The association between economic efficiency and competition goes back at least as far as Adam Smith’s ‘invisible hand’ metaphor. Indeed, a goodly portion of the vast body of subsequent work in value theory has dealt with the normative issues arising from the workings of the competitive economy. Thus any short essay on the topic must be somewhat idiosyncratic, focusing upon the points which are of greatest interest to the author. Therefore, I shall limit my attention to the properties of the (static, partial equilibrium) economic model of perfect competition and how its use has recently been extended to add to our understanding of a larger range of real world markets. I must leave to others the tasks of sorting out the importance of competition in, for example, the Schumpeterian process of ‘creative destruction’, the aggregation and transmission of society’s stock of information, or the evolutionary progress of technological advance. Fortunately for my purposes, the historical development of the competitive model has been thoroughly analysed by Stigler (1957). The formulation of the model, as we know it today, was completed in the work of Knight (1921). It is interesting to note that the last refinement to be added was the free mobility of resources across industries: i.e. the entry and exist of firms. In his insightful concluding section, Stigler points out that competition can flourish within a market without this last ingredient. (Consider an agricultural market with Ricardian rents.) He suggested that the term ‘market competition’ be used to describe such situations, and that the term ‘industrial competition’ be applied when mobility across industries is present. The work that I shall discuss deals with the converse possibility: perfectly contestable markets, situations in which competition may not necessarily exist within a particular market, but firms (and resources) are assumed to be perfectly mobile across industries.

The role of entry and exit in assuring the equalization of returns across markets is not logically limited those cases in which it is technologically feasible for the market to be populated by a large number of firms, each capable of achieving an efficient scale of operation. It may be expected that the lure of profits might serve to make relevant certain aspects of competitive theory even under conditions of ‘natural monopoly’. The most striking practical illustration of this point was the recent deregulation of airlines in the United States. This took place, in part, because the free mobility of resources (aircraft) across markets led policy makers to believe that satisfactory economic performance could be achieved without the stultifying effects of economic regulation. This, despite the fact that most city-pair airline markets are natural monopolies and none can be expected to support the large numbers of firms required by the perfectly competitive model. Thus the need to extend at least part of the competitive paradigm to incorporate such cases had become apparent.

In a classic article, Demsetz (1968) set forth one way to break the commonly perceived link between monopoly provision of certain increasing returns services and monopoly conduct on the part of the firm providing the service at any point in time. By pointing out that the impossibility of competition within the market need not preclude effective competition for the market, Demsetz raised a fundamental challenge to the conventional wisdom that the only effective ways to deal with a technological natural monopoly were through economic regulation or public enterprise.

Demsetz chose to elaborate this idea in the context of a franchise bidding scheme, in which the franchise was to be awarded, not to the firm offering the greatest lump sum payment to the municipal coffers, but to the firm offering to serve the market at the lowest price. Subsequent authors have criticized this as a policy proposal, focusing on the problems raised by considerations of sunk costs and incomplete contracts, from which Demsetz explicitly sought to abstract. However, there is another sense in which the franchise bidding example may have been an unfortunate expository choice. Because it introduced a new institution between the firm and the market – the franchise auctioneer – this illustration may have obscured the link between the analysis of competition for the market and the earlier notion of the role of free entry and exit in ensuring effective industrial competition. Furthermore, Demsetz’s simple bidding scheme cannot handle the realistic cases in which the monopolist produces two or more technologically related services.

The theory of contestable markets developed by Baumol et al. (1982) is most usefully viewed as an attempt to extend the neoclassical (partial equilibrium) theory of long-run competitive equilibrium to the case of increasing returns to scale. In so doing, they developed a model which achieved the Demsetz solution to the monopoly problem as the result of a market equilibrium process. This extension was accomplished by emphasizing the role played by potential entry in characterizing the role defining properties of long-run competitive equilibrium. To see this reinterpretation most clearly, the following definitions are necessary:

FormalPara Definition 1

A Feasible Industry Configuration (FIC) is a collection of firms, \( i=1,\dots, m \) output vectors for each, y1, …, ym; and a market price vector p such that each firm earns non-negative profits and the total quantity supplied equals the quantity demanded; i.e. \( p{y}^i-C\left({y}^i\right)\ge 0 \), for all \( i=1,\dots, m \) and \( \sum {y}^i=D(p) \), where C is the (multiproduct) minimum cost function and D the market demand function.

