UW Staff Merit Pay — Suggestions

Ray Butterworth — 2018 November 23


I've already written about UW's merit pay system for staff:

You don't need to read them to understand this report, but they do contain real examples and numbers should you want more specific details.

UW Merit Pay Calculations

Here I'll concentrate specifically on the merit pay calculations and completely ignore other problems, such as the inconsistent ratings from one department to another.

I have not researched the specific formulas used and might have some details wrong, but even so I believe that any such misunderstandings would not significantly affect my observations and conclusions.

Except where specifically noted, all graphs and figures have been adjusted for inflation.



Professor Salary Throughout Career

To the best of my knowledge, professors always receive the full annual cost of living allowance and then get their merit increase on top of that. A graph of a typical salary would start low, and increase slowly but exponentially throughout the professor's career. The graph can level off, but can never decrease.

This is not unreasonable: professors generally become more valuable with time. Throughout their career their reputations continually build. Research papers that were published decades ago are used as references, and still add prestige to the University.

Jack Edmond's Paths, Trees, and Flowers is still frequently referenced, 55 years after its publication. That title produces tens of thousands of hits on Google. (Coincidentally the paper is dated exactly 55 years ago today, 22 November 1963, which also coincidentally was the day JFK was assassinated.)

And a much more extreme and recent example is Donna Strickland. If she spent the next ten years doing nothing but sitting in the Grad House, drinking beer and reading comic books, her presence alone would still be worth millions to the University.


Staff Salary Throughout Career

Salaries for staff are completely different from those of professors.

A typical member of staff starts out with a very low salary and a promise that it will rise quickly at first and eventually level off at a target salary determined by their merit rating.

Again, this is not unreasonable: new staff are not very productive initially, but rapidly become more valuable as they learn their job and acquire and develop their skills. Their past achievements are of little value to the University.

That Robyn did amazing things with SGI systems, or that I did amazing things with Honeywell's software is long forgotten and added nothing to our worth as employees twenty-five years later. To be brutal about it, our past achievements didn't imbue us with any value to the University even one year later.

Staff members are only as valuable as the service they can provide now and in the near future. That value is already predetermined by the USG ratings of their jobs and their target salaries.

After a few years, when a salary is starting to level off, one can get a significant raise only by moving to a different job. Some of us do stay in the same position for our entire careers and as a result our salaries remain stagnant, but that is our choice. Again, this isn't an unreasonable way for the system to work.

Merit Ratings


As previously mentioned, a professor's salary can never decrease. Should professors ever start getting bad reviews, their graphs would level off, but the annual cost of living allowance would ensure that their salaries continue to approximately keep pace with inflation.


For staff, it would not make sense if salaries could never decrease. Otherwise, after a few years one could simply goof off and continue to receive a salary without providing any real benefit to the University. The merit system therefore must include a mechanism for producing real reductions in staff salaries.

Since it is somewhat demoralizing to receive a pay cut, UW uses the cost of living allowance to hide the actual reduction in salary. This cost of living allowance is not automatically added to staff salaries as it is for professors. Instead it's added to the nominal value of the USG rating for the position, which raises each employee's target salary, which increases the distance from the actual salary, which determines the annual raise.

This method means that staff will always receive at least a small annual increase, which keeps them happy, but it also allows the possibility of increases that are less than the cost of living allowance, which keeps the administration happy.

Problems for Staff

In what follows I refer to poor reviews and bad ratings, but those terms shouldn't be taken literally. A rating of 4.25 is considered good by most people, but for my purposes here I will still refer to it as bad because it is lower than the person's usual rating of 4.5.

Poor Merit Ratings

Staff members that receive a lower than usual merit rating are given a lower target salary. Since raises are calculated based on the distance from current salary to target salary, in most cases staff with a lower than normal merit rating simply get a smaller than usual increase, not ever realizing that their pay has actually been cut, in terms of inflation-adjusted dollars.

For new employees, at the steep increase beginning of their careers, this difference has very little effect; they still receive a substantial increase. And if the lower than normal ratings should become normal for them, the employee simply has a new target salary for the rest of their career.

But for long-time employees it can be much more serious, as previously discussed in detail in the two items listed at the top. Such people normally receive annual merit increases of ½% or less, otherwise they might exceed their targets. A one-time bad rating lowers their target for that one year, and they can lose not only their usual ½% merit increase, but also some or all of the cost of living allowance.


Consider two employees that have received consistently identical merit ratings for years and are now close to their target salaries. A few years ago Fred had personal troubles, got a one-time rating .5 below normal, and since then has recovered and has continued receiving his usual rating. The identical thing happened to Mary, though her bad year wasn't the same year as Fred's. In both cases the resulting new target salaries (5% less than before) were well below their current salaries.

In a sane world, a reasonable man would expect that the effect would be the same for each of them. But this isn't a sane world.

In Fred's case, the cost of living allowance for that year was ½%. Fred didn't get his normal ½% increase, and he lost the cost of living allowance of ½%. This meant that his one-time bad rating caused him to lose this year's normal ½% merit increase and to wipe out the previous year's ½% merit increase. One bad year moved his salary's path back to what it was two years ago.

