## Oren Bar-Gill

Print publication date: 2012

Print ISBN-13: 9780199663361

Published to Oxford Scholarship Online: September 2012

DOI: 10.1093/acprof:oso/9780199663361.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2019. All Rights Reserved. An individual user may print out a PDF of a single chapter of a monograph in OSO for personal use.  Subscriber: null; date: 17 September 2019

# Cell Phones

Chapter:
(p.185) 4 Cell Phones
Source:
Seduction by Contract
Publisher:
Oxford University Press
DOI:10.1093/acprof:oso/9780199663361.003.0005

# Abstract and Keywords

This chapter focuses on the failures of the cellular service market. It shows how carriers design their contracts in response to the systemic mistakes and misperceptions of their customers. In doing so, they impose welfare costs on consumers, reducing the net benefit that consumers derive from wireless service. The chapter focuses on three design features common to most cellular service contracts: three-part tariffs; lock-in clauses; and sheer complexity.

# Introduction

The cellular service market is an economically significant market that has substantially enhanced consumer welfare. From 1990 to 2009, the U.S. market grew from 5 million subscribers to 291 million subscribers. At the time of this writing, 93 percent of Americans have a cell phone, and an increasing number of households have given up their landlines and rely entirely on wireless communications. Annual revenues of the four national carriers—AT&T, Verizon, Sprint, and T-Mobile—total over $180 billion. While acknowledging these successes and welfare benefits, the focus of this chapter is on the failures of this market. We’ll see how carriers design their contracts in response to the systemic mistakes and misperceptions of their customers. In doing so, they impose welfare costs on consumers, reducing the net benefit that consumers derive from wireless service. We’ll focus on three design features common to most cellular service contracts: • three-part tariffs; • lock-in clauses; and • sheer complexity. As you have no doubt noticed, a major theme of this book is that the interaction between consumer psychology and market forces results in contracts that feature complexity and deferred costs. Lock-in clauses and three-part tariffs together generate cost deferral. Lock-in clauses enable bundling of handsets and cellular service. This bundling allows carriers to offer free or subsidized phones—an upfront benefit—recouping costs at the back end through the price of cellular service. This cost deferral is motivated by consumers’ demand for short-term perks and their relative inattentiveness to long-term costs. The underestimation of long-term costs is amplified by the (p.186) three-part tariff, which responds to and exacerbates the effect of misperceptions that lead consumers to underestimate the cost of cellular service. Sheer complexity is the third of the three design features that contribute to market failure. ## A. Three Design Features The basic pricing scheme of the common cellular service contract is a three-part tariff comprising (1) a monthly charge, (2) a number of voice minutes that the monthly charge pays for, and (3) a per-minute price for minutes beyond the plan limit. The three-part tariff is a rational response by sophisticated carriers to consumers’ misperceptions about their cell phone usage. Consumers choose calling plans based on a forecast of future use patterns. The problem is that many consumers do not have a very good sense of these use patterns—some underestimate whereas others overestimate their future usage. The three-part tariff is advantageous to carriers because it exacerbates the effects of consumer misperception, leading consumers to underestimate the cost of cellular service. The overage-fee component of the three-part tariff targets the under-estimators. These consumers underestimate the probability of exceeding the plan limit and incurring an overage fee. As a result, they underestimate the total cost of the cellular service. The other components of the three-part tariff, the monthly charge and the fixed number of minutes that come with it, target the over-estimators. These consumers think that they will use most or all of their allotted minutes. They therefore expect to pay a per-minute price equal to the monthly charge divided by the number of allotted minutes. In fact, the over-estimators use far fewer minutes and end up paying a much higher per-minute price. In this way, then, over-estimators also underestimate the cost of cellular service. Carriers seem to be aware of these misperceptions. As a pricing manager a top U.S. cellular phone carrier explained, “people absolutely think they know how much they will use and it's pretty surprising how wrong they are.”1 The prevalence of consumer misperception can be empirically confirmed by using a unique dataset of subscriber-level monthly billing and usage information for 3,730 subscribers at a single wireless provider. These (p.187) data enable calculation of not only the total cost of wireless service under each consumer's chosen plan, but also the total amount that the consumer would have paid had he chosen other available plans. Thus, one can determine the plan that best fits actual cell phone usage. The data show that over 65 percent of consumers chose the wrong plan. Some chose plans with an insufficient number of allotted minutes, whereas others chose plans with an excessive number of allotted minutes. Subscribers exceeded their minute allowance 17 percent of the time by an average of 33 percent, suggesting underestimation of use. And, during the 81 percent of the time when the allowance was not exceeded, subscribers used only 47 percent of their minute allowance on average, suggesting overestimation. In addition to the three-part tariff pricing structure, most calling plans come with a free or substantially discounted phone and a long-term contract with an early termination fee (ETF) that effectively locks the consumer in for a substantial time period—typically two years. These lock-in clauses and the accompanying ETFs can also be explained as a market response to the imperfect rationality of consumers. Imperfectly rational consumers underestimate the cost of lock-in, since they underestimate the likelihood that switching providers will be beneficial down the road. Switching providers may be beneficial, for example, if current service is not as good as promised, monthly charges are higher than expected (due to the misperception of use levels discussed above), or another carrier is offering a better deal. The lock-in that is enforced by the ETF also facilitates the common practice of bundling phones and service. The long-term revenue stream that lock-in guarantees enables carriers to offer free or subsidized phones. Rational consumers, knowing that they will pay for this “free” phone in the long term, would not be enticed by a free-phone offer. Imperfectly rational consumers, on the other hand, discount the long-term cost and seek out “free” phone offers. Finally, the third design feature that contributes to the behavioral market failure is the sheer complexity of the cell phone contracts. Cellular service contracts are complex and multidimensional. Choosing among numerous contracts can be a daunting task. The three-part tariff itself is complex. Lock-in clauses and ETFs add further complexity. In addition, the true cost of a calling plan depends on numerous other features. For example, most plans offer unlimited night and weekend calling, but carriers offer different definitions of “night” and “weekend.” Also, consumers must choose between (p.188) unlimited in-network calling, unlimited calling to five numbers, unlimited Walkie-Talkie, rollover minutes, and more. Finally, different carriers offer different ranges of handsets, handset subsidies vary, and so on. Complexity is further increased when family plans are added to the mix, data services are added to voice services, prepaid plans are considered in addition to postpaid plans, and so forth. According to one industry estimate, the cellular service market boasts over 10 million plan and add-on combinations. This level of complexity can itself be viewed as a contractual design feature that responds to the imperfect rationality of consumers. Complexity allows providers to hide the true cost of the contract. Imperfectly rational consumers do not effectively aggregate the costs associated with the different options and prices in a cell phone contract. Inevitably, consumers will focus on a subset of salient features and prices, and ignore or underestimate the importance of the remaining non-salient features and prices. In response, providers will increase prices or reduce the quality of the non-salient features. This, in turn, will generate or free up resources for intensified competition on the salient features. Competition forces providers to make the salient features attractive and the salient prices low. This can be achieved by adding revenue-generating, non-salient features and prices. The result is an endogenously derived high level of complexity and multidimensionality. Interestingly, consumer learning can exacerbate the problem. When consumers learn the importance of a previously non-salient feature, carriers have a strong incentive to come up with a new one, further increasing the level of complexity. ## B. Rational-Choice Explanations? Before we can draw normative and prescriptive implications from these behavioral theories, we must consider whether the more traditional rational-choice model can explain the same design features. If the rational-choice model comes up short, then we have good reason to appeal to behavioral economics to assess the appropriate policy response. The leading rational-choice explanation for three-part tariffs views these tariffs as mechanisms for price discrimination or market screening among rational consumers with different ex ante demand characteristics. The price-discrimination argument rests on specific assumptions about the distribution of consumer types—assumptions that are not borne out in the cell (p.189) phone market. With the distribution of types that we actually observe, providers selling to rational consumers would not offer three-part tariffs. Lock-in clauses can arise when consumers are rational. This happens when sellers incur substantial per-consumer fixed costs and liquidity-constrained consumers cannot afford to pay upfront fees equal to these fixed costs. However, in the cell phone market, while fixed costs are high, they are also endogenous. Carriers invest up to$400 in acquiring each new customer, but much of these customer-acquisition costs are attributed to the free or subsidized phones that carriers offer. This raises a series of questions. Why do carriers offer free phones and lock-in contracts? Why not charge customers the full price of the phone to avoid the lock-in? How many consumers cannot afford to pay for a phone up-front? For how many of these liquidity-constrained consumers is the carrier the most efficient source of credit? The rational-choice model can explain the presence of lock-in clauses, but only in a subset of contracts.

The rational-choice explanation for complexity is straightforward: Consumers have heterogeneous preferences, and the complexity and multidimensionality of the cellular service offerings cater to these heterogeneous preferences. But while this heterogeneity likely explains some of the observed complexity in the cell phone market, it cannot fully account for the staggering level of complexity exhibited by the long menus of multidimensional contracts available to consumers. Even for the rational consumer, acquiring and comparing information on the range of complex products is a time-consuming and costly undertaking. At some point, the costs exceed the benefit of finding the perfect plan. Comparison-shopping is deterred, and the benefits of the variety and multidimensionality are left unrealized. It seems that in the cell phone market, the optimal level of complexity has been exceeded.

## C. Welfare Costs

The design of cellular service contracts is best explained as a rational response to the imperfect rationality of consumers. Consumer mistakes and providers’ responses to these mistakes hurt consumers and generate welfare costs. For example, consumers who misperceive their future use patterns choose the wrong three-part tariff; that is, they do not choose the plan that would minimize their total costs. Extrapolating from the sample of 3,730 subscribers described above, the total annual reduction in consumer surplus from the three-part tariff (p.190) structure exceeds $13.35 billion. Moreover, while the average annual harm per consumer,$47.68, is small, this average masks potentially important distributional implications. The $13.35 billion harm is not evenly divided among the 250 million U.S. cell phone owners. Many of these subscribers choose the right plan. Even among those who choose the wrong plan, there is substantial heterogeneity in the magnitude of their mistakes. Each year, 42.5 million consumers make mistakes that cost them at least 20 percent of their total yearly wireless bill, or$146 per consumer annually. The distribution of mistakes implies a potentially troubling form of regressive redistribution, since revenues from consumers who make mistakes keep prices low for consumers who do not make mistakes.

Other welfare costs are a consequence of lock-in. Lock-in prevents efficient switching and thus hurts consumers. Switching is efficient when a different carrier or plan provides a better fit for the consumer. One survey found that while 47 percent of subscribers would like to switch plans, only 3 percent do so. The rest are deterred by the ETFs. Lock-in can also slow the beneficial effects of consumer learning and prolong the costs of consumer mistakes, since even consumers who learn from experience cannot benefit from their new-found knowledge by immediately switching to another carrier's plan. (Insofar as carriers allow consumers to switch among their own monthly plans, consumers can benefit from learning.) In addition to these direct costs, lock-in may inhibit competition, adding a potentially large indirect welfare cost. Since lock-in may prevent a more efficient carrier from attracting consumers who are locked into a contract with a less efficient carrier, it can deter new carriers from entering the market.2

Complexity is another detriment to welfare. The high level of complexity of cell phone contracts can reduce welfare in two ways. First, consumers tend to make more mistakes in plan choice when the menus are complex, and these mistakes reduce consumer welfare. Second, complexity inhibits competition by discouraging comparison-shopping. By raising the cost of comparison-shopping, complex contracts reduce the likelihood that a consumer will find it beneficial to carefully consider all available options. Without the discipline that comparison-shopping enforces, cellular service (p.191) providers can behave like quasi-monopolists, raising prices and reducing consumer surplus.

## D. Market Solutions and Their Limits

Do these behavioral market failures result from imperfect competition in the cell phone market? The simple answer is “no.” In fact, enhanced competition would likely make the identified design features more pervasive and the resulting welfare costs higher. If consumers misperceive their future use levels, competition will force carriers to offer three-part tariffs. If consumers are myopic, competition will force carriers to offer free phones and cover the cost of the subsidy with lock-in contracts. Finally, if consumers ignore less salient price dimensions of complex, multidimensional contracts, competition will force carriers to shift costs to these less salient price dimensions. When demand for cellular service is driven by imperfect rationality, competitors must respond to this biased demand; otherwise, they will lose business and be forced out of the market. Accordingly, given consumers’ imperfect rationality, ensuring robust competition in the cellular service market would not in itself solve the problem.

But it is a mistake to take the level of imperfect rationality as given. As we have seen in previous chapters, competition, coupled with consumer learning, can reduce levels of bias and misperception and thus trigger a shift to more efficient contractual design. In fact, the cellular service market has exhibited numerous examples of such market correction in recent years and now boasts a large set of products and contracts that cater to more sophisticated consumers.

At the same time, however, the evolution of the market demonstrates limits on the power of consumer learning to correct behavioral market failures. For example, the market has responded to greater consumer awareness of the costs of underestimated use among consumers who have experienced the sting of large overage charges. Since 2008, the major carriers have been offering unlimited calling plans that arguably respond to demand generated by this heightened consumer awareness. Yet, while overage fees make it easy to learn the cost of underestimated use, the costs of overestimated use are more difficult to learn since they are not so obviously penalized. The result of this uneven learning is unlimited plans rather than the optimal two-part tariff pricing scheme comprised of a fixed monthly fee and a constant per-minute charge.

