Contestability Theory and the Deregulation of US Airlines Free

Abstract

This essay discusses the development of the notion of contestable markets—the idea that optimal Pareto conditions can be induced by facilitating ultrafree entry into and exit from markets—set within the development of applied economics. The role of potential competition in influencing the behavior of incumbent suppliers had been discussed previously, but the work of William Baumol, Elizabeth Bailey, and others, they claimed, offered a more rigorous and internally consistent framework of analysis and a sounder basis for economic policymaking. This article outlines the history of the development of the theory and provides insights into the roles of key figures and institutions. It does this in the context of the role contestability played in influencing the deregulation of US airlines in the 1970s and in terms of the empirical findings of subsequent studies of the implications of this liberalization. Overall, it is also set against an intellectual background when applied and empirical research was becoming a more central part of political economy.

Introduction

The late 1970s and 1980s, the period often associated with notions of “Reaganomics” (Niskanen 1988) and “Thatcherism” (Skidelsky 1989) and their advocacy of freer markets and monetarism, saw considerable institutional deregulation (“liberalization” in European terminology) of economic markets. The changes, together with the forms they took, were at least partly due to the interplay between the theory and application of economics. The academy not only developed new economic approaches, but at the same time some of its members actively participated in industry and governance. The outcome included removal of many economic barriers to market entry and controls over price setting, privatization, withdrawal or restructuring of state subsidies, and reductions in tariffs and other impediments to international trade.

Regarding the links between theory and action, the period saw a greater appreciation of the need to include psychology in economic modeling, and serious consideration of the ideas of the new behavioral economists began (Thaler 2016). Further, the availability of larger computer capacity in the 1970s allowed for greater storage and manipulation of data. The period also witnessed more emphasis being placed on empirical (or applied) economics rather than the abstract mathematical modeling that had followed the publication of Samuelson's (1947) Foundations. There was a shift from focusing on systematic, rational decision-making based around homo economicus to more quantified and empirical methodologies, often involving inductive and experimental approaches and making use of large databases.¹

Here I focus on one thread of the economic history of that period by first looking at the intellectual development of the concept of contestability (sometimes called latent or potential competition) during the preregulatory reform debates of the early 1970s. But this is then set within the context of the subsequent outcomes of reforms to the US airline regulatory framework that some hoped would release the powers of potential competition in that industry.² The general idea put forward by William Baumol and others at the time was that in markets where there are no sunk costs—irreversibility in the market where there are no institutional barriers to entry or exit—the pressures of potential competition under specified conditions will force suppliers to price at marginal cost. This contestability results in a Pareto optimal outcome, even if the markets are natural monopolies.

Background

Much of the early thinking about contestability, and the necessary conditions for it to function, was in the context of US telecommunications. But, partly because of legislative timing, domestic US airline reform became the subject of the first experiment. A relatively small, and diverse, group of individuals were involved in defining the intellectual underpinnings of contestability theory during the late 1970s and 1980s. Some members of the group also subsequently became directly engaged in its implementation. These were mainly academics who often crossed the line between actively participating in intellectual discussions within the academy and being engaged in the realities of industry and government.

Baumol's (1982) presidential address to the American Economic Association (AEA) in Washington, D.C., highlighted the importance he gave to the concept. But his speech was something of a retrospective and synthesis. The notion of contestability was widely known by 1981 largely due to Baumol's earlier work and the studies of Elizabeth Bailey, Robert Willig, and John Panzar in the 1970s at AT&T's Bell Telephone Laboratories (e.g., Baumol 1981). The question being posed here is, Were the claims of Baumol and others in the early 1980s as to the importance of contestability theory, and its use as a guide to economic regulations, justified? In looking at this, the present article has somewhat flexible boundaries.

First, the idea of contestability evolved in the context of a number of markets, with the bulk of the historical debate surrounding its intellectual consistency and usefulness coming between the mid-1970s and the early to mid-1990s—a period sometimes referred to as the age of regulatory reform (Button and Swann 1989). Focusing on this era, our discussion of application largely relates to the 1978 US Airline Deregulation Act and its outcomes. Concentrating on a single, major domestic industry going through radical regulatory reforms, but experiencing less substantial technical changes than, say, telecommunications, also provides a relatively stable foundation for what amounts to a historical economic event study.

The international element of US airline markets is not covered. They had been broadly regulated at the United Nations' Chicago convention in 1944 that produced a set of very similar, essentially illiberal, bilateral arrangements between pairs of nations. Our assessment largely ignores the liberalization of these treaties following implementation of the United States’ Open Skies Policy from the early 1990s. While the latter made it easier for more US carriers to integrate their domestic networks with international services, as well as to form alliances with overseas carriers, these developments took the form of piecemeal changes rather than being a single, legislative event and also only became important in the late 1990s. The US domestic market remained the primary interest of its airlines and policymakers.

I will also say little about legal notions of potential competition and contestability. The concept of contestable markets has entered the legal vocabulary in a rather looser fashion than in economics. Bishop and Walker (2010: 29) summarize the situation thus: “Broadly speaking, the legal use of the term contestable applies to those markets in which potential competition is said to exert a significant competitive constraint on the behaviour of the incumbent firms, i.e., where entry is relatively easy and does not require large sunk costs.” While there are inevitable overlaps between economics and the governance of markets—the interpretation of property rights within a set of legal stipulations—this is a subject for another paper. Here the focus is on the stricter economic definitions of contestability.

Incorporating Potential Competition into Industrial Theory

Nineteenth-century economists such as Cournot, Bertrand, Walras, Edgeworth, and Marshall formalized the workings of perfectly competitive markets, monopolies, and some forms of oligopoly. The models involved situations where suppliers take into account the responses of other actors when making pricing and output decisions but where information on this is incomplete. George Shackle (1967) goes on to highlight the roles of Joan Robinson, Chamberlin, Harrod, and others in refining this analysis in The Years of High Theory toward the end of the interwar period, and the emergence of game theory enriched the study of imperfect markets further. The academic focus, however, remained largely on interactions between incumbent suppliers, as did the concerns of regulatory policymakers (Kahn 1988a).

