The literature I'm familiar with on the topic points out a few important lessons in the pay-for-performance debate:
There has to be money in the budget to honour the commitment or else things start to get twisted around and it blows up in your face. An exquisite paper in Public Personnel Management a few years ago documented some of the changes that happened in resonse to the Georgia Gain program in that state over a decade. Most notable was a progressive downward drift of performance appraisals as a means of coping with the costs of honouring improved performance. And after busting their hump to earn those performance bonuses, many employees were not particularly thrilled that their extra effort was rewarded with mediocre performance appraisals...so they left. Moral? If you can't afford to honour it, don't even attempt it.
It is the case that in the private sector greater productivity is assumed to generate greater revenue for the employer, and performance pay costs are offset by that greater revenue. In the public sector, better performance can often increase employer costs rather than reduce them. If social workers identify 50% more children encountering abusive homes who now need to be protected and placed somewhere at cost to the state, their performance may well be better, but that has resulted in more expense to the state, not more revenue. The bottom line is that many areas of the public sector simply cannot afford the consequences of improved performance, despite how much they crave and promise it.
In the private sector and in some areas of the public sector, performance is easily measured by objective outcomes. In many areas of the public sector "performance" is not so easily operationalized, and in some instances attempts to operationalize it at the individual level can end up rewarding those things that undermine performance at the organizational level.
Case in point. My wife works in an agency that oversees a form of chemical product safety. Manufacturers submit their safe-use instructions/advisories to the agency, and bottom rung evaluators check the documents to make sure that the information is complete, up to date, and coherent. Once out of their hands, the documents go to a reviewer whose role is to impose some degree of uniform quality control across such documents. A simple view of "performance" would suggest that an evaluator who can get through more manufacturer submissions and pass them on to the reviewer is doing a better job. As my wife has been finding out, though, emphasis on "the numbers" is effectively imposing greater burden on the reviewers who now have to make up for the slapdash job done by others at the lower level who are in a hurry to look good. The lower rung comes across as more efficient, but overall throughput is not improved one iota, simply because of the extra work that must be now done by the second rung.
So whatever the indices of performance are, they need to be thoughtful and not counterproductive, Because if you're going to pay people for more of it, it better damn well be exactly what you want and need more of. Just because you can count it objectively does not make it the right thing to reward.
Another more recent paper that surveyed a number of American state and municipal jurisdictions about their use of broader pay-bands and discretional bonuses for better performance revealed another critical element: parity across managers. Believe it or not, folks who work in the public sector talk to each other. And when the apparent criteria used for performance-pay in one work unit do not seem to line up with the stringency of criteria for performance pay in another, people learn about it fast. This can be very disruptive. The "referent cognitions" model of organizational justice proposes that people evaluate the fairness of individual/organizational transactions by comparing them to what could have happened instead, generally in an upward, rather than downward direction. So if I know that other places are awarding performance pay for what I consider to be something equivalent to my own level of performance....and I'm NOT getting any such bonus, then I interpret my own circumstance as unfair, because they could have implemented performance pay, but didn't. Unlikely to be healthy for retention or engagement.
The lesson from that is that it is not good enough to merely empower managers to reward good performance and give them the budget to do so. They have to operate under the same agreed-upon standards and monitor parity over the long haul. That will necessarily involve some training of managers.
In many levels of government and in many sorts of business lines, what "performance" consists of is open to question and even when you think you have a handle on it, it is often out of the individual performer's control. Governments and senior bureaucrats come and go. Projects get funding cut, and priorities of the day change in an eyeblink. So, um, what have you done lately? Well, I know I tried to do a lot, but was not able to finish it for reasons XYZ that "came from above". I know our government has attempted to use regular employee surveys (similar to the Federal Human Capital Survey) to assess senior bureaucrat performance for performance-pay purposes, and has encountered tremendous pushback, largely because much of what is being measured is often out of even senior officials' control, or is intrisically easier to do in agency A, than in agency B, simply because of their business lines, geographical distribution, and even current events. So, whatever it is about performance that you're rewarding, it has to be well within the peformer's control.
Finally, as far as I remember, bonuses and pay differentials need to be somewhere around 7%, at least, to motivate people to change their behaviour or job. I may be wrong about the specific percentage, but the basic principle is that there is a minimum benefit to the employee required before they will respond to the benefit and change some aspect of their behavior.
Put this all together and whaddya got? If there is something that individuals can do in their job, and accomplish, via discretionary effort, and you can measure it objectively, and it is something that can ONLY provide organizational benefit when accomplished, and the rules and standards can be equitably applied, and you can continue to afford to do it over the long haul, and the amount of earnable reward has some incentive value, then MAYBE your performance-pay program stands a chance.
It's a lazy day here today. I hope this counts as my "performance", but I suspect not.