very large coefficients in dif-in-dif

[Felix Chappuis]

Hi Statalist users

I run a study to get the effect of the minimum wage increases on the employment rate. Now I run a difference-in-difference analysis and got for some age category very large coefficients.
My method is that I built groups with individuals who have the same characteristics, for example are female and hold a master degree. For this groups I calculated the cap-measure.

I run the following code, to know the dif-in-dif coefficient of the groups containing only people who are 30-39 years old:

reg ln_unemployment_rate indicate avg i.numvar avg i.numvar##c.avg if anos2 == "03" , vce (cluster code)

numvar ist my time-variable, code is the group-id and anos2=="03" are the 30-39 year old people. avg is my gap measure. I tried to control for the interaction of the gap-measure with time, for the gap-measure itself and for time-fixed effects.
Did I do something wrong in coding or is this result realistic?
I would be very thankful if someone could help me, since I'm really a beginner.

Best

Felix


----------------------- copy starting from the next line -----------------------

* Example generated by -dataex-. For more info, type help dataex
clear
input float code int numvar float ln_unemployment_rate str2 anos2 byte anos2_n float avg_GAP_c str3 estu byte isdup str2 nuts1 byte nuts1_n str2 sexo byte sexo_n float(Anzahl_Betroffene post) long estu_n float(indicate Arbeitslosenrate all)
 8 222  -.6931472 "01" 1  .05259339 "2" 0 "2" 2 "1" 1  1 0 2          0        .5         2
 8 234  -.6931472 "01" 1  .05259339 "2" 0 "2" 2 "1" 1  1 1 2  .05259339        .5         2
 8 226 -.55961573 "01" 1  .05259339 "2" 0 "2" 2 "1" 1  1 0 2          0  .5714286      1.75
 8 230  -.4054651 "01" 1  .05259339 "2" 0 "2" 2 "1" 1  1 0 2          0  .6666667       1.5
 8 238  -.4054651 "01" 1  .05259339 "2" 0 "2" 2 "1" 1  1 1 2  .05259339  .6666667       1.5
 8 240 -.22314353 "01" 1  .05259339 "2" 0 "2" 2 "1" 1  1 1 2  .05259339        .8      1.25
 8 214 -.15415066 "01" 1  .05259339 "2" 0 "2" 2 "1" 1  1 0 2          0  .8571429 1.1666666
 8 218          0 "01" 1  .05259339 "2" 0 "2" 2 "1" 1  1 0 2          0         1         1
12 238 -.55961573 "01" 1  .04764649 "2" 0 "4" 4 "1" 1  2 1 2  .04764649  .5714286       3.5
12 226  -.5389965 "01" 1  .04764649 "2" 0 "4" 4 "1" 1  2 0 2          0  .5833333 3.4285715
12 240  -.4855078 "01" 1  .04764649 "2" 0 "4" 4 "1" 1  2 1 2  .04764649  .6153846      3.25
12 234   -.427444 "01" 1  .04764649 "2" 0 "4" 4 "1" 1  2 1 2  .04764649  .6521739 3.0666666
12 222  -.3254224 "01" 1  .04764649 "2" 0 "4" 4 "1" 1  2 0 2          0  .7222222  2.769231
12 214 -.27443686 "01" 1  .04764649 "2" 0 "4" 4 "1" 1  2 0 2          0       .76  2.631579
12 230 -.22314353 "01" 1  .04764649 "2" 0 "4" 4 "1" 1  2 0 2          0        .8       2.5
12 218 -.13353139 "01" 1  .04764649 "2" 0 "4" 4 "1" 1  2 0 2          0      .875 2.2857144
13 240  -.4054651 "01" 1   .3313497 "2" 0 "4" 4 "6" 6  1 1 2   .3313497  .6666667       1.5
13 238 -.22314353 "01" 1   .3313497 "2" 0 "4" 4 "6" 6  1 1 2   .3313497        .8      1.25
13 230 -.22314353 "01" 1   .3313497 "2" 0 "4" 4 "6" 6  1 0 2          0        .8      1.25
13 222  -.2006707 "01" 1   .3313497 "2" 0 "4" 4 "6" 6  1 0 2          0  .8181818 1.2222222
13 226  -.2006707 "01" 1   .3313497 "2" 0 "4" 4 "6" 6  1 0 2          0  .8181818 1.2222222
13 214 -.16705407 "01" 1   .3313497 "2" 0 "4" 4 "6" 6  1 0 2          0  .8461539 1.1818181
13 218          0 "01" 1   .3313497 "2" 0 "4" 4 "6" 6  1 0 2          0         1         1
13 234          0 "01" 1   .3313497 "2" 0 "4" 4 "6" 6  1 1 2   .3313497         1         1
14 226 -1.0116009 "01" 1  .30434015 "2" 0 "5" 5 "1" 1  6 0 2          0  .3636364      16.5
14 230 -.55961573 "01" 1  .30434015 "2" 0 "5" 5 "1" 1  6 0 2          0  .5714286      10.5
14 238 -.45198515 "01" 1  .30434015 "2" 0 "5" 5 "1" 1  6 1 2  .30434015  .6363636  9.428572
14 234  -.4054651 "01" 1  .30434015 "2" 0 "5" 5 "1" 1  6 1 2  .30434015  .6666667         9
14 218  -.3101549 "01" 1  .30434015 "2" 0 "5" 5 "1" 1  6 0 2          0  .7333333  8.181818
14 240  -.2876821 "01" 1  .30434015 "2" 0 "5" 5 "1" 1  6 1 2  .30434015       .75         8
14 222 -.24116208 "01" 1  .30434015 "2" 0 "5" 5 "1" 1  6 0 2          0  .7857143  7.636364
14 214 -.21622308 "01" 1  .30434015 "2" 0 "5" 5 "1" 1  6 0 2          0  .8055556  7.448276
15 226  -.6931472 "01" 1   .1741187 "2" 0 "5" 5 "6" 6  4 0 2          0        .5         8
15 238  -.2876821 "01" 1   .1741187 "2" 0 "5" 5 "6" 6  4 1 2   .1741187       .75  5.333333
15 214 -.25782913 "01" 1   .1741187 "2" 0 "5" 5 "6" 6  4 0 2          0  .7727273  5.176471
15 234  -.1823216 "01" 1   .1741187 "2" 0 "5" 5 "6" 6  4 1 2   .1741187  .8333333       4.8
15 218          0 "01" 1   .1741187 "2" 0 "5" 5 "6" 6  4 0 2          0         1         4
15 240          0 "01" 1   .1741187 "2" 0 "5" 5 "6" 6  4 1 2   .1741187         1         4
15 222          0 "01" 1   .1741187 "2" 0 "5" 5 "6" 6  4 0 2          0         1         4
15 230          0 "01" 1   .1741187 "2" 0 "5" 5 "6" 6  4 0 2          0         1         4
16 240  -.6931472 "01" 1  .06941935 "2" 0 "6" 6 "1" 1  1 1 2  .06941935        .5         2
16 238  -.6190392 "01" 1  .06941935 "2" 0 "6" 6 "1" 1  1 1 2  .06941935 .53846157 1.8571428
16 222  -.5389965 "01" 1  .06941935 "2" 0 "6" 6 "1" 1  1 0 2          0  .5833333 1.7142857
