« The Democrats Get A Moment | Main | Does The GOP Need Iraq? »

November 14, 2007

The Dangers of Performance Pay

James Surowiecki lays it out:    

a recent study of almost a thousand companies by the management professors W. Gerard Sanders and Donald Hambrick found that C.E.O.s whose compensation was made up mostly of stock options tended to “swing for the fences,” making investments and acquisitions that were riskier than those made by other executives. As a result, the performance of the companies run by the risk-takers was far more volatile, and not for the good of the companies: the risky strategies were more likely to end in a big failure than a big gain. Generous options grants may also encourage fraud; the business professors Jared Harris and Philip Bromiley, who have made a study of hundreds of firms forced to restate earnings after accounting irregularities, found that companies that paid out most of their compensation in stock options were far more likely to end up restating earnings. And, as with hedge funds, the perverse effects of performance pay are exacerbated by the fact that big bonuses are often based on short-term performance. Stanley O’Neal, who was recently forced to resign as the C.E.O. of Merrill Lynch, made eighty-four million dollars in 2005 and 2006, a figure based in part on the huge profits that Merrill booked as a result of its forays into the subprime market. This gets to larger issues with so-called "performance pay." 

Depending on what metrics you're using to evaluate performance (and thus dictate pay), you risk distorting a worker's incentives to approach their job in a balanced, prudent, way.  If a teacher's whole compensation is based on test scores, for instance, you'll not only have a lot of teaching to the test, you may simply see the test -- or at least past versions of it -- being taught.  That may make for comfortingly high test scores, but it may also lead to less actual learning, less critical thinking, less adaptive educations, etc.  If a CEO's pay is based on stock options, amping up the price of the stock -- even in the short-term -- becomes more attractive.  If a journalist were to be paid on a per word rate, you'd get a lot of block quoting and very few contractions.  And so on, and so forth.  This stuff isn't as easy as people would like it to be.

November 14, 2007 in Economics | Permalink

Comments

If a teacher's whole compensation is based on test scores, for instance, you'll not only have a lot of teaching to the test, you may simply see the test -- or at least past versions of it -- being taught.

This isn't a possibility. It's a reality, especially in Texas where Bush's people first implemented No Public School Left Standing.

I'll admit my evidence is anecdotal, but since it's my nieces and nephews getting treated like this, I'm a bit upset over it. Their moms are special-ed teachers, and they've started to test the special-ed kids and just fold those results into the rest of them.

But I'm sure there's no goal of making all public schools look like failures. I'm sure there's all sorts of really good reasons for kids who can't function in mainstream classrooms at all to be given the same tests as everyone else and their raw scores just dumped into the school average.

Posted by: Stephen | Nov 14, 2007 10:54:28 AM

That may make for comfortingly high test scores, but it may also lead to less actual learning, less critical thinking, less adaptive educations, etc.

That's true, of course, and it's a very important consideration.

That said, there isn't the same obvious divergence between short-term interests and long-term interests for teaching students that there is for stock options.

If the compensation packages of CEOs were based on where the stock is in 5 or 10 years, they'd be much more likely to do good things for the company.

Posted by: Elvis Elvisberg | Nov 14, 2007 11:02:35 AM

If teachers teach to the test, and that is bad (results in a lesser education in some way)? Then isn't the test bad. What should a teacher teach to? And how will it be evaluated?

As with professional compensation at any level, there needs to be a balanced approach.

Posted by: George | Nov 14, 2007 11:05:09 AM

One problem is that Congress, in reaction to ridiculous executive pay, decided to deny tax deductions for executive compensation over $1 million, but they added an exception for pay tied to performance. This had the effect not of putting any brakes on executive compensation, but on pushing it all to performance-based.

Posted by: Doh | Nov 14, 2007 11:05:44 AM

So why do they want to do the same to Doctors and Hospitals? To save money, instead of teaching to the test. Physicians and hospitals can pick what patients they want to inflate their numbers and dump complicated ones on someone else.

Posted by: jenga | Nov 14, 2007 11:27:43 AM

George,

I have thought a lot about what I learned at primary and high school and come to the conclusion that the most important thing was critical thinking and the ability to analyze and produce a coherent, well thought-out text. Yes, I did a lot of history tests, read many novels and learned my math, but while the dates and quotes and calculations have long disappeared, I am left with an ability--and even more important--a desire to look at the world around me and question and analyze what I see.

