This is an attempt to expose the essence of Samuelson's consumption loan paradox. It is maintained that the double infinity of traders and dated commodities allows for competitive equilibria that are not Pareto-optimal. While such models are most interesting in the dynamic setting, the fact that generations do not meet is not essential. The chain-letter aspect of the model reminds us that the appropriate form of the budget constraint is not obvious for the potentially infinitely long-lived economic entity (such as the corporation or the family). The analysis is related to recent contributions in the theories of general equilibrium, economic planning, and decentralization.

An imperfectly competitive economy is very prone to market uncertainty, including uncertainty about the liquidity (or “thickness”) of markets. We show, in particular, that there exist stochastic equilibrium outcomes in nonstochastic market games if (and only if) the endowments are not Pareto optimal. We also provide a link between extrinsic uncertainty arising in games (e.g. correlated equilibria) and extrinsic uncertainty in market economies (e.g. sunspot equilibria). A correlated equilibria to the market game is either a sunspot equilibrium or a non-sunspot equilibrium to the related securities games, but the converse is not true in general.

We analyze a banking system in which the class of feasible deposit contracts, or mechanisms, is broad. The mechanisms must satisfy a sequential service constraint, but partial or full suspension of convertibility is allowed. Consumers must be willing to deposit, ex ante. We show, by examples, that under the so‐called “optimal contract,” the postdeposit game can have a run equilibrium. Given a propensity to run, triggered by sunspots, the optimal contract for the full predeposit game can be consistent with runs that occur with positive probability. Thus the Diamond‐Dybvig framework can explain bank runs as emerging in equilibrium under the optimal deposit contract.

Can extrinsic uncertainty ("animal spirits," "market psychology," "sunspots,"...) play a significant role in rational expectations equilibrium models? We establish that extrinsic uncertainty cannot matter in the static Arrow-Debreu economy with complete markets. But we also establish that extrinsic uncertainty can matter in the overlapping-generations economy with complete markets but where market participation is limited to those consumers alive when the markets are open. Equilibrium allocations in which extrinsic uncertainty plays no role are Pareto optimal in the traditional sense. Equilibrium allocations in which extrinsic uncertainty does play a role are Pareto optimal in a (weaker) sense which is appropriate to dynamic analysis.

Can extrinsic uncertainty ("animal spirits," "market psychology," "sunspots,"...) play a significant role in rational expectations equilibrium models? We establish that extrinsic uncertainty cannot matter in the static Arrow-Debreu economy with complete markets. But we also establish that extrinsic uncertainty can matter in the overlapping-generations economy with complete markets but where market participation is limited to those consumers alive when the markets are open. Equilibrium allocations in which extrinsic uncertainty plays no role are Pareto optimal in the traditional sense. Equilibrium allocations in which extrinsic uncertainty does play a role are Pareto optimal in a (weaker) sense which is appropriate to dynamic analysis.