When running a regression to empirically test models like CAPM or the Fama and French Model, why do we test the statistical significance of the intercept? Do we ignore the residual error?
Why not analyze the residual instead of the alpha? Wouldn't the residuals be values not explained in the model?
