Imagine this scenario: Managers feel intense pressure from senior officials to demonstrate they’re meeting tough performance targets. They push hard on front-line workers to massage the numbers to demonstrate progress. But then stories leak out that managers have been sweeping problems under the rug by misreporting performance numbers. The top dog suddenly finds himself in a harsh spotlight.
Sound familiar? It should, because I’m not merely recounting the recent scandal at the U.S. Department of Veterans Affairs (VA) where managers are accused of keeping a secret list to disguise the wait times for appointments. This is also the tale of the New York City Police Department, which faced charges in 2010 that its famous CompStat performance tool was producing strong pressures on officers to cook the books.
Over the last 20 years, performance systems like CompStat have become the hottest strategy for improving governments’ results. But they’ve also presented real challenges: Are the risks of gaming the system so great that governments ought to walk away from the performance management strategy altogether? Does it work better at the local level where results are front and center than in federal agencies where the front lines can be thousands of miles away?
No one should be shocked that managers try to fake results. One of the first things I learned in my first semester as a professor was that all performance measures are gamed. I was reminded of that every time my students asked, “Will this be on the exam?” The trick for a professor -- and for a program manager -- is to realize that every measure creates incentives, and that good measures create the right incentives to get people to do what needs to be done in the right way.
That’s just what CompStat was intended to do in New York, and its success led to other “stats” like CitiStat in Baltimore and StateStat in Maryland. In a case of innovation flowing uphill, the feds came much later to the game. In recent years, the VA developed one of the federal government’s most robust performance management systems, centered on improving care for veterans. It’s had some significant successes, including reducing homelessness among vets.
But it broke down in the Phoenix VA and in many other hospitals across the country, as workers changed the measures to avoid signaling that vets had to endure unconscionably long waits for care.
The VA tales aren’t that different from the charges that surfaced in the New York Police Department. Former senior commanders reported that they felt intense pressure to drive crime rates down.
When a victim reported a theft, for example, officers would use eBay and other websites to find a lower value for the stolen item than the value the victim claimed. If they could drive the amount down to less than $1,000, it would become a misdemeanor instead of a felony, which are reported to the FBI as major crimes. Officers also claimed they were under pressure to convince victims not to report crimes at all or to adjust the facts to allow the police to downgrade the charge. Like the VA strategies to shorten the waiting list, the commanders said that they felt pressed to make the department look better.
Performance systems work best when top leaders use them to set goals and measure progress instead of using them to punish underlings who aren’t making the grade. The VA’s problem was that the measures imposed from the top down didn’t work so much as a way to manage as to play gotcha. The VA’s senior-level officials were too far away to notice that the numbers didn’t line up with reality, so the performance measures actually dragged down improvement.
Performance measures are meant to help front-line managers get what they need to get done. That’s because performance systems are much more about communicating -- being able to ask smart questions about what is working and what isn’t -- than about measuring. They’re the language that helps pave the road from ambitious goals to firm results, and then about how to drive successfully down the road again.
Local governments have a leg up in this game because top managers can quickly notice if the measures get out of sync with reality, according to NYPD Commissioner William Bratton. The farther senior officials are from the front lines, he says, the harder it is for this self-correcting system to work. That’s the lesson that then-VA Secretary Eric Shinseki, sitting 2,300 miles from the Phoenix VA hospital, so painfully learned.
These performance systems aren’t cheap, either in the money it takes to collect and analyze the data, or in the management time it takes to figure out what to do with it. But they are an essential tool for a 21st-century government that demands more accountability, efficiency and results. It’s here that local leaders have the advantage.