Sunday, May 04, 2014

Salaries and merit

People like to say that salaries should be based on merit. Employees should be rewarded by their ability and performance, and not by connections, or family, or race, or class, or luck.  

In a previous post, I argued that it's something we say, but don't really mean.  We don't really act as if we believe that "merit" is what counts.  In fact, outside of sports and school, we don't seem to want to measure merit at all.


The classic example is teachers.  We love to talk about how valuable teachers are and how important education is, and almost everyone has a story about a great teacher and what a difference he or she made to their life.  But, even so ... we don't make much effort to separate the better teachers from the worse ones.  

Sure, that's not necessarily easy to do.  Ratings are arbitrary, and subject to all kinds of manipulation and favoritism.  We can't let the students vote on merit, because they'll just reward the teachers who give the highest grades and the least homework.  How do you judge "good" and "bad", anyway?  Isn't it mostly subjective?

All of those issues are legitimate, but not insurmountable. The students themselves know who the best and worst teachers are, independent of the grades they give, right?  In my high school, we all pretty much agreed on who were the best.  A good analogy is MLB broadcasters.  Here's one poll of the best and worst.  You probably agree with it, for the most part, right?  (Except that Vin Scully should really be #1.)

There is a movement to pay teachers based on merit, but mostly from policy wonks.  It's not something that parents seem to care about, or students.  And even when it is, it's almost all about improving education. That's important, sure. But, what about from an ethical standpoint?  If we really believe in paying for merit, shouldn't the argument be that better teachers should be paid more on the principle of simple fairness?  Tellingly, the Wikipedia article on teacher merit pay doesn't mention that argument at all.


There are some situations where we *do* want to recognize merit.  It's obvious that someone who works 40 hours deserves --  "merits" -- twice as much pay as someone who works the same job for only 20 hours.  They do twice the work, so they deserve twice the pay.

But, what about when one employee can do twice the work in the same amount of time?  

In my field, software development, there's an adage that the difference between the best and worst programmer is 100 times the productivity: 10 times between best and average, and 10 times between average and worst.  Let's say it's really only 5 times instead of 10, and let's say we fire all the bad programmers.  That still leaves a 5x difference between best and worst.

Should a fast programmer make 500 percent of the salary of an average one?  If you believe in merit, then, you have to say yes.  Don’t you?  

You could even argue that the difference should be *more* than five times.  If you had to hire five slow employees to replace the one fast one, you'd need five times the desks, the heating bill, the parking lot, and lots of other fixed costs. Furthermore, critical situations often arise when something needs to be done very fast.  The single employee provides you that capability, but the five slow ones will just run into each other.  (Brooks's Law: "Adding manpower to a late software project makes it later.")

But no company could get away with that kind of salary differential.  There'd be a riot.  

Some of the average guys would feel underappreciated and quit.  Others would instead spend their time coming up with rationalizations that the fast programmers are producing lower-quality work, and perhaps even sabotaging them.  The managers would enviously rebel against their underling programmer making three times their salary (even though they'd have to grudgingly admit he's worth it).  The suits in head office would say, why is that geek with no social skills making more than me, when I have an MBA and a thousand dollar suit?  And, when the superstar programmer got older and slowed down, you'd have to find a way to cut his salary in half, without humiliating him.  

I’d argue that merit pay doesn't fail because merit is too difficult to measure.  It fails because merit CAN be measured, and we don't like the results.  


You're a teacher.  You've been doing it for a few years, and you're proud of your career choice.  The parents like you, and people show a lot of respect when they find out you're a teacher.  You have a satisfyingly high level of status in the community.

But, now, someone figures out a way to accurately evaluate all the teachers.  When they’re done, it turns out that you’re only 30th percentile.  Seven out of ten of your colleagues are better than you.  

Now, you look and feel like a bit of a failure.  Even if you kind of knew, before, that you weren't as wonderful a teacher as some of your colleagues, you didn't really have to face it. You could think, hey, I’m good in my own way.  I may not be able to hit home-run explanations like Barry Bonds, but I use time-tested approaches, so I don’t strike out much either!  

But now that "Teacher WAR" shows you’re barely above replacement, your rationalizations don't work any more.  Also, your pay gets cut.

