Extraction of R squared from a series of regressions

[Jakub Cirkowski]

Hi,

I was wondering if it possible to extract given value from a series of regressions.

I managed to generate 56 regressions (1962-2017) usind Stata 15 and now I need the R squared from every regression.

-> pyear = 2015

      Source |       SS           df       MS      Number of obs   =     3,935
-------------+----------------------------------   F(1, 3933)      =  13057.92
       Model |  21756541.1         1  21756541.1   Prob > F        =    0.0000
    Residual |  6552993.76     3,933  1666.15656   R-squared       =    0.7685
-------------+----------------------------------   Adj R-squared   =    0.7685
       Total |  28309534.9     3,934   7196.1197   Root MSE        =    40.819

------------------------------------------------------------------------------
         prc |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
        ibpr |   11.97655   .1048081   114.27   0.000     11.77107    12.18204
       _cons |   22.24909   .6580438    33.81   0.000     20.95895    23.53923
------------------------------------------------------------------------------

---------------------------------------------------------------------------------------------------------------------------------------------
-> pyear = 2016

      Source |       SS           df       MS      Number of obs   =     3,909
-------------+----------------------------------   F(1, 3907)      =   5940.46
       Model |  12809123.3         1  12809123.3   Prob > F        =    0.0000
    Residual |  8424470.45     3,907  2156.25044   R-squared       =    0.6032
-------------+----------------------------------   Adj R-squared   =    0.6031
       Total |  21233593.7     3,908  5433.36584   Root MSE        =    46.435

------------------------------------------------------------------------------
         prc |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
        ibpr |   14.08432   .1827367    77.07   0.000     13.72605    14.44259
       _cons |   22.48249   .7494967    30.00   0.000     21.01304    23.95193
------------------------------------------------------------------------------

---------------------------------------------------------------------------------------------------------------------------------------------
-> pyear = 2017

      Source |       SS           df       MS      Number of obs   =     3,809
-------------+----------------------------------   F(1, 3807)      =   7187.95
       Model |  22908960.9         1  22908960.9   Prob > F        =    0.0000
    Residual |  12133417.1     3,807  3187.13346   R-squared       =    0.6538
-------------+----------------------------------   Adj R-squared   =    0.6537
       Total |    35042378     3,808  9202.30515   Root MSE        =    56.455

------------------------------------------------------------------------------
         prc |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
        ibpr |   12.99276   .1532495    84.78   0.000      12.6923    13.29322
       _cons |   25.81245   .9221506    27.99   0.000     24.00449     27.6204
------------------------------------------------------------------------------

I found a code that link this coefficient to a company, but not for years.

I work on panel data and I have multiple observations for every year.


Any help or comments much appreciated. Sorry If I missed any important information.

Thanks!

[Nick Cox]

In your concurrent thread (https://www.statalist.org/forums/for...1-broken-loops) in #3 there I referred you to rangestat (SSC). Here's a dopey example of cross-section regressions for each year of a panel dataset.

. webuse grunfeld, clear

. rangestat (reg) invest mvalue, int(year 0 0)

. tabdisp year, c(reg_r2) format(%4.3f)

--------------------------------
     year |               reg_r2
----------+---------------------
     1935 |                0.865
     1936 |                0.696
     1937 |                0.654
     1938 |                0.635
     1939 |                0.787
     1940 |                0.820
     1941 |                0.844
     1942 |                0.855
     1943 |                0.835
     1944 |                0.872
     1945 |                0.905
     1946 |                0.895
     1947 |                0.891
     1948 |                0.787
     1949 |                0.862
     1950 |                0.857
     1951 |                0.864
     1952 |                0.843
     1953 |                0.886
     1954 |                0.856
--------------------------------

[Jakub Cirkowski]

Thank you very much for help I will try to use it as the result looks perfect to me.