(p.192) Another example: The shift from a time-invariant ETF to a time-variant, graduated ETF structure responds to consumers’ increased awareness and sensitivity to ETFs. This shift is not a pure market solution. Rather, it is an example of how consumer learning and legal intervention can work in tandem to change business practices. The change in ETF structure likely began with a small number of consumers who learned to appreciate the cost of ETFs and initiated litigation against the carriers. The threat of liability and greater consumer awareness of ETFs then pushed carriers to adjust their ETF structures.

Innovations like these suggest that the market has an impressive capacity to correct for consumer misperceptions. Yet, market solutions are imperfect. Not all biases are easily purged by learning. Not all consumers learn equally fast, as evidenced by the limited adoption of many design innovations. The speed of consumer learning and the market's response matter, since welfare costs are incurred in the interim period. Moreover, when consumers learn to overcome one mistake, or when a previously hidden term becomes salient, carriers have an incentive to trigger a new kind of mistake or to add a new non-salient term. Even if consumers always catch up eventually, this cat-and-mouse game imposes welfare costs on consumers.

## E. Policy Implications

While market solutions are imperfect and welfare costs remain, the potential for self-correction in the cellular service market merits a regulatory stance that facilitates rather than impedes market forces; disclosure regulation. The proposal we’ll explore deviates from existing disclosure regulation and from most other proposals for heightened disclosure regulation. Current disclosure regulation and other proposals focus on the disclosure of product-attribute information; namely, information on the different features and price dimensions of cellular service. The proposal we’ll explore, by contrast, emphasizes the disclosure of use-pattern information, which as you’ll recall from earlier chapters is information on how the consumer will use the product. To fully appreciate the benefits and costs of a cellular service contract, consumers must combine product-attribute information with use-pattern information. For example, to assess the costs of overage fees, it is not enough to know the per-minute charges for minutes not included in the plan, as proposed in the Cell Phone User Bill of Rights. Consumers must also know the probability that they will exceed the plan limit and by how (p.193) much. Use-pattern information can be as important as product-attribute information. The disclosure regime should be redesigned to ensure that consumers have access to both.

There are two possible approaches to disclosure regulation, approaches that are not mutually exclusive. The first approach focuses on designing simple disclosures that can be easily understood and utilized by imperfectly rational consumers. In particular, carriers should provide total-cost-of-ownership (TCO) information, which is the total amount paid by the consumer, given the consumer's specific use patterns. This information should be provided as an annual disclosure, such as on the year-end summary, to account for month-to-month variations in use. The TCO disclosure combines product-attribute information (pricing information) with information on the specific consumer's use patterns. This disclosure could be further supplemented by information on alternative service plans that would reduce the total price paid by consumers given their current use patterns.

The second approach to disclosure regulation re-conceptualizes disclosure, targeting the disclosed information not directly at consumers but rather at sophisticated intermediaries. Under this approach, carriers would provide comprehensive, individualized use information in electronic, database form. Imperfectly rational consumers will not try to analyze this information on their own. Instead, they will forward the information to sophisticated intermediaries. By combining the use information with the attribute information they collect on product offerings across the cell phone market, the intermediaries would be able to help each consumer find the plan that best suits his or her specific use patterns.

The remainder of this chapter is organized as follows:

• Part I provides background information on the cell phone and the cellular service market.

• Part II describes the key features of common cellular service contracts.

• Part III develops the behavioral-economics theory that explains these contractual design features, after concluding that rational-choice explanations fall short.

• Part IV discusses welfare implications.

• Part V considers the efficacy of market solutions.

• Part VI turns to policy, offering guidelines for enhanced disclosure regulation.

# (p.194) I. The Cell Phone and the Cellular Service Market

## A. The Rise of the Cell Phone

### 1. Technology

The key technological innovation that underpins cellular communications is the cellular concept itself. A cellular system divides each geographic market into numerous small cells, each of which is served by a single, low-powered transmitter. This allows the system to reuse the same channel or frequency in non-adjacent cells in order to avoid interference. Thus, multiple users can simultaneously make use of the same frequency. Sophisticated technology locates subscribers and sends incoming calls to the appropriate cell sites, while complex handoff technologies allow mobile consumers to move seamlessly between cells.3

High demand for cellular service has prompted the development of digital technology, which generates enhanced capacity without degrading service quality. Two kinds of capacity-increasing technological solutions have emerged. The first employs time-slicing technology; signals associated with several different calls are aggregated within the same frequency by assigning to each user a cyclically repeating time slot in which only that user is allowed to transmit or receive. Time-slicing techniques include Bell Labs’ time division multiple access (TDMA) and Global System for Mobile (GSM), which are used by AT&T and T-Mobile, and Integrated Digital Enhanced Network (iDEN), which is used by Nextel. Spread spectrum techniques, by contrast, spread many calls over many different frequencies while using highly sophisticated devices to identify which signals belong to which calls and decode them for end users. The family of digital (p.195) standards employing spread spectrum technology is known as Code Division Multiple Access (CDMA). CDMA standards are used by Verizon and Sprint. The introduction of these digital cellular technologies, starting in the early 1990s, marked the advance from first-generation (1G) systems to second-generation (2G) systems. Third-generation (3G) systems, which began to operate in the U.S. in 2002, incorporate more advanced technologies that provide the increased speed and capacity necessary for multimedia, data, and video transmission, in addition to voice communications. And now fourth-generation (4G) systems are being deployed.

### 2. History

Although the key concepts essential to modern cellular systems were conceived in 1947, the Federal Communications Commission's (FCC) refusal to allocate substantial frequencies to mobile radio service meant that significant development of cellular telephone services was delayed for several decades. It was not until the early 1980s that the FCC allocated 50 MHz of spectrum in the 800 MHz band to cellular telephone service. The FCC rules created a duopoly of two competing cellular systems in each of 734 “cellular market areas”—one owned by a non-wireline company and one owned by the local wireline monopolist in the area. Each carrier received 25 MHz of spectrum. The first set of cellular licenses, which pertained to the thirty largest urban markets (the Metropolitan Service Areas or MSAs) were allocated by comparative hearings. However, the FCC was so overwhelmed by the number of applicants that in 1984 Congress authorized the use of a lottery system to allocate spectrum in the remaining markets. By 1986, all the MSA licenses had been allocated, and by 1991 licenses had been allocated in all markets. As demand for cellular service rapidly increased over subsequent years, the FCC allocated more spectrum to wireless communications. New spectrum has been allocated by auction rather than lottery ever since Congress gave the FCC authority to issue licenses through auctions in the 1993 Budget Act, a move designed to raise revenues and cut down on delays associated with the lottery system.4

The more recent history of the cellular service market in the U.S. is one of consolidation. As noted above, the industry began with the local structural duopolies that were created by the FCC's lottery mechanism. (p.196) With different firms operating in different geographical markets, the national market initially included a large number of players. The number of firms increased further as the FCC auctioned off more and more radio spectrum for cell phone use. But this high level of market dispersion did not last long. The FCC placed few restrictions on the ability of firms to merge across markets, and a long history of voluntary merger and acquisition activity followed. Soon a handful of firms—AT&T Wireless, Cingular, Nextel, Sprint, T-Mobile, and Verizon Wireless—gained a dominant position as nationwide carriers. Consolidation activity increased in 1999, as national carriers sought to fill in gaps in their coverage areas and increase the capacity of their networks while regional carriers sought to enhance their ability to compete with the nationwide operators. Consolidation was further facilitated by the FCC's 2003 decision to abolish the regulatory spectrum cap that had limited the amount of spectrum that a company could own in any one geographical market, since this opened the door to mergers by companies with overlapping coverage areas. Most significantly, in October 2004, Cingular and AT&T Wireless merged to become AT&T Wireless, while in December 2004 Sprint and Nextel merged to become Sprint Nextel.5

### 3. Economic Significance

The FCC estimates that at the end of 2009, there were 291 million cellular service subscribers in the U.S., which corresponds to a nationwide penetration rate of 93 percent. The market has been growing rapidly, albeit with signs that the market is approaching saturation. Cellular service providers added 11.1 million new subscribers in 2009, 16.6 million in 2008, 21.2 million in 2007, 28.8 million in 2006, 28.3 million in 2005, 24.1 million in 2004, and 18.8 million in 2003. An historical perspective underscores the stellar growth of the market; 286 million subscribers were added between June 1990 and the end of 2009. While cell phones complement landline phones for most users, a significant and increasing number of users view the cell phone as a partial or even complete replacement for the traditional, landline phone. In the first half of 2010, an estimated 26.6 percent (p.197) of households used only wireless phones, up from 4.2 percent at the end of 2003.6

The high revenues enjoyed by carriers provide an indication of the magnitude of the cellular service market. In the second quarter of 2011, Verizon posted wireless revenues of $17.3 billion, AT&T$15.6 billion, Sprint $7.5 billion, and T-Mobile$5.1 billion.7 Total quarterly wireless revenues for the four national carriers were $45.5 billion, which potentially translates into total annual wireless revenues of$182 billion, ignoring seasonal variations. Wireless telecommunications have become the largest source of profit for nearly all major telecommunication providers. For example, Verizon's wireless services are about twice as profitable as its wireline offerings.8 Looking at revenues from spectrum auctions is also instructive. In 2006, the FCC's Auction No. 66 raised a total of $13.7 billion in net bids from wireless providers for 1,087 spectrum licenses in the 1710–1755 MHz and 2110–2155 MHz bands.9 In 2008, the FCC's Auction No. 73 raised a total of$19.0 billion in net bids from wireless providers for 1,099 licenses in the 698–806 MHz band (known as the “700 MHz Band”).10

Investment in telecommunications infrastructure in general—and one could argue cellular technology in particular—promotes economic growth (p.198) by reducing the costs of interaction, expanding market boundaries, and enhancing information flow. Specifically, cellular technology can create value by facilitating communication between individuals who are on the move, thus helping individuals to better coordinate their activities and respond to unforeseen contingencies. Wireless services also boost growth by expanding telephone networks to include previously disenfranchised consumers through prepaid service that is unavailable for fixed lines. Analysts estimate that the decades-long delay in the development of cellular networks after the discovery of the cellular concept cost the U.S. economy around $86 billion (measured in 1990 dollars).11 ## B. The Cellular Service Market ### 1. Structure The U.S. cellular service industry is dominated by four “nationwide” facilities-based carriers: AT&T Wireless, Verizon Wireless, Sprint Nextel, and T-Mobile. At the end of 2010, each had networks covering at least 250 million people. AT&T had 95.5 million subscribers, Verizon 94.1 million, Sprint Nextel 49.9 million, and T-Mobile 33.7 million.12 In addition to the national carriers, there are a number of regional carriers, including Leap, U.S. Cellular, and MetroPCS. There is also a growing resale sector, consisting of providers who purchase airtime from facilities-based carriers and resell service to the public, typically in the form of prepaid plans rather than standard monthly tariffs. ### (p.199) 2. Competition The overlapping geographic coverage of the national and regional providers gives rise to competition between cellular service providers. The FCC estimates that 97.2 percent of people have three or more different operators offering cell phone services in the census blocks where they live, 94.3 percent live in census blocks with four or more operators, 89.6 percent live in census blocks with five or more operators, 76.4 percent live in census blocks with six or more operators, and 27.1 percent live in census blocks with seven or more operators. The FCC measures market concentration by computing the average Herfindahl-Hirschman Index (HHI) across 172 Economic Areas (EAs)—aggregations of counties that have been designed to capture the “area in which the average person shops for and purchases a mobile phone, most of the time.” The HHI is a measure of market concentration that ranges from a value of 10,000 in a monopolistic market to zero in a perfectly competitive market.13 In mid-2010, the average HHI, weighted by EA population, was equal to 2,848. An industry with an HHI above 2,500 is considered highly concentrated by the antitrust authorities. And these figures might well underestimate market concentration, since the FCC's methodology gives equal weight to a mobile carrier assigning cell phone numbers in one county as it does to a carrier that assigns numbers in multiple counties in a given EA.14 Indeed, one analyst calculated an average HHI value exceeding 6,000 with 2005 data, using the amount of spectrum controlled by a carrier in a market as a proxy for market share.15 The relatively high level of concentration in the cell phone market is the product of an ongoing consolidation process. This consolidation is at least partly motivated by a desire to realize economies of scale and enlarge geographic scope. Broad coverage can be provided at lower cost by a single nationwide carrier than by regional carriers through roaming agreements with carriers operating in different geographic areas. In addition, extending (p.200) the national network spreads fixed costs, such as marketing expenditures and investments in developing new technology, over a wider base of customers. Economies of geographic scope arising from complementarities between markets may also provide an efficiency reason for consolidation.16 However, even if consolidation reduces certain costs, other costs may increase. Consolidation tends to reduce competition and facilitate collusion as the number of multi-market contacts between the dominant national carriers increases.17 The magnitude of entry barriers provides another important measure of competitiveness. If entry barriers are low, even a market with a small number of firms will behave competitively. Government control of spectrum—limiting the amount of spectrum allocated to wireless communications and requiring carriers to obtain a government-issued license—has the potential to create significant barriers to entry. However, the FCC has alleviated many of these concerns recently by increasing the amount of spectrum available for cellular communication services and allowing market forces to determine market structure through elimination of the old structural duopolies and abolition of the spectrum cap. Moreover, the Telecommunications Act and FCC regulations reduce entry barriers by imposing interconnection and roaming obligations. The ability to purchase spectrum on the secondary market further reduces entry barriers.18 Meanwhile, advertising expenditures—amounting to billions of dollars annually19—and the economies of scale and scope described above continue to impose substantial entry barriers. (p.201) Switching costs also affect the level of competition. Switching costs in the cellular service market are substantial, although recent developments are reducing these costs. Until recently, most consumers signed long-term contracts with fixed ETFs of approximately$200. But now major carriers are offering contracts with graduated ETFs that decline over the life of the contract. Likewise, historically carriers have allowed only certain approved phones to be used by their subscribers on their network and “locked” the phones they sold to render them incapable of being used on other networks.20 The recent trend, however, is toward open access, which allows more phones onto the network, and recent regulatory action by the Copyright Office clarified that phones can be unlocked.21 Being forced to change phone numbers was also a potentially significant switching cost until it was eliminated by the regulatory requirement that carriers provide local number portability.22 The high churn rates in the cell phone market—between 1.5 percent and 3.3 percent per month in 200923—suggest that switching costs, while potentially substantial, are not prohibitive for many consumers.

