proc ARIMA

[Golf]

Hi everyone,

I'm looking for help with the correct code to run the following model using "proc ARIMA":

dy = a0 + a1*trend + a2*coint(-1) + b1*dy(-1) + b2*dy(-2) +....+b5*dy(-5) + c1*dx(-1) + c2*dx(-2) + .... +c5*dx(-5) +error

In this model, dy is the first difference of y, dy(-1) represents the 1 period lag of dy, and dy(-5) represents the 5 period lag of dy. I generated the dy and dx variables from dif(y) and dif(x), respectively.

Additionally, I'm wondering if the backward selection can be applied to the lag of dy(-1) through dy(-5), dx(-1) through dx(-5).

Thank you for your help.

[sbxkoenk]

I moved this post to a more appropriate board:
SAS Support Communities > Analytics and Statistics > SAS Forecasting and Econometrics

What is coint(-1)?
An element from a cointegrating vector (cointegration)?

BR, Koen

[Golf]

@sbxkoenk Yes It is.

[sbxkoenk]

Hello,

I would first test for cointegration using PROC VARMAX.

VARMAX = Vector AutoRegression Moving-Average with EXogenous Regressors

• SAS/ETS 15.3 User's Guide
  The VARMAX Procedure
  MODEL Statement
  Cointegration Related Options
  https://go.documentation.sas.com/doc/en/pgmsascdc/9.4_3.5/etsug/etsug_varmax_syntax12.htm#etsug_varm...
• SAS/ETS 15.3 User's Guide
  The VARMAX Procedure
  Cointegration
  https://go.documentation.sas.com/doc/en/pgmsascdc/9.4_3.5/etsug/etsug_varmax_details46.htm

Examples of usage:

model y1 y2 / p=2 cointtest=(johansen=(iorder=2 normalize=y1));
model y1 y2 / p=2 cointtest=(sw siglevel=0.1);
model y1 y2 / p=2 cointtest=(sw=(type=kernel lag=3));

BR, Koen

[Golf]

@sbxkoenk Thank You.