Consider a time series generated by the following model:

$y_t = 0.8y_{t-1} + \epsilon_t$

$E(\epsilon_t) = 0$ and $Var(\epsilon_t) = 1$. Note that for any s < t, $y_s$ is independent with $\epsilon_t$. The above time series is a strictly stationary sequence and each element is normally distributed. Hence, $E(y_t)$ and $Var(y_t)$ are both constants. In particular, $E(y_t) = E(y_{t−1})$ and $Var(y_t) = Var(y_{t−1})$. Denote $E(y_t)$ by µ and $Var(y_t)$ by $σ^2$. Find the values of µ and $σ^2$.

Here is what I have so far:

$E(y_t) = 0.8E(y_{t-1}) + E(\epsilon_t) = 0.8y_{t-1}$

$Var(y_t) = 0.64Var(y_{t-1}) + Var(\epsilon_t) = 0.64σ^2_{y_{t-1}} + 1$

I do not know how to find the specific values. I thought at first the question was asking for formula answers but the specific values are required for future questions. Also, how can $E(y_t) = E(y_{t-1})$ when $E(y_t) = 0.8E(y_{t-1})$?


1 Answer 1



\begin{align} E(y_t) &= E(y_{t-1}) = \mu\\ Var(y_t) &= Var(y_{t-1}) = \sigma^2 \end{align}

Can you make the substitutions and solve for $\mu$ and $\sigma^2$?

  • $\begingroup$ Does $μ = 0$ and $σ^2 = 0.0278$? $\endgroup$ Oct 18, 2018 at 0:43
  • $\begingroup$ @EthanSmith Your $\mu$ is correct. Did you misplace a decimal point? $\sigma^2 = 2.78$. $\endgroup$
    – Almacomet
    Oct 18, 2018 at 2:22

Your Answer

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

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