This dissertation consists of three chapters on topics in the domain of behavioral and quantitative finance.;In the first chapter, I apply the model of salience theory of Bordalo et al. (2012) to asset prices and empirically test the prediction that a stock whose expected return distribution exhibits a high (low) salience theory value looks attractive (unattractive) to an investor, hence will be overvalued (undervalued), and thus earn a lower (higher) subsequent return. To test this empirically, I follow Barberis et al. (2014), whose objective is to test the predictive power of the prospect theory value of a stock's historical return distribution. In U.S. stock market data for the period from 1926 to 2014, I find that a stock's salience theory value does have significant predictive power, but that prospect theory works better in that it renders a stock's salience theory value insignificant as predictor when a stock's prospect theory value is included as a control in a Fama-MacBeth framework. In addition, I propose a formulation of salience theory that deviates from the original formulation in Bordalo et al. (2012) in that it does not only account for the salience ranking of payoffs, but also for the magnitude of their salience. I find that this alternative formulation empirically outperforms both the original formulation of salience theory and prospect theory in terms of predictive power.;In the second chapter, I revisit the Forward Rate Unbiasedness Hypothesis (FRUH). The FRUH states that forward exchange rates are unbiased predictors of future spot exchange rates. Early regressions of the log spot rate on the log forward rate supported this hypothesis, with estimates of the slope coefficient close to one. However, subsequent regressions of the log spot return on the log forward premium produced estimates of the slope coefficient that were not close to one, and often even negative. One explanation for this seemingly contradictory finding that the literature provides is a cointegrating relationship between the log spot rate and the log forward rate with cointegrating vector (1, ---1). I use a new inference procedure (IM-OLS, proposed by Vogelsang and Wagner (2011), with small-b and fixed-b asymptotics) to test this hypothesis; for the classical (small-b) case, I compare the results of IM-OLS to the established methods FM-OLS and D-OLS. The attractive feature of fixed-b asymptotics is that the choice of tuning parameters (bandwidth, kernel) are reflected in the asymptotics. To shed light on the influence of tuning parameters on the results, I execute all tests for a range of parameter combinations. I find that IM-OLS is robust to the choice of tuning parameters, and that it produces evidence in support of the FRUH. However, a nonparametric specification test (Kasparis and Phillips (2012)) strongly rejects a linear relationship between the log spot and the log forward rate, and thus the FRUH. This casts doubt on the validity of my results, as well as on the existing literature, which usually employs a linear specification.;The third chapter is joint work with Oliver Linton, Xiaohong Chen and Yapping Yi. We propose new methods for estimating the bid-ask spread from observed transaction prices alone. Our methods are based on the empirical characteristic function instead of the sample autocovariance function like the method of Roll (1984). As in Roll (1984), we have a closed-form expression for the spread, but this is only based on a limited amount of the model-implied identification restrictions. We also provide methods that take account of more identification information. We compare our methods theoretically and numerically with Roll's method as well as with its best known competitor, the approach of Hasbrouck (2004), which uses a Bayesian Gibbs methodology under a Gaussian assumption. Our estimators are competitive to Roll's and Hasbrouck's when the latent true fundamental return distribution is Gaussian, and perform much better when this distribution is far from Gaussian. Our methods are applied to the E-mini futures contract on the S&P 500 during the Flash Crash of May 6, 2010. We present extensions to models that allow for unbalanced order flow, or Hidden Markov trade direction indicators, or trade direction indicators having general asymmetric support, or adverse selection, all without requiring additional data.
展开▼
