Journal of Empirical Finance最新文献

We evaluate US market return predictability using a novel data set of several hundred ag- gregated firm-level characteristics. We apply LASSO, Elastic Net, Random Forest, Neural Net, Extreme Gradient Boosting, and Light Gradient Boosting Machine methods and find these models experience large prediction errors that lead to forecast failures. However, winsorizing and pooling machine learning model forecasts provides consistent out-of-sample predictability. To assess robustness, we apply machine learning methods to high-dimensional data for Canada, China, Germany and the UK as well as the Goyal–Welch data. All machine learning models we consider, except for the ensemble pooled methods, fail to significantly predict returns across our samples, highlighting the importance of pooling, evaluating additional economies, and the fragility of individual machine learning methods. Our results shed light on the sparsity versus density debate as the degree of sparsity and variable importance evolves over time.

This paper examines theoretically and empirically a variance-dependent pricing kernel in the continuous-time two-factor stochastic volatility (SV) model. We investigate the relevance of such a kernel in the joint modeling of index returns and option prices. We contrast the pricing performance of this model in capturing the term structure effects and smile/smirk patterns to discrete-time GARCH models with similar variance-dependent kernels. We find negative and significant risk premium for both volatility factors, implying that investors are willing to pay for insurance against increases in volatility risk, even if it has little persistence. In-sample, the component GARCH model exhibits a slightly better fit overall and across all maturity buckets than the two-factor SV model. However, the two-factor SV model reduces strike price bias, giving rise to the model’s ability in reconciling the physical and risk-neutral distribution. Out-of-sample, the two-factor SV model has better fit to data.

We examine the relationship between stock price synchronicity and stock liquidity using a comprehensive data set across 40 countries. Our local (within-country) empirical results reveal a positive relationship between local synchronicity and stock liquidity. The strength of this positive relationship depends on the quality of country-level institutions; the weaker the institutional environment, the stronger the synchronicity-liquidity relationship. Importantly, our global (across-country) findings mirror those at the local level. Overall, our study provides a comprehensive analysis of the synchronicity-liquidity relationship at both the local and global levels. In addition, our cross-sectional analyses provide new evidence on the institutional determinants of this relationship.

We use Markov chain Monte Carlo (MCMC) sampling to draw model coefficients to generate LGD distributions. We find that applying this Bayesian method on a sophisticated model, such as the zero-one-inflated beta (ZOIB) model, that accounts for the bi-modal distribution of the LGDs can generate LGD distributions that mimic the observed distributions well. By contrast, applying this Bayesian sampling approach on a simple model such as Tobit cannot capture the bi-modal LGD distributions accurately. Finally, we argue that this Bayesian sampling approach to generate LGD distributions is better fit for the stress testing purpose than the typical approach to estimate LGD model coefficients and then stress the macro variables. The latter approach yields stressed LGDs that may not be conservative enough, even if the macro variables are stressed to their worst historical values.