The summer of 1991 brought the riots of Crown Heights, N.Y., and a "low point in morale" for the New York City Police Department, where a former cop characterized policing efforts as "mass confusion."
"I was disgusted with the city," said Ed Reuss, a former captain. "I thought New York City was going down the tubes."
Reuss retired the following year, but glimpsed the future of New York City policing before he did. "What happened, thank God, was the computerization of data."
It started with an online booking system in 1992. When Bill Bratton was appointed New York City police commissioner in 1994, it swelled into CompStat, a system of computerizing and mapping crime data, then relentlessly pursuing solutions to the findings.
The result was a 39 percent decline in serious crimes, including a dramatic reduction in homicides. From 1993 to 1998, the number of murders dropped from 1,946 to 629. Although the plunge in violent crime was part of a national trend during that period, the decline in New York's crime rate was three times the national average.
The city later applied CompStat to other measurements of city service, such as emergency response time, police station conditions, police car availability for patrol and speed of repairs.
That effective, if straightforward, concept branched in numerous directions during the last decade. Many cities have implemented some variation of the CompStat idea, and some have broadened it far beyond its original intent.
Just south of New York, for instance, Baltimore was drowning in a sea of crime, drugs, failing schools and embarrassing budget deficits. Desperate to reclaim its city, Baltimore developed CitiStat, a copycat of CompStat. After three years of applying the concept to law enforcement, Baltimore cut its murder rate in half.
The city also transformed its version of CompStat from a law enforcement data collection technique into a management tool that measures effectiveness of city services, such as garbage collection, tree trimming and protecting kids from lead paint. Baltimore officials credit the concept with increasing government efficiency and saving taxpayers millions.
Finally, although CompStat itself relies on fairly basic technology, the concept has triggered the implementation of more complex systems. For instance, it was a key driver in Baltimore's creation of a centralized 311 system, which gives residents one number to call for virtually any city service.
Bratton, now chief of the undermanned Los Angeles Police Department, intends to take CompStat even further by adding new analytic functions, making the system even more powerful to combat violent street gangs.
And New York City is just as eager to recite statistics on crime reduction, and tout CompStat as the means to that end. Mayor Michael Bloomberg's administration has built on CompStat with new initiatives that make further use of the data.
The original CompStat model created a system that uses current, relevant data to direct police activities rather than relying on three- to six-month old information, which was previously the norm.
The concept is nothing new to the private sector, but it hadn't been applied to law enforcement until the mid-1990s, when CompStat was implemented in New York, Bratton said. "The dirty little secret in America is how poor a job we do at dealing with crime. It wasn't until the 1990s, when we accepted responsibility for it, that we began to be held accountable for it."
Bratton and his staff knew if law enforcement made decisions on old data, they'd hop from crime scene to crime scene, doing little more than cleaning up after criminals.
"In the '70s and '80s, there was this excuse that crime is caused by the economy; it's caused by this or that," Bratton said. "Police can't influence any of that. That's baloney."
Bratton said not only does law enforcement prevent crime, it can influence a city's economy by making it safe for businesses to invest and tourists to visit.
"We had it ass-backwards all those years," he said. "CompStat is the engine that effectively allows that type of revolutionary thinking to occur because it puts pressure on the police to deliver what they should deliver: crime reduction."
As New York police commissioner, Bratton demanded weekly "stick counts" of major crimes in the city. Deputy Commissioner Jack Maple had commanders deliver 5.25-inch floppy disks with the crime stats for uploading into a Microsoft FoxPro database.
The precinct commanders then explained the crime stats and what they were doing about them. Within a year, pin mapping was incorporated, and the CompStat team spent 20 minutes each day sticking pins into a physical map to point to where crimes were occurring. It was a new concept for New York, but not for Bratton.
"In the late '70s, as a lieutenant in Boston, I had huge maps to put cops on patrol where the crime was," Bratton said. "Maple [a former lieutenant with the New York transit police] was doing the same thing in transit, so that was the coming together."
The pair hooked up with Louie Anemone in New York. Anemone ran CompStat for six years as chief of patrol.