Feasibility surely reflects the minimal conditions one would expect to prevail in long-run industry equilibrium in a private enterprise economy: All firms must earn non-negative profits and the total quantity supplied by firms equals the amount demanded by consumers at the market price. While feasibility requires financial viability, it does not preclude the positive profits which may attract entry. Therefore, the neoclassical notion of long-run competitive equilibrium must encompass some additional restrictions. More specifically,

FormalPara Definition 2

A long-run competitive equilibrium is any FIC which also has the property that \( py-c(y)\le 0 \) for all y.

While this characterization of long-run competitive equilibrium may be unfamiliar, it is equivalent to the standard notion of price taking firms earning zero economic profits by equating marginal cost to price. (To see this, note that since profits are nonpositive for all output levels, the fact that \( p{y}^1-C\left({y}^i\right)\ge 0 \) means that output level yi maximizes the ith firm’s profits. This, in turn, implies that \( MC\left({y}^i\right)=p \) if firm i is producing.)

Characterizing competitive equilibrium via Definitions 1 and 2 has the advantage of focusing attention on the role played by potential entry. The strictures of Definition 2 can be interpreted to mean that the firms in an industry in long-run competitive equilibrium act as if they were policed by potential entrants prepared to enter the market in pursuit of any profit opportunity calculated at current market prices. While this lack of attention to the possibility of retailatory price responses by rivals reflects the noncooperative spirit of the competitive paradigm, it ignores the response of consumers to a change in the market price. Therefore it is useful to consider making potential entrants ‘less optimistic’ in the following sense:

FormalPara Definition 3

A Sustainable Industry Configuration (SIC) is any FIC which also satisfies the condition that \( {p}^{\mathrm{e}}{y}^{\mathrm{e}}-C\left({y}^{\mathrm{e}}\right)\le 0 \) for all \( {p}^{\mathrm{e}}\le p \) and \( {y}^{\mathrm{e}}\le D\left({p}^{\mathrm{e}}\right) \).

Thus firms in a SIC behave as if the market were policed by potential entrants that calculate the profitability of entry under the assumption that incumbent firms’ prices remained unchanged, but that do take account of the reality that consumers can be induced to purchase a larger quantity only at a lower price. Put another way, an FIC is also an SIC when no potential entrant can anticipate earning a positive profit by quoting a price at or below that prevailing in the market and serving all or a part of the resulting demand. The following semantic clarification completes the characterization of a contestable market:

FormalPara Definition 4

A perfectly contestable market is one in which perfectly free entry and exit ensure that the only possible long-run equilibria are SICs.

An immediate implication of Definitions 1, 2, 3, and 4 is:

FormalPara Proposition 1

Any long-run competitive equilibrium is an SIC, but not conversely. Thus all perfectly competitive markets are perfectly contestable, but not all perfectly contestable markets are perfectly competitive.

