Competition's Role in Efficient Contracts: Markets & Consumer Cognitive Errors, Study Guides, Projects, Research of Economics

Download Competition's Role in Efficient Contracts: Markets & Consumer Cognitive Errors and more Study Guides, Projects, Research Economics in PDF only on Docsity! *Sterling Professor of Law and Professor of Management, Yale University. Ian Ayres, Daniel Markovits and Stephen Morse made helpful comments on a prior draft. 1e.g., Crawford (2002). 2Smith (1991). 3The psychological experiments commonly test for one bias at a time. Subjects may perform well in economics experiments because (i) they are free from bias, (ii) their biases offset so they act as if they are free from bias, or (iii) the market mechanism ameliorates the subjects’ errors. Besharov (2004) analyzes the effect of offsetting biases. This paper is interested in the third potential explanation, so it supposes that bias can influence consumer choice. 4Gode and Sunder (1997) and (1993). How Much Irrationality Does the Market Permit? Alan Schwartz* 1. Introduction 1.1 Psychology and Economics Persons make systematic errors in laboratory experiments when asked to perform tasks that require cognitive skill, such as deciding whether different representations of a policy choice describe the same problem or different problems. Persons act as if they are means/ends rational, however, when asked to perform tasks in laboratory experiments that test economic theories.1 The disjunction between the psychological experiments, in which persons exhibit irrationality, and the economics experiments, in which persons exhibit rationality, once led Vernon Smith to speculate that the market institution somehow enables persons to overcome their mistakes.2 This paper is an early entrant in a literature that pursues Smith’s speculation.3 Gode and Sunder began this lit erature with two interesting papers arguing that market competition and budget constraints can overcome individual irrationality.4 These scholars created an 2 experimental auction market. Their subject buyers could resell any experimental “goods” they purchased from the subject sellers to the experimenters at a preset price; the buyers thus had an incentive to minimize the prices they paid. The sellers had purchased the experimental goods from the experimenters at a lower preset price, and were then permitted to sell the goods in the experimental auction, keeping the difference between what they paid the experimenters and what the buyers bid. The sellers thus had an incentive to maximize the auction prices. These human subjects coordinated on the equilibrium that economic theory predicted would have been reached by profit maximizing traders facing the same costs and prices. The experimental results thus were consistent with auction theory, but Gode and Sunder were interested in whether the outcome was driven primarily by the subjects’ rational choices or the market institution. To answer this question, they had computers play the same game. A computer “seller” could not agree to a sale below the price the experimenters charged the seller for the goods (the seller’s budget constraint); nor could a computer “buyer” pay more for goods than the sum the experimenters would pay to repurchase them (the buyer’s budget constraint). Between the sellers’ costs – the floor – and the buyer’s limit – the ceiling – the computers generated random bids and asks. A sale was concluded when a bid matched an ask. Gode and Sunder referred to their computer parties as “zero intelligence traders” because they bid randomly, had no memory and ignored market rules. The zero intelligence traders reached the equilibrium that the human subjects reached, however. Gode and Sunder concluded that competition and budget constraints (i.e., one can’t sell below cost) can 7Stanovitch’s extensive review summarizes the studies (at 66): “... the direction of all of the correlations displayed in Tables 2.1 and 2.2 is consistent with the standard normative models used by psychologists when interpreting tasks in the reasoning and decision-making literature. Individuals giving the normative response in one task tended to give it on another – even when the task requirements were quite different. Also, in every single case, cognitive ability was positively associated with giving the normative response – individuals of higher intelligence were relatively more likely to give the normative response. This was equally true for tasks where the normative response is the subject of great controversy as it was for the relatively uncontroversial tasks ....” 5 make mistakes in one context tend to make them in others.7 The correlation in performance across experiments is plausible. Subjects who give the normative response in an experiment score higher in intelligence, as measured by SAT scores, than subjects who fail to give the normative response; and the former subjects also test higher on such personal traits as a tendency to intellectualize problems. Since intelligence and personality are relatively invariant to context, smart, intellectual people make fewer mistakes in general than other persons. Also, persons do not come to the experimenter wearing signs that identify themselves as analytic or intuitive. Rather, when a psychology investigator is interested in who is who, she tests subjects ex post. The psychology results suggest that markets possess two features relevant to the questions pursued here. First, market participants also should be heterogenous: there will be “System 2 persons”, who are difficult to fool, and “System 1 persons”, who are more easily misled. Second, in mass transactions firms – sellers and lenders – cannot conveniently identify who is who before the firm offers contracts to consumers. Firms thus face a complex problem. A firm would like to present a deal in such fashion as to cue the System 1 response if consumers exhibiting that response would pay higher prices or accept less favorable terms than would more analytical consumers. For example, persons are said to be overly 8A good review is Frederick, et al (2002). 