Regulating for rationality.

• Author: Schwartz, Alan
• Position: Symposium: Festschrift in Honor of Richard Craswell

INTRODUCTION I. RATIONALITY AND TRADITIONAL REGULATION II. RATIONALITY AND OBSERVATIONAL EQUIVALENCE A. Mismatch Costs and Observational Equivalence B. The Many Bias Problem C. Consumer Heterogeneity Generally D. Reference Dependence and Observational Equivalence E. Summary III. NEXT STEPS A. Evidence B. Defaulting to Rationality C. Disclosure and Social Learning CONCLUSION INTRODUCTION

The United States enacted a large amount of consumer protection regulation in the 1960s and 1970s. (1) The national and state legislatures then did little, apart from changes to consumer bankruptcy law, for decades. Recently, a new wave of consumer protection legislation has been passed or is being proposed, largely in consequence of market failures during the Great Recession. (2) The new laws add a regulatory premise.

In the traditional view, consumer markets fail in consequence of monopoly power or imperfect information. Because monopoly power is the province of the antitrust laws, consumer protection regulators focused on imperfect information. Their standard response was disclosure. The Truth in Lending Act (TILA) is a good example. (3) Prior to its passage, consumers had difficulty choosing among the interest rates that sellers or banks charged because these firms quoted rates in different ways, all of which were complex. TILA required firms to disclose the cost of money in a single number: the annual percentage (interest) rate. As a consequence, consumers could more easily compare credit costs across firms.

Two assumptions led Congress, in TILA, to regulate the form rather than the substance of credit transactions. First, there was no externality concern. A regulator necessarily has to regulate contract substance when a contract creates a negative externality. The problem that consumer markets appeared to pose, however, was poor consumer decisionmaking, not third-party effects. Second, Congress assumed consumers were able to make rational choices. That is, a consumer could compare the expected gain from knowing the interest rate a particular seller charged, and from knowing the distribution of interest rates in the relevant market, to the cost of becoming informed. As a consequence, the consumer would minimize her interest bill unless it was too costly for her to acquire the necessary information. Consumers thus would make poor decisions when and because it was inefficient for them to search. It followed that the regulatory task was to reduce the consumer's cost of learning about interest rates. TILA's rationality premise is plausible even today because the consumer's cognitive problem is simple: she has only to compare the numbers that firms quote.

Assuming rationality when the consumer must evaluate other contract terms is less plausible, however. Psychologists, and more recently some economists, have shown that consumers exhibit numerous reasoning errors in laboratory tests. Reasoning errors are attributed to "cognitive biases": laboratory subjects, that is, make mistakes because they violate rationality in numerous ways. Because traditional regulation presupposes rationality, the new social science learning suggests different types of reform.

Home mortgages are a good illustration. These mortgages contain many complex terms. Because consumers take out mortgages infrequently, and because terms change over time, consumers may rationally not incur the costs of learning what their mortgages say. A traditional regulatory response would be to require firms to simplify the language in which mortgages are cast. But suppose that lenders offer variable rate mortgages and some consumers irrationally (overoptimistically) believe that future housing prices will much exceed current prices. Such consumers may take out variable rate mortgages because they expect to satisfy the contractually required increase in the interest rate by refinancing their homes. If housing prices turn out to be flat or fall, these consumers cannot refinance and so may lose their homes. Simplifying mortgage language is an inadequate regulatory response if consumers will make poor decisions when they know what their contracts say. Regulating mortgage content may be better.

This Article argues that regulators should take a second look before making such strong interventions. (4) Its more precise claim is that "regulating for rationality" poses different challenges than regulating for costly information. Scholars and regulators now pay insufficient attention to these challenges. The argument here does not hold that regulators should return to the 1970s and simplemindedly assume that everyone is rational. Rather, the regulator today needs new types of evidence, and new default normative premises when evidence is lacking, in order to intervene effectively in markets in which some consumers are making cognitive mistakes while others are not.

To introduce the new challenges, consider a current view of the regulatory task. Put schematically, the view holds that markets either offer one contract that it would be irrational of consumers to accept, or offer two contracts, one of which attracts irrational consumers (contract A) and the other of which attracts rational consumers (contract B). Because contracts A and B are facially distinguishable, the regulator can choose effectively from a standard set of regulatory responses: "nudge" consumers toward rational contract B, alter or ban the irrational terms in contract A, or ban contract A altogether.

This conceptualization of the regulatory task leads to error when, as often happens, there are no irrational contracts to regulate. Rather, in these cases there are contracts that are good deals for some and bad deals for others. A second look at the variable rate mortgage contract will show why. This mortgage may appeal to rational consumers who know what their probable future incomes are likely to be. A consumer in a steady job with plausible promotion prospects can hold a well-grounded belief that she will be able to pay a higher contract interest rate out of her income in the next period, if she cannot refinance. This consumer thus could rationally choose the variable rate mortgage because it enables her to buy a better home than her current income can support. But as shown, the variable rate mortgage also may appeal to consumers whose income prospects cannot support higher interest rates but who are irrationally optimistic about future housing prices.

This illustration shows that rational and irrational consumer contracting choices often present to the regulator in an "observationally equivalent" way: that is, both consumer types may prefer the same contract. As another salient example, consider a credit card contract with a low introductory rate and high late fees. A consumer with good impulse control may rationally prefer this contract either because she correctly anticipates that she will make timely payments or because she correctly anticipates that she may pay penalties but would rather borrow on her credit card (the late fees are the interest rate) than borrow from a payday lender or a pawnshop. The same contract may also appeal to a consumer who is prey to a myopia bias and to a present bias. This consumer focuses on the low introductory rate (the myopia bias) and fails to anticipate that she may overspend her current income (the present bias), and so she may be surprised by the substantial late fees her creditors charge.

It may help to be a little more formal about contracts that are both rationally and irrationally preferred. The substantive terms of these contracts generate positive utility for typical consumers. Thus, many consumers benefit from the additional current liquidity that a variable rate mortgage makes possible, or from a low introductory credit card rate. Consumer contracts have one universal cost and one possible additional cost. The universal cost is the price. Consumers who agree to a contract derive more utility from the substantive terms than they lose from parting with the price.

This Article denotes the possible additional cost a "mismatch cost." To understand what a mismatch cost is, reconsider the examples above. A consumer may lose her home because she cannot refinance. She probably would have escaped this problem had she not been excessively optimistic about future housing prices; for then she would have made a different contract under which she would borrow less and so have a better chance of keeping her home. A consumer faces heavy late fees, which she may have escaped had she not been myopic and present biased; for then she likely would have rejected the credit card contract with the low introductory rate and high penalties in favor of a contract she could more easily sustain. A mismatch cost is incurred when a consumer makes a contract that her rational self would have rejected', the consumer and the contract are mismatched. (5) To sum up this concept in economic language, a contract is rationally and irrationally preferred when, for every consumer who makes the contract, the utility that the terms yield exceeds the price but, for some consumers, the utility the terms yield is less than the sum of the price and the expected mismatch cost. These contracts create expected utility gains for rational consumers but expected utility losses, on net, for irrational consumers. The two consumer types are observationally equivalent to the regulator, however, because both types accept the same contract, which creates expected utility gains for them, the possibility of mismatch costs aside. (6)

Cognitive error is a concern in consumer markets when it causes many consumers to incur mismatch costs. The new regulatory challenge thus is to get behind observational equivalence in order to identify and ameliorate these costs. It is this Article's central claim that the challenge is much harder to meet than is commonly realized.

There are three major difficulties, the first and most...