Computer reservations systems: competition misunderstood.
The deregulation of the U.S. airline industry has had a major impact on both the structure of air travel markets and the pricing and availability of airline services. Although most observers agree that the removal of price and entry regulations has yielded significant consumer benefits,(1) the transition to a more competitive environment has not followed the path envisioned by the leading proponents of deregulation.(2)
One of the most cotroversial aspects of deregulated airline markets has resulted from the growth in importance of computer reservation systems (CRSs). CRSs have automated the travel agency business, making it possible for travel agency customers to receive the latest flight information and conveniently purchase tickets on numerous airlines. Private litigants,(3) government agencies,(4) and several commentators(5) have alleged that CRS vendors engage in a variety of anticompetitive practices. In response to these complaints, the Civil Aeronautics Board (CAB) in November of 1984 imposed regulations with the stated purpose of enhancing airline competition. These rules have not satisfied many of the complaining parties. Related litigation has continued, as have the requests for additional government intervention.(6)
CRS vendors have faced a long list of charges of anticompetitive conduct. This article will examine what appears to be the two most important of these allegations, the allegations that seem to be at the base of all the other claims. First, the article will examine whether because of the structure of the CRS industry each major CRS vendor, by itself, has market power. Second, the article will analyze the controversial issue of display preference ("bias") on CRS display screens. A close examination of the relevant issues indicates that the relevant anticompetitive allegations are not strongly supported by economic analysis. Rather, they appear to be based on significant misunderstandings of how competition works in the CRS industry. The resulting confusion has led to incorrect policy in this area.
II. The nature of CRS services
A. Airline marketing and sales options
Airline travel between two particular cities is generally provided by several competing airlines. The services offered by each carrier may exhibit price and service characteristics that vary significantly over time. In order to make an informed purchase decision, a prospective traveler desires access to timely and accurate information concerning the available options. The traveler will also desire the ability to reserve space and/or purchase tickets on his preferred flight. Thus, not only must an airline offer services that customers desire, but it must also market these services so that customers know the products are available and can purchase them conveniently.
Airlines use two general methods to disseminate price and schedule information and facilitate retail ticket purchases. First, each airline has an internal reservation system (IRS) that customers can telephone directly for ticket information and ticket purchases. Second, most airlines make information available to independent travel agents through the use of one or more CRSs. These systems also allow travel agents to process ticket purchases. (The individual traveler can thus typically purchase airline tickets either directly from the airline (using an IRS) or from a travel agency (using a CRS).
Each airline will thus attempt to develop an effective marketing strategy by choosing an appropriate mix of the available marketing techniques. Product information can be disseminated both through direct advertising and through travel agents, who are paid commissions to market the products offered by competing airlines. Similarly, ticket purchases can be processed both internally and through one or more of the competing CRSs. The appropriate mix of advertising, navel agent commission, and CRS marketing efforts can be expected to vary across airlines, as marketing expenses do across other firms and industries in the economy.
Airlines may also choose (absent government regulations) different levels of CRS participation. Some airlines may wish to pay CRS vendors more (less) money in order to improve (decrease) their flights' placement on screen displays. Other airlines may desire to limit their participation to a subset of the available systems or to avoid participation entirely. Some airlines may also desire to pay additional charges for "direct access," a service that gives CRS vendors and their travel agency subscribers more up-to-date access on carriers' flight offerings.(7) While most airlines are currently listed on each of the domestic CRSs, Southwest Airlines does not participate in any such system.(8) Similarly, People Express Airlines was not listed on any CRS for the first 4 years of its existence in the early 1980's.
B. A brief history and description of IRS and CRS
The IRS operated by any airline consists of three primary components: a large resident computer database containing current price, schedule, availability, and booking data for flights offered by the airline; a software interface that can be used to view and alter the database; and a set of display/printing terminals from which the interface program can be operated and tickets can be issued. In addition, the IRS requires personnel trained to operate the system. Airlines' use of IRSs date back to the mid-1950's.(9)
The first commercially successful CRS was American Airlines' SABRE, introduced to travel agents in 1976. SABRE's closest rival has been United's APOLLO system, also introduced in 1976.(10) A number of firms subsequently entered the market, including USAir, Frontier Airlines, Western Airlines, and American Express.(11) Many of these vendors eventually went out of business, leaving three competitors to APOLLO and SABRE: TWA's PARS system, Continental's System One (which was Eastern's SODA before Eastern was bought by Texas Air) and Delta's DATAS II system.(12) In 1990, after suffering serious financial difficulties, the DATAS II system was merged with PARS, leaving four U.S. vendors.(13) While market shares fluctuate, SABRE and APOLLO have always been the industry leaders. Outside the United States, other CRSs are being developed or already exist in Western Europe, Canada, Mexico, Asia, and Australia.(14)
A CRS is in many respects a natural extension of an IRS. Like an IRS, a CRS must provide access to relevant data concerning flights covered by the system. Similarly, both systems require an appropriate software interface, a set of terminals from which the system can be operated, and suitably trained personnel. There are, however, two primary distinctions between the two systems. First, a CRS is generally operated by independent travel agents rather than the airline's employees. Second, a CRS typically provides more extensive and timely price and schedule information and booking capabilities of other airlines that an IRS does. Since a CRS is designed for use by travel agents, the CRS software also generally includes an accounting support system that maintains sales records for individual travel agencies.
In response to mounting criticism, in November of 1984 the CAB enacted a series of regulations pertaining to the behavior of airline-owned CRSS.(15) Along with a variety of other provisions, vendors were prohibited from offering display preference and from signing contracts of greater than 5 years with travel agencies. Vendors were also required to offer access and services to every carrier on an equal basis.
III. The nature of CRS competition
There are currently only four CRS vendors in the U.S. Further, the sunk costs necessary to enter the CRS industry appear to be large, implying that barriers to de novo entry are high.(16) Given such market conditions, the industry may be subject to collusive activity among its members. Writers in this area, however, often assert that due to the nature of the CRS industry, each individual vendor, by itself, has substantial market power. This section will describe how this conclusion was reached and why it is unwarranted.
