quant: reading 12: page 425

why" if an omitted variable is correlated with variables already included in the model, coefficient estimates will be biased and inconsistent and standard errors will also be inconsisitent"… any variable suppose to be independent with other variable, if one is correlated with other variable, it shouldn’t included in the regression analysis, is it? I don’t understand this statement. thanks.

it’s just saying if you omit (leave out) a variable that is correlated with the other ones in the model, the estimates will be wrong. You are trying to include the relevant variables in the model, not leave them out.

Multicollinearity itself is not a problem in a model - in fact, there should be some - and it depends upon the degree of multicolliearity… Remember, the explanatory variables should have some correlation between each other or else it wouldn’t be very useful… But having TOO MUCH is the main problem. Think about explaining changes in GDP… Of course many factors in the economy have linkage between them, so it’s only when there is a high degree of multicollnearity that there is a problem and should consider dropping variables… Surprisingly the curriculum doesn’t use VIF for this as this what I learned in econometrics and forecasting courses… But oh well … Hope this helps