Provided that systems developed to capture individual productivity and performance yes it is possible to put in place differentiated remuneration schemes, regardless of career level. However, on practical level it is very difficult, if not impossible, when we come to academic matters. Otherwise if we can have that system why not?
There are institutions in 3rd world countries in which salaries of academic staff are determined by the number of research publications & not by the number of years in teaching service while salaries of administrative staff are determined by the number of years in administrative service only. Obviously, the system in such institutions is biased & unfair. How is that ?
You can see an administrative scholar holding master degree getting higher salary than an academic scholar holding PhD degree while both have the same years of service.
If the academic scholar is dishonest, s/he can fake research publications and gets points for promotion thereby increasing the salary. We cannot say "scientifically" that deception in research is indicative of higher productivity in such systems. In fact, you will detect a continuous decline in the quality of graduating students produced due to these systems since there has been no reward or regard for "good" teaching.
Ideally, Scientists/Faculty at the same career level should receive salaries based on their productivity and not years of service alone. A performance management system with fixed and variable pay components could be designed to capture the scientific/academic productivity. This would require an objective assessment of key results areas which should be incorporated into the appraisal system.
I do not think that scientists and professors at the same career level should receive the same salary . The main reason is that the true worth of a scientist or professor i s related to the extent to which they try to be professionally active.Therefore, as you have rightly observed, productivity should be an important criterion in determining the amount of salary, because as Martin Luther King, Jr. states, " human progress is neither automatic nor inevitable... Every step toward the goal of justice requires sacrifice, suffering, and struggle; the tireless exertions and passionate concern of dedicated individuals". Therefore , the toil and sacrifice that scientists or professors undergo toward extending the frontiers of the discipline to which they belong should be an important factor in deciding their amount of salary.
Two considerations. 1) It depends on the job description of the candidate. Some american Universities assign three parameters (duties) to their staff in agriculture. For example: 30% research, 50% extension, 20% teaching. On top of that, some scientists have a heavy Committee commitment. The salary system should take these parameters into account. 2) Universities must take quality of publications into account, not just quantity (number of papers).
I think we need a nuanced system. I could see a system with an annual performance pay, with performance assessed based on duties assigned (e.g., a professor with a heavier teaching load would not be expected to publish as much), but this should be a top-up to a career-stage base salary which ratchets up (or down) based on longer-term performance. In the Government of Canada, we do not quite approach this, with three factors determining pay for a reasearch cientist: long term performance determines a pay scale to be used (i.e., RES-01, RES-02, RES-03, RES-04 or RES-05), and years of service at that level determine the pay within that scale. There is no short-term performance pay on top of that, and the annual increments within the selected pay-scale are automatic. There is also no real possibility of falling back to a lower pay scale once promoted should one stop performing.
Defense Research and Development Canada has a similar system but more nuanced: good performance in the short term can result in faster in-range increments, and it is possible (albeit exceedingly rare) to be knocked back to a lower pay scale. It is very complex, but seems to work reasonably well, and provides most of the reward structure one would want to see, including varying the performance criteria according to the duties assigned.
We could take the problem from the other side: why can people keep their job if they don't satisfactorily fill the duties they were assigned? In some Canadian universities, professors are expected to teach, do research, take on administrative tasks (ex. department head, committee member), and sometimes be active in community outreach. If one duty is not filled, it should mean that others take significantly more time than "normal" (ex. the head of a large department often does not have time to do much research or teaching). That means the 3 (or 4) duties of a professor should be viewed as "communicating vessels". If one takes less time, another should take more. The problem is that there is absolutely no consequence for someone doing less (sometimes nothing) for one duty, while not compensating with another. Giving bonuses to those doing well will not bring those doing little to do more.
Researchers and teachers who are equally productive, innovative and creative should have the same high salary, extra money for their true and deep dedication to science, especially for those who do both science and teaching.
In some countries, Academicians and Researchers are promoted based on their outputs and contributions made to the community including teaching and problem solving researches which has implication in increasing their salaries. However, few scientists tend to produce articles for the sake of promotion which is not impact oriented to their community. I have a believe that salaries should be based on contributions but impact oriented to the community instead of counting the number of years of service.