Let's consider we have OLS model $Y=X \beta+\epsilon$ , where $E(\epsilon) = 0 $ and $Var(\epsilon)=\sigma^2 I$.

Here $\hat{\beta}$ is the least squares estimator of the parameter $\beta$ . Let $z \neq 0$ be $n$ dimensional vector. The goal is to find the best linear estimator $A$ for $z^T \epsilon$ , where $E(A)=0$ and MSE (mean squared error) is the smallest possible. Actually we have to show that $A=z^T$ ($Y-X\hat{\beta}$) is the best estimator with this properties.

It is clear to me that $E(A)=0$, but since this is not unbiased estimator, I can't use Gauss- Markof theorem and I don't know what I should use to show that this is the best linear estimator (actually that it has the smallest MSE).

Thanks for the help.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.