QUANT- please clear this up for me

CFA Level II Quant
#1

Hope studying is going well for everyone…

I wanted to clarify some things from quant. I will put my conclusions below, and if anything is wrong id love for some of you quant gurus to chime in (because i absolutely suck at quant)

  1. Heteroskedasticity- coefficients will still be consistent, but biased. standard errors will be biased and inconsistent

  2. Serial correlation- coefficients will still be consistent, but biased. standard errors will be biased and inconsistent

  3. multicollinearity- coefficients will be consistent, but biased. standard errors, again, will be biased and inconsistent

  4. omitted variable- coefficients will be inconsistent and biased. This is the only one in which coefficient estimates are actually inconsistent.

I have no idea if whatever i just typed makes sense, so im hoping you guys can help me out by explaining what all this means a little more. Im running into way too many mock questions about this to not have a deeper understanding

#2

Almost. To be more precise:

  1. Heteroskedasticity- coefficients will still be consistent. Standard errors will likely be too small (but could be too big) and may causes type I errors (when you think something is significant when it should be not significant)

  2. Serial correlation- coefficients will still be consistent. Standard errors will likely be too small, which causes type I errors (when you think something is significant when it should be not significant)

  3. multicollinearity- coefficients will be consistent. Standard errors will likely be inflated and likely causes type II errors (when you think something is insignificant when it should be significant)

#3

Coefficients are not unbiased when multicollinearity is present

#4

Yes, they are, assuming multicollinearity is the only issue. Bias is a calculable quantity. Coefficient estimates being “unreliable” does not make them biased in a technical sense.