A. The importance of CRSs in the air travel industry
CRS usage has grown rapidly since 1976. Today over 95% of navel agencies have CRS terminals installed.(17) Currently about 80% of airline tickets are booked through CRSs.(18) There are at least two reasons for the widespread use of CRSs in travel agency marketing. First, a CRS gives travel agencies access to an extensive and useful base of information with regard to available price and schedule options. This information allows travel agencies to reduce consumer search costs substantially. This attribute has become especially important since airline travel was deregulated in 1978, resulting in a dramatic increase in the number of flights and fares available to consumers. Second, travel agencies provide convenient purchase locations for many consumers. Installation of CRS access at a travel agency allows the consumer to purchase airline tickets, hotel accommodations, rental cars, and other travel needs with minimum inconvenience. Airline customers can then choose which travel agency to frequent based on these and other factors. Thus, a CRS offers benefits to both airlines and travel agencies. It allows airlines to market their products in an economical manner and travel agents to better serve their customers.
CRS vendors currently gain the largest share of their revenues through booking fees from participating carriers. In 1990 each carrier was charged a fee of $185 per "segment" booked on a particular CRS. Making "direct access" bookings cost airlines an additional 25 cents.(19) Travel agents pay fees for equipment and support services.(20) It appears that currently, in general, travel agents do not pay any marginal costs (a fee per booking) for the use of CRSs.
B. The pricing of CRS services
An airline that offers CRS services provides benefits to both the subscribing travel agents and the participating airlines. Because both parties benefit, both can be expected, in one way or another, to pay for the good. Since, however, a large portion of the costs of producing the good are common to both parties, it is not obvious how the market will allocate those Costs.(21)
Similar to credit card providers and real estate listing services, CRS operators help bring sellers of products together with buyers. For such products it is indeterminate a priori how the market will determine who will directly pay the cost of the common service. For most credit cards, the direct costs of the service are largely paid by retailers. These charges are then typically reflected in the prices paid by consumers through their effect on the cost of the products being sold. Some credit cards, however, such as those used at many gasoline stations, have the buyer pay directly for the service. Similarly, real estate listing firms sometimes have the buyer pay directly for the service and sometimes have the seller pay for the service.
CRS charges can be expected to be allocated to minimize transactions Costs.(22) If transactions between CRS vendors and travel agents are most costly than transactions between vendors and airlines, then charges are more likely to be paid directly by the airlines. Transactions costs may be the reason why credit card bills are usually borne by retailers. The retailer can add up all its credit card receipts when dealing with the credit card company. Buyers of goods and services who use credit cards may not be able to take advantage of such economies of scale. This may also explain why airlines, and not airline consumers, directly pay travel agent commissions. Transactions costs may also be reduced by CRS vendors sending bills for booking fees to eight or nine major airlines (arid a few dozen minor lines) as opposed to approximately 30,000 travel agencies. Another possible efficiency reason for this allocation of costs is that small travel agencies may find it difficult to pay CRS booking fees due to liquidity constraints.
It is the total price, not the price charged to particular types of participants, that is of primary relevance from an economic and antitrust perspective. The way these charges are allocated among groups of customers is unimportant for competitive analysis. Further, like various forms of taxes,(23) who directly pays CRS charges is unrelated to who directly bears the burden of these fees. (See the discussion of commission rates below in part C.)
The nature of CRS pricing has led to a number of misconceptions. For instance, the 1987 Department of Transportation (DOT) report has one chapter that concludes that the prices charged to travel agents are too low and the prices charged to airlines are too high. The Congressional Budget Office reached similar conclusions.(24) These conclusions fail to take into account the common nature of the CRS product,
C. The nature of competition in CRS markets
CRS vendors compete among themselves for travel agency placements along various margins such as price and quality. According to DOT, "there is vigorous competition among CRS vendors to induce agents to enter into contractual arrangements with them."(25) The nature of competition among CRS vendors for airline listings, however, is somewhat more complicated and has been widely misunderstood.
The source of this confusion arises from the fact that most travel agents have access to only one CRS. If an airline does not have access to that CRS, then travel agents using that CRS are unlikely to write tickets for that carrier, since they would have to process them by hand. It is therefore concluded that there exists no significant constraint on the prices CRS vendors can charge airlines for access. For example, the Department of Justice wrote that ". . . no nationwide carrier can afford to forego listings on any major CRS, and therefore, rivalry among the CRSs cannot be expected to serve as a check against the exercise of market power by any one of them."(26) The Congressional Budget Office concluded that "CRS owners face little competitive pressure in setting these booking fees [to airlines]: since travel agents do not have to pay the booking fees, they do not consider them in deciding which system to select."(27) Guerin-Calvert and Noll state "[no market] mechanism exists to bid down airline participation fees. One CRS is not a substitute for another as long as each agency uses only one CRS. Thus, nothing prevents a CRS from extracting excess profits from airline customers."(28)
To model the CRS market it is necessary to take into account the effect of CRS access, pricing and commissions on the marketing of airline tickets, and the importance of competition among travel agencies upon each other. To understand the importance of these factors, consider the following example. Suppose that a CRS vendor raises the booking price to an airline by $1 for a given itinerary. That airline may choose to remain on that CRS despite the higher price. The cost to that airline, however, of selling such an itinerary through a travel agent using that system is now more expensive by $1. Thus, the airline can be expected (other things being equal) to lower its commissions to a travel agent using that CRS by $1. This makes the particular CRS vendor less attractive to an agent, and reduces by a corresponding amount the price the agent is willing to pay for a particular system
Now suppose that the CRS vendor raises its price to an airline by $1 without a corresponding decrease in price to travel agents. As the discussion above indicates, the CRS vendor is raising the net price of its service by $1. In response, the airline will "buy" fewer bookings on that system by lowering commission rates to the travel agents that use that CRS. Fewer travel agents will sign up for the system because its effective price (the price agents pay for CRS access minus the difference in commission rates) will have risen. Travel agents who remain on the system will have fewer resources to spend in marketing airline tickets and their output will decline. The bookings of agents on the other systems will rise as airlines, travel agents, and consumers switch into using those systems.(29)
Thus, the final result is that if any vendor raises its net price, airlines, travel agents, and airline customers can be expected to substitute away from that system and toward the other vendors. This is conceptually the same as what would occur if, for instance, a clothing store were to increase the (implicit) price it charges to manufacturers for space in the store. Manufacturers would look for other stores to sell their products, and consumers would look for other stores that could offer them better services and prices.