When the CompStat team added MapInfo GIS software, which was unused from a previous city initiative, New York's twice-weekly CompStat meetings started featuring computer generated pin maps that plotted all crimes in each precinct.
No longer could precinct commanders "hide" from the numbers. Their crime stats, in map form, were displayed for fellow commanders, detectives, information specialists, officers and agency heads to see. And precinct commanders better have a plan to solve the problems.
That was part of Bratton's "middle down" theory of management, which increased responsibility and accountability on precinct and borough commanders. It gave them power to act as a sort of "mini-police chief."
Refining the Concept
Those principles were passed along to the current administration, which has continued to enhance CompStat.
Garry McCarthy was a New York precinct commanding officer from 1995 to 2000. The line of questioning during CompStat meetings was basic, he said. "What are your issues? Where are your issues? And what are you doing about it?"
Data collection and how it is used has become considerably more detailed, but the technology is simple. Cops record a multitude of statistics about every crime in an IBM AS/400 server. Microsoft Excel converts the numbers to charts and graphs, and MapInfo Professional maps the data.
So when a shooting occurs, for example, information is immediately entered into the computer system, in this case a database for shootings. Where police used to operate on hunches and scribbled crime reports, they now rely on updated, constantly analyzed data.
Better data collection and analysis can yield clues to solving cases that may otherwise collect dust. A shooting that might, on its face, appear narcotics-related because it occurred in a drug-infested area, may be related to a former girlfriend or previous dispute.
"We can push buttons so we know, right now, when our shootings were and where they were. Was it inside? Outside? Were the victims male, female, black, white and so on," said McCarthy, who is now deputy commissioner of operations for the New York City Police Department. "We measure it daily. That's important because the key components to crime reduction and CompStat are timely and accurate intelligence, rapid deployment, effective tactics, and relentless follow up and assessment."
Ask McCarthy about any crime in New York City and he'll rattle off statistics with near up-to-the-minute currency. "I can tell that in looking at last week's shooting incidents in the city, we were down 9 percent. But we had 38 shootings ..."
It's not just shootings, robberies and other major crimes, although they are the eventual targets. A major component of CompStat today is using statistics to determine troubled areas and focus resources on those areas. The cop standing on the corner is no longer in vogue -- proactive policing is. So is an emphasis on quality-of-life offenses such as panhandling, prostitution, parking violations, public drunkenness and other "nuisance" crimes.
Directing resources to those troubled areas is basic to CompStat, but the emphasis on enforcing quality-of-life crimes began as an initiative called Operation Clean Sweep, hatched by Bloomberg's administration.
Behind the initiative is a theory that arresting individuals for minor offenses keeps them from committing major ones. An officer also might find the person is wanted for more serious crimes. It puts crooks on notice that committing a crime, any crime, could cost them.
"If somebody goes to buy a $10 bag of marijuana, and they receive a $110 parking summons because they were double parked or parked on a hydrant, they might not come back to that place to buy drugs the next time," McCarthy said. "Proactive enforcement is incredibly important."
Under Operation Clean Sweep, New York keeps a running list of the 100 worst areas for quality-of-life offenses, and targets offenses in those areas and measures the success. When one area is cleaned up, it drops off the list and another area takes its place.
The administration -- including Police Commissioner Ray Kelly and John Feinblatt, the mayor's criminal justice coordinator -- recently developed two more initiatives.
Operation Impact redeployed 800 officers to especially troubled areas of the city. In 2003, crime in those areas was reported down 47 percent from the same period in 2002.
Operation Spotlight identifies repeat criminal offenders who are continually recycled through the criminal justice system to make sure they do time. "This is coordinated through the criminal justice coordinator because it's trying to get the district attorney and the judges to target these people to make sure they get some sentencing," McCarthy said.
And according to numbers, McCarthy said, it's working.
"I've been in this position almost four years now," McCarthy said. "I remember when I came here, people said that's a bad position to be in because crime is going to [go up]." Since then, the crime rate has had annual reductions of 6 percent, 7 percent, 12 percent and 8 percent. For the first six months of this year, the crime rate is down 7.5 percent, which he said is the result of an ever-evolving process.