16 230  -.3254224 "01" 1  .06941935 "2" 0 "6" 6 "1" 1  1 0 2          0  .7222222 1.3846154
16 226  -.2876821 "01" 1  .06941935 "2" 0 "6" 6 "1" 1  1 0 2          0       .75 1.3333334
16 218  -.2876821 "01" 1  .06941935 "2" 0 "6" 6 "1" 1  1 0 2          0       .75 1.3333334
16 214 -.14842002 "01" 1  .06941935 "2" 0 "6" 6 "1" 1  1 0 2          0   .862069      1.16
16 234 -.14310083 "01" 1  .06941935 "2" 0 "6" 6 "1" 1  1 1 2  .06941935  .8666667 1.1538461
17 230  -.4054651 "01" 1  .12124364 "2" 0 "7" 7 "1" 1  1 0 2          0  .6666667       1.5
17 214  -.3364722 "01" 1  .12124364 "2" 0 "7" 7 "1" 1  1 0 2          0  .7142857       1.4
17 222  -.2876821 "01" 1  .12124364 "2" 0 "7" 7 "1" 1  1 0 2          0       .75 1.3333334
17 218  -.2876821 "01" 1  .12124364 "2" 0 "7" 7 "1" 1  1 0 2          0       .75 1.3333334
17 226          0 "01" 1  .12124364 "2" 0 "7" 7 "1" 1  1 0 2          0         1         1
17 240          0 "01" 1  .12124364 "2" 0 "7" 7 "1" 1  1 1 2  .12124364         1         1
17 234          0 "01" 1  .12124364 "2" 0 "7" 7 "1" 1  1 1 2  .12124364         1         1
17 238          0 "01" 1  .12124364 "2" 0 "7" 7 "1" 1  1 1 2  .12124364         1         1
19 238  -1.446919 "01" 1  1.0230924 "3" 0 "1" 1 "1" 1  1 1 3  1.0230924  .2352941      4.25
19 234 -1.2685113 "01" 1  1.0230924 "3" 0 "1" 1 "1" 1  1 1 3  1.0230924    .28125 3.5555556
19 240 -1.1394343 "01" 1  1.0230924 "3" 0 "1" 1 "1" 1  1 1 3  1.0230924       .32     3.125
19 222  -.7672552 "01" 1  1.0230924 "3" 0 "1" 1 "1" 1  1 0 3          0  .4642857 2.1538463
19 230  -.7339692 "01" 1  1.0230924 "3" 0 "1" 1 "1" 1  1 0 3          0       .48 2.0833335
19 214 -.56798404 "01" 1  1.0230924 "3" 0 "1" 1 "1" 1  1 0 3          0 .56666666  1.764706
19 218  -.4480247 "01" 1  1.0230924 "3" 0 "1" 1 "1" 1  1 0 3          0  .6388889 1.5652174
19 226   -.427444 "01" 1  1.0230924 "3" 0 "1" 1 "1" 1  1 0 3          0  .6521739 1.5333333
20 238 -1.0647107 "01" 1  .09704527 "3" 0 "1" 1 "6" 6  1 1 3  .09704527  .3448276       2.9
20 226  -.9808292 "01" 1  .09704527 "3" 0 "1" 1 "6" 6  1 0 3          0      .375  2.666667
20 240  -.8472978 "01" 1  .09704527 "3" 0 "1" 1 "6" 6  1 1 3  .09704527  .4285714 2.3333333
20 234  -.6931472 "01" 1  .09704527 "3" 0 "1" 1 "6" 6  1 1 3  .09704527        .5         2
20 222  -.5260931 "01" 1  .09704527 "3" 0 "1" 1 "6" 6  1 0 3          0 .59090906 1.6923077
20 214  -.4964369 "01" 1  .09704527 "3" 0 "1" 1 "6" 6  1 0 3          0  .6086956 1.6428572
20 230  -.4924765 "01" 1  .09704527 "3" 0 "1" 1 "6" 6  1 0 3          0  .6111111 1.6363636
20 218 -.23361483 "01" 1  .09704527 "3" 0 "1" 1 "6" 6  1 0 3          0  .7916667  1.263158
21 240 -1.4271164 "01" 1 .036009464 "3" 0 "2" 2 "1" 1  4 1 3 .036009464       .24 16.666668
21 234  -1.060872 "01" 1 .036009464 "3" 0 "2" 2 "1" 1  4 1 3 .036009464  .3461539 11.555555
21 238  -.9808292 "01" 1 .036009464 "3" 0 "2" 2 "1" 1  4 1 3 .036009464      .375 10.666667
21 226  -.7827593 "01" 1 .036009464 "3" 0 "2" 2 "1" 1  4 0 3          0  .4571429      8.75
21 230    -.76214 "01" 1 .036009464 "3" 0 "2" 2 "1" 1  4 0 3          0  .4666667  8.571428
21 218 -.56798404 "01" 1 .036009464 "3" 0 "2" 2 "1" 1  4 0 3          0 .56666666  7.058824
21 214  -.5520686 "01" 1 .036009464 "3" 0 "2" 2 "1" 1  4 0 3          0 .57575756  6.947369
21 222  -.4855078 "01" 1 .036009464 "3" 0 "2" 2 "1" 1  4 0 3          0  .6153846       6.5
22 240 -1.0986123 "01" 1          0 "3" 0 "2" 2 "6" 6  0 1 3          0  .3333333         0
22 234  -.9555115 "01" 1          0 "3" 0 "2" 2 "6" 6  0 1 3          0  .3846154         0
22 226   -.737599 "01" 1          0 "3" 0 "2" 2 "6" 6  0 0 3          0  .4782609         0
22 238  -.4924765 "01" 1          0 "3" 0 "2" 2 "6" 6  0 1 3          0  .6111111         0
22 222   -.479573 "01" 1          0 "3" 0 "2" 2 "6" 6  0 0 3          0  .6190476         0
22 218  -.4700036 "01" 1          0 "3" 0 "2" 2 "6" 6  0 0 3          0      .625         0
22 214  -.3022809 "01" 1          0 "3" 0 "2" 2 "6" 6  0 0 3          0  .7391304         0
22 230  -.2719337 "01" 1          0 "3" 0 "2" 2 "6" 6  0 0 3          0  .7619048         0
23 218  -1.299283 "01" 1   .9331547 "3" 0 "3" 3 "1" 1 13 0 3          0 .27272728  47.66666
23 238  -.6931472 "01" 1   .9331547 "3" 0 "3" 3 "1" 1 13 1 3   .9331547        .5        26
23 240  -.6061358 "01" 1   .9331547 "3" 0 "3" 3 "1" 1 13 1 3   .9331547 .54545456  23.83333
23 234   -.597837 "01" 1   .9331547 "3" 0 "3" 3 "1" 1 13 1 3   .9331547       .55 23.636364
23 230 -.55961573 "01" 1   .9331547 "3" 0 "3" 3 "1" 1 13 0 3          0  .5714286     22.75
23 226 -.45198515 "01" 1   .9331547 "3" 0 "3" 3 "1" 1 13 0 3          0  .6363636  20.42857
23 222  -.3483067 "01" 1   .9331547 "3" 0 "3" 3 "1" 1 13 0 3          0  .7058824 18.416666
23 214  -.2876821 "01" 1   .9331547 "3" 0 "3" 3 "1" 1 13 0 3          0       .75 17.333334
24 226  -.8472978 "01" 1  .05983265 "3" 0 "3" 3 "6" 6  1 0 3          0  .4285714 2.3333333
24 240  -.8109302 "01" 1  .05983265 "3" 0 "3" 3 "6" 6  1 1 3  .05983265 .44444445      2.25
24 230  -.7731899 "01" 1  .05983265 "3" 0 "3" 3 "6" 6  1 0 3          0  .4615385 2.1666667
24 238  -.5108256 "01" 1  .05983265 "3" 0 "3" 3 "6" 6  1 1 3  .05983265        .6 1.6666666
end
format %tq numvar
label values estu_n estu_n
label def estu_n 2 "2", modify
label def estu_n 3 "3", modify

