The Magazine

Law and Disorder

The case for a police surge.

Feb 23, 2009, Vol. 14, No. 22 • By WILLIAM J. STUNTZ
Widget tooltip
Single Page Print Larger Text Smaller Text Alerts

We live in strange times. The federal budget deficit is higher than at any time since World War II as a percentage of GDP, yet the president and Congress are not in budget-cutting mode. Rather, they are seeking to make that deficit even larger by spending sums that, before now, seemed beyond the realm of possibility. The more obscene the amounts, the better. John Maynard Keynes's famous suggestion--pay some workers to dig ditches and others to fill them up--hasn't made it into any stimulus package, yet. But the night is still young, and politicians are still exercising their imaginations.

Sadly, in the face of record-breaking federal spending, one uncommonly good spending idea has gotten short shrift: Use federal budget dollars to pay for more cops on high-crime city streets. A modest version of that idea--$8.8 billion in federal money over six years--was enacted as part of the 1994 crime bill, and it contributed to the second-biggest crime drop of the last century.

Given that track record and the sheer magnitude of the stimulus spending, one might think that now is the time for an immodest version of the idea: say, $5 billion a year for four or five years. One would be wrong. The House stimulus package included one-shot spending of between $3 billion and $4 billion for urban policing, depending on how one counts. The Senate's more moderate proposal pared that sum down to $1 billion and change--less than "real money" by the Dirksen standard. At that level of funding, the stimulus package would pay for a mere 10,000-15,000 officers for one year only. That number is too small--America has 650,000 local police officers to patrol America's streets; 10,000 more is a drop in the bucket--and too short-term. If the recession lasts more than a year, the new officers will be out of jobs before they have time to do their productive work.

House and Senate alike are making a serious error. For $5 billion per year--five years' funding would be about 3 percent of the stimulus package--lawmakers could put another 50,000 cops on city streets. Doing so would likely both reduce crime and reduce the nation's swollen prison population--a rare combination--and would also help the economy in poor city neighborhoods by making investments in those neighborhoods safer. This is one policy that conservatives and liberals alike could support. If the Obama administration is looking for opportunities for bipartisanship, it should look hard at urban policing.


The case for more police spending begins in another part of the world, with another battle against mayhem. Not long ago, the Iraq war seemed a lost cause. Two decisions proved it wasn't. First, General David Petraeus changed tactics: Buy off or coopt all enemy fighters who are susceptible to such overtures, and bring violence to bear only on the irreconcilables. Give priority to protecting the population. Second, President Bush embraced the new approach and backed it up by sending another 30,000 soldiers to the front. Once an unattainable goal, victory in Iraq is now in sight.

Taken together, those two decisions were revolutionary. For 35 years, America's military had been looking for ways to avoid another Vietnam: a long war with heavy casualties that steadily lost public support. So, after Vietnam, military technology emphasized efficient violence: weapons systems that could take out targets more precisely and from longer distances with fewer soldiers. Military doctrine moved in a similar direction, emphasizing speed and surprise--"shock and awe"--in order to overpower enemies quickly. Both technology and doctrine aimed to get more bang for the buck, more and better-targeted violence with fewer soldiers.

The Iraq surge followed the opposite strategy: The goal was to get less bang for the buck, to use more soldiers to produce less violence. It worked. Post-Saddam Iraq is not, it turns out, ungovernable. All it needed was what any ruler needs in order to rule crime-ridden territory: armed men in uniform standing guard on violent street corners, in numbers enough to reassure local residents that they can walk the streets in peace.

That sounds like--and is--a job for a well-trained, well-funded force. The war in Iraq bears more than a passing resemblance to the battle against violent street gangs in the roughest parts of American cities. The tactics Petraeus used to win that war are eerily similar to the tactics the best police chiefs use to rein in gang violence. But better tactics alone cannot do the job. In Boston as in Baghdad, those tactics work only if the police forces that use them have enough personnel: lots of police boots on the most violent ground.

