Symmetric Information Won’t Be Perfect

Shortly after our Cato Unbound article appeared, The New York Times ran two stories about the end of asymmetric information. The first discussed a new form of insurance now being offered in the United States:

Andrew Thomas’s life insurer knows exactly when he arrives at his local gym. The company is notified when he swipes his membership card, and 30 minutes later, it checks that he is still there, tracking his location through his smartphone.

The insurance company has a vested interest in keeping Mr. Thomas alive and well. In return for sharing his exercise habits, his cholesterol level and other medical information, Mr. Thomas, a 51-year-old medical publisher who lives in Johannesburg, earns points, which translate into premium savings and other perks.

That is yet another example of the evolution of private information into common knowledge, in this case knowledge about Mr. Thomas’s proclivities for exercise.

The second example was the killing of Walter Scott by a police officer. After a cell phone video of the shooting surfaced, the claims of the officer were quickly shown to be false. Not long ago this major injustice would probably have been a neglected police report. Today it is a national scandal likely leading to greater adoption of police body cameras.

None of the commentators either at Cato Unbound or on the web appear to dispute the thrust of our argument that information about buyers and sellers, workers and firms, principals and agents is becoming more easily available and more evenly divided. Much more, in fact. Some wish to dispute the conclusions we draw or the conclusions they think that we draw from this phenomenon, but our actual central message stands unscathed.

We first clarify a few confusions evident in some readers’ comments on our paper.

It is important to remember that the opposite of asymmetric information is symmetric information, not perfect information. That is a simple distinction, yet it’s one that many commentators, such as David Auerbach writing at Slate, fail to recognize. Information will always be imperfect. Uncertainty and risk will never be banished. Uncertainty and risk, however, do not in general create market failure (indeed in the case of insurance and gambling, uncertainty and risk create markets).

We don’t even have perfect information about our own tastes. In this setting, when one of us orders a product that does not meet our expectations, what do we do? Most of the time, we return it. Rather than living in a world dominated by moral hazard, we live in a world dominated by firms so eager to sell quality products that they will often guarantee our satisfaction or take the product back for any reason with full refund (sometimes less postage).

Of course, some products such as durables may not develop observable problems until well into their lifetimes and so are not easy to return. Yet it’s important to realize that in a competitive marketplace this is a problem for firms as much as for consumers. Our model of reduced information asymmetry, however, predicts that durables will increasingly be provided with greater guarantees and bundled service plans. In the past, firms did not offer such plans because they feared our moral hazard. A firm that can monitor the use of the durable, however, will willingly offer more guarantees and return plans. Rolls-Royce, the British manufacture of jet engines, for example knows so much about how and when its engines are being used that it makes most of its money not by selling engines but by renting them by the hour, along with a promise to maintain and replace any engine that breaks down. Similar rental contracts already exist for consumer durables such as automobiles, and we predict they will become more attractive as the renting firms can monitor the use of the vehicle with greater ease.

A number of commentators pointed to extensive price discrimination as contradicting our argument. It’s true that more information and search capabilities on the consumer side are likely to lead to more competition, but more information on the seller side can lead to more price discrimination, such as personalized pricing. The latter effect is just another implication of the decline of privacy, as we discussed. There is no necessary contradiction, however, between greater competition and more price discrimination. Indeed, when fixed costs are high and marginal costs are low, personalized pricing will tend to increase output, consumer surplus, and welfare (Varian 2005, 2010).

Joshua Gans and a number of others suggest that we are predicting or promoting the end of regulation. But that is a misreading. In fact, government regulation has sometimes helped to overcome problems of asymmetric information. Our example was the Truth in Mileage Act of 1986, which requires that sellers disclose and record the odometer reading on the car’s title at every transfer. We would support further efforts to make odometer readings collected at inspections easily available online by Vehicle Identification Number. As another example, we think that calorie counts and ingredient labeling have had benefits, albeit small ones. How about requiring that firms provide the same labeling information tied to a barcode or QR code that consumers could easily scan with a cell phone? That would allow shoppers to easily access information via algorithm to create personalized recommendations (eat this, not that). Weight Watchers already offers a phone app with similar capabilities, but regulation could make this information more easily and universally available at lower cost.

More generally, our belief is that government regulation which helps buyers and sellers to make better choices is superior to government regulation that prevents choices. As another example of a government regulation ripe for improvement in the information age, consider the Food and Drug Administration’s (FDA) regulation of pharmaceuticals. Rather than following the old paternalism model of banning products unless they pass a one-size-fits-none rule, it may be better to think of a Consumer Reports model for the FDA. Consumer Reports rates and recommends but it doesn’t ban. An FDA that focused on information provision would collect, analyze and make information more easily available but it would leave more treatment choices to be made by patients and their physicians.

We agree with Shirley Svorny, by the way, that the asymmetric information argument has been used far too often to protect rents, including the protection of physicians and dentists from greater competition from nurse practitioners and dental hygienists.

The government also has access to a wealth of information about education and health from administrative records on school children and Medicare recipients. Making this information more easily available—in suitably anonymized form—could similarly improve market efficiency, education and health. Chetty, Friedman, and Rockoff (2014), for example, use school district and tax records from more than one million children and their families to show that value added scores for teachers do a good job measuring teacher quality and that high quality teachers increase the earnings of their classrooms by hundreds of thousands of dollars in present value.

We pointed to reviews, ratings and recommendation systems as new sources of information to overcome moral hazard problems. It’s true that these systems are themselves subject to attack and manipulation and signal jamming. Unlike physics, however, there is no theorem in economics that says that every action creates an equal and opposite reaction. The advance of information symmetry is a case of daily progress averaging 3 steps forward and 1 step back. If reviews were so gamed as to be useless, for example, people would stop using them and services like Airbnb would shut down as markets reverted to the old equilibrium. Yet we don’t expect services such as those offered by Airbnb, Yelp, and Angie’s List to shut down anytime soon. In fact those companies and others providing similar information services have high market valuations.