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

[Clyde Schechter]

I think for specific advice, more information is needed. First, show the actual output you got from Stata and explain which coefficients you think are too large and why. And also, show a different set of example data that matches the problem concerned. The example you posted has no observations with anos2 == "03". Show an example that will actually run the regression command that you are worried about--the problem is more likely to be with the data than with the code.

[Felix Chappuis]

Hi Clyde
Thank you very much for answering. Here is the stata output.

. asdoc reg Arbeitslosenrate_ln indicate avg i.numvar avg i.numvar##c.avg if anos2 == "03" , nest vce (cluster code) add(Time-Fixed-Effects, YES, Interaction GAP-Measure and Time, YES) cnames(30-39)
(File Myfile.doc already exists, option append was assumed)
note: avg_GAP omitted because of collinearity.
note: avg_GAP omitted because of collinearity.
note: 240.numvar#c.avg_GAP omitted because of collinearity.

Linear regression Number of obs = 312
F(15, 38) = 26.25
Prob > F = 0.0000
R-squared = 0.5522
Root MSE = .35741

(Std. err. adjusted for 39 clusters in code)
----------------------------------------------------------------------------------
| Robust
Arbeitslosenra~n | Coefficient std. err. t P>|t| [95% conf. interval]
-----------------+----------------------------------------------------------------
indicate | 15.36871 4.773936 3.22 0.003 5.704383 25.03304
avg_GAP | 12.83401 2.190158 5.86 0.000 8.400269 17.26776
|
numvar |
218 | -.2112056 .0489717 -4.31 0.000 -.3103436 -.1120676
222 | -.3640937 .0486977 -7.48 0.000 -.4626771 -.2655103
226 | -.4818032 .0685638 -7.03 0.000 -.6206034 -.343003
230 | -.6983471 .0562264 -12.42 0.000 -.8121715 -.5845226
234 | -.8680102 .0811457 -10.70 0.000 -1.032281 -.7037393
238 | -.9731426 .082945 -11.73 0.000 -1.141056 -.8052291
240 | -.8549269 .0896587 -9.54 0.000 -1.036431 -.6734224
|
avg_GAP | 0 (omitted)
avg_GAP | 0 (omitted)
|
numvar#c.avg_GAP |
218 | 6.654765 1.871522 3.56 0.001 2.866067 10.44346
222 | 6.140529 1.725625 3.56 0.001 2.647183 9.633874
226 | 8.943345 3.498238 2.56 0.015 1.861533 16.02516
230 | 11.92668 2.249964 5.30 0.000 7.371866 16.48149
234 | -1.572417 2.891235 -0.54 0.590 -7.425416 4.280581
238 | 3.160158 2.577931 1.23 0.228 -2.05859 8.378905
240 | 0 (omitted)
|
_cons | -1.522831 .0614625 -24.78 0.000 -1.647255 -1.398406
----------------------------------------------------------------------------------

And here is an adapted example which contains anos2=="03".