6 optimistic about their future earnings prospects. Such persons may be willing to accept default terms, such as security, that their more rational selves would reject. On the other hand, a firm that offers contracts that these System 1 persons will take could lose the business of System 2 persons (they would be put off by the harsh default terms). Since firms compete for the marginal consumer, and do not know who that is, competition among firms for the sophisticated perhaps could cause them not to satisfy the preferences of the naive. It will be helpful, in pursuing this possibility, further to partition the consumers who make mistakes. Some of these consumers likely know their flaws while others do not. Consider, for example, a person who develops a financial plan that requires her to save a certain amount each period. Persons are said – this is becoming controversial – to discount the future hyperbolically rather than exponentially.8 A hyperbolic discounter has a higher discount rate between tomorrow and today than she has between six and five months from now. To be concrete, the person thinks she should save $200 each month for the next twelve months, but when March arrives an iPod looks great to her compared to banking the money so she under-saves in March. Thus, she is time inconsistent. In January, she would like to save $200 in March, but when March arrives she over consumes. If this illustrative consumer is self-aware, she will attempt to pre-commit to her financial plan. The Government sometimes facilitates pre-commitment. Thus, Federal law helps persons to pre-commit both by using tax subsidies to encourage the creation of IRAs, and by creating penalties for early withdrawal from IRA accounts. The penalties reduce the attractiveness to consumers of departing from 9A Coasian analysis would take a broader focus, to ask when cognitive error is best corrected within firms or on markets. Market analysis is so undeveloped that it seems productive at this stage to pursue a market inquiry separately. 10A search model analyzes competition for “search goods”, which are goods all of whose features the buyer can observe before purchase. Color thus is a search good while durability is an “experience good”. This paper uses a search model because it is interested in the contracts consumers sign, and a contract, at least in theory, can be read before the consumer commits to buy. 7 a life time savings plan. The IRA example raises the question whether competition can cause firms also to offer contracts that will help mistake prone but self aware consumers to overcome their errors. 1.3 This Paper The literature suggests that markets may respond to cognitive error in two ways: by causing firms to offer the contracts that sophisticated consumers prefer to all consumers; and to offer contracts to self aware error prone consumers that permit these consumers to avoid trouble. The question whether actual markets behave in either of these ways poses theoretical and empirical issues. Regarding theory, the analysis must be more concrete. How would competition work when some consumers are fully rational and others are not? Does competition work in the ways that theory suggests?9 This paper takes a theoretical tack. Part 2 draws from a recently published paper to argue that firms with market power will offer contracts that maximize the utility of persons who do not make errors and persons who do but are self aware. Persons who make errors but are insufficiently self aware are exploited; they pay supracompetitive prices for bad contracts. Part 3 next adds competition to the story by creating a search equilibrium model of competition among firms for rational and irrational consumers.10 Part 4 uses the model to argue that, under plausible conditions, competitive markets can reduce the prices naive consumers pay for bad contracts and sometimes will drive bad contracts out 10 In English, this consumer knows that later she will care more strongly about saving the current cost t and care less about realizing the future benefit b than she would had she been time consistent. The smaller is $ the more present oriented the consumer is. A naive consumer believes that she is more able to resist current temptation than she actually is. This overconfident consumer thus discounts the future benefit b at the rate $’ where $’ > $. More precisely, when she signs the contract, at t0, she thinks she will consume in period one whenever t # $’*b - p (she thinks she will go to the gym every other day). The firm faces a start up cost of F and a per usage constant marginal cost of c. It thus earns P - F when a consumer signs the contract and p - c if the consumer engages in consumption. The firm will offer consumers a menu of contracts that maximize the firm’s expected profits subject to the constraint that consumers do at least as well signing contracts as they would do going without the good or service. 