[Statistical model selection criteria provide an informed choice of the model with best external (i.e., out-of-sample) validity. Therefore they guard against overfitting ("data snooping"). We implement several model selection criteria in order to verify recent evidence of predictability in excess stock returns and to determine which variables are valuable predictors. We confirm the presence of in-sample predictability in an international stock market dataset, but discover that even the best prediction models have no out-of-sample forecasting power. The failure to detect out-of-sample predictability is not due to lack of power.]

[An asymmetric information model of the bid-ask spread is developed for a foreign exchange market subject to occasional government interventions. Traditional tests of the unbiasedness of the forward rate as a predictor of the future spot rate are shown to be inconsistent when the rates are measured as the average of their respective bid and ask quotes. Larger bid-ask spreads on Fridays are documented. Reliable evidence of asymmetric bid-ask spreads for all days of the week, albeit more pronounced on Fridays, are presented. The null hypothesis that the forward rate is an unbiased predictor of the future spot rate continues to be rejected. The regression slope coefficients increase toward unity, however, indicating a less variable risk premium.]

Neoclassical finance assumes that investors are Bayesian. In many realistic situations, Bayesian learning is challenging. Here, we consider investment opportunities that change randomly, while payoffs are observable only when invested. In a stylized version of the task, we wondered whether performance would be affected if one were to follow reinforcement learning principles instead. The answer is a definite yes. When asked to perform our task, participants overwhelmingly learned in a Bayesian way. They stopped being Bayesians, though, when not nudged into paying attention to contingency shifts. This raises an issue for financial markets: who has the incentive to nudge investors?

[We derive and test a dynamic discrete-time model of asset returns. Both the risks of individual securities and equilibrium risk premia change predictably in the model, but these changes can be attributed to movements in the returns and prices of only two well-diversified portfolios. Any other components of returns should be unpredictable. Using the generalized method of moments, the model is estimated and tested on portfolios of equities. We find the data supportive of the model's restrictions, even when instruments designed to capture the January effect are employed.]

Procedures are presented that allow the empiricist to estimate and test asset pricing models on limited-liability securities without the assumption that thehistorical payoff distribution provides a consistent estimate of the market's priorbeliefs. The procedures effectively filter return data for unspecified historical biases in the market's priors. They do not involve explicit estimation of the market's priors, and hence, economize on parameters. The procedures derive from a new but simple property of Bayesian learning, namely: if the correct likelihood is used, the inverse posterior at the true parameter value forms a martingale process relative to the learner's information filtration augmented with the true parameter value. Application of this central result to tests of asset pricing models requires a deliberate selection bias. Hence, as a by-product, the article establishes that biased samples contain information with which to falsify an asset pricing model or estimate its parameters. These include samples subject to, "e.g." survivorship bias or Peso problems. Copyright The Review of Economic Studies Limited, 2004.

[We analyze theoretically and empirically the implications of information asymmetry for equilibrium asset pricing and portfolio choice. In our partially revealing dynamic rational expectations equilibrium, portfolio separation fails, and indexing is not optimal. We show how uninformed investors should structure their portfolios, using the information contained in prices to cope with winner's curse problems. We implement empirically this pricecontingent portfolio strategy. Consistent with our theory, the strategy outperforms economically and statistically the index. While momentum can arise in the model, in the data, the momentum strategy does not outperform the price-contingent strategy, as predicted by the theory.]

This article derives testable restrictions on equilibrium asset prices when investors have the option to time the realization of their capital gains and losses for tax purposes. The tax-timing option alters both the magnitude and timing of equity returns relative to those in a tax-free model. The tax-induced restrictions are empirically examined, and the tax rates and preference parameters are estimated. While the tax-free model can be rejected in favor of the tax-based model as the specified alternative, the tax-based model is still unable to adequately explain cross-sectional differences in asset returns.

Financial decision making under time pressure, though ubiquitous, is poorly understood; classical and behavioral finance are silent about the time required for a decision to be made. In an experiment, calibrating allowable decision times to 1, 3, and 5 s, we find that classical moment-based preferences reflect time-invariant sensitivity to expected reward, purchase impulsiveness under extreme time pressure, and decreased aversion to variance and increased aversion to skewness with decision time. These time-varying sensitivities translate into increased probability distortions and decreased risk aversion for gains under prospect theory (PT). Strikingly, moment-based theory provides a better fit than PT.

Speculators buy an asset hoping to sell it later to investors with higher private valuations. If agents are uncertain about the distribution of private valuations and about the beliefs of others about this distribution, a beauty contest with an infinite hierarchy of beliefs arises. Under Harsanyi's assumption of a common prior the infinite beliefs hierarchy is readily solved using Bayes' law. This paper shows that common knowledge of the “beliefs formation rule,” mapping the private valuation of each agent into his first-order belief, also simplifies the beliefs hierarchy while allowing for disagreement among agents. We analyse the resulting speculation in a stylized asset market. Several statistics, computed only from readily observable quote, return and volume data, are evaluated in terms of their power to discriminate between genuine disagreement and the Harsanyian case. Only statistics that relate volume and volatility, or volume and changes in best offers, have the necessary discriminatory power.

Statistical model selection criteria provide an informed choice of the model with best external (i.e., out-of-sample) validity. Therefore they guard against overfitting (“data snooping”). We implement several model selection criteria in order to verify recent evidence of predictability in excess stock returns and to determine which variables are valuable predictors. We confirm the presence of in-sample predictability in an international stock market dataset, but discover that even the best prediction models have no out-of-sample forecasting power. The failure to detect out-of-sample predictability is not due to lack of power.