----------------------- copy starting from the next line -----------------------

* Example generated by -dataex-. For more info, type help dataex
clear
input float code int numvar float(Arbeitslosenrate_ln GAP_ct) str2 anos2 byte anos2_n float avg_GAP str3 estu float gap_ct byte isdup str2 nuts1 byte nuts1_n str2 sexo byte sexo_n float(Anzahl_Betroffene post) long estu_n float(indicate Arbeitslosenrate)
70 230 -1.4053426 .012595055 "03" 3 .012595055 "2" .012595055 0 "4" 4 "1" 1  33 0 2          0   .245283
70 240  -1.293921 .012595055 "03" 3 .012595055 "2" .012595055 0 "4" 4 "1" 1  33 1 2 .012595055 .27419356
70 234 -1.2685113 .012595055 "03" 3 .012595055 "2" .012595055 0 "4" 4 "1" 1  33 1 2 .012595055    .28125
70 226 -1.2321438 .012595055 "03" 3 .012595055 "2" .012595055 0 "4" 4 "1" 1  33 0 2          0 .29166666
70 214 -1.2039728 .012595055 "03" 3 .012595055 "2" .012595055 0 "4" 4 "1" 1  33 0 2          0        .3
70 238 -1.1249295 .012595055 "03" 3 .012595055 "2" .012595055 0 "4" 4 "1" 1  33 1 2 .012595055 .32467535
70 222 -1.0473189 .012595055 "03" 3 .012595055 "2" .012595055 0 "4" 4 "1" 1  33 0 2          0  .3508772
70 218  -.8001193 .012595055 "03" 3 .012595055 "2" .012595055 0 "4" 4 "1" 1  33 0 2          0  .4492754
74 240  -1.740466 .012664276 "03" 3 .012664276 "2" .012664276 0 "6" 6 "1" 1  50 1 2 .012664276  .1754386
74 238  -1.323774 .012664276 "03" 3 .012664276 "2" .012664276 0 "6" 6 "1" 1  50 1 2 .012664276 .26612905
74 234  -1.044545 .012664276 "03" 3 .012664276 "2" .012664276 0 "6" 6 "1" 1  50 1 2 .012664276 .35185185
74 226 -1.0352427 .012664276 "03" 3 .012664276 "2" .012664276 0 "6" 6 "1" 1  50 0 2          0  .3551402
74 230  -.9777758 .012664276 "03" 3 .012664276 "2" .012664276 0 "6" 6 "1" 1  50 0 2          0  .3761468
74 222  -.7331525 .012664276 "03" 3 .012664276 "2" .012664276 0 "6" 6 "1" 1  50 0 2          0  .4803922
74 218  -.6526558 .012664276 "03" 3 .012664276 "2" .012664276 0 "6" 6 "1" 1  50 0 2          0  .5206612
74 214  -.5187938 .012664276 "03" 3 .012664276 "2" .012664276 0 "6" 6 "1" 1  50 0 2          0  .5952381
75 240  -.9162907 .072123796 "03" 3 .072123796 "2" .072123796 0 "6" 6 "6" 6  67 1 2 .072123796        .4
75 226  -.8700779 .072123796 "03" 3 .072123796 "2" .072123796 0 "6" 6 "6" 6  67 0 2          0  .4189189
75 234  -.8517522 .072123796 "03" 3 .072123796 "2" .072123796 0 "6" 6 "6" 6  67 1 2 .072123796  .4266667
75 222  -.7295148 .072123796 "03" 3 .072123796 "2" .072123796 0 "6" 6 "6" 6  67 0 2          0  .4821429
75 214  -.6931472 .072123796 "03" 3 .072123796 "2" .072123796 0 "6" 6 "6" 6  67 0 2          0        .5
75 230  -.6778796 .072123796 "03" 3 .072123796 "2" .072123796 0 "6" 6 "6" 6  67 0 2          0 .50769234
75 238  -.6641596 .072123796 "03" 3 .072123796 "2" .072123796 0 "6" 6 "6" 6  67 1 2 .072123796  .5147059
75 218  -.5250103 .072123796 "03" 3 .072123796 "2" .072123796 0 "6" 6 "6" 6  67 0 2          0  .5915493
76 234 -2.1690538 .009476304 "03" 3 .009476304 "3" .009476304 0 "1" 1 "1" 1  42 1 3 .009476304  .1142857
76 238  -2.095273 .009476304 "03" 3 .009476304 "3" .009476304 0 "1" 1 "1" 1  42 1 3 .009476304 .12303665
76 240 -2.0736105 .009476304 "03" 3 .009476304 "3" .009476304 0 "1" 1 "1" 1  42 1 3 .009476304   .125731
76 230  -1.915812 .009476304 "03" 3 .009476304 "3" .009476304 0 "1" 1 "1" 1  42 0 3          0 .14722222
76 222 -1.6206112 .009476304 "03" 3 .009476304 "3" .009476304 0 "1" 1 "1" 1  42 0 3          0  .1977778
76 226 -1.4900912 .009476304 "03" 3 .009476304 "3" .009476304 0 "1" 1 "1" 1  42 0 3          0  .2253521
76 218  -1.369487 .009476304 "03" 3 .009476304 "3" .009476304 0 "1" 1 "1" 1  42 0 3          0  .2542373