2.2 Analysis The contract intended for the time consistent consumer sets the user fee p* equal to the firm’s marginal cost c of providing the good or service. Marginal cost pricing ensures that this consumer engages in consumption whenever that would generate positive surplus (whenever t #*b - c). The firm offers the surplus maximizing contract because, being a monopolist, it is able to recapture the surplus in advance through the down payment. The consumer who is time inconsistent ($ < 1) but who is aware of her weakness, would like to pre-commit to consume appropriately in period 1 (to use the health club as often as she should). The contract intended for this consumer sets the user fee p below the firm’s marginal cost (p* < c), making up for the resultant loss by raising the down payment P. The sophisticated, time inconsistent consumer 12This consumer would not later switch to a low marginal cost provider because the seller is a monopolist; there is no other provider. 11 will make the higher down payment because she knows that the increment to the down payment buys the pre-commitment that she prefers. The lower per use fee, that is, encourages the consumer to engage in the appropriate amount of period one consumption (she will use the gym more as her optimal life plan would dictate). The firm offers the pre-commitment contract because it earns the same (monopoly) profit selling to the self aware consumer that the firm earns when selling to the time consistent consumer. The naive consumer does poorly, however. The contract intended for her sets the lump sum fee at its highest level and also prices use below marginal cost. This consumer believes, at t0, that she will consume more than she actually will and so she is willing to make the high down payment in order to profit from the low usage fee on the many uses she (incorrectly) expects to make. Put another way, she overpays up front but will not recover the overpayment through appropriate consumption later on. A time consistent consumer will choose the optimal consumption path for goods and services whose consumption generates benefits in period one but costs in period two. A self aware time inconsistent consumer, who suffers from weakness of will, knows that she is likely to over consume in period one (max out on a credit card or eat too much unhealthy food ) relative to the consumption she would choose were she time consistent. This consumer now wants to pre-commit not to consume excessively. The contract intended for her requires a lower down payment than the time consistent consumer pays, but sets the user fee above marginal cost. The high user fee reduces period one consumption and thus satisfies the consumer’s desire to pre-commit not to over consume.12 Under the 12 same logic, the contract intended for the naive consumer, who underestimates how much she will consume later, also sets a user fee that is above marginal cost but is lower than the fee the self aware consumer pays. The naive consumer thus makes a down payment that is higher than it should be. This analysis deals with two kinds of error: overconfidence and hyperbolic discounting (i.e., weakness of will). The results, however, likely generalize to some other forms of cognitive error. For example, consumers who know they are prey to the availability heuristic may pay firms to provide probability data; self aware boundedly rational consumers may pay for simple contracts. To summarize, when consumers face a seller with market power, the seller will offer efficient contracts to the rational consumers and to the consumers who are aware that they may make cognitive mistakes. In contrast, the seller exploits the naive consumers in two dimensions: it extracts all of the surplus that a contract can generate, and the contracts themselves are inefficient. Can the market do better when firms are added to the selling side? 3. A Search Model 3.1 The Identification Problem and the Social Goal There is an initial question what an “irrational contract” is. To understand the question, let the market offer two categories of product: safe and unsafe. A decision maker who can distinguish between them maker should ban the unsafe product. Similarly, if one contract type is always preferred by irrational consumers and never chosen by rational consumers, the decision maker should proscribe the irrational choice. There is no point to asking whether the market would eliminate an irrational contract because its existence alone answers the question. The contracts that are this paper’s subject, in contrast, would be purchased in a free market by 14Firms also risk nonenforcement or liability if their contracts too obviously exploit. 15 ambiguous welfare effects if the goal is to optimize actual consumer preferences. This issue is best considered after an analysis of how competition could work. 3.2 A Search Model A firm can sell a contract that is intended for sophisticated consumers, denoted Xs, or a contract that is intended for naive consumers, denoted Xn. A contract is a set of terms that define a loan or other consumer purchase. The monopoly model described above thus analyzed three categories of contract, each intended for a different consumer type, but it will be convenient here to put sophisticated and mistaken but self aware consumers in the same category; the Xs contract is intended for these types. Firms have a fixed cost Fi (i = s or n), produce at a constant marginal cost of ci over some range [0, zi] and produce at an infinite marginal cost thereafter (zi thus is the firm’s capacity constraint). A firm’s average cost for offering a package is AC(Xi) = (Fi/q) + ci, where q is