If I had been interested in science, I certainly would have had to study and retain a lot of facts. Then I might have become an engineer or a doctor. I was more interested in foreign languages; that is where I put in time and effort and I now work as a translator. But even in my chosen field, without critical thinking, I would not do my job properly. Translation is a lot more than replacing a series of words in Language A by another series in Language B.

This long introduction leads me to your question as to what's wrong with teaching to the test. This educational approach leaves no room for the type of analysis and critical thinking that helps us to become intelligent consumers, good parents and knowledgeable voters and citizens. I have nothing against learning "the facts" and memorizing dates, words and theorems but without critical thinking and debate, people turn into drones who, for instance, base their voting decision on the fact that they'd rather go out for a beer with George Bush than vote for (now Nobel prize winner) Al Gore.

Posted by: A Canadian Reader | Nov 14, 2007 11:34:44 AM

A. Canadian,

I do not disbute your educational gains or want to lessen their value or import, but how do we measure those gains - in some sense we test for them.

I certainly agree with too much focus on narrow testing. But that does not mean we can give up trying to determine how to measure effective teaching.

Posted by: George | Nov 14, 2007 12:07:54 PM

"If teachers teach to the test, and that is bad (results in a lesser education in some way)? Then isn't the test bad. What should a teacher teach to? And how will it be evaluated?"

No test can test everything a student should have learned in a class. A test is a measurement of proxy variables. We measure a few variables, and then we extrapolate that measurement to overall performance in the subject, which includes a number of non-measured variables.

In teaching to the test, teachers emphasize the skills and knowledge needed to raise performance on only the measured variables, and reduce or eliminate any teaching that relates only to unmeasured variables. Test scores go up, but the students' actual skill level in the subject as a whole stays the same or goes down.

For example: it's very hard to measure a student's ability to follow a written logical argument without also measuring the size of the student's vocabulary. Historically this is not a problem, because generally a large vocabulary corrolates with a high level of reading comprehension. Why? Because students obtain a good vocabulary by doing a wide range of reading and writing. So, if your reading comprehension test happens to require a large vocabulary, you nonetheless will generally get results that corrolate closely with reading comprehension.

But teachers can increase students' vocabulary, at least temporarily, by rote drill - especially if they have a list of the hard words that are likely to appear on the test. If they teach that list, the resulting scores can no longer serve as a reliable measure of reading comphresension. The testers will crow that scores in reading comprehension have gone up, although the students' ability to follow a logical argument hasn't changed at all.

Posted by: Bloix | Nov 14, 2007 12:08:54 PM

Ezra wrote: "If a CEO's pay is based on stock options, amping up the price of the stock -- even in the short-term -- becomes more attractive."

This isn't exactly what Suroweicki was saying. After all, what you are saying is also true of granting the CEO stock instead of stock options; if the stock price rises, the CEO benefits. The particular problem with options, which Suroweicki points out, is that there is no downside risk. That's why executives are more willing to take on riskier projects (i.e., swing for the fence) if they are being rewarded with options than if they were being rewarded with stocks.

Imagine you are the CEO of a company, and you have two investment opportunities before you. They both have the same net present value, but one is extremely risky and one is very safe. The CEO who is getting paid in options takes the risky investment because if it pays off, it pays off BIG, and if it doesn't pay off, comme ci comme ça. The CEO who gets paid with stock chooses the safer investment because his interests are perfectly aligned with the shareholders.

In both cases, their goal is to raise stock price. The difference is their downside exposure.

Now whether a CEO's interest should be perfectly aligned with shareholders is a different question. But the problem with options (and the 2 and 20 structure for hedge fund managers) is that these plans don't perfectly align their interests perfectly with the investors. They encourage riskier behaviors than most investors would want.

