PS 49 in Queens used to be an average school in New York City's decidedly below-average school system. That was before Anthony Lombardi moved into the principal's office. When Lombardi took charge in 1997, 37 percent of fourth graders read at grade level, compared with nearly 90 percent today; there have also been double-digit improvements in math scores. By 2002, PS 49 made the state's list of most improved schools. If you ask Lombardi how it happened, he'll launch into a well-practiced monologue on the many changes that he brought to PS 49 (an arts program, a new curriculum from Columbia's Teachers College). But he keeps coming back to one highly controversial element of the school's turnaround: getting rid of incompetent teachers.
Firing bad teachers may seem like a rather obvious solution, but it requires some gumption to take on a teachers union. And cleaning house isn't necessarily the only answer. There are three basic ways to improve a school's faculty: take greater care in selecting good teachers upfront, throw out the bad ones who are already teaching, and provide training to make current teachers better. In theory, the first two should have more or less the same effect, and it might seem preferable to focus on never hiring unpromising instructors—once entrenched, it's nearly impossible in most places to remove teachers from their union-protected jobs. But that's assuming we're good at predicting who will teach well in the first place.
It turns out we aren't. For instance, in 1997, Los Angeles tripled its hiring of elementary-school teachers following a state-mandated reduction in class size. If L.A. schools had been doing a good job of picking the best teachers among their applicants, then the average quality of new recruits should have gone down when they expanded their ranks—they were hiring from the same pool of applicants, but accepting candidates who would have been rejected in prior years. But as researchers Thomas Kane and Douglas Staiger found, the crop of new teachers didn't perform any worse than the teachers the school had hired in more selective years.
This unexpected result is consistent with the findings from dozens of studies analyzing the predictors of teacher quality. Researchers have looked at just about every possible determinant of teaching success, and it seems there's nothing on a prospective teacher's résumé that indicates how he or she will do in the classroom. While some qualifications boost performance a little bit—National Board certification seems to help, though a master's degree in education does not—they just don't improve it very much.
It's worth keeping in mind that economists study changes in test scores, not love of learning or comprehension of course material—it's possible that some of the teachers who look good to researchers are just good at teaching to the test. Needing some measure of success in the classroom, economists mostly rely on "value added" in test scores—that is, how much students' scores improve as a result of a year in a teacher's classroom. Since researchers study entire school systems over many years, they're able to separate out how much of an individual student's improvement is due to personal circumstance and how much is the result of inspirational teachers. If a student's test scores increase year after year, then no teacher gets any credit for it; similarly, no one's on the hook for a bad student's repeated failure to progress.
What economists have found is that only one thing tells us how much a teacher will boost his students' test scores next year: the amount he raised test scores in previous years. A good teacher this year will very likely be a good teacher next year. Unfortunately, when making hiring decisions, principals rarely have that information at their fingertips. Most hiring decisions are made before applicants have a teaching record. And an individual school has neither the necessary data nor the ability to run the complicated regression analyses needed to discern whether an experienced teacher has had a positive effect on his students in the past.