Many IO textbooks introduce the theory of oligopoly by presenting a series of models with little explanation of what their purpose is. When this happens, instructors are often faced with questions such as: Why so many models? Why are we not given the "right" model only, saving us the trouble of going through all the other ones?

The proliferation of IO models started in the 19th Century when, responding to the early work by Cournot, Bertrand proposed an alternative model of oligopoly competition that implied strikingly different predictions from the early model. As a result, the debate about which model is "best"--Cournot or Bertrand--has evolved into a cottage industry within IO.

The approach taken in this chapter is one that reflects recent
developments in IO theory. In particular, it views the Bertrand and
Cournot models as particular cases of a more general model of
oligopoly competition where firms choose prices *and* quantities
(or capacities). Seeing things from this perspective, the Bertrand
model arises as the "right" solution when firms can adjust capacities
faster than prices (e.g., software), whereas the Cournot model
corresponds to the opposite case, the case when prices can vary
faster than capacities (e.g., wheat, cement).

This way of looking at things is important for this chapter and, more generally, for the whole of oligopoly theory. There is no single model that is better than all other models. Each model is a better description of a certain type of industry. The art (and science) is then to determine what is the appropriate model for each situation.

Another important point that this chapter attempts to put across is that models are useful, operational tools to analyze a particular real-world situation--oligopoly competition, in the case of IO. For this reason, a fairly long section on comparative statics is included at the end of the chapter. Comparative statics is one of the main goals of economic modeling. It allows one to estimate the impact of an exogenous change in the equilibrium values of a given system--in the case of IO, prices, quantities, market shares, and so on.

Even when "theoretical" lectures are complemented by "applied" sections, it is recommended that some time is spent in the main lectures to go through the material in Section 7.5. Among other things, this will have the benefit of making students more confident in the value of the models--and theory, more generally. As Chesterton so aptly put it, the best help for practical life is a good theory.

The Bertrand model implies a very striking result: even if there are only two competitors, prices will be set at the level of marginal cost. From the perspective of positive economics, this implies what we might call the "Bertrand paradox" (Tirole, 1989): in reality, there are many industries that look like the Bertrand model but where prices are higher--or much higher--than marginal cost. Chapter 7 includes a brief description of the three main explanations for this apparent paradox: capacity constraints (Chapter 7), dynamic interaction (Chapter 8) and product differentiation (Chapter 12).

From the perspective of competitive strategy, the outcome of the Bertrand model has aptly been described as the "Bertrand trap" (Hermalin, 1993): an outcome that firms should avoid at all costs. The three ways of explaining the Bertrand paradox (positive analysis) can then also be interpreted as ways of exiting the Bertrand trap (normative analysis)--especially the "escapes" described in Chapters 8 and 12. Depending on the nature of the audience, one or the other view may be emphasized. For example, when teaching MBA students it may make more sense to stress the idea of the Bertrand trap.