"The key to crime reduction is like success in any corporation. It's an ongoing re-engineering process, and that's what CompStat provides for us," McCarthy said. "We're not sitting back and saying, 'This is the way we do things, period.' We're examining every day, every minute how to do things better."
That concept was foreign for the city of Baltimore until Mayor Martin O'Malley -- impressed with New York City's CompStat success -- initiated CitiStat in 2000, making it his signature issue.
O'Malley had his work cut out for him. Nearly 60,000 of Baltimore's 650,000 residents were hooked on drugs, and the city's crime rate hadn't fallen with the national trend. The city lost 12 percent of its population during the 1990s -- one of the largest losses in the country -- and tens of thousands of homes and more than 40,000 buildings were vacant.
Baltimore reported more than 300 homicides each year during the 1990s. With CitiStat, officials began tracking homicide rates and locations, concentrating more resources in high-rate areas. In 2000, the city recorded 260 murders. Then, in 2001, Baltimore posted the largest murder-rate reduction in the country and lowered overall crime by 30 percent. The city also enjoyed an increase in home sales, ending years of decline.
But CitiStat, as its name suggests, doesn't apply just to crime; it measures the effectiveness of a range of city services. And the technique transformed a bureaucracy that O'Malley described as having a "culture of failure" to the antithesis of what it once was.
CitiStat transcended traditional government-agency boundaries. An example of that is LeadStat. Prior to LeadStat, Baltimore had failed to hold landlords accountable for buildings where children were being poisoned by lead-based paint.
A LeadStat team was formed with officials from the city's housing, health and environmental departments, and Maryland state agency representatives. The team soon produced maps with red dots indicating where lead poisoning had occurred. In its first year, the team filed 121 cases against landlords. It filed 148 cases the next year and prevented 482 properties from being rented until landlords fixed lead problems. The City Council followed the team's work by passing legislation to require lead testing for one- and two-year-olds. Fifty percent more kids in that age group were tested after legislation passed, and most red dots were eliminated.
Return on Investment
Baltimore estimates it saved more than $13 million on an initial investment of around $20,000 during CitiStat's first year of operation, mostly by cutting overtime by 40 percent and absenteeism by 30 percent. During its first two years, CitiStat is credited with saving nearly $45 million and improving city services, such as garbage collection.
How does it happen? The simple answer is by holding people accountable.
The CitiStat team -- the mayor, his first deputy and other administrative types -- define customized performance measures for each agency. The main source of information for these measures is the human resources system data, such as payroll, overtime, absenteeism, accidents, light-duty time and comp time. Data that measures the quality of service to the public is garnered through the city's 311 call center as well. This information is combined and analyzed to produce comprehensive performance measures.
Then every two weeks, or once a month in a few cases, agency heads or their deputies face the CitiStat team for a progress report. The results of the meeting -- directives from the team and commitments made by the agency -- are recorded by CitiStat staff and provided to the agency by e-mail within 24 hours.
Answers to any questions or requests by the CitiStat team must be forwarded within a week after each meeting, so a CitiStat analyst can review and make sure issues are being addressed.
The analyst then drafts a briefing of the data and recommended questions to ask the agency head at the next CitiStat meeting. The briefing might include pictures, graphs, maps or charts, which are projected on a screen and serve as a backdrop during the meetings.
"We bring a department in for evaluation, but we kind of subject it to a number of different looks by putting other agencies in the room as well," said CitiStat Director Matt Gallagher. "The idea is to have decision-makers there who can help resolve issues quickly, as opposed to having to write a half dozen memos and have a back and forth for a couple of months."
The agency heads deliver at the podium every two weeks, and demand the same sort of accountability from their employees.
"Each level feeds off the next," Gallagher said, adding that the team, which includes the mayor, is not above showing up unannounced at work sites to observe.
"It's a huge part of what we do," he said. "You can show up at a fleet garage or a transfer station, and they know what CitiStat is. They know they've got information they've got to report, and their boss gets asked questions about it by his boss, and it goes all the way up to the mayor."
Up to Speed
After three years, the process works like clockwork, according to Gallagher. But it wasn't always that way.
"They've gotten regimented and adjusted to it. Before, it would take a long time to compile the information. Now people understand how to manipulate the data systems better to yield the information they need."
