There's some truth in the argument that making public services perform to a statistical target can distort they way the operate in favour of creating "good" statistical outcomes, I'd argue that this isn't happening here - or at Cambridge High.
In the latter case, the "traditional" education methods at the school were failing, as they usually do, with students of below average academic abilities. With School Cert, etc. this would show up less, as low ability students would be expected to fail. A more sophisticated system of examination, NCEA, made this problem more obvious. Instead of changing the methods of teaching, the principal chose to fiddle the assessment process.
With the "P" labs, the police are faced with the thankless task of enforcing drug prohibition. In conventional crime, success in enforcement does not increase the rewards for undetected crime - if the police arrest more burglars, it doesn't mean that those remaining are able to steal more or sell their stolen goods for more money. With drug dealing, the more drugs the police interdict, the higher the price goes and the greater the profits for those still in the business. So it's quite reasonable that the cops should concentrate on victimful crimes rather than drug enforcement.
So, in my biased opinion both NCEA and the model of police performance assessment are working - they are just throwing up unpalatable results to some (e.g. believers in drug prohibition or 'traditional' education).