Covariance Stationary

stanziolap · September 28, 2017, 10:42pm

Hi

Need help understanding this:

in the book says that an AR(1) needs b₁ to be less than one in order to be covariance stationary.

I know that b₁ can not be equal to one but why it cannot be greater than one and still be cov. Stat.

Thanks for ur help

S2000magician · September 29, 2017, 3:22am

Create a model in Excel with b₁ > 1 and see what happens.

Arthur_Yao · October 7, 2017, 4:49am

In Reading 11 4.1 you can find three requirements for a series to be covariance stationary. The first is:

First, the expected value of the time series must be constant and finite in all periods: E(y_t) = μ and |μ| < ∞, t = 1, 2, …, T.

If b₁>1, expect value will NOT be constant. It will grow larger and larger as time goes on.