Interaction term in fixed effects

[Julia Raciniewska]

Hello,

I would like to clarify the following uncertainty: I am analysing the impact of technology on different occupations and use panel data on 11 industries between 2006-2017. I have created a fixed effects model in order to look for the similar effect across all industries after accounting for individual effects, however, I would also like to look later into the different effect of computer_use for each industry. Hence, I decided to use the interaction of computer_use with industry dummies for the second model.
However, when I use the latter regression, other variables that were significant in the first model become now insignificant. Can I still use the first regression to interpret the effect of other variables, and from the latter just refer to different effects of computer_use?
Or does it make my results uncomparable?

. xtreg nonrout using_computer lngva  price_computer total_internet_access sharedegre
> e sharehigher shareother, fe vce(robust)

Fixed-effects (within) regression               Number of obs     =        120
Group variable: industry1                       Number of groups  =         10

R-sq:                                           Obs per group:
     within  = 0.3276                                         min =         12
     between = 0.4580                                         avg =       12.0
     overall = 0.4408                                         max =         12

                                                F(7,9)            =      16.27
corr(u_i, Xb)  = 0.5375                         Prob > F          =     0.0002

                                   (Std. Err. adjusted for 10 clusters in industry1)
------------------------------------------------------------------------------------
                   |               Robust
           nonrout |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------------+----------------------------------------------------------------
    using_computer |   .0014271   .0004206     3.39   0.008     .0004755    .0023786
             lngva |  -.0193869   .0317928    -0.61   0.557    -.0913072    .0525334
    price_computer |   .0014037   .0009901     1.42   0.190     -.000836    .0036434
total_internet_a~s |   .0041153   .0022304     1.85   0.098    -.0009303    .0091609
       sharedegree |   .0926562   .1112741     0.83   0.427    -.1590632    .3443756
       sharehigher |  -.2771514   .1359427    -2.04   0.072    -.5846752    .0303723
        shareother |   .1583427   .0836769     1.89   0.091    -.0309475    .3476329
             _cons |    .200577   .5024723     0.40   0.699    -.9360942    1.337248
-------------------+----------------------------------------------------------------
           sigma_u |   .1681399
           sigma_e |  .01373558
               rho |  .99337076   (fraction of variance due to u_i)
------------------------------------------------------------------------------------


Now - with the interaction term:

. xtreg nonrout c.using_computer#i.industry1 lngva  price_computer total_internet_acc
> ess sharedegree sharehigher shareother, fe vce(robust)

Fixed-effects (within) regression               Number of obs     =        120
Group variable: industry1                       Number of groups  =         10

R-sq:                                           Obs per group:
     within  = 0.4646                                         min =         12
     between = 0.0007                                         avg =       12.0
     overall = 0.0008                                         max =         12

                                                F(6,9)            =          .
corr(u_i, Xb)  = -0.7915                        Prob > F          =          .

                                   (Std. Err. adjusted for 10 clusters in industry1)
------------------------------------------------------------------------------------
                   |               Robust
           nonrout |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------------+----------------------------------------------------------------
         industry1#|
  c.using_computer |
    Accommodation  |  -.0019848   .0006782    -2.93   0.017    -.0035191   -.0004505
Administrative ..  |   .0011701   .0002696     4.34   0.002     .0005603    .0017799
     Construction  |   .0019143   .0003306     5.79   0.000     .0011663    .0026623
Financial and I..  |  -.0067784   .0034683    -1.95   0.082    -.0146244    .0010675
Information and..  |   .0032694   .0010298     3.17   0.011     .0009398     .005599
    Manufacturing  |   .0017933   .0013022     1.38   0.202    -.0011526    .0047392
Professional, S..  |  -.0004355   .0008249    -0.53   0.610    -.0023016    .0014306
      Real Estate  |   .0019032   .0002549     7.47   0.000     .0013266    .0024799
Transportation ..  |  -.0000783   .0002766    -0.28   0.784    -.0007041    .0005475
  Wholesale trade  |  -.0000112   .0014483    -0.01   0.994    -.0032876    .0032651
                   |
             lngva |   .0000949   .0355875     0.00   0.998    -.0804097    .0805995
    price_computer |   .0013431   .0009608     1.40   0.196    -.0008305    .0035166
total_internet_a~s |   .0037761   .0023165     1.63   0.138    -.0014641    .0090163
       sharedegree |   .0639118   .1022921     0.62   0.548     -.167489    .2953126
       sharehigher |  -.2078717   .1323905    -1.57   0.151    -.5073598    .0916164
        shareother |   .0901104   .0656804     1.37   0.203     -.058469    .2386898
             _cons |    .127795    .461287     0.28   0.788    -.9157087    1.171299
-------------------+----------------------------------------------------------------
           sigma_u |   .3108439
           sigma_e |  .01283018
               rho |  .99829925   (fraction of variance due to u_i)
------------------------------------------------------------------------------------

. 
.

[Clyde Schechter]

It isn't reasonable to mix findings from the two models in that way. It might be reasonable to report both models in their entirety if there is some doubt in your mind whether the interaction terms are really useful.

I think you are over-reading your results anyway. Most of your coefficient estimates have pretty wide confidence intervals. As far as I see, all of the changes in coefficients between models fall within the confidence intervals of the original model, and all of the coefficients in the original model fall within the confidence intervals of the expanded model. So I would say that nothing has really changed much anyway.

In both models, also, you are really pushing a large number of variables into a model that you are fitting to only 120 observations and 10 groups.

[Anne Nissen]

Hi Clyde,

Hope it is ok, that I continue this thread a bit. I was wondering if you can elaborate on the issues that might occur when using too many variables in the model and if you might have some literature on this?

I am researching energy poverty at the moment in EU countries, and I think I might be fitting too many variables into my model.
I have 130 obs, 17 groups, t=2007-2014 and I am trying to fit 17 variables into my model. I am using this amount of variables because I am trying to control for as much as possible.
I am researching if a state's welfare generosity (I have 4 variables for this) affects the degree of energy poverty observed and if the level of renewables consumed in a country also affects this (1 variable).
My main goal with including 12 control variables is to control for different circumstantial aspects. The biggest part of my control variables is energy prices, where I have included energy prices by source in my dataset and have 7 variables on this. Do you think, that I should consider using one 'total' variable instead of the seven source-specific ones and would this be a good enough adjustment, so I don't have to exclude more of my control variables?

Hope you can help me with making some sense of this.

- Anne

[Carlo Lazzaro]

Anne:
welcome to this forum.
How can interested listers help you out without reading what you typed and what Stata gave you back (as recommended by the FAQ, that first-time [as well as n-time] posters are kindly requested to read [and/or rehearse])? Thanks.

[Anne Nissen]

Hi Carlo,
I will try and upload my outputs as per the faq - thanks for the recommendation.

I am experiencing multiple issues as this is my first time dealing with panel data. I have, of course, tried following other threads in this forum to accommodate these issues, but I will describe them below.

As mentioned above, I am researching if welfare generosity affects energy poverty levels in EU countries. I have four generosity measures based on previous literature from the relative income poverty literature, seeing as energy poverty is quite a new subject to be addressed in an EU context in the political science world, which is my perspective.

Now as for the coding. I am not sure, i am doing this correctly, so please let me know if I should do it some other way or include more information.
First an output and summary of my data

