Book-to-market and size effects: compensations for risks or outcomes of market inefficiencies?

We employ the optimal orthogonal portfolio approach to investigate if the size and book-to-market effects in US data are related to risk factors beside the market risk. This method enables us to estimate the upper limit of the risk premium, due to observed as well as all possible unobserved factors, which can be derived from a linear asset pricing model. As a corollary, it is possible to divide the observed average asset return into three parts: one explained by the market factor, one due to the unobserved factors, and finally the non-risk-based (NRB) component. Our empirical results confirm the existence of latent risk factors, which cannot be captured by the market index. In particular, the size effect is related to some other background risk factors than the market portfolio, but a large part of observed book-to-market effect has a NRB explanation.