Research

Following research is ongoing.

• Predicting Equity Premium with Shrinkage Estimation Guided by Economic Theory
  with Xiaobin Liu, Yanchu Liu and Tao Zeng
  Abstract： We demonstrate that economic linkages between predictors and real economy, along with theory-implied parameter restrictions, can substantially enhance the out-of-sample forecasting performance of equity premium predictions. We introduce a novel empirical Bayes (EB) framework that effectively captures these economic relationships while accommodating theoretical constraints on predictor coefficients. Our proposed EB methodology is data-driven and optimally shrinks coefficient estimates toward an intermediate value between unrestricted OLS estimates and theory-implied restricted estimates. Empirical applications to equity premium prediction reveal improved forecasting performance across numerous established predictors. Using two comprehensive datasets, we document the superior performance of EB forecasting compared to traditional OLS methods. The improvements become particularly pronounced when implementing classic economic constraints following Campbell and Thompson (2008) and Pettenuzzo, Timmermann, and Valkanov (2014). These findings provide new insights into the crucial role of economic theory in time-series predictability and demonstrate how theory-implied parameter restrictions can enhance predictive accuracy. Our results underscore the importance of incorporating economic structure into forecasting models.
  This research is ongoing, and further developments will be provided in future versions.
  Keywords： Economic theory, Empirical Bayes, Macroeconomic predictors, Aggregate market excessive return