* Example generated by -dataex-. For more info, type help dataex
clear
year cntryid float(tincw hhep hhgp dipa ddpa noren noown povri ewmd epad epel epop epug wgswe wgsst wgge wgphs rec)
2007  1  4.13  17.4  6.1 25.2 35.4 1.4 1.7 16.7  9.1  92.33  78.12  93.29  93.35 25.09 19.4 49.2 74.37 29.03
2008  1  6.15 17.76 6.01 24.5 36.5 1.3 1.7 20.6   12  110.5  88.04 106.02 100.73 25.52 19.6 49.9 74.96 29.76
2009  1   6.2 19.09 6.35 25.4 35.9 1.4 1.8 19.1 19.7  86.85  92.16  86.96  87.01 27.47 21.2 54.1  75.1 31.29
2010  1   6.7 19.49 6.12   25 35.8 1.4 1.8 18.9  9.1  98.64   97.4 100.73  98.99 27.56 21.4 52.8 74.65 31.13
2011  1  5.83 19.76 7.08 25.4 36.6 1.4 1.8 19.2   11 118.67   98.2 117.36 113.12 26.74 20.7 50.9 74.61 31.64
2012  1   5.5  19.5 7.61 28.1 30.2 1.4 1.8 18.5 15.2  125.8  98.93  125.5  120.7 27.11 20.9 51.2 74.68 34.28
2013  1  5.63  20.5 7.61 29.5 29.7 1.4 1.7 18.8 12.8 121.06 102.57 120.34 115.88 27.52 21.3 51.6 74.01  34.7
2014  1  4.98 20.04 7.39 29.4 29.8 1.3 1.7 19.2  6.7 115.95 100.75 115.41  112.3 27.73 21.5 52.4 74.01 35.79
2007  7  8.25 24.01  8.5 41.7 33.3 1.8 1.9 16.8  9.5  88.42  88.47  87.15  87.93 25.91 21.5 49.6 83.71 17.68
2008  7  3.58  27.1 9.69 41.3 33.3 1.8 1.9 16.3 11.5  101.8  95.34     93  92.63 26.32 21.6 50.4 84.02 18.53
2009  7  3.35 26.27 8.27 41.4 33.6 1.8   2 17.6 18.5  83.77  92.33  84.52  86.01 29.28 24.3 56.5 84.46 19.32
2010  7  3.75 26.89  9.6 41.9 33.4 1.8   2 18.3 10.3  96.51  93.75  96.95  96.88 29.64 24.8 56.7 83.88 21.18
2011  7  6.08 29.42 10.3 39.9 33.6 1.8 1.9 17.6 11.2 112.07 102.67 108.44 107.56 29.78 24.7 56.4 83.61 23.81
2012  7  6.53 29.85 9.68 21.7 33.7 1.9   2 17.5 10.6 118.63 104.04 115.03 113.88 29.93 24.6   58 83.98 26.91
2013  7  6.95 29.68 9.85 20.3 34.2 1.8   2 18.3 19.1 121.29  98.48 115.84 114.07 29.89 24.5 55.8 84.26  27.1
2014  7  6.35 30.39 8.93 20.6 34.2 1.9   2 17.9  7.3  114.6 100.52 110.45  109.1 29.88   24 55.2 84.19 30.23
2007  8  8.38  7.86 2.63  2.6 46.2  .9 1.2   22   12  80.43  59.86   79.6  79.68 12.37    9 33.7 75.92 18.55
2008  8  6.68  8.32 3.53  2.9 46.1   1 1.2 21.8    9 106.26  63.69  94.62   94.5 15.12 11.3 39.5 76.98 20.24
2009  8   8.4  9.21 3.79  3.6 45.9   1 1.2 23.4  7.2  84.03  70.84  83.36  83.46 19.31   15 45.9 77.87 24.08
2010  8     9  9.87 3.82  3.4 45.7   1 1.2 21.7  6.1 102.53  76.29 100.72 100.77 17.99 13.8 40.3 76.33 25.39
2011  8  9.25 10.08 4.28    3 45.7 1.4 1.6 23.1 12.5 117.65  78.12 112.56 112.45 16.04 12.2 37.4 76.69 25.09
2012  8  9.68  11.1 5.09 16.9   42 1.4 1.6 23.4 15.5  127.3     86 123.52  123.4 15.64   12 39.2 76.65 25.02
2013  8  8.43 13.59    5 17.1 42.1 1.3 1.5 23.5   17  122.7 104.73 119.06 118.96 15.58 11.6 38.4 75.61 24.55
2014  8  7.33 13.16 4.93 15.7 43.1 1.3 1.7   26   10 117.44 101.27 115.18  115.1 15.95 11.5 37.8 75.68 25.49
2007 10   9.9    14 5.81 20.9 51.9 1.6 1.9 23.3   19   85.9  53.97  85.41  84.28  20.9 12.8 39.3  72.3  9.01
2008 10   8.3 14.62 6.14 22.1 51.2 1.6 1.9 23.8 18.3 101.32  59.03   94.4  90.22 22.33 13.8 41.4 73.28  9.74
2009 10  9.75 16.31 5.73 22.1 51.4 1.6   2 24.7 20.5  81.69  65.14  81.88  81.65 25.61 16.1 46.2 75.07 12.23
2010 10 11.08  17.9 5.37 22.3 51.2 1.7   2 26.1 14.2  96.37  70.65  96.25  94.87  24.9 16.7   46 74.44 14.41
2011 10   8.7 20.35 5.38 22.1 50.8 1.6 1.9 26.7 17.1 113.83  80.61 110.43 107.47 25.53   17 46.2 73.48 14.77
2012 10  9.43 22.33 7.62 22.9 50.4 1.6   2 27.2 29.2 122.35  88.53 119.68 116.01 25.55 17.8 48.7 72.11  15.8
2013 10 10.25 22.51 8.12 23.3 49.9 1.6   2 27.3 15.7 121.81  87.34 119.64  116.6  25.7 18.2 45.8 71.01 16.96
2014 10 12.13 22.66 8.56 22.5 51.1 1.6   2 29.2 20.1 117.13  97.31 115.44 112.97 25.37 17.9 45.1 70.26 17.37
2007 12  7.33 12.22 5.15 34.9 47.2 1.4 1.8   19 11.5  94.51   70.3  93.16  93.83 28.28 21.7 52.6 77.02  9.45
2008 12  7.38 12.08  5.5 35.5 46.9 1.5 1.9 18.5 13.5 109.97  69.17 105.01  99.99 28.52 21.8 53.3 76.49  10.6
2009 12  7.78 12.07 5.67 35.4 47.1 1.5 1.9 18.5 19.4  86.72  70.71  86.63  88.97 31.02 23.7 57.2 76.51 11.32
2010 12  7.75 13.17 5.48 35.1 46.7 1.5 1.9 19.2 12.6  99.12  76.61  99.44  99.09 31.08 23.7 56.9 76.26 11.99
2011 12   7.5 14.03 6.13 35.6 46.2 1.5 1.9 19.3 13.2 115.66  82.63 114.17 110.57 30.78 23.7 56.3    76 10.83
2012 12  7.88 14.47 6.59 19.7 45.8 1.5 1.9 19.1 18.3  120.9  85.12 121.31 115.51 31.21 24.2 57.1 76.09 12.53
2013 12  8.05  15.6 7.04   20   46 1.5   2 18.1 18.4 116.89  90.33  118.3 113.37 31.71 24.5 57.2 76.23 13.57
2014 12  7.88 16.44 7.32 19.8 45.2 1.5   2 18.5   11 111.53  94.49 112.47 109.58 32.03 24.5 57.2 76.47 13.35
2007 14  14.6 12.96 3.82 20.6 31.7  .7 1.1 29.4  8.3  73.64  96.64  77.29  77.29 22.42 17.2 49.8 68.87     9
2008 14  16.1 15.51 4.35 20.9 31.4  .7 1.1 28.2  9.6  86.18 108.07  82.02  82.02 22.66 17.4 48.7 68.87  8.96
2009 14 13.25 15.73 4.79 20.1 31.7  .8 1.1 29.6 13.3  75.16 116.48  79.49  79.49 23.25   18 50.5  68.3 12.29
2010 14 16.93 16.38 5.45 20.4 31.4  .8 1.1 29.9  8.1  89.76 126.96  94.49  94.49    23 17.2 48.8 67.08 13.46
2011 14  17.3 16.18 5.66 21.4 30.7  .8 1.1 31.5 12.3 106.07 122.72 106.97 106.97 22.31 16.8   49  66.5 14.74
2012 14 19.78 15.84    5 33.7 29.2  .9 1.1 33.5 13.5 121.23  128.3 119.19 119.19  22.6 16.6 49.1 65.51 16.47
2013 14 20.23 13.62 4.26 32.7 29.8  .9 1.2 34.8 12.1 118.92 113.75 116.62 116.62 22.22 16.4   50 66.62 17.18
2014 14  18.1 11.74 3.58 31.1 29.3  .9 1.2 31.8  9.5 116.11 102.88 113.59 113.59 21.46 15.3   50 67.08 15.72
2007 15     7 19.18 6.07 29.1   34 1.6 2.1 23.1 16.3  86.09  78.14  83.06  81.73 16.23 12.8 35.6 79.23  3.53
2008 15   6.9 19.01 5.97 28.4 34.2 1.7 2.1 23.7 14.9 101.41  80.32  92.08  89.69  19.2 15.1 41.6 79.32  4.14
2009 15  8.15 19.43 5.97 28.2   34 1.6 2.2 25.7 21.4  81.29  80.68  81.45  80.51 23.11 17.7 46.9 77.07  5.22
2010 15   9.7  18.4 5.12 27.7 33.1 1.7 2.2 27.3 15.2  96.09  77.18  97.02  95.24 23.78 17.5 64.9 76.18  5.24
2011 15  9.88 19.94 5.64 27.4 34.2 1.7 2.2 29.4 15.8 112.67  81.97 111.09 108.49 22.96   16 47.3 71.89  6.68
2012 15  11.8 22.22 6.43 25.6 35.1 1.7 2.2 30.3 15.7 123.98  92.44 121.69 118.39 22.57 15.8 42.5 71.75  6.58
2013 15 13.05  23.5 6.88 25.1 35.4 1.6 2.2 29.9  6.9 120.93   96.9 119.68 116.43 21.45   15 40.6 70.59  7.35
2014 15  12.6 24.72 7.13 24.4 39.7 1.6 2.2 27.7 17.3 115.95 101.13 114.05 111.75 19.65 13.6 37.6 71.01  8.44
2007 16 13.28     . 6.36   40 43.2 1.2 1.4   26 10.6  82.76  76.07  85.94  84.46 24.05 17.4 46.8 77.49  8.73
2008 16 14.33 21.29 6.75 39.4 43.9 1.2 1.4 25.5 17.3   95.6  83.07  91.83  89.78 24.99   18 47.8 77.68 10.82
2009 16 13.45 20.48 6.46 40.1 43.7 1.2 1.5 24.9 16.7  76.88  84.76  81.38  80.21 27.02 19.7 51.1 78.31 12.54
2010 16 13.73 19.43 7.02 39.9 43.6 1.2 1.5   25  9.3  86.46  80.77  90.06   88.7    27 19.7 49.9 78.45 12.79
2011 16 17.75 20.26 7.85 40.8 43.3 1.2 1.4 28.1 15.3 102.97  82.13 102.82 101.12 26.61 19.5 49.2 77.01  11.9
2012 16 18.93 22.15 8.69 41.2 43.1 1.1 1.4 29.9 21.3 121.29  91.43 117.12 116.15 27.29 20.3 50.6 75.88 14.39
2013 16  18.1 23.08  8.9 40.5 44.1 1.1 1.4 28.5 13.6  117.9  95.06 114.59 113.69 27.89 20.9   51 75.76 16.32
2014 16  18.3 23.92 8.74 41.1 43.4 1.1 1.4 28.3   14 114.48  98.82 111.93 111.33 28.16 21.1 50.9 75.43 17.13
2012 17  24.6 12.64 5.61 10.3 42.4   1 1.6 32.5  8.2 124.63 101.57 123.02 122.24  16.5 13.3 36.2 67.32 23.92
2013 17 22.88 13.81 6.09  9.8 42.5  .9 1.5 30.8   16 123.76 110.43 119.62 119.08 15.49 12.5 35.6 66.25 26.34
2014 17 20.58 13.25 5.29 10.5 42.8  .9 1.5 27.3  6.1  118.5 105.86 114.51 114.05 15.63 12.6 34.7 67.56 27.73
2007 18   4.4 16.65 3.64   32   46 1.5   2 15.9  9.9  91.76  99.06  96.22  95.57 20.21 15.1 37.4 84.19  3.64
2008 18  4.78 16.27 5.36 31.8 45.5 1.4   2 15.5  9.4 108.78  86.71 105.94  100.3 21.94 15.5 37.9 87.33  3.96
2009 18   5.1 18.82 4.77 31.5 46.4 1.5 2.1 17.8 19.7  82.95  99.88  87.57  88.15 24.04 17.7 42.7 85.75  3.73
2010 18  5.08 17.37 4.54 33.4 43.1 1.5 2.1 17.1 12.9  97.84  95.65 100.79  99.06 23.14 17.1   42 84.91  3.68
2011 18  4.88  16.7 5.45 34.9 42.6 1.5 2.1 16.8 12.5 114.52  93.47 114.12 109.71 21.19   17 41.5 83.16  3.76
2012 18  5.13 17.01 5.86 37.8 15.9 1.5 2.2 18.4  9.5 123.36  95.76 123.77 117.72 22.07 17.3 41.8 82.85  4.15
2013 18   5.4 16.56 5.95   38 14.6 1.5 2.1   19 19.8 119.53  92.29 119.14 113.71 22.27 17.7 41.2 83.09  5.72
2014 18  4.85 17.38 5.23 36.6 14.5 1.5 2.2   19  9.9 114.33  97.47 113.73 109.88 21.85 17.5 40.6 83.48  6.85
2007 19 19.05  7.29 3.11  1.7 46.3  .8   1 35.1 16.6  84.56  44.13  80.66  80.62 11.08  7.9 34.5 60.87 33.08
2008 19 17.78  9.23 4.07  3.2 47.1  .8   1 34.2  8.5  103.2  56.44  90.96  90.83 12.99    9 38.3 60.33 33.71
2009 19 19.13 10.53 4.51  2.9 46.7  .9   1 37.9  8.1  86.95  65.01  85.12  85.34 18.09 13.9 45.1 59.65  38.4
2010 19 21.35 10.49  3.6  2.3 46.7  .8   1 38.2  9.5 101.32  65.05  97.69  97.42 19.47 14.1 45.8 60.19 33.07
2011 19  23.6 12.55 4.22  2.9 46.1   1 1.2 40.1 10.8 120.72  77.27 115.13 114.13 16.58 12.6 42.3  63.5 35.54
2012 19  22.6 13.76 5.36    8 45.3   1 1.3 36.2   15 128.44  81.95 125.36  124.1 15.36 11.4 38.8 60.35 40.37
2013 19 22.65 13.68 5.05  8.9 43.3   1 1.3 35.1 20.1 123.25  81.59 120.64 119.66 15.53 11.6 38.5 60.01  39.6
2014 19 20.18 13.33 4.86  9.7 43.4   1 1.3 32.7  5.8  118.6  82.18 115.18 114.28 15.35 11.4 38.9 59.67 40.24
2007 21  5.93  17.5 7.16 34.8 62.8 2.3 1.9 15.7 10.3  88.32 111.52  93.66  93.66 15.74 14.5 42.5 84.08  3.68
2008 21  5.35 17.84 7.27 35.5 62.4 2.3 1.9 14.9 12.6 104.54  88.98  98.66  98.66 15.48 14.8 43.2 82.68  3.94
2009 21  4.73 19.43 7.42 34.6 63.2 2.3 1.9 15.1 17.8  80.97  99.57  86.19  86.21 17.07 16.4 47.7 83.18  4.38
2010 21  5.48 17.78    7 33.1 64.9 2.3 1.9 15.1 10.7   95.1  89.52  96.46  96.47 17.47 16.8 47.9 83.38  3.85
2011 21  5.05 18.02 7.38   32 66.1 2.3 1.9 15.7  8.2 109.53  91.67 105.24 105.26 17.54 16.8 46.8 82.66  4.69
2012 21  5.73 18.57 7.92   38 47.3 2.3 1.9   15 12.3 117.33  99.37 112.86 112.89 17.93   17 46.9 82.11   4.9
2013 21  5.95 19.02 8.27 38.5 46.8 2.2 1.9 15.9   16 115.45 103.87 111.43 111.44 18.13 17.2 46.6 81.15  4.94
2014 21     6 18.21 8.05 40.2 45.1 2.1 1.8 16.5  6.6 114.27    102 109.42 109.43 17.93   17 45.9 81.08  5.53
2007 22  24.9  13.8 4.01 13.1 40.1  .8   1 34.4  7.7  84.11  67.58  90.53  90.95 19.53 15.6 42.9 70.08  7.28
2008 22 18.25 12.77 4.66 14.1   40  .9   1 30.5 11.1  94.24   75.3  93.09  92.86 20.21 15.5 44.2 71.67  8.01
2009 22 15.68 12.11 4.25 14.7 39.3  .8   1 27.8 16.3  81.36  84.59  88.88  89.06  21.3 16.1   45 71.65  8.84
2010 22 14.78 13.62 4.65 14.9 39.8  .8   1 27.8  4.9  95.09  87.44  98.21  98.26 20.75 16.3 45.8 71.68  9.49
2011 22  12.9 14.11 4.82 15.2 39.8  .8   1 27.2 10.7 112.51  95.07 110.84 110.65 19.56 15.5 44.1 70.87 10.44
2012 22 12.75 14.74 5.23 24.5 33.9  .8 1.1 26.7  7.2 126.51  100.5 123.26 123.07 19.78 15.7 43.1 70.04 10.91
2013 22 11.73 14.59  4.9 24.3 33.3  .8 1.1 25.8 17.5 122.27 100.45 118.48 118.33 20.48 16.1   43 70.66 11.41
2014 22  10.4 14.15 4.95 24.5 33.2  .8 1.1 24.7    8 116.28  98.14 114.04 113.93 20.18   16 42.6 70.66 11.58
2007 23 27.13 15.62 6.53 31.1 44.3 1.3 1.5   25 28.2  91.04  67.09  92.21  92.44 21.73 14.9 44.5 62.76 23.15
end

