Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: run regression IF certain condition is given

From   "Braunfels, Philipp (Stud. SBE / Alumni)" <[email protected]>
To   "[email protected]" <[email protected]>
Subject   st: run regression IF certain condition is given
Date   Fri, 27 Jul 2012 10:58:49 +0200

Dear Statlisters,
I have the following question: I want to run a regression for a company over a number of rolling windows (window=24 months e.g. :01/2000-12/2002, 02/2000-01/2003, 03/2000-02/2003 ...), ONLY IF the y variables (=return) of each observation (company) are available for the FULL regression window. 
The reason why I need all data for the 24 month is, that I have an unbalanced panel and for some companies, especially in the beginning years, data are missing and thus the regression does not work.

My regression code for the 24-month window looks as follows (does not account for the fact that the data on x-variable have to be available for the respective 24-months, yet):

forvalues k = 1/1000 {            ///Defines ID of each company   
           matrix E = (0,0,0)                                     /// Matrix to safe regression coefficient estimates
	  forvalues i=528/627{                                            /// define time-window (my sample period)
		quietly reg return x1 if (id==`k' & month>=`i'-24 & month<`i'), r         ///regress return of comp "i" on x1 for a rolling window of 24 months
		matrix E = E\(`i',_b[msciwde], _se[msciwde])                                       ///save results to matrix

Now I would like to achieve the following:

For each ID check
IF return <. for the time period wheremonth>=`i'-24 & month<`i'
IF so, run the regression
if not, check (for the same ID), whether for the next time window all data on -return- are available (i.e. month>=(`i'+1)-24  & month<(`i'+1))
----do this for all time windows of an ID and then go to the next ID

I am very thankful for any help!!!

*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index