output. The competitive price for a contract, which has each firm pricing contracts at average cost and selling up to capacity, thus is pi* = ACi(Xi)(zi) = (Fi/zi) + ci. It is assumed equally costly to sell both contract types (cn = cs), but that it is more costly for a firm to set up an exploitative system than an efficient system. The motivation for this fixed cost assumption is that it apparently is more complex for a firm to find the standard form contract that best exploits the numerous, possibly partially offsetting biases to which consumers are prone than it is for a firm to choose the terms that a utility maximizing consumer would accept.14 Because Fn > Fs while all firms sell at the same marginal cost, the competitive price for the naive contract must exceed the competitive price for the sophisticated contract (i.e., pn* > ps*). There are Y total firms in the market, where Ys sell 16 the sophisticated contract Xs and Yn sell the naive contract Xn (Ys + Yn = Y, the total number of firms). Consumers are partitioned in two ways. First, some consumers make cognitive errors without realizing they do so – they are “naive” – and some do not make errors or are self aware – they are “sophisticated”. Using notation, there are Bs sophisticated consumers in the market and Bn naive consumers, where Bs + Bn = B, the total number of consumers. Persons also are partitioned according to their shopping behavior. Some consumers – B1 in number – visit n = 2 firms and then purchase the most attractive contract they see. A sample size of two is chosen for convenience. The B2 consumers – the nonshoppers – visit only one firm before purchasing. Let Prs be the probability that a sophisticated consumer shops, and Prn be the probability that a naive consumer shops. Then B1 = PrsBs + PrnBn, with B2 defined similarly. There seems no reason to suppose, however, that naive consumers are less interested in low prices than sophisticated consumers are. On this view, it is assumed here that Prs = Prn = P, the probability that any consumer engages in comparison shopping. For the reasons given in Part 1.2, firms cannot tell which consumers are naive and which not, nor can they tell which consumers shop or not. A consumer is said to prefer the contract Xi if she would choose Xi after seeing both contracts selling at their competitive prices. To understand the basis for this definition, begin by assuming that a sophisticated consumer will purchase the contract Xs at any price up to a common limit price of ls. Next recall that consumers will purchase their less preferred contract if the price of their preferred contract is too high. A sophisticated consumer thus also will purchase the naive contract Xn at any price up to a common limit price of lsn. Similarly, a naive consumer will purchase Xn at any price up to a common limit price lf ln, and will purchase Xs at any price up to a common limit of lns. These limit prices are referred to 17 as a consumer’s “willingness to pay” for a contract type. A sophisticated consumer who sees the contracts Xs and Xn priced competitively would purchase Xs if ls - ps* > lsn - pn*, where the left hand side of this inequality is the expected surplus the consumer would get from purchasing Xs and the right hand side is the expected surplus she would get from purchasing Xn. Since pn* > ps*, the sophisticated consumer would purchase Xs only if ls > lsn: if, that is, she has a greater willingness to pay for the sophisticated contract than for the naive contract. This is a plausible condition. Similarly, the naive consumer will buy Xn if she has a greater willingness to pay for the naive contract than for the sophisticated contract. Because the naive consumer is partly deluded, however, it cannot be said a priori whether her limit price for her preferred contract, ln,, exceeds the limit price of the sophisticated consumer for her preferred contract, ls. Part 4 considers both possibilities. A little more should be said about when consumers will purchase their less preferred contract. It is convenient to explain switching between contracts by focusing on the behavior of the naive consumer. She will purchase the sophisticated contract Xs if its price is low enough and she visits only firms that sell Xs, or if she sees both contracts but the price for Xn is too high. The cutoff price for Xn, above which the naive consumer would purchase Xs, is referred to as the “switching price”. To derive this price, recall that the discussion of consumer preferences just above showed that ln > lns. A naive consumer who purchases Xs at its competitive price of ps* earns an expected surplus of lns - ps*. She is assumed to reject Xn if she would earn at least as much surplus purchasing Xs at ps* than she would earn by purchasing Xn at the switching price. Letting this price be pn(a), the naive consumer will switch to Xs when , where the left hand side of this equation is the surplus the consumer wouldl p a l pn n ns s− = −( ) * 17It is assumed here that the willingness to pay of the sophisticated consumer for the naive contract is less than the willingness to pay of the naive consumer for the naive contract (i.e., lsn < ln). On this assumption, a firm that offers Xn at its limit price of ln could not sell to a sophisticated nonshopper. 