What’s driving New York’s crime rate down?
By Andrew Karmen
Murders are more than just frightening statistics; each and every one represents someone’s tragic, untimely death. As such, the recent dramatic drop in murders in New York City is all the more welcome a development. Whereas New Yorkers were killing each other at a rate of more than 2,000 a year at the start of the 1990s, this year’s death toll could fall below 1,000. All New Yorkers, especially those residing in the poorest neighborhoods, are now living under much safer conditions, as the number of assaults, robberies, burglaries and vehicle thefts falls along with homicides.
The key question is why? Who or what deserves the credit? If criminologists can solve the mystery behind the city’s across-the-board drop in crime, the lessons learned from the New York experience can be put into practice to benefit many other jurisdictions. Experts quoted in the press have suggested many leads worthy of follow-up investigations: the proportion of New Yorkers who are adolescents and young men in their crime-prone years has slipped; the level of violence surrounding the sale and abuse of crack cocaine has subsided; the number of hard-core offenders currently incarcerated in city jails and state prisons has grown, and an improving economy is drawing formerly unemployed persons back into the world of legitimate work. Last but certainly not least is the argument that the NYPD’s new leadership has greatly improved policing. In fact, Mayor Rudolph Giuliani and his police commissioners, William Bratton and Howard Safir, insist that a re-engineered, refocused department deserves all the credit for the sudden ebbing of the crime wave that peaked in 1990.
Criminologists, of course, cannot simply accept self-serving claims and then echo them. As social scientists, criminologists must try to evaluate claims for credit objectively, adopting a skeptical stance and saying: “I doubt it. Show me the evidence. Prove it!”
This preliminary analysis focused entirely on variables describing police performance as published in NYPD annual reports. The aim is to identify promising explanatory factors that can be scrutinized more carefully at a later date using more sophisticated methods of statistical analysis. Potential explanatory variables should roughly correlate (rise or fall in tandem) with the upward and downward trends in the number of murders. The empirical question is: As the murder rate went down, what measurable, demonstrable aspects of police operations also went down (or went up, if the relationship is inverse)?
The measurable aspects of police performance that were examined were clearance rates, patrol and enforcement strength, response time, and total misdemeanor and felony arrests. Unfortunately, the NYPD’s new strategies were not introduced one at a time and tested in some precincts but not others, as a social experiment. They were implemented citywide and all at once, as a package, and thus cannot be evaluated in isolation from each other. In other words, it is impossible to directly determine which innovations, if any, are working out well, either singly or in combination with other new strategies.
Department figures reveal that during the 1990s, murders changed in several ways. The number of murders fell sharply, far fewer were committed outdoors, and guns were used less often.
Since the New York City fiscal crisis of the mid-1970s, the murder rate has twice undergone downward trends: the four-year stretch from 1982 through 1985, when murders decreased by 24 percent (from 1,832 to 1,392); and the six-year period from 1991 through 1996, when murders dropped 54 percent (from 2,262 to a conservatively estimated year-end total of 1,030). Murders went up 21 percent from 1978 to 1981 and jumped 63 percent from 1986 (the start of the crack epidemic) to 1990. These four periods when murder rates rose, fell, then rose and fell once again, provide a basis for the search for correlations. [Figure 1.]
In what ways might an improvement in policing since January 1994, when Mayor Giuliani and Commissioner Bratton took over, contribute to this dramatic reduction in the level of violence? Perhaps new strategies are enabling detectives to solve a greater proportion of homicide cases. An improved clearance rate might prevent some future slayings in two ways: Bringing more murderers to justice might deter other would-be killers, and removing some dangerous predators from circulation will prevent them from striking again.
Clearance rates for murders and other crimes have been dropping across the country since the 1950s, when over 90 percent of all killings were solved. These days, detectives are fielding fewer “grounders” (easily solved cases, as when a husband kills his wife and then turns himself in), and are grappling with more difficult stranger-to-stranger crimes, such as murders arising from robberies.
