Contestable Markets: Long-Term Contracts and Market Performance, Lecture notes of Life Sciences

Download Contestable Markets: Long-Term Contracts and Market Performance and more Lecture notes Life Sciences in PDF only on Docsity! The Theory of Contestable Markets Stephen Martin Department of Economics Purdue University smartin@purdue.edu July 2000 2 Chapter 1 Contestable Markets “But he hasn’t got anything on,” a little child said. Hans Christian Anderson, The Emperor’s New Clothes 1.1 Introduction The literature on contestable markets emerged from a research program that claimed two principal achievements in advancing economic knowledge, and two important policy contributions. The theory of contestable markets was advanced as a generalization of the theory of perfectly competitive markets, and a generalization that (in con- trast with the previous literature) endogenizes the determination of industry structure. Thus (Baumol, 1982, p. 2)1 in the limiting case of perfect contestability, oligopolistic struc- ture and behavior are freed entirely from their previous depen- dence on the conjectural variations of incumbents and, instead, these are generally determined uniquely. . . by the pressures of po- tential competition. . . . Further, the theory of contestable markets was presented as suggesting an improved set of guidelines for determining when government intervention 1See also Bailey (1982, pp. xiii, xix), Baumol et al. (1982, pp. 13—4; 1986, pp. 340, 344); Baumol and Willig (1986, p. 11). 5 6 CHAPTER 1. CONTESTABLE MARKETS in the market is called for, and for the conduct of such activity when it is undertaken.2 We begin with a brief statement of the formal results of the theory of contestable markets. We then explore the assumptions that seem necessary to produce these results. This is followed by a review of the progress that the theory of contestable markets has made toward its goals, and of empirical work related to contestability theory. 1.2 Principal results The formal structure of the theory of contestable markets is disarmingly simple, particularly for the case of single-product firms.3 ,4 The essential def- initions are: (D1) an industry configuration is a vector (m, y1, y2, ..., ym, p). m is the number of firms. yi is the output of firm i (i = 1, 2, ..,m). p is the price that clears the market: Q(p) = y1 + y2 + ...+ ym). (D2) A configuration is feasible if production is sufficient to meet demand, and no firm is losing money. (D3) A configuration is sustainable if it is feasible and no potential entrant can cut price and make a profit supplying a quantity less than or equal to the quantity demanded at the lower price. (D4) A perfectly contestable market is a market in which sustainability is a necessary condition for equilibrium. (D5) A configuration is a long-run competitive equilibrium if it is feasible and there is no output level at which any firm could earn an economic profit at the prevailing price. 2Baumol (1982, p. 14); Baumol et al. (1982, pp. 476—83); Baumol and Willig (1986, p. 11). 3See Baumol et al. (1982; 1986, pp. 341—7). Our discussion follows Spence (1983). See also Shepherd (1984, 1987, 1988). 4The discussion that follows can be extended to multiproduct firms by interpreting y1 as a vector of outputs, p as a vector of prices, and Q(p) as a vector of demand functions. See Waterson (1987). 1.2. PRINCIPAL RESULTS 7 The main results follow almost immediately from the definitions. They are as follows. (R1) A long-run competitive equilibrium is sustainable. By (D5), in long-run competitive equilibrium there is no output level that earns an economic profit at the prevailing price, and the existing configura- tion is feasible. This satisfies (D3), the definition of a sustainable equilibrium. (R2) A sustainable configuration is not necessarily a long-run competitive equilibrium. p Q 1 2 3 4 5 6 7 7 • • • • • • • • • • ................................................................................................................................................................................................................. ............................................................................................................................................................................ ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... ..... .. ..... ..... ..... ..... ..... ..... ... Figure 1.1: Sustainability versus long-run competitive equilibrium It suffices to provide an example of a sustainable configuration that is not a long-run competitive equilibrium. Let the demand curve be p = 7 − Q. Let the cost function be c(q) = 4 + 2q. Then (m, y1, p) = (1, 4, 3) is a sustainable equilibrium. If one firm sells 4 units of output, average cost is 3. 3 is also the price that clears the market. In this configuration, output equals 10 CHAPTER 1. CONTESTABLE MARKETS as those available to the incumbent firms. . . . Second, the poten- tial entrants evaluate the profitability of entry at the incumbent firms’ pre-entry prices. and (Baumol, 1982, pp. 3-4) A contestable market is one into which entry is absolutely free, and exit is absolutely costless. . . . the entrant suffers no dis- advantage in terms of production technique or perceived quality relative to the incumbent, and that potential entrants find it ap- propriate to evaluate the profitability of entry in terms of the incumbent firms’ pre-entry prices. . . . The crucial feature of a contestable market is its vulnerability to hit-and-run entry. Much of the literature dealing with contestable markets has focused on the circumstances under which it is plausible for incumbents to believe that entrants could realistically engage in rapid and reversible (hit-and-run) entry. If incumbents do not find such behavior credible, potential entry does not constrain the actions of incumbents. 1.3.1 Entrants act as price takers The results of contestability theory require not only that rapid entry and exit be possible, but that potential entrants make their decisions taking the market price as given. Thus definition (D3) defines sustainability in terms of entrant profitability given the number of incumbents, their output, and the price at which that output clears the market. Under the definition of sustainability, the entrant is not permitted to take account of the price re- duction that its own output will produce when it assesses the profitability of entry. The entrant is not permitted to take account of possible reactions of incumbents. Hit-and-run entry is supposed to occur if the potential entrant could make a profit at the pre-entry price. If the potential entrant comes into the market only if it could make a profit at the expected post-entry price, hit-and-run entry is much less plausible. But if hit-and-run entry is implausible, there is no dynamic mechanism to enforce the static results of the theory of contestable markets. 