Network Externalities, Demand Inertia, and Dynamic Pricing in an Experimental Oligopoly Market

Many commodities are such that the utility they create for individual consumers depends positively on the number of people also consuming these goods. Prominent examples among others are mobile phones, game consoles, and computer software. The customers form a network, where the size of the network increases the usefulness of the product for consumers. In the real world we observe that prices for such products decrease over time. However, game theory predicts that producers should take into account that current sales in markets with network externalities increase the future demand. We show that optimal prices are increasing over time if a base model is used where all other factors are excluded that could lead to decreasing prices over time (increasing returns to scale, learning by doing or inter-temporal price discrimination). We use a laboratory experiment to test this prediction. We find that the observed price path in the experiment is consistent with the real-world observation of deceasing prices rather than with the game theoretic prediction. Even if we allow for learning (repeated markets) prices are decreasing in young markets. We attribute this pricing behaviour to the well-known fact that people are not able to conduct backward induction within long supergames, but rather use rules of thumb. Surprisingly, the rule subjects use, is such that aggressiveness of play is positively correlated with the market share throughout the game. So a high market share seems to be an objective for itself rather than a way of increasing future profits. Another interesting result is that we observe much less collusion in a market with network externalities than in markets without