The properties of the OLS estimator
Summary
Setup
The linear regression model with random sampling,
yi=β1+β2xi,2+β3xi,3+…+βkxi,k+εi,i=1,…,n
b1,b2,…,bk are the OLS estimators of β1,β2,…,βk .
Gauss-Markov assumptions
The following set of assumptions qualify as Gauss-Markov assumptions:
- All the x -variables are exogenous
E(εi|xi)=0,i=1,…,n
- The residuals are homoscedastic with a common variance σ2
Var(εi|xi)=σ2,i=1,…,n
OLS unbiased
- The OLS estimators b1,b2,…,bk are unbiased if all the x -variables are exogenous
OLS consistent
- The OLS estimators b1,b2,…,bk are consistent under mild conditions if all the x -variables are exogenous.
Under GM assumptions,
- It is possible to derive the variance of the of the OLS estimators (the OLS GM variance formulas)
Var(bj|x),j=1,…,k
- This variance will depend on all the x -data as well as σ2 but the formulas are messy when k>2 (without using matrices).
- We define
s2=RSSn−k=1n−kn∑i=1e2i
- Under GM, s2 (the OLS estimator of σ2 ) is an unbiased and consistent (under mild conditions) estimator of σ2 .
- s , the square root of s2 , is called standard error of the regression.
- If we replace σ2 with s2 in the OLS GM variance formulas, we get the estimated variance of b1,b2,…,bk :
^Var(bj),j=1,…,k
- The OLS standard errors
- SE(bj)=√^Var(bj),j=1,…,k
- are presented in all econometric packages.
- The OLS estimators will be BLUE (Best Linear Unbiased Estimator).