sum

    Variable    Obs    Mean    Std. dev.    Min    Max
                        
    year    131    2010.557    2.307714    2007    2014
    cntry    0
    cntryid    131    16.51145    7.224282    1    27
    bincw    131    20.72855    10.45256    5.4    45.18
    aincw    131    9.254809    5.393883    2.75    23.6
                        
    tincw    131    11.11542    6.319791    3.28    27.13
    hhep    130    17.01138    4.570348    7.29    30.39
    hhgp    131    6.375038    1.96457    2.63    12.26
    dipa    131    26.2374    10.93214    1.7    41.9
    ddpa    131    37.94122    11.07633    14.5    66.1
                        
    noren    131    1.317557    .4056298    .7    2.3
    noown    131    1.608397    .379481    1    2.2
    povri    131    23.11374    6.122151    13.9    40.1
    ewmd    131    13.47252    5.651195    3.5    36.7
    epad    131    105.3038    14.69675    73.64    128.44
                        
    epel    131    90.50809    15.57136    44.13    128.3
    epop    131    103.2085    13.43518    77.29    125.5
    epug    131    102.2399    12.74968    77.29    124.1
    wgswe    131    22.43557    4.77152    11.08    32.03
    wgsst    131    17.35649    3.666566    7.9    24.8
                        
    wgge    131    46.69924    6.179177    33.7    64.9
    wgphs    131    74.47069    6.988493    59.65    87.33
    rec    131    18.54664    12.05513    3.53    49.54

I then set my data as panel data

global id cntryid
global t year
global ylist tincw
global xlist hhep hhgp dipa ddpa noren noown povri ewmd epad epel epop epug wgswe wgsst wgge wgphs rec
xtset id year

Performing the hausman test on random and fixed models

 xtreg $ylist $xlist, fe

Fixed-effects (within) regression               Number of obs     =        130
Group variable: cntryid                         Number of groups  =         17

R-squared:                                      Obs per group:
     Within  = 0.5967                                         min =          3
     Between = 0.5160                                         avg =        7.6
     Overall = 0.4947                                         max =          8

                                                F(17,96)          =       8.35
corr(u_i, Xb) = -0.1052                         Prob > F          =     0.0000

------------------------------------------------------------------------------
       tincw | Coefficient  Std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
        hhep |   .0280269   .1932996     0.14   0.885    -.3556697    .4117235
        hhgp |   .4914672   .2721962     1.81   0.074    -.0488379    1.031772
        dipa |  -.0745511   .0314983    -2.37   0.020    -.1370748   -.0120274
        ddpa |  -.0081977   .0306862    -0.27   0.790    -.0691093    .0527139
       noren |   1.659452   2.747883     0.60   0.547    -3.795054    7.113957
       noown |  -.7908194     2.7713    -0.29   0.776    -6.291805    4.710167
       povri |   .7866492   .0971313     8.10   0.000      .593845    .9794533
        ewmd |  -.0224301   .0322808    -0.69   0.489    -.0865069    .0416466
        epad |  -.1133617   .0435624    -2.60   0.011    -.1998324   -.0268909
        epel |  -.0495486   .0315256    -1.57   0.119    -.1121263    .0130291
        epop |  -.1073209   .1337802    -0.80   0.424    -.3728724    .1582307
        epug |   .2627188   .1329301     1.98   0.051    -.0011454    .5265829
       wgswe |   .2540827   .2877072     0.88   0.379    -.3170116     .825177
       wgsst |  -.4646673   .3297627    -1.41   0.162    -1.119241    .1899066
        wgge |  -.0777319   .0663244    -1.17   0.244    -.2093848    .0539211
       wgphs |   .1234708    .098548     1.25   0.213    -.0721455    .3190872
         rec |  -.0151661    .085388    -0.18   0.859    -.1846599    .1543277
       _cons |  -11.29999   8.207012    -1.38   0.172    -27.59078    4.990802
-------------+----------------------------------------------------------------
     sigma_u |  4.5369775
     sigma_e |  1.3341949
         rho |  .92040526   (fraction of variance due to u_i)
------------------------------------------------------------------------------
F test that all u_i=0: F(16, 96) = 22.35                     Prob > F = 0.0000

. estimate store fixed

. xtreg $ylist $xlist, re

Random-effects GLS regression                   Number of obs     =        130
Group variable: cntryid                         Number of groups  =         17

R-squared:                                      Obs per group:
     Within  = 0.2475                                         min =          3
     Between = 0.9074                                         avg =        7.6
     Overall = 0.8444                                         max =          8

                                                Wald chi2(17)     =     607.60
corr(u_i, X) = 0 (assumed)                      Prob > chi2       =     0.0000

------------------------------------------------------------------------------
       tincw | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
        hhep |  -.1116863   .1118598    -1.00   0.318    -.3309275    .1075549
        hhgp |   1.168678   .2655854     4.40   0.000     .6481401    1.689216
        dipa |   .0720898    .037054     1.95   0.052    -.0005346    .1447143
        ddpa |   .0392789   .0362982     1.08   0.279    -.0318642    .1104221
       noren |  -3.542154   2.285047    -1.55   0.121    -8.020763     .936456
       noown |  -.2928582   2.041896    -0.14   0.886    -4.294901    3.709185
       povri |   .5071164   .0942246     5.38   0.000     .3224396    .6917933
        ewmd |    .106643   .0570682     1.87   0.062    -.0052086    .2184946
        epad |  -.1702194   .0705698    -2.41   0.016    -.3085338   -.0319051
        epel |  -.0689553   .0248606    -2.77   0.006    -.1176812   -.0202295
        epop |  -.2736536   .2108813    -1.30   0.194    -.6869733    .1396661
        epug |   .4909351   .2067978     2.37   0.018     .0856187    .8962514
       wgswe |  -.3989374   .2502493    -1.59   0.111     -.889417    .0915422
       wgsst |   .6348091   .3525678     1.80   0.072    -.0562111    1.325829
        wgge |  -.0935146   .1064199    -0.88   0.380    -.3020937    .1150645
       wgphs |  -.4297557   .0750673    -5.72   0.000    -.5768849   -.2826264
         rec |  -.0801897   .0334849    -2.39   0.017    -.1458188   -.0145606
       _cons |   32.17264   8.239852     3.90   0.000     16.02283    48.32246
-------------+----------------------------------------------------------------
     sigma_u |          0
     sigma_e |  1.3341949
         rho |          0   (fraction of variance due to u_i)
------------------------------------------------------------------------------

. estimate store random

. hausman fixed random

                 ---- Coefficients ----
             |      (b)          (B)            (b-B)     sqrt(diag(V_b-V_B))
             |     fixed        random       Difference       Std. err.
-------------+----------------------------------------------------------------
        hhep |    .0280269    -.1116863        .1397132        .1576455
        hhgp |    .4914672     1.168678       -.6772108        .0596249
        dipa |   -.0745511     .0720898       -.1466409               .
        ddpa |   -.0081977     .0392789       -.0474766               .
       noren |    1.659452    -3.542154        5.201605        1.526245
       noown |   -.7908194    -.2928582       -.4979612        1.873702
       povri |    .7866492     .5071164        .2795328        .0235842
        ewmd |   -.0224301      .106643       -.1290731               .
        epad |   -.1133617    -.1702194        .0568578               .
        epel |   -.0495486    -.0689553        .0194067        .0193859
        epop |   -.1073209    -.2736536        .1663327               .
        epug |    .2627188     .4909351       -.2282163               .
       wgswe |    .2540827    -.3989374        .6530201        .1419533
       wgsst |   -.4646673     .6348091       -1.099476               .
        wgge |   -.0777319    -.0935146        .0157827               .
       wgphs |    .1234708    -.4297557        .5532265        .0638484
         rec |   -.0151661    -.0801897        .0650236        .0785485
------------------------------------------------------------------------------
                          b = Consistent under H0 and Ha; obtained from xtreg.
           B = Inconsistent under Ha, efficient under H0; obtained from xtreg.