20 4.1 Pricing Decisions It is helpful to begin by assuming that firms are selling both contract types at their competitive prices. Would a firm selling the naive contract Xn deviate from the competitive equilibrium? Two deviations are possible: in the price dimension, from pn* to a higher price; and in the “quality” dimension, from Xn to the sophisticated contract Xs. Beginning with price deviations, first consider a deviation to the limit price ln. The firm would then sell only to naive nonshoppers. A sophisticated nonshopper would reject Xn at its limit price.17 A naive shopper would either see the package Xn selling elsewhere at its competitive price pn* or the package Xs selling at its competitive price ps*. Since the latter is below the switching price of pn(a), neither naive nor sophisticated shoppers would purchase Xn at ln. The firm would not deviate from the competitive equilibrium if a deviation would earn it a non- positive profit. This equilibrium condition is expressed as ( ) (1 )l c Fn n P B Y n n− − ≤− 0 The first term on the left hand side of this expression is the surplus the firm would earn from the deviation (ln - cn); the second term is expected demand from the naive nonshoppers. The expression can be rewritten as ( ) (1 )1 − −≤ P B Y F l c n n n n A deviation to the highest price for the naive contract thus would not occur if (a) Many naive consumers 21 comparison shop (P is high); (b) Consumers have a low willingness to pay for the naive contract (ln is low); or (c) The fixed costs of setting up an exploitative system are high (Fn is big). The intuition underlying condition (a) is obvious. The intuition underlying condition (b) is that the lower is the willingness to pay of naive consumers the less the firm gains from charging them the limit price. Regarding condition (c), the greater the fixed costs of creating the naive contract, the more naive nonshoppers the firm needs to recover these costs. The presence of sophisticated consumers in the market also reduces the likelihood that firms will charge excessive prices for the naive contract. To see why, realize that when firms sell the sophisticated contract Xs at its competitive price, they restrict the ability of other firms to price Xn at its limit. This is because the limit price is less than the switching price of pn(a) so a naive shopper will either buy Xn if she visits a firm that prices it competitively, or she will buy Xs if her other draw is from a firm offering it. Firms could not sell Xs, however, unless there were sophisticated consumers to buy it. Hence, the presence of sophisticated consumers helps to protect naive consumers from being maximally exploited in the price dimension. The illustrative firm also could deviate to the lower switching price pn(a). The firm would not sell to sophisticated shoppers. This consumer’s two draws will reveal at least one firm selling the contract Xs priced competitively, or one firm selling the contract Xn priced competitively. Whether this firm would sell to sophisticated nonshoppers depends on whether the most such consumers would pay for the naive package (lsn) exceeds pn(a), the switching price for naive consumers. The firm would not sell to a naive shopper whose other visit was to a firm selling the naive contract at its competitive price. On the other hand, the deviant firm would sell to naive nonshoppers and to naive shoppers whose other visit was to a 18This possibility may be of theoretical interest only, as it is realized when few sophisticated consumers comparison shop and the sophisticated consumers have a high willingness to pay for the naive contract. 22 firm selling Xs. A firm selling Xn would earn non-positive profits from a deviation to the switching price if it did not sell to sophisticated nonshoppers and if ( ) ( )(1 ) ( )2 2 − −+ • ≤ P B Y n Y Y F p a c n n n n n P B The existence of a second term on the left hand side of Expression (2) apparently makes (2) harder to satisfy than (1): that is, a firm selling Xn at pn* is more likely to deviate to the switching price for Xn than to the limit price. This result is strengthened if the firm also would sell to sophisticated nonshoppers.18 On the other hand, if the naive consumers’ willingness to pay for Xs, lns, is relatively high, then pn(a) will be small, making the right hand side of Expression (2) large. Then (2) will become easy to satisfy; the firm likely will do better pricing Xn competitively than deviating to a higher price. The intuition for this result was introduced above: When the switching price for Xn is low, the gain to the firm from charging it – (pn(a) - cn) – is more likely to be below the cost – the loss of the naive shoppers who visit another firm selling Xn competitively. Thus, as consumer naivety falls, comparison shopping is more effective at producing competitive prices. This analysis of the seller’s pricing decision can be summarized in Proposition One: The presence of sophisticated consumers in a market, and the penchant of both sophisticated and naive consumers to comparison shop, both increase the likelihood that firms will price naive contracts competitively. 20A firm selling the naive contract at its competitive price would not deviate to selling the sophisticated contract at its competitive price because the firm realizes no surplus in any competitive equilibrium. 