In Mary's case, the cost of living allowance for that year happened to be 3½%. Mary didn't get her normal ½% increase, and she lost the cost of living allowance of 3½%. This meant that that one-time bad rating caused her to lose this year's normal ½% merit increase and to wipe out the previous seven years of ½% merit increases. One bad year moved her salary's path back to what it was eight years ago.

For the rest of his career and on into his pension, because of that one bad rating, Fred will receive 1% less than he would have.

For the rest of her career and on into her pension, because of that one bad rating, Mary will receive 4% less than she would have.

Is there anyone that thinks this inconsistent situation is in any way fair, or that a punishment of 4% for life is reasonable?

Now consider Joe, who like Fred and Mary has received consistent ratings for years, but has never had a bad year. One day he gets a new boss. At annual review time, the boss tells Joe that he can see that Joe is a good worker, but they haven't worked together long enough that he can with any honesty and confidence give the same excellent rating that Joe always gets. Instead he gives a slightly lower, but still above average rating. He tells Joe that if he lives up to his reputation, he'll get the usual excellent rating next year (which is what happens). Both Joe and the new boss agree that this is fair, as would any reasonable man. What they don't realize is that Joe's salary and pension have just been permanently reduced by 4%.

Note that it is very likely that Fred, Mary, Joe, and the supervisors that assigned the ratings have no idea that any of these financial effects even happened, much less why. At best they would have noticed that their annual raises were smaller this year, something they were already expecting to happen. And unlike me, they would all continue in ignorant bliss.

Two Serious Problems

  1. Salaries are affected by the specific year in which an employee is penalized.
  2. A one-time drop in performance from excellent to very good can inflict significant financial punishment.

There is no way that the effects on their salaries should have been so different for Fred and Mary, or even any different at all. Had either of these incidents happened in 2010 or 2011, when there was a cost of living freeze in effect, there would have been no reduction at all, simply a loss of merit increase.

Not receiving the usual merit increase when one gets an occasional bad review is perhaps to be expected, but also having the previous seven years of merit increases removed doesn't help employees to improve; it's far too punitive. Wouldn't a penalty of something like ½% (or less) be more appropriate?

And in many cases, should there even be a punitive effect at all? The purpose of the annual performance reviews and merit ratings are to ensure that employees know what they are supposed to be doing and how they can improve or at least stay on track. Would a reasonable man expect that a long-time employee be financially punished for getting a 4.25, which is better than what most of the other people in the department normally get? In Mary's case, it's very doubtful that her supervisor would have given her a lower rating knowing that it would remove eight years of previous merit raises, reducing her income by 4% for the rest of her life.

Fundamental problem

What makes the current merit system so fundamentally flawed has little to do with its design, which actually works quite well in achieving its purpose. The real problem that causes such inequities is the effect of inflation, which is both outside the system and unpredictable. The current year's inflation rate is an external factor and should be irrelevant when punishing an employee. The only way to correct the system is to handle inflation differently.

It's also not obvious that it is even appropriate to impose significant financial punishments on excellent employees that are occasionally only very good.


I could suggest various changes to alleviate the situation (e.g. rather than using this year's merit rating directly, the formula should instead use a running average of the previous five years of ratings in order to even out the effects of the variability of inflation). But they would simply be treating the symptoms, not the causes.

Instead I have three suggestions that can eliminate the actual causes of the sometimes serious consequences of the current system. With a little fine tuning of the merit formula, these changes will be revenue-neutral for the University.

Remove inflation from the equation

The unpredictable effects of externally imposed inflation must be eliminated. The strongest suggestion I can make is that the cost of living allowance should be handled for staff just as it is for professors. All employees should be given the full cost of living allowance each year, before calculating their merit rating. That will totally remove the influence of this uncontrollable external variable from the system.

But doing so would also remove the administration's ability to reduce salaries for consistently poorly performing employees, so it is hardly an acceptable solution by itself.

Allow negative pay adjustments

Once separated from cost of living increases, merit increases become far more understandable and predictable. Negative annual pay adjustments should be allowed for when the cost of living allowance is small, though with a limit to how much, say -1%. For consistently bad ratings, employees will still see their salaries drop down to the level of their new target salary.

Give supervisors awareness and control of the effects of merit ratings

Very few supervisors actually understand the merit rating system and how it affects salaries. I suspect that those few that do understand it simply ignore the rules and simply give consistent ratings year after year in order to avoid its failings.

Supervisors should be required to explicitly indicate how a rating may affect an employee's salary. I would suggest that the rating value be accompanied by a check-box selection something like this:

If this rating potentially decreases the employee's merit raise, how serious an effect should be allowed?

  • Retain the usual merit raise.
  • Decrease the merit raise.
  • Cancel the merit raise.
  • Cause a negative merit raise.

This explicit selection not only allows supervisors to say Your performance was down a little this year, but I'm not going to financially punish you for it this time., it also ensures that supervisors are well aware of the effects of the ratings they give.

And lest anyone think that some of these choices are too harsh, note that under the current system it is the last and most punitive option that is already in effect.


Even if nothing else is taken from this report, please take Joe's case. It is a perfect example of where everyone did everything correctly, everyone thought it was reasonable, and Joe still ended up getting screwed for no reason other than that he had been assigned a new supervisor. Regardless of what changes are made to the system, it needs to handle this case, and it must do it naturally, without treating it as a special case.