Last month the Kennedy School of Government and the U.S. Conference of Mayors convened a seminar for newly elected mayors. It came as no surprise to the conveners that performance measurement was paramount in the minds of the participants. The spectacular success of CompStat in New York City -- CompStat is credited with dramatically decreasing crime rates -- and Baltimore's CitiStat, the gold standard in civil performance measurement, guaranteed as much. Indeed, few government reforms over the last decade have garnered as much acclaim, or inspired as much emulation, as these two programs.
Recognizing the importance of performance measurement to new Mayors, Bob Behn, the country's leading expert on performance management, used last week's Management Insights column to continue the discussion initiated at the Mayors' Conference. Bob's column poured cold water on the idea that implementing a stats program is a panacea, and he delineated what he sees as "The Five Big Errors of PerformanceStat."
In this column, I respond to Behn's list of five "big errors." Bob's list, while interesting and helpful, describes relatively minor obstacles. New officials would be well advised to focus on truly big issues with the goal of turning "performance" into public value. Correcting the errors on Bob's list will insure the integrity of the mechanics of the programs, which is critical but also insufficient.
Let's imagine a situation involving the now famous CompStat, which took the vision of Rudy Giuliani and the leadership of Police Commissioner Bill Bratton, and transformed a crime wave into relative tranquility. What if the mayor and commissioner had not begun by sharply changing the vision of the department? What if they had not started by deciding that problems like broken windows matter? What if they had not realized that citizens would not participate when the bad guys owned the neighborhood and the police arrived only to make an arrest or take a report from a victim? There would be plenty of stats to manage in that situation -- mostly bad. Or what if Bratton had rarely shown up to meetings and Giuliani had only paid lip service to the concept, and neither backed the process with resources and personal support? Clearly, even if the "five Behn errors" had been avoided or vanquished, the crime rate probably would have remained troublingly high.
So let's look the truly big issues that need to be tackled before managers can move on to address the five little errors.
Leadership, leadership, leadership: The elected official, or top official in an agency, has only so much time in a day. Since it's inevitable that a crisis will arise that justifiably takes that official away from a regularly scheduled process meeting like performance stat, this eventuality must be addressed beforehand. If the mayor attends and actively participates in the performance stat meetings, change will happen. If not, it won't. A drove of brilliant young data analysts and bombastic senior managers cannot make up for the absence of hands-on executive leadership.
Measure value, not activities: Only measuring activities is too simplistic. What if the activities don't result in much of the public value that is their intended goal? Here, the very important opportunity and problem is to first define public value. Perhaps the performance unit is not speed and cost of the emergency room's treatment of indigents, but rather how the neighborhood health clinics were able to improve the outreach and treatment to those in need, that they were able to effectively reduce the number of emergency room admissions. After all, it's public health, not operating controlling hospital costs, that provides true public value.
A numerator without its denominator: Improving performance requires not only leadership dedication, but the dedication of resources. No matter how excited the mayor might be about the process, all management priorities require tradeoffs. Similarly, better performance can typically be achieved -- in short bursts -- by temporary spikes in resources. Think here of saturation policing: you have a rash of neighborhood burglaries, so you redeploy officers on walking patrols and stakeouts, but for how long? This can even apply to routine services like trash pickup, which can be improved with dollars: shorter routes; and, more laborers working more slowly, making sure the cans are all set upright and no litter remains from the process. Measuring performance in most areas is meaningless without an understanding of the cost per unit of activity. If the numerator is performance, the denominator should be the cost of the activity -- at least for any performance that needs to be sustained.
Speed-walking in place: Marginal improvements in the numbers resulting from much harder work may obscure the fact that the system being used is so antiquated and bogged down with bureaucracy that performance stat efforts may obscure the much more urgent need for redesigning the process from the ground up.
The curse of professionalism: Performance stat is a wonderful way to bring out the best in professional managers, but what if they do not really have all the necessary knowledge? What if, in fact, the citizenry know important details not sought after by the manager, or they have a very different reaction to quality? I am reminded when, during my term as mayor of Indianapolis, we relentlessly measured how well and how frequently the grass in the parks was cut (and indeed measured the cost), only to be told via customer surveys that users thought the parks looked worse because the bathrooms were inadequate and dirty. Obvious you say, but it still took the crowd to remind us. Moreover, sometimes the crowd can be a mayor's own line workers, often ignored by their own managers.
