Building Better Econometric Models Using Cross Section and Panel Data

Chapter 3

Misspecifications for the More Advanced Reader

As implied earlier, the next three topics are not for the faint of heart but often rear their ugly heads in most regression-based studies. The problem is that most researchers either don’t know these issues could exist, or if they believe these issues can exist they don’t know how to check for them outside of time-series analyses, and they certainly never correct for these problems if they believe they can exist and know how to check for them.

Variance Heterogeneity

Variance heterogeneity is a strange beast. It is perhaps the most widely misinterpreted and misrepresented misspecification issue in the history of econometrics. Here’s why. From earlier we know that heteroskedasticity is represented ...