Paul Oyer: What Online Dating Can Teach About Economics
When I set up my dating profile, I was upfront about my teenage children and my sweet but impish golden retriever. But I left out the YouTube videos my children introduced me to under “Things I Find Entertaining,” and in the section of OkCupid.com where they ask some very personal questions, I checked a few boxes that were not technically accurate.
I admit it. I left out details — and lied. What led me to be honest on some parts of my profile and not others?
We can find the answer in a branch of game theory known as cheap talk. A cheap talk framework considers the potential conflict between my preferences and those of the women I am trying to attract and lets us analyze, in a given situation, when and if it is sensible to hide information or lie outright. Since what was true and what I thought would appeal to people were often the same, I could quickly fill in most answers. But sometimes there was a conflict between an honest answer and what I thought would make me attractive. Game theory would say it all came down to utility: the degree to which I was forthcoming depended on what I thought the people were looking for, as well as the probable cost to me of lying about myself.
As much as we would all love to be loved for the people we are, things are more complicated. A woman and I can find each other attractive, but at the same time, she finds my favorite Internet video extremely unfunny, or feels she could never go out with anyone who checked a certain box regarding politics. If I revealed my video and views, that woman would never agree to meet me in the first place.
So I, like many others, hide these minutiae. I rationalize that even if a woman eventually finds these things out, she’ll accept these small negative traits as part of the whole package. And I justify my minor deception by arguing that I’m doing myself and the woman a favor. It’s cooperative game theory. Our interests are aligned, and I’ve simply removed some minor hurdles.
Research shows that minor lying is prevalent on dating sites, with a typical person claiming to be an inch taller, about 5 pounds lighter, and a year or two younger. OkCupid’s blog found that the average heights claimed by men on their profiles are suspiciously greater than that of American men, with an inexplicable number claiming to be exactly 6 feet tall. Similarly, they report that there are four times as many people making $100,000 per year as there should be. Finally, they found that people often posted dated pictures that likely made them look more attractive than they really were, reaching the very scientific finding that “hotter photos were much more likely to be outdated than normal ones.”
Unfortunately, profile inflators have a major impact on those of us who would like to tell the truth. Their lies lead all of us to discount claims as cheap talk. If everybody knows that many people who claim to be “Athletic and Toned” on Match.com are closer to the “A Few Extra Pounds” category, then to claim “A Few Extra Pounds” would mean one is actually significantly overweight. If you insist on always telling the truth, “profile inflation” will make everyone assume you are fatter, poorer and uglier than you are.
But I prominently displayed two features on my profile — my teenage children and the big, friendly dog that is not so familiar with the concepts of personal space and hygiene — that would be big turnoffs to many people. When it comes to children and dogs, my interests need to be perfectly aligned with those of potential partners. This is key to cheap talk models: The cooperative side suggests that the more genuinely aligned the interests of the provider and consumer of information, the more accurate the information will be. Lying — the noncooperative part of game theory — occurs far more often with baseline data we all share, like looks, income and age, where everyone wants to seem as attractive as possible. But not everybody has teenagers or a hairy canine sidekick, which are non-negotiable.
Companies’ Talk Is Cheap, Too
The logic that drives our online profiles also leads companies and their top managers to stretch the truth. One example was documented by Dartmouth economists Jonathan Zinman and Eric Zitzewitz, who found that ski resorts exaggerate their snowfall, especially during periods (generally weekends) when they have more to gain by doing so.
But just as Internet daters will exaggerate less if they think they will get caught, ski resorts tell the truth more when skiers can catch their lies. The proliferation of smartphones has made it possible to question snow reports in real time. One SkiReport.com user post in 2009, for example, read, “Jackson Hole/Teton Village DID NOT get 15 inches today. More like 0.” This immediate feedback had an effect; Zinman and Zitzewitz show that snow exaggeration shrank noticeably at a typical resort as iPhone reception reached that resort.
Corporate cheap talk is so common it extends all the way to top executives. Over the last couple of decades, there has been a rapid increase in firms’ use of stock and stock options to pay CEOs, justified by the need to align the interests of CEOs and their shareholders. But sophisticated shareholders are a lot like skeptical Internet daters. When the CEO makes a pronouncement about the prospects of the company, the stockholders know she might well be engaging in cheap talk and, as a result, they’re likely to discount the CEO’s statements. The CEO, knowing the market will discount what she says, really has no choice but to inflate expectations. Harvard economist Jeremy Stein analyzed the cheap talk of CEOs and how markets react, concluding that careful analysis “clearly exposes the fallacy inherent in a statement such as ‘since managers can’t systematically fool the market, they won’t bother trying.’”
Similarly, stock analysts have also been widely identified as potential providers of cheap talk. When a company goes public, analysts at investment banks evaluate the company’s prospects and make recommendations about the stock. By convention and SEC rules, the people who do these analyses are supposed to be isolated from people at the bank who handle the stock offering. However, the bank can gain overall when analysts inflate their estimates, because this makes the value of the bank’s other services (especially underwriting securities) greater.
