APPENDIX 7.9 (referred to in paragraph 7.

181)

Models of price leadership

1. This appendix considers various models of oligopoly that address price leadership, defined as a situation where one firm in an industry sets a price which others follow. It distinguishes the three classes of model identified by Scherer,1 ie (a) barometric; (b) dominant firm and a competitive fringe; and (c) collusive price leadership. 2. The following section comprises a brief summary of each of these taken in turn.

Barometric price leadership

3. This version postulates that the price leader is a firm that responds more quickly than its rivals to changing cost and demand conditions, but does not in itself have significant market power. In such circumstances the price leader acts, in effect, as a barometer of market conditions for the rest of the industry, with its price levels set close to those that would emerge even under competition. Barometric price leadership may be indicated by a number of market characteristics, for example occasional switching between firms in the role of price leader; the occurrence of upward price leadership only in response to increased industry costs or demand; occasional and sometimes substantial time lags in the price response of follower firms; and occasional rejection by the rest of the market of price changes initiated by the price leader.

Dominant firm and competitive fringe

4. Other models characterize price leadership in terms of industries where the distribution of firm sizes is highly skewed, resulting in a dominant firm that exists alongside a competitive fringe of much smaller firms, typically supplying a relatively standardized product. The fringe suppliers are small individually but may have a significant share of the market collectively. The dominant firm sets its own price on the basis of industry demand not served by the fringe suppliers, thereby providing a price umbrella for the latter. Market performance then depends on relative costs and ease of market entry. If there are barriers to market entry, the dominant firm will be able to charge supra-competitive prices, though this power will be moderated by the presence in the industry of the competitive fringe. Since it acts as pricesetter, the dominant firms price cuts will be matched by the competitive fringe. On the other hand, if barriers to entry are low, the price leaders ability to exercise market power will be constrained by the entry (or threat of entry) of additional fringe suppliers.

Collusive price leadership

5. Early static models of oligopoly suggested that attempts to coordinate pricing would often break down, because of the incentives facing an individual firm to deviate from the agreement. The idea was that individual firms would often face too strong an incentive to cheat by selling additional output at the higher (collusive) price, thereby leading to the breakdown of the collusive pricing policy. Coordination of pricing was viewed as likely to be feasible only in industries that are highly oligopolistic, where products are close substitutes, where barriers to entry exist, and where firms face similar cost conditions (making it easier to detect cheating). 6. However, it is likely that collusive price leadership can only be understood properly in a dynamic context. More recent work, much of it based on game-theoretic foundations, examines the conditions under which pricing behaviour may be collusive when firms are viewed as taking decisions over a number of time periods, in contrast to the single-period approach of earlier static models.
1

F M Scherer, Industrial Market Structure and Economic Performance, 1990, Houghton Mifflin.

139

7. A flavour of the approach is given by Rotemberg and Saloner.1 They observe that the barometric and dominant firm models are often inappropriate for industries in which there are a number of more or less equally-sized players selling differentiated products. In such industries one would expect at least some degree of strategic behaviour. They model a form of oligopolistic competition in which price announcements by one firm are quickly matched by competitors, pricing behaviour that they argue is typical of many industries. In choosing whether to follow the leaders price change, the follower trades off the expected one-off gains from deviating from the price-matching policy against the losses it would incur were the leader to revert to non-cooperative behaviour (ie a permanent price war). They show that collusive behaviour can be sustained in a repeated game provided that the threat of reversion to noncooperative behaviour is credible. The following features emerge from their analysis. 8. The model attempts to predict which firm will be the price leader. Given the presence of differentiated products, each firm will face different cost and demand conditions and therefore have different preferences in terms of the (matching) price level. Other things being equal, the firm that acts as price leader is likely to obtain greater profits from the pricing strategy than the firm that assumes the role of follower. However, if one firm possesses superior information about demand, then the less informed firm may find it profitable not to take on the price leadership role. 9. Some degree of price stickiness, though not completely rigid pricing, would be expected to emerge from the price leadership regime. The price leader has an incentive to changes prices in response to shifts in relative demands for the two products and in response to changes in its own costs. In so doing it can make profits at the expense of the follower. However, the price leader has to balance the gains from opportunistic price changes against the possibility that the follower will revert to non-cooperative behaviour if there are too many such changes.

Multi-market contact
10. The game-theoretic models of collusive behaviour assume that firms determine their pricing strategy over time. However, it is also possible to consider the possibility of collusive behaviour across markets, either product or geographic. Bernheim and Winston2 provide a theoretical framework for assessing the degree to which multimarket contact facilitates collusive pricing behaviour. The intuition is that, if a firm faces the prospect of retaliation in other markets because of, say, a price war fought in a local market, it may moderate its pricing behaviour accordingly. Bernheim and Whinston identify the following conditions under which multimarket contact may give rise to collusive behaviour: (a) If firms face different cost conditions in different markets, this may lead to mutual recognition of their relative cost advantages in these markets. This can result in the creation of spheres of influence which serve to keep prices high in these markets. (b) When markets are non-identical in terms of factors such as growth rates or fluctuations in demand. The costs of a breakdown in cooperative pricing behaviour are greater in expanding markets. If firms operate across markets with different levels of growth and profitability, they may refrain from competing aggressively even in low-growth markets in order to avoid damaging price wars in expanding markets.

