Do we have to know how to actually detect heteroskedasticity and autocorrelation, ie: actually perform the BP and DW tests?? According to the LOS’s, no where does it say we have to, however, in the Schweser books and in the real CFA books, there are examples/problems involving these calculations. Thought???..tks
Technically we know how to calculate the BP test. Take the independant variables and regression them against the squared errors of the regression. Then take the R^2 of the regression and multiply by N. There are much more important concepts to test you then this. I would know the basic form of the test N x R^2 of the second regression and test by Chi-test. If you regret the null then you have a problem. Just remember that if the regression does explain a large portion of the squared residuals errors then we do have a problem and then use adjusted white-t-test to test the orginal coeffecients. At most they will have one problem on the exam on this.