Jeff Ely writes about genetic tests and how the information advantage may shift to insurers. Tabarrok (1994) long ago discussed potential problems for insurance with genetic information and suggested a solution, genetic insurance. Let’s assume this isn’t possible. Ely’s argument is that instead of asymmetry on the buyer side, the superior information will be held on the insurer side: “health insurance companies will have exclusive access to a wealth of data that predicts the future costs of an applicant with a given array of genetic markers.” That outcome is not obvious, as consumers will be able to access information about their own genes from services like 23andMe, again assuming the government doesn’t prevent such markets.  Furthermore it may be possible to provide basic information about costs based on genes in an intelligible form.

But still, let’s assume Ely’s premise is correct. If insurance buyers have greater information than sellers, markets may fail to exist, a negative outcome. But if, as in this hypothesized case, insurance sellers have more information than buyers, the result is not usually so grim. Indeed, contra Ely, with competition in insurance the insurers won’t even earn an information rent. Honda knows much more about the cost of building a car than we do, but they don’t earn a significant information rent from that knowledge. Even with some degree of monopoly, information rents on the sell side won’t cause markets to shut down. As long as prices are free to adjust, something close to actuarially fair insurance still should be available, albeit with segmented rates for different classes of buyers according to risk. There will be less pooling, and some welfare losses as a result, but note we discussed that shortcoming in our initial essay.

Ely also suggests, or seems to suggest, that the end of asymmetric information will mean the end of markets and thus should be rued (we are not quite sure how to read him here). We should keep in mind a distinction between asymmetric information and dispersed knowledge. Dispersed knowledge is the broader concept, and it includes knowledge of particular circumstances of time and place, such as tradeoffs in local production possibilities or knowledge of local demands and consumer preferences. As Hayek emphasized, the market does not require perfect knowledge to function, rather it is the means by which imperfect knowledge is made to function in the social interest. The case for the market in minimizing production costs, incentivizing the satisfaction of consumer demands and spurring innovation remains robust. But when it comes to dealing with the very specific problem of systematically asymmetric information, there we see the informational playing field leveling in many sectors. That blend of developments won’t render a market economy obsolete anytime soon, though it should change the proper scope of regulation.

Ask yourself the following question. If the amazing proliferation of information, monitoring, and quality evaluation in recent times does not render at least some regulation obsolete, or in need of revision, than what kind of information developments might do so? Once the question is phrased in such terms, it becomes obvious that a) information asymmetries are diminishing significantly, and b) the nature and extent of government regulation really does need to change.



Chetty, R., Friedman, J. N., & Rockoff, J. E. 2014. Measuring the Impacts of Teachers II: Teacher Value-Added and Student Outcomes in Adulthood. American Economic Review, 104(9): 2633–79.

Varian, H. R. 2004. Competition and market power. In H. R. Varian, J. Farrell, & C. Shapiro (Eds.), The Economics of Information Technology: An Introduction (pp. 1–47).

Varian, H. R. 2010. Computer Mediated Transactions. The American Economic Review, 100(2): 1–10. Retrieved from

Tabarrok, A. 1994. Genetic testing: An economic and contractarian analysis. Journal of Health Economics, 13(1): 75–91.


Also from this issue

Lead Essay

  • Tyler Cowen and Alex Tabarrok argue that the age of asymmetric information is coming to an end. In the traditional account, asymmetric information has some unfortunate effects: when the seller has much information and the buyer has little, only goods of poor quality generally appear on the market; they tend to be few in number, and they are bought mostly by rubes. Cowen and Tabarrok argue that this state of affairs, which may once have characterized the market for used cars, no longer obtains there – or in very many other places at all. Replacing it is a world of ubiquitous information, in which many regulations are no longer needed, but in which privacy concerns loom large.

Response Essays

  • Technology may or may not be eliminating asymmetric information. But either way, says Joshua Gans, regulation is here to stay. Whether it operates as a first line of resort or a last one, we ought not to do away with it entirely. From this perspective, technology should not so much abolish regulation as push it into the background and make less often invoked. Gans goes on to complicate the picture of technology delivering better information in the used car market: As cars become more reliable, they last longer, and they are more likely to go through multiple owners. The expansion of this market may have prompted technological developments in car-related information, rather than vice versa. Gans then turns to police body cameras, noting grimly that they only work when police themselves turn them on. He recommends pervasive citizen monitoring of authorities as the appropriate way to end this information asymmetry.

  • Shirley Svorny looks at the ways that patients, doctors, and hospitals are using information in new ways - and calling old regulations into question. Whereas state medical licensure may have served a valuable purpose in the past, it is much less clear that it maintains the same importance today. Better information technology as well as changes to liability, certification, and malpractice underwriting have combined to offer much better information to all parties in medical care. Innovations such as retail clinics, telemedicine, and increased roles for non-physician clinicians all stand to improve the quality of medical care while also reducing costs. But a big obstacle commonly remains: the state licensing laws themselves.

  • Jeff Ely takes a skeptical and Hayekian look at the idea that the age of asymmetric information is ending. He notes that improvements in information about health outcomes will not help solve the fundamental problems of risk sharing that the insurance market addresses. In Hayek’s way of thinking, markets exist above all because information is asymmetrical; if all players had all relevant information, market prices would be wholly unnecessary, and allocations would happen both voluntarily and without prices. That world is so utterly different from our own that it’s hard even to think about it for very long. It also shows no signs of materializing anytime soon.