THE stimulus package will more than double the fed eral money being spent on K-12 education for the next two years. But will local districts spend the cash productively?
Flash back to 2002, when school reform was in the air in Philadelphia, one of the nation's largest and most troubled school districts. The district, under state pressure, revamped its governing structure and contracted out its 46 most challenging schools.
Taking charge of 16 schools were local non-profits - the University of Pennsylvania, Temple University, Universal (a community-development corporation) and Foundations (a local entity started by a former school-district administrator). All four had good local political connections, but none had operated a public elementary or middle school.
The other 30 schools were given to for-profit firms - Education Learning, Victory and a third which operated schools for only one year.
With my colleague Matt Chingos, I've now analyzed all available test scores from the Philadelphia school district for 2002 through 2008. To be specific, we looked at all available data for each student in all the for-profit- and nonprofit-managed schools, as well as for the lower-performing half of the regular district schools.
Using a standard difference-in-difference econometric model, we sought to answer to the key question: Did students do better over this period at the for-profit and nonprofit schools than they would have if the schools had remained under district management?
We had 68,000 test observations but they were grouped together at just 46 schools under private management and 71 district schools. As a result, we could only detect large impacts at accepted levels of statistical significance.
Nonetheless, our most surprising result was large enough to pass the significance test: The success in Philadelphia depended upon whether the schools were assigned to for-profit or nonprofit managers.
At the schools run by the for-profits, students learned in math each year an estimated 60 percent of a year's worth of learning more than they would have learned under district management. This was true, on average, for each of the six years after 2002, the year the new management took over.
At the schools run by the nonprofits, students learned in math 50 percent of a year's worth of learning less each year than if district control had remained. But that finding, though large, fell short of statistical significance.
But the numbers clearly show that the difference in for-profit and nonprofit performance was extraordinarily large and significant. Our findings suggest that the students at the for-profits gained more than a year's worth of learning each year than they would have if the nonprofits had run those schools.
The pattern for reading was similar but less extreme: Students at the for-profit-run schools learned two-thirds of a year more each year than if the school had been run by a nonprofit. (We did not detect a significant difference between progress under for-profit managers and what would have happened under district management.)
Experience may explain the results as much as the profit motive. The for-profits had run many public elementary and middle schools, while the nonprofits were doing this for the first time. Yet it was disappointing to see no obvious sign of nonprofit improvement over the six-year period.
The decisions taken by the Philadelphia school district last year are also disappointing. The district took back management of five schools from for-profit managers, but just one from the nonprofits. Nothing in our data validates those decisions. One is left to wonder just how Philadelphia's leaders measure success - and, indeed, how school systems nationwide will spend their stimulus windfalls.