The system is based on Microsoft Office. Excel data templates are developed for each agency, which complete the templates and e-mail them to the CitiStat team. The team worked with the existing technology of each agency to develop reporting requirements.
At first, the process was tedious and there were complaints, Gallagher said. "It took an extraordinary amount of work to put on these types of meetings and for managers to be prepared. A popular refrain among managers early on was, 'If I wasn't spending my time collecting data, I'd be out doing my job.'"
Union leaders weren't excited about the concept either, and they wanted to be involved in all CitiStat meetings, according to Gallagher. "The view we took was that it's a management meeting, and the labor unions have other forums in which to voice their opinions to the mayor."
Gallagher said union leaders were invited to some meetings, and generally left satisfied that CitiStat was a management tool and not a union issue.
There were early misconceptions, however, that CitiStat was about getting people fired or disciplined. Some workers lost jobs -- mostly malcontents -- Gallagher said, but most bought into the concept, partly because they had to and partly because they were motivated by the "carrot and the stick" approach.
The city created incentives such as free tickets to Baltimore Ravens football games and Baltimore Orioles baseball games for good CitiStat performance.
"It can't be just about punishment," he said. "There has to be the carrot and the stick, and you have to go out and get great people to do this."
That has become a lot easier after three years of CitiStat. "We probably interview eight to 10 people for every one we hire now," Gallagher said. "The credentials of the people who work [in the mayor's office] are unmatched in Baltimore city government. We're turning away people from the Brookings Institute."
Now CitiStat attracts thousands of visitors from around the globe to Baltimore each year -- many of whom hope to replicate its success. The CitiStat model is even being piloted in two cities in Serbia.
Can it be done anywhere?
Yes, said Gallagher, but with a caveat: Executive commitment is necessary. In Baltimore, that commitment originated with the mayor. "It was really his signature management issue for the whole government, and we were lucky to find some likeminded people to share in the mayor's vision."
Hoping to share the vision with their constituents are the mayors from Chattanooga, Tenn., and Providence, R.I., who have made commitments to developing CitiStat models.
"It's a very, very exciting initiative and one of the most important things from my perspective that we have begun in my administration," said Providence Mayor David Cicilline, who is spearheading the creation of ProvStat in his city.
Two principle functions of ProvStat are to hold people accountable and provide a decision-making tool for city managers.
"Directors of a department are charged with making decisions every day, and very often they make those decisions on what they think is happening in their department and in government," Cicilline said. "Often they don't have the data they need to make informed decisions. ProvStat gives them that data, so decisions are well informed and driven by evidence."
Cicilline said department managers haven't necessarily had the tools and information they need to boost efficiency and cut costs. He cited the Forestry Division of the city Parks Department, which is responsible for cutting down or pruning trees.
The department was not scheduling its tree maintenance in an organized manner, sometimes directing services to whoever screamed the loudest. With the aid of GIS mapping, maintenance now is scheduled geographically by street to cut down on driving time.
The ProvStat team -- directed by Pamela Cardillo, a marketing consultant brought in from the private sector to guide the ProvStat operation -- is in the process of evaluating city agencies for needs, deficiencies and their readiness to collect and present data.
"We are working toward a common way to collect data," Cardillo said. "We're examining what technology [agencies] have available to them in terms of hardware, software applications and personnel resources to manage the data."
Why hire a marketing consultant?
"Frankly, we want to brag about this," Cicilline said. "This is really unusual for cities, and we are taking a leadership role. We want people to know Providence is different in that respect."
So far, agency heads have embraced ProvStat because it has worked elsewhere and is a mayor-driven initiative.
The same can be said in Chattanooga where Mayor Bob Corker developed an Office of Performance Review in preparation for the launch of its own version of CitiStat called Chattanooga Results. The office will measure departmental performance with help from data gathered from a new 311 call center.
The office will focus on assessing the city's ability to meet citizens' expectations, as well as issues such as overtime, cell phone bills, telecommunication costs, grants management and procurement.
The office staff includes six members from the city's 311 call center, two from internal audits and two from Chattanooga's grant staff. No additional staff has been hired and no new costs incurred, except for the call center.
