Essays on Bayesian Econometrics and Asset Pricing

My dissertation is composed of three chapters. These three chapters contribute to at least one of two areas, Bayesian Econometrics or Asset Pricing.;The first chapter of my dissertation, "Asset Pricing with Adaptive Learning and Internal Habit Persistence," investigates the extent to which the assumption of rational expectations contributes to the failure of production-based asset pricing models with internal habit persistence to match asset pricing facts. The chapter concludes minor deviations from rational expectations are not sufficient to fully match pricing statistics or create predictable returns. However, deviations push the model's statistics closer to matching the data. The work contributes to the existing macroeconomic literature by evaluating the assumption of rational expectations in asset pricing models.;The second chapter, "Contribution of a Rational Bubble to Stock Prices," uses a Bayesian perspective to decompose the S&P500; stock price index into a market fundamental and bubble component. Results indicate the contribution and role of the bubble depends on prior specification of market fundamentals. Assuming fundamental log price-dividend ratio is stationary, the bubble explains over 90 percent of the variation in prices and bubble-switching behavior is consistent with expected bubble dynamics. Moreover, this paper helps to construct a flexible econometric framework that can accommodate multiple prior beliefs while decomposing the stock price index.;The final chapter, "Probability of an Instrument Being Excludable," estimates the posterior probability of satisfying the exclusion restriction in the instrumental variables model. Typically, practitioners justify the exclusion restriction because it is sufficient for identification. Relaxing the assumption causes the model to be partially identified. The chapter takes advantage of the Bayesian perspective to study the exclusion restriction through posterior probabilities. Results illustrate that posterior probabilities are well-defined, data dependent, and take into account prior beliefs about exclusion, even without the property of identification. This implies, by incorporating information through proper prior distributions, it is possible to determine if the resulting posterior distribution supports the exclusion restriction.