Date Added: May 2009
The authors conduct an extensive analysis of return predictability for a variety of component portfolios using a large number of potential predictors from the literature on aggregate market return predictability. Focusing on three sets of component portfolios sorted on industry, size, and book-to-market, in-sample and out-of-sample tests both point to important differences in predictability across component portfolios. More specifically, they find that returns are more predictable for particular industries, including construction, textiles, apparel, furniture, printing, automobiles, and manufacturing; small-cap in contrast to large-cap stocks; and high as opposed to low book-to-value stocks. Employing a forecast combination approach, the predictability they find is robust to the use of individual predictors and particular sample periods.