Abstract:
We compare three implementation schemes of an infinite-horizon monetary economy with discounting. Under the standard random termination scheme and its block variation, the economy lasts for an indefinite number of periods and the discounting factor is captured by the probability that the economy continues to the next period. These schemes rely on the belief that the experimenter can credibly implement a game that lasts an arbitrarily long time. We also propose a new method that does not rely on such a belief. Under this scheme, subjects participate in an experiment for a fixed number of periods where the discount factor is captured by a weighting factor that shrinks the payoffs over time. Dynamic incentives are preserved by paying subjects their continuation value, which is based on past market prices. The results show that dynamic incentives are preserved, and behavior is similar in all three implementations. Researchers may decide among these approaches, depending on the research question of interest and more practical concerns, such as the ease of implementation and the need to collect data for multiple supergames when the discount factor is high.