In an experiment, a group of strangers was randomly divided in pairs to play a prisoners’ dilemma; this process was indefinitely repeated. Cooperation did not increase when subjects could send public messages amounting to binding promises of future play.

In a trust game experiment, we elicited choices using either the so-called game or strategy method. While the two methods yield similar rates of trust, the strategy method reveals a significantly lower rate of trustworthiness.

Survey data show that subjects positively discount both gains and losses but discount gains more heavily than losses. This holds for monetary and non-monetary outcomes. These results do not confirm the findings of two earlier studies about negative time preferences for non-monetary outcomes.

We study a decentralized trading model as in Peters (1984a), where heterogeneous market participants face a trade-off between price and trade probability. We present a novel proof of existence of a unique demand vector in Nash equilibrium, based on a recursive approach that exploits the...