The proof follows from the fact that the conditions which an FIC must satisfy in order to be a long-run competitive equilibrium are stronger than those required of an SIC. Thus a long-run competitive equilibrium is, by construction, an SIC. To see that the converse is not true, consider the case in which the average costs of production fall throughout the relevant range; i.e. at least as far as the intersection of the average cost curve and the market demand curve. The point of intersection, the Demsetz outcome, characterizes a sustainable industry configuration, since profits are non-positive and no point on or below the demand curve can yield non-negative profits at a lower price. However this outcome is clearly not a long-run competitive equilibrium because price is equal to average cost which, by hypothesis, is strictly greater than marginal cost. The above demonstration points out the fact that the concept of contestable markets can be applied beyond the large-numbers case of perfect competition. However it also raises questions about the efficiency properties of such markets. The fact that equilibrium may involve a price greater than marginal cost means that the First Best optimality properties of the competitive model need no longer apply. What efficiency properties, then, can be associated with equilibria in contestable markets? In the case of single product markets it is intuitively clear (and straightforward to prove) that, when they exist, sustainable industry configurations are solutions to the Second Best optimization problem: maximize welfare (as measured, for example, by the sum of producers’ and consumers’ surpluses) subject to the constraint that firms earn non-negative profits. Clearly, when increasing returns to scale render marginal cost pricing unprofitable, the best that can be done, in the absence of lump sum transfers and discriminatory or non-linear pricing, is to set price equal to average cost.

However, even this level of performance can no longer be guaranteeed once one moves to the realistic realm of multiple products. For example, a monopolist producing two or more products can, in general, find an infinite number of price combinations which will yield it exactly zero economic profits. Some of these prices and resulting market demand quantities may represent SICs. Call this set P. If the underlying cost and demand functions are sufficiently well-behaved, there will exist a unique constrained welfare maximizing price vector p*. The most desirable efficiency result would be for the set P to consist of the single element p*. Unfortunately, it is easy to construct examples in which P does not contain p*, as well as cases in which P is empty. What efficiency properties does this generalized process of industrial competition possess when extended beyond the realm of perfect competition? The results that pertain generally lie entirely on the cost side.

FormalPara Proposition 2

In any SIC, the industry’s output is divided among the firms in a way that minimizes total industry costs.

The proof is by contradiction. Consider an initial SIC composed of m firms producing output vectors, y1, …, ym, at market prices p. Suppose, contrary to hypothesis, that there exists an alternative group of k firms with output vectors, z1, …, zk, that could produce the current industry output at a lower total cost. That is \( {\sum}_j{z}^j=D(p)={\sum}_i{y}^i \), but \( {\sum}_jC\left({z}^j\right)<{\sum}_iC\left({y}^i\right) \). Then the new group, in total, would earn positive economic profits at the initial price p. This is true because, by hypothesis, total revenues would be equal, but total costs would be lower for the alternative group, while the initial group of firms must have been earning non-negative profits. Therefore at least one firm, say firm j, in the alternative group would anticipate earning strictly positive profits at the price vector p. But then there exists an entry plan \( {p}^{\mathrm{e}}=p\le p \) and \( {y}^{\mathrm{e}}={z}^j\le D\left({p}^{\mathrm{e}}\right) \) such that \( {p}^{\mathrm{e}}{y}^{\mathrm{e}}-C\left({y}^{\mathrm{e}}\right)>0 \), which contradicts the hypothesis that the initial group of firms constituted a SIC.

Additional efficiency results for contestable markets are presented in chapter 11 of Baumol, Panzar and Willig. Here, I shall mention specifically a class of results which are relevant only in the multiproduct context. One implication of the fact that equilibrium in a contestable market presents no profit opportunities for potential entrants is that no subset of services of a multiproduct enterprise can generate revenues in excess of the cost of providing them alone. Thus, equilibrium in perfectly contestable markets cannot involve one gorup of services subsidizing another. Whether or not this property is a desirable efficiency result is unclear. Consider, for example, a situation in which a monopoly firm uses common facilities to produce two services, one of which has a very elastic demand curve while that of the other is very inelastic. Maximizing total surplus subject to a break-even constraint leads to the well-known inverse elasticity rule: the markup of price over marginal cost is greater for services whose demand is least elastic. However, this pricing policy may easily lead to revenues from the inelastic service in excess of the cost of providing it alone. Such an outcome would not be an SIC and could not persist in a perfectly contestable market. Thus while the mobility of firms and resources can, even without the presence of market competition, ensure productive efficiency, it cannot in general guarantee that an optimal relationship of output prices in a multiproduct industry will prevail outside the perfectly competitive realm.

See Also