"We're beginning to look at the way we contract for goods and services and levels of competition," said David Eichenthal, a former New York City official who now heads Chattanooga's Performance Review Office.
He said the biggest surprise so far has been the level of cooperation from agency heads. "There's always the fear that when you set up something like this office, and when you set up a 311 center, that it's the performance police coming to town. That really hasn't been the case here."
The executive leadership, a driving force in all of these implementations, has made that possible, Eichenthal said. "This is a serious mayoral initiative. It's got strong support from the mayor, and that's made all the difference in the world in terms of bringing department heads to the table to understand this is something that can help them do better."
Stiff Test Out West
Out West, Bratton has new challenges as he leads the Los Angeles police force against an entrenched culture of gang violence.
"Gangs are our biggest problem, both symbolically as well as perception and reality," Bratton said. "The gangs here are very different from the gangs back East. They're literally focused very heavily on violence. Violence is the principle business of the gangs. Ostensibly it's all around drugs and the control of territory, but the engine that fuels it is the sheer focus on violence."
Bratton, who was joined in L.A. by Anemone, his former chief of patrol in New York, will try to tighten the grip on gangs with his version of CompStat. His predecessors in Los Angeles already tried to emulate CompStat with a program called FASTRAC, but it was nondescript when Bratton took over.
"The chief of operations was not driving it," Bratton said. "It was delegated to somebody else in the organization who had no functional responsibility for operations. It just became an exercise every week. It was not focused and it was not prioritized."
Now it is.
"CompStat will eventually, in addition to dealing with crime, be involved in consent decree and anti-terrorism issues," said Bratton, who personally attends each of the city's weekly CompStat meetings, which began this spring. "It's the engine that literally drives the department."
Aside from the renewed focus, Bratton also is promoting the team concept of CompStat -- rather than putting all the pressure on police captains, the burden is on the entire staff.
"CompStat functions best when the whole team is up there," Bratton said. "The captain is assisted by all the detectives who head up the various investigative tables, as well as the special operations units. It's much more of an emphasis on the team approach. That way we get to determine if they're working together."
That type of cooperation is even more necessary in Los Angeles -- where the department is outgunned and underfunded -- than it was in New York, where Bratton had a police force of between 37,000 and 41,000 officers.
"New York had a ton of cops, and with CompStat, we could literally hit a lot of different problems at the same time," he said. "Here, once I get beyond what's assigned to the various areas, I don't have much in the way of resources."
Bratton said the city's 9,300-officer police force is 3,000 to 5,000 cops short of what it needs. "So I would argue CompStat is more critical here. With fewer resources, it's more critical that you use them in a more timely fashion."
Indeed, Compstat already is altering how Los Angeles deploys police resources.
Gangs were fought in the past by small gang units within geographic areas. That will change because gangs don't have geographic boundaries; they're mobile, moving about the city while police were tethered. CompStat will enable police to track gang movement and mobilize resources to take advantage.
"The evolution here will be to take CompStat beyond the idea of jurisdiction and into a multijurisdiction phenomenon," Bratton said. "That's when we can really have impact."
Along the way, the culture of the police force -- which is considered arrogant and resented by at least some of the populace -- will have to be changed. Bratton described the Los Angeles Police Department as "standoffish," as one that wouldn't listen and wasn't really interested in dealing with crime.
"That's why most detectives in this department work Monday through Friday, nine to five," he said. "If they were really serious about crime, they might have thought to have some detectives on nights or weekends."
Bratton is confident he will cut into L.A.'s crime with CompStat, but maybe not to the extent he did in New York due to lack of resources. "It really comes down to resources. We know what to do about crime now. And what's behind it all is the technology -- what originally started with push pins and flip charts," he said.
The LAPD has applied for a $2 million federal COPS grant, which would be used to strengthen CompStat by adding software to automate much of the analysis now performed by humans. The new technology would walk commanders through a series of exercises, prompting the officer about what has and has not been done on a particular case, Bratton said.
"What L.A. would have within a year to 18 months is the next generation of CompStat, in terms of truly using computer technology to maximize what literally began as a couple of cops in New York City sitting around a table with flip charts and acetate easels to track where crime was occurring."