76 214 -1.2174904 .009476304 "03" 3 .009476304 "3" .009476304 0 "1" 1 "1" 1  42 0 3          0   .295972
77 234   -1.56181  .03839993 "03" 3  .03839993 "3"  .03839993 0 "1" 1 "6" 6  94 1 3  .03839993  .2097561
77 240 -1.5017105  .03839993 "03" 3  .03839993 "3"  .03839993 0 "1" 1 "6" 6  94 1 3  .03839993  .2227488
77 230  -1.437828  .03839993 "03" 3  .03839993 "3"  .03839993 0 "1" 1 "6" 6  94 0 3          0 .23744294
77 238 -1.4271164  .03839993 "03" 3  .03839993 "3"  .03839993 0 "1" 1 "6" 6  94 1 3  .03839993       .24
77 226  -1.372811  .03839993 "03" 3  .03839993 "3"  .03839993 0 "1" 1 "6" 6  94 0 3          0 .25339368
77 222  -1.335742  .03839993 "03" 3  .03839993 "3"  .03839993 0 "1" 1 "6" 6  94 0 3          0 .26296297
77 218 -1.2335316  .03839993 "03" 3  .03839993 "3"  .03839993 0 "1" 1 "6" 6  94 0 3          0 .29126215
77 214  -1.227926  .03839993 "03" 3  .03839993 "3"  .03839993 0 "1" 1 "6" 6  94 0 3          0 .29289943
78 234   -2.61496 .006512205 "03" 3 .006512205 "3" .006512205 0 "2" 2 "1" 1  31 1 3 .006512205 .07317074
78 230 -2.4107985 .006512205 "03" 3 .006512205 "3" .006512205 0 "2" 2 "1" 1  31 0 3          0  .0897436
78 238  -2.410216 .006512205 "03" 3 .006512205 "3" .006512205 0 "2" 2 "1" 1  31 1 3 .006512205  .0897959
78 240 -2.1113348 .006512205 "03" 3 .006512205 "3" .006512205 0 "2" 2 "1" 1  31 1 3 .006512205 .12107625
78 226 -2.0249534 .006512205 "03" 3 .006512205 "3" .006512205 0 "2" 2 "1" 1  31 0 3          0      .132
78 222 -1.8021222 .006512205 "03" 3 .006512205 "3" .006512205 0 "2" 2 "1" 1  31 0 3          0 .16494846
78 218    -1.7492 .006512205 "03" 3 .006512205 "3" .006512205 0 "2" 2 "1" 1  31 0 3          0 .17391305
78 214 -1.6518586 .006512205 "03" 3 .006512205 "3" .006512205 0 "2" 2 "1" 1  31 0 3          0  .1916933
79 238  -1.683546   .0200352 "03" 3   .0200352 "3"   .0200352 0 "2" 2 "6" 6  49 1 3   .0200352  .1857143
79 222 -1.5612358   .0200352 "03" 3   .0200352 "3"   .0200352 0 "2" 2 "6" 6  49 0 3          0 .20987655
79 214 -1.5170646   .0200352 "03" 3   .0200352 "3"   .0200352 0 "2" 2 "6" 6  49 0 3          0 .21935484
79 240  -1.491655   .0200352 "03" 3   .0200352 "3"   .0200352 0 "2" 2 "6" 6  49 1 3   .0200352      .225
79 230 -1.4593195   .0200352 "03" 3   .0200352 "3"   .0200352 0 "2" 2 "6" 6  49 0 3          0 .23239437
79 234  -1.445135   .0200352 "03" 3   .0200352 "3"   .0200352 0 "2" 2 "6" 6  49 1 3   .0200352  .2357143
79 226 -1.3796936   .0200352 "03" 3   .0200352 "3"   .0200352 0 "2" 2 "6" 6  49 0 3          0 .25165564
79 218 -1.1769441   .0200352 "03" 3   .0200352 "3"   .0200352 0 "2" 2 "6" 6  49 0 3          0  .3082192
80 238 -2.3125355 .011943158 "03" 3 .011943158 "3" .011943158 0 "3" 3 "1" 1  43 1 3 .011943158  .0990099
80 234 -2.1535494 .011943158 "03" 3 .011943158 "3" .011943158 0 "3" 3 "1" 1  43 1 3 .011943158 .11607144
80 240 -2.1058748 .011943158 "03" 3 .011943158 "3" .011943158 0 "3" 3 "1" 1  43 1 3 .011943158 .12173913
80 230 -2.0522907 .011943158 "03" 3 .011943158 "3" .011943158 0 "3" 3 "1" 1  43 0 3          0 .12844035
80 226 -1.8430527 .011943158 "03" 3 .011943158 "3" .011943158 0 "3" 3 "1" 1  43 0 3          0 .15833333
80 222 -1.6711315 .011943158 "03" 3 .011943158 "3" .011943158 0 "3" 3 "1" 1  43 0 3          0  .1880342
80 214 -1.3862944 .011943158 "03" 3 .011943158 "3" .011943158 0 "3" 3 "1" 1  43 0 3          0       .25
80 218 -1.1267833 .011943158 "03" 3 .011943158 "3" .011943158 0 "3" 3 "1" 1  43 0 3          0  .3240741
81 238   -2.00148 .016390441 "03" 3 .016390441 "3" .016390441 0 "3" 3 "6" 6  56 1 3 .016390441 .13513511
81 240 -1.9715526 .016390441 "03" 3 .016390441 "3" .016390441 0 "3" 3 "6" 6  56 1 3 .016390441  .1392405
