Roth/Shapley Nobel

Lynne Kiesling

I have little to add to today’s congratulations to Al Roth and Lloyd Shapley for this year’s Nobel; Peter Klein has a useful roundup of links to good commentary, with a namecheck to us, in his link to his discussion of market design back in 2007 (thanks!). Mike’s frequent posts on price gouging and ticket scalping as “repugnant” transactions are an interest shared with Roth, who is himself a prolific economics blogger.

This year’s prize rewards their work in what’s called market design or market engineering, and those names are sufficiently vague to be confusing. As Peter notes in his post and Alex Tabarrok illustrates helpfully, the Shapley-Gale deferred choice algorithm provides a rule by which parties can be matched that yields stable, mutually preferable outcomes. The Wikipedia entry for it also provides a piece of code for implementing the algorithm in a computer simulation. It’s an elegant piece of formal cooperative game theory, which Al Roth applied to situations requiring coordination but lacking prices (i.e., lacking organic markets). His work doing so has yielded substantial improvements in the processes of matching medical residents and medical schools, kidney donors and recipients, and other situations in which coordination means matching. In fact, his kidney swap design has meant many successful kidney transplants, and makes programs like the kidney swap program at Northwestern’s Feinberg School of Medicine possible. While I would argue in favor of kidney markets as a preferable way to coordinate and match (and it’s not clear from any of Roth’s writings that he shares that position), if you are in a situation where better matching can increase total welfare, then designs-rules-institutions such as these are an improvement on worse rules or institutions, such as random assignment or using a lottery.

And that’s an important context in which to think about, and expand upon, the work being rewarded with this prize. This sort of institutional design (because make no mistake, that’s what this is, because it’s designing a set of rules) pertains to situations in which, for whatever reason, organic markets and price signals have not emerged or have not been allowed to emerge. In some cases, such as kidneys, some of us decry the absence of markets while others classify such transactions as repugnant. In other cases, such as medical residency positions, strong social norms and perceptions of fair play preclude the use of cash transactions and prices to coordinate the match; it’s also not necessarily clear that a cash market for such positions would yield a superior match on compatibility grounds. So yes, it’s top-down institutional design that’s meant to address a deliberate set of matching problems. This is a good opportunity for comparative institutional analysis — compare the designed institution that’s been created and implemented top-down with a more decentralized, generalized set of market rules, and/or an actual market with prices.

Methodologically, Roth tests his institutional designs using experiments, a methodology that both Mike and I find quite congenial, and that is well-suited to comparative institutional analysis, although in experiments you do in some ways lose the organic and emergent nature of the bottom-up market process with prices.

Some of my Austrian economics friends are not impressed with this body of work, but I don’t share that opinion. Roth and Shapley are in the mechanism design strand of game theory, which means they take as given the decentralized nature of preferences and information. They are not Austrian, and their approach is certainly prone to Hayek’s (and Vernon Smith’s) “pretence of knowledge” critique that they are constructivist. That said, the Shapley-Gale algorithm and the way Roth has applied it to medical residency and other problems are explicit in recognizing that economic problems are problems of coordination in the presence of diffuse private knowledge. And if we’re being honest we have to recognize that in reality, most institutions/rules systems involve both organic evolution and conscious design.

About these ads

4 thoughts on “Roth/Shapley Nobel

  1. I hope to post something later this week, but in the meantime the interesting reader could follow up Lynne’s post by checking out related commentary after 2007′s mechanism design Nobel awards to Hurwicz, Maskin, and Myerson:

    Also related, from earlier in 2007, these posts on Hayek and market design:

    And earlier this year:

  2. Pingback: Recomendaciones « intelib

  3. Mark Thoma at CBS MoneyWatch:

    Mario Rizzo at ThinkMarkets:

    While I agree with Rizzo that perhaps there are not enough fundamentally important contributors in economics to justify an annual prize, I am left in the uncomfortable position of favoring Thoma’s reaction to this year’s award over that of Rizzo.

  4. Pingback: Matching and a static environment « Knowledge Problem

Comments are closed.