IO Chapter 8

When we did the simple game theory models we found a couple of examples where firms would be better off if they were to plan together. For instance Figure 4.1. These situations are known as prisoner's dilemmas, because when firms act alone they cannot do as well as if they act together. In other words the best joint strategy is collusion!

Reexamining Figure 4.1 we find that while collusion is the best solution, it is not a Nash equilibrium. Each firm has the incentive to cheat. Of course if they do, they again end up in the suboptimal solution. Is this a realistic model of the likely outcome of a market where collusion is the best strategy? Yes and no.

In another example, (the Cournot model), we saw from the graph of the reaction functions (and the algebraic calculations) that the oligopoly solution was somewhere between the perfectly competitive and the monopolistic outcome, as well as being a Nash equilibrium.

Cartels are one form of collusion. Examples include the oil and diamond cartels. Following these examples a number of developing countries have tried to establish cartels around various commodities, but these have been less successful. What commodities do you think are most likely to be successful as cartels?

Collusion in price setting is generally considered illegal. In the US the Sherman Anti-Trust Act addresses this issue. In Europe the Treaty of Rome covers it. Look at the wording of each.

See p. 144 in text and p. 6-7 of de Beers study. Not only is collusion in price and quantity fixing prohibited, but at least in the case of the US, activity anywhere in the world which may affect US markets is covered.

Why is it illegal? From an efficiency standpoint what can we say? Total welfare is lost when firms collude. The solution is allocatively inefficient. It also implies a transfer of welfare from consumers to producers. In other words there is a distributional issue here as well.

Of course enforcement of anti-trust legislation is far more complex. It is not always easy to identify when violations of anti-trust law, such as collusive behavior, are taking place or when instead firms are simply following a price leader.

Then there is the question of whose rights are most important and the question of jurisdiction.

Is the US justified in arguing that even activity which takes place outside the US should be subject to US anti-trust laws? We'll address this question next week when we discuss the de Beers case.

Back to the question of models of firm behavior-

Why was the Bertrand model unrealistic? Because it assumed that price setting is a one shot deal. Instead firms have the chance to react to other firms and play 'repeated games.' For instance, the outcome of the prisoner's dilemma may be different if it is played more than once, since the players may learn something. In the long run, with repeated games, two additional features are added: firms can learn from the other firm's behavior, firms can use the threat of retaliation.

The book hypothesizes that collusion is not very common, but does not provide much evidence. This question is to some degree still open to debate. What factors might make collusion difficult?

1. Collusive behavior may not be a Nash equilibrium in some cases.

2. The legal structure may inhibit firms from colluding. (Although the history of the enforcement of anti-trust has been spotty, so the strength of this threat may vary depending on the government in power.)

3. The instability of some markets may lead to firms not cooperating, because they don't know how long a particular equilibrium will hold.

4. Firms may not have complete information about their opponents.

5. Markets with fluctuating or unpredictable demand may make price collusion more difficult.

On the other hand, why is it that prices are so similar in many cases?Are these the outcome of price collusion or market competition. How can that question be answered? One way is by looking at profit margins. If prices look similar and profits are high, this is evidence of collusion. If prices are similar and profits are low, it is much less likely.

What are some factors which may increase the chance of collusion?

1. Concentrated industries (not surprisingly) are more likely to lead to collusive behavior. Coordinating among a small number of players is of course easier. Also, market power of individual firms is greater.

2. Similar firms (in terms of size, technology, etc.) are more likely to collude. Otherwise questions about the appropriate distribution of profits may arise. (Of course the opposite may also be the case, if you have a dominant firm, this may also be an incentive for collusion.)

3. Firms competing in more than one market are more likely to collude. Potential profit gains are higher if colluding in multiple markets. Basically what this means is that firms carve up their markets and agree that each will dominate in one market.

4. Ironically some regulations (such as restricting firms to offer the same price to all customers) may encourage collusion. Although these laws may have been put in place to protect consumers, they may end up with higher pricing. An example from Denmark was that making prices a matter of public record may make enforcing a cartel easier (rather than protecting consumers.)

We will talk in more depth about 2 examples of cartels. Next Monday we will discuss the de Beers case, using the Harvard Business School Case study material, and for your group projects you will also be looking into the oil industry.

The opposite of collusion is price wars.

As the book points out, if firms agree on a price, but do not know what demand will be or what their rivals actual prices are, then if their sales are lower than expected, they do not know whether it is due to a secret price cut by their opponent or a dip in demand. What should the firm do? Cut prices based on the assumption that the other firm is cheating? Or wait and see? It could be that firms alternate between colluding and price wars (since price wars may be a way of punishing those who do not hold up their end of the collusive agreement.)

Another possible model is one where firms do know demand, but demand fluctuates. How should the cartel price if demand fluctuates? When will the incentives to cheat be highest? When the demand is highest. So it might make sense for the cartel to lower prices during periods of high demand, to reduce the temptation to cheat.

Which firms are the leaders in the case of collusion/price wars? In the case of OPEC, (which can be seen both as a cartel and a dominant firm example) Saudi Arabia often increases production, which in turn reduces the world price, as a punishment when it sees that other countries are cheating on their quota of oil. Saudi Arabia in this example is the dominant firm and can use its power to punish other players. But at other times smaller countries, such as Iran or Iraq, may 'cheat' increasing production because they are in need of foreign exchange. As the book points out in some cases it is the small, weak companies which begin price wars, because they need revenues. In other cases it is the larger companies. The danger of course for small companies is that if there is a dominant firm, that firm can flood the market undercutting all other firms. The example the book provides is the case of Murdoch and UK newspapers. Murdoch basically cut the price of newspapers by more than half. Although one interpretation is that it was simply a price war due to a scramble for revenues, another theory is that Murdoch did it to boost his market power, hoping to run other firms out of business.