In practice, industries are not citadels of incumbents but vary with respect to the ease with which new sellers can enter markets. Barriers to entry consist of advantages that incumbent sellers have over a potential entrant. In our context, they are generally measurable by the extent to which established sellers can persistently elevate their prices above minimal average costs without attracting new entrants. Hurdles may exist because costs for established sellers are lower than they would be for new entrants or because buyers have some intrinsic preference for the products of incumbents. An industry's economics may also be such that new entrants would need to be able to command a substantial share of similar markets before they could operate profitably—scale barriers.

The idea of workable competition as developed by J. M. Clark (1940) can be seen as a way to operationalize a framework for minimizing distortions associated with such cases. It assumes that because the economic performance of industries varies, along with their market characteristics including entry barriers, then it should be possible to devise some standards for identifying the sorts of market structure that engender socially satisfactory performance in a given industry even if it falls short of maximization. The features that Clark thought germane may include the number and size distribution of sellers, the degree of product differentiation, the long- and short-run cost conditions of firms, the geographical market covered, the existing output of incumbents, and whether price making is quoted or governed by supply.

A problem with this, as Sosnick (1958) puts it when critiquing the method, is that Clark found eighteen writers who had proposed sophisticated criteria of workability, but these were long and varied lists of norms for performance, conduct, and structure with no clear way of weighting each. The studies also placed an emphasis on competition between incumbent actors with little concern with potential competition from outside, the point Clark had touched upon earlier.

Joseph Bain (1950), however, had argued that although the existing price need play no direct role in deciding market entry because the expected industry price after entry and the entrant's anticipated market share are strategic considerations, a potential entrant may regard the expected industry price as an indication of postentry profitability. To incorporate the effects of new entry and the reactions of incumbents, Bain (1956) developed the notion of a limit price (or quantity). He maintained, as had others, that analyses of business competition placed a disproportionate emphasis on competition among firms already established in an industry, whereas in reality incumbent firms limited their prices to make it unprofitable for competitors to enter the market.

From an economic policy perspective, a practical problem in rectifying this is the difficulty in modeling both the established firm and the entrant as strategic agents; that is, once an entrant has entered the market, the quantity used by the established firm as a threat to deter entry is no longer its best response. For limit pricing to be an effective entry deterrent, the threat must be made credible. Bain circumvents this by treating the situation as a game of complete information so that limit pricing cannot emerge in equilibrium. When information is incomplete, although preentry actions by the incumbent supplier may not influence postentry profitability, information may act as signals for some unobservable determinants of profits. However, in equilibrium, because the entrant will recognize the incentives for limit pricing, its expectations of the profitability of entry will not be consistently biased by the incumbent's behavior. Depending on the parameters of the model, the probability of entry may fall short of or exceed what it would be if there were complete information and no limit pricing.

The Bell Laboratories, NYU, Princeton, and Contestability Theory

Until the late 1960s, interest in the role of potential competition was mainly motivated by the intellectual curiosity of economists seeking a better understanding of industrial organization. It was also driven to a lesser extent by practical matters involving the regulation of specific industries, and most notably public utilities (Kahn 1988a). Advances in technology, transformations in management practices, and the lessons of experience, combined with changes in political priorities in the mid-1970s, however, brought about pressures for reforms in traditional governance.

Against this background, a major strand of the debates emerged concerning the role of competition found in the US telecommunications industry. The industry had been in private hands since its inception but subjected to economic regulation; for example, the Mann-Elkins Act of 1910 imposed common carrier obligations on all telegraph and telephone companies and put the communications industry under the jurisdiction of the Interstate Commerce Commission (ICC). The Federal Radio Commission took over the regulation of radio in 1927 until the Federal Communications Commission (FCC) succeeded it in 1934 with the remit to make “available so far as possible, to all the people of the United States, without discrimination on the basis of race, color, religion, national origin, or sex, rapid, efficient, nationwide, and world-wide wire and radio communication services with adequate facilities at reasonable charges.” The 1950s began a period of deregulation as technology changed, and it was clear that more substantive reforms could serve the public interest. The 1960s and 1970s saw, for example, FCC actions in relaxing regulation of telephone terminal equipment to stimulate competitive markets for handsets and fax machines.

AT&T, the perceived natural monopoly in the industry, sought to preempt any further regulation. It feared that the monopoly power of the company would be diluted by regulatory reforms and this, in turn, would damage the technological integration of the Bell system. At this point William Baumol became engaged (Beker 2022). From 1966, for three decades he acted as an adviser and expert witness to the company. AT&T had already lost a predatory pricing case before the FCC and had decided that this was because the government had more sophisticated economists. Partly on Baumol's advice, a team of experts led by Edward Zajac was gathered to meet future challenges. This Council of Economic Advisors included as a founding member the Cornell economist Alfred Kahn, who was later a central player in US airline deregulation (Button 2015).

Baumol had studied economics and art at the City College of New York before joining the US Department of Agriculture. Following military service, he returned to the USDA but then enrolled in the London School of Economics and was rapidly hired as an assistant lecturer in American economy and economic dynamics. After completion of his PhD he moved to Princeton, where he remained as the Joseph Douglas Green 1895 Professor of Economics, and in 1971 he also joined the faculty of New York University half-time as the Harold Price Professor of Entrepreneurship.³

His approach to economics was heterodox, with his analysis often deviating from standard neoclassical norms. He saw his work largely as intellectual pragmatism—“I've always been oriented towards applied theory” (Baumol, quoted in Krueger 2001: 224)—and, regarding contestable markets, he observed, “I found myself embarked almost accidently on what I consider to be the most fruitful piece of research in which I have ever participated” (Baumol 1989: 227). His approach, however, differed in several ways from the behavioral economists of the time such as Simon, Cyert and March, and Leibenstein, with whom he has been grouped (Gilad, Kaish, and Loeb 1984). For example, he retained neoclassical assumptions of maximization and rationality within a marginalist framework when considering motivations of sales-revenue maximization. Much the same can be said about his work on contestability and with regard to his interest in entrepreneurship that saw him adding flesh to the role of the entrepreneur in a broadly neoclassical framework.