The percentage of murder cases solved by arrests (Figure 2) fell as low as 57 percent in 1979 and 1981 and rose as high as 74 percent in 1983. Clearance rates, which had slipped to 59 percent in 1992, rebounded to 62 percent in 1993, to 64 percent in 1994, and then to 73 percent in 1995, when 861 of the 1,180 murders were solved. This sudden improvement placed the NYPD’s clearance rate way ahead of the national average (about 68 percent for the other 16,000 police departments across the country, according to the FBI’s Uniform Crime Reports).
One possible interpretation is that the beneficial effects of the new Giuliani/Bratton strategies seem to have really “kicked in” and produced very favorable results during 1995. But an opposite spin can be given to the same statistics. The impressive 73-percent solution rate was achieved after the collective workload dropped by 19 percent, nearly 400 cases. Maybe this improvement was a result, not a cause, of the dramatic drop in the murder rate. Perhaps detectives were able to solve many more cases in 1995 because they had much more time than in prior years to devote to thoroughly investigating each one.
For the year 1993, 1994 and 1995, the clearance rate rose as the murder rate fell. When the ups and downs of the murder rate are juxtaposed against the changes in clearance rates [seen in the lower line in Figure 2], there appears to be some indication of a weak inverse relationship between the two variables: When clearance rates go up, murder rates tend to go down; when fewer murders are solved, more murders are committed. The relationship is weak because this pattern does not hold for certain years (for example, clearance rates fell in 1991 and 1992, yet murders decreased as well).
It seems reasonable to hold the police more accountable for crimes that are committed right out in the open, in public places like streets, corners, stoops or parks, often in broad daylight, in front of bystanders and in locations that are visible to officers on patrol. Understandably, the police can do less to stop crimes committed out of their sight, in back alleys, or indoors in hallways and elevators or behind closed doors.
In its annual statistical report “Complaints and Arrests,” the Police Department notes the percentage of the crime scenes that were deemed to be “visible” to officers on patrol. In 1994, as compared to the previous year, 359 fewer outdoor “visible” murders and 26 fewer indoor “not visible” murders were committed [Figure 3]. Therefore, almost the entire decline during that first year of the Giuliani/Bratton administration can be accounted for by this reduction in “visible” murders. The surge in fatal confrontations in outside locations has been the primary cause of the high murder rates that have burdened the city since the late 1980s. Prior to 1988, “visible” murders did not far exceed “not visible” murders, except in 1983.
The even greater drop in murders committed in 1995 as compared to 1994 was more complicated. “Visible” murders dropped by 22 percent (192 fewer); “not visible” murders fell by 28 percent (also 192 fewer). Clearly, murders were not merely being driven indoors. Nor was criminal activity simply displaced geographically from some neighborhoods to others, according to an analysis of murder rates by precinct.
Has the Police Department’s campaign to “take back the streets” prevented many “visible” murders during 1994 and 1995? The NYPD has had a greater presence on the streets that may have intimidated criminally inclined persons who hang out in parks and on corners and stoops from acting on their desires.
Police presence can be measured in terms of the average daily patrol strength of uniformed officers on their tours of duty, riding in squad cars and walking their beats. Another measure is to add in those plainclothes officers and detectives who are out making arrests, as indicated by the average daily enforcement strength. Trends in both these manpower head-counts for recent fiscal years are depicted in the lower two lines in Figure 4. Superimposed on the same axes, and represented by the line connecting the triangles, is the rise and fall of murders carried out in “visible” locations.
Taken together, the three lines give the impression that a weak inverse relationship exists. When both patrol strength and enforcement strength went up, murder went down during the early 1980s. In the late 1980s, patrol strength and enforcement strength declined slightly, while outdoor murder rates soared. After 1990 and 1991, the two worst years in recent history, patrol strength and enforcement strength picked up again, while outdoor killings plunged. There are a few years that don’t fit this pattern. Furthermore, in fiscal 1996, both measures of police strength declined, yet murders also declined, in a break with the generally inverse relationship detected in other years.