1.3. CRITICAL ASSUMPTIONS 11 In their original discussion, Baumol et al. make a comparison between the entry assumptions of contestability theory and those of the standard model of perfect competition (1982, p. 5): That is, although the potential entrants recognize that an expan- sion of industry output leads to lower prices-in accord with the market demand curves-the entrants nevertheless assume that if they undercut incumbents’ prices they can sell as much of the corresponding good as the quantity demanded by the market at their own prices. This is an extension of the axioms on entry behavior in the classical model of perfect competition that makes it possible to deal with the small-numbers case. The assumption of price-taking behavior is plausible for a model that describes the behavior of many small buyers and sellers. It is not plausible for markets with a small number of firms (in the limit, one or two actual firms with one potential entrant). It is clear that Baumol et al. recognize this (1982, p. 11): Bertrand-Nash [price-taking] expectations are not always fulfilled, and in some cases they are unlikely to be. For. . . competitive entry can impose significant losses upon incumbents and thereby force a change in their prices. However, if an entrant’s output is “small” relative to that of the industry, the magnitude of these required adjustments may also be “small”, and hence it may be justifiable for the entrant to ignore them. On this rationale, the theory of contestable markets can no longer be said to apply where technology requires firms to be large relative to the market. The theory of contestable markets applies where efficient firms can be so small that they make decisions taking price as given. This, of course, is the usual size condition imposed for applicability of the theory of perfectly competitive markets. If this size condition must be imposed before the theory of contestable markets is to be valid, it is difficult to see how it can be claimed that contestability theory extends the theory of perfect competition to markets where economies of scale are important. Thus (Baumol and Willig, 1986, p. 17) 12 CHAPTER 1. CONTESTABLE MARKETS the critical issue that remains is the determination of the circum- stances under which the Bertrand-Nash [price-taking] assumption holds or at least is assumed by the participants to hold approxi- mately. Most economists would reject the claim that price-taking behavior is a reasonable assumption to make about oligopolistic markets (Friedman, 1982, pp. 527—8): The entry or exit of one firm from such a market is likely to be followed by a large discrete change in the policies being pursued (prices, etc.) by the firms which are active before and after the change. Even if none changed its behavior, amounts demanded of each firm would be much different after the alteration in the number of firms. That is to say, for the competitive firm it is reasonable for it to suppose that the prices, profitability, etc. of a given market are independent of whether it is in the market. For an oligopolist, however, it is abundantly clear that these variables must depend, in part, on whether or not it is active. 1.3.2 Exit lag Baumol, Panzar, and Willig point out that an entrant need not believe prices are fixed forever in order for the results of contestable market theory to hold. The entrant need only believe that prices will not change for the duration of its stay in the market (Baumol et al., 1983, p. 493; emphasis in original): To produce its results, even the limiting case of perfect contesta- bility does not require entry and exit to be instantaneous. Rather, it is sufficient that the process be rapid enough so that the entrant does not find his investment vulnerable to a retaliatory response by the incumbent. The length of this time period is not exclusively a technological datum, but is also the result of business practice and opportunities in the market in question. The observation that the length of time before an entrant would expect incumbents to respond reflects business practice and opportunities would seem to conflict with the claim that performance in contestable markets is determined in a way that is independent of oligopolistic interactions. 1.3. CRITICAL ASSUMPTIONS 15 If scale economies are large, an entrant would have to inform and nego- tiate contracts with a large fraction of the market to insulate itself from a retaliatory price response. Transactions costs rooted in imperfect and im- pacted information make negotiations difficult. The transaction costs reflect the fact that consumers have no experience with the potential entrant’s prod- uct, and they have no way of knowing whether or not the claims the entrant makes about its ability to perform can be relied upon. Such transactions costs — an investment in information and reputation — are largely sunk in the market, irrecoverable upon exit.10 Furthermore, in a market where potential entrants could negotiate con- tracts in advance of entry, the same option would be available to incumbents. Incumbents could defeat a potential entrant’s strategy by negotiating con- tracts promising to “undercut any legitimate price” (Shepherd, 1984, p. 576, fn. 12). The possibility that potential competition in the offering of contracts insures optimal market performance seems unlikely.11 1.3.4 Absence of sunk costs Suppose rapid entry and exit is possible, that incumbents will not alter price in the event of entry, and that potential entrants know this. Suppose also that to enter at all, a firm would have to make an irrecoverable investment of 1,000,000. The investment might reflect the acquisition of information about the market before the decision to enter is taken. The investment might reflect the purchase of physical assets that could not be resold or transferred to other markets in the event of exit. So long as the firm remains in the market, the assets (tangible or intangible) are productive, and the firm enjoys the value of the marginal product of the assets as part of its income stream. If the firm exits, the investment is left behind. It follows that incumbents, for which any such sunk investments lie in the past, can earn a combination of rent and economic profit with a present discounted value of up to 1,000,000 without inducing entry. Very early in their book, Baumol et al. emphasize the importance of sunk costs for market contestability (1982, p. 7):12 10Essentially the same arguments rule out long-term contracts as a device to combat predatory pricing. 11Schwartz and Reynolds (1983, p. 490); Brock (1983, pp. 1061—4); Schwartz (1986, pp. 52—5). 12Cairns and Mahabir (1988) present a sharply different analysis of the impact of sunk 16 CHAPTER 1. CONTESTABLE MARKETS Clearly, when entry requires the sinking of substantial costs, it will not be reversible because, by definition, the sunk costs are not recoverable. However, if efficient operation requires no sunk out- lays, then entry can, by and large, be presumed to be reversible, and the market can be presumed to be contestable. Various definitions of sunk cost appear in the contestability literature Thus (Baumol et al., 1982, pp. 280-1) Definition 10A1: Long-Run Fixed Cost Long-run fixed cost is the magnitude F (w) in the long-run total cost function CL(y,w) = δF (w) + V (y,w) δ = ½ 0 if y = 0 1 if y > 0 where lim y→0 V (y,w) = V (0, w) = 0. V () is nondecreasing in all arguments, and y and w are, re- spectively, the vectors of output quantities and input prices. Definition 10A2: Let C(y, w, s) represent the short-run cost function, applicable for the flow of production, that occurs s units of time (years) in the future. Then, K(w, s) are the costs sunk for at least s years, if C(y, w, s) = K(w, s) +G(y,w, s) G(0, w, s) = 0. Here, since in the long-run no costs are sunk, lim s→∞ K(w, s) = 0. costs on market performance. They argue that the most likely potential entrants are firms with sunk assets in related markets, and suggest that contestability theory best describes competition among multiproduct firms. 