Baumol's early theoretical interests also morphed into his later approach to economic regulation. He, for instance, sought to deal with reasonable rules for rate regulation and to establish guidelines for second-best optimization pricing for public enterprises legally constrained to a maximum profit. Additionally, in the mid-1960s, he was diagnosing his “cost disease,” explaining how technological advances raise productivity and push up wages as workers are able to produce more goods at lower cost but that these same increases in productivity do not apply to technologically unprogressive, labor-intensive activities, thus raising the relative expense of health care, education, and other essential services.

Alongside Baumol at AT&T were his “tiny band of rebels” (Baumol 1982). First among these was Bailey, whose own program, although wide ranging, specifically embraced US airlines for several years. After graduating in economics from Radcliffe College, she worked for Bell Laboratories, and studied at Stevens Institute of Technology (Landau 2017). One summer she gave a presentation on regulatory distortions to the Council of Economic Advisors at AT&T, which included Kahn. Baumol was impressed and helped her gain admission to Princeton, from which she graduated in 1972, the first woman to be awarded an economics doctorate. She subsequently retained her links with Bell, and with Baumol she produced work looking at conditions whereby contestability can serve as a “weak invisible hand” limiting the power of multiproduct natural monopolies (Baumol, Bailey, and Willig 1977). In 1977, President Carter named Bailey one of the five Civil Aeronautics Board (CAB) commissioners, and in 1981 she became vice-chair of the board. Later she became dean of Carnegie Mellon University's Business School, served as a director of companies such as CSX Corp., and was chair of the National Bureau of Economic Research.

The “tiny band” also included Robert Willig, later deputy assistant attorney general in the Antitrust Division of the US Department of Justice. Alongside him was John Panzar, who became Louis W. Menk Professor of Economics at Northwestern University. Of the others, Thijs ten Raa held academic posts at York, Erasmus, and Tilburg Universities, Gerald Faulhaber served as chief economist at the FCC, and Dietrich Fischer taught at New York University and then became academic director of the European University Center for Peace Studies in Stadtschlaining. Bailey (1982: xxii) summarizes the main roles each played:

Contestable Markets and Public Policy

His work at Bell led Baumol to conclude that under certain conditions even the largest incumbent businesses can be influenced by the threat of new market entry. In this context, the concept of contestability provides an analytic framework within which the features of demands and production technology determine the shape of industries’ structures and many of the characteristics of their prices. It does this by stripping away through its assumptions all barriers to entry and exit, along with the strategic behavior that goes with them in theory and in reality.

In terms of where the assumptions of contestability sit within the traditional neoclassical market extremes of perfect competition and monopoly, table 1 offers comparisons. While contestability is seen to limit the market power of monopolies and oligopolies, there need not be a large number of suppliers to fulfill this function, nor need there be a U-shaped cost curve to ensure only normal profits are earned. This was used by Baumol to argue that minimalist approaches to antitrust policy are generally appropriate.

Several steps led him and his team to the notion that the markets served by AT&T can be both efficient and sustainable. The “company lore,” as Colton (2022) describes it, of contestability theory can be traced back to a timeframe from the early 1970s and the legal cases pitting the United States against AT&T. AT&T was charged with engaging in uncompetitive practices, and notably with exploiting its scale to monopolize the acquisition of licenses needed to serve newly emerging telecommunications markets.

Baumol at that time was engaged in somewhat different work but was drawn into the debate. The National Science Foundation (NSF) was seeking advice on the financing of academic journals and was specifically interested in whether it was less costly to have independent journals or several journals put out by one publisher. Baumol in his own words thought,

The series of legal actions saw Baumol, together with Bell's economists, seeking to prove that monopolies were not necessarily suboptimal. In particular, there were pressures to break up the AT&T Corporation following legal action by Microwave Communications Inc. (MCI) in 1974; MCI wanted a license to run telecommunications services between Chicago and St. Louis. Baumol offered evidence that MCI's action was de facto cream skimming to extract profits from dense markets. AT&T, however, lost the case, leaving it with problems of how to proceed in the future and how to handle situations when competitors with new technologies sought licenses (Colton 2022). The initial questions involved asking, What represents a barrier to market entry other than government-imposed regulation, and do scale effects play a dominant role?

Tackling these questions, Baumol harked back to the task set by the National Science Foundation (NSF), tying together his own thoughts with those of others. For example, Faulhaber (1975) introduced the concept of economies of scope (sometimes called cost complementarity), later refined by Baumol (1977) and Panzar and Willig (1977, 1981) and measured econometrically in the airline context by Caves, Christensen, and Tretheway (1984). Such economies occur when a monopoly is the efficient form of production because a single firm is able to produce the current industry output at lower cost than two or more separate firms.

Linked to this, incorporating the notion of sunk costs offered the Bell group a way of isolating situations when economies of scope may exist by separating them from fixed costs. Sunk costs are those excluded from future business decisions because they will remain the same regardless of the outcome of a decision—they are sunk and nonrecoverable (Baumol and Willig 1981). In contrast, sizable costs like a fleet of aircraft are deemed fixed, rather than sunk, because planes can be sold off or redeployed to another route or market. Using this delineation, an entry barrier can be delineated as anything requiring expenditure by an entrant into an industry but imposing no equivalent cost upon incumbents. Economies of scale and large fixed costs are thus compatible with contestability.

