How Long is Forever in the Laboratory? Three Implementations of an Infinite-Horizon Monetary Economy

Standard monetary models adopt an infinite horizon with discounting, where the infinite horizon helps preserve the dynamic incentive to accept money. Since people cannot directly derive utility from money, subjects have no incentive to value money in the last period of a finite-horizon game. Testing monetary models in the lab requires the implementation of the infinite horizon within a limited time frame. We compare three approaches to do so.

The first two approaches are based on random termination, where the game continues to the next period with a probability equal to the discounting factor. Under random termination, the game may last for an arbitrarily long time. Therefore, these approaches are subject to the criticism that they cannot be credibly implemented within a limited time frame. We propose a third approach that is free from this criticism, as subjects play a fixed number of periods and their payoffs are discounted over time. With this new approach, the dynamic incentives at the end of the game are preserved by paying subjects a continuation value calculated from past outcomes.

We find that all three approaches preserve the dynamic incentives, and subjects use money to trade throughout the game. In most experimental economies, the observed outcomes are close to what the theory predicts. As a result, researchers can choose any of the three approaches depending on the question of interest.