Facilitating practices
11. Another strand to the theory on tacit collusion has been to focus on some of the practices that may facilitate coordination of pricing behaviour by firms. However, many of the practices identified may also be consistent with benign explanations of firm behaviour, and some may be welfare-enhancing. For example, there is a considerable amount of theoretical research into the competition and welfare effects of sales tactics such as public commitments to match or beat the prices of competitors (sometimes known as price-matching guarantees). Such tactics have been shown in a number of theoretical papers to be
1 J J Rothemberg and G Saloner, Collusive price leadership, The Journal of Industrial Economics, vol xxxix, no 1, September 1990. 2 B D Bernheim and M D Whinston, Multimarket contact and collusive behaviour, RAND Journal of Economics, vol 21, no 1, 1990.

140

capable of producing tacit collusion under certain conditions (for example, Logan and Lutter;1 Zhang2). Intuitively, a price-matching policy could be seen as a way of signalling information about a desired price level to competitors. It could also be seen as a way of deterring price cutting by rivals, since it increases the prospect of immediate retaliation if a particular firm tries to undercut the price-matching firm. 12. On the other hand, the welfare effects of such tactics can be ambiguous, in that they may sometimes bring consumer benefits, for example reduced consumer search costs, or price discrimination that benefits at least some groups of consumers. It is also possible, of course, that a firm will implement such tactics purely on competition grounds, because it believes it has inherent cost advantages that allow it to match or beat any price offered by its rivals. 13. Hess and Gerstner3 provide an empirical assessment of the pro-competitive versus collusive pricing theories of price-matching policies, using data collected from a regional US grocery market where several stores had announced that they would match (some) prices of a low-price supermarket. They find evidence that price-matching policies helped supermarkets in the study to avoid price competition; the price-matching policy induces the targeted store (low-cost supermarket) to raise its prices for products included relative to those excluded from the policy. They also suggest that there may be greater incentive to employ price-matching policies in the grocery sector than in other markets, because total demand for food is price inelastic. In such circumstances increased market share can only be gained at the expense of rivals, rather than by generating additional market sales, so firms may be keener to avoid price competition. 14. Another potentially collusive tactic, identified from game theory, is the trigger price strategy.4 This is a strategy designed to prevent cheating on a tacit price agreement. Firms agree that, if price falls below a certain level, a price war will be triggered for a predetermined amount of time, after which they revert to a higher (collusive) price. Even though the price war could be triggered by an exogenous shock (such as a fall in demand that pushes down the market price), the existence of a trigger price is also sufficient to prevent individual firms from gaining by cheating on the tacit pricing agreement. This suggests that periodic price wars may not always be evidence of competitive pricing, but perhaps indicative of some form of trigger price strategy in play. 15. There may be other mechanisms of enforcing tacit collusion, such as investment in unnecessary low-cost capacity, so that the threat of output expansion (and ultimately lower market prices) serves to enforce cooperative behaviour. Overall, the message of game-theoretic approaches is that the sustainability of collusive price leadership ultimately depends on the credibility of the threat of reversion to non-cooperative behaviour in the event of cheating by individual firms. Tactics or policies that signal likely future behaviour, or commit the firm to predetermined actions in response to price cutting by rivals, may well give added force to this threat and make tacit collusion more likely.

Application to the UK supermarket sector

16. Why would it be in the interest of the multiples to have a tacit agreement of price leadership? What may make such an arrangement advantageous to all the multiples and prevent an outbreak of unbridled competition? For many of the goods sold by supermarkets the demand is inelastic and a change in price will not bring about an increase in the total sales of the product. In such conditions any one supermarket can increase its share of the total by cutting its price but this is likely to cause a response by other supermarkets also cutting their price. Such competition will not increase total sales but will cut profits of all the multiples. Under these conditions all the multiples can increase their profits by reaching a tacit agreement as to the optimal, or near optimal, price level and keep to that level for fear of retaliation if they break rank. Price leadership is one way of signalling the appropriate price level.
1 J W Logan and R W Lutter, Guaranteed lowest prices: do they facilitate collusion?, Economics Letters, vol l31, pp189192, 1989. 2 Z J Zhang, Price matching policy and the principle of minimum differentiation, The Journal of Industrial Economics, vol xliii, no 3, September 1995. 3 J D Hess and E Gerstner, Price-matching policies: an empirical case, Managerial and Decision Economics, vol 12, pp305315, 1991. 4 D A Yao and S S DeSanti, Game theory and the legal analysis of tacit collusion, The Antitrust Bulletin, vol xxxviii, no 1, Spring 1993.

141

17. Given that the possibility of price leadership is enabled by the structure and practices of the UK supermarket sector, we can look for behaviour consistent with the three types of price leadership set out above. 18. In the supermarket industry barometric price leadership would be a situation where one supermarket, and not necessarily the same supermarket each time, reacts more quickly to changing market conditions, for example a change in costs, and other supermarkets then follow. The outcome in such a situation could be the equilibrium competitive price but it could also be a higher price. The outcome would be higher prices if the market changes facing the leader are not the same as those facing other supermarkets and if the barometric price leadership is combined with some other features of price leadership described below. 19. If dominant price leadership operated in the supermarket industry leadership would be expected from the largest chains, ie Tesco or Sainsbury. However, dominant price leadership tends to operate in industries where there is great disparity in size between the leader and the followers. These conditions do not match closely the state of the supermarket industry where there are several large operators. Prices set by a dominant price leader could, and mostly are, higher than competitive prices. 20. Overtly collusive price decisions are legally unacceptable but a tacit agreement, or actual practice, of following price changes determined by a price leader can also result in prices that are higher than competitive prices. Again, price leadership in this context may involve more than only one price leader. Different operators could be price leaders at different periods or in different market segments.

142