81 226 -1.7107904 .016390441 "03" 3 .016390441 "3" .016390441 0 "3" 3 "6" 6  56 0 3          0  .1807229
81 230  -1.686399 .016390441 "03" 3 .016390441 "3" .016390441 0 "3" 3 "6" 6  56 0 3          0  .1851852
81 222 -1.4294665 .016390441 "03" 3 .016390441 "3" .016390441 0 "3" 3 "6" 6  56 0 3          0 .23943663
81 218 -1.3328056 .016390441 "03" 3 .016390441 "3" .016390441 0 "3" 3 "6" 6  56 0 3          0 .26373628
81 234  -1.332227 .016390441 "03" 3 .016390441 "3" .016390441 0 "3" 3 "6" 6  56 1 3 .016390441  .2638889
81 214 -1.0593916 .016390441 "03" 3 .016390441 "3" .016390441 0 "3" 3 "6" 6  56 0 3          0  .3466667
82 238 -2.1077263 .011992987 "03" 3 .011992987 "3" .011992987 0 "4" 4 "1" 1  68 1 3 .011992987 .12151393
82 230  -2.036882 .011992987 "03" 3 .011992987 "3" .011992987 0 "4" 4 "1" 1  68 0 3          0 .13043478
82 240  -1.980471 .011992987 "03" 3 .011992987 "3" .011992987 0 "4" 4 "1" 1  68 1 3 .011992987 .13800424
82 234 -1.9530276 .011992987 "03" 3 .011992987 "3" .011992987 0 "4" 4 "1" 1  68 1 3 .011992987 .14184397
82 226  -1.770074 .011992987 "03" 3 .011992987 "3" .011992987 0 "4" 4 "1" 1  68 0 3          0  .1703204
82 222  -1.625967 .011992987 "03" 3 .011992987 "3" .011992987 0 "4" 4 "1" 1  68 0 3          0 .19672133
82 218  -1.309177 .011992987 "03" 3 .011992987 "3" .011992987 0 "4" 4 "1" 1  68 0 3          0 .27004218
82 214 -1.2997397 .011992987 "03" 3 .011992987 "3" .011992987 0 "4" 4 "1" 1  68 0 3          0 .27260274
83 234 -1.5799155  .02804787 "03" 3  .02804787 "3"  .02804787 0 "4" 4 "6" 6 102 1 3  .02804787  .2059925
83 238   -1.57372  .02804787 "03" 3  .02804787 "3"  .02804787 0 "4" 4 "6" 6 102 1 3  .02804787 .20727272
83 240  -1.418043  .02804787 "03" 3  .02804787 "3"  .02804787 0 "4" 4 "6" 6 102 1 3  .02804787  .2421875
83 230 -1.3926035  .02804787 "03" 3  .02804787 "3"  .02804787 0 "4" 4 "6" 6 102 0 3          0 .24842767
83 226 -1.3049487  .02804787 "03" 3  .02804787 "3"  .02804787 0 "4" 4 "6" 6 102 0 3          0 .27118644
83 222 -1.1279399  .02804787 "03" 3  .02804787 "3"  .02804787 0 "4" 4 "6" 6 102 0 3          0  .3236994
83 214  -1.063273  .02804787 "03" 3  .02804787 "3"  .02804787 0 "4" 4 "6" 6 102 0 3          0  .3453237
83 218  -.9852836  .02804787 "03" 3  .02804787 "3"  .02804787 0 "4" 4 "6" 6 102 0 3          0  .3733333
84 238 -2.1788418 .009505322 "03" 3 .009505322 "3" .009505322 0 "5" 5 "1" 1 109 1 3 .009505322 .11317253
84 234  -2.151088 .009505322 "03" 3 .009505322 "3" .009505322 0 "5" 5 "1" 1 109 1 3 .009505322  .1163575
84 240 -2.1139276 .009505322 "03" 3 .009505322 "3" .009505322 0 "5" 5 "1" 1 109 1 3 .009505322 .12076273
84 230 -1.9074117 .009505322 "03" 3 .009505322 "3" .009505322 0 "5" 5 "1" 1 109 0 3          0 .14846416
84 226 -1.8442775 .009505322 "03" 3 .009505322 "3" .009505322 0 "5" 5 "1" 1 109 0 3          0 .15813954
84 222  -1.613808 .009505322 "03" 3 .009505322 "3" .009505322 0 "5" 5 "1" 1 109 0 3          0  .1991279
84 218 -1.4056587 .009505322 "03" 3 .009505322 "3" .009505322 0 "5" 5 "1" 1 109 0 3          0 .24520548
84 214 -1.3392884 .009505322 "03" 3 .009505322 "3" .009505322 0 "5" 5 "1" 1 109 0 3          0 .26203206
85 238 -1.8645188 .021509806 "03" 3 .021509806 "3" .021509806 0 "5" 5 "6" 6 137 1 3 .021509806 .15497077
85 240 -1.6897603 .021509806 "03" 3 .021509806 "3" .021509806 0 "5" 5 "6" 6 137 1 3 .021509806 .18456376
85 234  -1.533378 .021509806 "03" 3 .021509806 "3" .021509806 0 "5" 5 "6" 6 137 1 3 .021509806 .21580547
85 230 -1.4999537 .021509806 "03" 3 .021509806 "3" .021509806 0 "5" 5 "6" 6 137 0 3          0  .2231405
end
format %tq numvar
label values estu_n estu_n
label def estu_n 2 "2", modify
label def estu_n 3 "3", modify

