Regressions with a fixed time period

[Edward Wu]

Hi!

I have monthly panel data since 2008. Only 4 variables: year and month, company_id, y, x

From January 2011, I would like to run panel regression in the past 36 months period and record the estimated coefficient as beta of that month for that company. I need to do so every month and record the betas for all the companies, until December 2018.

Anyone can help?

Great thanks!

[Edward Wu]

Up ：））

[Phil Bromiley]

rangestat or statsby probably

[Edward Wu]

Thanks! Could you provide the code? I need to generate a new variable beta to record all the estimated coefficients of each company in every 36 months regression period )

[Edward Wu]

Up ：））

[Nick Cox]

Please read https://www.statalist.org/forums/help#adviceextras #1 on bumping. The short advice on bumping is: please don't, and the longer advice is constructive.

The implication of -- or rather your first guess on -- getting no reply is that your question was not clear or detailed enough to get an answer out of interested people.

In this case, your second guess is, or should be, that you are irritating the heck out of people who might be interested but don't respond well to impatience or any hint that you deserve an answer when you are not obviously trying very hard. .

You're not new to the forum. After nearly 50 posts you should have a sense of how it works.

#1 is not obscure, but it shows minimal effort. Many of us who answer questions are, or have been, teachers, and regardless those who answer questions are teachers in that sense and respond best to effort and enthusiasm. No one is paying us anything for any of this!

Phil Bromiley gave a really good hint in #3, so did you search the forum following that hint? There are hundreds of posts on this topic using rangestat (SSC), statsby and indeed other commands too.

if you gave a detailed data example, following longstanding advice at https://www.statalist.org/forums/help#stata , you would make it easier for someone to give you detailed advice. Otherwise, what you get out of Statalist depends on what you put in.