The most compelling evidence usually cited(30) for the existence of CRS market power are events occurring in the early 1980's, soon after the advent of airline deregulation. Newly former "upstart" carriers that competed with CRS vendor owners were required to pay much higher booking prices (from $2.00 to $3.00 a segment), as opposed to lower prices for established carriers (from $0.25 to $0.75 a segment). These differential prices were said to be the result of market power being exercised in the nature of price discrimination.
These events, however, are entirely consistent with how a competitive CRS market would operate.(31) Price discrimination refers to charging firms different prices for the same service at the same time. If the same service is not involved, prices will naturally be different. If the service in question is sold at a different time, the relevant price will reflect the different demand and supply conditions that apply to each period of time.
The upstart carriers, being small, would have been unable to take advantage of any economies of scale that exist in CRS access. Further, many of the contracts supplied different levels of display preference (as discussed below in section IV). These factors imply that each carrier was obtaining a different service when it acquired CRS access.
More importantly, the price of CRS access appears to have largely been a function of when that access was acquired. The established carriers obtained low fee contracts generally during the early years of CRS marketing, often prior to the enactment of airline deregulation.(32) During this time CRSs may not have been a proven product and it may not have been known how valuable CRSs services would be. These contracts may have been sold as "loss leaders" in order to make CRSs more attractive to travel agents and other airlines.(33)
In addition, these contracts were signed in a period prior to when the current distribution of CRS costs was established. It appears that vendors' initial marketing strategy was to charge travel agents the larger share of CRS costs. This system gradually was replaced by the current system, where airlines pay the larger share of CRS fees. During the late 1970's travel agents paid a much larger share of CRS costs than they did in the early and mid-1980's.(34) This would imply that during this period of disequilibrium airlines were paying a smaller part of CRS costs and thus would be awe to gain CRS access at a lower price.(35)
It may well be that each major airline has a strong desire to be listed on each CRS, as the Justice Department states. That does not imply, however, that an airline cannot vary its output of tickets on a particular system. If a CRS vendor charges a supracompetitive price for its service (without lowering the price charged to travel agents), then the airline can be expected to reduce commissions paid to the relevant travel agents and therefore reduce output on that CRS. Therefore, absent collusion, rivalry among different CRSs can be expected to reduce or eliminate the market power of any single CRS vendor, just as rivalries among retail establishments and credit card companies can be expected to reduce or eliminate the market power of any one firm in those markets.
D. Observational tests of the hypotheses
The competitive hypothesis presented here can be tested against the "market power" hypothesis generally presented by advocates of CRS regulation. Examining events in the CRS market and reactions to suggested regulatory provisions can shed light on which model is appropriate.
Under the market power hypothesis, any established CRS, even the DATAS II system prior to its merger with PARS, should be able to capture a large degree of monopoly profits.(36) Yet one of the reasons for the DATAS II-PARS merger was the poor financial condition of DATAS II. It seems odd that DATAS II did not alleviate its financial troubles by raising its booking fees above the $1.85 it charged.
Indeed, it would appear odd that in these circumstances booking fees are so low (about one-two percent of airline revenues).(37) If each CRS had monopoly power, it would seem to be in position to capture a good deal of monopoly profits associated with airline travel. Given the amount of political pressure CRS vendors have received, it could be that the threat of action from DOT or Congress has implicitly regulated CRS access prices.(38) If this were the case, however, it would imply that the CRS vendors with the most political difficulty (American and United) would charge the lowest prices. In fact, in the post-1984 era, CRS vendors have largely charged the same booking fee. At times when booking fees were different, SABRE and APOLLO always charged the highest fee. Any change in booking fees was led by these two vendors.(39)
One of the implications of the "competitive" hypothesis is that, because CRSs are joint goods, if something limits booking fees to airlines, fees to travel agents would rise. Guerin-Calvert and NOll(40) bear this out. According to them, "In the course of the [In re Air Passenger Computer Reservation Systems] litigation, the arguments of both sides made clear that a core issue was whether CRS vendors should charge a booking fee, and that should plaintiffs prevail [resulting in lowering or eliminating booking fees to airlines], prices to travel agents would likely increase." Similarly, in CRS regulatory proceedings the Department of Justice has advocated the elimination of booking fees to carriers, only to receive opposition from travel agency groups that believe that eliminating booking fees for carriers would prompt airlines to charge them booking fees.
IV. Display preference
Display preference (or "bias," as A is often called in the industry) refers generally to the practice of systematically improving the position of particular flights on CRS display screens based on the identity of the supplier. Prior to November of 1984, airlines that offered CRS services typically "bought" for their own flights display preference relative to the flights of other participating carriers.(41) Vendors would then generally offer other airlines the opportunity to buy "co-host" status (with a somewhat lower display ranking) or non-host status (with the lowest available display ranking).(42)
Display preference has been the focus of significant controversy. In response to these complaints, the CAB prohibited display preference as pan of As 1984 rules.(43) While there has been substantial criticism of display preference, there has been little explanation of precisely how it impaired competition. This section will explore the nature of display preference and its implications for competition.
A. Markets for display preference
Display preference is not unique to CRSs. h is ri common phenomenon in marketing, and arises in contexts where there is a limited supply of preferred display space. Newspapers charge higher prices for advertisements on more prominent pages. Television stations charge more for "prime time" commercials. Other examples of display preference can be found in telephone books, the product placement on supermarket shelves, the location of selection buttons on soft drink machines, and in the Official Airline Guide for air travel services.(44)
The value that firms place on display preference is derived from at least three sources. First, display preference may reduce the cost to consumers of searching; for their preferred supplier. Second, the willingness of a firm to purchase display preference may provide a signal of product quality. Third, a preferred display position may increase customer brand awareness for new or relatively unknown suppliers. Each of these considerations can be applied to the airline industry.