1.3. CRITICAL ASSUMPTIONS 17 What is called “costs sunk for at least s years” in Definition 10A2 looks very much like short-run fixed cost, where the short run lasts s years.13 Nor is it clear why some costs might not be sunk even in the long run.14 A quite different definition of sunk costs, that permits costs to be sunk over the long run, is given by Baumol et al. (1983, p. 494): suppose that a unit of capital purchased at a price of β per unit could be sold or utilized elsewhere. . . for a unit salvage value of α ≤ β. Thus it is possible to parametrize continuously the degree of sunkenness of capital from zero (α = β) to absolute sunkenness (α = 0). According to this definition, sunkenness depends on the nature of resale markets for capital assets. But if this is what determines whether or not costs are sunk, the possibility that sunk costs are completely absent — as perfect contestability requires — seems extremely limited. Consider first physical assets. Many businesses require highly specific physical assets that might be resold at a substantial loss upon exit or not at all. A firm that wished to leave a retail food market because of low or negative profitability might recover some of its investment in kitchen and other equipment by selling it in that market. If the market is so depressed that the firm has decided to exit, it will most likely sell at a loss. The exiting firm might get a better price by shipping the physical assets to another market, which would involve the expense of transportation. The more specific the physical assets, the greater the extent to which the investment in the assets is sunk. Other businesses involves assets that are not specific — they can be used as inputs to produce many different goods. Trucks and delivery vans are examples. Resale markets for such assets, however, suffer from the “lemon problem” of Akerlof (1970) — a consequence of imperfect and impacted infor- mation. 13Weitzman (1983) argues that without sunk costs there cannot be truly fixed costs. In the absence of sunk costs, an entrant could come into a market, operate very briefly at minimum efficient scale, and exit. Provided the good can be stored, any production that occurs will take place at minimum efficient scale. But then returns to scale are effectively constant. 14Investment in research and development creates an asset — knowledge — that is, at least potentially, productive forever. Much of the value of such knowledge, if highly specific and tied to the operations of the firm, would be lost upon exit. 20 CHAPTER 1. CONTESTABLE MARKETS 1.3.6 No product differentiation Baumol et al. argue (1982, pp. 329—32; 1986, pp. 355—9) that the the- ory of contestable markets applies to monopolistically competitive markets in which products are differentiated.16 This claim is based on an unusual interpretation of the theory of monopolistic competition. • p q D D d d ............................................................................................................ .................................. ......................................................................................................... Average cost .................................... .............................................................................................................. Figure 1.2: Long-run equilibrium, monopolistic competition Figure 1.2 is a standard illustration of long-run equilibrium in Chamber- linian monopolistic competition. Compare Baumol et al. (1982, figure 11F3). dd is the demand curve for a single variety in a product group if all firms hold price constant. DD is the demand curve for a single variety if all firms match price changes. It is customary to describe the situation depicted in Figure 1.2 as a long- run equilibrium. Each firm is maximizing profit: marginal cost equals mar- ginal revenue along the demand curve dd (for simplicity, the marginal revenue and marginal cost curves are omitted from Figure 1.2). Since each firm is 16We discuss horizontal product differentiation; see Lambertini (1992, 1996) for a dis- cussion of contestability theory and vertical product differentiation. 1.3. CRITICAL ASSUMPTIONS 21 maximizing profit, there is no incentive for any firm to alter its own behavior. At the same time, economic profit is zero (price equals average cost). There is no incentive for entry or exit. With no incentive for entry or exit, and no incentive for incumbents to alter behavior, the configuration illustrated in Figure 1.2 will persist until some other factor alters demand or cost condi- tions. This is usually thought of as an equilibrium. Baumol et al. disagree, if the market is contestable (1982, p. 332; footnote omitted; emphasis added): However, such a position cannot be an equilibrium in a con- testable market. An entrant can closely or exactly duplicate the product design of the firm depicted, and enter at a lower price. In a contestable market, with an arbitrarily short lag in incum- bents’ price reactions, he would expect demand to behave in ac- cord with curve dd during the period of the lag. Since dd is more elastic than the DD curve that is tangent to the A[verage]C[ost] curve, the dd curve necessarily cuts into and above the AC curve, as illustrated. Consequently, there exist (temporarily) profitable entry opportunities, and so “high-tangency equilibria” cannot be equilibria in contestable markets. It should be noted that Cham- berlin’s discussion implies strongly that the markets of which he was thinking satisfied the free-entry requirements of contestabil- ity. Thus, it is of some significance for his analysis that his high- tangency solution turns out to be unsustainable. But the distinguishing characteristic of monopolistic competition is prod- uct differentiation. Under monopolistic competition, it is impossible for an entrant to duplicate exactly the variety of an existing firm (Chamberlin, 1933, p. 56): A general class of product is differentiated if any significant basis exists for distinguishing the goods (or services) of one seller from those of another. Such a basis may be real or fancied, so long as it is of any importance whatever to buyers, and leads to a preference for one variety over another. . . . Differentiation may be based upon certain characteristics of the product itself, such as exclusive patented features; trade- marks; trade names; peculiarities of the package or container, if any; or singularity in quality, design, color, or style. It may 22 CHAPTER 1. CONTESTABLE MARKETS also exist with respect to the conditions surrounding its sale. In retail trade. . . these conditions include such factors as the conve- nience of the seller’s location, the general tone or character of his establishment, efficiency, and all the personal links which attach his customers either to himself or to those employed by him. In so far as these and other intangible factors vary from seller to seller, the “product” in each case is different. . . . As product differentiation has usually been characterized by economists, it would be impossible for a hit-and-run entrant to duplicate the variety of any existing firm exactly. Close approximation might be possible, but it would not produce the results of the theory of contestable markets, as Bailey and Baumol acknowledge (1984, p. 117, fn. 10): A problem can arise if products in a perfectly contestable industry are heterogeneous, each supplier offering his own special brands with their own special features. However, it can be shown. . . that if each variant