Today, that condition is not satisfied. Most American cities are underpoliced, many of them seriously so. Instead of following the Bush/Petraeus strategy, the United States has sought to control crime by using small police forces to punish as many criminals as possible. As all those who have even a passing familiarity with contemporary crime statistics know, that approach--call it "efficient punishment"--does not work. Like the Army in pre-surge Iraq, the nation's criminal justice system is in a state of crisis. America needs another surge, this one on home territory.


Readers may be excused for wondering how crime could be a large problem in 21st-century America. Crime fell substantially in the 1990s. In this decade, crime rates have been mostly level; the gains from the 1990s remain intact. Where is the crisis?

The short answer is: in urban neighborhoods that readers of this magazine rarely visit. The most important crime trend of the last half of the 20th century wasn't a rise or fall in overall crime levels. Rather, the key trend was a change in the distribution of serious crime, as Table 1 illustrates:

In 1950, violent crime was mostly a southern phenomenon; of the six non-southern cities listed above, three had murder rates lower than the nation's, and the other three had rates only modestly higher than the figure for the nation as a whole. Today, violent crime is an urban problem--everywhere, not just south of the Mason-Dixon line. Atlanta and Philadelphia are comparably homicidal, as are Houston and Chicago. All but one of the non-southern cities listed above--New York is the exception--saw their murder rates double since 1950; all but two saw murders triple, while the nation's rate rose slightly. And remember: These figures are calculated after the 1990s crime drop.

Those high contemporary murder rates understate the rise in urban violence. Thanks to advances in emergency medicine, a sizable fraction of 1950 murder victims would survive today. For accurate comparison, one must cut the 1950 figures by at least one-fourth. Do that, and a clear picture emerges: Outside the South, American cities are at least several times more violent than they were in the mid-20th century. The same is not true of the small cities and towns where many Americans live: In New York state, 3.2 million people inhabit jurisdictions that saw zero murders and manslaughters in 2007. In most of the United States, violent crime is something that happens elsewhere. In the poorer parts of American cities, crime stories are much closer to home. The 1990s narrowed that gap modestly, but the gap remains large.

Even within crime-ridden cities, violent offenses are geographically concentrated. The safest city neighborhoods are about as safe as those small towns in New York where no murders happen. In the most dangerous city neighborhoods, violent crime reaches levels common in Mexico, Russia, and South Africa--three of the highest-crime countries on the planet. Most of our dangerous communities have two things in common: They are poor, and a large share of their population is African American, as one more pair of statistics suggests. In the United States in 2005, the homicide rate among whites stood at 3.5 per 100,000. Among blacks, the figure was 26.5. Urban homicide rates are strongly correlated with the size of cities' black populations.

This is the core of 21st-century America's race problem. The poorest black neighborhoods are frighteningly crime-ridden. The social cost of that crime, and of the criminal punishment that seeks to hold it in check, is colossal, measured in families never formed and investments unmade, lives ended murderously and other lives slowly crushed by long prison terms. Plainly, the public policy status quo is not working. To stay the course--to resist change in a setting that so needs changing and in which so many lives are at stake--is morally indefensible. That proposition holds true for those of us on the political right as well as for those on the left. Conservatives are not anarchists; we believe governments should be small, not absent. Even the smallest governments seek to maintain a decent level of order and safety in public places. Today, in much of urban black America, that goal is not being met.


Over the past 35 years, the United States has run the equivalent of a controlled experiment in alternative means of reducing violent crime. A massive, disproportionately black prison population has had surprisingly little effect on urban violence. As we shall see, modest increases in the size of urban police forces appear to have achieved much better results. More-than-modest increases might do better still.