Watch what you're measuring: Stat programs do a terrific job (if the errors identified by Behn are avoided) of motivating public employees to perform better. But what if the solution to the problem requires government to manage a network of public, private and not-for-profit providers? The performance measured should include how the public managers leverage results across the network, not just in their own domain. Take the goal of reducing teen pregnancy: The performance being measured should not be just how much counseling or outreach the public-health nurses have accomplished, but also how effective the schools, neighborhood organizations and houses of worship are in their outreach. Performance stat might find difficulties in measuring the performance of network participants, but it can and should measure its own employees on their management of the relationships within the network, and then compare their results.
Federal, state, and local officials are flocking to stat programs. They hold great promise, and indeed the warnings voiced by Behn need to be heard. However, radical improvement requires a city or state leader to first address the truly big challenges.
For resources related to this article and to Behn's article, please visit the Government Innovators Network.
The Right Question
Stephen Goldsmith raises important issues about what you should "chase" in a performance stat program. In Bob Behn's recent article for mayors ("What All Mayors Would Like to Know About Baltimore's CitiStat Performance Strategy"), he extols the improvement in the number of potholes that the city's public works department was able to attain. Indeed, given the level of service previously provided, this is a significant improvement in work produced.
However, Baltimore's relentless pursuit of outputs left me wondering if it was the right question to begin with. Shouldn't the real question be "Why are we getting so many complaints about potholes? Why are we having to fill so many potholes? Shouldn't the number of potholes requiring filling be going down over time?" Ultimately, the more strategic questions needing to be addressed are: "What are the conditions of our roads? Is our overlay strategy effective and sufficient? How do residents perceive the quality of the roads?"
Until there is a stronger conceptual and management connection between the outputs we produce and their intended outcomes, a performance management system in place to measure both, as well as a strategic perspective on how to best make improvements (and I might add that filling potholes is not that strategy) we will be left providing accurate and timely answers to the wrong questions.
Performance Management Director
King County Executive's Office
Posted December 21, 2007
Focusing on Core Measures
Over the past 12 years, the Dallas County Tax Office has implemented performance-incentive programs in two different departments that have produced very successful results: highest productivity per staff person in the state; significantly increased customer compliments/reduced complaints; reduction of total staff; significant improvement in response times (reduced customer wait time in lines, faster telephone response time, faster response to mailed requests); overall improved productivity per staffperson of 12 percent; two state-level recognitions and one national recognition for quality management; in 2007 the only government agency in the state of Texas to be recognized by the Texas Baldrige Award (congrats to Coral Springs, Florida, for being the first government agency ever to earn a national Baldrige Award).
In short, our program has been very successful. I believe that it has been successful because 12 years ago, we created, fine-tuned and tweaked it based on three core principles: (1) we would focus on only three or four core measures (based on the Southwest Airlines model in which they have one core measure, profit per seat, and all other measures, analyses and decisions in the entire organization can ripple out from that one core source that serves as an indicator of the health of absolutely every other aspect of the organization); (2) we would make sure that the three or four measures were easy to gather and monitor (to this day department managers can create their monthly performance report in an hour and a half per month); and (3) we created a performance reward program, not a performance punishment program (solving Dr. Behn's observed dilemma of brutal vs. bland).
From the beginning, the program was designed by the staff and managers as a tool to create challenging, but obtainable, performance standards, then provide monthly bonus pay to each employee who exceeded the performance standards during that month. This creates buy-in and excitement.
I have presented these concepts at several national conferences in recent years (most recently just two weeks ago in Las Vegas) and always receive extreme interest from other agencies. I have been a student of the performance measurement movement since it began, I suppose, in Sunnyvale 25 years ago, moved to Indianapolis with Mayor Goldsmith, then has grown somewhat since then. It amazes me that we are now 25 years into this movement and it has not progressed any more than it has. Of course, the two core obstructions have been the fact that too often measures have been presented as a "gotcha" punishment and government's fear of accountability. That is why I have believed that the key to greater success is to present measures in a positive light, as a way for the department head to report successes at improving the agency, and a way for the staff to be recognized for their personal efforts and contributions to the organization.
Such an approach has certainly created significant improvements in my organization. There is absolutely no reason that measures cannot begin being presented as an opportunity and that a quality group of "core" measures cannot be developed (for fire departments, street departments, etc.) for use and adaptation by government agencies. These should be the two missions of the performance-measurement movement for the next 10 years.
David Childs, Ph.D.
Dallas County Tax Assessor
Posted December 21, 2007
You may use or reference this story with attribution and a link to