25 is harder to satisfy than Expression (1) (i.e., whether a firm is more likely to offer the sophisticated contract at its limit price than the naive contract at its limit price). The left hand side of (3) would be larger than the left hand side of (1) -- deviations to Xs would be more likely -- if there were more sophisticated consumers than naive consumers. Also, the right hand side of Expression (1) is 8n, a firm’s comparative advantage at selling the naive contract, and the right hand side of Expression (3) is 8s, a firm’s comparative advantage at selling the sophisticated contract. Hence, if 8s < 8n, the right hand side of (3) would be smaller than the right hand side of (1). In this event, a firm originally offering the naive contract would be more likely to deviate in both the price dimension and the quality dimension : the firm, that is, would rather sell the sophisticated contract at its limit price than the naive contract at its limit price. The marginal cost of selling both contract types is assumed to the same, but the fixed costs of selling the naive contract are higher than the fixed costs of selling the sophisticated contract. Thus, firms will have a comparative advantage at selling the sophisticated contract if sophisticated consumers have a higher willingness to pay for Xs than naive consumers have a willingness to pay for Xn (i.e., ls > ln); or if the two willingnesses to pay are sufficiently close. Next consider a deviation from Xn to Xs at the switching price of ps(a.)20 The analysis of this deviation is similar to the analysis for the naive switching price above. A firm that deviates to selling the sophisticated contract Xs at its switching price would sell to sophisticated nonshoppers, to sophisticated shoppers whose other draw was at a firm selling the naive contract Xn and to naive nonshoppers if their 26 willingness to pay for the sophisticated contract (lns) exceeded the sophisticated switching price. The firm thus is more likely to deviate to selling the sophisticated contract at its switching price than at its limit price. This analysis of contracting decisions is summarized in Proposition Two: Competition may drive naive contracts from the market. This outcome is more likely to occur if there are many sophisticated consumers, if both naive and sophisticated consumers comparison shop, if sophisticated consumers have a relatively high willingness to pay for contracts intended for them and if naive consumers have a relatively low willingness to pay for contracts intended for them. Remark 4: Firms selling Xn may deviate to Xs at its limit or switching price. Since firms in the original assumed equilibrium were pricing Xs competitively, the analysis here predicts the existence of price dispersion in the Xs market. Equilibria with price dispersion are common when search is costly. Remark 5: A number of biases, such as over confidence and the endowment effect, dissipate with experience. This suggests that naivety is less common or weaker in markets in which consumers buy or borrow frequently. The less naive a consumer is, the lower is her willingness to pay for the naive contract and the higher is her willingness to pay for the sophisticated contract. Common transactions thus are more likely to be conducted under sophisticated contracts. Remark 6: In the model, sophisticated consumers will buy contracts intended for the naive. A plausible alternate specification holds that sophisticated consumers never would purchase naive contracts because they would recognize the contracts’ exploitative nature. On this view, naive contracts are more likely to disappear than in the analysis here because firms offering naive contracts could not sell to sophisticated consumers at any price. A possibly more realistic assumption is that sophisticated consumers will purchase some naive contract types but not others. The rejected contracts may be the most 27 exploitative. Remark 7: Eliminating naive contracts, however, would be undesirable if actual consumer preferences are the normative benchmark. When firms sell the sophisticated contract at its limit price, no naive consumers would purchase; and when firms sell the sophisticated contract at its switching price, naive consumers would buy it only if their limit price for that contract exceeds the switching price (lns > ps(a)). Recalling that naive consumers get surplus from the contract intended for them (this is ln - pn*), competition that eliminates the naive contract creates welfare losses. 5. Conclusion 5.1 Summary and Normative Analysis Consumers may be partitioned in three ways: some consumers do not make cognitive errors; some consumers are error prone, but know they are; and some consumers are error prone but think they are not. Theory shows that firms with market power will offer contracts that are optimal for the first two consumer types but which exploit the third. Naive consumers are offered naive – that is, wrong – contracts, and pay monopoly prices for them. The first result carries over when market competition is introduced on the selling side: firms will offer optimal contracts to rational and to self aware consumers. The second result may change, however. When naive consumers are offered the wrong contracts, those contracts are less likely to be priced supracompetitively. Also, sellers sometimes do better offering good contracts to every consumer type. The intuition underlying these results now is easy to state: If many firms exist, if some consumers are sophisticated while others are naive, if firms cannot tell into which class a consumer falls, and if all consumer types will shop for low prices and preferred contracts, then competition among firms for the marginal consumer will drive prices down for all contract types and, if there are enough sophisticated consumers, will of itself, lead a decision maker to conclude that people are time inconsistent when they make real world choices. It would be better for the decision maker to obtain real world evidence. 