The reputation and market position of the affiliated air carriers may play an important role in the ability of the CRS vendors to place their systems with travel agents. As Guerin-Calvert and Noll(45) put it, "the reputational value of the best airlines spills over into their marketing of CRS, and then the great benefits of a CRS to the travel agent spill over to increase the likelihood that the travel agent will sell tickets on that airline." It follows that the agents on any particular system are more likely to have a preference for the host carrier's "name brand," and may prefer a display that gives priority to that carrier. The travel agency's choice of a CRS may reflect the strength of the agency's confidence in the various competing airlines, and thus its preferences with respect to the competing products each airline offers for sale.(46) Agents may also prefer that carrier because its vertically integrated CRS offers more accurate information on its own flights. Agents may also prefer display preference for the host carrier if they receive higher commissions from that carrier. Such travel agents would be likely to book more tickets on the host carrier, even if there was no display preference. Guerin-Calvert and Noll(47) also note that a travel agency's choice of CRS may also result from "synergy in marketing--if a carrier already had extensive dealings with an agency, it faced an easier time (lower costs) in selling the agency its CRS." Thus, display preference for a host carrier can be viewed as an additional form of efficient partial vertical integration between carriers and agents.(49)
This hypothesis is consistent with the way airlines structure their own IRSs. These systems are heavily "biased" toward the carrier that operates them. The customers who use these systems, having chosen a particular airline to contact, have expressed a desire for that name brand and it is therefore efficient for an IRS to generate display preference for its host carrier. Similarly, it may be efficient for a vendor to generate display preference in its external CRS to meet the demands of the travel agent network it has established, agents who may be likely to prefer the vendor's air service name brand.
This hypothesis is also consistent with the events that followed the November 1984 CAB rule prohibiting display preference on primary display screens. SABRE and APOLLO subsequently offered secondary display screens with display preference to travel agents. A large number of agents apparently bypassed the initial "neutral" displays offered by the two vendors in favor of the "biased" displays. These travel agents evidently preferred using the screens with display preference, even when they had easier access to "neuntral" displays.(49) It follows that those agents had a preference for the American and United brand names. The reduction in search costs made possible by the "biased" screens thus represents an efficiency associated with display preference.
The argument that display preference provides a signal of product quality is less straightforward. The direct consumers of CRS services are travel agents rather than the traveling public. Thus, any inference of product quality drawn from the display position of We airline would affect demand only to the extent that the signal is relayed to the traveler by the travel agent. (The frequent repeat customer, who may be more likely to use a travel agent, could be expected to be more likely to understand this signal.) This is similar to the phenomenon of firms investing in advertising to signal to customers the high quality of their goods(50) and complementary to the argument that the demand for displays is driven by the preferences of the travel agent, who chooses a vendor/airline that gives that agent and his customers high quality service.
A, related explanation of the relationship between the demand for display preference and the airline quality can be derived from the observed characteristics of "low quality" (reduced service) airlines. Low cost, reduced service airlines typically focus more on direct price advertising compared to their relatively higher priced, higher quality competitors. These airlines attempt to attract cost-conscious customers most interested in obtaining low fares. Such customers may be less likely to be influenced by schedule differences and other flight characteristics used to determine display position. Display preference thus may represent an unnecessary expense for reduced service airlines. By avoiding the cost of display preference, discount airlines are able to reduce their fares, providing benefits to highly cost-conscious passengers. The customers of the full service airlines (perhaps in large part business travelers) appear more sensitive to flight characteristics other than price. For this reason, display preference may have a much greater value to full service airlines.
This hypothesis is also consistent with the behavior observed in the CRS marketplace. Discount airlines such as World Airways, Midway, Muse, and America West generally purchased the lowest available level of display preference service on CRSs prior to November of 1984.51 More dramatically, the two most aggressive price cutting carriers prior to 1985, People Express and Southwest, chose not to offer their products through any CRS. This would imply that their price-conscious customers were willing to engage in even higher search costs to take advantage of the lower fares these two airlines could offer by not paying CRS booking fees. A similar argument can be made with respect to airlines that currently choose not to pay for direct access.
B. Display preference as anticompetitive exclusion
Numerous parties have asserted that display preference inhibited competition by diverting traffic to "host" carriers from other airlines who were "excluded" from higher display orderings. The CAB attached considerable importance to American Airlines' estimate that if a travel agent used SABRE, American's market share was from 8% to 15% over its national average, while if the agent used the APOLLO system (on which American did not have co-host status) American's market share was 10% below its national average.(52) Estimates also indicate that between 70% and 90% of all CRS tickets were booked from the primary display screen.(53) It is to be expected, however, that the first screen would produce a large number of bookings, since ordering was and is largely based on the service criteria desired by customers. Furthermore, as discussed above, travel agents who select a particular CRS vendor may have a customer-induced preference for the flights offered by the host airline.
Morrison and Winston(54) state that CRS display preference is anticompetitive because it increases search costs for consumers. Although this point is not fully elaborated, its essence appears to be that the presence of display preference induces customers to buy their tickets from the favored airline when there are better options available. Krattenmaker and Salop(55) have similarly argued that CRS display preference increases the relative cost of marketing tickets for nonvendor airlines generating anticompetitive exclusion. The Department of Justice also concludes that CRS vendors use display preference to "leverage" market power in CRSs into the airline market. That is, CRS vendors have used their market power in CRSs to increase their share of the air travel market.(56)
As discussed above, while display preference increases the costs of locating products offered by low preference firms it reduces the cost of locating other products offered by high preference firms. The combined effect on "aggregate" search costs is ambiguous. However, economic theory suggests that, absent any market failure, the market will allocate "shelf space" on CRS screens, like any other scarce resouce, to its highest valued use, and therefore reduce net search costs.
Krattenmaker and Salop do not fully describe their analysis as to why display preference is anticompetitive. It would appear, however, from the context of their analysis that they believe that display preference improperly excludes low preference airlines from distributing their products. As Liebeler notes, however, the Krattenmaker and Salop argument appears consistent with the idea that carriers should receive CRS access at no charge.(57) As developed, both positions would appear difficult to either prove or disprove.
