In one of the questions, 1. " Conditional Heteroskedasticity will result in consistent coefficient estimates but both the t-stastics and F-statistics will be biased" resulting in False Inferences 2. If an omitted variable is correlated with variables already included in the model, coefficient estimates will be biased and incosistent and standard errors will also be inconsistent" For Q1. We have probably come across zillion times that if there is a conditional heteroskedasticity then standard errors will be small and t-statistics will be high and biased F-statistics also. But the interesting thing is, the answer says the standard erros could be up or down? Does this make sense For Q2. Once you omit the correlated variable, why would they be any errors. May be I am intrepreting the question wrong. Thanks again

q1: what happens if the standard errors were too low? You would have higher T-stats, and choose you to accept your null hypotheses more. if the std. errors were too high -> lower T-Stats and you would end up rejecting your null hypotheses. std. error is distance between the predicted value and the actual value of dependent variable. in a conditional heterosked. case - values could be high or low, more often than not, it would be high. q2: The regression equation is only a model, with which you are trying to predict things. Just because you removed a variable - errors are not going to totally disappear. They would at best reduce, and help you make a better prediction than before.

Thanks CPK. you are the best