If performance pay is properly designed, I think you can avoid both problems. In Queensland, every quarter each school sends representatives for each subject to a district "teacher's meeting," and as a component of this, the teachers must assess each other, from test content to top students. I'm not sure of the exact details, but I know it was designed to ensure each school was testing their students equally. I remember that my physics teacher was made to make the final question in our final physics test relatively complicated in order to justify giving certain students top grades.
There is also a state-wide curriculum (or is it national now?), which in my opinion is quite broad, and teachers are made to test many aspects of it.
On top of this district-based system, there is a state-wide test which gives a weighting factor to the grades of students in each school and also to the students within each subject within each school. If a teacher in one school was to somehow bypass the district-based system and give students easier tests, one would assume the students would perform worse in this national test, which would then impact their overall grade regardless. The opposite it also true.
As an interesting side-note, because of the implications bad students have on good students as a result of this weighting system, some schools pressure underperformers (e.g. weed smokers, kids who always skip class, etc) to leave and find work. I'm sure there are varied opinions about whether this is good or bad.
Anyway, to conclude, if performance pay took into consideration the systems that are designed to make assessment fair, then I think it would have a better chance at benefiting students. As with most systems, design is important.