The relevant facts, however, indicate that neither American nor United made a systematic effort to buy excessively high display preference from the other CRS vendors. Neither of these carriers had a high level of display preference on the four other systems. In addition, both American and United had the lowest level of preference ("non-host") on each other's systems.(58)
The Department of Justice's leverage argument is subject to the standard critiques in the economic literature. Generally, there is only one monopoly profit. If the relevant input is used in fixed proportions, vertical integration can be expected to have no effect on economic welfare (absent efficiencies from such integration) and market power cannot be "leverage" from one market to another. If the input is used in variable proportions (as in CRSs) then the welfare implications of vertical integration are generally ambiguous.(59)
Even if display preference decreases output in CRSs or airline services in the short run, it may not imply that CRS vendors have behaved "anticompetitively." Suppose an airline created its own CRS, placed only its own flights on it, and then supplied the system to travel agents, similar to what Southwest Airlines has done. It would be difficult to accuse that firm of acting to harm competition. Now suppose that the vendor decided to add other airlines' flights, but with a lower display priority than its own flights.(60) It would seem more difficult to argue that this action was anticompetitive. Indeed, given that CRSs are the result of internal development and innovation, it would appear counterproductive to regulate them so closely.
C. Display preference as deception
Two reasons have been suggested(61) for why display preference may represent a deceptive trade practice. First, consumers may not fully comprehend the effects of commissions on travel agent incentives, and thus incorrectly believe that travel agents generally act in the best interest of the customer. Second, there may be no good alternative source of information on airline flights.
The travel agent market appears to be highly competitive, with low levels of concentration and insignificant barriers to entry or expansion.(62) Competition alone, however, as Darby and Karni point Out,(63) may be insufficient to eliminate deception. Even in a competitive market, firms may lack the necessary incentives to give their customers the optimal amount of information. In effect, firms compete for "fraud rents," resulting in an inefficient market outcome.
Despite this result, it remains unclear why consumers are less able to understand the motives of travel agents relative to those of other similarly situated sales personnel such as insurance and shoe salesmen. Because they are paid largely on commission, agents may have incentives to sell customers more expensive goods than the customers desire, in the same manner that insurance and shoe salesmen do. Further justification, however, is generally needed to warrant public policy action.(64)
Deception is most likely to occur in environments with three general characteristics. First, deception is more likely in markets where the good (or service) in question cannot be easily evaluated, sometimes even after consumption, by the consumer. Deception is also more likely when the good is purchased infrequently, so that the reputation gained from a purchase is not an important constraint on the agent's activities. Finally, successful deception is likely only where there are no easily accessible alternative sources of information. Two examples of the types of products where deception might be expected to occur is the servicing of major car repair problems by automobile mechanics and medical care by physicians.
It is not clear that any of these conditions characterize the travel agent market. Airline tickets are easily evaluated as to their own price and quality. Further, the critics of display preference, as well as the other literature in this area, generally assume that CRS airline customers are likely to be repeat buyers, often frequenting the same travel agent or airline several times. Finally, consumers have many other sources of flight information. They can call an airline directly, check advertisements,(65) visit or call other travel agents, examine an Official Airline Guide, or perhaps simply ask their usual agent if other flights are available. In the case of business clients, firms can also use auditing to evaluate their agent's performance or vertically integrate and set up their own in-house travel agency. Thus, it seems unlikely that travel agents will act to deceive customers because such deception would threaten the stream of "quasi-rents" travel agents would expect to receive from their long-term arrangements with their customers.(66) Put another way, it does not appear logical that customers would enter into long-term arrangements with travel agents only to be systematically deceived by them.(67)
If display preference does indeed generate a deception problem, there is a much less restrictive solution than an outright ban on display preference. Travel agents and CRS vendors could merely be required to inform customers (perhaps through labeling) that the relevant CRS has a display algorithm that grants preference to certain airlines. Consumers could then decide for themselves which type of preference they desired.
D. Display preference as an aid to CRSs and airline entry
Display preference could also generate new opportunities for entry in the airline industry.(68) Selecting the desired level of display preference may help airlines choose the marketing technique that best fits their corporate strategy. An airline could buy high display preference, or perhaps even advertise its product on the first lines of a CRS display screen as a substitute or a complement for other forms of marketing. On the other hand, a low cost airline could buy low cost, low quality CRS service as part of offering a low cost product for consumers willing to search. In effect, CAB regulation may be limiting product differentiation in the market for airline services.
Antipreference regulations may also decrease competition in the CRS vendor market. A firm will enter the CRS market only if it expects to make at least a competitive return on its investment. Product differentiation through offering co-host status to airlines was an important source of revenue for CRS vendors. With this option foreclosed by regulation, CRS vendors have less opportunity to offer services that can earn a competitive return. This may retard entry into the CRS market.(69)
Analogously, if a newspaper could only run classified ads of a fixed size, and not full page ads, its revenue base would be substantially diminished. Not being able to run full page advertisements would not only reduce the information content of the newspaper but would also force subscribers to pay a greater share of the newspaper's costs. This would result in reduced circulation or even the newspaper ceasing publication. The same analysis appears to apply in the CRS industry. By reducing opportunities to generate revenue a ban on display preference reduces the possibilities for competitive entry.
Because there are only four CRS vendors in the U.S. the CRS market may be subject to collusion. It does not appear correct, however, to conclude that each CRS vendor, by itself, has market power. CRSs are joint goods, and competition in this industry is governed not only by the prices vendors charge airlines, but also by the prices vendors charge travel agents and the commissions that airlines pay travel agents. The current alignment of CRS charges, with airlines directly paying most of the costs of CRS services, would appear to be the arrangement that minimizes transactions costs.
Display preference is a common phenomenon in the economy. There are several procompetitive possibilities why display preference might exist in CRSs. One logical explanation is that many travel agents and their customers preferred their display screens to be ordered in that manner. "Shelf space" on display screens is a scarce resource that, barring market failure, economic theory suggests can be expected to be allocated to its highest valued use.
The anticompetitive explanations of display preference do not appear consistent with the nature of the CRS industry. In particular, it appears unlikely that travel agents are systematically deceiving their customers. Further, a simple remedy to any problem in this area might be to merely require labeling on CRS systems that use display preference.