Pulling these elements together provides a foundation for latent competition (Baumol 1989; Bailey 1982). If there are no sunk costs, then entities outside of existing suppliers can move into and out of a market with no fear of losing assets. Even a contestable monopolist is constrained by the potential of new actors attracted by supernormal rents. This framework may be treated as a generalization of the perfectly competitive market concept, and in particular it avoids the strong assumption that economic production exists only when there are large numbers of active suppliers. It was seen as offering a standard welfare maximizing structure in which efficiency may call for a very limited number of firms. It involves neither extreme interventions nor extreme noninterventions but rather a way of sharpening the roles that government market interventions should, in the public interest, play.

The Academic Debates

Reservations over the influence of contestability appeared soon after the Washington AEA meeting and the publication of the more expansive Contestable Markets and the Theory of Industry Structure (Baumol, Panzar, and Willig 1982).⁵

At this stage, the fencing over the place of contestability theory in economic analysis centered around technical issues and ended in an indeterminate outcome. We find Weitzman (1983), for example, arguing that increasing returns to scale are inconsistent with ultracompetition and that constant returns are needed when demand can be met by instantaneous accumulation of inventories that are created by constant production when hit-and-run actions take place. At the same time, Schwartz and Reynolds (1983) expressed concern that even a modest lag in entry can impede the influence of potential competitors if incumbents can earn monopoly rents before competition bites and then leave without cost.

A number of lunges were also made regarding policy perspectives. Schwartz (1986), for instance, asked if perfect contestability was being advanced only as a theoretical benchmark and what, from a practice standpoint, is “imperfect contestability” and whether many markets actually fall into this category. Supporting this, Brock (1983: 1062) poses a similar question of whether “perfect contestability is robust in some meaningful sense,” citing studies on market-penetration issues involving “reputation effects” as barriers to entry. He also expresses a more general concern that Baumol's analysis, being comparative statics, pays insufficient attention to games that participants, both those within markets and potential entrants, may play. Perrakis (1982), from an industrial organization perspective, concurs with this but adds an emphasis on the need to include consideration of the reactions of multiproduct incumbent suppliers.

Baumol, Panzar, and Willig's (1983) ripostes included the argument that inventories cannot be accumulated even in the shortest period if there is contestability and that the theory does not require the capacity to start and stop production costlessly. Further, they parried Weitzman by arguing that he had confused technical with economic sunk costs, leading to a misunderstanding regarding how rapid the anticipated entry and exit is: it does not even have to be instantaneous with perfect contestability. Unconvinced, Shepherd (1995) made the point that while Baumol offers ultrafree entry as a new general system of industrial organization, his analysis relates only to an extreme set of conditions seldom found in real markets that exhibit significant internal market power, concluding that “as a theory [it] has been mainly a detour” (299). Spence (1983: 982), however, offers some support for Baumol and his coauthors when pointing out that “prior to the work of our authors, economists and business strategists did not have a language or a set of concepts with which to talk precisely about scale economies in a multiproduct setting. . . . The contestable market theory . . . provides a more robust welfare standard [than perfect competition].”

Airlines and the Empirical Testing of the Theory

While the intellectual dueling over the internal consistency of contestability theory may be seen as ongoing, as far as empirical testing is concerned the seventeenth-century saying that “the proof comes out in the pudding” offers firmer conclusions. This is where applied economics becomes important, with the pudding being the outcomes of US domestic airlines’ deregulation. While the theories underlying contestability were largely developed at AT&T in the context of telecommunications, the removal of regulatory constraints on market entry and price setting in the passenger airline market after 1978 occurred at a time when empirical analysis was gaining momentum—see Backhouse and Cherrier 2017a, Backhouse and Biddle 2000, Deaton 2007, and Littlechild 2008—and provided the basis for significant quantitative work assessing the implications of these reforms.⁶

The 1920s had seen the Hoover administration introduce bidding for airmail services. In turn, this was replaced by the creation of the Civil Aeronautics Authority (later the CAB following a merger with the Air Safety Board in 1949) with the objective of developing existing trunk carriers to serve new, nonstop passenger markets with some competitive services. The board controlled US interstate scheduled carriers, deciding which routes would be serviced by which airlines and setting minimum limits on passenger fares using a standard industry fare level (SIFL) formula, and it required equality of price for services over equal distances no matter whether these involve sparsely used and expensive rural routes or routes heavily used and less costly. The CAB also rejected most applications from new carriers; the challenge of proving effective demand beyond that already being met was almost insurmountable.

Fares were based upon the airlines’ estimates of the cost of providing service plus a markup aimed at ensuring sustainable profits. Only economy and, at a 50 percent premium, first-class fares were offered, and competition was in terms of service quality, basically, flight frequency, scheduling, onboard services, and so on. While aiming to create a workable system, the rate-of-return fare regulations, with their explicit averaging and inevitable delays in fare adjusting, meant that normal profits were seldom achieved. When a carrier experienced difficulties, the CAB facilitated merger with more financially robust airlines rather than have new market entry.

This system broadly applied into the 1970s, albeit with most safety matters having been transferred to the National Transportation Safety Board by then. The CAB had de facto become a pure economic regulatory agency rather than a regulatory body for the airline industry in general. By the mid-1970s, however, a number of factors were converging to bring pressure for legislative reform. And this is one instance when academic thinking and analysis became directly pertinent to policymaking (Bailey 1986, 2002).

At the theoretical level, economists in the 1970s began to question the overall efficiency of regulating what were then loosely called public utilities. George Stigler (1971), for instance, had argued that given the size and economic and political lobbying power of the utilities, they had in effect captured the regulatory system to serve their own, and not the public, interest. These theories of regulatory capture were further extended by Peltzman (1976), who also questioned the interests of those responsible for administrating regulations.

