Conditional Asset Pricing with a Large Information Set

Dynamic factors summarize the information in a large number of variables and are therefore intuitively appealing proxies for the information set available to investors. This paper demonstrates that conditioning on dynamic factors instead of commonly used instruments substantially reduces the pricing errors implied by conditional models. Dynamic factors are further shown to exhibit incremental explanatory power over benchmark conditioning variables. The results withstand a number of robustness tests and carry important implications for the specification of conditional asset pricing models in applied research and practice.