How to Construct Fundamental Risk Factors?

Our paper reexamines the methodology of Fama and French (1993) for creating US empirical risk factors, and proposes an extension on the way to compute the mimicking portfolios. Our objective is to develop a modified Fama and French (F&F) methodology that could be easily implemented on other markets, and that could also easily price other risk fundamentals. We raise three main problems in the F&F methodology. First, their annual rebalancing is consumptive in long timeseries which sometimes simply do not exist for small exchange markets. Moreover, this does not match with the investment horizon of the investors. Second, the independent sorting procedure underlying the formation of the 6 F&F twodimensional portfolios causes moderate level of correlation between premiums. Finally, sorting the stocks into portfolios according to NYSE stock returns tend to over-represent the proportion of small stocks in small and value portfolios. We estimate, along our technology, alternative premiums for the size, book-to-market and momentum risk fundamentals. We compare these three risk premiums to the Fama and French and Carhart benchmarks that Kenneth French make available on his website. In an analysis framework without data snooping bias, we show evidence that although they are correlated, the original F&F premiums and our versions of the F&F premiums bring complementary information. Furthermore, we find that our empirical model better complements the market model for explaining cross-sectional dispersion in returns than the F&F premiums.