A0872
Title: Asset-pricing factors with economic targets
Authors: Sicong Allen Li - London Business School (United Kingdom) [presenting]
Victor DeMiguel - London Business School (United Kingdom)
Svetlana Bryzgalova - London Business School (United Kingdom)
Markus Pelger - Stanford University (United States)
Abstract: A method is proposed for estimating latent asset-pricing factors that incorporate economic information. Our estimator generalizes Principal Component Analysis (PCA) by penalizing deviations of the factors from economically motivated cross-sectional and time-series targets. Using a high-dimensional dataset containing decile-portfolio returns for 37 characteristics, we show that a cross-sectional target that aligns factor composition with firm characteristics and a time-series target that nudges factors to explain decile-portfolio alphas help to span pricing kernels with substantially higher Sharpe ratio and lower pricing error than conventional PCA factors.