is sold by at least two different suppliers, perfect contestability will lead to marginal cost pricing. By definition, however, if products are differentiated it is impossible for any variant to be produced by two suppliers. Kellogg’s Corn Flakes and Martin’s Corn Flakes are not the same product, even if they are physically identical (Caves, 1971, p. 5):17 In the nature of differentiation, a successful (rent-yielding) prod- uct variety is protected from exact imitation by trade markets, high costs of physical imitation, or both. Hit-and-run entry is not possible where products are physically identical but bear different brand names. Aspirin is the generic example. As Far- rell (1986) points out, if varieties produced by entrants and incumbents are identical but buyers are uncertain about the quality of entrants’ products, buyers will be reluctant to patronize an entrant, all else equal. The result is an entry barrier that impedes hit-and-run entry.18 17See also Friedman (1982, p. 501): “At the heart of differentiated products models are the assumptions that no two firms produce identical products and that firms can be grouped according to the type of product they make.” 18See also Seabright (1990, pp. 20—35). 1.4. GOALS 25 But the theory of competitive markets produces the usual efficiency re- sults only in long-run equilibrium. Economic profits or losses are quite pos- sible in the short run in competitive markets. The theory of competitive markets has a well-established theory about passage from the short run to the long run. If in an initial long-run equilib- rium there is an exogenous outward shift in the demand curve, incumbent firms earn economic profits in the short run; these profits attract entry; the industry supply curve shifts out, price falls, profit is reduced. This process continues until profits are eliminated. There is a similar story with exit if in the short run incumbent firms suffer losses. The theory of contestable mar- kets, which is acknowledged to be strictly static in nature, lacks the short-run and dynamic implications of the theory of competitive markets. In this sense, the theory of contestable markets is less general, not more general, than the theory of competitive markets. Contestability theory is particularly defended as providing a welfare stan- dard when the technology mandates an oligopolistic market structure (Bailey and Baumol, 1984, p. 119): while perfectly competitive and perfectly contestable markets are both ideals, the latter is more ideal than the former. After all, one must be tempered in one’s praise of the many-firm structure of perfect competition in those cases in which the availability of economies of scale and scope means that an oligopoly structure can (perhaps) achieve far lower costs and offer far lower prices to consumers. Two remarks ought to be made. First, it seems clear by now that the assumption of costlessly reversible entry, by itself, is insufficient to produce the results of long-run equilibrium in a perfectly contestable market. Either entrants must make decisions on the assumption that price will not change after entry, or it must be possible for entrants to negotiate contracts, before entry, that insulate them from post-entry responses by incumbents. There must be no sunk costs, not even sunk costs associated with the process of collecting information about the target market. Incumbents must believe that entrants who resort to financial markets would be able to raise capital on the same terms as incumbents. Products cannot be differentiated. Where these assumptions fail-which seems likely to be almost everywhere in the economy-one must be tempered in one’s praise of the costless entry-and-exit requirement of perfect contestability. 26 CHAPTER 1. CONTESTABLE MARKETS Second, the bulk of empirical evidence is that economies of scale are not important in modern economies. Average cost curves in most industries appear to flatten out at relatively small market shares, and observed levels of market concentration exceed, often substantially, that which can be explained in terms of economies of scale.19 Yet the theory of perfect contestability does generalize a received body of economic theory – Bertrand’s (1883) model of price-taking oligopoly with standardized products.20 Bertrand obtains the efficiency results of long-run competitive equilibrium under assumptions strikingly similar to those of the theory of contestable markets. In Bertrand’s model, products are standard- ized. Firms set price taking rivals’ prices as fixed. The entire market is assumed to shift from one supplier to another in response to tiny price dif- ferences. If there are at least two suppliers in the market, performance is optimal. The theory of imperfectly contestable markets, on the other hand, is now acknowledged to be an extension of the mainstream structure-conduct- performance school of industrial economics (Baumol et al., 1983, p. 494):21 models which support the robustness of contestability analysis follow a relatively long tradition going back at least to the work of J. S. Bain. This tradition holds that increased ease of entry and exit improves the welfare performance of firms and industries. On this subject, the theory of contestable markets has only sought to contribute insights on the underpinnings of that judgment. The tradition referred to also holds that difficulty of entry allows incum- bent firms to exercise some market power, and that market performance depends on oligopolistic interactions as well as potential competition. 1.4.2 Endogenization of industry structure One of the primary claims of the theory of contestable markets is that, in contrast with the previous literature, it endogenizes industry structure. This claim seems doubtful on two counts. 19Scherer (1974). Baumol et al. recognize that average cost curves are typically found to flatten out (1982, fn. 50). 20The relation of contestability theory to Bertrand’s work is emphasized by Knieps and Vogelsang (1982). 21See also Schwartz (1986, p. 38). 1.4. GOALS 27 First, the literature before contestability did not treat market structure as exogenously given. Second, the sense in which perfectly contestable markets provides a theory of market structure is extremely limited. Analysis of market structure before contestability theory There is a large literature in which economists develop theoretical and empiri- cal models of market structure. A common simplifying assumption in models of oligopoly is that firms are symmetric — that they produce the same out- put in equilibrium. In this case, the critical element of market structure is the number of firms in the industry. Long before contestability theory, economists modeled the long-run equilibrium number of symmetric firms in a market on the assumption that entry occurs in response to excess profits.22 The literature on dynamic limit pricing23 analyzes the optimal pricing strategy for a dominant firm that faces the possibility of fringe entry, over time, if price exceeds a critical level. An essential result of dynamic limit pricing models is that a dominant firm without a cost advantage will set a high short-run price and gradually lower price to a level that no longer induces entry. A high short-run price generates short-run profits but future loss of market share. Models of dynamic limit pricing yield predictions about the time path of the dominant firm’s market share and about the equilibrium shares of the fringe and the dominant firm. When firms are not symmetric, market share is a critical element of market structure. There is a large empirical literature that endogenizes the determination of industry structure (Weiss, 1963a; Shepherd, 1964; Carter, 1967; Ornstein et al., 1973; Mueller and Hamm, 1974; Strickland and Weiss, 1976; Martin, 1979; Caves 1981b). Indeed, one of the few unchallenged empirical regulari- ties in industrial economics is that, roughly 2 years after the appearance of a new Census of Manufactures, an econometric study will be published analyz- ing changes in US concentration. Nor is this literature limited to examining US data (Shepherd, 1966; Jenny and Weber, 1978). This literature is much older than contestability theory. The exponents of contestability theory have admitted that with respect to the determination of industry structure, the empirical literature has contributions to make which 22Howrey and Quandt (1968); Okuguchi (1972). See also Prescott (1973). 