On the empirical side, even before this, Lucile Keyes (1949, 1951) at the US Board of Investigation and Research and Richard Caves (1962) had been using traditional numerical analysis and had found no justification for limiting entry to airline markets on the basis of entrants eroding incumbents’ economies of scale. Adopting somewhat more rigorous statistical methods, Jordan (1970), Keeler (1972), and Levine (1965), when comparing fares and service coverage in the regulated interstate system with more liberal intrastate markets, found the latter offered fares about 50 percent of the CAB certified carriers and provided more services over comparable routes. Econometric work by Douglas and Miller (1974), comparing hypothetical competitive counterfactual fares to those available in CAB-regulated markets, found the latter higher, with resultant excess capacity and suboptimal low-load factors.

On the supply side, the postwar period had been one of changing technology. There had been the jetliner revolution from the late 1950s, with gains from jumbo jets being reaped in the 1960s. In addition, the range of aircraft allowed them to fly over cities and offer direct services whereas previously competition had been indirect. For example, while there were initially competitive services between New York and San Francisco via Chicago or St. Louis, the new hardware allowed direct services.

On the demand side, disposable real incomes had risen, and the US population had increased from 150 million in 1950 to 215 million in 1975 and was projected to grow further. The amount of business travel had grown as the high-technology and service sectors engaged in more networking activities than traditional industries and as the supply chains in production and distribution grew longer and more complex. With the advent of wide-bodied jets, demand for international tourism had swelled.

While the existing regulations had served to increase capacity in line with revealed demands, there was considerable evidence that the full benefits of advances in technology and management were not being realized, and pressures for regulatory review mounted. Basically, airline fares were too high, and networks did not conform to the needs of the new spatial economy.

Stemming from this, Kahn, who chaired the CAB when deregulation came, put forward three reasons for its enactment and a raison d’être for its timing. First, the carriers and some other parties wanted change.⁷ Airlines had never been very profitable, even under rate-of-return regulation. This was in part because of their restricted abilities to compete. Rate-of-return regulation, for example, prevented them from competing in price, although it did not stop competition in terms of frequency of flights. Because costs could be passed on, if airlines held fares artificially high, they would schedule more flights, resulting in poor load factors—potentially profitable seats were not filled (table 2).

Second, airlines had engaged in a binge of acquiring new hardware, and when the economy moved into a recession between 1974 and 1975, they found themselves with a tremendous number of empty seats (see again table 2). There were very strong motivations to offer discount fares to fill those seats, and pressure built up to make this legal. Finally, general inflation in the economy in the late 1970s (running at 13.5 percent at the end of Carter's administration), and its embeddedness in the stagnation of the time, led to calls for freeing up markets to reduce cost-push forces. The idea was that deregulation of key industries would compete prices down. The airline sector was important in this context—it was a prominent industry.

One of Kahn's first actions in 1977 was to establish an Office of Economic Analysis at the CAB (Eisner 1991). The economic approaches pursued by Kahn's CAB are clearly spelled out in his Ely Lecture to the AEA (Kahn 1979) in which he argued that the relevant principles are easy to characterize: economic efficiency calls for prices to equal marginal social opportunity costs, and whenever it is possible, competition is the best institutional mechanism for achieving this, as well as for minimizing X-inefficiency and ensuring the optimum rate of innovation. Simply put, as he said at a conference of airline executives, airplanes were “marginal costs with wings.” The issue then became one of whether deregulation of the US airline market would produce the economic efficiency sought and, if so, what part contestable forces would play.

Parallel to this, the CAB's economic personnel was not only boosted by the appointment of Bailey but, in 1977, by the arrival of Darius Gaskins, an expert on dynamic limit pricing, as director of the Office of Economic Analysis. Michael Levine became general director from 1978 (and later executive vice president of Northwest and Continental Airlines as well as dean of the Yale School of Management), and although not always an ardent believer in the concept of potential competition, he was strongly supportive of measures that formed part of the theory, including the removal of regulated fares (Levine 1987).

The situation within the CAB was, even prior to these appointments, changing. In 1975, Roy Pulsifer, assistant director of the CAB's Bureau of Operating Rights, had been tasked by the then chair, Richard O'Melia, to do a “self-study” of the board. The exercise involved three CAB staff economists and Lucile Keyes. The resultant special report looked at the issues “from an economic perspective. . . . The political implications of various policy alternatives have been ignored” (Pulsifer 1975: iii); and it recommended a phased elimination of entry and exit control and public-utility-type rate regulation. Contestability was not explicitly mentioned, but the report noted that there were no structural traits inherent in US air transportation that indicated superior performance by large-size firms, nor were there traits that would significantly inhibit the entry of new firms.

The US Department of Transportation (USDOT) took note of this, discerning a slightly different message, the deputy undersecretary, John Snow (1975: 648), accepting that

Bailey spent time injecting the concept of contestability into a model comprehensible to nonspecialists and to show its relevance for the airline industry (Bailey 1981; Bailey and Panzar 1981).⁸ Mapping out the features of the underlying airline market showed many similarities to the telecommunications sector, as well as traits of contestable markets. Both are transportation industries, one moving people and cargo and the other information. Both are network industries involving sunk infrastructure but with relatively costless entry and exit to and from individual links in the networks. In the context of airlines, for example, their regulation was separate from that of the airports and the air traffic control system to which the carriers required access. Their discussion centered on the degree to which services would be subject to actual and potential competition in a deregulated environment given reasonable access to the infrastructure. Debates over predatory behavior, as well as the potential and realizable threat of ultrafree entry if incumbent airlines charged excessive fares, formed part of the policymaking process.

