Creating a loop for time regression

[Jakub Cirkowski]

Hello,

I am investigating trends in R2 over time by regressing it over time. Because I need to do it for 16 variables and for 12 industries I was trying to improve the following code so that it generates 16 regression outputs for each variable. I tried to do it by creating a varlist and using loop foreach but I got lost.

I was wondering if somebody knows how to do it quickly and easily?

gen R2ibpr =.
forvalues i = 1962(1)2017 {
display "`i'"
reg prc ibpr if  pyear==`i'
replace R2ibpr = e(r2_a) if pyear == `i'
}
keep pyear R2ibpr

duplicates drop
sort pyear

gen time = pyear-1961
reg R2ibpr time

      Source |       SS           df       MS      Number of obs   =        56
-------------+----------------------------------   F(1, 54)        =      1.80
       Model |  .053632868         1  .053632868   Prob > F        =    0.1859
    Residual |  1.61346458        54  .029878974   R-squared       =    0.0322
-------------+----------------------------------   Adj R-squared   =    0.0142
       Total |  1.66709745        55  .030310863   Root MSE        =    .17286

------------------------------------------------------------------------------
      R2ibpr |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
        time |   .0019147   .0014291     1.34   0.186    -.0009505    .0047798
       _cons |   .3718075   .0468233     7.94   0.000     .2779325    .4656825
------------------------------------------------------------------------------

Thank you for taking your time to help me,

Jakub

[Raffaele Grotti]

Hi,
you need to nest several loops for: year, variable and industry.

local variables "va1 var2 ... var16"
forvalues i = 1962(1)2017 {
    foreach var of local variables {
        if `i'==1962 gen R2`var' = .
        levelsof industry if pyear==`i' & !mi(prc, `var'), local(industries)
        foreach ind of local industries {
            reg prc `var' if pyear==`i' & industry==`ind'
            replace R2`var' = e(r2_a) if pyear == `i' & industry==`ind'
        }
    }
}

I haven't test the code, hope it works and does what you need.

Raffaele