There seems to be no end to the amount of regulatory and antitrust difficulties CRS vendors have had with their innovations. Yet the two most important antitrust charges, that of market power and display preference, appear theoretically incorrect and derived from important misunderstandings about how competition works in the CRS industry. These misunderstandings have led to unwarranted policy actions in this area. (1) See Congressional Budget Office, Policies for the Deregulated Airline Industry (July 1988) (hereafter cited as CBO); Council of Economic Advisors, Maintaining the Momentum, in The Economic Report of The President 199-229 (1988); Jon Ogur, Micahel Vita & Curtis Wagner, The Deregulated Airline Industry: A Review of the Evidence (Bureau of Economics, Federal Trade Commission, 1988); and Morrison & Winston, Enhancing the Performance of the Deregulated Air Transportation System, 1 Brookings Papers on Economic Activity: Microeconomics 61-112 (1989). (2) See, for example, Kahn, Surprises of Airline Deregulation, 78 am. Econ. Rev. 316 (May 1988). (3) See, for example, In re Air Passenger Computer Reservation Systems, 694 F. Supp. 1443 (S.D. Cal. 1988), 948 F.2d 536 (9th Cir. 1991) and United Air Lines, Inc. v. Austin Travel Corp., 867 F.2d 737 (2d Cir. 1989). To date, it appears that antitrust plaintiffs in CRS cases have not received any court decisions in their favor. (4) See Civil Aeronautics Board, Report to Congress on Airline Computer Reservation Systems (June 1983) (hereafter cited as CAB); Department of Justice, 1985 Report of the Department of Justice to Congress on the Airline Computer Reservation Systems Industry (December 20, 1985) (hereafter cited as DOJ); Department of Transportation, Study of Airline Computer Reservation Systems (May 1988) (hereafter cited as DOD; CBO, supra note 1; Department of Justice, Comments on Advance Notice of Proposed Rulemaking--Computer Reservation Systems Regulations, Before the Department of Transportation, Docket No. 46494 (November 22, 1989) (hereafter cited as DOJ 1989); and Department of Transportation, Airline Marketing Practices: Travel Agencies, Frequent-flyer Programs, and Computer Reservation Systems, Secretary's Task Force on Competition in the U.S. Domestic Airline Industry (February 1990) (hereafter cited as DOT 1990). (5) Saunders, The Antitrust Implications of Computer Reservation Systems, 51 J. Air L. & Com. 157 (1985); Krattenmaker & Salop, Analyzing Anticompetitive Exclusion, 56 Antitrust L. J. 71 (1987); Krattenmaker & Salop, Exclusion and Antitrust, 11 Regulation 29-33, 40 (1987); Levine, Airline Competition in Deregulated Markets: Theory, Firm Strategy, and Public Policy, 4 Yale J. Reg. 393 (1987); Steven Morrison & Clifford Winston, The Economic Effects of Airline Deregulation (1986); Fisher, Pan American to United: the Pacific Division Transfer Case, 18 Rand J. Econ. 492 (1987); Margaret E. Guerin-Calvert, Vertical Integration as a Threat to Competition: The CAB's Investigation of the Airlines' Computer Reservation Systems, in The Antitrust Revolution 338-70 (White & Kwoka eds. 1988); and M. Guerin-Calvert & Roger Noll, Computer Reservation Systems and Their Network Linkages to the Airline Industry, in Electronic Services Networks: A Business and Policy Challenge, chap. 8, 145-87 (Wildman & Guerin-Calvert eds. 1991). (6) See, for example, Computer Reservation System Regulations, Department of Transportation, 57 Fed. Reg. 43780 (September 22, 1992). (7) For a more detailed discussion of direct access, see DOT, supra note 4, at 29. (8) DOT 1990, supra note 4, at 44. Over the past several years Southwest Airlines has found a different way of marketing its product. On a limited basis it has sold software ("Southwest Airlines Dial Access Network") that allows travel agents to dial directly into Southwest's own IRS and book ticket. (9) CAB, supra note 4, at 7. (10) SABRE and APOLLO appear to be common textbook examples of successful innovation in the management information systems area. See, for example, Kenneth C. Laudon & Jane Price Laudon, Management Information Systems: A Contemporary Persoective (71-72, 81-82, 1988). (11) CAB, supra note 4, at 37. (12) Half ownership of PARS was sold in late 1986 to Northwest Airlines. 49.9% of Covia (APOLLO'sparent company) was sold in 1988 to a consortium of USAir, British Airways, Swissair, KLM, and Alitalia (DOT 1990, supra note 4, at 57 n.4). (13) For a discussion of DATAS II's financial problems, see DOT, supra note 4, at 111-12. (14) A more extensive discussion of the history of the CRS industry is contained in Guerin-Calvert & Noll, supra note 5, at 157-73. (15) See 14 CFR 255. (16) See DOT, supra note 4, at 22-27. Entry into the U.S. market by foreign CRS firms may be more likely. Guerin-Calvert & Noll, supra note 5, at 160, appear to believe, however, that sunk costs are not particularly large in this industry. (17) DOT 1990, supra note 4, at 21. (18) DOT, supra note 4, at 13. It has been estimated, however, that from 20% to 25% of these tickets are booked using airlines' IRSs and then printed using travel agents' CRS access (DOT, supra note 4, at 15 1), implying that from 60% to 64% of all tickets are booked using information from a CRS. (19) See DOT 1990, supra note 4, at 56-57. A segment, or "leg" is part of a passenger's itinerary that is on one individual flight and thus calls for an individual ticket coupon. For instance, if a passenger's itinerary calls for him to fly USAir flight #100 from New York to Pittsburgh and then flight #101 to Milwaukee, such an itinerary has two segments. If the booking the is $1.85. USAir will pay the relevant CRS vendor $3.70 for booking this itinerary. If direct access is used at a cost of 25 cents per segment to book both segments, USAir will pay $4.20 in booking fees for this itinerary. (20) DOT, supra note 4, at 3 1; DOJ 1989, supra note 4, at 4. (21) The classic discussion of common, or "joint" goods is Alfred Marshall, Principles of Economics (8th ed. 1936). (22) For general discussions of the economics of transactions costs see Alchian & Woodward, The Firm Is Dead, Long Live the Firm: A Review of Oliver Williamson's "The Economic Institutions of Capitalism", 26 J. Econ. Lit. 65 (1988); Coase, The Nature of the Firm, 4 Economica (n.s.) 386 (1937); Coase, The Problem of Social Cost, 3 J. L. & Econ. 1 (1960); and Williamson, Assessing Vertical Market Restrictions: Antitrust Ramifications of the Transactions Cost Approach, 127 U. Pa. L. Rev. 953 (1977). (23) See Antony B. Atkinson & Joseph P. Stiglitz, Lectures on Public Economics 160 (1980). (24) DOT, supra note 4, at 91-111; CBO, supra note 1, at ix & 30-31. Guerin-Calvert & Noll, supra note 5, at 177-78, note that in the CRS antitrust case In re Air Passenger Computer Reservation Systems the plaintiffs alleged that the defendants engaged in predatory pricing on sales to travel agents and monopoly pricing when selling access to airlines. (25) DOT 1990, supra now 4, at 50. This form of competition seems to be well understood by both DOT and DOJ. See, for example, DOT, supra note 4, at 31-38; and DOJ 1989, supra note 4, at 21-22. (26) Department of Justice, Comments on Advance Notice of Proposed Rulemaking--Computer Reservation Systems Regulations, Before the Civil Aeronautics Board, Docket No. 41686 (November 17, 1983) at 46. See similar statement in DOJ 1989, supra note 4, at 10-11. (27) CBO, supra note 1, at ix. (28) Supra note 5, at 151 As a result of this and similar conclusion individual CRSs are often referred to as "essential facilities." See, for example, Gorinson, Overview: Essential Facilities and Regulation, 58 Antitrust L. J. 871, 874 (1990) and Department of Transportation, supra note 6. For a critique of the "essential facilities doctrine," see Reiffen & Kleit, Terminal Railroad Revisited: Foreclosure of an Essential Facility or Simple Horizontal Monopoly? 