Good morning to the community,

I would like to ask an econometric question on my dataset which is shown below:

Code:
Code:
* 
 Example generated by -dataex-. To install: ssc install dataex
clear
input double czone float(e_rank_b_n expof_euro5_qo93_07 expof_euro5_qo93_07_middle_pct1 expof_euro5_qo93_07_middle_pct2 expof_euro5_qo93_07_middle_pct3) byte gini99_pct
 100   -.9960554  3.840633         0  3.840633         0 3
 200  -1.1039523 2.8010066         0 2.8010066         0 3
 301   -.8791488  3.026863  3.026863         0         0 3
 302  -1.0925575  2.719768  2.719768         0         0 3
 401  -1.2466074  1.666586  1.666586         0         0 3
 402  -1.1890154 2.7771695         0 2.7771695         0 3
 500  -1.3335255  1.479456  1.479456         0         0 3
 601   -1.003652  1.641976         0  1.641976         0 2
 602   -.5675598  1.641976         0         0  1.641976 1
 700   -1.346113 1.6473615         0 1.6473615         0 3
 800   -1.517785 1.8004093         0 1.8004093         0 2
 900  -1.4469873 1.3499714 1.3499714         0         0 3
1001   -.7082276   2.45536         0   2.45536         0 3
1002  -1.1438781 1.9775038         0         0 1.9775038 2
1100  -1.4961878 1.7257783         0 1.7257783         0 2
1201   -.4924781 2.1177337         0 2.1177337         0 2
Code:
 list czone e_rank_b_n  expof_euro5_qo93_07 expof_euro5_qo93_07_middle_pct1 expof_euro5_qo93_07_middle_pct2 expof_euro5_qo93_07_middl
> e_pct3 gini99_pct in 1/10, clean noobs abbreviate(20)

    czone   e_rank_b_n   expof_euro5_qo93_07   expof_euro5_q~e_pct1   expof_euro5_q~e_pct2   expof_euro5_q~e_pct3   gini99_pct  
      100    -.9960554              3.840633                      0               3.840633                      0            3  
      200    -1.103952              2.801007                      0               2.801007                      0            3  
      301    -.8791488              3.026863               3.026863                      0                      0            3  
      302    -1.092558              2.719768               2.719768                      0                      0            3  
      401    -1.246607              1.666586               1.666586                      0                      0            3  
      402    -1.189015              2.777169                      0               2.777169                      0            3  
      500    -1.333526              1.479456               1.479456                      0                      0            3  
      601    -1.003652              1.641976                      0               1.641976                      0            2  
      602    -.5675598              1.641976                      0                      0               1.641976            1  
      700    -1.346113              1.647362                      0               1.647362                      0            3
Basically I am running a regression:

Code:
reg e_rank_b_n expof_euro5_qo93_07 expof_euro5_qo93_07_middle_pct1 expof_euro5_qo93_07_middle_pct2 expof_euro5_qo93_07_middle_pct3 where the dependent variable is Absolute Mobility and the main indepdendent variable it's a tech shock taken from Acemoglu (2019). I would like to create an interaction variable between the tech. shock and gini coefficient. I tried different strategies from a direct "shock * gini" interaction to the one presented above where I divided Gini in tertiles, and then by creating 3 dummies I interacted them with my main covariate.
i.e.
Code:
  "expof_euro5_qo93_07_middle_pct1 expof_euro5_qo93_07_middle_pct2 expof_euro5_qo93_07_middle_pct3"
I am not quite sure of the robustness of my strategy for the interaction variables and also how to interpret the coefficient, if somebody could give me some feedback or some references I would be very thankful. Moreover should I control for the continous variable Gini or for the categorized Gini?

Thank you again
Guido
Tags: None