Test of H0: Difference in coefficients not systematic

   chi2(17) = (b-B)'[(V_b-V_B)^(-1)](b-B)
            = 103.59
Prob > chi2 = 0.0000
(V_b-V_B is not positive definite)

As the hausman test suggested, i ran the fixed effects model. And then performed some post-tests on this as i read about this on this forum.

. xtserial $ylist $xlist

Wooldridge test for autocorrelation in panel data
H0: no first order autocorrelation
    F(  1,      16) =      7.774
           Prob > F =      0.0132

. quietly xtreg $ylist $xlist, fe

. xtcsd, pesaran abs
 
 
Pesaran's test of cross sectional independence =    -0.433, Pr = 0.6652
 
Average absolute value of the off-diagonal elements =     0.400

. xttest3

Modified Wald test for groupwise heteroskedasticity
in fixed effect regression model

H0: sigma(i)^2 = sigma^2 for all i

chi2 (17)  =     826.11
Prob>chi2 =      0.0000

As i could observe my data both suffered from autocorrelation and groupwise heteroskedasticity, I ran the xtreg with cluster() as mentioned in "Robust standard errors for panel regressions with cross-sectional dependence" by Daniel Hoechle.

. xtreg $ylist $xlist, fe cluster()

Fixed-effects (within) regression               Number of obs     =        130
Group variable: cntryid                         Number of groups  =         17

R-squared:                                      Obs per group:
     Within  = 0.5967                                         min =          3
     Between = 0.5160                                         avg =        7.6
     Overall = 0.4947                                         max =          8

                                                F(17,96)          =       8.35
corr(u_i, Xb) = -0.1052                         Prob > F          =     0.0000

------------------------------------------------------------------------------
       tincw | Coefficient  Std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
        hhep |   .0280269   .1932996     0.14   0.885    -.3556697    .4117235
        hhgp |   .4914672   .2721962     1.81   0.074    -.0488379    1.031772
        dipa |  -.0745511   .0314983    -2.37   0.020    -.1370748   -.0120274
        ddpa |  -.0081977   .0306862    -0.27   0.790    -.0691093    .0527139
       noren |   1.659452   2.747883     0.60   0.547    -3.795054    7.113957
       noown |  -.7908194     2.7713    -0.29   0.776    -6.291805    4.710167
       povri |   .7866492   .0971313     8.10   0.000      .593845    .9794533
        ewmd |  -.0224301   .0322808    -0.69   0.489    -.0865069    .0416466
        epad |  -.1133617   .0435624    -2.60   0.011    -.1998324   -.0268909
        epel |  -.0495486   .0315256    -1.57   0.119    -.1121263    .0130291
        epop |  -.1073209   .1337802    -0.80   0.424    -.3728724    .1582307
        epug |   .2627188   .1329301     1.98   0.051    -.0011454    .5265829
       wgswe |   .2540827   .2877072     0.88   0.379    -.3170116     .825177
       wgsst |  -.4646673   .3297627    -1.41   0.162    -1.119241    .1899066
        wgge |  -.0777319   .0663244    -1.17   0.244    -.2093848    .0539211
       wgphs |   .1234708    .098548     1.25   0.213    -.0721455    .3190872
         rec |  -.0151661    .085388    -0.18   0.859    -.1846599    .1543277
       _cons |  -11.29999   8.207012    -1.38   0.172    -27.59078    4.990802
-------------+----------------------------------------------------------------
     sigma_u |  4.5369775
     sigma_e |  1.3341949
         rho |  .92040526   (fraction of variance due to u_i)
------------------------------------------------------------------------------
F test that all u_i=0: F(16, 96) = 22.35                     Prob > F = 0.0000

As you can see, I only get a few significant results on some of my control variables (dipa - dwelling in intermediately populated areas; povri- poverty risk; epad - energy price automotive diesel; energy price - unleaded gasoline, borderline significant).
I am not sure if I am doing something wrong in my data or if this is just the results, hence why i was curious about what the number of variables included in the xtreg could do to my results.

Please let me know, if you need more information. I have both been in doubt if i should include i.year when I perform my xtreg and whether this is the right way to go about control variables that might confound with each other as well?

And apologies for this long response, hope it makes sense and thank you for having a look at it!

- Anne

[Carlo Lazzaro]

Anne:
thanks for abiding by the FAQ.
As far as your query is concerned:
1) your -xset- seems correct;
2) it is not correct to run -.hausman- with deafault standard errors and then impose them. You should investigate the presence of heteroskedasticity and serial and/or across panels correlation before -hausman-. Therefore, while it's correct to impose cluster-robust standard errors to take both heteroskedastcity and autorrelation into account, you should switch from -hausman- to the community-contributed module -xtoverid- (-search xtoverid. from within Stata to spot and install it) to check with specification is the best one for your pane data regression model, as -hausman- deos not support non-default standard errors;
3) especially with -xtreg,fe- is a good idea to include -i.year- in the right-hand side of your regression equation.

[Anne Nissen]

Hi Carlo,
Thank you for your response, i wasn't aware of my flaw with testing after the hausman test so i greatly appreciate your response!

When I perform xtoverid in stata I get the following error message: Error - saved RE estimates are degenerate (sigma_u=0) and equivalent to pooled OLS
I read in a previous thread (sigma_u=0 in random effects model - Statalist), that it might be caused by one of the variables, and found that sigma_u=0 when I include welfare generosity public health spending (wgphs) and rec (consumption of renewables). It would hurt my analysis a bit to exclude the two.
In the thread I linked to, I could read that the person when removing his faulty variable, the rho was still low - however, mine looks good I think (?)
I performed the xtoverid on the new xtreg excluding the two variables

. xtreg $ylist hhep hhgp dipa ddpa noren noown povri ewmd epad epel epop epug wgswe wgsst wgge i.year, fe cluster()

Fixed-effects (within) regression               Number of obs     =        130
Group variable: cntryid                         Number of groups  =         17

R-squared:                                      Obs per group:
     Within  = 0.6238                                         min =          3
     Between = 0.5366                                         avg =        7.6
     Overall = 0.5225                                         max =          8

                                                F(22,91)          =       6.86
corr(u_i, Xb) = -0.0263                         Prob > F          =     0.0000

------------------------------------------------------------------------------
       tincw | Coefficient  Std. err.      t    P>|t|     [95% conf. interval]
-------------+----------------------------------------------------------------
        hhep |   .0825271   .1990451     0.41   0.679    -.3128517    .4779058
        hhgp |   .7093138   .2905711     2.44   0.017     .1321301    1.286498
        dipa |   -.077868   .0320761    -2.43   0.017    -.1415832   -.0141528
        ddpa |  -.0199812   .0321717    -0.62   0.536    -.0838865     .043924
       noren |   .7698163   2.898833     0.27   0.791     -4.98836    6.527993
       noown |   .3374957   2.907585     0.12   0.908    -5.438065    6.113056
       povri |   .6901674   .0941287     7.33   0.000     .5031923    .8771424
        ewmd |   -.010719   .0416958    -0.26   0.798    -.0935426    .0721046
        epad |  -.1314238   .0664569    -1.98   0.051    -.2634323    .0005848
        epel |  -.0756988   .0343182    -2.21   0.030    -.1438678   -.0075299
        epop |  -.0928882   .1380903    -0.67   0.503    -.3671875    .1814111
        epug |   .3574038   .1535379     2.33   0.022     .0524196    .6623881
       wgswe |   .2504394   .3322971     0.75   0.453    -.4096279    .9105067
       wgsst |  -.3365769    .353228    -0.95   0.343    -1.038221    .3650671
        wgge |  -.0707352   .0669977    -1.06   0.294     -.203818    .0623476
             |
        year |
       2008  |  -.2284542   .9580592    -0.24   0.812    -2.131521    1.674613
       2009  |  -.2159023   .8067717    -0.27   0.790    -1.818455    1.386651
       2010  |   -.915309   .8247258    -1.11   0.270    -2.553526    .7229075
       2011  |  -2.514369   1.259642    -2.00   0.049    -5.016494   -.0122451
       2012  |  -3.410177   1.721717    -1.98   0.051    -6.830157    .0098036
       2013  |  -2.888527   1.633085    -1.77   0.080     -6.13245    .3553956
       2014  |  -2.729789   1.450819    -1.88   0.063    -5.611663    .1520857
             |
       _cons |  -10.46765   5.312598    -1.97   0.052    -21.02048    .0851698
-------------+----------------------------------------------------------------
     sigma_u |  4.4383739
     sigma_e |  1.3233972
         rho |  .91835264   (fraction of variance due to u_i)
------------------------------------------------------------------------------
F test that all u_i=0: F(16, 91) = 28.13                     Prob > F = 0.0000

. est store fixed

. xtreg $ylist hhep hhgp dipa ddpa noren noown povri ewmd epad epel epop epug wgswe wgsst wgge, re

Random-effects GLS regression                   Number of obs     =        130
Group variable: cntryid                         Number of groups  =         17

R-squared:                                      Obs per group:
     Within  = 0.5874                                         min =          3
     Between = 0.6246                                         avg =        7.6
     Overall = 0.6001                                         max =          8