To sum up, while there is reason to believe that the cellular service market is less than perfectly competitive, cellular service providers are actively competing to attract consumers. Declining prices, albeit with a leveling off in recent years, are evidence of such active competition.24 Competition is (p.202) also observed on non-price dimensions. Competition to attract and retain customers appears to be driving carriers to improve service quality. Carriers pursue a variety of strategies to improve service quality, including network investment to improve coverage and quality and acquisition of additional spectrum.25 While an economic conclusion reached by politically appointed regulators should be taken with a grain of salt, it is noteworthy that the FCC described the cellular service market as one characterized by healthy competition with carriers engaging in “independent pricing behavior, in the form of continued experimentation with varying pricing levels and structures, for varying service packages, with various handsets and policies on handset pricing.”26

### 3. Related Markets

The cellular service market interacts with other markets, specifically with the market for phones/handsets and with the market for cell phone applications.

#### a. The Handset Market

The market for handsets is controlled by five firms: Samsung, LG Electronics, Motorola, Apple, and RIM. In the U.S., Samsung enjoys the largest market share, controlling 25.5 percent of the handset market in the second quarter of 2011. LG placed second with 20.9 percent of the market, and Motorola followed with 14.1 percent. Apple and RIM, the maker of BlackBerry, lag behind considerably with 9.5 percent and 7.6 percent of the market, respectively.27

In the U.S., the major cellular service providers exert significant control over the handset market. Internationally, about half of handsets are (p.203) purchased through carriers and about half are sold directly to consumers through other channels. In the U.S., the vast majority of cell phones—nine out of every ten phones according to one estimate—are sold through a service provider.28 The practice of subsidizing handset prices for consumers who sign long-term service contracts is at least partially responsible for the competitive disadvantage suffered by handset makers looking to sell directly to consumers.

Carriers in the U.S. determine which devices consumers can operate on their networks. The result of this control by service providers is that only a fraction of any given manufacturer's total line of products is offered. For example, in 2006, of the fifty new products Nokia introduced into the market, U.S. cellular service providers offered a scant few. By allowing only certain approved phones on their networks, carriers influence the design of handsets. Moreover, as a condition of network access, carriers require that developers disable certain services or features that might be useful to consumers, such as call-timers, photo sharing, Bluetooth capabilities, and Wi-Fi capabilities.29

But the balance of power is shifting. Handset brands and models are an increasingly important determinant of a consumer's choice of service provider. Apple's launch of the iPhone is a significant example of a handset manufacturer successfully overcoming carrier pressure. More generally, the rapid expansion of the “smartphone” market is enhancing the power of handset makers and companies who provide software for these handsets. An example is the Android operating system, developed by Google.30

In addition, the open-access trend is starting to limit carriers’ control over the handset market.31 Regulation is also playing an important role: (p.204) One-third of the recently auctioned spectrum comes with a requirement that “cellular networks allow customers to use any phone they want on whatever network they prefer, and be able to run on it any software they want.”32 And, perhaps sensing the inevitable, carriers are beginning to embrace the new open-access business model, reasoning that they can cut costs by eliminating handset subsidies and letting handset manufacturers bear most of the development and customer service costs.33

## B. Lock-In Clauses

In addition to the three-part tariff pricing structure, most postpaid calling plans share the following two features; a free or substantially discounted phone and long-term contracts with ETFs. At the time of this writing, AT&T gave customers the option to buy the LG Phoenix for $379.99 without a contract. With the signing of a two-year data contract, however, AT&T charged$0.01 for the same phone. T-Mobile customers (p.207) who signed a new contract had the opportunity to receive, for free, a number of phones with suggested retail values of over $200, including the LG Optimus, the HTC Wildfire, and the Samsung Gravity. Similarly, AT&T and Apple heavily subsidized the iPhone, sacrificing short-term revenues, and Sprint sold Samsung's music phones far below cost at only$149.45 The free or heavily subsidized phone strategy pervades the U.S. cell phone market. A 2011 survey by J.D. Power found that 42 percent of customers receive a free cell phone when subscribing to a wireless service.46

Of course, free phones are not really free. Carriers recoup the costs of the phones through subscription fees. To make sure that they collect enough subscription fees to cover the cost of the phone, carriers lock consumers into long-term contracts.47 Such lock-in is secured by substantial ETFs. For example, in June 2007, T-Mobile charged a fixed termination fee of $200, AT&T charged$175, and Sprint charged up to $200, depending on the service selected. Historically, the same termination fees were charged regardless of when the agreement was broken, meaning that a consumer would have paid a$200 termination fee for ending a two-year contract just one month early. In the wake of a number of class action lawsuits challenging the legality of these fees, providers have begun to offer contracts with termination fees that decline over the life of the contract. Verizon led this transition when, in June 2007, it started charging customers a termination fee of $175, less$5 for each full month that the customer remains on the initial contract. By the end of 2008, all the major carriers were offering similar graduated ETFs.48

## (p.208) C. Complexity

The complexity and multidimensionality of cellular service contracts can be viewed as a contractual design feature. Most cellular service contracts are highly complex in and of themselves. This high level of complexity increases substantially when the perspective shifts from a single-contract to the many different multidimensional contracts being offered. According to one industry estimate, the cellular service market boasts “more than 10 million different plans and add-on combinations.”49

### 1. Postpaid Plans—The Basics

Even the basic components of the common postpaid calling plan are complex. As described above, the basic pricing scheme is three-dimensional. Each dimension of the basic pricing scheme is one of the tariffs that make up the three-part tariff. Moreover, each provider offers a long menu of different three-part tariffs. To make things even more complicated, the menus of three-part tariffs vary among providers.50 Further complexity is introduced by the diversity of additional service features covered by the fixed monthly fee. Some of these features are offered by all carriers in the exact same way. Others are offered by some carriers but not others or are offered in varying formats by the different carriers.

For example, all four major carriers offer unlimited calls during off-peak times, usually nights and weekends. There is, however, some potentially significant variation. Nights are defined differently by different carriers. For AT&T and Verizon, the night begins at 9 p.m. and ends at 6 a.m. For T-Mobile, the night begins at 9 p.m. and ends at 7 a.m. For Sprint, the night begins at 7 p.m. and ends at 7 a.m., except for its more basic plans, where the night begins at 9 p.m. and ends at 7 a.m. By varying the definition of “night,” providers can offer up to three extra hours of unlimited calling. (p.209) These three extra hours represent an additional 33.3 percent of unlimited calling time. But since most consumers probably talk more during the three hours between 7 and 9 p.m. and between 6 and 7 a.m. than they do during the three hours between, say 1 and 4 a.m., these extra three hours of unlimited calling probably represent much more than a 33.3 percent increase in value.

To take another example, consumers might also consider whether to select Verizon's Friends and Family program, offering unlimited calls to five (or ten) phone numbers selected by the user. Sprint, T-Mobile, and AT&T offered similar programs, but as of this writing their plans were no longer available to new users.

### 2. Family Plans

We have thus far focused on individual calling plans. The four major carriers also offer family plans, adding another layer of complexity. The identifying feature of a family plan is the ability to share the allotted minutes between up to five users, each operating on a different line. For example, Verizon offers family plans with monthly charges ranging from $69.98 to$119.98, allotted minutes ranging from 700 to unlimited, and overages ranging from $0.45 to$0.35. These monthly prices include two phone lines, and families can add up to three more lines for an additional $9.99 per month per line (or$49.99 under the unlimited plan).

Cell phones can be used for much more than voice communication. Carriers offer advanced communication services, including text messaging, multimedia messaging, and Internet and email data services. Carriers also offer applications, such as ringtones and games, as well as monthly mobile Internet access packages. These services and applications are marketed to consumers primarily as add-ons to their voice services.

Pricing of these services adds additional complexity. Providers offer advanced communication services to consumers in one of three modes:

1. (1) Pay-as-you-go, applied mainly to text and multimedia messaging, where the consumer pays per message sent or received

2. (2) Fixed-quantity monthly packages, where the consumer pays a monthly fee for a fixed number of allotted messages or megabytes of data

3. (p.210) (3) Unlimited-quantity monthly packages, where the consumer pays a monthly fee for unlimited messaging or data transmission.51

Entertainment applications, specifically ringtones and games, can be purchased for a one-time download rate. Advanced applications, such as GPS location services and music and TV applications, are now also available from some providers—usually for an additional monthly or daily fee.

# (p.212) III. Explaining the Cellular Service Contract

What explains the contractual design features described in Part II? Why does the common cellular service contract look the way it does? In this Part, we’ll explore possible rational-choice, efficiency-based theories for each of the three design features: three-part tariffs, lock-in clauses, and complexity. But as we’ll see, these theories provide only a partial account of the contractual outcomes in the cellular service market. We’ll fill the explanatory gap by developing a behavioral-economics theory that explains contractual design in the cellular service market as a market response to consumer mistakes.53

## A. Three-Part Tariffs

### 1. Rational-Choice Theories and Their Limits

The leading rational-choice explanation for three-part tariffs views these pricing schemes as a mechanism for price discrimination or market screening of rational consumers with different ex ante demand characteristics. For expositional purposes, let's focus on two dimensions of demand heterogeneity: average (or mean) monthly minutes of use and variance of minutes used.

Suppose that consumers vary only on the first dimension, average monthly minutes used. Here, the rational model cannot explain three-part tariffs. To discriminate between heavy users with high average usage and light users with low average usage, carriers would use a menu of two-part, not three-part, tariffs. A two-part tariff includes a fixed monthly fee and a constant per-minute charge. Carriers can discriminate between heavy users and light users by offering an “H” tariff with a higher monthly fee and a lower per-minute charge and an “L” tariff with a lower monthly fee and a higher per-minute charge. The heavy users care more about the per-minute charge, and will thus prefer the H tariff. The light users care more about the monthly fee, and will thus prefer the L tariff.

While two-part tariffs provide a mechanism for discriminating between consumers based on their mean usage, three-part tariffs can provide a mech (p.213) anism for discriminating between consumers based on variance of use. Assume that there are two types of consumers: one type with highly variable—or High-Variation (HV)—demand, and another type with more predictable, Low-Variation (LV) demand.54 Specifically, suppose that the HV type's demand will either be H + h or Hh with equal probability, while the LV type's demand will be either L + l or Ll with equal probability, where hl and H + hL + lLlHh. A monopolistic carrier can discriminate between the HV types and the LV types using a menu of three-part tariffs. This is because HV types are more concerned than LV types about: (i) using a very large number of minutes, which makes them more inclined to choose a tariff with a larger allocation of minutes to reduce the risk of paying substantial overage fees; and (ii) using only a very small number of minutes, since then, for any given fixed monthly fee, they end up paying a higher per-minute price.

The monopolist can exploit this difference by designing a three-part tariff for LV types with steep overages above L + l, LV's highest possible demand, and offering HV types a tariff with at least H + h free minutes for a higher monthly price. LV types like the LV tariff more than the HV types since: (i) they are indifferent to the possibility of overages as they never consume minutes above L + l, while the HV types will end up paying overages half of the time if they choose the LV tariff, and (ii) the zero marginal price on allotted minutes within their plan is worth relatively more to LV types, since, unlike the HV types, they always consume at least L – l of those minutes, whereas HV types still end up paying a high average price per minute in the event that their demand is low.

While a three-part tariff pricing structure can facilitate price discrimination, the assumptions required for this rational-choice explanation are often unrealistic. In the price-discrimination model, the HV type chooses a plan with a high number of allotted minutes and the LV type chooses a plan with a low number of allotted minutes. Moreover, their highly variable use levels imply that HVs are more likely than LVs to use a very low number of minutes. Using the data described in subsection 2 below, Figure 4.1 shows the cumulative distribution functions of usage for consumers choosing different three-part tariff plans: Plan 1 with 200 included minutes, plan 2 with 300 included minutes, plan 3 with 400 included minutes, and plan 4 with 500 included minutes.

(p.214)

Figure 4.1. Cumulative distribution functions of cell phone usage

Figure 4.1 confirms that the cumulative distribution function corresponding to a plan with a higher number of allocated minutes first-order stochastically dominates the cumulative distribution function corresponding to a plan with a lower number of allocated minutes.55 In other words, consumers who choose plans with a higher number of allotted minutes are less likely to end up using a very low number of minutes. These findings are inconsistent with the price-discrimination theory.56

An alternative rational-choice explanation is based on risk aversion. In theory, the use patterns revealed in the data (see subsection 2 below) are consistent with the behavior of perfectly rational but risk-averse subscribers. Such subscribers would choose plans with more allotted minutes than they expect to use to reduce the risk of paying substantial overage fees. As a result, most of these subscribers will end up using much less than their allotted minutes. This explanation fails for three reasons. First, given the sums of money involved, the observed plan choices are not consistent with risk aversion under the rational-choice Expected Utility Theory.57 Second, as demonstrated below, subscribers often (p.215) choose the wrong plan. And, importantly, these mistakes in plan choice are not only ex post mistakes as the risk-aversion explanation would imply; many of them are ex ante mistakes. Finally, while risk aversion may explain the observed usage patterns, it cannot explain the emergence of the three-part tariff as the equilibrium pricing structure. With rational, risk-averse subscribers, we should expect to see two-part tariffs.

