Hi everyone, The following assumptions are made for the Market Model in reading 66 (page 370 CFAI): 1) The expected value of the error term is 0 => E(e)=0 2) The market return (Rm) is uncorrelated with the error term => Cov(Rm,e) =0 3) The error terms are uncorrelated among different assets. Can someone connect the dots for me… these assumptions have been used before I think. In the reading on regression models? Are there specific names/concepts for these assumptions I’m supposed to know? Quant was February for me…I have a feeling I’m being stupid/my brain is fried and I’m not connecting things like I should. Thanks! PS: Unrelated question, has anyone counted the LOS’s… how many are there? Do we know?
1.) If the error terms have an expected value, i would assume that would mean you have serial correlaton 2.) If your independent variable (return on the market) is correlated with the error term of your regression, then you potentially have conditional heteroskedasticity 3.) not sure
If the error terms are correlated then the model could be subject to Serial Correlation
“uncorrelated across assets” - I take this to mean across various regressions (since you would regress each stock against the index separately)
- Kind of covariance stationary 2. Conditional Heteroskedasticity 3. Serial Correlation (AutoCorrelation)