Some history is in order. Between 1950 and 1975, in the midst of the largest crime wave in American history, prison populations fell sharply throughout the Northeast, Midwest, and West Coast. By the early 1970s, imprisonment rates in those parts of the country were comparable to the lowest imprisonment rates in the 21st-century Western world. In 2001, Denmark, Norway, and Sweden imprisoned 58, 58, and 69 inmates per 100,000 population, respectively. (The European average was 87 per 100,000.) In 1972, the imprisonment rates in Massachusetts, Illinois, and New York stood at 32, 50, and 64, respectively. Between the mid-1970s and 2000, those same prison populations skyrocketed. New York's imprisonment rate sextupled; Illinois's multiplied more than seven times, while Massachusetts's rose eight-fold. Measured per unit crime, the increase in punishment was greater still. Nationwide, the number of prisoner-years per homicide rose more than 800 percent between the mid-1970s and the beginning of the 21st century. In the span of three decades, Americans first embraced punishment levels lower than Sweden's, then built a justice system more punitive than Russia's.

In criminal justice as in business and finance, price stability is one of the keys to successful governance. Criminals are what economists call risk-preferring; they enjoy taking chances. When the consequences of a given crime vary widely across time and place, criminal punishment takes on the character of a lottery. Criminals like lotteries. Fluctuating "prices" thus rob criminal punishment of its deterrent power.

The same is true when the prices offenders pay for their offenses are opaque, such that members of the public and would-be offenders alike cannot tell which consequences flow from which crimes. Over the last generation, America's justice system has increasingly embraced the use of certain criminal prohibitions as means of punishing other, harder-to-prove crimes. Federal prosecutors could not convict Martha Stewart of insider trading, but no matter: Obstruction of justice could substitute for the more serious crime that prompted Stewart's prosecution. Convicting members of urban gangs of violent felonies is often difficult: The government needs witnesses to convict, and gang members are skilled at silencing those who might testify against them. Often, prosecutors respond by charging gang members--all of them, not just the ones with the most homicides or assaults to their names--with drug crimes or with possessing unregistered weapons. In such a system, even defendants cannot easily tell what price they pay for the violent crimes that prompt their prosecutions.

There is one more reason why skyrocketing prison populations in the 20th century's last quarter did not produce plummeting crime rates: The number of inmates rose too far and too fast, especially the number of black inmates. Table 2 tells the story

The black imprisonment rate in 2000 was roughly one-third higher than the rate at which citizens of the Soviet Union were confined to the gulag's camps in Stalin's last years (when the camps' population peaked). If America's jail population is included, the black incarceration rate is nearly double the Soviet rate of confinement circa 1950.

Such massive incarceration rates cannot effectively deter crime. At percentages like these, a term in the state penitentiary becomes either an ordinary life experience or a badge of honor, not a scarlet letter that stigmatizes those who must wear it. Over the course of the past generation, incarceration has become part of the cultural furniture of black America. When criminal punishment is at once routine, opaque, and variable--three adjectives that fit American punishment practices to a T--it cannot serve the purposes for which it is designed.

Given those characteristics, it should be unsurprising that the tripling of America's imprisonment rate in the 1970s and 1980s produced no drop in urban crime. On the contrary, in high-crime cities, the level of criminal violence rose during those two decades. True, crime fell in the 1990s--but as it did so, the rate of increase in the prison population slowed. The notion that increased criminal punishment accounts for the bulk of the crime drop is implausible: If so, why didn't crime fall more, and sooner? The murder rate in the United States is about the same as in 1966; the imprisonment rate is almost five times the rate in 1966. Apparently, it takes five times as much punishment to achieve the deterrent effect prison terms had more than 40 years ago.

As it was in Iraqi streets before the surge, so it is in violent American neighborhoods: Killing more insurgents and locking up more criminals seem only to produce more insurgents and criminals. A different approach is needed.


