Adverse Selection in the Dating 'Market'
By Spencer Hill
Fall 2006 Issue

Why did the economist reject the man who asked her out?

Because he was single.

Broadly speaking, adverse selection describes a situation in which the mere fact that a good is available suggests its undesirability – the classic “lemon” problem. Consider a new car’s resale value, which plummets as soon as the car leaves the lot. Just as a prospective buyer has to wonder why the previous owner wants to sell, those of us fortunate enough to find ourselves in a romantic pursuit might ask, "If this guy is so great, why isn’t he already taken?" A 35-year-old bachelor practically begs this question.

At first glance, economics paints a rather unattractive portrait of the romantic search process, and indeed, the colors bleed deeper.

The reality that the group of potential dates has been adversely selected may encourage some people to exit the market, and if the most desirable people are most inclined to exit, the adverse selection problem worsens, potentially collapsing the market entirely. “In the language of Sex and the City, all the good men are taken,” said Professor of Economics and Management Benjamin Polak.

Game theoretical economics suggests two methods with which to confront the adverse selection problem: signaling and screening. Market participants would prefer to reveal to the other side of the market the high quality of their offering – perhaps regardless of its true quality – and to interpret the signals received by other market participants. Unfortunately, one of the lessons of economic theory is the difficulty of signaling, and therefore of screening as well.

Truth be told, any action or statement early in the courtship process remains of dubious veracity. “There’s not much information in the guy opening the door or seeing the latest movie version of a Jane Austen novel,” Polak said.

With used cars as well as new relationships, there is no substitute for due diligence, but one unique aspect of the market for dates and relationships remains the dearth of available references. While information flow can often remedy problems arising from asymmetric information, the ideal people to question regarding the quality and desirability of a given romantic partner – his or her past partners – will most likely respond to such queries with incredulity if not contempt.

The advent of online and newspaper-based dating exacerbates the problem of adverse selection. In order to lure more desirable partners, embellishing one’s own desirability seems to represent a reasonable strategy, however ethically questionable. By the time the disappointing real-life introduction occurs, each partner has invested so much time into the internet courtship that he or she may be willing to accept someone who does not meet expectations. This in turn renders the strategy of embellishment even more attractive to the participants of such services.

And embellish they do.

In the January 2005 article “What makes you click: An Empirical Analysis of Online Dating,” University of Chicago Economics Professor Ali Hortacsu, University of Chicago Graduate School of Business Professor Günter Hitsch, and MIT Sloan School of Management Professor Dan Ariely analyze a data set compiled from a popular online dating site. According to their research, 69% of men and 73% of women reported better than average looks, while only 1% admitted to being below average.

Is the online dating world inhabited solely by movie stars and modern Casanovas? Considering that 40 million Americans logged on to online dating sites in 2003, according to ComScore Networks, it seems more likely that online dating services have been overrun by an ugly population of exaggerators.

According to Polak, the very existence of dating reflects the fact that extracting information remains difficult. As such, dating and courtship can themselves be considered partial solutions to the problem of stymied information flow. Surely superior to the arranged marriages of more draconian days, modern courtship arguably functions as a trial period, intended to reduce the risk inherent in a world of adverse selection and informational asymmetry.

But stepping back from the adverse selection problem to consider an agent-matching problem in a broader sense, economics can deliver a much more robust result. Perhaps not surprisingly, economists turn out to be better matchmakers than dates.

Suppose some equal number of men and women can rank the members of the opposite sex in terms of their romantic desirability. Is there a way to place each of them into stable couples such that no two people could benefit by abandoning their respective partners in order to be with each other? In a 1962 publication entitled “College Admissions and the Stability of Marriage,” former University of California at Berkeley Economics and Mathematics Professor David Gale and former Professor of Mathematics at University of California at Los Angeles Lloyd Shapley solve this problem, proving by construction the existence of at least one divorce-proof equilibrium.

Given a set of men and women with any complete and transitive pattern of preferences, the imaginatively titled Gale-Shapley Algorithm proceeds as follows: each man proposes to his favorite woman, who accepts if and only if she prefers him to her current partner, or lack thereof. Then, each remaining unmatched man proposes to his next most favorite woman, whether or not she has already been matched. Iterating this process results in a stable-matching equilibrium in which no man or woman can improve his or her situation.

The obvious shortcoming of this computerized cupid is that it applies the rather bold assumption that we know what we want. But without an intimate knowledge of every possible partner – no pun intended – problems relating to adverse selection and asymmetric information threaten to leave this algorithm, and the spouses it selects, with much to be desired.

Fortunately for the inventors, in certain analogous situations, the necessary assumption becomes considerably more realistic. Consider job placements. After a large number of candidates has interviewed at a large number of firms, it seems reasonable to assume that the rankings that each firm and candidate report more or less reflect their true preferences. Indeed, a version of the Gale-Shapley Algorithm is used to place medical students into their residency programs. While this game is not “strategy proof,” because misreporting one’s preferences can potentially secure a more desirable matching, Polak emphasizes the difficulty of such a Machiavellian scheme. In medical school as well as in relationships, honesty may indeed be the best policy.

But the above applications by no means preclude the Gale-Shapley Algorithm from assisting us in our amorous activities. The algorithm could certainly be applied to small groups of people who know each other well enough to participate openly in matchmaking games and to rank their preferences accurately. Even with much larger pools of hopeful singles, such as those on an online dating or social networking site, the algorithm could match individuals using an ordinal set of preferences declared by each participant after browsing all other members’ profiles.

In this case, however, not only are the rankings generated by each member likely to be inaccurate, but the adverse selection problem itself might worsen. The fact that participants whose profiles are of lower desirability will become paired with similarly undesirable dates will predispose more people to lie and to do so to a greater degree. Hardly a match made in heaven. Even so, Polak notes that a version of the algorithm could suggest multiple matches for each individual, perhaps calming the feelings of desperation aroused by the one-shot game, and in turn mitigating the adverse selection problem.

While Polak doubts that adverse selection poses much of a problem for undergraduates, the informational asymmetries inherent in the market for romance ought to inspire some degree of skepticism in us all. So the next time you find yourself doting on a potential suitor, whether online or off, be forewarned. You may be picking up another lemon.