Consider the following stochastic process:

$$Y_i=c\cdot Y_{i-1}+\varepsilon_i,$$ where $$Y_0\sim \mathcal N(0,\sigma^2)$$ is independent of the white noise $$\varepsilon_i \overset{\text{iid}}\sim \mathcal{N}(0,\sigma^2).$$

(Here iid means independent and identical and $\mathcal N$ denotes the Normal distribution. Also, $\sigma>0$ is a parameter.)

What can be said about the conditional expectation $$\mathbb E(Y_{i+1}\mid Y_1,Y_2,Y_3,\dots,Y_i)$$? What about the conditional Variance $$\mathbb V(Y_{i+1}\mid Y_1,Y_2,Y_3,\dots,Y_i)$$?

My attempt: We can solve the recurrence: For any $n\geq 1$, we have

$$Y_{i+n}=c^n Y_i + E,$$

where $$E=\sum_{i=1}^n c^{n-i} \varepsilon_i \sim\mathcal N\left(0,\sigma^2\frac{1-c^{2n}}{1-c^2}\right).$$

Now I'm stuck.


Since the errors are iid, the process can be considered also as a Markov process (see Is AR(1) a Markov process?).

It follows that $Y_{i+1}$ conditional on $Y_1, Y_2, \ldots, Y_i$ is the same in distribution as $Y_{i+1}$ conditional on only $Y_i$. Therefore, conditional expectation and variance can be found like so: $$\mathbb E \left( Y_{i+1} | Y_1, Y_2, \ldots, Y_i \right) = \mathbb E \left( Y_{i+1} | Y_i \right) = \mathbb E \left(c Y_i + \varepsilon_{i+1} | Y_i \right) = c Y_i$$

$$\mathbb V \left( Y_{i+1} | Y_1, Y_2, \ldots, Y_i \right) = \mathbb V \left( Y_{i+1} | Y_i \right) = \mathbb V \left( c Y_i + \varepsilon_{i+1} | Y_i \right) = \sigma^2.$$

Regarding your attempt, you're on the road to determine the conditional expectation and variance of $Y_{i+n}$ conditional on $Y_i$, which doesn't seem to be what you are looking for based on your initial two questions.

| cite | improve this answer | |

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.