We show how multivariate GARCH models can be used to generate a time-varying “information share” (Hasbrouck, 1995) to represent the changing patterns of price discovery in closely related securities. We find that time-varying information shares can improve credit spread predictions.

With moral hazard and anonymous asset trade, first-order conditions need not characterize effort and portfolio choices. The standard procedure for establishing validity of the first-order approach in economies with one hidden asset is not fruitful when multiple assets are hidden.

In a global game, larger ambiguity is shown to decrease the amount of coordination each player perceives. Consequently, small uncertainty tends to select the Pareto dominated equilibrium of the game without uncertainty. Implications for models of financial crises are drawn.

Subjects who overestimate their performance in experimental tasks unrelated to travel are less willing to insure against failing in the task and also less inclined to buy travel insurance. This suggests intrinsic optimism influences insurance demand and diminishes adverse selection.

For gold, moving from clandestine to official trading does not significantly change informational efficiency. Both markets are inefficient suggesting that efficiency is linked more to the type of asset than to the legal status of the market.

We analyze a strategic trading model where an overconfident insider is required to publicly disclose his trades after the fact. We find the more confident insider is more concerned about the effect the initial trading has on the future.

Estimates of agents’ risk aversion differ between market studies and experimental studies. We demonstrate that these estimates can be reconciled through consistent treatment of agents’ propensity for narrow framing.

We model a reputation game, in which a sequence of short-run players chooses if to interact with a long-run player. Although beliefs may be identical, choices may be different, as not-interacting can lead the long-run player to improve on effort.

We use extreme value theory to analyse the tails of a momentum strategy’s return distribution. The asymmetry between the fat left tail and thin right tail strongly reduces a momentum strategy’s prospective utility levels.