Merit pay for teachers - rambling thoughts on the issues

One answer to the issue of bad performing schools/students in the US has been the idea of merit pay for teachers. The Economist  reported on the topic recently using the case of New Jersey state, which, according to the article, has a real problem in its hands already since more than 15 years. Ugh..that's a long time, a whole childhood, if you wish..

Although they have been under the supervision of New Jersey’s state government since 1995, there has been little improvement since then. Only 40% of students read to the standard prescribed for their age, and in the 15 worst-performing schools the figure is less than 25%. More than 30% of pupils do not graduate. Few of those who do are ready for higher education. Of those who entered one local establishment, Essex County College, in 2009, a whopping 98% needed remedial maths and 87% had to take remedial English. As a result, fed-up parents are taking their children out of Newark’s public high schools and placing them in independent charter schools. Many public-school buildings now stand half-empty. The best teachers often leave in despair.

Merit pay for teachers Bonus time
A new contract for teachers is shaking up New Jersey’s largest city

It's impossible to understand the whole magnitude of the problem, but for now, I'm only wondering how such merit pay system works in an educational establishment? What is it based on? What does it really evaluate? Does it take into account how the students' perform, say, during the last academic year? Or how did they perform with other teachers for that matter? Or does it only concentrate on the teacher? What is the baseline for the evaluation? How well did you teach last year? How many hours you taught for a give group of kids? How well the kids that you teach perform in a standardised test over a period of time (which period;)? Is your "merit" based on your average performance with different group of kids if you for example teach math to several groups of kids?

Creating metrics for teacher performance has gotta have been a hec of a job, they probably spent lots of resources trying to figure that out. And then, I would also like to see how does it plays out in the school itself. Is the performance of all the teachers fitted on the normal curve to give the "grades"/merit pay? What if all the teachers' started teaching better, does it mean that non of them get any extra money?

I pay lots of attention to the TALIS study (OECD, 2009) where teachers' appraisal and feedback mechanisms as well as the feeling of self-efficacy are studied and discussed. The whole issue isn't only about the "quality" of teachers'  performance that is delivered to the kids, but the Professional Development and teacher cooperation opportunities in general play an important role too. Placing more emphasis on those factors might be a more helpful way forward..