This preview shows pages 1–2. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: 9/9/2011 1 Heteroskedasticity Copyright Dick Startz 1 Goals Meaning of heteroskedasticity What happens to OLS? GLS as an alternative Robust standard errors ARCH (if time) Copyright Dick Startz 2 Assumptions 1. ? ? = ? + 2 ?2 + + ? ?? + ? , true model of DGP . nonrandom (or exogenous), not perfectly collinear 3. ? = 0 4. Homoskedasticity ? = 2 , = 0, 5. (sometimes) ~? Copyright Dick Startz 3 Terminology General: Homoskedasticity ? = 2 , = 0, Heteroskedasticity ? 2 , = 0, Specific: Homoskedasticity ? 2 = 2 , Heteroskedasticity ? 2 = ? 2 Copyright Dick Startz 4 What facts about OLS are still true? Still best fit Still unbiased Which now fail, and is there a fix-up within OLS?...
View Full Document
- Fall '08