### 2. A Behavioral-Economics Theory

#### a. Theory

Basic voice services are commonly priced using three-part tariffs. To choose the optimal three-part tariff from the menu of tariffs offered by the carriers, consumers must accurately anticipate their future cell phone usage. But many consumers, when asked to choose a calling plan, are not armed with accurate estimates of how they will use their cell phones. According to a pricing manager at a top U.S. cell phone service provider, “people absolutely think they know how much they will use and it's pretty surprising how wrong they are.”58 The three-part tariff responds to consumers’ misperceptions about their future use.

Consumers both overestimate and underestimate their use levels. A carrier who is aware that consumers suffer from such misperceptions can make its service plan appear more attractive to consumers than it really is by using a three-part tariff that charges a low (zero) per-minute price for minutes up to the plan limit and a high per-minute price thereafter. Consumers who overestimate their usage overestimate the value of the low prices because they overestimate the probability that they will use most of these free minutes. Conversely, consumers who underestimate their usage pay insufficient attention to the high overage fees because they underestimate the probability of exceeding the plan limit. For a monopolistic carrier, the three-part tariff creates opportunities for increased profits, while carriers operating in a competitive market will adopt the three-part tariff because it maximizes perceived consumer surplus.59

These ideas can be illustrated using a simple numeric example. Assume that several carriers are operating in a highly competitive market. All carriers (p.216) face the same cost structure: a $10 per-consumer fixed cost and a$0.10 per-minute variable cost. Consumers have the following preferences: they value each minute of airtime at $0.40 per minute up to a certain saturation point, s, while minutes beyond the saturation point are worth zero to the consumer. There are two types of consumers: heavy users and light users. Fifty percent are heavy users with a saturation point of 300 minutes, and 50 percent are light users with a saturation point of 100 minutes. If consumers are rational and accurately predict their saturation points, then the carriers will set a two-part tariff with a fixed monthly fee of$10 and a constant, per-minute marginal price of $0.10. Heavy users will pay 10 + 300 · 0.1 = 40, light users will pay 10 + 100 · 0.1 = 20, and the carriers will just cover their costs, as expected in a perfectly competitive market. Under this two-part tariff, heavy users enjoy a surplus of 300 · (0.4–0.1) –10 = 80 and light users enjoy a surplus of 100 · (0.4–0.1) –10 = 20.60 Let's introduce consumer misperceptions into the mix. We assume that light users overestimate their saturation point, mistakenly perceiving a saturation point of 200 minutes instead of the actual 100 minutes. And heavy users underestimate their saturation point, mistakenly perceiving a saturation point of 200 minutes instead of the actual 300 minutes. With such misperceptions, a three-part tariff becomes more appealing than the two-part tariff. Consider the following three-part tariff: a fixed$10 monthly fee, 200 allotted minutes (at a marginal price of zero), and an overage charge of $0.40 per minute beyond the 200-minute allocation. The 200-minute allocation tracks the common perceived saturation point, the$0.40 overage is the maximal marginal price that would not deter usage beyond the plan limit, and the $10 fixed fee is calculated to exactly cover the carrier's expected costs: $10 + ( 1 2 100 + 1 2 300 ) 0.1 - 1 2 ( 300 - 200 ) 0.4 = 10$.61 Under this tar (p.217) iff, heavy users will pay $10 + ( 300 - 200 ) 0.4 = 50$10+(300–200). 0.4=50. They will enjoy a surplus of $300 0.4 - ( 300 - 200 ) 0.4 - 10 = 70$300 0 .04–(300-200). 0.4-10=70, less than the surplus of 80 under the two-part tariff. But since they underestimate their use, they misperceive the surplus. The perceived surplus under the three-part tariff is $200 0.4 - 10 = 70$200 0 .04–10=70, greater than the perceived surplus of $200 ( 0.4 - 01 ) - 10 = 50$200. (0.4–01)–10=50 under the two-part tariff. Light users will pay$10 under the three-part tariff. They will enjoy a surplus of $100 0.4 - 10 = 30$100 0 .04–10=30, more than the surplus of 20 under the two-part tariff. More importantly, the perceived surplus under the three-part tariff is $200 0.4 - 10 = 70$200 0 .04–10=70, greater than the perceived surplus of $200 ( 0.4 - 0.1 ) - 10 = 50$200. (0.4–0.1)–10=50 under the two-part tariff.

The three-part tariff extracts payments in the form of overage fees that are invisible to consumers,62 while reducing or eliminating payments that are visible to consumers—specifically, fixed fees and charges for minutes within the plan limit. Notice that the heavy users, who underestimate their usage levels and end up paying overage fees, are subsidizing the light users. But since the heavy users do not anticipate paying the overage fees, a competitor cannot lure them away ex ante by, for example, offering a different tariff with lower overage fees. The three-part tariff maximizes the perceived consumer surplus for both types of consumers, and thus will be selected as the equilibrium tariff in a competitive market.

#### b. Data

A unique dataset of subscriber-level monthly billing and usage information for 3,730 subscribers at a single wireless provider was used to test the misperception theory. These data provide information on which calling plan (of four) subscribers chose, as well as monthly consumption of peak minutes for the period of September 2001 to May 2003. Each of the four calling plans offered a standard three-part tariff with a fixed allocation of peak minutes and steep overages for additional peak minutes consumed, as described in Table 4.1 below.63

(p.218)

Table 4.1. Menu of  Three-Part Tariffs

Plan 1

Plan 2

Plan 3

Plan 4

Market share (%)

22.15

2.00

73.28

2.57

Monthly fixed charge ($) 30 35 40 50 Number of included minutes 200 300 400 500 Overage rate ($)

0.40

0.40

0.40

0.40

The data reveal substantial variance in usage. Summary statistics are provided in Tables 4.2a–e below. For plans 1, 2, 3, and 4, Tables 4.2a–d present the overall mean and standard deviation of minutes used. To gain an initial sense of usage underestimation versus overestimation of usage, average figures for under-usage (unused minutes per month) and over-usage (minutes over the plan allocation) are also presented. This information is aggregated across all plans in Table 4.2e.

In aggregate, subscribers exceeded their minute allowance 11.4 percent of the time by an average of 34.3 percent. In the 88.4 percent of the time when the allowance was not exceeded, subscribers used, on average, only 45.4 percent of their minute allowance. Notice that “underages” and overages do not, in and of themselves, imply overestimation and underestimation of use. A perfectly rational consumer with variable use will experience both underages and overages.

Let's now look at consumers who arguably chose the wrong plan, and the cost of the mistake to those consumers. A plan choice is a mistake when, given the consumers’ usage, the selected plan is different from another available plan that would have cost less. The unit of analysis is the consumer's tenure with a plan, and only the 3,456 consumers who stayed with a plan for at least ten months are considered. Given the variance in usage from month to month, identifying mistakes over shorter time horizons is less reliable. For each of the 3,456 consumers, the total cost of (p.219)

Table 4.2a. Summary Statistics—Plan 1

Share

Usage/Allowance

Mean

Std. Dev.

Under allowance

0.819

0.45

0.294

Over allowance

0.179

1.46

0.624

All consumers

1

0.633

0.538

Table 4.2b. Summary Statistics—Plan 2

Share

Usage /Allowance

Mean

Std. Dev.

Under allowance

0.872

0.51

0.274

Over allowance

0.126

1.25

0.279

All consumers

1

0.607

0.368

Table 4.2c. Summary Statistics—Plan 3

Share

Usage/Allowance

Mean

Std. Dev.

Under allowance

0.911

0.45

0.287

Over allowance

0.089

1.284

0.324

All consumers

1

0.524

0.375

Table 4.2d. Summary Statistics—Plan 4

Share

Usage/Allowance

Mean

Std. Dev.

Under allowance

0.718

0.573

0.296

Over allowance

0.279

1.259

0.29

All consumers

1

0.766

0.425

Table 4.2e. Summary Statistics—Aggregate

All Plans

Share

Usage/Allowance

Mean

Std. Dev.

Under allowance

0.884

0.454

0.289

Over allowance

0.114

1.343

0.457

All consumers

1

0.557

0.422

(p.220) wireless service under the consumer's chosen plan was compared to the total amount that this consumer would have paid had each of the other three plans been chosen. The magnitude of the mistakes is measured by the difference (in both percentages and dollars) between the consumer's actual wireless costs and the lowest possible cost—the cost that the consumer would have paid if the consumer could have predicted usage with certainty.64

The results are summarized in Tables 4.3a and 4.3b. In these tables, each row represents the group of subscribers who chose a certain plan. (Note that there is no row for Plan 2, since no Plan 2 subscriber remained with the plan for more than 10 months.) This group is then divided into four subgroups according to the plan that these subscribers should have chosen. For instance, the cell located at the intersection of the Plan 3 row and the Plan 1 column represents the subgroup of subscribers who chose Plan 3 but should have chosen Plan 1. Table 4.3a presents the size, in percentage terms, of these subgroups. Table 4.3b presents the magnitude of the mistakes or cost-savings, both in percentage terms and in annual dollar terms, for each subgroup.65

The results for one group of subscribers, those who chose Plan 3, are presented in Figure 4.2. This group of subscribers is noteworthy for its significant numbers of both under-estimators, who should have chosen Plan 4, and over-estimators, who should have chosen either Plan 2 or Plan 1. Figure 4.2 displays the share of Plan 3 consumers who should have chosen each of the four plans (the dark gray bars). For those who should not have chosen

Table 4.3a. The likelihood of mistakes

Optimal Plan

Plan 1

Plan 2

Plan 3

Plan 4

Chosen plan

Plan 1

74.09%

21.79%

1.49%

2.49%

Plan 3

27.20%

35.61%

21.19%

16%

Plan 4

9.00%

10.66%

8.00%

73.33%

(p.221)

Table 4.3b. The magnitude of mistakes

Optimal Plan

Plan 1

Plan 2

Plan 3

Plan 4

Chosen plan

Plan 1

0%

9.56%

26.97%

28.22%

$0$54.16

$203.58$341.71

Plan 3

21.09%

6.55%

0%

11.34%

$101.58$32.59

$0$102.98

Plan 4

36.71%

12.38%

7.00%

0%

$220.27$75.31

$39.90$0

Plan 3, Figure 4.2 shows the amount of money they would have saved, both in percentage terms (the light gray bars) and in dollar figures.

These figures underestimate the number and cost of mistakes, especially for plans with a lower allocation of minutes. For example, for subscribers who chose Plan 1, the data only reveal mistakes arising from underestimation of use (selection of Plan 1) when the subscriber should have chosen Plan 2, Plan 3, or Plan 4. But it is likely that many Plan 1 subscribers who overestimated their use could have done better by choosing a prepaid plan that is not included in the dataset. A conservative estimate of the number and magnitude of the cost of such overestimation can be generated by adding a hypothetical prepaid plan with a high per-minute charge of $0.40 (equal to the overage charges in our data). An estimated 24.4 percent of Plan 1 subscribers would have saved$149 annually on average had they chosen the prepaid plan.66

The mistakes in plan choice that we have discussed are ex post mistakes. Since consumers face ex ante uncertainty about their future use, even consumers who forecast their use ex ante in a rational manner will make predictions that turn out to be incorrect ex post. To assess the extent of ex ante mistakes in our data and again focusing on subscribers who stayed with the same plan for at least 10 months, each subscriber's tenure with the plan was divided into two: the first-half and the second-half periods.

(p.222)

Figure 4.2. Plan 3 subscribers—likelihood and magnitude of mistakes

Consider a subscriber who, given her usage in the first-half period, should have chosen Plan 1 but chose Plan 3 instead. This subscriber, who could have easily switched to Plan 1 at the end of the first-half period (subscribers were not locked in to a plan), stayed with Plan 3 in the second-half period even though Plan 1 remained the optimal plan based on the second-half period usage. Since the same plan was optimal in both half periods, her usage must have remained roughly the same during the two half periods, which means that her first-half period usage provided her with a basis to update her predictions about her usage in the second-half period. Since subscribers were free to switch plans and their switching costs were likely to be low, we can surmise that the subscriber's decision to choose Plan 3 was probably an ex ante mistake—at least for the second-half period. The data reveal that a substantial percentage of subscribers who chose the wrong plan in the first-half period also chose the wrong plan in the second-half period.67

(p.223) To sum up: Many consumers fail to accurately anticipate their use patterns, and the three-part tariff design is a market response to such misperceptions.

Consistent with this story, providers do not seem to be troubled by consumers’ use-pattern mistakes. In fact, until recently they actively foster these mistakes by requiring, as a condition for network access, that handset manufacturers disable the call-timer feature that would make it easier for consumers to monitor their usage.68 However, consumers are becoming more aware of their use-pattern mistakes and more frustrated with carriers who take advantage of them. As elaborated in Part V below, the market is responding to the demand generated by these more sophisticated consumers.