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

Sorry that I included such a wrong example. Hope someone can help me. My result would mean that in the 10 % most exposed groups to the minimum wage relative to the 10 % least exposed to the minimum wage, would increase very much. I think that there may be a difference, but not so much.

[Clyde Schechter]

Well, it looks to me like you are trying to make a complicated model out of data that is inadequate to support it. At least in your example data, both the variable indicate and the variable avg_GAP are more or less proxies for intervals of the time variable numvar:

           |                                         numvar
   avg_GAP |    2013q3     2014q3     2015q3     2016q3     2017q3     2018q3     2019q3     2020q1 |     Total
-----------+----------------------------------------------------------------------------------------+----------
  .0065122 |         1          1          1          1          1          1          1          1 |         8
  .0094763 |         1          1          1          1          1          1          1          1 |         8
  .0095053 |         1          1          1          1          1          1          1          1 |         8
  .0119432 |         1          1          1          1          1          1          1          1 |         8
   .011993 |         1          1          1          1          1          1          1          1 |         8
  .0125951 |         1          1          1          1          1          1          1          1 |         8
  .0126643 |         1          1          1          1          1          1          1          1 |         8
  .0163904 |         1          1          1          1          1          1          1          1 |         8
  .0200352 |         1          1          1          1          1          1          1          1 |         8
  .0215098 |         0          0          0          0          1          1          1          1 |         4
  .0280479 |         1          1          1          1          1          1          1          1 |         8
  .0383999 |         1          1          1          1          1          1          1          1 |         8
  .0721238 |         1          1          1          1          1          1          1          1 |         8
-----------+----------------------------------------------------------------------------------------+----------
     Total |        12         12         12         12         13         13         13         13 |       100

