Easy way to run multiple regressions at the same time

[eaghdasi]

I'm using Eviews 7.1 with the web interface. Is there a quick way to run a handful of least-squares regressions in a single step? Or do I really have to open up a new object for every regression I want to run? I'm sure there's an answer already posted somewhere for this but I couldn't find anything. Thanks very much to anyone who can help.

[EViews Gareth]

Could you define what you mean by "using EViews 7.1 with the web interface"?

[eaghdasi]

Sorry, I meant Eviews version 7.1, which has a WINDOWS (not web) interface. (I'm new to Eviews and not sure what type of interface previous versions had.)

[EViews Gareth]

ok. Well, I'm still not 100% certain I understand what you're asking...

But if you want to estimate a handful of new regressions (as opposed to re-estimating existing ones), the quickest way to do it would be in a program. Although, of course, that would require learning the EViews programming language.

A second method would be to use the command window and type the regressions one at a time, which is still quicker than using the mouse. The command for a least-squares regression is something like:

Code: Select all

ls y c x1 x2

where Y is the dependent variable, C is the constant and X1, X2 are independent variables.

[damien]

Hi,

I think I understand Eaghdasi's problem as I have the same issue. I'm using Eviews 7 and I am analysing 40 portfolios using Fama-French. I think the question is: how can I run the 40 simple linear regression without typing 40 times the equation for the 40 portfolios? Is there any way to program that? Is this the so called rolling regression???

Thanks a lot in advance for your answers!

Cheers

[EViews Gareth]

Yes, you can program it. Check out the programming guide for a good start:
http://forums.eviews.com/viewtopic.php?f=5&t=1638

[alexma]

I am having the same problem, and i have came up with a possible solution, simple but seemed to work out just fine.

What if I open all portfolios I am analysing and make estimates using the respective independent variables as "equation specifics coeficients", as opposed to using fixing the same coefficients for the same equations?

I dont know if there is any computacional limitations, but for the number of portfolios I am using seemed to work just fine.

Is there any mistake in this approach?

Appreciate any comments!

Rgds from Brazil.

[EViews Gareth]

I presume you're talking about in a Pool object.

That works fine, as long as all you are interested in is the portfolio specific coefficients and standard errors. If you want to know more things about each individual estimation, such as the individual R-squareds, you're out of luck in a pool.

[SA1]

Hi alexma,
Could you please elaborate further on how to run this simultaneously on a pool? I'm a novice here, and would appreciate if you could give me some specifics!! Thanks!