Panel data regression results

[Adam Klaas]

Hello everyone,

I was wondering if my model is correct the way I have set it up?
I do have 2 years omitted, but as others have said that its a common issue when working with panel data.
If anyone have suggestions how I can improve my regression feel free to put your input.


I get these as my regression results

. xtreg log_realHP logReal_income real_interest Unemployment_rate logrealconsind i.high_dev i.Year,fe cluster(GM_code) noom
> itted
note: 2018.Year omitted because of collinearity
note: 2019.Year omitted because of collinearity

Fixed-effects (within) regression               Number of obs     =      2,250
Group variable: GM_code                         Number of groups  =        282

R-sq:                                           Obs per group:
     within  = 0.8397                                         min =          7
     between = 0.6998                                         avg =        8.0
     overall = 0.3075                                         max =          8

                                                F(10,281)         =    1116.74
corr(u_i, Xb)  = 0.3154                         Prob > F          =     0.0000

                                   (Std. Err. adjusted for 282 clusters in GM_code)
-----------------------------------------------------------------------------------
                  |               Robust
       log_realHP |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
------------------+----------------------------------------------------------------
   logReal_income |   .2636083   .0951621     2.77   0.006     .0762873    .4509293
    real_interest |   .3547894   .0173299    20.47   0.000     .3206764    .3889024
Unemployment_rate |   .0375936   .0089651     4.19   0.000     .0199464    .0552408
   logrealconsind |   9.613054   .3593735    26.75   0.000     8.905648    10.32046
       1.high_dev |  -.0261825   .0058349    -4.49   0.000    -.0376682   -.0146968
                  |
             Year |
            2013  |   .5348951   .0266802    20.05   0.000     .4823767    .5874134
            2014  |   .9567037   .0388947    24.60   0.000     .8801417    1.033266
            2015  |   1.067336   .0439418    24.29   0.000     .9808392    1.153833
            2016  |   .5203723   .0207381    25.09   0.000     .4795506    .5611941
            2017  |   .7161938   .0286912    24.96   0.000     .6597168    .7726707
                  |
            _cons |   -101.474   4.764903   -21.30   0.000    -110.8535    -92.0946
------------------+----------------------------------------------------------------
          sigma_u |  .20620563
          sigma_e |  .02854951
              rho |  .98119167   (fraction of variance due to u_i)
-----------------------------------------------------------------------------------

[Michael Schuster]

How did you xtset you data? Something like that:

xtset GM_code year

[Carlo Lazzaro]

Adam:
as already commented on, with a sky-rocketing -within Rsq- and a similar -rho-, what kind of improvement are you expecting from your regression results?

[Adam Klaas]

How did you xtset you data? Something like that:

xtset GM_code year

Yes sorry forgot to add that

[Adam Klaas]

Adam:
as already commented on, with a sky-rocketing -within Rsq- and a similar -rho-, what kind of improvement are you expecting from your regression results?

Well I thought maybe dropping some year because I have a lot of missing values for some variables:

* Example generated by -dataex-. For more info, type help dataex
clear
input float(real_interest logReal_income log_realHP logrealconsind) double Unemployment_rate
        .         . 11.066638         .                  .
        .         . 11.448387         .                  .
        .         . 11.449027         .                  .
    3.355         . 11.436362         .                  .
 3.809167         . 11.425336         .                  .
 3.126667         .  11.88968         .                  .
 3.594167         . 11.878828         .                  .
 3.216667         .   11.9539         . 1.1709528024032914
 2.404167         . 12.027326         . 1.0189597064731957
3.6008334         . 12.061186         .   1.65181669133134
 3.459167         . 12.058692         . 1.7168838356396807
 2.470833 10.228337  12.03159         . 1.8364521815338959
2.2066667  10.21136 11.994463 11.449235 2.3129111842105266
2.0708334 10.195252 11.930356  11.39064 2.5779070224387945
 3.365833 10.225975   11.8488  11.28985 2.5307801523129543
   3.5075 10.213326 11.817344 11.270553 2.3942860219851227
3.4741666 10.241876 11.832868  11.32823 2.2004945873548722
2.0316668 10.274782 11.876635 11.364115 1.9359284664104206
1.4991666 10.283296 11.936114  11.46233 1.5915721050958196
    .4025  10.30509 12.051608 11.507878  1.363041781627516
    1.435         . 12.139122 11.534215                  .
        .         . 11.373663         .                  .
        .         .   11.8328         .                  .
        .         . 11.792617         .                  .
    3.355         . 11.767217         .                  .
 3.809167         . 11.753452         .                  .
 3.126667         . 12.143667         .                  .
 3.594167         .  12.12681         .                  .
 3.216667         . 12.205215         . 1.3994621962031346
 2.404167         .  12.26741         . 1.2629286880783888
3.6008334         . 12.281482         . 1.9913041609661408
 3.459167         . 12.280895         . 1.9178043271348002
 2.470833 10.303476 12.225747         . 1.9098898465745313
2.2066667 10.286084 12.191173 11.449235   2.37366700042712
2.0708334  10.26936 12.096637  11.39064 2.7690231192558423
 3.365833 10.301238  12.01052  11.28985 2.6149684400360687
   3.5075 10.293784 11.996036 11.270553 2.3745448788982113
3.4741666 10.326927 12.025772  11.32823 2.2898609680403648
2.0316668  10.35463 12.065701 11.364115  2.007377765032522
1.4991666 10.368253 12.124705  11.46233 1.7314613663629586
    .4025 10.389135  12.19029 11.507878 1.5859610214620499
    1.435         . 12.223978 11.534215                  .
        .         . 11.289782         .                  .
        .         . 11.818513         .                  .
        .         .   11.7855         .                  .
    3.355         . 11.753032         .                  .
 3.809167         . 11.739367         .                  .
 3.126667         . 12.134277         .                  .
 3.594167         .  12.11742         .                  .
 3.216667         . 12.226357         .               1.33
 2.404167         .  12.26741         .               1.05
3.6008334         . 12.273849         . 1.6038806086922506
 3.459167         . 12.269553         . 1.6676362226100447
 2.470833  10.33514 12.249003         . 1.7095345403683269
2.2066667 10.327867 12.199078 11.449235 2.3100113542930973
2.0708334 10.308275  12.11769  11.39064 2.4056473557597506
 3.365833 10.347644 12.024005  11.28985 2.3806689679800024
   3.5075 10.330508 11.982306 11.270553 2.2976666798716474
3.4741666 10.366885 11.976323  11.32823 2.2146642410820214
2.0316668 10.391165  12.01838 11.364115 2.0086648286727056
1.4991666 10.399435  12.01325  11.46233  1.615063420783109
    .4025 10.417065 12.068002 11.507878  1.218314010611122
    1.435         . 12.101523 11.534215                  .
        .         . 11.552146         .                  .
        .         . 12.104395         .                  .
        .         . 12.069604         .                  .
    3.355         . 12.047832         .                  .
 3.809167         . 12.032354         .                  .
 3.126667         . 12.387163         .                  .
 3.594167         . 12.373962         .                  .
 3.216667         .   12.3918         .  .7958766012279239
 2.404167         . 12.434464         .  .6784260515603799
3.6008334         . 12.436175         . 1.1649021824785006
 3.459167         . 12.440403         . 1.1593626817715061
 2.470833 10.482217 12.424276         . 1.2185482808589119
2.2066667  10.48165  12.37217 11.449235 1.4044026389705402
2.0708334 10.450358 12.310374  11.39064 2.0481810201892126
 3.365833  10.50418  12.23276  11.28985 2.0915789812401457
   3.5075  10.46103 12.226426 11.270553 1.9489440557206301
3.4741666 10.506447 12.241873  11.32823  1.880597966404233
2.0316668 10.546596  12.27701 11.364115  1.492110860662243
1.4991666  10.54604 12.296556  11.46233 1.2607160867372669
    .4025  10.57182 12.403312 11.507878 1.1927555792711635
    1.435         . 12.442365 11.534215                  .
        .         . 11.127263         .                  .
        .         . 11.796694         .                  .
        .         . 11.756512         .                  .
    3.355         . 11.724045         .                  .
 3.809167         . 11.710588         .                  .
 3.126667         .  12.07601         .                  .
 3.594167         . 12.059152         .                  .
 3.216667         . 12.111186         . 1.0519442832269297
 2.404167         . 12.145667         .  .9472802127737094
3.6008334         .  12.14305         . 1.6363636363636365
 3.459167         . 12.139817         . 1.4213345967418638
 2.470833 10.279052  12.10502         .  1.501556491485076
2.2066667 10.261792  12.04164 11.449235  1.957019422494646
2.0708334 10.239828 11.984159  11.39064  2.443268056121127
 3.365833  10.27592 11.880217  11.28985 2.4531668153434434
   3.5075 10.250465 11.833517 11.270553 1.9693816884661117
end

------------------ copy up to and including the previous line ------------------

for example for the variables:
logrealconsind(construction cost) I only have values from 2012 until 2020 everything before 2012 are missing(the same for log real income)
unemployment rate missing values all years before 2007

[Carlo Lazzaro]

Adam:
there's no gain in -drop-ping observations with missing vakues, as Stata applies listwise deletion by default.
In addition, please consider that making-up your panel dataset is the best way to end up with a subsample of observations that is expected to show a very poor relationship with your original dataset, making all the regression estimates sadly unreliable.