Eric Maskin

Eric Maskin Honorary Degree Speech at Universidad Carlos III de Madrid

February 16, 2018

I feel both lucky and privileged to be receiving an honorary doctorate from the University Carlos III of Madrid. Thank you so much for this great honor.
I’d like to take this opportunity to tell you a little about a subject to which I devoted much of my career, mechanism design. Briefly, mechanism design is the engineering part of economic theory. Most of the time in economics we look at existing economic institutions and either try to predict what outcomes they will give rise to or look back on past outcomes and try to understand why they occurred. This is the so-called positive or predictive part of economic theory. But what I am personally most interested in - -and what mechanism design concentrates on - - is just the opposite: we start with the outcomes. That is, we first identify what outcomes or goals we want to achieve, and then work backwards and ask what institutions or mechanisms or procedures could be designed to achieve those goals. This is the normative or prescriptive part of economics. It’s a smaller part of economics than the positive part, but it’s the part that I love best.

This is all pretty abstract, so let’s get more concrete. Here’s an example that might be familiar to you from your own experience at home. Let’s imagine that there’s a parent – a mother – who has a cake that she wants to divide between her two children, Alice and Bob so that each child is happy with the piece that he or she has got. That means that Bob should think that he’s got at least half, and Alice should think she’s got at least half. If the mother achieves this goal, then we say she achieved a fair division.
Well, how does the mother do this? If she knows the kids see the cake the way she does, then there’s a simple solution. She just takes a knife and cuts the cake exactly equally in her view, and gives each child a piece. Because we’re assuming that the kids see the cake in the same way, they each realize that they have half, and that’s the end of the story.

But, what if Bob or Alice sees the cake differently from their mother? In fact, this is, inevitably, what’s going to happen in reality. The mother might think that she has divided it exactly equally, but Bob is likely to think that his piece is smaller than Alice’s. The problem is that the mother doesn’t have enough information to achieve a fair division on her own. She, in effect, doesn’t know which division is fair because she doesn’t know how the kids see the cake. So here is the mechanism design question: Is there an indirect way of proceeding that will generate a fair division, even though the mother herself doesn’t know what is fair? It turns out that this is a very old question.

In fact, essentially this question was discussed in the Bible by Lot and Abraham when they were trying to divide grazing land fairly between the two of them – it’s the same question of fair division. So the question goes back at least a couple of thousand years, and as usual the Bible doesn’t just pose the question, it provides an answer. Here is the Bible’s answer.
The mother should have one of the children (say, Bob) divide the cake in two, and have the other child (Alice) choose which piece she takes for herself. Very simple. Why does it work? It works so because it exploits Bob’s incentive to divide the cake equally. This is the crucial word in mechanism design: incentive. When Bob is cutting the cake, he has a strong reason to divide the cake so that in his eyes the pieces are equal. Why? Because if one of the pieces is bigger, he knows Alice is going to take that one. So he should try to cut the cake in such a way that, whichever piece Alice takes, he’s happy with the other one. So Bob will be happy. And Alice will be happy because she gets to choose the piece she prefers. Thus, the divide and choose procedure – the mechanism from the Bible – solves the mother’s problem.

Now, cutting a cake is a relatively simple problem, but it’s still rich enough to exhibit some of the key features of mechanism design. First, the problem arises because the mechanism designer (the mother, in our example) doesn’t have enough information to know which outcomes are best (which outcomes are fair, in this case), so she has to proceed through a mechanism. In effect, the mechanism (the divide and choose procedure, in this case) gets the participants – through their actions – to generate the information needed to identify the optimal outcome. But there’s a complication: the participants (Alice and Bob) couldn’t care less about the mother’s goal – she wants to achieve a fair division, but Alice and Bob just want more cake. So the mechanism has to take into account that their goals are not the same as the mother’s. To use the term from mechanism design, the mechanism must be incentive compatible, compatible with Alice’s and Bob’s objectives.

Cutting a cake is probably not the most important problem in the world. But mechanism design can be used to solve some very important problems indeed. One is climate change. This is a global mechanism design problem. Scientists are in agreement that not only are temperatures rising, but that humans are responsible for this by releasing greenhouse gases like carbon dioxide into the atmosphere. The obvious solution to stopping climate change is to reduce our emissions of these gases.
But there is a conflict between that global goal (which we can agree on) and the individual goals of the countries of the world. Countries are interested in making sure their economies thrive, and unfortunately reducing greenhouse emissions is economically costly: it requires developing new technologies, or shutting down old factories, or reducing output, all of which are expensive.

So, each country would like the other countries to reduce emissions, but it would prefer not to do so itself. How do we solve this conflict of goals? Through an international treaty in which countries pledge to reduce emissions if other countries do the same. But designing an international treaty on greenhouse gas emissions is a problem in mechanism design - - how can we design a treaty that both reduces emissions adequately and also gives countries the incentive to sign and adhere to the treaty. We don’t have the solution to that yet. But I am confident that we will. And when we do, mechanism design will play a big part in that solution. Thank you.