This preview shows page 1. Sign up to view the full content.
Unformatted text preview: t;
population parameters ( βo , β1 ); then I sort the estimates so obtained
and graph them into 2 histograms. Melissa Tartari (Yale) Econometrics 25 / 27 Part II: NonNormal Disturbances II ˆ
ˆ
Observe that, both the histograms for β1 and βo are centered about
the true value of the parameters (namely 1 and 3): this is as it should
since we know that unbiasedness of the OLS estimator holds under
LR.1 through LR.4 without any role played by distributional
assumptions on the disturbance u . Melissa Tartari (Yale) Econometrics 26 / 27 Part II: NonNormal Disturbances II ˆ
ˆ
Observe that, both the histograms for β1 and βo are centered about
the true value of the parameters (namely 1 and 3): this is as it should
since we know that unbiasedness of the OLS estimator holds under
LR.1 through LR.4 without any role played by distributional
assumptions on the disturbance u .
However, we also see that the two histograms resemble a Gaussian
distribution: this is the implication of Theorem 5.2 on asymptotical
normality of the OLS estimator (since the histogram estimates the
distribution of the estimator). Melissa Tartari (Yale) Econometrics 26 / 27 Part II: NonNormal Disturbances II ˆ
ˆ
Observe that, both the histograms for β1 and βo are centered about
the true value of the parameters (namely 1 and 3): this is as it should
since we know that unbiasedness of the OLS estimator holds under
LR.1 through LR.4 without any role played by distributional
assumptions on the disturbance u .
However, we also see that the two histograms resemble a Gaussian
distribution: this is the implication of Theorem 5.2 on asymptotical
normality of the OLS estimator (since the histogram estimates the
distribution of the estimator).
Notice that by assumption/construction the disturbances are iid with
…nite variance so the conditions for theorem 5.2 to apply are met. Melissa Tartari (Yale) Econometrics 26 / 27 Asymptotic E¢ ciency of OLS ˆ
We know that under LR.1 through LR.5 βj is BLUE. Melissa Tartari (Yale) Econometrics 27 / 27 Asymptotic E¢ ciency of OLS ˆ
We know that under LR.1 through LR.5 βj is BLUE.
ˆ
Under the same ass.s βj is also asymptotically e¢ cient among a
certain class of estimators. Melissa Tartari (Yale) Econometrics 27 / 27...
View Full
Document
 Fall '10
 DonaldBrown
 Econometrics

Click to edit the document details