33 J. L. & Econ. 419 (1990). (29) Conceptually, airlines could also take unilateral action in response to high CRS access fees from one vendor by including access fees as part of ticket fees to consumers. (30) For instance, by Guerin-Calvert & Noll, supra note 5, at 166, and Department of Transportation, supra note 6. (31) The analysis here is similar to that contained in In re Air Passenger Computer Reservation Systems, 694 F. Supp. 1443, 1462 (198%. For a general discussion of why price discrimination and market power cannot always be inferred from price differences, see Lott Jr. & Roberts, A Guide to the Pitfalls of Identifying Price Discrimination, 29 Econ. Inquiry 14 (1991). (32) CAB, supra note 4, at 53-57. (33) See the discussion along these lines in CAB, supra note 4, at 11-12. (34) See DOT, supra note 4, at 51-53; CAB, supra note 4, at 53-54; and Guerin-Calvert & Noll, supra note 5, table 8.1, at 158-59. (35) Indeed, the MARS PLUS system continued to charge different prices after the 1984 CAB regulations. DOJ, supra note 4, at 44, n.59, states that the price differences were a function of the time period that carriers signed their booking fee contracts with MARS PLUS. MARS PLUS was a small CRS whose market share apparently never exceeded about five percent. It was not owned 19 an airline and hence the 1984 CAB rules did not apply to. MARS PLUS went out of business in 1986. (36) See, for example, DOT, supra note 4, at 90 and DOJ, supra note 4, at 32. (37) See DOT 1990, supra note 4, at 59 n.5, DOJ, supra note 4, at 53 n.65. Guerin-Calvert & Noll, supra note 5, at 180, point out that, "[i]n general, neither airline nor CRS profits are particularly high, except for APOLLO and SABRE, which arguably are the best CRS[s] in terms of capabilities." (38) As discussed in DOJ, supra note 4, at 13-14. (39) See DOJ, supra note 4, at 45 and DOT, 1990, supra note 4, at 56. It is not clear why political factors deterred DATAS II from raising its prices. It seems likely that political actors would have preferred that DATAS II stay in the market at a higher price rather than merge with the competing PARS system. (40) Supra note 5, at 180. Guerin-Calvert & Noll also note that "[a]n apparently insurmountable problem for plaintiffs was to explain why the antitrust laws should care whether airlines or travel agents paid for CRS services," a question that is also discussed above in part B of this section. (41) The one exception among vendor-owning airlines was Delta, which placed no explicit display preference in its DATAS II system displays. See CAB, supra note 4, at 35. (42) The nature of display preference varied over both time and among the various CRS vendors. For example, in 1983 the TWA PARS system gave TWA flights a 30 "point" or minute advantage over a co-host's flight and 135 point advantage over non-co-hosts. All other criteria being equal, this meant that if a travel agent asked for a flight to a certain city leaving at 9 a.m., a TWA flight leaving at 11:14 would appear before a non-co-host flight leaving at 9. A co-host's flight at 10:44 would appear before a non-co-host flight at 9, but the non-co-host's 9 a.m. flight would appear before a co-host's 10:46 flight. (43) While the CAB rules have eliminated explicit carrier "bias," there are continuing allegations of "implicit bias." Implicit bias is said to occur when a carrier-vendor uses service related factors to improve its flights' screen position. Allegations of such behavior include the use of additional penalty points for the use of competitor's hub cities for connections and penalty points if a competitor does not offer first class service. (44) For discussions of examples of display preference, see Reuben H. Donnelley Corp., 95 F.T.C. 54 (1980); Official Airline Guides, Inc. v. FTC, 630 F.2d 920 (1980); Kellogg Co. (Cereals), 99 F.T.C. 141-50 (1982); Wilkinson, Paksoy & Mason, A Demand Analysis of Newspaper Advertising and Changes in Space Allocation, 57 J. Retailing 30 (Summer 1981); and Airline Reservation Systems: The Curse of the Mummy's Tomb, 9 Regulation 8-9, 55-56 (Jan./Feb. 1985). (45) Supra note 5, at 162. (46) This "halo effect" is discussed in DOT, supra note 4, at 116-21. (47) Supra note 5, at 162. (48) For discussions of efficiencies generated by vertical integration, see Williamson, supra now 2@ and Comments of the Department of Justice, In the Matter of Evaluation of the Syndication and Financial Interest Rules, Before the Federal Communications Commission, MM Docket No. 90-162 (June 14, 1990) at 15-20. (49) After receiving pressure from Congress and the Department of Transportation, SABRE and APOLLO ceased offering secondary screens with display preference. See Guerin-Calvert, supra note 5, at 361-62. (50) See Kihistrom & Riordan, Advertising as a Signal, 92 J. Pol. Econ. 427 (1984). (51) See CAB, supra now 4, at F4; and Identification of Gary J. Dorman as Expert Witness for American Airlines, In Re Air Passenger Computer Reservation Systems Antitrust Litigation, U.S. District Court, Central District of California, Master File No. MDL 667 ER (Tx), 1987. (52) Civil Aeronautics Board, Advanced Notice of Proposed Rulemaking, Airline Computer Reservation Systems, 48 Fed. Reg. 41176 (September 9, 1983). (53) Statement of Dan McKinnon, Chairman, Civil Aeronautics Board, Review of Airline Deregulation and Sunset of the Civil Aeronautics Board Before the Aviation Subcommittee of the House Public Works and Transportation Committee, 98th Cong., 1st Session (1983) at 9 and Statement of Jay B. Rea, Chairman, Automation Committee, Association of Retail Travel Agents, LTD, id. at 353. (54) Supra note 5, at 68-70. (55) Supra note 5, Analyzing Anticompetitive Exclusion, at 80 and Exclusion and Antitrust, at 31. (56) DOJ, supra note 4, at 11-13. It should be noted that increasing share may be the result of increasing output, which would lower price and thus be procompetitive.