The 1978 Airline Deregulation Act

The 1978 Airline Deregulation Act was the first thorough dismantling of a comprehensive system of US government control since the Supreme Court declared the National Recovery Act unconstitutional in 1935. It was part of a broader movement that, with varying degrees of thoroughness, transformed industries such as trucking, railroads, buses, cable television, stock exchange brokerage, oil and gas, telecommunications, financial markets, and even local electric and gas utilities. Given that most of the US federally regulated industries, unlike their Europe counterparts, were in private hands, the changes were predominantly directed at relaxing pricing, market entry, and quality of service controls, rather than whether the industries should be state or privately owned (Button and Swann 1989).

Political bipartisan support emerged for the airline legislation, which was introduced into the Senate by Howard Cannon (Schiller 2019). From 1975, Stephen Breyer (1982), the future Supreme Court justice and who was on leave from the Harvard Business School as a Senate staffer, structured a series of hearings held by Edward Kennedy to consider if airline fares were too high, whether market entry was being excessively blocked, and what would be the implications of regulatory reform on smaller communities.⁹ Breyer (1999) sums up the role he saw economists had played in bringing up the case for reforming airline economic regulation:

The hearings were presented with considerable amounts of factual information. For instance, it was found that East Coast Boston fares were twice those for comparable-length flights and to similar-sized cities from San Francisco on the West Coast. That there had been no new entry into the industry over the previous forty years was highlighted. Studies comparing intrastate airline routes and regulated interstate airline routes were brought up, and the former's lower fares, higher load factors, and more seats were linked to the poor competitive performance of the legacy airlines. As Bailey (2002) points out, the case for deregulation was also enhanced by the free market arguments made by cargo airlines during the debates over the 1977 Air Cargo Deregulation Act and by the positive experiences of the CAB from 1975 in liberalizing charter services, expanding competitive route authorities, and experimenting with fare competition, including the off-peak “peanuts fare” by Texas International and American Airlines’ “Super Saver” tickets.

The outcome of the committee's opining was that the interstate airline market became deregulated. In addition to the intellectual arguments offered, air travel was highly visible and in the 1970s was seen as a glamorous industry largely reserved for the wealthy, and the theoretical underpinnings of its regulation were particularly questionable. Economically, the industry was also seen as underperforming. In political terms, it had a relatively simple vested-interest structure involving firms and unions with limited political influence, and the success of unregulated intrastate airlines offered a powerful demonstration effect for politicians to use against this. Airlines were also seen as a stand-alone industry, unlike, for example, the railroad, where operations and track were combined in single firms, making the airlines an easier subject for deregulation.

There was little transition time between the passing of legislation and its phased enactment. In this, the US big-bang approach differed significantly from the gradualism later deployed in many other airline markets, such as that of the European Union, which involved three separate “packages” of reform (Button and Johnson 1998). The US approach, however, gave limited opportunity for airlines to adjust and inevitably left them with considerable stranded cost. But equally, it gave them limited time to lobby for dilution of the reforms. The extant aircraft fleets, service features, land-based infrastructure, etc., were designed for a largely fragmented series of services rather than an integrated network generating economies of scope and density. The CAB was already allowing some discounted fares beyond the traditional coach and first-class options, and in 1977 downward flexibility of 50 percent in the standard fare was permitted without justification, and 70 below for 40 percent of a carrier's weekly capacity. After this, fares were set by market forces. The act formalized this and also circumscribed upward flexibility within a ceiling.

The opening up of markets created potential for new entrants, including low-cost carriers, and the expansion of intrastate and regional carriers into the national market. The legislation meant that the CAB would automatically certify entry, unless doing so damaged the public interest. In addition, antitrust policy was modified, with consideration of mergers no longer being based on the combined market share of the airlines involved. The Department of Justice, although allowed to comment, had no direct involvement in this. The CAB ceased to exist in 1985.

Examining the Outcomes of Deregulation

A slew of empirical studies were conducted on the implications of the 1978 act. The general views regarding the overall effects of the 1978 act can be summed up by Kahn (2004: 3): “I begin by baldly stating my essential conviction: airline deregulation has been a nearly unqualified success, despite the industry's unusual vulnerability to recessions, acts of terrorism, and war.” A similar position was echoed by Bailey (2002), who argued that the move from economic regulation delivered on its efficiency promises, generating significant benefits to consumers in terms of lower fares and more convenient schedules. At the international level, comparing overall levels of productivity, Good, Röller, and Sickles (1993) found US airlines were far more productive than their European counterparts.

Quantification in the short term was somewhat hampered because of the volatility often associated with an intermediate product. This was especially an issue in the late 1970s, when a sudden fuel cost shock, accompanied by a global economic recession, hit the US aviation market at a time of its overexpansion in the wake of deregulation. Nonetheless, in other ways the natural deregulation experiment was a good practical subject for study. The transition period was relatively short, and the authorities maintained important datasets after the removal of pricing and entry controls. Not only were data available from the sunsetting phase, but immediate, official data were later dovetailed into a compulsory 10 percent, quarterly ticket sample from 1993.

The quantitative analysis of airline deregulation may also be seen as a segue into the wider use of empirical economics and the types of natural experiments associated with the difference-in-differences work on labor markets by Card and Krueger (1994). Although the context inevitably affected the detail, much of the airline analysis, and notably that at the Brookings Institution, used time-series data to compare individual airline markets disrupted by regulatory changes with a counterfactual where the background conditions essentially stayed the same.

In the context of simple trends, the immediate effect of the 1978 act was that real air fares declined for the first time since 1966, and revenue passenger miles expanded faster than they had in over a decade. Load factors rose from around 60 percent in the 1970s to over 70 percent by the 1990s, as passenger numbers increased from 254 million in 1978 to 582 million in 1999. Some routes previously supported by the CAB's cross-subsidy fare-setting policy lost services, or experienced rate increases, but these were small in number. Table 3 compares actual post-deregulation fares by distance and market size in the fourth quarter of 1983 with fares estimated had the SIFL formula been applied. Besides an overall fall, the benefits were distributed in favor of longer distances and larger markets.

