Essays in Competition and Externalities

This dissertation consists of three papers. A common feature of these papers is the interest in how externalities affect consumers and firms’ behavior. In the first paper, I study one type of contractual externalities called exclusive dealing, whereby one firm cannot deal with the competitors of the other. More specifically, I propose and estimate an empirical structural model to investigate the effects on prices of upstream mergers in markets with exclusive dealing contracts. The second paper is concerned with markets for a good with network externalities, i.e. a good that generates higher utility the higher the number of consumers purchasing it. The third paper studies externalities of investments on quality improvement. When more than one firm is active, the product improvement externality occurs because as firms chose different quality levels, competition is relaxed and consumers get some consumer surplus from product variety. In the case of winner-take-all markets, the business-stealing externality occurs because as one firm invests in quality upgrade, the competitors become more likely to lose all customers. The first chapter examines the incentives for price increase in upstream mergers when the supplier has a network of exclusive dealers (ED). The incentives explored in this paper come from changes in the threat point of the bargaining between the supplier and exclusive retailers. The bargaining power of the exclusive dealer comes from local market power of the dealer or due to reputation aspects (when dealers know that the supplier behaves opportunistically after the ED contract is signed, they will be reluctant in becoming exclusive of that supplier or renewing the contract). The change in the threat point post merger is due to the larger network of exclusive retailers, which enables the merged supplier to recapture a larger portion of the consumers that will be diverted from any specific exclusive dealer in case of disagreement on the wholesale price negotiation. The empirical application explored in this paper uses a unique and comprehensive dataset from the Brazilian fuel industry, with information that includes retail and wholesale prices as well as quantities at the station level. Aside from the good quality, this dataset is adequate for the intended analysis because in Brazil fuel stations can either operate independently (in which case they can purchase from any distributor) or sign an ED contract, when they can only purchase from a specific distributor. Moreover, the data spam a period that includes an important merger. I estimate the model using pre-merger data and simulate the effects of combining the networks of exclusive dealers of the merging companies. The simulation shows that the incentives for price increase are sizable, and the mechanism studied in the paper captures a large fraction of the actual price increase observed in the data. The second chapter, joint with Ilwoo Hwang, studies adoption and pricing when consumers can delay their purchase of a good with network effects. In those cases, price alone does not convey sufficient information for consumers to make their purchase decision and they need to infer about current and future adoption in order to make their decisions. This feature implies that some consumers might find optimal to delay their purchases in order to make their decisions better informed about the success of the network. The multiplicity of equilibria that is typical in the coordination game played by consumers implies that the demand is not well defined for a given price, creating a problem for the firm's pricing decision. We consider a two-period model in which a monopolist sets prices and consumers can delay their purchases to the second period when they will receive information about early adoption. The dynamic coordination problem with endogenous delayed purchases is modeled as a global game, for which we derive conditions for uniqueness of equilibrium. The model is capable of exploring many issues in the economics of network effects such as introductory pricing and early critical mass for platform survival. Our specification nests the pure durables goods and herding models. Numerical results illustrate the amplitude of possible outcomes in the dynamic model with delay. Substantial differences can arise in terms of pricing, adoption and profits when we compare the full specification with multiple benchmarks. In the third chapter, joint with Michael Riordan, we develop a duopoly model of product quality competition that focuses on how information structure determines equilibrium outcomes. When we introduce private and correlated signals about the fundamental uncertainty about quality differences, each firm can form a more educated guess about what the opponent must be doing, which is the key for uniqueness of equilibria. Equilibrium product improvement decisions are unique if and only if market uncertainty is sufficiently high relative to strategic uncertainty, except in a non-generic special case. A unique equilibrium takes the form of threshold strategies, whereby each firm improves its product upon receiving a sufficiently favorable signal of brand advantage. We show that the unique equilibrium depends on the fundamentals as well as on investment costs and that the probability of miscoordination vanishes as strategic uncertainty decreases. In the type of competition studied here, firms have no incentive to choose the same quality as the competition arising in the marketplace would bring prices to equalize marginal cost. Interestingly, this information structure alleviates substantially the problem of miscoordination observed in the no “information game” and also dominates the complete information game for a large range of parameters in the model