slides_Ch2_W[1]

ng 3 slr2 e u jx 0 2 slr1 slr3 random

This preview shows page 1. Sign up to view the full content.

This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: f β, ˆ namely on E β . Melissa Tartari (Yale) Econometrics 49 / 93 Unbiasedness of the OLS Estimator I - the Roadmap ˆ We start by focusing on the …rst moment of the distribution of β, ˆ namely on E β . ˆ A natural question to ask is whether we have E β = β; more precisely, under what assumptions is the OLS estimator unbiased. Melissa Tartari (Yale) Econometrics 49 / 93 Unbiasedness of the OLS Estimator I - the Roadmap ˆ We start by focusing on the …rst moment of the distribution of β, ˆ namely on E β . ˆ A natural question to ask is whether we have E β = β; more precisely, under what assumptions is the OLS estimator unbiased. We next state four assumptions and then verify that they are su¢ cient for unbiasedness of OLS. Melissa Tartari (Yale) Econometrics 49 / 93 Unbiasedness of the OLS Estimator II - Assumptions 1 bivariate linear model y = βo + β1 x + u Melissa Tartari (Yale) Econometrics (SLR.1) 50 / 93 Unbiasedness of the OLS Estimator II - Assumptions 1 bivariate linear model y = βo + β1 x + u 2 (SLR.1) f(yi , xi ) ji = 1, ..., ng (SLR.2) random sampling Melissa Tartari (Yale) Econometrics 50 / 93 Unbiasedness of the OLS Estimator II - Assumptions 1 bivariate linear model y = βo + β1 x + u f(yi , xi ) ji = 1, ..., ng 2 3 (SLR.1) (SLR.2) E [u jx ] = 0 (SLR.3) random sampling zero conditional mean Notice that 2. and 3. imply that nothing is lost (in derivations) in treating xi as non-random or …xed in repeated samples. Melissa Tartari (Yale) Econometrics 50 / 93 Unbiasedness of the OLS Estimator II - Assumptions 1 bivariate linear model y = βo + β1 x + u f(yi , xi ) ji = 1, ..., ng 3 (SLR.2) E [u jx ] = 0 2 (SLR.1) (SLR.3) random sampling zero conditional mean Notice that 2. and 3. imply that nothing is lost (in derivations) in treating xi as non-random or …xed in repeated samples. 4 sample variation in the eVar xi for i = 1, ..., n are not all equal Melissa Tartari (Yale) Econometrics (SLR.4) 50 / 93 Unbiasedness of the OLS Estimator III Recall ¯ ¯ ¯ ∑n= yi (xi x ) ∑n (yi y ) (...
View Full Document

Ask a homework question - tutors are online