Moving to a more analytical analysis, in terms of finances, Morrison and Winston (1986) used data from 1981 through 1983 to backcast what 1977 profits would have been without regulation. Their findings suggested that they would have exceeded actual regulated profits by $4.5 billion in 1988 dollars. In terms of traveler benefits, to back predict 1977 deregulated fares, the relationship between fares and fuel and labor costs was also estimated for the deregulated period 1980 to 1982 to calculate a fare-deflator for 1983. Using this, fares in 1977 would have been nearly 30 percent lower if they had been unregulated.

While early the evidence indicated that deregulation produced significant economic gains, the role of contestability in influencing airline behavior was murky. Certainly, there was evidence that markets were relatively easy to enter and exit, for instance, between July 1, 1978, and July 1, 1979, there was a gain of 117 new nonstop trunk and local services, but this involved 449 additions and 332 deletions. Such results may have provided evidence to support the power of potential contestability, but they also fitted with the neoclassical model of actual competition, for example, the average number of airlines serving a route rose from 1.5 in 1978 to between 1.9 and 2.0 in the late 1990s.

But this pattern was to be expected in the short term, according to Bailey and Baumol (1984), who point out that the condition of contestability that incumbents’ prices are relatively sticky is not met in the airline industry. In many cases, incumbent carriers respond immediately to meet a new competitor's lower fares. Hence is not surprising that the evidence on pricing policy following deregulation is more complex. It would be only after a transition period that potential rather than actual competition should serve to police markets. Allowing for this, Bailey and Panzar (1981: 145) concluded that “we do feel that the admittedly scanty evidence during the first year after deregulation is consistent with our theory that airline markets are basically contestable. . . . The empirical evidence of late 1979 and early 1980 does, in fact, bear us out.”

This positive position, however, changed as the transition phase ended. The views of several who had been engaged in formulating the legislation began to focus much more on the dominant influence of actual, rather than potential, competition. Kahn, always somewhat skeptical of the influence of potential competition, observed,

This view was echoed by Levine (1991), who felt that while much of the underlying theory of deregulation was fairly closely related to notions of the theory of contestability, contestability did not seem applicable to the airline industry. Even Bailey accepts that when there are economies of scope, this does not automatically imply free market entry and exit in a network industry. This also depends upon the residual effects that regulation has imposed on incumbents disadvantaging them compered to potential new entrants. As to the nature of the imperfections, she found that while deregulation did deliver lower prices and provided more convenient schedules, airline competition, and notably that at hub airports, remained imperfect (Bailey 2002). The magnitude and forms of these imperfections in airline markets were more in line with models of oligopolistic behavior than of contestability.

Capping it all, two of the people most closely associated with the contestable concept, while not abandoning the notion of potential competition entirely, concluded by the mid-1980s that, although they initially cited the airline industry as a case where in a deregulated environment, hit-and-run entry was possible because investments in aircraft do not incur any sunk costs, airplanes (to paraphrase Alfred Kahn) constitute capital on wings. However, “reconsideration has led us to adopt a more qualified position on this score. We now believe that transportation by trucks, barges and even buses may be more highly contestable than passenger air transportation” (Baumol and Willig 1986: 24).

These observations were largely based on simple data analysis, and often represented comparative cross-sections, rather than being time series. Initially, it had proved difficult to conduct more rigorous statistical analysis. Additionally, it did not help that the theoretical work on contestability had provided limited insights into ways of determining empirically whether a market is imperfectly contestable, and certainly not on how to measure various forms of imperfections. Separating the effects of potential competition from actual competition was, thus, a major problem.

By the mid-1980s, the market had settled enough for more rigorous examination of the airline concentration ratios and the like, and more refined methods of analysis were emerging. The argument here was that if markets were contestable, then concentration of services should have little effect on fares, a condition, however, that was largely found not to be met. Moore (1986), for example, found a substantial price effect when the number of carriers reached five; the effect of having more than four carriers in a market reduced rates by 24 to 29 percent when 1983 fares were compared to those in 1976. Graham, Kaplan, and Sibley (1983) found that market concentration had a positive bearing on fares in relatively unconcentrated markets but not much incremental effect on others. Call and Keeler (1985) also found airline markets not fully contestable, with greater concentration being associated with higher fares, although the size of the difference, 5 to 10 percentage points, was not large. Bailey's (1985) work supported these findings.

An alternative approach to considering the role of latent competition draws upon an insight provided by Bailey and Baumol (1984). They point out that potential threats to incumbents are unlikely to appear as manna from heaven but rather from “the availability of a pool of potential entrants able to respond quickly to an entry opportunity and to choose the timing, place, and manner of entry that best suits the circumstances” (120). The effects on an incumbent of having an airline already operating from airports at one or both ends of a route, but not the route in question, is a guide to the power of potential competition. The cost for the latter carriers of adding competitive services on the incumbent's route is minimal; at the extreme, it is just its “ramp-up” costs. Examining changes in the behavior of incumbents after deregulation offers guidance to the effects of contestability on outcomes.

Deploying a synthesized measure of welfare change and studying some seven hundred routes, Morrison and Winston (1989) find that the pure theory of contestability is not applicable to the airline industry. But they do find support for Bain's notion of imperfect contestability in that potential carriers can affect welfare. Some combination of some version of the dominant form model and imperfect contestability seemed to them to characterize competition in the liberalized US industry. Later, Morrison (2001) again deployed a form of difference-in-differences approach to study fares of incumbent carriers on routes where Southwest operated from one or both the airports the incumbent served, but neither directly competed on the routes of interest nor offered service on adjacent routes. If Southwest served both airports, it had a 33 percent effect on incumbents’ fares compared with just 6 percent when only serving a single airport. It became 46 percent when there was actual competition, findings largely supporting earlier work by Dresner, Lin, and Windle (1996).