23See Gaskins (1971) and Ireland (1972) for seminal contributions. 30 CHAPTER 1. CONTESTABLE MARKETS need some time to calculate and implement price changes. Perfect contestability is the judgment that the third lag is the longest. . . .The traditional presumption in industrial organization is the opposite, that is, that prices can be changed more quickly than sunk capacity.. . . In practice, careful empirical work in each specific context will have to be undertaken before we can say whether an industry is contestable and sustainable, and decide whether and what regu- latory attention it requires. Industrial economists early recognized the need for careful study before it would be possible to conclude that policymakers could prudently treat any particular industry as if it were contestable. This call for cautious application contrasts with the sometimes casual way in which it has been asserted that the prescriptions of contestability theory can be applied. 1.4.4 Contestability and the deregulated US airline in- dustry The airline industry is a case in point.26 The early contestability literature used passenger airline travel as the flagship example of a contestable market. Thus in 1981 (Bailey, 1981, pp. 179-80): The new policies are based on the theory that both trucking and aviation markets are, in the absence of regulatory intervention, naturally contestable. Capital is highly divisible in the truck- ing industry, and there is every reason to suppose that market mechanisms will work. . . . Even in nondense city-pair markets in aviation. . . potential competition should be able to act as a potent force. This is true because the major portion of airline capital costs, the aircraft, can readily be moved from one market to another. and (Bailey and Panzar, 1981, pp. 128—9) 26It has also been argued that the theory of contestable markets applies to the barge transport industry. For discussion, and a negative assessment, see Tye (1985). 1.4. GOALS 31 Thus, there is no reason, a priori, to expect that economies of scale should lead to substantial barriers to entry in the airline industry because airline capital costs, while substantial, are not sunk costs. . . . the major portion (i.e., aircraft) can be “recov- ered” from any particular market at little or no cost. Such factor mobility makes for ease of potential entry and exit in such mar- kets. . . .Thus, despite substantial natural monopoly attributes, most airline markets are likely to be readily contested. This fact ensures that, even if actually operated by a single firm, their per- formance should approach the competitive norm. . . . The paper just cited is one of the earliest studies of the contestability of the passenger airline industry. It notes a number of factors that might impede contestability of a deregulated industry 1. State and local governments often find it convenient to lease airport facilities to particular airlines on a long-term basis. The airlines so favored can control rivals’ access to airport facilities. 2. Local authorities in some areas limit airport access or growth for rea- sons of noise or pollution control. 3. Authorities ration slots (takeoffs and landings per hour) at the most congested airports. 4. Incumbents that operate many connecting flights from a single airport offer a quality of service that an entrant into a single city-pair market cannot duplicate. The authors report a regression analysis that suggests (Bailey and Panzar, 1981, p. 143) that actual competition of trunks was. . . an effective check on the pricing policies of local service carriers at mileage bands under 400 miles, while potential competition was the check in mileage bands over 400 miles. Potential competition between locals or between locals and commuters was not an effective check on the pricing policies of the locals. . . . the gap in perceived quality of service (jets versus commuter aircraft) meant that commuters were not perceived by locals as potential entrants of sufficient stature to cause them to lower prices. 32 CHAPTER 1. CONTESTABLE MARKETS A number of these findings seem to conflict with the predictions of con- testability theory. It was actual competition by trunk airlines, not potential competition, that limited the exercise of market power by local service car- riers flying 400 miles or less. Potential competition by locals did not limit the exercise of market power by locals. Because of product differentiation — differences in the quality of service — potential competition from commuter airlines did not limit the exercise of market power by local service airlines. The only aspect of the results consistent with contestability theory is that potential competition from trunk airlines appears to limit the exercise of market power by local service airlines flying more than 400 miles. The authors conclude (Bailey and Panzar, 1981, pp. 145): We cannot claim to have done an exhaustive empirical analysis of airline markets in transition. However, 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. . . . In a perfectly contestable natural monopoly market, actual entry is redundant. The mere threat of entry will discipline the market even if it is a natural monopoly. We have argued that long haul airline markets served by local service carriers most closely fit this theoretical ideal. The empirical evidence of late 1979 and early 1980 does, in fact, bear us out. Local service monopolists have been pricing more or less competitively on their long-haul routes. Airlines are again presented as a contestable market in Contestable Mar- kets and The Theory of Industry Structure (Baumol et al., 1982, p. 7): A clear example is provided by small, and therefore naturally monopolistic, airline markets. . . . because airline equipment (vir- tually “capital on wings”) is so very freely mobile, entry into the market can be fully reversible. In principle, faced with a prof- itable opportunity in such a market, an entrant need merely fly his airplane into the airport, undercut the incumbent’s price, and fly the route profitably. Then, should the incumbent respond with a sufficient price reduction, the entrepreneur need only fly his airplane away. . . . Thus, it is highly plausible that air travel provides real examples of contestable markets. 