How does it work? Here's Adnot et al in a new NBER working paper:
DCPS [District of Columbia Public Schools] began evaluating teachers under IMPACT, a new performance-assessment and incentive system, during the 2009-10 school year. The design of IMPACT appears consistent with virtually all of the emerging best-practice principles. First, all teachers are evaluated on a multi-faceted measure of teacher performance (e.g., clearly described standards, the use of multiple teacher observations made by different observers and the use of student outcomes). Second, these evaluations are linked to high-powered incentives that include the potential dismissal of low-performing teachers and very large financial incentives for high-performers. Third, in addition to the feedback associated with the evaluations, teachers are provided with various supports, including instructional coaching to assist in improving their teaching practice.It isn't just based on standardised test scores. Rather, it combines outcome measures with other evaluations of teacher performance for a better overall picture: students with poor absolute outcomes could still be doing far better than expected given their backgrounds.
The departure of poorly performing teachers improved teaching quality (the average replacement was of higher quality than the departing teacher). It also improved student achievement by a fifth of a standard deviation in maths, and 0.14 standard deviation in reading.
And, the benefits are concentrated in the poorest schools:
Forty percent of teacher turnover in high-poverty schools is among low-performing teachers (Figure 3). Our estimates indicate that there are consistently large gains from the exit of low-performing teachers in high-poverty schools. In math, teacher quality improves by 1.3 standard deviations and student achievement by 20 percent of a standard deviation; in reading these figures are 1 standard deviation of teacher quality and 14 percent of standard deviation of student achievement. In DCPS, virtually all low-performing teacher turnover is concentrated in high-poverty schools: on average, 1 percent of students in low-poverty schools experience low-performing teacher turnover.15In the D.C. data, there was a lot of turnover of poor teachers in schools serving poorer communities. There was less of it in richer communities - somehow, those communities seem to have had smaller proportions of poor-quality teachers to start with.