Heteroskedasticity vs Serial correlation

Can anyone help why autorcorrelation is a bigger of a deal breaker here (or am i missing something here?) ?!


An analyst wishes to test whether the stock returns of two portfolio managers provide different average returns. The analyst believes that the portfolio managers’ returns are related to other factors as well. Which of the following can provide a suitable test?

A) Model misspecification.

B) Heteroskedasticity. C) Serial correlation.


One of the primary assumptions of linear regression is that the residual terms are not correlated with each other. If serial correlation, also called autocorrelation, is present, then trend models are not an appropriate analysis tool.

Thanks in advance guys!!

I’m not too sure how well I understand that explanation, however, I noticed that answer B lists Heteroskedasticity, NOT Conditional Heteroskedasticity. Unconditional Heterskedasticity is not an issue (at least in the context of the CFA exam).

that makes sense actually, thanks!

answer here is C?


I think this one it’s C because with autocorrelated residuals, C., you can still run their returns against a benchmark and get an R-squared to attempt to explain their results. The variance size of residuals doesn’t change, just their alignment. It makes projections damn near impossible, but explanations of goodness-of-fit are typically still okay.

With B., we can’t trust our estimators because the variance of our residuals keeps changing, and we can’t average it out… with this data we can’t explain much, but can do a little forecasting still; that won’t help us here.

If the model is misspecified, A., we’re kinda in trouble before we even get started.