1.4. GOALS 35 If contestability theory applies to passenger airlines, they suggest, it is most likely to do so in the long run. Hurdle et al. (1989) examine the determinants of revenue per passenger mile for a sample of 867 city-pair markets for 1985. They find that the number of likely potential entrants is a significant factor explaining differences across markets. But they find that the number and size distribution of incumbents is a factor as well, unless there are neither economies of scale nor economies of scope. By 1986, contestability advocates had abandoned the position that the airline industry is inherently contestable (Baumol and Willig, 1986, p. 24): In the initial enthusiasm with which we described contestabil- ity analysis we agreed with this assessment, and more than once cited the airline industry as a case in point, using the metaphoric argument that investments in aircraft do not incur any sunk costs because they constitute “capital on wings.” Reconsideration has led us to adopt a more qualified opinion on this score. . . . trucks, barges, and even buses may be more highly contestable than passenger air transportation. Barges and trucks have business firms. . . as their primary customers, and that facilitates the pro- vision of services via contracts on which potential entrants can effectively bid against incumbents. . . . trucks and buses do not face the heavy sunk costs involved in the construction of airports or the shortage of gates and landing slots at busy airports. . . . By this time, it is acknowledged that several features of airline markets make them imperfectly contestable (Baumol and Willig, 1986, p. 24): First, . . . there have been constraining shortages of facilities and services of air traffic control at several pivotal airports. . . . Second, technological advances, changes in relative prices of jet fuel and equipment, and changes in the desired configurations of route networks have significantly altered the types and mix of aircraft demanded by the industry. As a result, there have been shortages in the availability of aircraft demanded. . . .Third, newly certified airlines have been able to avoid the costly labor contracts that pervaded the industry before deregulation, so that their labor costs have been substantially lower than those facing the older established carriers. 36 CHAPTER 1. CONTESTABLE MARKETS To these industry characteristics may be added the fact that incumbent airlines can respond to competitors’ rate changes instantaneously (Stockton, 1988, p. 1; see also Evans and Kessides, 1994): With fare regulation gone and airlines now possessing sophisti- cated computer systems that monitor their own and competitors’ fares, literally hundreds of thousands of fares and the availability of discounts can change with lightning speed-even hourly. If, say, Midway Airlines lowered its fare between Cleveland and Omaha by $10, United, American, and Northwest would learn of it al- most immediately through their computer connections and could react quickly if customers began defecting. Some airline executives claim pricing is so flexible that on some days the industry introduces as many as one million fare changes. In such an industry, it is hardly reasonable to expect that entrants will make decisions in the belief that incumbents’ prices are fixed. The role of actual rather than potential competition in determining mar- ket performance is now admitted (Baumol and Willig, 1986, p. 25):28 Several econometric studies have confirmed the imperfection of the contestability of the airline market. . . . there is a significant positive correlation between profits and concentration in airline markets. Thus the threat of entry does not by itself suffice to keep profits to zero. . . when new entry does occur, established carriers do reduce their fares in response, something one would expect in a conventional oligopolistic market. 1.4.5 Contestability theory as a guide to areas for in- tervention In principle, the message of contestability theory is that policy intervention in market processes is unnecessary, if entry and exit are free and easy. In this case, potential competition as well as actual competition will influence market performance. 28For later studies that confirm the imperfect contestability of the airline industry, see Morrison and Winston (1987) and Bailey and Williams (1988). 1.4. GOALS 37 Pre-merger market shares Post-merger market shares National 0.27 0.51 Texas International 0.24 Delta 0.23 0.23 Continental 0.17 0.17 Eastern 0.07 0.07 Approximate Herfindahl index 0.22 0.35 Numbers equivalent = 1/(approx- imate Herfindal index) 4.6 2.9 Table 1.2: Shares refer to the Houston—NewOrleans market for the 12 months ending 30 June 1978. Remaining firms supply 2 percent of the market. Their contribution will affect the H index only in the fourth decimal place and is ignored here. Source: Bailey (1981, p. 181). In practice, the prescriptions of contestability theory seem often to have been applied in a free and easy manner, without the kind of detailed analysis necessary before it would be safe to conclude that a particular market could be treated as “workably contestable.” The airline industry was early and easily asserted to be contestable. Real- world policy decisions were influenced by this position. Bailey (1981, p. 181) comments on a merger case that passed before the Civil Aeronautics Board in 1978, long before definitive evidence on the contestability of airlines was in hand. The prospective merger, as shown in table 11.2, involved market shares of the size that traditionally evoke policy concern. Thus (Bailey, 1981, p. 181) The share of the two leading firms was therefore 51 percent and would be almost 75 percent after a combination of Texas Interna- tional and National. This number was greater than comparable figures in mergers declared unlawful by the Supreme Court. The C[ivil]A[eronautics]B[oard] countered by arguing that concentra- tion ratios were not instructive in this case since with the passage of the Airline Deregulation Act of 1978. . . there was now relative ease of entry, even for small carriers, into such markets. Subsequent research has demonstrated that the premise of this decision 40 CHAPTER 1. CONTESTABLE MARKETS 1.5 Empirical tests of contestability theory Tests of the contestability of the passenger airline industry, which are typical of industry-level studies of contestability, have been reviewed above.32 Two other types of empirical tests deserve mention. 1.5.1 Experimental evidence Various studies have carried out experimental tests of markets that either conform closely to the assumptions of perfectly contestable markets or depart from those assumptions in controlled ways. Experimental subjects are given the roles of sellers or buyers, and make price, output, and purchase decisions. The experiments simulate the operation of the market for several periods. It is possible to examine how closely, if at all, performance conforms to the predictions of contestability theory.33 Coursey et al. (1984a) examine an experimental market which, by con- struction, satisfies the assumptions of perfect contestability. The market is endowed with a technology that satisfies the conventional34 definition of nat- ural monopoly: the average cost curve falls throughout the range of market demand. In four experiments, there is a single supplier. In six experiments, there are two suppliers, each able to supply the entire market demand. Sell- ers enter prices and quantities that will be offered at those prices; buyers select the suppliers from whom they will purchase. In four of the six duopoly experiments, price and output moved to com- petitive levels. In the remaining two duopoly experiments, price remained between the monopoly and competitive levels, although appearing to tend toward the competitive level. These results support the theory of perfect contestability. If there exist at least two firms, each capable of supplying the entire market, and the entire market moves from one supplier to another in response to small price differences, performance approaches the competitive extreme, even if average cost declines throughout the relevant output range. 