Andrew Biggs of the American Enterprise Institute offers a good answer in a National Review Online column.

Desperate times demand desperate measures. And so I propose an education-reform plan that is simple and achievable, and that peer-reviewed academic research indicates would improve student achievement and future earnings, save government budgets billions from its very first year, and boost long-term economic growth.

Here’s the plan: Fire the worst teachers. That’s it. Terminate their employment. Don’t replace them. Simply reallocate their students to other classrooms. This approach isn’t merely a sign of desperation, although a bit of desperation is hardly unwarranted after decades of unfruitful efforts. The “fire the worst teachers” strategy may in fact be the optimal way to improve education, a first-choice approach rather than a last resort.

Economists Raj Chetty and John Friedman of Harvard and Jonah Rockoff of Columbia University have documented the cost to students of having a poor schoolteacher. Chetty, Friedman, and Rockoff estimate that terminating the worst 5 percent of teachers and reassigning their students to average-performing teachers would increase those students’ lifetime earnings by between $130,000 and $190,000. Put another way, to convince a rational child to willingly accept a poor-performing teacher, you’d have to write that child a check for up to $190,000 on the spot. These figures indicate that eliminating bad teachers may be the quickest way to improve the job prospects of low-income Americans, reduce income inequality, and boost our future economy.

But this fire-the-worst policy could have benefits beyond those calculated by Chetty and his co-authors. Economists Thomas Dee of Stanford and James Wyckoff of the University of Virginia recently analyzed Washington, D.C.’s IMPACT program, which financially rewards the best teachers and threatens the worst with dismissal. Dee and Wyckoff compared teachers who were near the thresholds for bonuses or dismissal threats with teachers who actually crossed the thresholds. They found that the highest-graded teachers performed even better as a result of IMPACT’s bonuses. The lowest-graded teachers who received dismissal threats, by contrast, were more likely to either improve their performance or to quit than nearly-as-poor teachers who weren’t threatened with firing. In other words, the threat of dismissal caused the worst-performing teachers to either shape up or ship out. In either case, the quality of teachers in the classroom improves.