Even the *good* teachers don’t necessarily benefit from evaluations.  When the world gets a reminder that some teachers are worse than others, everyone’s status takes a hit.  Then the mediocre teachers quit, better ones replace them, there's more competition, and suddenly you’re middle of the pack.  And what happens when you have a bad year? Whatever it is you’re doing right, you have to keep it up in order to keep your place in the pecking order.  In fact, you probably have to get substantially better.  It's lose/lose for almost everyone.  

Except: if they only rate the best.  Then, nobody else really gets hurt.  Sure, you didn't win, and that shows that you're not the very, very best, but that’s no real shame.  And, for all the world knows, you might still be the second best, or in the top 1%.  

I think that’s why, in many professions, you'll see awards given to the best in the field, while nobody else gets rated or mentioned.  (Google any job description followed by "awards," and, usually, you'll find something.)  The awards aren't just to benefit the winners ... they’re also there to raise the status of the profession in general.  When a local employee wins a national award for best mechanic, all mechanics gain a little bit of additional respect.  ("Geez, I thought that mechanics were just people too dumb to graduate high school, but this guy seems really smart!  Now that I think about it, my guy might be pretty good too.")

In the everyday world, the equivalent to an award is a promotion.  Instead of saying, "Bob, you fix cars only half as fast as Joe, so we're going to pay you only half as much," we say, "Bob, as you know, Joe is the very best mechanic here, so we're promoting him to senior technician."  You get to recognize the best without having to formally call out the worst.

Another benefit is we tend to compare ourselves to our peers, rather than people at different positions in the hierarchy.  So, after the promotion, Bob is less likely to become resentful, because he's less likely to compare himself to Joe.  Also, if Joe is doing a different job, Bob doesn't get reminded every day that Joe is twice as good.

But ... it's a promotion to a different job.  You're taking the best performers out of the environment where they've excelled, and moving them to something unproven.  In software in the government, they take the best programmers, and move them to something that’s very different -- like “project leader,” which is a middle-management job.  Sure, you need strong technical expertise to do that job, but almost any decent programmer can do it competently.  When you have a programmer who’s five times as good as average, why put him in a job that’s he’s less suited for?  It’s like telling Mike Trout, "You’re so good at baseball that we’re promoting you to bench coach."

And there are only so many promotions to go around.  We've all seen lots of low-paid workers who are excellent at what they do, and lots of their colleagues who aren't that good.  I’m pretty sure the range of their salaries is much narrower than the quality and quantity of their output -- Walmart cashiers, for instance, seem to all be within a 25% range

So if you're good, and you want to be rewarded on merit, you have to count on promotions.  But how many of the best can be promoted?  Walmart has 2.2 million associates, and promotes 170,000 of them annually.  That’s less than 8 percent.  That means you're going to have a lot excellent cashiers making only slightly more money than average cashiers.  

If we really cared about merit, we'd be outraged by that.


In Canada, we have "pay equity" legislation, designed to close the gap between men’s and women’s earnings, by reclassifying male-dominated and female-dominated jobs in such a way that their pay scales become more equal.  

The rationale is "equal pay for work of equal value."  As in, "a secretary is just as valuable as a carpenter and should be paid no less."

But ... we only seem to care that the pay in question becomes literally equal.  We demand that equals be paid equally, but we don’t really care that non-equals are paid non-equally.

If you really believe in equal pay for work of equal value, the faster programmer should be earning much more than the slower programmer, and the better teacher should be earning much more than the worse teacher, and the industrious janitor should be earning much more than the average janitor. But we don’t take to the street to say, "Jan is working twice as hard at the ice cream parlor as Marsha, but being paid the same!  Stop the exploitation!"

When we do accept pay differences, we don’t seem to care much about whether they're proportional to ability.  We may agree that doctors are more "meritorious" than, say, welders. But, how much more?  Should a doctor be paid twice what a welder makes?  Ten times?  Twenty times?  Nobody cares, really, do they?  

The rare times that we do care, it’s almost always on the side of making everyone more equal.  Much is made of the ratio between CEO and worker pay, how it's too high.  Why do we think CEOs make too much?  For egalitarian reasons that have nothing to do with merit.  I have no idea, honestly, what the ratio should be, how much more valuable a fast-food CEO is worth than the guy who flips the burgers.  Is it 100 times? 1000 times?  10,000 times?  If we did care about equal pay for equal value, it would matter, wouldn't it?

