1
$\begingroup$

So I have this AR data series, AR1, mostly (for the argument). I also have an exogeneous regressor, only the DGP I suspect is something in the line of: $ y_t=y_{t-1}+\alpha x_ty_{t-1}+\varepsilon_t $

Can an ARIMAX estimate this? What's the specification? If not, what?

$\endgroup$
2
  • $\begingroup$ Your equation does not have an error term, so $y_t$ is a deterministic function of $y_{t-1}$ and $x_t$. Is that right? Regardless of that, this is not within the class of ARIMAX. $\endgroup$ May 10, 2022 at 5:58
  • $\begingroup$ Meant the stochastic version. Will use AR(1). Thank you Richard $\endgroup$
    – BlackNinja
    May 11, 2022 at 7:39

1 Answer 1

1
$\begingroup$

A simple ARIMAX(1,0,0) should be of the form $y_t = α_1y_{t-1} + α_2x_t + e_t$. From your question it seems that is what you wanted to write, isn't it?

As @Richard pointed out, the equation you wrote is not an ARIMAX, but solving it would give $log(y_t/y_{t-1}) ≈ αx_t$, which a different but well known type of model. Please, clarify.

$\endgroup$
1
  • $\begingroup$ Yeah, I mean the stochastic version. Thank you very much :) $\endgroup$
    – BlackNinja
    May 11, 2022 at 7:38

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.