30 and difficult questions as to which legal institution would be good at error reduction and how that institution should proceed. That markets may help suggests a different mode of response, which is to improve market performance. There are two well known tools: to facilitate comparison shopping by requiring common terms in consumer contracts to be cast in standard forms; and to require the language in those forms to be accessible to the average reader. A possible third tool is to require firms to provide consumers with bias reducing information. This paper briefly illustrates these possible policy responses by analyzing how the model applies to credit card contracts. Every extant credit card contract could be preferred by every consumer type, but not every consumer should prefer every contract. Some companies offer credit cards with high introductory charges, relatively low monthly interest rates and a variety of ancillary services. Other companies offer cards with no initial charges, low interest rates for an introductory period, high interest rates thereafter and few ancillary services. Letting income be a rough proxy for sophistication, sophisticated consumers could prefer the former card because they are likely to use the services (easier reservations at expensive restaurants and clubs, for example) and are relatively indifferent to the interest rate on an outstanding balance, which these consumers plausibly expect never to have. This credit card contract would be an Xs contract. A naive consumer could prefer the stripped down version with high later interest rates – the Xn contract. This contract is attractive to many consumers because the contract is costless to make (there are 24See Bar Gill (2004). 25Put more precisely, if C(L) is the total cost of a loan, then C(L) = r + tc where r is the interest rate and tc are transaction costs. Then C(L)card may be less then C(L)bank even if rcard > rbank. 31 no initial charges), relatively less affluent consumers may have little interest in the ancillary services, and consumers may be attracted to the feature that later borrowing will be quick and easy. To see why this contract may be wrong for a naive consumer, let a person develop a financial plan at time zero (t0). The plan requires her to save the fraction & of her income in each period and to spend the fraction (1 - &) in that period. This consumer, however, is prone to two cognitive errors: she discounts the future hyperbolically (she suffers from weakness of will) and she is overconfident. As a consequence of the former error, in period t3 she succumbs to temptation, and chooses to spend the fraction ' > & of her period three income. As a consequence of the latter error, she uses her credit card conveniently to borrow a sum she mistakenly believes she will promptly repay. Firms offering the Xn credit card contract specialize in selling to such naive types. This contract type is said to be unfair, however. The low initial rate lures the naive consumer into the arrangement and the later high rate exploits her.24 . A contract with these features, however, could be rationally preferred by consumers free from cognitive error, and also preferred by self aware error prone consumers. Nominal credit card interest rates are high relative to other credit sources, but the total cost of a credit card loan may be lower than the total cost of alternate financial sources because credit card borrowing has low transaction costs. The debtor gets the money without providing the lender with a credit history or income and employment data, or having an interview.25 Hence, sophisticated consumers who are not interested in the ancillary services associated with Xs may prefer to save the introductory fee and use the Xn card. Also, a self aware error 26See DellaVigna and Malmender (2004). 32 prone consumer may prefer this card because of the high later interest rate. This consumer wants to make future borrowing difficult because she knows she is likely otherwise to borrow too much. The high rate thus is a form of pre-commitment: the higher it is, the more difficult it is for the consumer to borrow in the course of deviating from an optimal financial plan.26 As just said, however, naive consumers perhaps should not prefer Xn: under it, they may make excessive purchases early in their consuming lives because they mistakenly overestimate their ability to pay credit card debt promptly. To summarize, sophisticated consumers likely prefer the Xs card, but some of these consumers also prefer borrowing under the Xn card and more of them would have this preference if the introductory charge for Xs was increased too much. Naive consumers likely prefer the Xn card but they could come to prefer the Xs card with its low monthly rate if the introductory charge fell sufficiently. Since all consumer types can prefer all card types, no type should be banned. This market would be performing poorly, however, if too many naive consumers borrowed under the Xn card relative to these consumers’ circumstances and if the prices for both contract types were too high. If good and bad contracts are partly person relative, there is a question how a decision maker would know when to intervene. The model suggests that problems may exist if particular contract types sold at widely disparate prices. Considerable price dispersion indicates that some consumers are paying supracompetitive prices. Also, considerable price dispersion suggests that too few consumers comparison shop. The less comparison shopping there is, the more likely it is that the naive contract is profitable to sell.