## B. Lock-In Clauses

### 2. Rollover Minutes

Consumer use varies from month to month. A consumer may talk 350 minutes one month and 550 minutes the next. With a standard 450-minute plan, this consumer will waste 100 minutes in the first month and pay overage (p.235) charges for 100 minutes in the second. With AT&T's 450-minute plan, which includes the rollover minutes feature, the 100 spare minutes in the first month are not wasted. Rather they are “rolled over”—added to the available minutes for the second month.85 This means that in the second month, the consumer has 550 minutes instead of 450 minutes and thus will not pay any overage.86 The rollover feature, which predates the unlimited calling plans described above, can also be seen as a response to consumer learning about the costs of underestimated use and overage charges. But unlike unlimited plans that directly respond to underestimation of use, the rollover feature seems to respond to a different bias—overconfidence about use levels, which implies underestimation of use in some months and overestimation of use in others. By enabling the consumer to smooth uneven use over time, the rollover feature mitigates the costs of overconfidence.

### 3. Prepaid Plans

Prepaid, no-contract plans are the natural choice for a sophisticated consumer who wants flexibility and has learned the costs of lock-in. This flexibility, however, comes at a cost. Not only do prepaid, no-contract subscribers forgo the phone subsidies offered to postpaid, locked-in subscribers, but they also pay higher per-minute charges (at least as compared to postpaid subscribers who use all the allotted minutes under their plans). As a result, even sophisticated consumers would be reluctant to choose a prepaid plan. In fact, prepaid, no-contract plans, with their lower profitability, were designed for distinct segments of consumers—specifically younger and poorer consumers who have low credit scores and do not qualify for a postpaid plan.87 The numbers confirm this: In 2008, only 16 percent of U.S. cell phone users had prepaid plans; among households with incomes above $75,000, only 6 percent of cell phone users had prepaid plans.88 This is starting to change. With the growth of unlimited prepaid offerings and the reduction in per-minute rates, prepaid plans are now attracting (p.236) consumers from segments of the market previously controlled by postpaid plans. Prepaid is becoming a real alternative to postpaid.89 While having a prepaid alternative is valuable, prepaid plans are not a panacea. While solving the lock-in problem and avoiding underestimation of lock-in costs, prepaid plans trigger other consumer mistakes. Misperceptions about future use may still lead consumers to choose the wrong monthly prepaid or pay-as-you-go plan. In fact, expiration dates on minutes purchased under pay-as-you-go plans—important design features of such plans—may be a response to consumers’ overestimation of use. ### 4. Graduated ETFs As described in Part II.B, carriers have been moving from a time-invariant ETF to a time-variant, graduated ETF structure. This shift is a response to consumers’ increased awareness and sensitivity to ETFs. The change in the design of ETF provisions is not a pure market solution. Rather, it is an example of how consumer learning and legal intervention can work in tandem to change business practices. The ETF story likely began with a small number of consumers who learned to appreciate the cost of ETFs and initiated litigation against the carriers. The threat of liability probably pushed carriers to adjust their ETF structure. But the litigation also facilitated greater awareness and sensitivity to ETFs among consumers. This adjusted demand was something that carriers could not ignore. ### 5. Open Access The open-access movement in wireless telecommunications is a market-driven development that could reduce the costs of lock-in and handset-service bundling. While carriers are still the leading handset retailers, recent developments are diminishing their power such that it is likely that handset manufacturers will increasingly sell their products directly to consumers, who can use the phone on any network. Open access is not a response to consumer learning about biases and the cost of lock-in. Nevertheless, it is an important development that can reduce the costs of consumer biases. ## (p.237) B. Market Solutions and Consumer Welfare The cellular service market seems quite responsive to demand generated by increasingly sophisticated consumers who learn from their mistakes. From a policy perspective, the question is to what extent market solutions mitigate the welfare costs identified in Part IV. As we have seen, the market promptly responds when consumers quickly learn about the implications of their mistakes, as they do when underestimated use leads to overage charges. But we have also seen that the market responds more sluggishly when learning is slower because the feedback mechanisms are weaker, as is the case with overestimated use. While the market solutions described above have the potential to minimize the welfare costs of the identified behavioral market failure, in practice their impact is more limited. The reason is that many consumers do not take advantage of these market solutions. For example, unlimited plans with their high monthly fees are attractive only to a small fraction of heavy users. Prepaid plans are chosen by a minority of consumers. If consumers are not aware of their mistakes, they will not search for products that reduce the likelihood and consequences of those mistakes. It is evident, then, that consumers learn and that the market responds to the demand generated by these more sophisticated consumers. But welfare costs are incurred during the interim period. To assess the magnitude of welfare costs, we need to ascertain the speed of consumer learning and of the market response to changing demand. Moreover, when consumers learn to overcome one mistake or when one hidden term becomes salient, carriers have an incentive to add a new non-salient term and to trigger a new kind of mistake. Even if consumers always catch up eventually, this cat-and-mouse game imposes welfare costs. Wireless operators are among the leading generators of consumer complaints.90 Market solutions, while important, are clearly imperfect. # (p.238) VI. Policy Implications The identified behavioral market failure imposes substantial welfare costs. Consumer learning coupled with market forces works to reduce these welfare costs but do not eliminate them. Can legal intervention help? In this Part, we’ll focus on disclosure mandates which, we believe, can help. We’ll start with a brief survey of existing rules and regulations. We’ll then outline several potential reforms. ## A. Existing Regulations Regulation of the consumer–carrier relationship is largely limited to the information that the provider must disclose to its consumers.91 The FCC exercised its powers under the Communications Act by promulgating rules intended to prevent fraudulent behavior by telecommunications providers and by increasing the transparency of providers’ billing practices. Providers must clearly identify the name of the service provider associated with each billed charge and prominently display a toll-free telephone number that customers can call to inquire about or dispute any charges.92 Most importantly, since 2005, charges must “be accompanied by a brief, clear, non-misleading, plain-language description of the service or services rendered” that is “sufficiently clear in presentation and specific enough in content so that customers can accurately assess that the services for which they are billed correspond to those that they have requested and received, and that the costs assessed for those services conform to their understanding of the price charged.”93 The underlying rationale is “to allow consumers to better understand their telephone bills, compare service offerings, and thereby promote a more efficient competitive marketplace.”94 Further disclosure (p.239) requirements are imposed at the state level. In particular, state laws regulate wireless line item charges—discrete charges that are separately identified on a consumer's bill.95 There have been calls for more stringent disclosure requirements. For instance, in 2003, Senator Charles Schumer introduced a bill—The Cell Phone User Bill of Rights—designed to improve disclosure and make it easier for consumers to choose among providers and plans. The bill sought to ensure that marketing materials and contracts clearly spell out the terms and conditions of service plans by requiring that all wireless contracts and marketing materials display a box containing standardized information on a number of key price dimensions, including the monthly fixed charge, per-minute charges for minutes not included in the plan, and the method for calculating minutes charged. Information on included weekday and daytime minutes and nights and weekend minutes, long-distance charges, roaming charges, incoming call charges, and charges for directory assistance would also have to be displayed. Termination and start-up fees and trial periods would have to be outlined as would any taxes and surcharges. In addition, the bill would authorize the FCC to monitor service quality industry-wide and make the resulting data publicly available to enable consumers to make informed choices among providers.96 The bill has not been enacted into law. In 2004, the California Public Utility Commission (CPUC) promulgated a similar set of rules. These regulations required wireless providers and other telecommunications operators to (1) ensure that subscribers receive clear and complete information about rates, terms, and conditions when customers sign up for the service; (2) produce clearly organized bills that only contain charges that the subscriber has authorized; and (3) list all federal, state, and local taxes, surcharges, and fees separately.97 The regulations were suspended by the CPUC less than a year after their (p.240) adoption, after the terms of two commissioners who supported the rules expired.98 The drive for improved disclosure, however, is continuing: Twenty-two states have introduced some form of a Cell Phone User Bill of Rights.99 ## B. Rethinking Disclosure ### 1. From Product Attributes to Use Patterns Consumers in the cellular service market learn, often quite effectively, to appreciate the implications of their biases and mistakes. Carriers respond with products that reduce resulting costs to consumers. While these market solutions are imperfect, the market's responsiveness suggests that the regulation best suited for the cellular service market would facilitate rather than inhibit market forces. It is, therefore, not surprising that many of the existing and proposed laws and regulations focus on the provision of information. We too focus on rules governing information provision; specifically, on disclosure regulation. Our proposals, however, deviate from existing disclosure regulation and from other proposals for heightened disclosure regulation in an important way. Current disclosure regulation focuses on the disclosure of product-attribute information; in other words, information on the different features and price dimensions of cellular service. Our proposal emphasizes the disclosure of use pattern information—information on how the consumer will use the product. The proposed Cell Phone User Bill of Rights illustrates the current exclusive focus on product-attribute information. It requires comprehensive disclosure of fees and charges. However, a truly informed choice cannot be based on product attributes alone. To fully appreciate the benefits and costs of a cellular service contract, consumers must combine product-attribute information with use-pattern information. To assess the costs of overage fees, it is not enough to know the per-minute charges for minutes not included in the plan (as proposed in the bill); consumers must also know the probability that they will exceed the plan limit and by how much. Likewise, to assess the benefit of unlimited night and weekend calling, consum (p.241) ers must also know how many “night” and “weekend” minutes they will use as well as the precise contractual definition of “night” and “weekend.” Use-pattern information can be as important as product-attribute information. The disclosure regime should be redesigned to ensure that consumers have both categories of information. ### 2. Disclosing Use-Pattern Information Conventional wisdom assumes that sellers have better information on product attributes while buyers have better information about use patterns. If a buyer has better information about how she will use the product, then it makes no sense to require sellers to disclose use-pattern information. The best that sellers can do is to provide general statistical information on product use. The buyer, on the other hand, has specific information on how he or she, not the average consumer, will use the product—or so the conventional account goes. While the conventional wisdom is correct in many markets, it is not correct in the cellular service market. Carriers have valuable statistical use-pattern information that is not available to subscribers. More importantly, they have individualized use-pattern data, collected over the course of their relationships with subscribers. As suggested below, disclosing this information can empower consumers and facilitate the efficient functioning of the cellular service market. #### a. Average-Use Disclosures Carriers collect and analyze enormous amounts of use-pattern information. They know how the average subscriber will use his or her cell phone. This statistical use information is not limited to averages taken across the entire subscriber population. Carriers have, and can be required to disclose, average-use information for subgroups of consumers who are similar (in terms of demographic characteristics, product choices made, and so forth) to the consumer receiving the use-pattern disclosure. As the subgroup over which the averaging takes place becomes smaller, intra-group heterogeneity decreases, and the value of the average-use information to the individual consumer increases. However, excessively small subgroups may also be undesirable. Averaging over large numbers has the benefit of reducing randomness. Reducing the size of the subgroup reduces this benefit. The optimal size of a subgroup is the product of a tradeoff between the benefit of reducing heterogeneity and the benefit of reducing randomness. (p.242) One potentially beneficial average-use disclosure would target the misperception of use levels that underlies three-part tariffs by requiring carriers to disclose the average overage charges that consumers pay. Carriers could also be required to disclose the percentage of consumers who use, for example, 50 percent or less of their allotted minutes, or the percentage of consumers who would save money if they switched to a lower fixed-fee, lower-limit plan. Consumers’ underestimation of the cost of lock-in could be addressed by requiring carriers to provide information about the percentage of consumers who stop using their phones before the end of the lock-in period but continue paying for them. Carriers could also be required to disclose the percentage of consumers who broke the contract and paid the exit penalty.100 #### b. Individual-Use Disclosures Despite their potential benefits, average-use disclosures suffer from important shortcomings. Even when averaging across smaller subgroups of consumers, substantial heterogeneity remains. Heterogeneity limits the value of average-use information to any individual consumer. Moreover, heterogeneity allows optimistic consumers to further discount the value of average-use information. Most people think that their driving skills are above average, even though most people cannot be better than average (given a symmetrical distribution of ability about the mean). Similarly, optimistic consumers might all think that they will never exceed the plan limit, even when provided with information that the average consumer pays$50 a month in overage fees.

Fortunately, use-pattern disclosure in the cellular service market need not be limited to average-use information. The long-term relationship between carriers and consumers allows for the provision of individualized use-pattern information.101

Individual-use disclosure can reduce consumers’ misperception of use levels. Carriers already provide consumers with individualized information on overage charges. Arguably, this disclosure reduced consumers’ underestimation of use and contributed to the demand to eliminate overage fees—a (p.243) demand satisfied by unlimited calling plans. A parallel disclosure would help reduce the costs consumers incur due to overestimation of use. Carriers should be required to disclose the number of minutes used and the effective per-minute price, calculated as the monthly fixed fee divided by the number of minutes used.

Individual-use disclosure can also help consumers evaluate the costs and benefits of other plan features. Carriers could be required to disclose the number of night and weekend minutes used and the costs saved by the unlimited nights and weekends feature. They could also be required to disclose the number of minutes used in in-network calling and the associated savings. Similarly, Verizon, which offers unlimited calls to five numbers, could be required to disclose the number of minutes used calling these five numbers, along with the costs saved by this feature.

The proposed individual-use disclosures should be provided on the monthly bill and in aggregate form on a year-end summary to account for month-to-month variations. Lawmakers should also revisit another key feature of the proposed Cell Phone User Bill of Rights. This bill focuses on disclosures provided at the time of contracting, which makes perfect sense when carriers are disclosing product-attribute information. Individual-use information, on the other hand, is not available to carriers when a new subscriber signs up for service. Continuous disclosures throughout the life of the contract are equally important.

### 3. Combining Use-Pattern Information with Product-Attribute Information

We have focused on the disclosure of use-pattern information as opposed to product-attribute disclosures. But the more appealing proposals argue for total cost disclosures, which combine both. For example, the disclosure of effective per-minute prices combines product-attribute information (the monthly fixed fee) and use-pattern information (the number of minutes used).

