College football fans may strenuously argue the specifics of play, but there is little doubt the quality of championship bowls they watch is higher than in years past. Does that mean players now pass, run, and tackle with more skill, strength, and precision? Perhaps—but the biggest change in the quality of games is that, thanks to tweaks in the design of postseason matchups, teams at the highest championship level more often find themselves facing their true competitive counterparts.
It was not always so. Until 1992, as HBS economist Alvin E. Roth and colleagues Guillaume R. Fréchette and M. Utku Ünver explain in a paper, factors such as contracts, control of stadiums, and corporate sponsorship effectively tied the hands of the National Collegiate Athletic Association (NCAA) into a system of early matching of teams. Hard-charging football teams were paired off for postseason play up to several weeks before the regular season had even ended. The 1990–1991 postseason was particularly dire: Premature matching led, predictably, to games that left undecided the pressing question of which team was best.
Faced with these institutional constraints and real-time consequences, over the next six years the bowls and athletic conferences went through three major reorganizations in the apparently dysfunctional system of matching teams.
The upshot, in economist terms, is that the market has since become increasingly thick—larger numbers of teams are potentially matched after their final rankings are known. The resulting implications for college football are immense: "When the bowls have matched later, the quality of the teams matched to bowls has improved, the likelihood of a championship game has increased, and the total viewership of all the bowls in the late-matching consortia has increased," write Roth, Fréchette, and Ünver.
Their paper, "Unraveling Yields Inefficient Matchings: Evidence from Post-Season College Football Bowls," is available for download [pdf]. According to the authors, "The present paper provides, as far as we know, the first direct evidence and measurement of the inefficiency due to early transaction times in a naturally occurring market."
Lessons about the value of better matching go far beyond the football field, however. As Roth describes in this e-mail Q&A with HBS Working Knowledge, efficient matching improves opportunities for everyone from schoolchildren to attorneys—and even carries the potential to help sustain life itself.
Sarah Jane Gilbert: What led you to research football teams? Are you a sports fan?
Al Roth: I'm a matching fan.
Q: What is matching, and why should business executives take notice of research in this field?
A: Matching is economist-speak for how we get the things we choose in life that also choose us. You can't just choose a selective school or a job—you have to be admitted or hired—just as you can't simply choose your spouse, you also have to be chosen.
Matching is one of the big things that markets do. Markets don't just determine prices, they also determine who gets what. For commodities, the price does most of the work, but many markets, like labor markets, don't clear by price alone. You don't hire just anyone who is willing to work at a given wage, you interview applicants carefully and try to find, and woo, the best ones.
Q: You and your colleagues studied the postseason matchups of college bowl games and their impact on television viewership. What were your key findings?
A: Having a good match between teams is a big predictor of how many people will tune in to watch a particular bowl game. The best match is one between the teams ranked first and second in the coaches' or sportswriters' polls. If those two teams play each other, the winner can claim to be the national champion. But for many years, athletic conferences used to sign contracts with bowls that would prevent a national championship game from being played if the number 1 and 2 teams were in conferences committed to different bowls. While the current organization of bowl games leaves much to be desired, it does a much better job of producing good matches than the system that preceded it.
Q: Would you explain the concept of "unraveling"?
A: Unraveling is the process by which some markets start to clear earlier and earlier, before important information becomes available. In the case of football bowls, the NCAA used to try to prevent bowls from signing independent teams (i.e., those that weren't already committed to a bowl) too early. But the NCAA gave up trying to enforce this after 1990, when Notre Dame very publicly agreed to go to the Orange Bowl when there were still four games to play in the regular season.
Unfortunately for Notre Dame and the Orange Bowl, Notre Dame lost one of those games, and instead of what had looked like a game between the two top-ranked teams, the Orange Bowl that year had the number 1 team and the number 5 team, while the number 2 team ended up matched with the number 19 team in the Florida Citrus Bowl. So the quality of the bowl matches suffered a lot from being made too early.
It turns out that a number of very competitive labor markets have suffered from similar problems. In some recent years, law students who apply to clerk for federal appellate judges were hired almost two years before they would graduate and begin work. Doctors are recruited into some specialty training programs well over a year in advance. Athletes are often recruited into coaching contracts while still very young. This can interfere with making good matches, if the qualities that will determine a good match haven't been realized at the time the contract is made.
Often this unraveling comes about for strategic reasons. For example, the HBS recruiting office regularly has to deal with investment banks and others that would like to make offers that expire before other offers become available.
Q: What particular changes do you see in the design of matching?
A: For football bowls, the Bowl Championship Series helps to delay bowl matchups until the completion of all the games in the regular season, so that the top teams can more often be matched with each other in a championship game.
For doctors, the marketplace for new graduates has been organized into a centralized clearinghouse that makes efficient matches in students' last year of medical school. (I had the privilege of designing the current clearinghouse algorithm.) Medical marketplaces for subspecialties like gastroenterology are increasingly able to use similar clearinghouses. Some of the same matching technology is now also used for assigning children to New York City high schools and to Boston schools at all levels. And many professional organizations, and universities and business schools, try to help set the recruiting "rules of engagement" so as to help level the playing field in a way that will lead to good matches between job applicants and employers.
Q: What lesson can business leaders take away from your research on matching and unraveling?
A: Hiring top talent—finding the people who are the best match for your firm—is part of the competitive environment. You have to know your market and help shape it.
Q: What are you working on next?
A: I'm making a matching market for live donor kidneys for transplant, a kidney exchange. This is a market that gives market designers an opportunity to help save lives a little more directly than other kinds of work in economics.