Determining the Rank of the Beta Matrix in a Factor Model with Factor-Candidate Regressors

Seung C. Ahn, Alex R. Horenstein, Na Wang
2010 Social Science Research Network  
We consider the estimation methods for the rank of a beta matrix generated by risk factors and study which method would be appropriate for data with a large number (N) of risky assets. We find that a restricted version of Cragg and Donald's (1997) Bayesian Information Criterion (BIC) estimator is quite accurate even if N is large. Using twenty-six empirical factors for U.S. stock returns, we show that beta matrices from many multifactor asset-pricing models fail to have full column rank,
more » ... column rank, suggesting that risk premiums in these models are under-identified and that many empirical factors are redundant.
doi:10.2139/ssrn.1567933 fatcat:k2emqj567vb4nmybbzmzu5e22a