                                                Wald chi2(15)     =     165.25
corr(u_i, X) = 0 (assumed)                      Prob > chi2       =     0.0000

------------------------------------------------------------------------------
       tincw | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
        hhep |  -.0168247   .1578861    -0.11   0.915    -.3262758    .2926263
        hhgp |   .4983082    .253358     1.97   0.049     .0017356    .9948807
        dipa |  -.0629293   .0282748    -2.23   0.026    -.1183468   -.0075117
        ddpa |   .0005462   .0282073     0.02   0.985    -.0547391    .0558315
       noren |   1.209543   2.280141     0.53   0.596     -3.25945    5.678537
       noown |  -1.249435   2.449793    -0.51   0.610    -6.050941    3.552072
       povri |   .7504647   .0764499     9.82   0.000     .6006257    .9003038
        ewmd |  -.0162191   .0311648    -0.52   0.603    -.0773009    .0448628
        epad |  -.1115036   .0418541    -2.66   0.008    -.1935362    -.029471
        epel |  -.0462431    .026253    -1.76   0.078    -.0976979    .0052118
        epop |  -.0989915   .1293674    -0.77   0.444    -.3525469     .154564
        epug |   .2500325   .1254859     1.99   0.046     .0040847    .4959803
       wgswe |   .2026179   .2509482     0.81   0.419    -.2892315    .6944672
       wgsst |  -.3751621    .305643    -1.23   0.220    -.9742114    .2238872
        wgge |  -.0636361   .0633462    -1.00   0.315    -.1877924    .0605203
       _cons |   -1.14471   3.275348    -0.35   0.727    -7.564274    5.274854
-------------+----------------------------------------------------------------
     sigma_u |  4.4425074
     sigma_e |  1.3312634
         rho |  .91760037   (fraction of variance due to u_i)
------------------------------------------------------------------------------

. est store random

. xtoverid

Test of overidentifying restrictions: fixed vs random effects
Cross-section time-series model: xtreg re   
Sargan-Hansen statistic  12.445  Chi-sq(15)   P-value = 0.6450

I am a bit unsure of what my results mean in this case, i see in the description of xtoverid that the null hypothesis is that the excluded instruments are valid instruments which i am unable to reject - but practically, what does it mean?

[Carlo Lazzaro]

Anne:
-xtoverid. outcome tells you to go -xtreg, re-.
You can find clue of that in -xtoverid. helpfile, section that discusses the compruson of -re- (the null) and -fe- (the alternate) as a test of overidentfying restrictions.

[Anne Nissen]

Thank you very much for your help, I will take a look at it. Hope it's okay that I get back to the thread if I run into new problems.

[Carlo Lazzaro]

Anne:
yes, it is.

[Anne Nissen]

Hi Carlo, I am back with some new issues :D

I found a way to get a more extensive time period, as I had previously overlooked some available data, so my dataset has changed slightly. Thus, I now have a somewhat larger time period from 2007-2017.
I will produce the same outputs as last to give you an initial overview, just in case:

* Example generated by -dataex-. For more info, type help dataex
clear
input int year byte cntry float(bincw tincw hhep hhgp dipa ddpa noren noown povri eptot epad epop wgswe wgsst wgge wgphs) byte dc float(rec gcr ocr ccr ecr hcr rcr)
2007  1  9.38  4.13  17.4  6.1 25.2 35.4 1.4 1.7 16.7  85.93  92.33  93.29 25.09 19.4 49.2 74.37 2 29.03  23.22555 20.395636  1.0081159 24.711527   8.450013 22.205816
2008  1 11.93  6.15 17.76 6.01 24.5 36.5 1.3 1.7 20.6  95.71  110.5 106.02 25.52 19.6 49.9 74.96 2 29.76  23.48639  20.40695  1.0044297 24.268173   8.637357 22.193394
2009  1 11.65   6.2 19.09 6.35 25.4 35.9 1.4 1.8 19.1  89.71  86.85  86.96 27.47 21.2 54.1  75.1 2 31.29 23.490437 18.633514   .7448246 24.449175   8.977472 23.701353
2010  1 12.15   6.7 19.49 6.12   25 35.8 1.4 1.8 18.9  96.81  98.64 100.73 27.56 21.4 52.8 74.65 2 31.13  23.57059 19.024532   .7526115  23.09149   9.330699 24.226913
2011  1 12.38  5.83 19.76 7.08 25.4 36.6 1.4 1.8 19.2 105.57 118.67 117.36 26.74 20.7 50.9 74.61 2 31.64 22.565424  17.80501   .4971629 24.402155   9.786998 24.939806
2012  1 10.48   5.5  19.5 7.61 28.1 30.2 1.4 1.8 18.5 110.59  125.8  125.5 27.11 20.9 51.2 74.68 2 34.28  22.61912  16.47214   .5034489 24.117697  10.415233  25.86904
2013  1  12.5  5.63  20.5 7.61 29.5 29.7 1.4 1.7 18.8 110.32 121.06 120.34 27.52 21.3 51.6 74.01 2  34.7 21.627825  15.71511   .3932869  22.14183  11.863143  28.25569
2014  1   9.2  4.98 20.04 7.39 29.4 29.8 1.3 1.7 19.2  107.2 115.95 115.41 27.73 21.5 52.4 74.01 2 35.79  21.21227 15.412713   .3686959  24.30238  11.299107 27.401283
2015  1 10.85   5.1 19.96 7.21 30.1 29.9 1.3 1.8 18.3    100    100    100 27.74 21.2 51.1 74.06 2 34.88 21.683344 15.606082  .28485712  23.15494  11.703347 27.564137
2016  1 11.25   5.2 20.22 6.82 29.6 30.1 1.3 1.8   18  95.97   91.9  90.82 27.72   21 50.1 73.96 2 34.61  22.07724 15.520995  .27891415 22.616787  11.916186  27.58989
2017  1 11.75  5.08 19.64 6.87 30.3 30.6 1.3 1.8 18.1  98.14  98.64  97.58 27.26 20.6 49.3 74.02 2 33.96  21.97039   15.8495  .27847612  22.54122  11.797644 27.562754
2007  2 25.33 12.03 16.83    5 42.5 53.5   2 2.2 21.6  85.31  89.39  98.39 24.98 16.8 48.6 73.97 1  3.34  39.41136   32.3906  1.0942321  22.58899  .17656232 4.3382716
2008  2 18.53  8.98 20.62 6.57 42.7   53   2 2.2 20.8 102.95 102.33 115.36 26.47 17.5 50.8 75.54 1  3.91  36.69792 37.938927  1.4350085 19.111605   .1546265  4.661909
2009  2 17.63  7.83  18.9 5.61 42.1 53.8 2.1 2.3 20.2  87.12  83.43   89.7 28.76 19.2 54.7 76.09 1  4.83  38.80354 33.091167    2.46416 20.354927  .14660493  5.139606
2010  2    20     9 19.67 5.67 41.5 54.3   2 2.2 20.8  94.42  98.22 105.75 28.45 18.9 53.9 76.07 1  5.82  41.83622 32.886734  1.1185629 18.298855  .11681983  5.742823
2011  2  22.2 10.35 21.28 6.83 41.4 54.3 2.1 2.3   21 109.05  117.7 124.82 28.89 19.1 55.3 76.13 1  6.63  38.98323 33.042038  1.2313198  20.73156   .1617904  5.850071
2012  2 19.65   9.5 22.75 7.12   52   30 2.1 2.3 21.6  115.1 125.18 135.19  28.8 19.6 56.5 76.59 1  7.76  39.81797 31.316317  1.1497375 20.412014  .07826941  7.225688
2013  2 19.28  8.68 21.94 6.64 53.1 29.4 2.1 2.3 20.8 111.38 120.63 129.86  29.2 20.1 56.1 76.15 1  8.09  41.05736 31.490515   1.384388  18.77888  .04581846  7.243032
2014  2 19.63  8.53  20.7 6.54 53.2 28.8 2.1 2.2 21.2 105.59 115.71 121.11 29.11 19.8 55.6 76.36 1  9.12  39.00589  31.26771  1.1056286 21.777443  .05340703  6.789936
2015  2 19.78  8.43 22.39 6.03 52.8 29.1 2.1 2.3 21.1    100    100    100 29.52 19.4 53.7 75.94 1  9.38  38.83195 33.158264  1.1430725 19.595497 .023987735  7.247225
2016  2 20.23  8.45 26.45 5.41   54 28.1   2 2.2 20.9 101.49  97.02  89.93 28.85 19.4 53.1 77.26 1  9.21  40.94687 30.284184  1.0148362  19.48147 .010659258  8.261998
2017  2 20.15  8.53 28.38 5.36 54.6 27.4 1.9 2.2 20.6 108.11 108.51  99.21 28.72 19.3   52 77.35 1  9.64  40.82359 30.988714   .9711155  19.19027  .01105012  8.015184
2007  5 19.85   7.9  11.6 3.62 23.7 34.9   1 1.3 15.8  83.17  92.04  94.11 17.43 12.1 40.4 84.69 2  8.32  31.73323  .3921629  10.007526   19.6273  17.728952  20.51083
2008  5 17.48   7.1 14.15 4.84 24.7 34.8   1 1.3 15.3  90.26 101.77  96.74 17.72 12.2 40.9  82.1 2  8.99  31.42914 .27316406   9.902892   19.4067  17.588236  21.39985
2009  5 16.33  7.25 14.94 4.84 23.8 35.3   1 1.4   14  90.44  84.05  86.95    20 13.4 44.4 83.27 2 10.25 30.927687 .09430304   10.86059 18.966566  16.736694  22.41416
2010  5 15.03   6.6 15.23 4.93 24.6 35.5 1.1 1.4 14.4   94.8  98.42 101.62 19.57 13.5 43.6 83.33 2 10.95  32.01096  .3234589  12.777144  17.36242  16.621796  20.90422
2011  5 16.08  7.25 16.43  5.7 24.6 35.2 1.1 1.4 15.3 102.94 109.84 110.37 19.62 13.5 43.2 83.88 2 12.16  29.17096  .4573459  12.951976 17.787163   16.17372 23.458824
2012  5  17.5     7 16.66  6.6   31 30.2 1.1 1.5 15.4 110.59 116.86 117.05 19.83 13.7 44.7 83.74 2  12.8  28.45086  .6178253  13.187037 17.629211  16.451874 23.663197
2013  5  17.1   6.6 16.63 6.08 32.8 29.9 1.1 1.5 14.6 109.98 115.83  115.4 20.68 13.8 42.7 83.61 2 13.96 27.960726  .5331555  13.464274 17.429077  16.449474 24.163265
2014  5  17.7  6.53 13.84 5.56 31.2 31.5 1.1 1.5 14.8 105.53 116.31 115.23 20.07 13.4 42.6 82.69 2 14.86  25.13905  .6874444    12.6856 18.560434  15.374326 27.553144
2015  5  14.3  5.48 13.97 5.79 32.8 30.3 1.1 1.5   14    100    100    100 19.27 12.9 41.9 82.37 2 14.85  26.48091  .6657068  12.290433  18.30056  15.038042 27.224337
2016  5 12.83   4.7 14.21 5.73 32.3 30.2 1.1 1.5 13.3   96.2  87.76  91.21 18.89 12.7 39.8 81.98 2 14.78 28.174265  .6357044  11.588123 18.147747  14.936806  26.51734
2017  5 10.63  4.08 14.63 5.58 34.6 30.1 1.2 1.5 12.2  97.87  94.55  96.69 18.55 12.3   39 82.09 2 14.47  27.89005  .6259038  11.654857 18.198517   14.83082  26.79985
2007  6 14.43  6.95 20.65 6.13 34.5 50.7 1.6 1.9 20.6  85.67  98.35  96.33 24.21 18.9 43.4 74.92 2 10.47  38.70176  20.65273   1.317827 22.125376  8.0620575  9.140249
2008  6 16.78  7.43 21.72 7.02 34.5 50.6 1.6 1.9 20.1  93.17  112.1 106.48 24.35 18.8 44.2 75.05 2 10.25 36.572247  26.62609  1.0752764  19.79081    7.27728  8.658301
2009  6 16.58  7.15 22.88 6.19 35.3 49.7 1.6 1.9   20  87.45   91.7  91.63  26.8 20.7 48.2 83.34 2 10.72  37.33253 24.080725    1.08467  20.45085   7.200607  9.850622
2010  6 15.63   6.7 24.07 5.68 36.1 48.9 1.6 1.9 19.7  91.68 103.16 102.82 26.05 20.1 48.1 83.24 2 11.61   38.2658  22.94968  1.5880527  19.50209   7.221112 10.473257
2011  6 17.43     7  25.3 6.14 36.7 48.6 1.6 1.9 19.9  99.73 119.89 116.47 24.75 18.9 45.2 83.11 2 12.54 36.201603 22.658813   1.586333   21.5207   7.192253 10.840303
2012  6  14.8  6.58 26.36 6.43 39.6 35.3 1.6 1.9 19.6 105.09 125.36 124.73 24.65 18.9 44.9 82.98 2 13.64  38.17933  23.00872  1.3118844  20.80359   5.962346 10.734136
2013  6 15.28  6.83  29.2 6.75 41.1 34.8 1.6 1.9 20.3 107.65 120.25  120.2 24.83   19 44.9 83.84 2 13.63 37.696846 23.795984  1.1582955 19.584124   7.379486 10.385265
2014  6 14.05  6.58 29.78  6.8 41.4 35.4 1.6 1.9 20.6 106.17 114.66 113.74 24.77 18.8 44.3 84.21 2 14.02 35.512333  23.96168  1.1116024   21.6271   7.108074 10.679215
2015  6  13.8  6.25 29.49 6.79 41.8 35.8 1.6 1.9   20    100    100    100    25 19.1 44.1 84.18 2 14.55   37.1238  22.04626  1.1299255 20.812754   7.654703  11.23256
2016  6  13.5  5.88 29.73 6.52 40.8 36.1 1.6 1.9 19.7  96.28   92.5   91.1 25.25 19.5 44.4 84.32 2 14.24  39.42286  20.57042   .9271631 19.578287   7.886722  11.61455
2017  6 12.08  5.48 30.48  6.1 40.4 36.6 1.6   2   19  99.18  99.29  97.66 25.37 19.4 44.2 84.53 2 15.22  38.19159  20.06627   .8848022 19.492075   7.835474 13.529788
2007  7    14  8.25 24.01  8.5 41.7 33.3 1.8 1.9 16.8  89.46  88.42  87.15 25.91 21.5 49.6 83.71 2 17.68  13.86752 11.427213          0  19.70684   32.76115 22.049236
2008  7  6.95  3.58  27.1 9.69 41.3 33.3 1.8 1.9 16.3   96.6  101.8     93 26.32 21.6 50.4 84.02 2 18.53 13.886897 10.708405 .009711055 19.642576   33.79596 21.767275
2009  7   5.4  3.35 26.27 8.27 41.4 33.6 1.8   2 17.6  89.56  83.77  84.52 29.28 24.3 56.5 84.46 2 19.32 13.925094 10.106245 .009691102 19.251364  34.679955  21.81312
2010  7  7.98  3.75 26.89  9.6 41.9 33.4 1.8   2 18.3  98.33  96.51  96.95 29.64 24.8 56.7 83.88 2 21.18 14.800093  9.516583 .008732644 17.850822  36.578526  21.03346
2011  7 10.63  6.08 29.42 10.3 39.9 33.6 1.8 1.9 17.6 108.21 112.07 108.44 29.78 24.7 56.4 83.61 2 23.81  14.20326  8.911214 .009719974  19.33741  35.851624 21.447266
2012  7  12.1  6.53 29.85 9.68 21.7 33.7 1.9   2 17.5 110.99 118.63 115.03 29.93 24.6   58 83.98 2 26.91  14.43431  7.609967 .009832406  19.32504   37.10734  21.32586
2013  7  13.2  6.95 29.68 9.85 20.3 34.2 1.8   2 18.3 108.88 121.29 115.84 29.89 24.5 55.8 84.26 2  27.1 14.285147  7.111837 .009747467    19.768   36.71273 21.901087
2014  7   9.6  6.35 30.39 8.93 20.6 34.2 1.9   2 17.9  106.1  114.6 110.45 29.88   24 55.2 84.19 2 30.23 13.422942  5.639378          0  20.95508  36.426983  23.32076
2015  7 14.78  6.68 30.55 7.83 20.5 34.2 1.8   2 17.7    100    100    100 29.87 23.5 54.5 84.18 2 32.98 13.160286  5.406931          0 19.694605  36.463696 25.050444
2016  7 10.23  5.95 30.86 7.29 34.9 31.9 1.8   2 16.8  98.79  92.75  93.66 29.31 22.9 52.5 84.12 2 32.13 13.328979  5.040223 .003656468  19.26852  36.201275  25.91835
2017  7  11.4  5.95  30.3 8.42 34.6 32.3 1.7   2 17.2 101.41  99.34 100.04 29.16 22.2 50.5 84.02 2 35.51  12.79836  4.743867          0 18.660427   36.81436 26.797646
2007  8  15.8  8.38  7.86 2.63  2.6 46.2  .9 1.2   22  72.86  80.43   79.6 12.37    9 33.7 75.92 2 18.55  5.080664  .8757363   .6744272 15.852159  37.685673  39.39431
2008  8    12  6.68  8.32 3.53  2.9 46.1   1 1.2 21.8  84.18 106.26  94.62 15.12 11.3 39.5 76.98 2 20.24  5.163007 1.1109046   .6130034 16.660522  35.064487  40.78599
2009  8 13.93   8.4  9.21 3.79  3.6 45.9   1 1.2 23.4  79.28  84.03  83.36 19.31   15 45.9 77.87 2 24.08  5.222933  .8568472   .4681115 16.699345  34.081116  42.39589
2010  8 17.63     9  9.87 3.82  3.4 45.7   1 1.2 21.7  91.12 102.53 100.72 17.99 13.8 40.3 76.33 2 25.39   5.34655  .8127778  .50542974  16.93988  34.824986  41.23547
2011  8 17.58  9.25 10.08 4.28    3 45.7 1.4 1.6 23.1   98.9 117.65 112.56 16.04 12.2 37.4 76.69 2 25.09  5.525457  1.000371   .6958545 17.780643   35.56952  38.97876
2012  8 17.38  9.68  11.1 5.09 16.9   42 1.4 1.6 23.4 108.61  127.3 123.52 15.64   12 39.2 76.65 2 25.02  5.699362  .8561764   .6032994 17.323975   35.01156  40.09433
2013  8 13.58  8.43 13.59    5 17.1 42.1 1.3 1.5 23.5 113.38  122.7 119.06 15.58 11.6 38.4 75.61 2 24.55  5.600286  .7822124   .6251701 17.178358  35.359707  40.02663
2014  8 11.93  7.33 13.16 4.93 15.7 43.1 1.3 1.7   26 109.92 117.44 115.18 15.95 11.5 37.8 75.68 2 25.49  5.879008  .9444425    .511179 16.835915  33.594364  41.82626
2015  8  11.5  6.33 12.97  4.2 15.1 43.8 1.3 1.7 24.2    100    100    100 17.15 12.5 39.5 75.62 2 27.56  5.751106 1.1050401   .3025302 17.328545    33.0621  42.15422
2016  8  11.4   6.8 12.23 3.37 15.2   44 1.4 1.7 24.4  94.32  95.94  96.07 17.51 12.8 39.4 75.66 2 26.91  6.218501 1.1431605   .2090099 17.675896   33.51261  41.24082
2017  8  11.8   6.5 12.63 4.16 15.3 43.9 1.4 1.7 23.4 102.74 109.09 107.58 17.24 12.6 39.2 73.59 2 27.03  5.877591 1.0215439  .22302505 17.698767  33.755405  41.42388
end

sum

    Variable    Obs    Mean    Std. dev.    Min    Max
                        
    year    242    2012    3.168832    2007    2017
    cntry    242    14.27273    7.638622    1    27
    bincw    242    20.40496    10.30543    5.4    47.58
    tincw    242    10.8007    6.466974    2.98    29.2
    hhep    241    17.42664    5.016967    7.29    30.86
                        
    hhgp    226    6.239469    1.906045    2.35    12.26
    dipa    237    27.6308    11.36708    .8    54.6
    ddpa    242    37.58802    10.31794    13    66.1
    noren    242    1.345455    .3944638    .6    2.3
    noown    242    1.650826    .3823103    1    2.3
                        
    povri    242    22.68058    5.998838    12.2    40.1
    eptot    242    96.93992    10.64845    66.06    122.27
    epad    242    103.1803    13.17705    73.64    130.26
    epop    242    102.0201    12.28347    71.03    135.19
    wgswe    242    22.82558    4.769228    11.08    32.03
                        
    wgsst    242    17.52603    3.756922    7.9    25.5
    wgge    242    46.65702    6.510901    26.2    64.9
    wgphs    242    74.73178    7.258142    55.87    87.33
    dc    242    1.5    .893731    0    3
    rec    242    18.95603    11.96193    3.34    52.89
                        
    gcr    242    24.99163    19.51676    .3802139    77.83798
    ocr    242    13.65882    13.29408    .094303    49.49794
    ccr    242    2.990839    7.282563    0    35.87513
    ecr    242    23.84931    10.17473    10.58211    51.69038
    hcr    242    13.93679    13.24902    0    40.24403
                        
    rcr    242    20.49344    13.06915    .7375578    52.39993

My first question regards the random effects model:
(1): Are there any other tests I should perform besides the Breusch and Pagan Lagrangian multiplier test for random effects on this kind of model before doing my Hausman test as there is with the fixed effects model?

