[The existence of perfect equilibrium is demonstrated for a class of games with compact space of histories and continuous payoffs, and in which the set of actions feasible at any given period is a lower hemicontinuous correspondence of the previous history of the game. The proof is by construction. A set of histories is constructed, each of which is the equilibrium path of some perfect equilibrium point of the game. Also, any equilibrium path is a member of this set. The construction therefore provides a characterization of perfect equilibrium.]

Extending Barro (

Laboratory and field studies of time preference find that discount rates are much greater in the short-run than in the long-run. Hyperbolic discount functions capture this property. This paper solves the decision problem of a hyperbolic consumer who faces stochastic income and a borrowing constraint. The paper uses the bounded variation calculus to derive the Hyperbolic Euler Relation, a natural generalization of the standard Exponential Euler Relation. The Hyperbolic Euler Relation implies that consumers act as if they have endogenous rates of time preference that rise and fall with the future marginal propensity to consume (e.g., discount rates that endogenously range from 5% to 41% for the example discussed in the paper).

We describe two auction forms for search engine advertising and present two simple theoretical results concerning i) the estimation of click-through rates and ii) how to adjust the auctions for broad match search. We also describe some of the practical issues involved in implementing a VCG auction.

The paper presents two models of races in which there is both technological uncertainty and strategic interaction between competitors as the race unfolds. Most of the existing literature examines one or other of these features, but not the two combined. Our aim is to see how the efforts of competitors in a race vary with the intensity of rivalry between them. In our principal model, whch is of a one-dimensional race, it is shown that the leader in the race makes greater efforts than the follower, and efforts increase as the gap between competitors decreases. Under certain conditions the same results hold in our second, related model, which is of a two-dimensional race.

This paper investigates perfect equilibrium in a model of a race in which two players are competing for an indivisible prize. The winner is the first player to reach the finishing line. It is shown that the behaviour of the winner of the race is often exactly as if he were the only player: the rival makes no difference. Even if competition does affect the winner's behaviour, it does so only in the first stage of the race and not thereafter. It is shown how several factors combine to determine which player will win: relative valuations of the prize, discount rates, efficiency at making progress and initial distances from the finishing line. Insofar as the model applies to patent races, it suggests that the potential competition faced by one firm in a patent race (e.g. an incumbent monopolist) may be of little or no consequence.

This paper extends the classic two-armed bandit problem to a many-agent setting in which N players each face the same experimentation problem. The main change from the single-agent problem is that an agent can now learn from the current experimentation of other agents. Information is therefore a public good, and a free-rider problem in experimentation naturally arises. More interestingly, the prospect of future experimentation by others encourages agents to increase current experimentation, in order to bring forward the time at which the extra information generated by such experimentation becomes available. The paper provides an analysis of the set of stationary Markov equilibria in terms of the free-rider effect and the encouragement effect.

This paper is an attempt to identify some of the factors that affect the evolution of market structure in a model of dynamic competition between two firms. The stochastic evolution of the state of competition depends on the respective effort rates of the firms. The question is whether the current leader works harder than the laggard—does the ‘gap’ between firms tend to increase or decrease? We show that several effects are at work. The state tends to evolve in the direction where joint payoffs are greater. Since joint payoffs are related to joint product-market profits less joint effort costs, there are two classes of effect: the joint-profit effect and various joint-cost effects. The latter result in part from the pattern of profits, and in part from endpoint effects that give relief from efforts. Asymptotic expansions illuminate these influences. Moreover, we show by numerical simulation that there is another kind of joint-cost effect. The pattern of joint effort costs can influence the pattern of evolution of market structure, and the evolution of the pattern of market structure can influence the pattern of efforts, in a mutually self-reinforcing manner. In particular, there may be equilibria in which this last effect means that the laggard works harder than the leader even though all the other effects work in favour of the leader.