It certainly matters in sports.  We know Mike Trout is a bargain at 600 times the US minimum wage, but overpaid at 6000 times the minimum wage.  Why don’t we have the same argument for CEOs?  I think it's because we don’t care about merit for corporate bigwigs; we're just uncomfortable that they make millions. Maybe I'm wrong -- maybe we really do believe CEOs make more than they deserve on merit.  But, if that's the case, why does hardly anyone argue numbers about how much they're actually worth?

I stumbled onto this article that profiles Marie Sanders, a single mother who works for McDonald's.  After more than two years of full-time work at Mickey D's, she still earns only $7.75 an hour and has trouble making ends meet.  

There's nothing in the article about merit.  The article never asks: "Is Ms. Sanders worth more than $7.75?  Is she worth the $15 she implies she deserves?  How were her job evaluations?"  If the writer *did* ask that, there would be outrage.  "How dare you imply that she's not competent enough to deserve a living wage!  Don't big corporations owe their workers a decent enough paycheck to allow them the dignity of feeding their family?"

Actually, Ms. Sanders sounds intelligent and competent.  I bet she’s well above average among her peers.  If that's true, does that bolster her case for $15 an hour?  I think it does, and I think most everyone would agree.

But if her excellence is an argument for paying her more, that's the same thing as saying her co-workers' averageness (or even mediocrity) is an argument for paying them less.  If Marie should earn more than Andy, the reasonably competent fry cook, that means Andy should earn less than Marie. 

How much less?  Maybe even ... $7.75?  If not that low, then, what?  $10?  OK, say $10.  But, now, what if there's someone even worse than Andy, someone who's just marginally competent?  

I think we don't want to bite the "merit" bullet and face that some workers are worth a lot less than others.  We're uncomfortable if Marie earns the $15 she's worth, and Andy earns the $7 he's worth.  Instead, we're happier if Marie and Andy both make $11 -- even if Marie is a lot better at her job.  

We say we want salaries to be merit-based, but what we really seem to want is for salaries to be equal.  

Labels: ,


At Tuesday, May 06, 2014 2:12:00 PM, Anonymous Anonymous said...

I love your stuff - but here I think you are completely off-base.

First of all - your entire perspective is, what can I say? socialist? Canadian? I don't know.
You write -
People like to say that salaries should be based on merit. Employees should be rewarded by their ability and performance, and not by connections, or family, or race, or class, or luck.

What does the word "should" mean in this context? Is that some sort of ethical/moral judgement? If I'm an employer I want to hire who I want and pay them as little as possible. If I hire my nephew I might want to pay him more (or less). I might discriminate by "connections, or family, or race, or class, or luck". If someone doesn't like the atmosphere or the conditions they should leave. This is not slavery and I'm not running a charity. It's not a public institution and I'm bearing the risk.
As an employee, I try to strike the best deal that I can - regardless of what others are earning. Where does that leave 3rd parties? - out in the cold..

--I stumbled onto this article that profiles Marie Sanders, a single mother who works for McDonald's. After more than two years of full-time work at Mickey D's, she still earns only $7.75 an hour and has trouble making ends meet.
There's nothing in the article about merit. The article never asks: "Is Ms. Sanders worth more than $7.75? Is she worth the $15 she implies she deserves? How were her job evaluations?" If the writer *did* ask that, there would be outrage. "How dare you imply that she's not competent enough to deserve a living wage! Don't big corporations owe their workers a decent enough paycheck to allow them the dignity of feeding their family?--

"Outrage" - why? I'm sorry she has trouble making ends meet. But if the labor pool is so overflowing with these kind of skills -then she'd better find a new line of work. --Don't big corporations owe their workers a decent enough paycheck to allow them the dignity of feeding their family?"-- No. They owe what they've promised. It's not a charity.
If John Q. Public wants to help the working poor - and to help everybody for that matter - I strongly advocate for free (or minimal cost) national post-secondary education/training. Probably delivered in large part over the web. That would be a public policy that I could strongly endorse.

