For a many years I have thought about offering a graduate level class in market design. Obligations to this, that and the other have prevented the thought becoming action. Thinking that now might be the time, I decided to set out a syllabus. As always, one first looks to copy another syllabus. There are many graduate courses on market design offered. They descend, either from Alvin Roth’s course or Paul Milgrom’s (there was a version taught by both Milgrom and Roth but I don’t know if it comes before or after their separate versions). So, market design is either matching or auction theory (if you subscribe to Hatfield and Milgrom, matching theory = auction theory! By the way, in a screech against the Matthew effect, I mention Eriksson and Karlander, Discrete Applied Mathematics (1992)). The matching theory variants appear to have reproduced without mutation hewing closely to interns, kidneys and school choice. The auction theory variants incorporate empirics and field experiments.
Is market design no more than matching, auctions and empirical IO? Shouldn’t all that has been scribbled about the regulation of markets also be called market design? However, discussions of disclosure rules, regulation of utilities and insurance providers and so forth are rarely to be found. When discussing the issue with a colleague he suggested that market design be relabeled regulatory micro-economics. A title while more accurate conjures images of dreary bureacrats laboring in musty offices. Market design, as a label, however, has a `Randian’ quality to it. One can imagine John Galt studying it. Apparatchiks study regulatory micro-economics.
Well, when in doubt, return to the source, Roth’s manifesto (The Economist as Engineer………..Econometrica, 2002). He writes:
……….design involves a responsibility for detail; this creates a need to deal with complications. Dealing with complications requires not only careful attention to the institutional details of a particular market, it also requires new tools, to supplement the traditional analytical toolbox of the theorist.
Which reminded me of Macaulay:
……he who has actually to build must bear in mind many things never noticed by D’Alembert and Euclid.
It seems, from Roth, the objective is to strike a balance between the perspective of the parachutist (i.e. theorist) and that of the truffle hunter (domain expert). I cannot yet see how to pull this off in a class. To illustrate, consider the literature on mechanisms for assigning students to schools. There have been a flurry of papers in the last decade on this subject that are `technically sweet’. Some of the findings have changed how students are assigned to schools. However, if one felt that the market for pre-college education was `broken’ is the method by which school districts assign students to schools, the first piece one would pick up to study? More narrowly, is there any evidence that how students are assigned to schools makes a difference to educational outcomes? Thus, were one to follow Roth’s injunction to pay attention to institutional details, one is forced to become an economist who specializes in education. Nothing wrong with that. But it means that a course on market design either focuses on one topic, say, the economics of education. Or, a vehicle to introduce theorists to issues in education, healthcare, labor, financial markets etc. Much like the European bus tours of old where one spent no more than a day in any capital city.
I still wrestle with my syllabus and welcome suggestions.
6 comments
October 2, 2011 at 3:51 pm
Anonymous
here is a paper to begin with:
http://onlinelibrary.wiley.com/doi/10.1111/j.0741-6261.2007.00121.x/abstract
October 9, 2011 at 8:24 pm
rvohra
Thank you for the reference.
Rakesh
October 6, 2011 at 9:34 am
Daniele
My perspective on this is the following. After writing it down I realize that this is conventional wisdom, but nevertheless let me post it.
Definitions first. A market is any institution where individuals enter into binding contracts (i.e. a game that produce enforceable outcomes based on the interaction among participants). Design refers to the study of how the rules of the game affect the outcome and, more important, of how to achieve given goals if there is freedom in designing the rules.
Hence, if I were an outsiders of economics (or if I wanted to stick to the literal meaning of the words), I would interpret market design as a very large field, encompassing, among others, works on competition policy, contract law and regulation of labour markets.
However, it is true that when we speak of market design the first contributions that come to mind are auction and matching models. I believe that this is so because we give prominence to the set of models that have been successfully applied to the design of markets from scratch, as opposed to models that have been used to improve a market (e.g. regulation and competition policy models).
In particular, matching models have been useful in studying and setting up markets where monetary exchanges are banned, while auction theory has been used to study markets with money.
Moreover, in an important unification effort Kelso-Crawford and Hatfield-Milgrom have showed us that (under complete information) both auction and matching models can be nested in a more general setting where utility is more or less transferable depending on the set of contracts that is available to the agents.
Hence, I believe it make sense to organize a market design course around this two big applications: (i) how to match people without money (interns to hospital, students to school, etc) and (ii) how to match buyers and sellers — in a way that certain goals are obtained (efficiency, revenue maximization and what not). One can then show that, within the same general model, there is a nice algorithm (call it ascending auction or deferred acceptance) that, roughly speaking, leads to core outcomes in many cases (e.g. to stable matching in (i) and to competitive equilibrium in (ii)). After this, one can introduce complete information and move into classical mechanism design.
Alternatively, rather than looking at mechanism design theory (which would be treated elsewhere in the graduate curriculum), on can look at some finer design examples where details matters, and also at some empirical evidence.
October 9, 2011 at 8:36 pm
rvohra
Dear Daniele:
I think you’ve made a good argument for narrowing the focus of a market design course to matching and auctions. But, always, a but. Note the assumption in the phrase: design of markets from scratch. That markets are designed in a vacuum I would dispute. There is usually an `alternative’ already in place and so part of the issue with design is a competition between an existing set of trading patterns and one to be introduced `ex-cathedra’. This issue is not fully addressed in the kind of course you have in mind (though stability in matching touches upon the issue).
Rakesh
October 12, 2011 at 12:41 am
Anonymous
Hi Professor Vohra,
Is there any chance you will be teaching a course on market design this year? I did an independent study on auction theory, but I am interested in learning more about the matching side of market design.
Cheers
October 15, 2011 at 6:06 pm
fotoğraf
thanks you very mach.thanks you very mach.