Monday, October 15, 2007

The 2007 Nobel Prize: Mechanism What?

The 2007 Nobel Prize in Economics went to Leonid Hurwicz, Roger Myerson, and Eric Maskin for “having laid the foundations of mechanism design theory.”

Mechanism design isn’t covered in the typical introductory economics class. The narrative of your first econ class usually goes something like this: “The ‘invisible hand’ of the free market is the most efficient way of answering the fundamental economic questions: what to produce, how to produce it, and who consumes it. Sometimes the market doesn’t work—for example, in the case of externalities or public goods.”

In short, a single mechanism—the “market”—is usually the topic of discussion for intro courses. But there are lots of other mechanisms for answering these fundamental questions. And unlike the market, which is a decentralized mechanism (meaning it is not run by a central authority), there are plenty of man-made institutions that are centralized mechanisms. One example of such a mechanism is an auction, which allocates goods according to bids. Another is a political election, which allocates political power according to the preferences of the electorate. Both auctions and elections have rules, and these rules determine the optimal behavior of bidders and politicians.

One of the biggest challenges of designing an economic mechanism is that people have private information about their own preferences. One of the most famous examples of a mechanism design problem is the provision of public goods. Suppose a small town is considering the establishment of a public park in the town square. Should it ask the citizens how much each of them would value the park, and ask them to contribute that amount? Clearly, each of the citizens would have an incentive to “free ride” on their neighbors by understating their own value of the public good—so as a mechanism, just asking for voluntary contributions leaves a lot to be desired.

We will be creating a news analysis assignment about mechanism design for professors who use Aplia in their classrooms. In the meantime, here are some discussion questions to get the ball rolling.

Discussion Questions

1. “Market failure” often occurs when dealing with things other than purely private goods—for example, public goods, common resources, or goods with externalities. One solution to market failure can be broadly categorized as “market solutions.” An example of such a market solution is the levying of a Pigovian tax, which keeps the basic mechanism of the market but alters the incentives of participants. Another solution to market failure would be to replace the market with another institution entirely. For example, the right to use a specific frequency of the wireless spectrum is allocated by the Federal Communications Commission using an auction mechanism. Can you think of other examples of market failure that we address by using centralized mechanisms? What are the advantages and disadvantages to centralized mechanisms as opposed to market solutions?

2. The term “asymmetric information” refers to cases in which parties hold private (or hidden) information about their preferences or costs. One of the core challenges of mechanism design is to encourage people to reveal their private information in a truthful and credible way. For example, it is easy to show that the optimal strategy for a bidder in a Vickrey auction like eBay is to bid one’s true value. Think of a situation in which asymmetric information causes problems. What kind of mechanism could you design to elicit truthful information from the participants in that situation?

3. A recent Washington Post article has provoked a fair amount of discussion about the effectiveness of torture in acquiring information from prisoners. The most heavily quoted passage of the article reads:

“We got more information out of a German general with a game of chess or Ping-Pong than they do today, with their torture,” said Henry Kolm, 90, an MIT physicist who had been assigned to play chess in Germany with Hitler’s deputy, Rudolf Hess.
What do you think the economic study of mechanism design would have to say about torture? Is it an effective method for eliciting private information? How would an economist interrogate a suspected terrorist?

Labels: , , , , ,


  • At 2:32 AM, October 23, 2007, Blogger Tomasz Ludwik Górski said…

    I must say I am disappointed with the quality of this entry.

    You mention market failure and immediately go wireless spectrum licensing by FCC. Where do you see this market failure in hypothetical spectrum market?

    Second, Ebay auction, as any Ebay user would be able to say, is NOT a Vickrey auction and it is NOT an optimal (weakly dominant) strategy to bid your value there.

    Third, what made you link mechanism design, such a wide and interesting topic, with torture?

    And last, you seem to imply that market as described in basic economic courses with its invisible hand is one of possible mechanisms observed in the real world and MD studies also other mechanisms. While many mechanisms are not similar to a standard market, I would rather say that they all are kinds of markets, even markets for votes. And more importantly the standard market model is just a model which in mechanism design is studied in more detail, where the invisible hand of the market becomes visible actions of the market agents, the sellers and buyers in decentralized or centralized markets.

  • At 8:19 AM, October 23, 2007, Blogger Chris Makler said…


    Thanks for writing in! Let me see if I can respond to each of your points in turn.

    First, the wireless spectrum is a classic example of a common resource: it is rival but not excludable. Once property rights are established over the wireless spectrum, then the market for spectrum does indeed function well, as I think is your point. But the initial allocation of spectrum rights demands some sort of mechanism.

    Second, why is eBay not a Vickrey (i.e., second-price, sealed-bid) auction? The highest bidder pays the second-highest price, right?

    Third, I link mechanism design with torture because torture is, indeed, a mechanism whose purpose is to elicit private information. Most of MD theory revolves around how mechanisms can be designed to do just that - albeit in a much more benign environment.

    Finally, this point about markets is a wording preference. If you want to call everything a market, that's fine - but then the word "market" loses its meaning. I use the word "market" to describe a decentralized system of voluntary exchange; I think that's the standard definition.

    Thanks again for writing in!


  • At 2:27 AM, October 24, 2007, Blogger Tomasz Ludwik Górski said…

    So property rights (or other type of excludability) for spectrum work fine, if only they are introduced, right? Then wouldn't you agree that the same applies to all other goods? Of course this is not easy to see now, when they are all protected by property rights we have, but if we go back to the point when they were not yes present or where just being introduced, the situation becomes very similar. As an example you could recall the home-staging of land period that occurred in the US.

    In short, what I want to say about spectrum is that the same rules of excludability could be introduced in spectrum as they were for apples, cars, land etc. Instead licensing was introduced because of certain issues (mainly Constitutional freedom of speech in case of the US I would say) - but it does not mean that we, the economists, should take it as given, but rather question it and look for better solutions. Here I see the fault of your original entry that enforced the view of spectrum usage as some special case where property rights would fail.

    For eBay, one could say it is a Vickrey auction if it is approached naively. But if one observes how the bidders bid on eBay, or if you read any economic papers on internet auctions, one immediately realises that something else is being played - people bid several times and often in the last minute. It turns out that although highest bidder pays the second bid, several other conditions render other strategies more profitable than bidding one's valuation. Here, I believe, it would be important to demonstrate mechanism design care for details such as auction ending rules (see eBay vs Amazon). I believe that this care for detail is the special value in MD which makes it quite different from most of the economics where we try to generalize and avoid details.

    Now I get the torture part, although still I would say it is rather misleading for somebody just introduced to MD but I appreciate the approach that you take at this Blog.

    Finally, wording preference is only a part of the issue when it comes to discussing markets. What I was trying to say is that the decentralized voluntary exchange can be generalized by employing the notion on invisible hand, but the same exchange can be analyzed in more detail when we try to look at the ways in which the decentralized trade takes place, how sellers meet buyers, set prices, are these prices observed by others etc. The same applies to centralized exchanges as well.

    What I understood from your post and that I believe to be wrong is that: (a) decentralized markets are described sufficiently enough by the invisible hand notion and (b) MD analyzes other mechanisms only.


Post a Comment

Links to this post:

Create a Link

<< Home