We study a long-horizon, oligopolistic market with random shocks to demand that can be arbitraged by two large storage operators with finite capacity. This problem applies to any storable commodity — that is, most commodities. Because the arbitrage spread is so sensitive to market power, storage operators face strong incentives to restrain quantities by tacitly colluding. This cooperation takes new forms thanks to the multiplicity of actions they must take: selling, buying or both. We construct payoff-maximizing equilibria of this stochastic game, and uncover a new form of Partial Cooperation that trades off quantities and delay. While collusive, Partial Cooperation delivers higher consumer surplus thanks to the market power effect. Head-on competition is not always an equilibrium of the long-horizon game, unlike many standard games, when market power becomes large enough. We present some robustness checks. We also draw implications for policy and suggest poorly competitive storage is a negative externality to the development of the underlying commodity — for example, renewable energy.

Download

We consider a dynamic model of an oligopolistic market with demand shocks, in which a storage unit buys and sells over time subject to a capacity constraint. To make progress in this stochastic game with constraints, we restrict attention to simple heuristics, and we can characterise the optimal policy of a storage unit in this restricted class of heuristics. The heuristics, the exogenous stochastic process and the capacity constraint interact to induce rich dynamics. The optimal policy is sensitive to the nature of demand shocks and to storage capacity. For a fixed capacity, the storage unit internalises its unilateral market power; it acts like a monopolist on its arbitrage spread. We uncover a new phenomenon that we call continuation risk. It is a corollary of market power and induces the optimal capacity to be interior even absent investment cost. We discuss some implications. This work applies to any storable commodity such as crops, raw materials or fuels, and more recently, electricity.

Download

We consider decentralized matching in a two-sided market of employers and employees (universities and students) with application costs and limited budgets. Students choose whether they should take the risk of applying to more prestigious school (with some probability of rejection) or make a safe choice. We show that application costs set by universities may be treated as a screening instrument in order to attract only strong students or to avoid the competition. Also, we find conditions for equilibria when costs are negligible, costs are a substantial part of the budget or even the entire budget, in addition to other equilibria.

Download

We consider multi-period decentralized matching in a two-sided market of employers and employees. Because of dynamics, asymmetry, and private information, we always observe delays and coordination failures with nonzero probability even in a simple case of two agents from each side. For this, we obtain conditions for different strategies to be in equilibrium. We show that problems of miscoordination and delay may be solved by signaling or by incentivizing immediate response that turn a dynamic problem into a static one and make the matching stable. In the general case, we obtain sufficient conditions for assortative matching to be and not to be in equilibrium.

Download

The concept of iterated regret minimization (IRM) provides solutions that are more reasonable than ones offered by Nash equilibrium (NE) for many games of interest, such as the Traveler's Dilemma, the Centipede Game, and many others. For them, we analyzed new data — the IRM approach prescribes the most profitable strategy better than NE. Also, we apply the IRM to the games with continuous strategy sets, such as Colonel Blotto game and Bertrand-Edgeworth duopoly, and obtain reasonable pricing and allocating policies.

Download

We study the monopoly problem of a large-scale storage unit that faces a periodic but uncertain demand over multiple cycles. Time is continuous and strategies are functions of time expressed in terms of power rates (rather than quantities). Storage buys in periods of below-average demand and sell when demand exceeds the mean. We characterize the selling and buying strategies exactly as a pair of (time-varying) intensity and stopping time for different information structures. This flexibility enables the operator to smooth the impact of its market power over time; in turn, this limits demand smoothing. When the capacity is not too large, the storage operator trades that capacity in full every cycle, even under rate (dis)charge constraints.

In many situations of economic interest, economic agents decide whether to undertake a costly action with an uncertain payoff, but they can learn about the profitability of the action by waiting and observing others' behavior. Examples include investment decisions, market entry, and product adoption. To model these scenarios, we propose a dynamic endogenous-entry game with uncertain payoffs, where individuals decide whether and when to enter. While delaying entry incurs costs, postponing decisions allows agents to benefit from an informational externality, as the actions of others provide additional information about the state of the world. We demonstrate that the presence of this informational externality and the tension between early entry and strategic, costly delay can lead to market failures — excessive entry in unfavorable conditions and insufficient entry in favorable ones — and rational herding in equilibrium, suggesting a role for public intervention. We then show that subsidizing entry does not enhance welfare. In contrast, subsidizing waiting — and thus facilitating learning about the state of the world — can improve welfare across a wide range of parameter constellations.

We study how the structure of a network in terms of number of agents, feasible actions, and absorbability of actions of others affects the agent's trade-off between the incentive to delay actions in order to infer information by observing actions of others and the incentive to move early to avoid costs of delay.

We consider decentralized matching in a two-sided market of employers and employees, where each side has two types of agents, high and low. With limited capacity of employers and if both employees' types are weak enough, we observe an interesting result: in order to avoid competition with high-type employees, low-type employees send applications to high-type employers.

We analyze the dynamic application process where the probability of success is determined by the type of an applicant. In the case the type is private and may be revealed to jury only with some random noise, we prove that agents maximize their profits by aiming higher in the first rounds. However, this strategy may change if an applicant herself misinterprets her type with a positive (overconfidence) or a negative (underconfidence) bias.