. xtreg tincw dipa ddpa noren noown povri wgswe wgsst wgge wgphs rec epad hhgp hhep epad epop, re
note: epad omitted because of collinearity.

Random-effects GLS regression                   Number of obs     =        220
Group variable: cntry                           Number of groups  =         21

R-squared:                                      Obs per group:
     Within  = 0.6009                                         min =          6
     Between = 0.6740                                         avg =       10.5
     Overall = 0.6322                                         max =         11

                                                Wald chi2(14)     =     319.08
corr(u_i, X) = 0 (assumed)                      Prob > chi2       =     0.0000

------------------------------------------------------------------------------
       tincw | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
        dipa |  -.0392455   .0176258    -2.23   0.026    -.0737915   -.0046996
        ddpa |   .0063827   .0171118     0.37   0.709    -.0271557    .0399212
       noren |   -.738471   1.540009    -0.48   0.632    -3.756832     2.27989
       noown |  -1.935602   1.678005    -1.15   0.249    -5.224431    1.353227
       povri |   .7672141   .0519291    14.77   0.000     .6654349    .8689932
       wgswe |   .2051212   .1863756     1.10   0.271    -.1601682    .5704107
       wgsst |  -.4157811   .2131102    -1.95   0.051    -.8334694    .0019073
        wgge |  -.0286019   .0484703    -0.59   0.555    -.1236019    .0663981
       wgphs |    .054728   .0482755     1.13   0.257    -.0398903    .1493463
         rec |  -.0161759   .0357048    -0.45   0.651     -.086156    .0538042
        epad |  -.0628722   .0263954    -2.38   0.017    -.1146063   -.0111381
        hhgp |   .1789633   .1438991     1.24   0.214    -.1030737    .4610002
        hhep |  -.0303706   .0580303    -0.52   0.601    -.1441079    .0833666
        epad |          0  (omitted)
        epop |    .079994   .0302316     2.65   0.008     .0207412    .1392467
       _cons |  -3.495252   4.364361    -0.80   0.423    -12.04924    5.058739
-------------+----------------------------------------------------------------
     sigma_u |  3.9200088
     sigma_e |  1.2617715
         rho |  .90612001   (fraction of variance due to u_i)
------------------------------------------------------------------------------

. xttest0

Breusch and Pagan Lagrangian multiplier test for random effects

        tincw[cntry,t] = Xb + u[cntry] + e[cntry,t]

        Estimated results:
                         |       Var     SD = sqrt(Var)
                ---------+-----------------------------
                   tincw |   40.23271       6.342926
                       e |   1.592067       1.261771
                       u |   15.36647       3.920009

        Test: Var(u) = 0
                             chibar2(01) =   346.47
                          Prob > chibar2 =   0.0000

Again, just in case, below i've copied my tests on fixed effects and the result of the community contributed robust Hausman test xtoverid by Mark E Schaffer & Steven Stillman (2010).

xtreg tincw dipa ddpa noren noown povri wgswe wgsst wgge wgphs rec hhgp hhep epad    epop    i.year,    fe

Fixed-effects (within) regression               Number of obs     =        220
Group variable: cntry                           Number of groups  =         21

R-squared:                                      Obs per group:
Within  = 0.6302                                         min =          6
Between = 0.5214                                         avg =       10.5
Overall = 0.4915                                         max =         11

F(24,175)         =      12.43
corr(u_i, Xb) = -0.0296                         Prob > F          =     0.0000


tincw  Coefficient  Std. err.      t    P>t     [95% conf. interval]

dipa    -.036817      .0186    -1.98   0.049    -.0735262   -.0001077
ddpa   -.0090984   .0200867    -0.45   0.651    -.0487419     .030545
noren   -1.370826   1.696216    -0.81   0.420    -4.718498    1.976847
noown    .4609595   1.879974     0.25   0.807     -3.24938    4.171299
povri    .7192726   .0563931    12.75   0.000     .6079744    .8305707
wgswe    .2981975   .2082641     1.43   0.154    -.1128351    .7092302
wgsst   -.4727493   .2299917    -2.06   0.041    -.9266638   -.0188347
wgge   -.0624191   .0514873    -1.21   0.227     -.164035    .0391968
wgphs    .1106438   .0526128     2.10   0.037     .0068066     .214481
rec    .0991194   .0580104     1.71   0.089    -.0153708    .2136095
hhgp    .2474524   .1686136     1.47   0.144    -.0853255    .5802303
hhep   -.0051598   .0614677    -0.08   0.933    -.1264733    .1161537
epad   -.0374935   .0334188    -1.12   0.263    -.1034493    .0284623
epop    .0650862   .0369807     1.76   0.080    -.0078995    .1380718

year 
2008    -.8931288   .6070067    -1.47   0.143    -2.091125    .3048672
2009    -.9229855   .5476195    -1.69   0.094    -2.003774    .1578032
2010    -.7910677   .6125346    -1.29   0.198    -1.999974    .4178383
2011     -1.57604   .9059493    -1.74   0.084    -3.364032    .2119533
2012    -1.766482    1.21608    -1.45   0.148    -4.166554     .633589
2013    -1.721908    1.17242    -1.47   0.144     -4.03581     .591995
2014    -1.839581   1.088597    -1.69   0.093    -3.988049    .3088872
2015    -1.838962   .8266178    -2.22   0.027    -3.470385   -.2075388
2016    -1.861555    .737822    -2.52   0.013    -3.317729   -.4053801
2017    -2.473105   .8276078    -2.99   0.003    -4.106482   -.8397274

_cons   -11.62313    5.80563    -2.00   0.047    -23.08119   -.1650633

sigma_u   4.6806048
sigma_e   1.2522541
rho   .93320298   (fraction of variance due to u_i)

F test that all u_i=0: F(20, 175) = 46.01                    Prob > F = 0.0000

. xttest3

Modified Wald test for groupwise heteroskedasticity
in fixed effect regression model

H0: sigma(i)^2 = sigma^2 for all i

chi2 (21)  =     372.36
Prob>chi2 =      0.0000

xtserial tincw dipa ddpa noren noown povri wgswe wgsst wgge wgphs rec hhgp hhep epad epop

Wooldridge test for autocorrelation in panel data
H0: no first order autocorrelation
    F(  1,      20) =     14.880
           Prob > F =      0.0010

quietly xtreg tincw dipa ddpa noren noown povri wgswe wgsst wgge wgphs rec hhgp hhep epad epop i.year, fe

xtcsd, pesaran abs
 
 
Pesaran's test of cross sectional independence =    -2.051, Pr = 0.0403
 
Average absolute value of the off-diagonal elements =     0.388

As last, i can conclude my data both has autocorrelation AR(1) and heteroskedasticity as last time.

. quietly xtreg tincw dipa ddpa noren noown povri wgswe wgsst wgge wgphs rec epad hhgp hhep epop, re

. xtoverid

Test of overidentifying restrictions: fixed vs random effects
Cross-section time-series model: xtreg re   
Sargan-Hansen statistic  14.219  Chi-sq(14)   P-value = 0.4335

As last, the results tells me to go with the random effects model.

Last we spoke about me using an xtreg with cluster() as explained by previously mentioned Daniel Hoechle to accommodate the complications shown in my data after the tests when looking at the fe model.
(2) Should i also apply this to the re model - this probably relates to my first question regarding more tests to run on random effects models.

On another note, i ran into some different youtube videos, which led me to the stata manual on xtgls (xtxtgls.pdf (stata.com)). As i am reading it, i could maybe perform an xtgls ylist xlist, panel(heteroskedastic).
(3) Do you think this could be a plausible solution, i get way better results using xtgls, but dont want to use it if it is methodologically wrong. I am a bit in doubt of this since i have no presence of cross-section correlation per the tests displayed above, but i read in the maunal that when applying panel(heteroskedastic), its not taking cross-section correlation into account.

Results from xtgls:

xtgls tincw dipa ddpa noren noown povri wgswe wgsst wgge wgphs rec epad hhgp hhep epop, panel(heteroskedastic)

Cross-sectional time-series FGLS regression

Coefficients:  generalized least squares
Panels:        heteroskedastic
Correlation:   no autocorrelation

Estimated covariances      =        21          Number of obs     =        220
Estimated autocorrelations =         0          Number of groups  =         21
Estimated coefficients     =        15          Obs per group:
                                                              min =          6
                                                              avg =   10.47619
                                                              max =         11
                                                Wald chi2(14)     =    1090.94
                                                Prob > chi2       =     0.0000

------------------------------------------------------------------------------
       tincw | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
        dipa |  -.0011577   .0184459    -0.06   0.950     -.037311    .0349956
        ddpa |  -.0112832   .0161919    -0.70   0.486    -.0430187    .0204523
       noren |   1.277538   1.115021     1.15   0.252    -.9078631     3.46294
       noown |  -5.726547   1.168278    -4.90   0.000    -8.016331   -3.436763
       povri |   .4950312   .0526464     9.40   0.000     .3918462    .5982163
       wgswe |   .2374913   .1230344     1.93   0.054    -.0036517    .4786343
       wgsst |  -.1309246   .1315063    -1.00   0.319    -.3886721     .126823
        wgge |  -.0093227   .0628512    -0.15   0.882    -.1325088    .1138634
       wgphs |  -.2632611   .0421939    -6.24   0.000    -.3459597   -.1805626
         rec |  -.1121734   .0179515    -6.25   0.000    -.1473577    -.076989
        epad |   -.033142   .0318757    -1.04   0.298    -.0956173    .0293333
        hhgp |   .4942622   .1229442     4.02   0.000     .2532961    .7352283
        hhep |   .0109311   .0455445     0.24   0.810    -.0783344    .1001966
        epop |   .0361175   .0344903     1.05   0.295    -.0314822    .1037172
       _cons |    22.8291   4.590343     4.97   0.000     13.83219    31.82601
------------------------------------------------------------------------------

And, to my final question - so sorry to spam you with all of this - i had to exclude 3 countries due to so much missing data in those 3 countries. I was thinking of still testing them seperately from this batch of countries, on the available variables. However, is 3 countries (52 obs in total) too little to work when using these kinds of models. I have had no significant results from the three so i thought it might be an issue of too little data?


Hope you can shed some light on some of my issues yet again!

