1. Consider a simple regression model with no intercept:
$$yt = \beta xt + u_t.$$
You have two observations (T=2). Suppose $x1$ and $x2$ are fixed; $E(u1) = E(u2) = 0$; $\text{var}(u1) = 1$, $\text{var}(u2) = 2$, $\text{cov}(u1 u2) = 0$.
1-1 (10%) Now consider the estimator $\hat{\beta} = (x1 y1 + \frac{1}{2} x2 y2) / (x1^2 + \frac{1}{2} x2^2)$.
Whether this estimator is unbiased. (6) Calculate its variance. (7)