Besides the "leveraging" argument in its 1985 report, the Department of Justice appears to present no economic rationale for opposing display preference. Fisher, supra note 5, at 503, states that "[o]bviously, diversion of traffic by providing biased or incomplete information is destructive of competition on its merits" and provides no further explanation. (57) Liebeler, Exclusion and Efficiency, 11 Regulation 34, 35 (1987). The thrust of Liebeler's article is that the Krattenmaker & Salop thesis on exclusionary conduct would seem to be observationally equivalent to the efficient allocation of property rights. For another critique of Kratten-maker,& Salop, see Malcolm B. Coate & Andrew N. Kleit, Exclusion, Collusion, and Confusion: The Limits of Raising Rivals' Costa (Federal Trade Commission Working Papers Series No. 179, October 1990). (58) CAB, supra note 4, at F-1. In addition, the Krattenmaker & Salop "exclusion" argument is more properly termed "overbuying." This is because American and United are alleged to have bought "too much" space on CRS displays rather than having parts of display screens go unused. The welfare implications of "overbuying" are ambiguous. See Salop & Scheffman, Cost-Raising Strategies, 36 J. Indus. Econ. 19, 26 (September 1987). (59) See, for example, Warren-Boulton, Vertical Control with Variable Proportions, 82 J. Pol. Econ. 783 (1974) and Reiffen & Kleit, supra note 28. (60) This sequence of events describes how Air Canada marketed its "Reservac" CRS to Canadian travel agents in the mid-1970's. See P. Nikolai Ehlers, Computerized Reservation Systems in the air Transport Industry 7-8 (1988). (61) See Frederick R. Warren-Boulton, Director, Economic Policy Office, Antitrust Division, Department of Justice, Computer Reservation Systems, December 16, 1983 (submitted as part of the Department of Justice's Reply Comments to CAB Docket 41686) and Borenstein, Hubs and High Fares: Airport Dominance and Market Power in the U.S. Airline Industry, 20 Rand J. Econ. 344 (1989). (62) See the discussion in DOT 1990, supra note 4, at 8-12. (63) Darby & Karni, Free Competition and the Optimal Amount of Fraud, 16 J. L. & Econ. 67 (1973). (64) See Statement of be Staff of We Bureau of Economics of the Federal Trade Commission on Proposals to Relax the Interpretation of Section 8 with Regard to Home Mortgages, Before the Department of Housing and Urban Development Docket No. R-88-1256 (July 15, 1988). (65) Guerin-Calvert & Noll, supra note 5, at 150, briefly argue that "the economic theory of efficient consumer search teaches that consumer shopping is unlikely to police efficient behavior by agents" and refer to Wilde & Schwartz, Equilibrium Comparison Shopping, 46 Rev. Econ. STUD. 543 (1979). The Wilde & Schwartz model, however, assumes no advertising and no repeat purchases, conditions that do not appear to apply to the airline travel agent market. Wilde & Schwartz note (id. at 552) that "advertising increases the likelihood that markets will behave competitively." They also suggest that the policy proscriptions that result from their model "should be taken somewhat tentatively." For a discussion of how advertising acts as a constraint of the behavior of travel agents, see Alison Leigh Cowan (with Edward A. Gargan), Mirage of Discount Air Fares Is Frustrating to Many Fliers, N.Y. Times, April 22, 199 1, at A- 1. (66) See Klein & Leffler, Nongovernmental Enforcement of Contracts: The Role of Market Forces in Assuring Quality, 88 J. Pol. Econ. 615 (1981); Shapiro, Consumer Information, Product Quality, and Seller Reputation, 13 Bell J. Econ. 20 (1982); and Von Weizsacker, The Costs of Substitution, 52 Econometrica 1085 (1984). (67) If indeed travel agent customers are prone to deception, it is not clear why that eliminating display preference would reduce the problem. Travel agents could seek other margins upon which to deceive customers. (68) See, for example, Statement of Fred L. Smith, Jr., President, Competitive Enterprise Institute, Hearings on Computer Reservation Systems Before the Aviation Subcommittee of the Senate Committee on Commerce, Science and Transportation 115-27 (1985). (69) According to Copeland & McKenney, Airline Computer Reservation Systems: Lessons From History, 12 Mis Q. 353, 363 (1988) "[p]aradoxically, the Civil Aeronautics Board's guidelines against display bias dealt the final blow to smaller carriers still harboring hopes of mounting retail automation strategies. It was financially impracticable, if not impossible, for all but the largest trunks to modify their internal reservation systems to eliminate bias and comply with the Board's guidelines. Further, generating incremental passenger revenues sufficient to offset the cost of reprogramming would be unlikely without a biased display that favored the host carrier."
ANDREW N. KLEIT Economist, Bureau of Economics, Federal Trade Commission and Visiting Assistant Professor, Louisiana State University.
AUTHOR'S NOTE: I would like to thank Gary Roberts, Mike Vita, and Paul Pautler for their help with this article. The views presented here are soleb, those or the author and not those g( the Federal Trade Commission, any of its members, or the Bureau of Economics of the FTC.
|Printer friendly Cite/link Email Feedback|
|Author:||Kleit, Andrew N.|
|Date:||Dec 22, 1992|
|Previous Article:||Future money and banks: 1990-2010.|
|Next Article:||Untwisting the strands of Chicago antitrust.|