Rolling regression estimated coefficients

[Abi Den]

Hi all,

I have an unbalanced panel data t = 2000-2014 , I want to plot the estimated coefficient using rolling regression. Can someone tell me where the process or command example for panel data. It is my first time using stata I am just following the forum and help rolling stata, I get empty plots using rolling regression.( where are the estimated coef stored?)
Thank you in advance for your help
I saw this
http://www.statalist.org/forums/foru...ssion-in-stata
not clear to me .

Best,

Abi.

[Nick Cox]

Saying that something is "not clear" is no doubt sincere but it is not a question. What don't you understand?

Two of the three examples at help rolling concern regression. They contain code you can run to see how it works.

The example in the link in #1 contains code you can try to see how it works.

What syntax did you try? As you don't show your data or give any code, we have no way to explain your empty plots.

[Abi Den]

Thank you.
I did run the examples. They work fine. When I try my code there is no estimated coefficients . My question is where are my estimated coefficients stored?
Thank you again.
Abi

[Nick Cox]

Sorry, no idea. You didn't answer #2 about your code or your data, so I don't know what else to add.

[Marcos Almeida]

The information I share below may not be fully helpful, because I believe the "empty plots" related to the window you selected has nothing to do with the way Stata handles coefficients.

Under rolling, the larger the window and the fewer the time periods, the less "appropriate" would be the data so as to perform a rolling regression. What is more, the initial period won't appear in the graph, since it is mainly to "start up" the rolling mechanism.

Anyway, just to clarify the "whereabouts" of the coefficients, you may wish to read this excerpt from the Stata Manual:

Stored results
rolling sets no r- or e-class macros. The results from the command used with rolling, depending on the last window of data used, are available after rolling has finished

[Abi Den]

Thank you for your response. I used smaller window it works. Now I can see the estimated coeff. but I have a lot of zeros in the results I am thinking that it is because I have unbalanced data. Any suggestion how I can adjust that? May be to get rid of the zeros so that I can plot the results. Obs # = 2800 and T = 14 Thank you again

[Marcos Almeida]

The odds of getting helpful advice will increase, provided you act according to the advice shared in #2.