Under Freedom of Information, the government published that information but they never did a randomized trial to see whether people who served the full term would have any lower rate of offending, when they get out, in the first 14 days than the people who are let out 14 days early.

I told them so in 2001, 2002 and 2003. And when they finally did this in desperation, they let themselves up for what the Tories have driven all over London which is the Prime Minister of England saying, “Vote for me because I let 80,000 criminals out early.”

I don't think the governor of Michigan, who is trying to reduce the prison population, is going to benefit from the politics of that according to the New York Times story on this last month. Oregon has suspended its program after critical radio ads. The Illinois governor has called their program for reducing prison populations a big mistake. Colorado hoped to save 14 million by letting out safe people early but they couldn't find enough safe people. How were they looking? “Aw, let's sit down and look at this guy's rap sheet and decide whether we can let him out.” That's not actuarial science and certainly not what criminology can provide.

So what we need to do, I think, is to provide both carrots and sticks in which cost-saving is a stick at the front end, not at the back, I think, but much better at the front end, and then adding police with the savings on prison to be the main carrot. Cohen has shown that people prefer to spend tax money, want more taxes to be spent on police and less to be spent on prisons. So this is evidence-based politics if nothing else.

But what about the federal carrot? What about, perhaps, incentives for hiring more police?

But perhaps even more important is the Federal leadership, the research and development that would do two things: the first is to help put into place the kind of automated system that Jeff Barnes has given to the Probation and Parole Department in Philadelphia, where in 15 seconds — this is under an NIJ grant, incidentally — in 15 seconds, every new case walking into Probation and Parole will be classified as high, medium or low risk. — has been, in fact, for the better part of a year. And once that tool is available, the second thing we can do is randomize trials of introducing that information at every point of the criminal justice system, starting with prosecution and pretrial assessments and perhaps even investigating cases that are marked red with more resources to make sure that everything is done to keep the next Willie Horton off the street.

That needs to be part of the argument. We can't back away from it. One of our funders in this effort didn't like that aspect of it. But I'm sorry, it's not about the ideology; it's about getting the overall harm from crime down and the harm from spending too money much on prisons down, not to mention the harm that prison does itself.

So clearly we've got to get beyond the crisis, and bankruptcy of the states is a bad reason to switch the entire focus of the criminal justice system. But it's a good opportunity to take advantage of the attention that it's getting. And if the result turns out to be less crime and people are persuaded from controlled trials that that's what's happening, then the cost issues may fade away after we have changed the paradigm and started to operate criminal justice around the purpose of punishment being to prevent crime and not just to punish for punishment's sake.

So the Obama administration has already made history in healthcare. The question for you is whether we can also make history in criminal justice as well. My answer for one is, of course. Yes we can.

Thank you very much.

[Applause.]