Program to run the same equation on different observations

[nawal]

Hi,

I want to run the same equation on different observation of a data set. The objective of this study to check at which point of time there is change in the model parameters.
For example suppose if I have 100 observations. I want to run atleast 50 models on the same dataset and each time the last observation will be removed.
So, from this exercise I will check at what point of data my model parameter are changing.

Can somebody please suggest me how to program this exercise in Eviews. I would be very thankful to you.

[xprimexinverse]

Hi nawal,

If you're estimating the equations using OLS then there is a built-in feature that you could use instead of trying to write a program from scratch.

Here's how to do it. Open an equation object. Click "View". Under "Stability Diagnostics", you will see an option called "Recursive Estimates (OLS only)". If you select the "Recursive Coefficients" option you will be presented a graph from which you can assess the stability of the parameters in your model.

If this option doesn't suit your need then search the forum for "rolling regression". Programming it yourself will involve setting up a loop where you change the sample period on each iteration. See thread results from searching "rolling regression".

Graeme

[nawal]

Thanks for your reply..

I am able to write the code which take different sample and run different model. I have attached the code also for the reference.

But now, My problem is that I am not able to get the AIC, SBC, Constant Variance and Normality test in the output table.

I tried to use table_abc(!j,5)=@aic

But after that the value of AIC remain same for all the models.

Could anyone please help me on the same? I need this very urgently :(