Posted by: RWB | Nov 14, 2007 12:58:30 PM

Regardless of the merits of either idea, I don't think that the idea of performance pay for teachers is similar to stock options for executives. The big problem of executive perfomance compensation in the form of options is that there are positive consequences for stock price increase but no negative consequences for stock price decrease. This creates a pro-volatility bias that isn't really analogous to anything that might come out of an educational performance pay system

Posted by: Julian Elson | Nov 14, 2007 1:03:07 PM

People need to learn that the parents should and in fact must decide what should be taught and if it was taught well. You can do this by having an excess of teachers and allowing parents to sign there children up for the schools and teachers that they want. Then you can pay the teachers on the number of students that they teach.

If the parents want the students taught un-scientific stuff like:

Organic food is better than regular food. The consensus among scientists on this is that organic food is no better than non-organic. The consensus is very strong and we have experience with it.

Nuclear power is dangerous. The consensus among nuclear scientists on this is that Nuclear power is safer than all but natural gas generated power. The consensus is very strong and we have years of experience with it.

Creation science. The consensus among biologists on this is that we came for evolution. The consensus is very strong.

There is no global warming. The consensus among climatologists is global warming is real and man is contributing to it.

Kindergarteners not taught that there is no Santa Claus. No one but 5 year olds believe in Santa Claus

That is how it must be. If you have a problem with having the schools support parents in un-scientific things you must convince the parents. It is too difficult to reward good teachers anyother way.

Posted by: Floccina | Nov 14, 2007 1:12:51 PM

I think teacher pay and CEO pay are related. It is the balance of the compensation package that matters. The problem is - to my mind - a squeaky wheel issue.

Student scores are low - so focus on test scores, the stock price is stagnant, get the stock price up.

It seems to me it is hard to design an effective merit based compensation package in both cases.

Posted by: George | Nov 14, 2007 1:58:40 PM

The key point isn't that "performance pay" is impossible or worthless, it's that it is, generically, costly to implement and monitor and not easily translated into "media friendly" league tables.

I personally am for "performance pay" when done right, but against when it's done wrong. The difficulty is that 99% of the groups proposing it for teachers don't want to spend the money/time/effort to do it right. Hence, the suspicion that they have a different agenda.

Of course, maybe I'm crediting them with too much intelligence and they just haven't studied any evidence about how performance pay really works...

Posted by: Meh | Nov 14, 2007 3:30:06 PM

"As a result, the performance of the companies run by the risk-takers was far more volatile, and not for the good of the companies: the risky strategies were more likely to end in a big failure than a big gain."

This shouldn't be a surprise to anyone who's ever glanced at the Black-Scholes formula for pricing options. The value of an option goes up with volatility. So the CEOs who receive compensation in options decide to stake the future of the company on the equivalent of the little ball going into Red #31 are behaving in the way their incentive packages reward them to.

Posted by: Sock Puppet of the Great Satan | Nov 14, 2007 6:10:09 PM

And when bloggers are paid (or see their reputations climb) by the hit, they tend to write stuff that's controversial and will drive the debate, rather than stuff they actually think is true.

Posted by: brooksfoe | Nov 15, 2007 4:59:32 AM

The consensus among scientists on this is that organic food is no better than non-organic.

Nutritionally maybe, but you're forgetting the negative externalities of conventional farming. What about the consequences of fertiliser use? Nitrate runoff? Algal blooms? Eutrophication? What about pesticide runoff? Damage to local ecologies by destruction of the insect populations?

Posted by: ajay | Nov 15, 2007 9:14:34 AM

One additional problem with CEO performance incentives -- a company's stock price isn't based solely on CEO performance. That is but one of many variables impacting stock price. The overall perception of the economy and the performance of the stock of similar companies probably dwarf most CEOs' impact on the stock price. But I suppose if you're a CEO it's good to take the credit for it in a time of booming equity markets.

Posted by: demisod | Nov 15, 2007 9:56:22 AM

Ajay wrote:

Nutritionally maybe, but you're forgetting the negative externalities of conventional farming. What about the consequences of fertiliser use? Nitrate runoff? Algal blooms? Eutrophication? What about pesticide runoff? Damage to local ecologies by destruction of the insect populations?

Wrong the consensus goes the other way on that too. Organ growing takes more land. Manure and rotenone are more dangerous than the non-organic alternatives.

Posted by: Floccina | Nov 15, 2007 10:52:13 AM

The comments to this entry are closed.