Heteroskedasticity: Studenmund Chapter 10
Violations of Homoskedasticity Assumption
The assumption of constant variances for different observations of the error term (homoskedasticity) is not always realistic. For example in a model explaining heights, compare a one-inch error in measuring the height of a basketball player with a one-inch error in measuring the height of a mouse. It’s likely that error term observations associated with the height of a basketball player would come from distributions with larger variances than those associated with the height of a mouse…. OLS, when applied to heteroskedastic models is no longer the minimum variance estimator (it is still unbiased, however). From Studenmund ‘ s Using Econometrics Ch.10
So, the assumption of homoskedasticity says that the variance (or standard deviation, or common) of the error term is exactly the same for each observation. Often we have heteroskedasticity, where the variance is a function of the explanatory variables: That is: