15 October 2007

2007 Nobel Prize in Economics ~ Mechanism Design, or Why Technocratic Dreams of Dispensing with Politics Don't Work

The prize was announced today in Stockholm and has been awarded to a troika of American economists who work in the arcane area of "mechanism design." Although it is arcane, it also is quite important, mostly because, unintentionally, it establishes the fundamental importance of politics. None of the actual interest in the work seems to have come out in the article in today's New York Times (or in the story on npr). It was not at all clear that the reporters quite understood what the work is about.

Here is the problem. Markets work well (read produce efficient outcomes) only when all participants are parametric. That means that no participant has undue influence over the terms of market exchange. Unfortunately, most of life is strategic, meaning that what is best for any given individual depends on what she expects relevant others to do. Others therefore have influence, whether intentionally or otherwise, over her decisions. So, markets are useful institutional tools only in a quite restricted range of social life. They are very useful in that restricted domain. But they may wreak havoc in domains of interaction where the basic assumption of parametric action does not hold.

I cannot offer the whole argument here*, but it runs roughly like this. In situations of strategic interdependence such as decisions on whether to provide a public good (where, due to strategic interdependencies, markets are not going to function well), a mechanism is an institution designed to mitigate the effects of that interdependence. So, mechanism design models will postulate an impartial mediator who can determine the actual cost of providing the good and elicit honest information from members of the group facing the decision about the level at which they individually value the public good. Such an arrangement is termed incentive compatible.

In some instances honesty then becomes a dominant strategy for all members of the group - except the mediator. This means all members of the group will be best off telling the truth regardless of what other members of the group do when the mediator asks them how much they value the public good. Unfortunately, since the total amount members of the group reveal to the mediator will not be budget balancing (there will always be a deficit or a surplus), she will have an incentive to overstate the cost of the good and pocket the surplus. This obviously is not efficient. Moreover, because the task of monitoring the mediator is itself a public good the group faces a regress.

In other instances honesty will be a (weaker) Nash equilibrium strategy for everyone, including the mediator. This means honesty will be the best policy for each individual given what she expects all relevant others to do. Unfortunately, as is typically the case in such circumstances, there will be multiple equilibrium outcomes and hence another decision problem since those outcomes will distribute benefits and costs to various individuals differentially. Hence again the group faces a regress.

So, in either instance there is a residual political problem - how to monitor the mediator or how to select one among the available, but distributionally non-equivalent equilibrium outcomes. Technocrats weep - there is not a market-mimicking way to elicit honest information about provision public goods that can eliminate politics. We thus face the task of making politics more attractive - hence the importance of democratic theory.**
___________
* So here are two citations for the curious. Neither paper is technical despite the fact that work in mechanism design enters mathematical hyperspace very quickly. Only one of the citations is directly self-promoting!

Gary Miller & Thomas Hammond. 1994."Why Politics is More Fundamental Than Economics"
Journal of Theoretical Politics Vol. 6, No. 1, pages 5-26.

Jack Knight & James Johnson. 1999. "Inquiry into Democracy: What Might a Pragmatist Make of Rational Choice Theories?"
American Journal of Political Science 43: 566-589.

** Since this will seem a leap, you might have a look at this second, self-promoting reference:

Jack Knight & James Johnson. 2007. "The Priority of Democracy: A Pragmatist Approach to Political-Economic Institutions and the Burden of Justification," American Political Science Review 101:47-61.

Labels:

6 Comments:

Anonymous trane said...

Thank you for a clear explanation.

Question: How's your book with Knight coming along?

Best,
trane

15 October, 2007 13:11  
Anonymous Heinz from Mögenburg said...

Wow, how did you reach this conclusion from the prize winners research? Isn´t it rather that their research tells us when it is in order to privatize, and when it is not?

16 October, 2007 07:53  
Anonymous heinz... said...

i.e.

http://www.marginalrevolution.com/marginalrevolution/2007/10/mechanism-des-1.html

16 October, 2007 11:41  
Blogger Jim Johnson said...

Heinz,

I reached this conclusion by reading the work as part of a borader project on institutional creation/design/reform. I won't repeat my view, since I state it in the post. But the basic issue is about how we might coordinate ongoing interacitons. We do so with a variety of institutions. We might have a preference of decentralized insititutions - markets, social norms, etc. - the mechanism design literature on my understanding asks whether we can have decentralized, market-mimicing arrangements that allow us to deal with public goods. The answer it provides is 'no.'

Markets do not work in public goods provision for the standard reasons. We rightly resist turning immediately to the state. So we try various decentralized alternative - Coasian bargainin, mechanism design, community a la Michael taylor. Each fails or work only under very restrictive conditions. Only then should we move to more centralized insititutions.

Since the conclusion is that we are driven to more centralized arrangements, it seems a no brainer that we prefer democratic accountability to not.

Jack & I spell this out in detail in our papers.

16 October, 2007 11:53  
Anonymous Heinz said...

Ok, I just might have misunderstood your post as referral of the work of the nobel prize winners. But it´s your own twist, if I get it right. Than everything´s ok.

16 October, 2007 12:19  
Blogger Jim Johnson said...

I just had occasion to re-read this post and my reply to a comment. My reply, especially, looks as though it were written while I was abusing some substance. So I am re-posting it here with corrected spelling, etc.
------

Heinz,

I reached this conclusion by reading the work as part of a broader project on institutional creation/design/reform. I won't repeat my view, since I state it in the post. But the basic issue is about how we might coordinate ongoing interactions. We do so with a variety of institutions. We might have a preference of decentralized institutions - markets, social norms, etc. - the mechanism design literature on my understanding asks whether we can have decentralized, market-mimicking arrangements that allow us to deal with public goods. The answer it provides is 'no.'

Markets do not work in public goods provision for the standard reasons. We rightly resist turning immediately to the state. So we try various decentralized alternatives - Coasean bargaining, mechanism design, community a la Michael Taylor. Each fails or work only under very restrictive conditions. Only then should we move to more centralized institutions.

Since the conclusion is that we are driven to more centralized arrangements, it seems a no brain-er that we prefer democratic accountability to not.

Jack & I spell this out in detail in our papers.

14 July, 2008 23:54  

Post a Comment

Links to this post:

Create a Link

<< Home