Besides a beefed-up police presence, a quicker response time might be a factor in reducing slayings carried out in the open. It is reasonable to assume that lives could be saved if the police intervene more quickly when summoned by a 911 emergency call about a crime in progress, such as a street robbery, a fight in the park, shots fired, or a man with a gun. Figures for “annual average response time to a 911 call about a crime in progress” are available, for the 1990s only, from the Mayor’s Management Report. The data show the average response time, starting from a baseline of 9.9 minutes in the 1992 fiscal year, dropping to 8.0 minutes in the 1993 fiscal year, 7.9 minutes in 1994, 7.7 minutes in 1995, and then reverting to an average of 9.1 minutes in fiscal 1996.
The generally downward trend in response times tracks with the downward trend in murders, as expected under the improved-policing hypothesis. However, the reductions in response times are so minor in 1994 and 1995 that they are probably statistically insignificant. Furthermore, the big jump in 1996 is puzzling, since murders continued to fall sharply even though response time rose sharply. Response time, thus, seems unimportant.
Homicides caused by blunt instruments, stabbings, strangulations, poisonings, burnings and other means generally have been going out of fashion since the early 1970s [Figure 5]. Unfortunately, at the same time, guns have grown in popularity as the weapon of choice. The number of people who died from bullet wounds rose in the late 1970s, fell in the early 1980s, soared shortly after the emergence of the crack epidemic in 1986, peaked in 1991, and then began to drop, at first slowly in 1992 and 1993, and then sharply in 1994 and 1995.
The centerpiece of the Giuliani/Bratton/Safir approach to “take back the streets” has been a policy of aggressive order maintenance, which rests upon “zero tolerance” for any lawbreaking. Even the smallest quality-of-life infractions are grounds for arrest. According to the “Broken Windows” theory, small signs of disorder send out signals that attract offenders, who then feel emboldened to commit even more serious crimes. Stepped-up arrests for minor violations are deemed the antidote to a slide into disorder.
Police Department spokesmen have asserted that the wave of misdemeanor arrests has contributed to the drop in gun-related crimes because the “bad guys” are now afraid to go around carrying concealed weapons, since they might be stopped, frisked and questioned for committing some minor quality-of-life offense.
Juxtaposing the number of people killed by gunfire against the number of people arrested for minor offenses [Figure 6], it is obvious that in 1994 and 1995, as misdemeanor arrests soared to record levels, homicides carried out with firearms (mostly handguns) plummeted. In general, the patterns of the two lines seem to indicate a weak inverse relationship. When misdemeanor arrests rise, gun deaths tend to drop; when misdemeanor arrests fall, murders committed with firearms rise. The relationship is weak because some years do not fit the pattern.
The search goes on
In sum, then, murders are plummeting, slayings in visible locations are declining, and deaths from gunfire are dropping. Clearance rates are recovering, police strength is up, and misdemeanor arrests are soaring. At first glance, these performance measures look like they may correlate with murder rates, and thus it would appear that the NYPD deserves some of the credit for bringing down the city’s murder rate. But the observed relationships are weak, not striking, and some of the data can be interpreted in a different light.
Advocates of the “NYPD deserves all the credit” point of view now have the obligation to put forth their case in greater detail. They must provide measurable indicators of improved policing, and need to demonstrate the specific linkages between the new strategies and the drop in crime rates. More data are needed from NYPD homicide files (about what percent of recent murders are outgrowths of conflicts between drug dealers, street gangs, family members, etc.). Furthermore, since the drop in murders predates the new police strategies and is also taking place in many other large cities, the search must go on for the additional reasons for this most welcomed outbreak of better behavior on the part of New Yorkers, especially the city’s poor young men.