I'm struggling with the derivation of this known result. As a prerequisite, consider the following results. Suppose we have time-series data {$x_i$ for $i = 1, \dots, N$} from a stationary process having mean $\mu$ and variance $\sigma^2$ and theoretical acf $\rho(k)$. The sample mean is $\bar{X} = \sum_{i=1}^N \frac{X_i}{N}$. For correlated observations, it can be shown that the sample mean variance is not $\sigma^2/N$ but is given by:
$$ Var(\bar{X}) = \frac{\sigma^2}{N}\left[ 1 + 2 \sum_{k=1}^{N-1}\left( 1 - \frac{k}{N} \right) \rho(k) \right] $$
I'd like to find the variance of the sample mean for a stationary AR(1) process. The book states that $Var(\bar{X})$ for AR(1) reduces to $ Var(\bar{X}) = \frac{\sigma^2}{N}\left( \frac{1+\alpha}{1-\alpha} \right)$. To show this, I start with the autocorrelation function of an AR(1) process with parameter $\alpha$ which is $\rho(k) = \alpha^{|k|}$ where $\alpha \lt 1$. So the above equation becomes
$$ Var(\bar{X}) = \frac{\sigma^2}{N}\left[ 1 + 2 \sum_{k=1}^{N-1}\left( 1 - \frac{k}{N} \right) \alpha^k \right] $$
Then let $N \rightarrow \infty $. The complication then is ultimately in solving $\sum_{k=1}^\infty (1 - \frac{k}{N}) \alpha^{|k|}$. If I split it then:
$$ \sum_{k=1}^\infty \alpha^k = \sum_{k=0}^\infty \alpha^k - 1 = \frac{\alpha}{1-\alpha} $$
Now I need to solve the second part which is where I get stuck. I don't know how to evaluate:
$$ \sum_{k=1}^\infty k \alpha^k $$
Ultimately, I'd like to show that $ Var(\bar{X}) = \frac{\sigma^2}{N}\left( \frac{1+\alpha}{1-\alpha} \right)$ but I believe by solving the above sum which I'm stuck on I'll be able to show this result from the general equation for $Var(\bar{X})$ above.