. tab indicate numvar

           |                                         numvar
  indicate |    2013q3     2014q3     2015q3     2016q3     2017q3     2018q3     2019q3     2020q1 |     Total
-----------+----------------------------------------------------------------------------------------+----------
         0 |        12         12         12         12         13          0          0          0 |        61
  .0065122 |         0          0          0          0          0          1          1          1 |         3
  .0094763 |         0          0          0          0          0          1          1          1 |         3
  .0095053 |         0          0          0          0          0          1          1          1 |         3
  .0119432 |         0          0          0          0          0          1          1          1 |         3
   .011993 |         0          0          0          0          0          1          1          1 |         3
  .0125951 |         0          0          0          0          0          1          1          1 |         3
  .0126643 |         0          0          0          0          0          1          1          1 |         3
  .0163904 |         0          0          0          0          0          1          1          1 |         3
  .0200352 |         0          0          0          0          0          1          1          1 |         3
  .0215098 |         0          0          0          0          0          1          1          1 |         3
  .0280479 |         0          0          0          0          0          1          1          1 |         3
  .0383999 |         0          0          0          0          0          1          1          1 |         3
  .0721238 |         0          0          0          0          0          1          1          1 |         3
-----------+----------------------------------------------------------------------------------------+----------
     Total |        12         12         12         12         13         13         13         13 |       100

In fact, avg_GAP and numvar jointly identify unique observations in the data. So the interaction i.numvar#c.avg_GAP which is the key term in your regression model, uniquely identifies observations. The only reason you aren't getting R² = 1 in your regression is that the relationship between your outcome and avg_GAP is not exactly linear. Anyway, with such complete explanation, it is not surprising that something in your regression model will end up having a huge effect attributed to it.

That said, backing away from complicated models and just looking at a scatterplot of Arbeitslosenrate_ln against avg_GAP does show a very strong relationship, where the difference between the highest and lowest values of Arbeitslosenrate_ln is 3.13, which corresponds to a roughly 23 fold ratio of values of Arbeitslosenrate itself. And in this simple analysis, avg_GAP does seem to be associated with much of that variation.

[Felix Chappuis]

Thank you very much Clyde, soo happy that you help me. Now I would like to construct a graph with numvar on the x-achsis and the coefficients of avg_GAP on the Y-Achsis. The reference-time should be at 2019, where the minimum wage increasing took place. And I would like to have on every dot a 95% confidence interval. But how can I constract that? I would be so thankful if you could help me. It should look like that.

[Felix Chappuis]

[Clyde Schechter]

By coefficients of avg_GAP I take it you mean the marginal effects of avg_GAP in each year. You won't be able to get that from your current model because of the colinearities that exist among numvar (time) and indicate and avg_GAP. You will have to simplify your model to eliminate those.

regress Arbeitslosenrate_ln i.numvar##c.avg_GAP, cluster(code)

margins numvar, dydx(avg_GAP)
marginsplot, xdimension(numvar)

-marginsplot- accepts most -graph twoway- options, so you can customize the appearance of the graph to your preferences.

Note: You can add more covariates to the -regress- command, but only if they are not proxies for time, nor colinear with c.avg_GAP. The variable indicator, in particular, is not permissible here.

[Felix Chappuis]

Thank you very much. Yes the graph looks great, only one thing, shouldn't there be a confidence interval of zero at the reference time 2019? Is there also a possibility to show in the time trend how it would be without the minimum wage increases in 2019? Thank you very much.

[Clyde Schechter]

Why do you think there should be a zero-width confidence interval in 2019? I can't think of a reason.

I don't understand your proposed counterfactual. That may just be because I don't understand what the variables mean.

[Felix Chappuis]

My goal would be that the coefficient is normalized so that it is zero in 2019. How can I do that?

[Clyde Schechter]

That makes no sense to me. Please explain what your variables are and what the underlying process connecting them that you are trying to model is.

[Felix Chappuis]

Thank you very much for the answer. Sorry for not having explicated clearer. Now I try to explain more. Thank you so much that you help me.
My figure should show how the unemployment rate in Logs (Arbeitslosenrate_ln) evolve in groups differentially exposed to the minimum wage, relative to the pre-policy year 2018. avg_GAP is my exposure measure. It is the same figure as in the figure 7c of the paper "Reallocation Effects of the Minimum Wage" of Christian Dustmann.
The code variable contains individuums with same age (anos2), region (nuts1), sex(sexo), education(estu). Unemployment_rate for this groups is (Arbeitslosenrate_ln). Post is a dummy which takes the value 1 after the treatment and before the treatment equals 0. Indicate the dif-in-dif-variable (post x avg_GAP).

[Felix Chappuis]

The same with the reference period 2019 would be my goal.

[Clyde Schechter]

I do not understand the graphs you are showing. They make no sense to me. They talk about plotted coefficients and effects in the caption, but the axis titles say they are plotting levels of outcome variables, not coefficients. They make reference to an equation (4) which is not shown, so who knows what that means. The graphs shown look self-contradictory to me. I suspect it could be puzzled out from reading the full text of the article. I do not have access to economics journals through my institution, and probably would not have time to read it even if I did. I don't think I can help you with this.

I also still cannot make sense of fixing the 2019 outcomes at zero. To do that, you have to be able to, for each observation in a given year, identify the corresponding observation in other years so you can calculate the difference. But there is no clear way I can see in your data to decide which observation of a given year corresponds to which observation in 2019. Does the variable code serve that purpose?

Post is a dummy which takes the value 1 after the treatment and before the treatment equals 0. Indicate the dif-in-dif-variable (post x avg_GAP).

The variable post has not appeared anywhere in your model. I'm just finding this more confusing as we go.

[Felix Chappuis]

Yes, exactly, the code serves for that. The post is not important in my case, sorry for having described it.