"Research is mixed on merit pay's success."

The latest report from a New York City teacher incentive plan sheds more doubt on the efficacy of merit pay for educators.  However, this experiment mostly rewarded schools instead of individual teachers.  The successful incentive arrangement cited in the article rewards individual teachers based on student performance.

From The Wall Street Journal:

In 2007, New York City and its teachers union launched an experiment to determine whether rewarding teachers with extra cash would boost student performance.
Four years and $57 million later, the answer appears to be no. Backers of incentive pay are blaming the way New York's program was structured, and school and union officials are pointing fingers at each other....
But unlike traditional incentive programs, New York didn't identify stand-out teachers and shower them with money. Instead, in a compromise reached with the union, the city gave bonus pools to schools that had performed well.
Teachers could vote on whether to distribute the money evenly or to specific staffers. Most schools chose to distribute the money evenly, according to a Harvard University study. Backers of merit pay for teachers say that eliminated the direct link between individual performance and rewards....
Research is mixed on merit pay's success. A rigorous and closely watched study of a Nashville incentive-pay program found it didn't improve student test scores, while a study of Denver's merit-pay initiative found it attracted higher-quality teachers and kept them in hard-to-staff schools.
But there is a key difference between those programs and New York's. Both Nashville and Denver directly linked performance pay of teachers to the performance of students in their classes. The Denver program also considers classroom evaluations as part of the bonus pay and allows teachers in non-tested subjects to get cash based on schoolwide improvements.
From the start, the New York approach had its skeptics. Some said the bonuses needed to go to the teachers whose classes made the biggest improvements, not be spread among all the teachers of a school....
Two studies—one by Columbia University and the other by Harvard—found recently that student test scores at schools that received bonus money didn't improve at any better rate than at schools that received no money. In fact, both studies found that in some subjects and grades, the program resulted in slightly lower achievement.
Researchers posited that because the bonuses were based on how well entire schools performed, and how well they performed compared to similar schools citywide, the money didn't offer much incentive to individual teachers to excel.
"It was clear in 2007 that this plan wouldn't enable the best teachers to earn dramatically more, and therefore would likely be limited in long-term effect," said Bryan Hassell, co-director of Public Impact, a research and consulting organization that is often at odds with the teachers union. He wasn't involved in the studies.
"This plan paid chump change compared to what the best teachers should be earning for reaching more kids successfully," Mr. Hassell said.