Merit Pay
24 May 2011

UW's merit pay system has many real and perceived problems. Many of these are a result of abuse or misuse caused by a lack of understanding by individual managers, while others are caused by policies at higher levels. The system also has one fundamental flaw that can seriously affect those that are nearing retirement.

Note that this is not intended as a grievance against any of my mangers, past or present.

Inconsistency

The most obvious problem with this system is the inconsistency with which it is applied. Two people holding the same position and USG value in different departments can get very different merit ratings even though their actual performances are similar.

This is generally a result of inconsistent policies between the various departments and faculties. One unit might decide that its employees deserve to be rewarded, while another unit might decide to save money by keeping the merit values low. These local policies are not how this system is supposed to work, but in practice such policies significantly affect how it does work.

Human Resources Guidelines

The requirements specified by the guidelines are very much out of line with reality. A manger new to the system, or one that believes in strictly following rules, will find it very difficult to assign even a value of 4 to their staff because of its strong requirements. And anything higher than that is nearly impossible to achieve.

Meanwhile, other managers recognize that most staff should be getting around 4.0, and that the campus average is yet another step higher, near 4.25, and so they ignore the guidelines and give much more realistic ratings.

New managers, especially those hired from outside the UW community are unaware of the difference between policy and practice. They follow guidelines given them by Human Resources or requests from Deans to keep expenses down, and as a result end up penalizing their staff.

Others, out of ignorance or laziness, will decide that the evaluations really aren't worth the time and effort and will arbitrarily give very large or very small merit ratings to their staff.

The curve

Over their careers at UW, employees can plot their salaries as a percentage of the job value (midpoint), effectively eliminating inflation from the picture. Typically this is a steep curve up to the midpoint, and then a more gentle curve slowly approaching the horizontal limit defined by their merit rating. An employee that gets consistent ratings year after year will have a very smooth and predictable curve, leveling off at a value that will eventually determine their pension.

Changes in merit ratings from one year to the next can cause variations in this curve. For employees still a long way from retirement, these fluctuations don't have much long-term effect. An occasional rating that is higher than their normal value is nice, but next year's raise will be a little smaller. Similarly an occasional lower rating isn't good, but next year's raise will be a little higher than it would have been. An employee with generally consistent ratings isn't affected much by minor changes to their rating every now and then, and their salaries will generally follow the same curve.

This can work well for everyone. An occasional good or bad year can result in a small reward or punishment, while employees are inspired to keep their usually consistent rating high for long-term benefits.

Although it is more natural to think of these variations from the curve as being vertical, in fact they are horizontal. As far as I know, the system doesn't have any memory of how employees got to their current salary, so a slight increase above the usual target's curve really is a shift to the right, and a slight decrease below that target is a shift to the left. Staff that are well below their targets are positioned on steep sections of the curve, where the curves for adjacent targets are quite close horizontally. That's why these early-career variations actually have a very small effect.

Even for employees nearing retirement though, a single year's rating above their normally expected value can be nice, but again the next year's will be a little smaller and it averages out again.

Unpredictability

But for employees nearing retirement, even one year's rating that is below their normally expected value can have a very serious impact on their salaries and on their pension. When the adjacent target curves are nearly horizontal, as they are for long-time employees, a slight vertical change can mean a very large horizontal change. A reduction of a few percent vertically ($) can mean a huge horizontal change (years). (Despite what I've heard more than several times, salaries do not rise to meet the target in only a few years. I've been permanent staff for nearly 30 years, and except for the one time my rating dropped below my actual salary I've never been closer than 4.5 percent of the target.) A single instance of a bad review can instantly eliminate many years of merit increases, increases that will take many many years again to recover. And that bad review isn't necessarily bad, it could mean one year of excellent rather than very excellent. Yet that one blip on the record can severely punish the employee.

This is a very major flaw with the current system.

Consider an employee with a long history of 4.5 ratings at a job with a midpoint value of $60,000. The salary will be approaching $69,000. Now imagine that next year a new manager arrives and follows the Human Resources guidelines. The employee might ask Do you consider my work to be below average compared with all the other UW staff? and the manager would reply Definitely not, but according to the guidelines I can't possibly give you anything higher than 4.0. I've discussed it with the Dean's office and they agree. They also feel that this faculty has inflated merit values and want to move them down.. The manager is being completely reasonable, but in practical terms, the employee is now in serious difficulty.

A 3% mid-point increase is scheduled for the next few years. This employee is earning near $69,000, but the system says that with a 4.0 merit rating his target salary is actually below that, $67,980 ($60,000 × 1.03 × 1.10). His raise for the year will be effectively zero. And next year, if the manager can't be convinced of how things really should work, the same thing will happen. And to make it worse, even if the manager does raise later ratings, it will take many many years to regain those lost 3% job-value raises.

Now if that employee is about to retire, those bad years are what are going to determine his pension. The rest of his career's high merit ratings will have been for nothing. His expected pension, which would have been based on about $73,000 (71,000 + 73,200 + 75,400) will now be based on perhaps $69,450 (69,000 + 69,350 + 70,000). This employee, who for 30 years had a reasonable expectation of what his eventual pension would be, now discovers that that expectation is more than 5% larger than what he will actually receive for the rest of his life (and it's many times worse than that when measured in terms of disposable income), and all because three years before retiring he was unfortunate enough to be assigned a manager that decided to follow the rules. (And even if that merit reduction had been legitimate (e.g. reduced productivity), should that one bad instance in a long career have such a tremendous effect on his pension? Such punishment far exceeds the crime.)

In fact, consider two employees with long almost identical careers with a history of almost identical merit ratings and salaries. One of them is in a department that traditionally gives their employees excellent ratings for the last few years before retirement in order to boost their pensions. The other gets hit by a new manager that significantly lowers his salary during his last few years. Even though both have contributed almost the same amount to their pension plans over the previous 30 years, one will receive tens or even hundreds of thousands of dollars more in pension than the other. Is there perhaps grounds for a lawsuit there?

A Real Example

This table represents my merit ratings relative to the one I normally received, and the resulting changes in my salary relative to the USG midpoint.

  Year Rating Change Difference from 2003
  1998 X 1.63  
  1999 X 0.66  
  2000 X 0.75  
  2001 X 0.65  
USG aligned with IST positions 2002 X -2.69  
  2003 X 0.87 0.00
new manager 2004 X-.25 0.49 0.49
  2005 X-.25 0.42 0.91
  2006 X 0.64 1.55
new manager 2007 X-.50 -0.02 1.53
  2008 X-.50 -1.71 -0.18
  2009 X-.50 -1.04 -1.22
  2010 X-.25 0.69 -0.53
  2011 X-.25 0.56 0.03

In practical terms, my current salary as a percentage of midpoint is where it was in 2003, eight years ago. It is .65 less than what for years I viewed as my eventual pension target. That's a 6.5% life-time penalty for nothing that I had any control over, simply because I happened to be assigned a new manager. Who knows how many more percent I've been missing because I happen not to work for a unit that routinely gives much higher ratings than what I normally used to get. And of course I'm also approaching retirement age, so it's too late to recover from this no matter what ratings I get in the next few years. The best I can hope for is that it doesn't go down again.

I've been hit by just about everything that can go wrong with this system (you don't want to know about the things that went wrong around the time of my hiring). Had I known ten years ago that the eventual pension that I had very reasonable expectations of receiving was an illusion, perhaps I would have looked for a different position. I was significantly deceived by this system, but do I have any recourse?