Question

a) Briefly explain the problem of heteroscedasticity in the context of OLS estimation. Its description, how it affects OLS output and inference and how it might arise should be in your answer. (10 marks) b) Describe in detail how the White test for heteroscedasticity is performed. (10 marks) c) Briefly explain the issue with using White-corrected variance estimators when they ee,e whe f orginl iceda is Mark

Answer 1

a) In an econometric model, in order to apply OLS method for estimating the parameters we need to make some assumptions. One of the assumption is that the variance of the error or disturbance term must be constant, i.e., Var(u)= σ². This is known as homoskedastic assumption. When this assumption is not satisfied then the model is said to have Heteroskedasticity problem. It causes the estimated parameters to be inefficient and arises in cross section studies.

b) The White's Test for Heteroskedasticity first computes the Chi Square Statistic from the errors which are obtained from the estimation procedure. Then the Chi Square statistic is multiplied by the number of observations, n. Then if this value exceeds the critical value there is said to be considerable heteroskedasticity.

c) White's Test for heteroskedasticity corrects for heteroskedasticity when it is in fact an issue. However, in many cases if we adjust the variables using White's measure then we might not end up with desired results say because the original variables with which we were interested are now redefined as something else. Then the test is counterproductive.