Taking total cost disclosure one step further, carriers could be required to disclose a comprehensive TCO figure for their calling plans—the total amount paid, or to be paid, by a consumer, including overage charges, on a yearly basis or over the duration of a plan. For new subscribers, this TCO figure can be based on average-use information. For existing subscribers, who are considering whether to renew or switch plans or even switch carriers, the TCO figure can be based on individual-use information.

(p.244) TCO information for a single plan, specifically for the subscriber's current plan, may be insufficient. To effectively compare different plans, the subscriber needs TCO information on all plans. Carriers could be required to provide TCO information for their entire menu of plans or, at least, for several main offerings. Perhaps a better solution would be to require carriers to disclose only the plan with the lowest TCO for the prospective subscriber and for the existing subscriber whose use patterns have changed. For example, the monthly bill or yearly summary can include a notice if an alternative plan would have a lower TCO than the subscriber's current plan.102

TCO disclosures are simple and thus useful even to the imperfectly rational consumer. An alternative paradigm focuses on more complex and comprehensive disclosures for sophisticated intermediaries, and carriers rather than consumers. Specifically, carriers could be required to provide comprehensive use-pattern information in electronic form. Consumers would not use this information themselves. Rather, they would forward it to intermediaries that provide comparison-shopping services. Such intermediaries already exist. Companies like BillShrink and Validas promise to find the right plan for each consumer.103 But they currently do this based on minimal, usually self-reported, use-pattern information, which, as we have seen, is often inaccurate. If carriers were required to provide comprehensive use-pattern information in electronic form, intermediaries such as BillShrink or Validas could combine this information with the product-attribute information that they already have and find the carrier and plan with the lowest TCO for each individual consumer.

Comprehensive use-pattern information disclosed in electronic form can be helpful even without the intervention of intermediaries. The consumer could provide the information to competing carriers, soliciting individualized TCO figures from each. Consumers could then choose the carrier and plan with the lowest TCO, given their individual use patterns.

(p.245) To sum up, consumer choice should be guided by information about the total cost of the product. Conventional wisdom assumes that consumers have better information about their own use patterns and thus need only product-attribute disclosures to calculate total cost. We have seen that carriers may well have better use-pattern information, as well as better product-attribute information. Carriers can more easily combine the two categories of information into a total cost disclosure. Alternatively, carriers can provide comprehensive use-pattern information in electronic form, so that intermediaries, or competing sellers, can use it to calculate TCO figures.

### 4. Real-Time Disclosure

In addition to disclosures made at the time of purchase and after-the-fact disclosures in the monthly bill and/or in a year-end summary, individual-use information can and should be provided in real time. The challenge of keeping track of cumulative use has increased with the invention of multiple-limit plans. For example, plans with different limits for peak and off-peak minutes (as well as limits for messaging and data services) have increased the chance that consumers inadvertently exceed their plan limits. To help consumers avoid this, carriers should be required to notify their subscribers when they are about to exceed the plan limit. A consumer receiving such notification may well decide to cut the conversation short, switch to a land line, or postpone the conversation until off-peak hours.

In late 2010, the FCC proposed new regulations that mandated such real-time disclosures. Concerned about what it terms “bill shock”—unexpected increases in the monthly bill that come from high roaming fees or exceeding a monthly allotment of voice minutes, texts, or data consumption—the FCC, in its proposed rules, would require carriers to provide usage alerts in real time.104 The proposed regulations were put on hold when the carriers, through their trade group CTIA, agreed to voluntarily implement the real-time disclosures.105

### (p.246) 5. Mobile Disclosure

Traditional disclosure mandates require sellers to provide information printed on a piece of paper. Mobile technology opens the door to a variety of additional, more innovative disclosure methods. For example, carriers can provide information via voice messages, text messages, and even multimedia messages. These modes of disclosure may be more effective than the traditional paper disclosure.

### 6. Enhanced Disclosure in Action

Some carriers are already providing product-use information. “Usage analysis” functions are beginning to appear on some carriers’ secured websites. AT&T's capped data plans offer usage tracking and alert options. And there are other examples.106 Certain carriers even provide information that combines product-use and product-attribute information. For instance, AT&T offers Personal Plan Review, which lets you know if your plan fits your usage or if you should switch to a different AT&T plan. Moreover, there are smartphone applications that can track usage. Smartphones can even serve as virtual intermediaries, providing a recommendation on the best available plan in the market by combining the subscriber's individual use information with rate information available online. These developments should be applauded. They provide further evidence that market forces can work to the benefit of consumers. Still, as the FCC observed in the “bill shock” context, these voluntarily provided, enhanced disclosures vary widely between service providers and types of service and, in many cases, are insufficient. A regulatory nudge is probably required.107

# Conclusion

The cellular service market, boasting annual revenues exceeding $180 billion, is one of the largest and most important consumer markets in the United States. While cell phones provide obvious benefits to consumers, cellular service contracts, designed in response to consumer biases, hurt (p.247) consumers and reduce social welfare. Mistakes in plan choice, triggered by a key contractual design feature—the three-part tariff—cost consumers over$13 billion annually. Consumer welfare and market efficiency are further reduced by the ETF-enforced lock-in feature and by the sheer complexity of the cell phone contract, which also respond to the imperfect rationality of consumers. Since consumer mistakes often result from consumers’ misperceptions about their own future use patterns, disclosure mandates should require carriers to provide consumers with use-pattern information. This information should be combined with product-attribute information in TCO or total-annual-cost disclosures and made available in electronic, database form to facilitate the work of intermediaries.

## Notes:

(1.) Michael Grubb, “Selling to Overconfident Consumers”, Amer. Econ. Rev., 99 (2009), 1770, 1771 (note 2).

(2.) A carrier's relative efficiency depends on its costs of providing service and the quality of service that it offers. Thus, a carrier that provides the same quality of service at lower cost than another or a higher quality service at the same cost as another is a more efficient carrier.

(3.) The information presented here, and below, on cellular technology and on the history of the cellphone market is based, in large part, on: FCC, FCC 06–142, “Annual Report and Analysis of Competitive Market Conditions with Respect to Commercial Mobile Services, Eleventh Report” (2006) 21 F.C.C.R. 10947, 62 (hereinafter “FCC Eleventh Report”); Jonathan E. Nuechterlein and Philip J. Weiser, Digital Crossroads (MIT Press, 2005); Mischa Schwartz, Mobile Wireless Communications (Cambridge University Press, 2005); William Stallings, Wireless Communications and Networking (Prentice Hall, 2002); SRI International, “The Role of NSF's Support of Engineering in Enabling Technological Innovation, Final Report Phase II 94–97” (1998), 〈http://www.sri.com/policy/csted/reports/sandt/techin2/contents.html〉 (hereinafter “SRI-NSF Report”); Theodore Rappaport, Wireless Communications (Prentice Hall, 1996).

(4.) See Omnibus Budget Reconciliation Act of 1993, Pub. L. No. 103-66, Title VI, § 6002(a), 6002(b)(2), 197 Stat. 312, 387–93 (1993) (codified as 47 U.S.C. § 309(j) (2006)).

(5.) FCC, FCC 05–173, “Annual Report and Analysis of Competitive Market Conditions with Respect to Commercial Mobile Services, Tenth Report” (2005) 20 F.C.C.R. 15908, 15930 ¶ 58 (2005); Jeremy T. Fox, “Consolidation in the Wireless Phone Industry” (2005) 3, 7, 9 (Net Inst. Working Paper No. 05-13), available at 〈http://www.netinst.org/Fox2005.pdf〉.

(6.) FCC, FCC 11–103, “Annual Report and Analysis of Competitive Market Conditions with Respect to Commercial Mobile Services, Fifteenth Report” (2001) 8, 9, 207, available at 〈http://wireless.fcc.gov/index.htm?job=cmrs_reports〉 (hereinafter “FCC Fifteenth Report”; FCC, DA 09-54, “Annual Report and Analysis of Competitive Market Conditions with Respect to Commercial Mobile Services, Thirteenth Report” (2009) 24 F.C.C.R. 6185, at 6279–80 ¶ 197, 6301 ¶ 230 (hereinafter “FCC Thirteenth Report”); SRI-NSF Report, above note 3, at 94.

(7.) Verizon, “Verizon Wireless—Selected Financial Results” (July 22, 2011), 〈http://www22.verizon.com/idc/groups/public/documents/adacct/2011_2q_fs_pdf.pdf〉; AT&T, “Investor Briefing 2nd Quarter 2011,” at 4 (July 21, 2011), 〈http://www.att.com/Investor/Financial/Earning_Info/docs/2Q_11_IB_FINAL.pdf〉; News Release, “Sprint, Sprint Nextel Reports Second Quarter 2011 Results” (July 28, 2011), 〈http://newsroom.sprint.com/article_display.cfm?article_id=1990〉; Financial Release, “T-Mobile USA,” T-Mobile USA Reports Second Quarter 2011 Results” Bibliog cites July 28, 2011, 〈http://www.t-mobile.com/company/InvestorRelations.aspx?tp=Abt_Tab_InvestorRelationsandViewArchive=Yes〉 (follow “T-Mobile USA Reports Second Quarter 2011 Results” hyperlink).

(8.) George Gilder, “The Wireless Wars,” Wall Street Journal, April 13, 2007, at A13 (stating that Verizon's mobile phones generated $804 million in profits, whereas its wired phones generated$393 million in profits).

(9.) “Auction of Advanced Wireless Services Licenses Closes: Winning Bidders Announced for Auction No. 66” (2006) 21 F.C.C.R. 10521.

(10. ) “Auction of 700 MHz Band Licenses Closes: Winning Bidders Announced for Auction 73” (2008) 23 F.C.C.R. 4572.

(11.) See Robert Jensen, “The Digital Provide: Information (Technology), Market Performance, and Welfare in the South Indian Fisheries Sector”, Q.J. Econ., 122 (2007), 879, 881–3 (describing how the introduction of cell phones revolutionized the fishing industry in Kerala, leading to dramatic reductions in price dispersion, the complete elimination of waste (previously 5–8 percent of the daily catch), an 8 percent average increase in fishermen's profits, a 4 percent decline in consumer prices, and a 6 percent increase in consumer surplus); Leonard Waverman, Meloria Meschi, and Melvyn Fuss, “The Impact of Telecoms on Economic Growth in Developing Countries,” in The Vodafone Policy Paper Series no. 3, Africa: The Impact of Mobile Phones (2005), 〈http://www.vodafone.com/etc/medialib/attachments/cr_downloads.Par.78351.File.tmp/GPP_SIM_paper_3.pdf〉 (“We find that mobile telephony has a positive and significant impact on economic growth, and this impact may be twice as large in developing countries as compared to developed countries.”); Nuechterlein and Weiser, above note 3, at 268.

(12.) Information presented here, and below, on the cellular service market is based, in large part, on FCC Fifteenth Report, above note 6.

(13.) Formally, the HHI is given by , where si is the fractional market share of firm i, and I is the number of firms in the market. Thus a monopolistic market has an HHI of 10,000, a market that is equally divided between two firms has an HHI of 5,000, a market that is equally divided between three firms has an HHI of 3,333.33, a market that is equally divided between four firms has an HHI of 2,500, etc.

(14.) FCC Thirteenth Report, above note 6, at 6212 ¶ 45 n. 87.

(15.) Fox, above note 5, at 15–17. Moreover, this figure excludes data on Nextel, and so the Sprint Nextel merger does not contribute to the high HHI, suggesting that this figure may underestimate the true concentration. Id. at 16 n. 11.

(16.) See Patrick Bajari, Jeremy T. Fox, and Stephen Ryan, “Evaluating Wireless Carrier Consolidation Using Semiparametric Demand Estimation” (2006) 5 (Nat’l Bureau of Econ. Research, Working Paper No. 12425), available at 〈http://www.nber.org/papers/w12425〉; Fox, above note 5, at 10.

(17.) Fox, above note 5, at 12. Multi-market contact was an important factor in explaining above-competitive prices in the early mobile telecommunications industry. See Philip M. Parker and Lars-Hendrik Röller, “Collusive Conduct in Duopolies: Multi-Market Contact and Cross-Ownership in the Mobile Telephone Industry”, Rand J. Econ., 29 (1997), 304, 320. There were also significant cross-ownership effects, i.e., if operators co-own an operating license elsewhere, they tend to collude more. Id.

(18.) On the FCC actions—making more spectrum available, eliminating the structural duopolies and abolishing the spectrum cap—see FCC Thirteenth Report, above note 6, at 6220 ¶¶ 65–6. On interconnection and roaming obligations—see 47 U.S.C. § 251(a)(1) (2006); “Reexamination of Roaming Obligations of Commercial Mobile Radio Service Providers, Final Rule” (2007) 72 Fed. Reg. 50064, 50064–65 (hereinafter “Reexamination of Roaming Obligations”). On the role of the secondary market—see FCC Thirteenth Report, above note 6, at 6220 ¶ 67.

(19.) Estimated advertising spending for wireless telephone services totaled between $4.1 billion and approximately$5.1 billion in 2007, $3.7 billion in 2008, and$3.4 billion in 2009. See FCC Thirteenth Report, above note 6, at 6261 ¶ 158 (for the 2007 estimates); FCC Fifteenth Report, above note 6, at 50 (for the 2008 and 2009 estimates).

