2. (10%) A linear stationary time series model relating a variable, $yt$, to the rational expectation of the, $xt$, is given by

$$yt = \beta E(xt | I{t-1}) + ut,$$

where $E(xt | I{t-1})$ is the expected value of $xt$ given all information known up to time t-1. Note that, at a minimum, $(y{t-1}, x{t-1}, y{t-2}, x{t-2}, \ldots)$ is contained in $I{t-1}$. Throughout this problem assume that $E(ut | I{t-1}) = 0$. For simplicity, there is no intercept in the model. Let $\hat{\beta}$ be the OLS estimator from the regression $yt$ on $xt$, $t = 1, 2, \ldots, T$. Whether $\hat{\beta}$ is a consistent estimator of $\beta$?__(9)__