We analyze horizontal mergers in Cournot oligopoly. We find general conditions under which such mergers raise price, and show that any merger not creating synergies raises price. We develop a procedure for analyzing the effect of a merger on rivals and consumers and thus provide sufficient conditions for profitable mergers to raise welfare. We show that traditional merger analysis can be misleading in its use of the Herfindahl Index. Our analysis stresses the output responses of large firms not participating in the merger.

We analyze incomplete long-term bilateral contracts when buyers incur relationship-specific setup costs and sellers choose some aspect of quality that is not verifiable to third parties. If setup costs are observable, the first best can be achieved even though contracts cannot enforceably specify quality; this does not even require long-term contracts. If setup costs are unobservable, however, then long-term price contracts can outperform short-term contracts, although they are not constrained Pareto efficient.

A good is often more valuable to any user, the more others use compatible goods. We show that this effect may inhibit innovation. If an installed base exists and transition to a new standard must be gradual, early adopters bear a disproportionate share of transient incompatibility costs. This can cause "excess inertia." The installed base, however, is "stranded" if the new standard is adopted: this may create "excess momentum." These dynamic effects have strategic implications.

We argue that although decentralization has advantages in finding low-cost solutions, these advantages are accompanied by coordination problems, which lead to delay or duplication of effort or both. Consequently, decentralization is desirable when there is little urgency or a great deal of private information, but it is strictly undesirable in urgent problems when private information is less important. We also examine the effect of large numbers and find that coordination problems disappear in the limit if distributions are common knowledge.

Although Friedrich A. von Hayek, Leonid Hurwicz, and others stressed how the price system communicates private information (on preferences, technology, and endowments), our main means of communication is surely words. But it is unclear whether mere words can be trusted: absent outside sanctions for lying (such as perjury law or a valuable reputation for honesty), talk is cheap. The perspective of the standard theory of signaling seems to suggest that cheap talk should be uninformative, but this is misleading: cheap talk can sometimes convey information and affect real (payoffrelevant) actions. Yet cheap talk's power is limited: for example, even given an unlimited chance to talk, rational people may not reach an equilibrium, or may not escape a bad one. A growing literature in game theory, political science, and economics, which I cannot survey here, studies what cheap talk can and cannot do. Here, I merely sketch some themes.

A central problem in organization of society is to choose rules for making choices when people's interests conflict. Economists have had much to say about such problems when compensating payments are possible. In practice, however, such payments are often not made, and decisions are either rigidly imposed by a central authority or taken unilaterally by one of parties concerned, who we say has the to choose outcome. In this paper, without asking why these rules are common, I ask which is most efficient. I assume that interested parties know more about their preferences than does central authority (the government). This is a reason to give one of them right to decide outcome: in this model, rights are a decentralization device (see Partha Dasgupta, 1980), and since I assume a benevolent government, there is no role for decentralization without private information. But, while this private information can make delegated decision more efficient than an undelegated and rigid central decision, there is a countervailing problem: an interested party chooses selfishly and ignores even what is common knowledge about others' preferences, which benevolent though ignorant government would take into account. Thus we find that giving an interested party right to choose is more desirable when he has important private information, but less desirable when he and others are very much in conflict.