32See Baumol and Willig (1986) for references to empirical studies of the contestability of other industries. 33In addition to the experiments discussed below, see Harrison and McKee (1985). 34Contestability theory identifies subadditivity as the defining characteristic of natural monopoly; see Baumol et al. (1982). 1.5. EMPIRICAL TESTS OF CONTESTABILITY THEORY 41 Coursey et al. (1984b) extend the original work by adding sunk costs to the experimental design of the duopoly markets. One player, the incumbent, is compelled to purchase a permit that allows it to supply the market for ten periods. After the tenth period, the incumbent must purchase an additional permit, valid for five periods, if it is to continue to operate. The other player can purchase such a permit, if it wishes, in or after the sixth period. The price of the permit is an entry cost that is sunk for five periods. They ran 12 trials of this experimental duopoly with sunk cost. In six of the trials, the tendency was for both firms to operate and for price to approach the competitive level. Four trials exhibited unstable pricing, in the sense that low prices would induce one supplier to exit, the survivor would raise prices, re-entry would occur, and price would fall. One trial showed an episode of limit pricing: five periods during which one player exited the market, with the remaining firm raising prices to the top of the competitive range. Two trials exhibited periods of tacit collusion: two firms in the market, with prices rising toward and falling from the monopoly level. Taken as a group, the experiments seem far from supporting the hy- pothesis that markets that are almost perfectly contestable behave nearly as predicted by the theory of perfectly contestable markets. Contestability theory predicts that potential competition determines market performance. In 11 of the 12 trials reported, it was actual competition, or the lack of it, that determined market performance. Only the experiment that produced an episode of limit pricing conforms closely to what one would expect if potential competition alone determines market performance. If actual competition must be examined to explain market performance, then an important goal of contestability — to free the analysis of market performance from dependence on analysis of oligopolistic interactions — fails. 1.5.2 Cross-section studies In a study of entry into 266 US industries, Kessides (1986) finds evidence that the need to invest in advertising is a sunk cost of entry.35 This is plausible, as advertising creates an asset — goodwill or reputation — that is subject to extremely limited resale upon exit. At the same time, net entry is shown 35The industries are defined at the four-digit Standard Industrial Classification (SIC) level. The study examines net entry — the number of entrants minus the number of exits — between 1972 and 1977 for industries where net entry was positive. 42 CHAPTER 1. CONTESTABLE MARKETS to rise with the industry advertising-sales ratio. This is consistent with the hypothesis that advertising is a device that entrants can employ to inform customers and carve out a niche in the marketplace. Kessides also reports that net entry is less, all else equal, in more prof- itable industries. He interprets this result as suggesting that incumbents are more likely to react to entry in an aggressive way the greater the profits that would be lost under a more competitive regime. Net entry is less, the greater the initial level of market concentration. The greater is initial concentration, all else equal, the more certain a single incumbent will be that it will benefit from its own efforts to deter entry. In an unconcentrated market, the benefits of entry-deterring activity accrue mainly to rivals. Net entry, in short, is less, the greater the profits to defend and the fewer the firms that benefit from defending them. If such markets were perfectly contestable, incumbents would act in the belief that entrants treated price as given, and incumbents would set price at sustainable optimal levels. Market concentration would have no effect on net entry. Kessides’ evidence is against this proposition. Actual entry is affected by elements of market structure in ways that suggest that potential competition will have limited effect, in general, in tempering the exercise of market power. Kessides (1988) examines the implications of sunk cost for market struc- ture.36 He finds concentration to be greater, all else equal, the greater is minimum efficient scale, and lower, all else equal, the more firms in the in- dustry are able to use rental or used capital equipment. Rental and used capital markets reduce the sunkenness of capital investments, so these re- sults are consistent with the argument that it is the sunkenness of assets that creates barriers to entry. If a market is perfectly contestable, its structure will be efficient in long run equilibrium. It follows that variance in market shares ought to be less, all else equal, the more contestable a market. Kessides also finds that the variance of market shares is less the more firms are able to use rental or used capital equipment. This is consistent with the hypotheses that market structure is endogenous, that market forces will tend to produce an efficient market structure, and that this tendency will be greater the less important are sunk costs. 36The study uses a sample of 400 1982 US four-digit SIC industries. References Adelman, Morris A. “Comment on the ‘H’ concentration measure as a num- bers equivalent,” Review of Economics and Statistics 51 February 1969, pp. 99—101. Akerlof, George A. “The market for ‘lemons’: quality uncertainty and the market mechanism,” Quarterly Journal of Economics 89(3) August 1970, pp. 345—64. Bailey, Elizabeth E. “Contestability and the design of regulatory and an- titrust policy,” American Economic Review 71(2) May 1981, pp. 179— 183. – “Foreword,” in Contestable Markets and the Theory of Industry Struc- ture. New York: Harcourt Brace Jovanovich, Inc. 1982, p. xiii, p. xix. – “Price and productivity change following deregulation: the U.S. experi- ence,” Economic Journal 86 March 1986, pp. 1—17. Bailey, Elizabeth E. andWilliam J. Baumol “Deregulation and the theory of contestable markets,” Yale Journal of Regulation 1 1984, pp. 111—137. Bailey, Elizabeth E., David R. Graham and Daniel P. Kaplan Deregulating the Airlines. Cambridge, Massachusetts: MIT Press, 1985. Bailey, Elizabeth E., Daniel P. Kaplan, and David S. Sibley. “On the con- testability of airline markets: some further evidence,” in Jörg Finsinger, editor, Economic Analysis of Regulated Markets. New York: St. Mar- tin’s Press, 1983, pp. 48—64. 