The Handling of Sunk Costs

An assortment of factors have been cited as potentially thwarting the strong contestability of US airline markets, some already in place pre-1978 but others that emerged in the dynamics of carriers responding to the new situation. Such impediments included contractual wage agreements that incumbents had entered into before 1978 when new entry was not a threat (Meyer and Tye 1988) and the costs sunk in advertising service networks, with potential newcomers having to devote similar nonreturnable resources to compete with incumbents. The use by travel agents of computer reservation systems (CRS) supplied by carriers such as American Airlines (Sabre system), developed pre-deregulation for handling booking, was also problematic. Multivending agents who began using them after 1978 tended to be biased toward booking from early pages of information containing flights offered by CRS's airline owners.¹¹

The advantage of incumbency is further extended if the airline in situ operates over a large network of services (Borenstein 1991). When passengers have to make a connecting flight, they prefer that the connecting flight be with the same airline as the first flight; thus as the proportion of online connections grows, so perceived service quality increases. The emergence of airline loyalty (frequent flyer) programs magnified this capture effect (Cairns and Galbraith 1990), and this was enhanced from the 1990s as US carriers began to form strategic alliances with foreign airlines (Lin 2008).¹²

Perhaps the main factor stymieing the forces of latent competition, however, was that the 1978 legislation did nothing to remove imperfections in the allocating of sunk costs inherent in other parts of the aviation supply chain (Kahn 1988b) and, notably, those associated with airports and air traffic control. This, in particular, helped create fortress-hub airports whereby an incumbent carrier dominates landing and take-off slots and gates at key locations, leading to hub premiums being levied on passengers.¹³ Bailey (2002), for example, complained that airline competition at most hub airports remained imperfect, and as a result the airline market may be better explained by oligopolistic behavior than by contestability theory.

Theorists had not been unaware of the sunk cost nature of landing slots and the like. They tackled this problem by embodying Harold Demsetz's (1968) notions of creating competition for the market rather than competition in the market. This, in the runway slot case, would involve the owner of the capital auctioning off capacity to carriers for a designated period. In operational terms, much of the devil is in the detail of the auction mechanism adopted.

Traditionally, slots had been allocated on a first-come, first-served basis, with fees based mainly on aircraft weight, a reflection of the damage done to runways and aprons by the movement of planes. The fees were less reflective, however, of the congestion emerging as the industry grew. The federal authorities were not unaware of this and of the public's concern over their effects on flight punctuality. In 1961 scheduling committees were established to institutionally match supply with demand for runway capacity at congested airports. By the 1970s, the CAB was cognizant that this was not working well. Given the lack of historical experience, early experimental economics was applied to assess various auction options.

A group from the California Institute of Technology involving Charles Plott, David Grether, and Mark Isaac, and later Vernon Smith, was commissioned by the CAB to study the ways scheduling committees were affecting slot-allocation processes (Grether, Isaac, and Plott 1981). A number of economic experiments were conducted, and their outcomes provided a basis for proposing a more efficient, smart computer-assisted exchange institution (Svorenčík 2017). These experiments suggested a separate market for slots at each airport allocated through regular sealed-bid auctions. To allow for the auctions’ lack of provision for slot complementarities, after-market computerized, oral, or double auctions would follow (Rassenti, Smith, and Bulfin 1982).

Auctions for slot use, however, were not part of the high-density rule introduced at some airports in April 1986. Instead, slots could be bought, sold, or leased for any consideration and time period. Potential entrants argued, though, that this impeded new entry, and, as a consequence, in 1994 the USDOT sanctioned the granting of exemptions from the high-density rule. However, this had little impact because the slot transactions were usually conducted between financial institutions and incumbent carriers, or between related carriers, rather than involving new entrants (Fukui 2010). The specifics of the institutional reforms needed to release the forces of potential competition, despite the evidence from the experiments, were not adopted.

Conclusions

One objective of this article has been to provide a case study that partially fills gaps that Backhouse and Biddle (2000) diagnosed in their history of applied empirical economics, namely, the role of economists on the crafting of policy (including the regulation of utilities) and the relations between, on the one hand, economists working in government and business, whose role was solely to “do” applied economics, and, on the other hand, academic economists. The development of the theory of contestability, with the supposed greater realism of its behavioral assumptions and the subsequent empirical analysis of its implications for airline deregulation, forms the basis for study.

Our experiences with the development of contestability theory from the 1970s to the early 2000s seem to confirm George Box's view that all models are wrong, but some are useful. While contestability theory, developed by an eclectic group of theorists and practitioners, initially attracted considerable attention, it did not live up to Baumol's expectations as an “uprising in the theory of industry structure.” The notion of contestability had arisen in the US telecommunications industry with the dominant supplier questioning the legal and de facto constraints being imposed on its behavior. AT&T sought to broaden industrial theory by suggesting a way in which the benefits of competitive markets can be combined with those of economies of size and multiproduct production. In this way the theory was designed as part of a critical, abstract examination of the ways that governance was then being exercised by regulatory US authorities over the telecommunications sector. In this it was not entirely successful.

Contestability theory did, however, subsequently exert influence on the regulation of US airlines and the passing of the 1978 Airline Deregulation Act. The policy prescription was to favor measures that redirected attention from market share to matters involving the ease of entry to, and exit from, markets, appreciating the practical problems in creating conditions that facilitated this. The natural experiment, and a not insignificant amount of subsequent applied empirical economic analysis, indicates, however, that the power of potential competition, in the US airline context at least, is considerably less than many of the theorists had thought it would be.

I would like to thank the editors of HOPE, and especially Kevin Hoover and Paul Dudenhefer, for their extremely valuable help in preparing this article, and the journal's referees for very constructive comments that improved the article considerably.

Notes

References