1
$\begingroup$

How do you show that the white noise process is strictly stationary?

Let's consider the i.i.d. white noise process $a_t$: \begin{align} E[a_t] &= 0\\ Var[a_t] &= \sigma_a^2 \end{align}

The weak stationarity is obvious. But I don't know how to show the strict stationarity. Basically I would like to show that for all $t_1, \dots, t_n, k \in \mathbb{Z}$, the joint distribution of $a_{t_1},\dots, a_{t_n}$ is the same as $a_{t_1 -k},\dots, a_{t_n-k}$.

$\endgroup$
1
  • $\begingroup$ Hi: the joint distribution of the $n$ $X_{i}$ random variables at time $t_{n}$, ( the $n$ element vector ) is multivariate normal with mean zero and covariance matrix with diagonal elements $\sigma_{a}^2$ and zero everywhere else. So, you just need to show that the same thing is true for the $n$ random variables at time $t_{k+n}$ (i.e: the $n$ vector $X_{t+k+1}, X_{t+k+2}, \ldots, X_{t+k+n}$. ). I hope that helps. $\endgroup$
    – mlofton
    Apr 16, 2019 at 2:20

2 Answers 2

2
$\begingroup$

That cryptic initialism i.i.d. is the key; it stands for independent and identically distributed which says to the cognoscenti that

  • Every random variable in the process/time series has the same distribution, and so for any given distinct integers $t_1, t_2, \ldots, t_n$ and any $k \in \mathbb Z$, the $2n$ random variables $A_{t_1}, A_{t_2}, \ldots, A_{t_n}, A_{t_1-k}, A_{t_2-k}, \ldots, A_{t_n-k}$ all have the same distribution.
  • The random variables $A_{t_1}, A_{t_2}, \ldots, A_{t_n}$ are independent, and so their joint distribution is just the product of their $n$ identical marginal distributions. Similarly, $A_{t_1-k}, A_{t_2-k}, \ldots, A_{t_n-k}$ are also independent and since the have the same marginal distributions as the $A_{t_1}, A_{t_2}, \ldots, A_{t_n}$, the joint distribution of $A_{t_1-k}, A_{t_2-k}, \ldots, A_{t_n-k}$, which is the product of their marginal distributions, is the same as the joint distribution of $A_{t_1}, A_{t_2}, \ldots, A_{t_n}$.
$\endgroup$
1
$\begingroup$

IID is a stronger condition than strict stationarity, and it implies the latter. If you have a sequence of IID values then the sequence is exchangeable, which implies strict stationarity.

$\endgroup$

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.