BR Anne

[Carlo Lazzaro]

Anne:
1) as you have a short panel (N>T), stick with -xtreg, re vce(cluster idcode) and forget -xtgls-.
As you know, both -robust- and -vce(cluster clusterid)- standard error option take both heteroskedasticity and autocorrelation into account;
2) no need to search for other tests, as the outcome of the community-contributed module -xtoverid- points you clearly towards the -re- specification (BTW: with non-default standard errors, you can safey forget -hausman-);
3) I would not exclude myself the three countries you mention, because Stata applies listwise deletion by default (ie, they're automatically ruled out from your regression code).

[Anne Nissen]

Carlo:
Thank you yet again for your help - it works perfectly fine!
I had thought to include a dummy variable where the countries are grouped into four different groups according to climate zone. I have tried to add it with an i. and also tried to add it as my id in the xtset option after rearranging the data. The latter option gives me a sigma_u=0 and rho=0. I looked through the forum and saw some have had the same issue, I have thus checked my dataset to see if I had extra rows included from excel that would create missing variables. After having removed those I still have the same issue. Is it too few groups and would you recommend me just running it as with the i.? I am just a bit apprehensive about it, since it is a difficult thing to list in a perfect order, so I am worried it won't be meaningful. The issue with the meaningfulness would probably be an issue no matter which of the two options I chose, but i just thought i would run it by you, just in case you have some good inputs to which way i should chose to do this?

sum

    Variable |        Obs        Mean    Std. dev.       Min        Max
-------------+---------------------------------------------------------
        year |         44        2012    3.198837       2007       2017
          dc |         44         2.5     1.13096          1          4
       bincw |         44    19.29182     9.00865       6.08      34.21
       tincw |         44    10.61432    5.511808       3.13      20.28
        hhep |         44    15.21614    3.931078       6.24      20.33
-------------+---------------------------------------------------------
        hhgp |         44    7.050682     2.50455       4.31      12.26
        dipa |         44    27.81591    6.506897       14.9      36.45
        ddpa |         44    37.53977    7.601973      23.25       48.9
       noren |         44    1.432955    .2444793       1.04       1.82
       noown |         44    1.753864    .2694198        1.3        2.1
-------------+---------------------------------------------------------
       povri |         44    22.03932    4.202019      15.65      30.25
        epad |         44    102.5293    12.42466      80.76     124.02
        epel |         44    92.46114    10.98931      63.18     106.35
       eplfo |         44    106.8518    18.25087      78.88     140.83
        epng |         44    80.53068    22.17318      39.14     111.25
-------------+---------------------------------------------------------
        epop |         44    101.8343    11.23656      80.13      122.8
       eptot |         44    96.69932    9.654911      73.75     111.08
        epug |         44    100.8861    10.42345      79.22     119.31
       wgswe |         44    22.92364    4.176442         15      28.52
       wgsst |         44    16.89727    3.039919      13.34      23.05
-------------+---------------------------------------------------------
        wgge |         44    44.51341    5.524516      36.12       54.2
       wgphs |         44    69.95568    6.232086       64.4      81.21
         rec |         44    21.02886    13.79211       4.25      48.67

. xtset dc year

Panel variable: dc (strongly balanced)
 Time variable: year, 2007 to 2017
         Delta: 1 unit

xtreg tincw povri dipa ddpa noren noown wgswe wgge wgphs wgsst rec hhep hhgp epop epad, re vce(robust)

Random-effects GLS regression                   Number of obs     =         44
Group variable: dc                              Number of groups  =          4

R-squared:                                      Obs per group:
     Within  = 0.8242                                         min =         11
     Between = 1.0000                                         avg =       11.0
     Overall = 0.9917                                         max =         11

                                                Wald chi2(3)      =          .
corr(u_i, X) = 0 (assumed)                      Prob > chi2       =          .

                                     (Std. err. adjusted for 4 clusters in dc)
------------------------------------------------------------------------------
             |               Robust
       tincw | Coefficient  std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
       povri |   .7945952   .1738942     4.57   0.000     .4537688    1.135422
        dipa |  -.0095748   .0394767    -0.24   0.808    -.0869477     .067798
        ddpa |   .0044669   .0422884     0.11   0.916    -.0784168    .0873506
       noren |  -1.252093   5.567381    -0.22   0.822    -12.16396    9.659774
       noown |  -5.699108   5.256033    -1.08   0.278    -16.00074    4.602527
       wgswe |   -.146217   .1816426    -0.80   0.421      -.50223    .2097961
        wgge |   .2447543     .22556     1.09   0.278    -.1973352    .6868438
       wgphs |   .0110702   .3804342     0.03   0.977    -.7345671    .7567075
       wgsst |  -.4529311   .3681628    -1.23   0.219    -1.174517    .2686547
         rec |  -.0952029   .1446041    -0.66   0.510    -.3786216    .1882159
        hhep |   .1277079   .1482868     0.86   0.389    -.1629288    .4183447
        hhgp |   .4353815   .2975463     1.46   0.143    -.1477986    1.018562
        epop |   .0180998   .0848851     0.21   0.831    -.1482719    .1844714
        epad |  -.0220373   .0774462    -0.28   0.776    -.1738289    .1297544
       _cons |   1.731472   29.16066     0.06   0.953    -55.42238    58.88532
-------------+----------------------------------------------------------------
     sigma_u |          0
     sigma_e |  .58232569
         rho |          0   (fraction of variance due to u_i)
------------------------------------------------------------------------------

when including climate zone as i.dc:

xtreg tincw povri dipa ddpa noren noown wgswe wgge wgphs wgsst rec hhep hhgp epop epad i.dc, re vce(robust)

Random-effects GLS regression                   Number of obs     =        220
Group variable: cntry                           Number of groups  =         21

R-squared:                                      Obs per group:
     Within  = 0.6025                                         min =          6
     Between = 0.7360                                         avg =       10.5
     Overall = 0.6942                                         max =         11

                                                Wald chi2(17)     =    1413.72
corr(u_i, X) = 0 (assumed)                      Prob > chi2       =     0.0000

                                 (Std. err. adjusted for 21 clusters in cntry)
------------------------------------------------------------------------------
             |               Robust
       tincw | Coefficient  std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
       povri |   .7602694   .1169691     6.50   0.000     .5310141    .9895246
        dipa |  -.0422104   .0177573    -2.38   0.017    -.0770141   -.0074068
        ddpa |    .004894   .0166294     0.29   0.769    -.0276991     .037487
       noren |   -.985936    1.81294    -0.54   0.587    -4.539234    2.567361
       noown |  -1.568028   2.505599    -0.63   0.531    -6.478912    3.342855
       wgswe |   .1636614   .1562385     1.05   0.295    -.1425604    .4698832
        wgge |  -.0287188   .0394154    -0.73   0.466    -.1059715    .0485339
       wgphs |   .0772432    .054414     1.42   0.156    -.0294062    .1838927
       wgsst |   -.385553   .2478628    -1.56   0.120    -.8713552    .1002492
         rec |  -.0071723   .0379036    -0.19   0.850     -.081462    .0671175
        hhep |  -.0409066      .0716    -0.57   0.568      -.18124    .0994269
        hhgp |   .1787298   .1385794     1.29   0.197    -.0928809    .4503405
        epop |   .0809009   .0358393     2.26   0.024     .0106571    .1511447
        epad |  -.0628535   .0357465    -1.76   0.079    -.1329153    .0072083
             |
          dc |
          1  |  -4.178242   2.922705    -1.43   0.153     -9.90664    1.550155
          2  |  -4.408024   3.107393    -1.42   0.156     -10.4984    1.682355
          3  |  -7.329449   3.401392    -2.15   0.031    -13.99605   -.6628436
             |
       _cons |  -1.141493   7.007214    -0.16   0.871    -14.87538    12.59239
-------------+----------------------------------------------------------------
     sigma_u |  4.6499065
     sigma_e |  1.2617715
         rho |   .9314169   (fraction of variance due to u_i)
------------------------------------------------------------------------------

BR Anne

[Carlo Lazzaro]

Anne:
your second code is the way to go (I do not see how you -xtset- your dataset, though).
Clusters=21 might be a bit low, but you found both heteroskedsticity and autocorrelation so you should stick with 21 clusters.

[Anne Nissen]

Hi Carlo,

Again thank you for your answer.
I am experiencing some weird results on some of my variables, which to me and theoretically doesnt make any sense. My energy prices (index=2015) for automotive diesel and oil products are producing negative coefficients and are significant - which in my case would mean that if the price of automotive diesel increase by 1 then energy poverty decrease by 0.21, when we move one unit on country and year. This makes absolutely no sense to me, and I am quite puzzled by what could explain it.
Have you experienced weird prefixes before, and do you have an idea to what I can do about it?

Can it possibly be explained by the bias variance tradeoff when chosing a random effects instead of a fixed effects model? Or could it be caused by the one unit change in the country variable?

 
xtreg bincw dipa ddpa noren noown hhgp hhep epad epop epug wgge wgphs wgsst rec, re cluster()

Random-effects GLS regression                   Number of obs     =        209
Group variable: cntry                           Number of groups  =         20

R-squared:                                      Obs per group:
     Within  = 0.1957                                         min =          6
     Between = 0.7145                                         avg =       10.4
     Overall = 0.6331                                         max =         11

                                                Wald chi2(13)     =     100.08
corr(u_i, X) = 0 (assumed)                      Prob > chi2       =     0.0000

------------------------------------------------------------------------------
       bincw | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
        dipa |    -.11896   .0437347    -2.72   0.007    -.2046785   -.0332416
        ddpa |   .0269258   .0433534     0.62   0.535    -.0580454     .111897
       noren |   -1.45799   3.400739    -0.43   0.668    -8.123315    5.207335
       noown |  -8.542249   3.724434    -2.29   0.022    -15.84201   -1.242492
        hhgp |   .3165467   .3465723     0.91   0.361    -.3627226     .995816
        hhep |  -.0073165   .1588096    -0.05   0.963    -.3185776    .3039445
        epad |  -.2064444   .0724658    -2.85   0.004    -.3484749    -.064414
        epop |  -.2397357    .136564    -1.76   0.079    -.5073962    .0279248
        epug |    .611444   .1521178     4.02   0.000     .3132985    .9095894
        wgge |   .1650103     .11456     1.44   0.150    -.0595232    .3895438
       wgphs |  -.4320382    .108138    -4.00   0.000    -.6439849   -.2200916
       wgsst |  -.1977692   .3066784    -0.64   0.519    -.7988478    .4033094
         rec |  -.1316739   .0720012    -1.83   0.067    -.2727936    .0094458
       _cons |     51.801   8.780813     5.90   0.000     34.59092    69.01108
-------------+----------------------------------------------------------------
     sigma_u |  4.1444811
     sigma_e |  3.0344652
         rho |   .6510107   (fraction of variance due to u_i)
------------------------------------------------------------------------------

I am not sure if this can be caused by something methodological, which is why i wanted to run it by you.

BR Anne