Given that $a_t \sim WN(2, 0.5)$, I have generated the process defined by

$$Y_1 = a_1$$

$$Y_t = \theta Y_{t - 1} + a_t$$

to be: $$Y_t = \theta^{t - 1}a_1 + \theta^{t - 2}a_2 + \cdots + \theta a_{t - 1} + a_t$$

I am then asked to determine the range of the parameter $\theta$ for which the representation of $Y_t$ in terms of $a_t$ exists.

Is the idea to write $Y_t$ as: $$\sum\limits_{i = 0}^{t - 1} \theta^ia_{t - i}$$

and treat as a finite Geometric Series?

I have also found the mean of this series to be: $$E[Y_t] = 2\sum\limits_{i = 1}^t \theta^{t - i}$$

But in finding the variance, I am confused on how to find $E[Y_t^2]$ and computing $(E[Y_t])^2$ seems difficult given the form I have.

Any help would be greatly appreciated.

  • $\begingroup$ Is there any information regarding $Y_t$ is a stationary process or not? $\endgroup$ Commented Feb 19, 2021 at 8:51

1 Answer 1


I'm fairly new to stack exchange so I am not able to comment... hopefully someone will edit this for clarity.

To solve for the variance of Y(t) you need to represent it as a geometric series, then use the property of variance, namely Var(aX) = a^2 Var(X)


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.