Conditioning Information, Out-of-Sample Validation, and the Cross-Section of Stock Returns

摘要: Empirical research on conditional asset pricing has been built several standard return-predictive variables. However, recent studies have raised serious doubts these variables that typically serve as the instruments to capture relevant conditioning information. In stochastic discount factor framework, we propose and implement a new approach assess value of instruments. We compare out-of-sample performances models are different subsets widely-used find some combinations instruments, after adjusting for effect horse-race over all subsets, can significantly improve performance cross-section stock returns. contrast, other give rise drastically underperform unconditional model. The results affirm cross-sectional highlight importance instrument selection.