Page 2 of 2

Re: Errors instruction

Posted: Mon Apr 15, 2019 11:45 am
by ateeb
Dear Tom,

When we use Errors instructions to compute the VDCs we get a standard error in output, is it the standard error for the residuals from the VAR or for point estimates of the VDCs?

Regards,

Re: Calling CATs

Posted: Mon Apr 15, 2019 12:16 pm
by TomDoan
Standard error of forecast (computed at the point estimates).

https://estima.com/docs/RATS%2010%20Use ... f#page=250

Re: Calling CATs

Posted: Fri May 24, 2019 9:27 am
by ateeb
Dear Tom,

I am getting the following error:

## EQ4. Equation %M(AICLAG,1) Has At Least One Undefined or NA Coefficient
## EQ4. Equation %M(AICLAG,2) Has At Least One Undefined or NA Coefficient
## EQ4. Equation %M(AICLAG,3) Has At Least One Undefined or NA Coefficient
## EQ4. Equation %M(AICLAG,4) Has At Least One Undefined or NA Coefficient

when i am trying to modify equations 2 to 5 of the system previously estimated. Can you please help me explain what is the problem, it does work properly for the first equation.

Regards,
Ateeb

Re: Calling CATs

Posted: Tue May 28, 2019 2:29 pm
by TomDoan
Sorry. "Trying to modify" how?

Re: Calling CATs

Posted: Tue May 28, 2019 3:05 pm
by ateeb
So this is the situation:

I have x1, x2, x3, x4 and x5. x4 and x5 are stationary.

x1, x2 and x3 are cointegrated of order 1. so i use Engle-Granger 2 step estimation.

I form e11 as my equilibrium error.

I run system with all stationary variables with x1, x2 and x3 as stationary. Then i modify the system by replacing x4 and x5 by themselves as they are stationary in levels and do no differences, but x1, x2 and x3 in levels with differencing. so it is a vecm with everything specified properly. However, it works for equation 1 and as soon as i do that for second one it does not.

Furthermore, the system has 6 lags as picked up by AIC but i can see x1 comes with 7 lags as a result of first modification (i mean coefficients) and same is the case with x2 and x3. whereas x4 and x5 comes with 6 lags.

So I dont know whats going on. furthermore, i could not find %M in manual, user guide or reference manual. What does it show?

Re: Calling CATs

Posted: Tue May 28, 2019 5:10 pm
by TomDoan
The LAGS option in CATS (and in @JOHMLE and pretty much all of the cointegration procedures) are the number of augmenting lags which is one less than the number of lags in the expanded model.

Re: Calling CATs

Posted: Thu May 30, 2019 5:29 pm
by ateeb
Thank you I got the idea!