(20.) Tim Wu, “Wireless Net Neutrality: Cellular Carterfone and Consumer Choice in Mobile Broadband” (2007) 1 New Am. Found. Wireless Future Program Working Paper No. 17, available at 〈http://www.newamerica.net/files/WorkingPaper17_WirelessNetNeutrality_Wu.pdf〉; see also Spencer E. Ante, “Verizon Embraces Google's Android”, Business Week, December 3, 2007, 〈http://www.businessweek.com/technology/content/dec2007/tc2007123_429930.htm?campaign_id=yhoo〉 (“Verizon Wireless has created the most profitable U.S. cellular business by tightly restricting the devices and applications allowed to run on its network.”).

(21.) See 37 C.F.R. § 201.40(b)(5) (2008). Carriers are embracing the new open-access business model. See Ante, above note 20. (“But over the past year, [Verizon's] leadership came to conclude that it was time for a radical shift. Such a move, they reckoned, might help Verizon Wireless keep growing while holding down costs.”) Sprint Nextel and T-Mobile also support the shift to an open-handset environment, as members of the Google-led “Open Handset Alliance.” Id.; see also Amol Sharma and Dionne Searcey, “Verizon to Open Cell Network to Others’ Phones,” Wall Street Journal, November 28, 2007, at B1.

(22.) FCC Eleventh Report, above note 13, at 11012 ¶ 146. Wireless local number portability began on November 24, 2003. In re Telephone Number Portability, 19 F.C.C.R. 875, 876 (2) (order). The FCC reports that from December 2003 to December 2007, 49.93 million consumers took advantage of the right to retain their phone number while switching from one wireless carrier to another. FCC Thirteenth Report, above note 6, at 6272 ¶ 183.

(23.) FCC Fifteenth Report, above note 6, at 154–5. A “churn rate” is the rate at which users cancel their cellular service in a given period of time.

(24.) FCC Fifteenth Report, above note 6, at 12–13. On the other hand, there is substantial similarity between the pricing schemes offered by the major carriers. See below Part II. This price matching may reflect tacit collusion among the major carriers. Cf. Meghan R. Busse, “Multimarket Contact and Price Coordination in the Cellular Telephone Industry” (2008–9) 9 J. Econ. and Mgmt. Strategy, 9 (2008-9) 287, 313–16.

(25.) FCC Thirteenth Report, above note 6, at 6262–3. Carriers’ marketing campaigns emphasize their “superior network coverage, reliability, and voice quality.” Id.

(26.) FCC Eleventh Report, above note 3, at 10987 ¶ 90. Yet, since this is an industry characterized by high network costs, this phase of apparently intense competition may be nothing more than a price war designed to squeeze out smaller carriers that will ultimately result in an increase in the market power of the remaining large carriers and an attendant rise in prices.

(27.) Press Release, comScore, “comScore Reports July 2011 U.S. Mobile Subscriber Market Share” (August 30, 2011), 〈http://www.comscore.com/Press_Events/Press_Releases/2011/8/comScore_Reports_July_2011_U.S._Mobile_Subscriber_Market_Share〉. The market shares of the leading handset firms are quite different outside the United States. Nokia is the global market leader, with 33.3 percent of the global market in 2010, followed by Samsung with 20.6 percent, LG with 8.6 percent, RIM with 3.6 percent, and Apple with 3.5 percent. Press Release, “Strategy Analytics, Global Handset Shipments Reach 400 Million Units in Q4 2010” (January 28, 2011), 〈http://www.strategyanalytics.com/default.aspx?mod=pressreleasevieweranda0=5001〉.

(28.) Marguerite Reardon, “Unlocking the Unlocked Cell Phone Market,” CNET News, July 2, 2009, 〈http://news.cnet.com/8301-1035_3-10277723-94.html〉.

(29.) See Wu, above note 20, at 10–13; Reardon, above note 28; Ante, above note 20 (“Verizon Wireless has created the most profitable U.S. cellular business by tightly restricting the devices and applications allowed to run on its network.”).

(30.) On the increasing power of handset manufacturers—see Rita Chang, “Proof That Handset Brands Help Sell Wireless Plans,” RCR Wireless, October 28, 2008, 〈http://www.rcrwireless.com/article/20081028/WIRELESS/810289995/1081/proof-that-handset-brands-help-sell-wireless-plans#〉; Press Release, Nielsen, “In US, Smartphones Now Majority of New Cellphone Purchases” (June 30, 2011), 〈http://blog.nielsen.com/nielsenwire/?p=28237〉. On power struggles between carriers and handset manufacturers, as well as with application developers—see generally Jessica E. Vascellaro, “Air War: A Fight over What You Can Do on a Cellphone,” Wall Street Journal, June 14, 2007, at A1; see also Miguel Helft and Stephen Labaton, “Google Pushes for Rules to Aid Wireless Plans,” New York Times, July 21, 2007, at A1.

(31.) See George S. Ford, Thomas M. Koutsky, and Lawrence J. Spiwak, “Wireless Net Neutrality: From Carterfone to Cable Boxes”, Phoenix Ctr. Pol’y Bull. 17, (April 2, 2007), at 2, 〈http://phoenix-center.org/PolicyBulletin/PCPB21Final.pdf〉.

(32.) Editorial, “A Half-Win for Cellphone Users,” New York Times, August 6, 2007, at A18; In re Service Rules for the 698–746, 747–762, and 777–792 MHz Bands, 22 F.C.C.R. 15289, 15367, 15370–71 (2007) (second report and order) (hereinafter “Service Rules Second Report and Order”).

(33.) See Ante, above note 20; see also Sharma and Searcy, above note 21.

(34.) FCC Eleventh Report, above note 3, at 11007 ¶ 136–37.

(35.) FCC Fifteenth Report, above note 6, at 13–14.

(36.) Wu, above note 20, at 13–14, 22–5.

(37.) Vascellaro, above note 30.

(38.) One feature that we will not address is the definition of call types for which the subscriber is charged (or that count toward the plan limit). Specifically, while in most countries subscribers are charged only for outgoing calls, in the U.S. subscribers are also charged for incoming calls. This feature of the U.S. cellular service market seems to fit nicely within the general behavioral theory, as subscribers probably find it even more difficult to accurately estimate the number/length of incoming calls along with outgoing calls than outgoing calls alone.

(39.) The description of products and prices provided in Part II, and in other Parts, is largely based on information available through carriers’ websites focusing on services available in the New York area. See AT&T, “Cell Phones, Cell Phone Plans, and Wireless Accessories—from AT&T”, 〈http://www.att.com/shop/wireless/#fbid=n6la6zd-YLG〉 (last visited September 29, 2011); Sprint, “Cell Phones, Mobile Phones, and Wireless Calling Plans from Sprint,” 〈http://www.sprint.com〉 (last visited September 29, 2011); T-Mobile, “Cell Phones | 4G Cell Phone Plans | Android Tablet PCs | T-Mobile,” 〈http://www.t-mobile.com/〉 (last visited September 29, 2011); Verizon Wireless, “Cell Phones—Smartphones: Cell Phone Service, Accessories—Verizon Wireless,” 〈http://www.verizonwireless.com/b2c/index.html〉 (last visited September 29, 2011). It should be noted that some variation exists between online and offline (retail store) offerings and between different geographical markets across the U.S. This variation is mentioned explicitly only when it is relevant to the analysis.

(40.) See Elizabeth Douglass, “The Cutting Edge Special Report: Wireless Communications; ‘Prepaid’ Idea is Catching On in U.S. Market,” L.A. Times, March 15, 1999, at C1 (discussing trend away from long-distance and roaming charges).

(41.) Roger O. Crockett, “The Last Monopolist,” Business Week, April 12, 1999, at 55; Dan Meyer, “Coverage Problems Trigger Headaches for Carriers,” RCR Wireless News, July 9, 2001, at 16.

(42.) Andrew M. Odlyzko, “The Many Paradoxes of Broadband” First Monday, September 1, 2003, 〈http://firstmonday.org/htbin/cgiwrap/bin/ojs/index.php/fm/article/view/1072/992〉.

(43.) See Rebecca Blumenstein, “The Business-Package Plan: AT&T Sees Wireless as the Key to its Broader Strategy of Bundling Its Services,” Wall Street Journal, September 20, 1999 at R26; see also Peter Elstrom, “Wireless with All the Trimmings,” Business Week, November 16, 1998.

(44.) Peter Elstrom, “Mike Armstrong's Strong Showing,” Business Week, January 25, 1999, at 94; Elstrom, “Wireless with All the Trimmings,” above note 43.

(45.) See Amol Sharma and Roger Cheng, “iPhone Costs Prove a Drag for AT&T,” Wall Street Journal, October 23, 2008, at B4 (“The company said $900 million in customer-acquisition costs related to the iPhone shaved 10 cents off its earnings,” but “AT&T executives said the investment will pay off because iPhone users are lucrative in the long-term, spending about$95 a month on average, or about 1.6 times the amount other customers do.”); Cliff Edwards and Roger O. Crockett, “New Music Phones—Without the i,” Business Week, April 16, 2007, at 39.

(46.) Press Release, J.D. Power and Associates, “The Right Blend of Design and Technology is Critical to Creating an Exceptional User Experience with Smartphones and Traditional Mobile Devices” 2 (September 8, 2011), 〈http://businesscenter.jdpower.com/JDPAContent/CorpComm/News/content/Releases/pdf/2011146-whs2.pdf〉.

(47.) When no-contract plans are offered, phone subsidies disappear. For example, a customer with no contract would be required to pay an additional $400 beyond the contract price for the same iPhone. “AT&T Plans to Offer No-Contract iPhone,” Wall St. J., July 2, 2008, at B5. (48.) See Verizon Wireless, Customer Agreement, 〈http://www.verizonwireless.com/b2c/index.html〉 (follow “Customer Agreement” hyperlink at the bottom of the page); AT&T, Plan Terms, 〈http://www.wireless.att.com/cell-phone-service/legal/plan-terms.jsp#gsm〉; Press Release, Sprint Nextel Corp., “Sprint Launches One of the Industry's Most Customer-Friendly Policies on Pro-Rated Early Termination Fees” (October 31, 2008), 〈http://newsroom.sprint.com/article_display.cfm?article_id=771〉; T-Mobile, T-Mobile Terms and Conditions, 〈http://www.t-mobile.com/templates/popup.aspx?passet=ftr_ftr_termsandconditions〉. (49.) See BillShrink.com, Cell Phone Plans, Compare Best Cellular Service Carrier Deals on BillShrink, 〈http://www.billshrink.com/cell-phones/plans.html〉 (last visited September 29, 2011). (50.) We briefly mention two additional dimensions: (1) The directionality of the calls that consume allotted minutes, and (2) the one-time activation charge. Along dimension (1), allotted minutes are typically used up on both outgoing calls and incoming calls. As for (2), AT&T and Sprint charge a$36 activation fee while Verizon and T-Mobile charge $35. (51.) For an example of pay-as-you-go pricing—see Verizon Wireless, “Cell Phones—Smartphones: Cell Phone Service, Accessories—Verizon Wireless,” above note 39 (as of September 29, 2011, Verizon charged$0.20 per text message and $0.25 per multimedia message). Verizon also offers fixed-quantity monthly packages (as of September 29, 2011, Verizon charged$5.00 per month for 250 text or multimedia messages and $10.00 for 500 messages). For an example of unlimited-quantity monthly packages—see AT&T, “Cell Phones, Cell Phone Plans, and Wireless Accessories—from AT&T,” above note 39 (AT&T charged$20.00 per month for unlimited messaging as of September 29, 2011.) Unlimited messaging and even data are covered by the monthly fee component of the basic three-part tariff in some premium plans. Id.

(52.) None of the four major operators charges for roaming in its postpaid pricing plans.

(53.) The analysis of the three-part tariff, in Section A, relies heavily on Grubb, above note 1. The behavioral explanation described in Section A is also closely related to the one developed in K. Eliaz and R. Spiegler, “Contracting with Diversely Naive Agents”, Rev. Econ. Stud., 73 (2006), 689; K. Eliaz and R. Spiegler, “Consumer Optimism and Price Discrimination”, Theor. Econ., 3 (2008), 459.

(54.) Formally, the cumulative distribution function (“c.d.f.”) describing the priors over the demand parameter of the predictable type must cross that of the variable type once from below. Grubb, above note 1.

(55.) The data aren’t completely consistent with a strict FOSD ranking, however, since Plan 2's cumulative distribution function crosses Plan 3's towards the bottom of the graph. But it is likely that the Plan 2 data are less reliable.

(56.) Grubb's analysis of a different dataset yields the same conclusion. Grubb, above note 1.

(57.) See Matthew Rabin, “Note, Risk Aversion and Expected-Utility Theory: A Calibration Theorem”, Econometrica, 68 (2000), 1281. However, they may be consistent with certain behavioral accounts of risk aversion. See id., at 1282 n. 3.

(58.) Grubb, above note 1.

(59.) Three-part tariffs can arise also when consumers are overconfident about their ability to predict their future use. Namely, when the same consumers exhibit a tendency to both over- and underestimate future use. See Grubb, above note 1.

(60.) Price calculations add the fixed monthly fee to the number of minutes multiplied by the per-minute price. Surplus calculations take the number of minutes multiplied by the difference between the per-minute benefit and the per-minute price and subtract the fixed monthly fee.