Secondly, you admit --
There are some situations where we *do* want to recognize merit. It's obvious that someone who works 40 hours deserves -- "merits" -- twice as much pay as someone who works the same job for only 20 hours. They do twice the work, so they deserve twice the pay.--

And I would say if I would get paid by the piece - then all things being equal, if you made a hundred widgets/hour while I made 50 widgets/hour you should hope to get paid twice as much. Hopefully, you would strike a deal that actually gives you a per widget salary.
But, since in most cases we can't measure output so precisely ( teachers, software developers, etc.,), it's simply not possible to contract based on output. Teacher A cannot say I produced 100 education units this year while teacher B only produced 50.
On the other hand - I would love to see metrics that were so precise. And as a member of John Q. Public I feel that I am the employer of public teachers. And I would have no hesitations at all at paying the going rate for each education unit delivered.

I just have to disagree from top to bottom. I would love to hear your response. And keep the articles coming. Thanks.

At Wednesday, May 07, 2014 1:28:00 PM, Blogger Phil Birnbaum said...

Hi, Anonymous ... thanks for the comments!

I'm trying to summarize what people generally argue ... it seems like you think those are my views. Not necessarily.

I think people DO believe salary should be based on merit. Suppose a company pays blacks less than whites, and someone objects. You ask why. They will probably say, "it's unfair to pay someone less just because he's black -- people should be paid by how good an employee they are."

Ask any employer if they pay by merit, and they will say yes. No employer will say, "no, we pay as little as we can get away with, and we deliberately pay good employees less so the mediocre employees won't get jealous." That's because people will complain about any explicit rationale that isn't merit-based.

At Thursday, May 08, 2014 11:47:00 AM, Anonymous Anonymous said...

Pay should be based on supply and demand?

At Thursday, May 08, 2014 1:27:00 PM, Anonymous PG said...

PB, I’m anonymous from the first comment. Thanks for the response.
I gotcha. I’ll just take the word “we” that appears several times in the opening paragraphs as a literary affectation – or just a friendly approach to people.. No problem.

And I guess I can’t relate to this topic because I just don’t see at all what “fairness” has to do with this sort of business contract willingly entered into by the two parties.

To support your thesis, I wonder if the “fairness” doctrine of merit pay, to be consistent, would require all vendors of similar items, (e.g., Vendor1 sells oranges and Vendor2 sell oranges of similar quality) to be paid the same amount for their products – under the theory that equal value requires equal pay .

On the other hand, I still don’t see your objection from professions that are more difficult to rate, e.g., teachers and software developers. It’s not sufficient to know that teacher1 is better than teacher2. You need not just an ordinal scale but a ratio scale. As I said before, there are no measurable deliverable education units that can be counted like widgets – are there ?

Also, regarding software developers: if there is such an obvious disparity between productivity – it makes no economic sense that the highly skilled and lesser skilled are paid so similarly. If that were the case, any enterprising employer would simply offer more and hire only the highly skilled. Then there would be a bidding war until the employee was better valued. I wonder if you can cite a more complete and recent study that suggests otherwise.

Anonymous2: yes, that is what I was suggesting.

At Tuesday, May 20, 2014 3:16:00 PM, Anonymous Anonymous said...

Enjoy your work, Phil. I generally agree that we don't like the results of merit based pay. The reality is that it would negatively affect a majority of people.

The only thing I wanted to say is that it can and does work in some industries. I worked for a high frequency trading firm where the top traders and developers are making close to 10x as those at the bottom. We did not see a single employee leave in 2 years. I felt no resentment when being below the mean number. It is nice to have something to strive for.

I wonder if this only works because the industry can be high paying/upside and would be skeptical of these differentials being a net benefit in other industries. My point is simply that it definitely can happen and at least appears to have been beneficial for the firm I am at.

At Wednesday, May 21, 2014 2:36:00 PM, Blogger Phil Birnbaum said...

Hi, Anonymous,

Sorry so late replying!

It also seems to work in sports and in commission sales. What all three have in common is that (a) it's easy to measure output, and (b) the *quality* of the output is not a major consideration. (One trader's dollar earned is the same quality as another's.)

Another thing those have in common is that they're almost all zero-sum -- one trader's gain is someone else's non gain. Selling the customer a Buick means a different salesman doesn't get him to buy a Chrysler instead (assuming he'd buy SOME car regardless). And, of course, sports.

It seems to me that the first feature is the important one, but I'm wondering if the second also matters. Hmmm ...


Post a Comment

<< Home