That different approach has already succeeded. Between 1989 and 1999, the number of urban police officers per unit of population rose 17 percent. Arrests fell by a little more than 20 percent; arrests of black suspects fell by one-third. Crime fell too, and it fell most in the jurisdictions that hired the most cops. In 41 pairs of neighboring states, one jurisdiction increased its policing rate more and its punishment rate less than its neighbor during the 1990s. In the higher-policing, lower-punishment states, violent crime fell by an average of 24 percent. In the lower-policing, higher-punishment group, the average crime drop was only 9 percent. Higher-policing, lower-punishment states outperformed their more punitive, less well-policed neighbors in all parts of the country. The city that saw the nation's largest crime drop--New York--increased the size of its police force the most. The state that includes that city increased its prison population the least.

The results--more cops, less crime, fewer prisoners--sound too good to be true. What explains them? The answer is the same in American cities as in Iraqi ones. General Petraeus saw the process: Increase soldiers' street presence, make civilians feel safe, and local traffic increases; before long, the streets are inhospitable to terrorists and militia members. Over time, the real work of pacification is done by civilians reclaiming their own streets.

Best of all, putting more boots on violent ground tends to reduce the incidence of inappropriate violence and abuse by those whose job is to maintain order. In Abu Ghraib, there were 15 prisoners per guard on average; at some points the ratio reached 75 to 1. By comparison, prisons in the United States ordinarily have a ratio of 6 prisoners per guard. Violence and coercion are substitutes for numbers; when the number of guards is high enough, the need for physical force and the temptation for cruelty both drop sharply.

The same principle applies in police work. The Washington Post Magazine recently ran a story about the mayor of Berwyn Heights, a small town in Prince George's County, Maryland. The county police executed a drug raid on the mayor's house; the raid turned up no evidence but left Mayor Calvo and his wife traumatized; among other things, the police shot and killed the couple's two dogs. Even the best police forces sometimes act on bad tips. But those mistakes are fewer when officers are numerous enough to know the communities in which they work. And the errors that remain are less costly when the police force is sufficiently well staffed that an ordinary house search does not resemble a military action. Nationwide, the number of local police officers per 100,000 population stands at 245; in New York City at its peak size in 1999, the NYPD employed 561 officers per 100,000. In Prince George's County, the number is 195. Given a larger police force, Calvo's dogs might still live. So might his trust in the decency of his community's law enforcement personnel.


One more question must be answered: If American cities need more policing, why shouldn't the cities pay for it themselves? The same question might be asked about urban public schools, and the same answer given: Big-city public schools disproportionately serve the poor. Local tax bills are mostly paid by wealthier city dwellers. If cities tax them too heavily, those taxpayers may move to friendlier jurisdictions. Which is why federal and especially state governments pay most of the tab for urban schools. By contrast, cities pay more than 90 percent of their own police budgets--though urban police spending, like spending on urban public schools, disproportionately benefits the poor.

Actually, the authors of state and federal budgets have gotten these two subjects backward. There is little evidence that marginal budget dollars make for better schools, but a great deal of evidence that marginal dollars mean better policing. Throwing money at problems is rarely good public policy. Urban policing may be the exception to that rule.

That is the liberal case for more federal funding for local police. There is also a conservative case. Conservatives are skeptics of governments that seek to fine-tune individual incentives. For the past generation, state legislators and members of Congress have acted as though, if only the right criminal prohibitions and sentencing rules are enacted, behavior will change in just the ways lawmakers wish. Like other regulated actors, criminals stubbornly resist such direction. Conservatives believe in the power of culture. Policing surges free residents to take control of their neighborhoods and build the street cultures the locals want. Those local cultures do more to control crime than any government-tailored incentives. Last but not least, conservatives believe in the traditions that capture the wisdom of the past. Before the last half-century and outside the South--at the times and places where criminal justice worked best--the normal ratio of police officers to prisoners was two to one. Today, in most of the country, the ratio is one to two.

Again, we live in strange times. Over the next couple of years, money will be spent in fantastic sums, and much of it will leave nothing good in its wake. All the more reason to invest in a service that makes life better for those whose lives most need help. That is the liberal thing to do--and also the conservative thing to do. More important, it is the right thing to do.

William J. Stuntz is the Henry J. Friendly professor at Harvard Law School.