Features

May/ June 2013 A Short History of Data-Driven Government

By Haley Sweetland Edwards

If it can be said that there is a father of data-based government, it is a famous and controversial one: Robert McNamara. As a young captain during World War II, McNamara was assigned to the Army Air Corps Office of Statistical Control, where he applied the ideas about efficiency and cost-effectiveness that he’d learned at Harvard Business School to questions of defense: What should the schedules be for the B-29 bombers? When is it most efficient for them to carry fuel and cargo?

In 1946, shortly after leaving the service, McNamara and nine of his former colleagues were hired by Henry Ford to help make the ailing Ford Motor Company profitable again. The “Whiz Kids,” as they were called, applied their relentless statistical strategies and quickly turned the company around. But, foreshadowing the problem with relying too heavily on measurements of efficiency—and not, say, waste—workers at the Ford factory began chucking their leftover parts into the river, giving rise to the joke that a man could “walk on water—atop rusted pieces of 1950 and 1951 cars,” write Viktor Mayer-Schoenberger and Kenneth Cukier in their book, Big Data: A Revolution That Will Transform How We Live, Work and Think.

Still, McNamara rocketed up through the ranks of Ford, and in 1960 was named president. A few weeks later, President John F. Kennedy appointed him secretary of defense. It was from that perch, under Kennedy’s successor Lyndon B. Johnson, that McNamara, applying those same ideas about efficiency, instituted the now-infamous “body count policy” in the quickly escalating Vietnam War. Daily newspapers published the number of enemy killed as shorthand for progress in the war effort, and soldiers were, in turn, rewarded in kind: the more people killed, the better. Years later, McNamara’s policy was roundly maligned by liberals and conservatives alike, underscoring the problem of data-driven policies that, by failing to quantify holistic goals, sometimes incentivize appalling behaviors. But in the mid-1960s, it still seemed like the future. And in many ways, it was.

Embracing McNamara’s vision whole hog, Johnson insisted that all of the federal agencies adopt data-driven decisionmaking strategies as well. His policy, the Planning, Programming and Budgeting System (PPBS), required agencies to create and report multi-year plans outlining measurable results. Around this time, the trend of data-driven management also began to catch on at the local level. In New York, Mayor John Lindsay began producing the mayor’s “Management Report,” which included performance indicators for each of the city’s departments, a system that is still around today. But at the federal level, despite LBJ’s enthusiasm, the data-based decisionmaking trend was short-lived. By the late ’70s and ’80s, it had almost entirely petered out.

It wasn’t until the early ’90s, under the banner of President Bill Clinton’s National Partnership for Reinventing Government and, specifically, the Government Performance and Results Act (GPRA), that the idea was resurrected. As part of those programs, agencies were again required to create operating plans and annual productivity reports. Representatives from the private sector were brought in to teach government employees how to do things like write an annual report. While GPRA did not require agencies to participate until 1999, about seventy pilot projects got off the ground earlier, with some success. The Coast Guard, for instance, after using data to map the problem of crew casualties in the towboat industry, was able to reduce fatalities by two-thirds from 1997 to 2000.

During the 2000 presidential race, Al Gore’s transition team planned to keep and re-up GPRA, borrowing some of the ideas from then Baltimore Mayor Martin O’Malley’s new CitiStat program to create a comprehensive, government-wide, data-driven performance management tool. “But it was one of those transitions that never took place,” says Harry Hatry, a fellow at the Urban Institute, who was one of the authors of the plan. When George W. Bush took over in 2001, he ignored most of GPRA’s requirements in an effort to reduce paperwork and, in its place, implemented the Program Assessment Rating Tool. PART, as it was known, had some good ideas—senior officials, not lowly bureaucrats, were required to do the goal setting, for example—and some bad ones. Budget evaluators could easily rank programs according to whether or not they agreed with their ideological ends.

When Barack Obama came to power, he kept the valuable aspects of PART, dusted off Clinton’s version of GPRA, lifted the managerial strategies from Baltimore’s CitiStat (and now Maryland’s StateStat), and put them all together to produce the GPRA Modernization Act (GPRAMA), which passed Congress in 2010. One of the key differences in Obama’s plan was that it adopts what is known as the “Stat” model: agencies are now required to submit quarterly progress reports and gather for regularly scheduled, structured, data-driven meetings.

GPRAMA, or any data-driven performance management tool, is not, of course, immune to the pitfalls made infamous by McNamara. “Every measure, every indicator, can be gamed,” said Donald Kettl, dean of the School of Public Policy at the University of Maryland. “You have to try to set up the game so people play it to win the way you want it to be won.” Skeptics tend to regard the effort as Sisyphean. (When I asked one longtime government watcher about the prospect of implementing the “Stat” model at the federal level, he laughed aloud: “That’s the funniest joke that will ever be published in the Washington Monthly,” he said.)

While it’s too early to tell if GPRAMA has made a significant impact, the U.S. Government Accountability Office released a report in February calling the program “promising” and citing some early successes. For example, the Treasury’s Bureau of Engraving and Printing used the “Stat” model to transition to a centralized accounting system—a move that the Financial Management Service estimated could save the government $400 million to $500 million annually. If, over the next three years, the Obama administration is able to institutionalize elements of GPRAMA, the next president, or VP, whoever he or she is, could inherit a truly valuable tool. —H.S.E.

Haley Sweetland Edwards is an editor of the Washington Monthly.