Hsiou-wei Lin and Maureen McNichols studied in detail the recommendations of investment bank analysts at the time of new stock offerings. They compared the recommendations made by analysts at banks that underwrite a firm’s securities relative to recommendations made by analysts at independent banks. Lin and McNichols showed that independent analysts were considerably less generous with their forecasts than analysts whose bank had a relationship with a company. As we would expect, however, the market assumes this exaggeration. As a result, the stock market is less responsive to the recommendations made by an analyst whose bank has an underwriting relationship with the company he analyzes relative to those made by a truly independent analyst.
In addition to hanging on the words of analysts and CEOs, the stock market waits breathlessly for statements by one person in particular — the chairman of the Federal Reserve. Remember how Alan Greenspan could move the market seemingly with the mere raise of an eyebrow and how, during the financial crisis, Ben Bernanke’s every word was parsed for meaning? Statements by the Fed chairman have the potential to be cheap talk. The Fed can always say he or she plans to take certain actions regarding interest rates just to try to calm the markets, or that things look better than they do. But the Fed is often somewhat cagey about its intentions, providing ranges — rather than exact numbers — for certain financial targets. Jeremy Stein (who has subsequently become a Federal Reserve member himself) analyzed the Fed’s incentives to reveal information. He found that if the Fed announced a precise target, such as that the inflation rate should be 2 percent, there might be situations in which it made this announcement when its true goal was 4 percent inflation. But then markets would back out the Fed’s true intentions and undermine its goals.
There is less scope for manipulation when announcing a target range, such as 1 percent to 3 percent inflation. So cheap talk is more believable when a range is provided than when someone pins himself down with an exact figure. Perhaps, then, I should update my online profile to say that I am between 45 and 55 years old and between 5 feet 8 inches and 6 feet 2 inches tall.
Signaling You Really Mean It
So, how can you overcome cheap talk? An online dating site in Korea tried to find out. The site, essentially the Korean equivalent of Match.com, ran a special event: Over a five-day proposal period, participants browsed online profiles as in standard online dating, but could show only up to 10 people that they were interested in a date. In addition, some participants could offer a virtual rose along with two of their date requests. The rose, in effect, told the recipient that he or she was one of the person’s top choices. Next, there was a four-day period during which people responded (essentially yes or no) to the proposals they received. The company then matched up the mutually interested pairs. Participants paid the equivalent of about $50 to take part, which is a little less than the one-month fee for this particular dating site.
Why did the site add the element of the virtual rose, and did it affect the outcomes of the dating arrangements? The answers are that a couple of economists talked them into it, and, yes, it had large effects. The idea of signaling something to someone you are trying to impress was modeled by Michael Spence in the early 1970s (and won a Nobel Prize in 2001), and these economists wanted to try it out.
In the promotion, some daters received two “I really mean it” signals that were completely credible. But note that what makes the signal work in this case is that it costs something. Participants who use the virtual rose have to give up something very important — the ability to show special interest in others. That’s what elevates talk from cheap to credible. Signals become meaningful only if they are costly.
When Michael Spence originally explained signaling, online dating had not yet been invented and he had to think of another venue for his idea. He imagined a world where colleges exist only so that prospective employers can figure out whom they want to hire. In the model, there are exactly two types of people — those who are talented and those who are innately unskilled. Talented individuals cannot simply tell people “I’m talented” because such cheap talk proves nothing.
But suppose that only the talented people will be able to graduate from college. They may learn nothing useful, but they show employers that they are talented and, as a result, they are eligible for higher-level jobs. In this model, education has solved our cheap talk problem. A potential employee puts his money where his mouth is by spending a lot of time and money on his education to prove (rather than just say) that he is talented.
As the Korean dating site results showed, the signaling idea applies quite nicely to the virtual rose. If a man or woman sent a standard proposal, the recipient accepted about 15 percent of the time. But about 18 percent of proposals that came with a virtual rose were accepted, which means sending a rose increased the chances of acceptance by about one-fifth.
Looking at who accepted which proposals provides more evidence for the credibility of virtual rose proposals. Tracking height, earnings, education and other characteristics, the company that runs the dating site can determine which participants will be viewed as more or less desirable. The virtual roses do not matter that much for the most desirable people. That’s no great surprise — those people already expect to be among the most sought after.
But the effect of a virtual rose is largest on the middle desirability group. They are almost twice as likely to accept a proposal with a rose than one without. To them, being told in a credible manner that they really are particularly attractive is very meaningful. They have heard a lot of cheap talk in their lives, and they value someone backing it up. Or, put another way, the rose is a meaningful investment in the person because the sender had to give up other opportunities in order to send it.
The basic principles behind the Korean dating site’s virtual roses apply to other situations, too, such as early admission college applications and companies’ underpricing shares at an IPO to signal quality and make it easier to raise more cash in the future. These are effective ways to signal that you really mean what you say.
Adapted from Everything I Ever Needed to Know About Economics I Learned From Online Dating (Harvard Business Review Press). Copyright 2014 Paul Oyer. All rights reserved.