(61.) The carrier's costs include a fixed cost of $10 and an expected variable cost of$0.1 per minute multiplied by the expected number of minutes—100 minutes for light users (50 percent of users) and 300 minutes for heavy users (50 percent of users). The total cost is $30. The carrier gets$20 from overage charges that the heavy users pay on their last 100 minutes. The remaining $10 is collected as a fixed monthly fee. (62.) In a more general model, overage charges would be underestimated, but not completely invisible. (63.) The database was provided by the Center for Customer Relationship Management at Duke University. The description of the data in the text is based on the description provided by the Center. See The Center for Customer Relationship Management, Telecom Dataset, available at 〈http://www.fuqua-europe.duke.edu/centers/ccrm/index.html#data〉; see also Raghuram Iyengar, Asim Ansari, and Sunil Gupta, “A Model of Consumer Learning of Consumer Service Quality and Usage”, J. Mktg. Res., 44 (2007), 529, 535–7. It is not entirely clear from the data that all four plans were offered at all dates in all markets. We acknowledge this limitation of the data and qualify our results accordingly. Our empirical strategy builds on Grubb, above note 1, who tested a related behavioral explanation, the overconfidence theory, using a different dataset. (64.) This analysis assumes risk neutrality. The sums involved are small enough that this assumption seems reasonable. (65.) The magnitude of the mistake (or cost saving) is calculated as follows (this calculation is performed for each cell in Table 4.3b): We calculate the actual total price paid by a consumer under her chosen plan over the period that she stayed with the plan. We then calculate the hypothetical total price that this consumer would have paid had she chosen the best plan given her actual usage. The difference between these actual and hypothetical total prices is the magnitude of the mistake and, when reporting dollar values, we transform this number into annual figures (by dividing the difference by the actual number of months under the plan and then multiplying by 12 months). The mistakes thus calculated are then averaged out across all consumers in the specific Table 4.3b cell, e.g., across all consumers who chose Plan 3 but should have chosen Plan 1. (66.) These conclusions are tentative, since prepaid plans may differ from postpaid plans on other dimensions. In particular, while the service quality offered by prepaid plans is improving, in the period when the data were collected there was still a non-negligible difference in quality between prepaid and postpaid plans. (67.) This analysis is performed in Oren Bar-Gill and Rebecca Stone, “Pricing Misperceptions: Explaining Pricing Structure in the Cell Phone Service Market” forthcoming in J. Empirical Legal Stud., Vol. 9. Evidence of “bill shock”—a sudden increase in the monthly bill that is not caused by a change in service plan—also suggests that many consumers are making ex ante mistakes. A 2010 FCC survey study found that one in six mobile users have experienced “bill shock,” and that for a sizeable proportion of users the magnitude of the “shock” was substantial—more than a third reported an increase of at least$50, and 23 percent reported an increase of $100 or more. See FCC, News Release: “FCC Survey Confirms Consumers Experience Mobile Bill Shock and Confusion about Early Termination Fees,” May 26, 2010, available at 〈http://www.fcc.gov/document/fcc-survey-confirms-consumers-experience-mobile-bill-shock-and-confusion-about-early-termin〉; last visited November 17, 2011. (68.) Wu, above note 20, at 9. For an example of the carrier-imposed difficulty customers face in determining their unused plan-minute allowances—see Sherrie Nachman, “Cranky Consumer: How to Check Up on Your Cell Phone Minutes,” Wall Street Journal, June 18, 2002, at D2. (69.) Lauren Tara Lacapra, “Breaking Free of a Cellular Contract—New Web Sites Help Customers Swap or Resell Phone Service; Avoiding$175 Termination Fee,” Wall Street Journal, November 30, 2006, at D1 (“It costs a cell phone company approximately $350 to$400 to acquire a new customer, according to Phil Doriot, a partner in the consulting firm CFI Group, who has studied company performance and customer satisfaction for major cellular service providers.”);  Jane Spencer, “What Part of ‘Cancel’ Don’t You Understand?—Regulators Crack Down on Internet Providers, Phone Companies That Make It Hard to Quit,” Wall Street Journal, November 12, 2003, at D1 (noting that customer acquisition costs are approximately “$339 per new customer, according to Yankee Group, a technology research firm”). (70.) Jared Sandberg, “A Piece of the Business,” Wall Street Journal, September 11, 1997, at R22. (71.) The fact that termination fees were initially structured such that consumers paid the same fee regardless of when they terminated the contract raises doubts about the argument that ETFs were necessary to cover the cost of the free or subsidized phones, either by inducing consumers to stay on and pay the monthly subscription fees or by replacing the subscription fees of consumers who leave. (72.) See Thomas J. Tauke, Executive Vice President, Verizon, “Testimony at FCC Early Termination Hearing” 2 (June, 12, 2008) (hereinafter Verizon Testimony), available at 〈http://www.fcc.gov/realaudio/presentations/2008/061208/tauke.pdf〉; see also CTIA, “Early Termination Fees Equal Lower Consumer Rates”, CITA.org, April 2006, 〈http://files.ctia.org/pdf/PositionPaper_CTIA_ETF_04_06.pdf〉. (73.) The ETF effectively deters switching. See Lacapra, above note 69 (according to a July 2005 survey by the U.S. PIRG Education Fund, “[r]oughly 47 percent of cell customers would switch or consider switching cellphone companies if early-termination fees were abolished,” but “because of the fee, only 3 percent of customers go ahead with terminating the contract”). (74.) Caroline E. Mayer, “Griping about Cellular Bills; Differences from ‘Regular’ Phones Take New Users by Surprise,” Washington Post, February 28, 2001, at G17; see also Fawn Johnson, “FCC Head Seeks Rules on Cell-Termination Fees,” Wall Street Journal, June 13, 2008, at B7 (“Wireless carriers argue that the termination fees are used to subsidize the cost of cellphones to customers. People who sign up for one- or two-year contracts receive discounts on phones and their monthly wireless rates.”); CTIA, above note 72, at 1 (arguing that prohibiting carriers from charging ETFs will cause prices for wireless services to increase). (75.) See Verizon Testimony, above note 72, at 1. (76.) See Paul Wagenseil, “That ‘Cheaper’ iPhone Will Cost You More,” FoxNews.com, June 11, 2008, 〈http://www.foxnews.com/story/0,2933,365347,00.html〉. (77.) AT&T Plans to Offer No-Contract iPhone, above note 47. (78.) See Ante, above note 20. (79.) A market for “comparison shopping services” is emerging, with vendors such as BillShrink and Validas offering to find the best product/plan for any consumer who is willing to pay a fee. See below note 102 and accompanying text. The availability of comparison-shopping services reduces the cost of comparison-shopping and increases the optimal level of complexity in a rational-choice model. However, it seems that most cell phone users do not avail themselves of the services offered by BillShrink and Validas. The emergence of a market for “comparison shopping services” suggests that complexity makes it difficult for consumers to comparison-shop by themselves. But since the majority of consumers do not seek help from professional comparison-shoppers and thus do not benefit from the high level of complexity, the rational choice explanation for complexity is less convincing. (80.) Andrea Petersen and Nicole Harris, “Hard Cell: Chaos, Confusion and Perks Bedevil Wireless Users,” Wall Street Journal, April 17, 2002, at A1. (81.) Lacapra, above note 69. (82.) Joseph Farrell and Paul Klemperer, “Coordination and Lock-In: Competition with Switching Costs and Network Effects” in Armstong, M. and Porter, R. (eds.), Handbook of Industrial Organization, Vol. 3, pp. 1967, 2005 (North-Holland, 2007). (83.) Roger Cheng, “Business Technology: Virgin Mobile to Join Flat Rate Phones Frenzy,” Wall Street Journal, June 24, 2008, at B4. (84.) See Jeff Blyskal, “Mostly Talk: New Unlimited Cell Plans Won’t Pay for Most,” ConsumerReports.org, February 26, 2008, 〈http://blogs.consumerreports.org/electronics/2008/02/mostly-talk-new.html〉. (85.) Unused minutes do not roll over forever. They expire after a year. (86.) In this example, the rational non-AT&T customer will switch to a 900-minute plan and pay an additional$20 per month because this charge is smaller than the average overage paid in the seemingly cheaper plan: $45/2 months =$22.50.

(87.) FCC, FCC 08–28, “Annual Report and Analysis of Competitive Market Conditions with Respect to Commercial Mobile Services, Twelfth Report” 2297–98 ¶¶ 116–18 (2008), available at 〈http://wireless.fcc.gov/index.htm?job=cmrs_reports〉.

(88.) Opinion Research Corporation, “Prepaid Phones in the U.S.: Myths, Lack of Consumer Knowledge Blocking Wider Use” 4, 10 (2008), 〈http://www.newmillenniumresearch.org/archive/120408_prepaid_myths_survey_report.pdf〉.

(89.) FCC Fifteenth Report, above note 6, at 67;  Jenna Wortham, “Cellphones Without Strings,” New York Times, February 20, 2009, at B1 (describing the growing attraction of prepaid plans and citing Pali Research, an investment advisory firm, reporting that, in 2008, sales of prepaid plans grew 13 percent in North America, nearly three times faster than traditional postpaid plans).

(90.) See Spencer E. Ante, “The Call for a Wireless Bill of Rights,” Business Week, March 20, 2008, at 80, available at 〈http://www.businessweek.com/magazine/content/08_13/b4077080431634.htm?campaign_id=rss_tech〉 (noting that, according to the Better Business Bureau, for each of the past three years, the wireless sector has received more complaints than any other industry). In the fourth quarter of 2011, the FCC received 21,076 complaints about wireless telecommunications. FCC, Quarterly Report on Informal Consumer Inquiries and Complaints for Fourth Quarter of Calendar Year 2010 August 15, 2011, available at 〈http://transition.fcc.gov/Daily_Releases/Daily_Business/2011/db0815/DOC-309057A1.pdf〉.

(91.) Prohibitions against unfair or deceptive advertising should also be mentioned. On one important dimension, early termination fees, the law has moved beyond the regulation of information provided by carriers. See Oren Bar-Gill and Rebecca Stone, “Mobile Misperceptions” Harv. J.L. & Tech. 23 (2009) 49.

(92.) 47 C.F.R. § 64.2401(a)(1), (d) (2008).

(93.) 47 C.F.R. § 64.2401(b) (2008).

(94.) In re Truth-in-Billing and Billing Format, Second Report and Order, Declaratory Ruling, and Second Further Notice of Proposed Rulemaking, FCC CC Docket No. 98–170, 20 F.C.C.R. 6448, 6450 (2005) (hereinafter Truth-in-Billing 2005).

(95.) Id., at 6462.

(96.) Cell Phone User Bill of Rights, S. 1216, 108th Cong. (2003). A similar bill, the Wireless Consumer Protection and Community Broadband Empowerment Act, was proposed more recently by Representative Edward Markey. See Press Release, Office of Rep. Edward Markey, “Markey Holds Hearings on Draft Bill to Address Wireless Customer Protections,” February 27, 2008, 〈http://markey.house.gov/index.php?option=com_contentandtask=viewandid=3281andItemid=241〉.

(97.) See Press Release, California Public Utilities Commission, PUC Sets Protection Rules for Consumers through Telecommunications Bill of Rights, May 27, 2, 〈http://docs.cpuc.ca.gov/published/NEWS_RELEASE/36910.htm〉; Robert W. Hahn et al., “The Economics of ‘Wireless Net Neutrality’” J. Competition L. and Econ. 3 (2007), 399, 413.

(98.) “California Suspends Wireless Bill of Rights,” ConsumerAffairs.com, January 28, 2005, 〈http://www.consumeraffairs.com/news04/2005/cpuc_wireless.html〉.

(99.) See Ante, above note 20.

(100.) Both of these disclosures are incomplete measures of the cost of lock-in since they do not capture consumers who continue using their phones only because they are locked in.

(101.) Of course, consumers have access to the same use-pattern information. But while providers save the information and analyze it, consumers tend not to notice it and even if they do notice it, they tend to forget it.

(102.) Utility companies in Germany have voluntarily adopted an even more pro-consumer policy. At the end of the year they retroactively match each consumer to the service plan under which the consumer pays the lowest total price given her use over the past year. See Ian Ayres and Barry Nalebuff, “In Praise of Honest Pricing”, M.I.T. Sloan Mgmt. Rev. 45 (2003), 24, 27. A similar idea is already being applied by cell phone companies in other countries. See, e.g., Orange.fr, “Forfait Ajustable Pro,” 〈http://sites.orange.fr/boutique/files/html/pe_packpro_forfait_ajustable.html〉 (Orange in France offers to charge the subscriber at the end of the month according to the plan that best fits the subscriber's usage during that month).

(103.) “What is BillShrink?,” 〈http://www.billshrink.com/about/〉 (last visited September 20, 2011); “About Validas,” 〈http://www.validas.com/about.aspx〉 (last visited September 20, 2011).

(104.) See FCC, News Release: “FCC Proposes Rules to Help Mobile Phone Users Avoid ‘Bill Shock’,” October 14, 2010, available at 〈http://www.fcc.gov/rulemaking/10-207〉; (last visited: September 21, 2011). Similar rules already exist in Europe. While some U.S. carriers voluntarily provide certain usage alerts, the FCC found that “[t]he tools in place to eliminate bill shock vary widely between service providers and type of service, and can be difficult to find. Most of the alerts that are offered do not cover all services or are not sent before the overage charges are incurred.” Id.

(105.) See FCC Chairman Julius Genachowski, Remarks at Bill Shock Event, Brookings Institution, Washington, DC, October 17, 2011, available at 〈http://www.fcc.gov/document/chairman-genachowski-remarks-bill-shock-event〉; (last visited): November 17, 2011).

(106.) See Amy Schatz and Sara Silver, “A Plan to Ease the Shock of Cellphone Bills,” Wall Street Journal, May 12, 2010; David Pogue, “AT&T's Capped Data Plan Could Save You Money,” New York Times, June 3, 2010.

(107.) Indeed, in the “bill shock” context, such a regulatory nudge, or the threat of a regulatory nudge, spurred the industry into action. See above Sec. VI.B.4.