45 46 CHAPTER 1. CONTESTABLE MARKETS Bailey, Elizabeth E. and John C. Panzar “The contestability of airline mar- kets during the transition to deregulation,” Law and Contemporary Problems 44 Winter 1981, pp. 125—145. Bailey, Elizabeth E. and Jeffrey R. Williams, “Sources of economic rent in the deregulated airline industry,” Journal of Law and Economics 31(1) April 1988, pp. 173—202. Baumol, William J. “Contestable markets: an uprising in the theory of industry structure,” American Economic Review 72(1) March 1982, pp. 1—15. Baumol, William J., Panzar, John C., and Willig, Robert D. Contestable Markets and the Theory of Industry Structure. New York: Harcourt Brace Jovanovich, Inc. 1982. – “Contestable markets: an uprising in the theory of industry structure: reply,” American Economic Review 73(3) June 1983, pp. 491—6. – “On the theory of perfectly contestable markets,” in Joseph E. Stiglitz and G. Frank Mathewson, New Developments in the Analysis of Market Structure. Cambridge, Massachusetts: The MIT Press, 1986, pp. 339— 65. Baumol, William J. and Willig, Robert D., “Pricing issues in the deregula- tion of railroad rates,” in Jorg Finsinger, Economic Analysis of Regu- lated Markets. New York: St. Martin’s Press, 1983, pp. 11—47. – “Contestability: developments since the book,” Oxford Economic Papers 38 Supplement November 1986, pp. 9—36. Bertrand, Joseph “Review,” Journal des Savants 68 1883, pp. 499—508; reprinted in English translation by James W. Friedman in Andrew F. Daughety, editor, Cournot oligopoly. Cambridge: Cambridge Univer- sity Press, 1988, pp. 73—81 and by Margaret Chevaillier in an Appendix to Magnan de Bornier (1992). Bhaskar, V. “Quick responses in duopoly ensure monopoly pricing,” Eco- nomics Letters 29(2) 1989, pp. 103—7. 1.6. CONCLUSION 47 Borenstein, Severin “Hubs and high fares: dominance and market power in the U.S. airline industry,” Rand Journal of Economics 20(3) Autumn 1989, pp. 344—65. Brock, William A. “Contestable markets and the theory of industry struc- ture: a review article,”Journal of Political Economy 91(6) December 1983, pp. 1061—4. Call, Gregory D. and Keeler, Theodore E. “Airline deregulation, fares, and market behavior: some empirical evidence,” in Andrew F. Daughety, editor, Analytical studies in transport economics. Cambridge: Cam- bridge University Press, 1985, pp. 221—47. Carter, Anne P. “Changes in the structure of the American economy, 1947 to 1958 and 1962,” Review of Economics and Statistics, 49(2) May 1967, pp. 209—24. Caves, Richard E. “International corporations: the industrial economics of foreign investment,” Economica 38(149) February 1971, pp. 1—27. – “Intra—industry trade and market structure in the industrial countries,” Oxford Economic Papers, N.S. 33(2) July 1981b, pp. 203—23. Chadwick, Edwin “Results of different principles of legislation and admin- istration in Europe of competition for the field, as compared with com- petition within the field, of service,” Journal of the Royal Statistical Society 22 September 1859, pp. 381—420. Cairns, Robert D. and Mahabir, Dhanayshar “Contestability: a revisionist view,” Economica 55 May 1988, pp. 269—76. Chamberlin, Edward H. The Theory of Monopolistic Competition. Cam- bridge, Massachusetts: Harvard University Press, 1933. Coursey, Don, R. Mark Isaac, Margaret Luke, and Vernon L. Smith “Market contestability in the presence of sunk (entry) costs,” Rand Journal of Economics 15(1) Spring 1984, pp. 69—84. Crandall, Robert, “Vertical integration and the market for repair parts in the U.S. automobile industry,” Journal of Industrial Economics 16(3) July 1968, pp. 212—34. 50 CHAPTER 1. CONTESTABLE MARKETS Pashigian, B. Peter “Limit price and the market share of the leading firm,” Journal of Industrial Economics 16 July 1968, pp. 165—77. Prescott, Edward C. “Market structure and monopoly profits: a dynamic theory,” Journal of Economic Theory 6 1973, pp. 546—57. Scherer, F. M. “Research and development resource allocation under ri- valry,” Quarterly Journal of Economics 81(3) August 1967, pp. 359— 94. – Industrial Market Structure and Economic Performance. Chicago: Rand McNally, 1970. Schwartz, Marius “The nature and scope of contestability theory,” Oxford Economic Papers 38 Supplement November 1986, pp. 37—57. Schwartz, Marius and Reynolds, Robert J. “Contestable markets: an upris- ing in the theory of industry structure: comment,” American Economic Review 73(3) June 1983, pp. 488—90. Seabright, Paul “Can small entry barriers have large effects on competi- tion?,” Department of Applied Economics, Cambridge University, Eco- nomic Theory discussion Paper 145, February 1990. Shepherd, William G. “Trends of concentration in American manufacturing industries,” Review of Economics and Statistics 46 May 1964, pp. 200— 12. – “Changes in British industrial concentration,” Oxford Economic Papers 18(1) March 1966, pp. 126—32. – “‘Contestability’ vs. competition,” American Economic Review 74(4) September 1984, pp. 572—87. – “On the core concepts of industrial economics,” in Henk W. de Jong and William G. Shepherd, editors Mainstreams of Industrial Organization. Dordrecht: Kluwer Academic Publishers, 1987, Book I, pp. 23—67. – “Entry barriers, contestability, and predatory pricing,”Revue d’Economie Industrielle 46(4) 1988, pp. 1—20. 1.6. CONCLUSION 51 Spence, A. Michael “Contestable markets and the theory of industry struc- ture: a review article,” Journal of Economic Literature 21 September 1983, pp. 981—90. Stiglitz, Joseph E. “Technological change, sunk costs, and competition,” Brookings Papers on Economic Activity 1987:3, pp. 883—937. Stockton, William “When eight carriers call the shots,” New York Times November 20, 1988, Section 3, p. 1. Strickland, Allyn D. and Leonard W. Weiss “Advertising, concentration, and price—cost margins,”Journal of Political Economy 84(5) October 1976, pp. 1109—21. Tye, William B. “The applicability of the theory of contestable markets to rail/water carrier mergers,” Logistics and Transportation Review 21(1) March 1985, pp. 57—76. Weiss, Leonard W. “Factors in changing concentration,” Review of Eco- nomics and Statistics 45 February 1963a, pp. 70—7. Weitzman, Martin L. “Contestable markets: an uprising in the theory of industry structure: comment,” American Economic Review 73(3) June 1983, pp. 486—7. Werden, Gregory J., Andrew S. Joskow, and Richard L. Johnson “The effects of mergers on economic performance: two case studies from the airline industry,” U.S. Department of Justice, Antitrust Division, Discussion Paper EAG 89—15, October 2, 1989. Index Bertrand model theory of contestable markets, 26, 43 concentration seller Herfindahl index, 33 conjectural variations Bertrand-Nash, 11, 12 contestable markets, theory of assumptions, 9 Bertrand model, 26 empirical tests of, 40 generalization of structure-conduct- performance paradigm, 26 goals, 24 guide to areas for public inter- vention, 36 guide to conduct of public in- tervention, 38 guide to suitability of public in- tervention, 29 hit-and-run entry, 9 long-term contracts, 14 market performance, 29 market structure, 26 oligopolistic interactions, 12 passenger airline industry, 30 price-taking entrants, 10 product differentiation, 20 results, 6 sunk costs, 15 transaction costs in financial mar- kets, 19 Herfindahl index seller concentration, 33 limit pricing dynamic, 27 in experimental markets, 41 monopolistic competition contestable markets, 20 long-run equilibrium, 20 Names Adelman, Morris A., 33, 45 Akerlof, George A., 17, 45 Anderson, Hans Christian, 5 Arrow, Kenneth J., 48 Bailey, Elizabeth E., 5, 13, 22— 25, 30—34, 36, 37, 39, 45, 46 Bain, Joe S., 26, 28, 29, 43 Baumol, William J., 5, 6, 9—17, 20—22, 25, 26, 28, 29, 32, 35, 36, 38—40, 44—46 Bertrand, Joseph, 26, 46 Bhaskar, V., 13, 46 Borenstein, Severin, 38, 47 Brock, William A., 15, 47 Cairns, Robert D., 15, 47 52