Mind Your Decisions

Outcome Obsession: The Problem of Mechanism Design, as Demonstrated in Baseball

Every Tuesday is a Game Theory article at Mind Your Decisions.

I discussed mechanism design recently and suggested how it can be applied to solve routine problems. I was thrilled to see that one reader intends to use mechanism design as a parenting tool. I hope the reader will design a good mechanism and succeed in its application. But it would be foolish for me to expect that all well-designed mechanisms succeed. Today, I take a step back and discuss why good mechanisms can fail.

I’ll describe the drawback in an area that mechanism design can be applied: in writing contracts.

Economically, a contract is a mechanism that specifies legal obligations and provides incentives for a worker to perform well for the company. Most companies would probably like compensation to be completely variable and tied to performance, but that would not sit well with most of us who like stable income. To balance company and worker interests, compensation is usually set up as combination of a guaranteed salary and a performance-based bonus. The bonus is the carrot that the company uses to extract effort from workers.

Do bonuses really change how employees work? I tend to think they do. A study at Cornell showed that giving a 1 percent raise increases worker productivity by 2 percent, but offering the same money in a performance bonus boosts productivity by almost 20 percent. The result suggests that businesses can improve company performance by putting more of employee compensation in a performance-based bonus.

Sports are businesses too, and in theory, contract design is important to getting players and managers to perform. The New York Yankees appeared to have taken a note from mechanism design theory in writing Joe Torre’s recent contract offer. Although Torre did win four World Series in five years, he recently has managed six years of teams that failed to win the World Series. Management was unsatisfied to pay him top dollar if he did not meet their goal of winning it all.

So it should come as no surprise that Yankee management offered him a contract with high performance-based bonuses. The Yankees devised a one-year contract with large payoffs for team success. Torre would get $5 million in guaranteed pay, plus $1 million for each level of the playoffs the team reached. This is a possible $8 million payoff if the Yankees won it all.

On the surface, the contract has some nice economic features. First, the guaranteed $5 million, while technically a $2.5 million pay cut from his previous contract, is nonetheless a complement to Torre’s managing skill. It is still a whopping $2 million larger than the next-highest paid manager, Lou Pinella. And second, the bonus incentives show the organization’s focus on winning it all. If Torre could do any thing to increase team performance, he probably would to get his bonus.

But what happened next was not a success. Torre quickly rejected the contract and there was little negotiation. And Torre admitted it was not about money. He made that clear last week as he accepted a position with the Dodgers for less money per year.

What went wrong with the Yankees offer?

It has to do with a bigger problem of mechanism design, as described by Paul Milgrom, a Stanford economics professor and an expert in auction theory:

Mechanism design theory is outcome-oriented. A central assumption of the theory is that people care only about the outcomes, not how they are achieved. In the real world, processes sometimes succeed or fail based on whether they are perceived as fair, simple, or open—attributes that are hard to evaluate in a formal model.

(passage from Putting Auction Theory to Work)

We do not play the mechanisms drawn up on paper; we make choices about our lives.

The contract might have led to the outcome where the Yankees won the World Series and Torre got lots of money. But it was how the contract was presented that was the problem.

Torre called the contract an “insult” (ESPN story):

I just felt the contract offer, the terms of the contract, were probably the thing I had the toughest time with — the one year for one thing, the incentives for another thing. I’ve been there 12 years and I didn’t think motivation was needed.

In other words, a one-year deal did not signal commitment, and the bonus was somewhat insulting in the face of his history of effort. The contract decision depended on more than the outcome it might have achieved.

What can you take away? Mechanism design is very useful as long as you do not offend participants during the process of determining outcomes. Unfortunately, things like contract negotiations and parenting inherently deal with very emotional participants. Even if the outcomes will be desired by everyone, much larger care is needed in presenting the mechanism so that people participate.

The flip side is that if participants primarily care about outcomes, then mechanism design will be successful and should be swiftly applied. One application I thought up is how you can design a mechanism to improve how you save for retirement.

I ask you try mechanism design in your own life and share your results, but while doing so, keep in mind how the Yankees failed at signing Torre.