Hello everyone,
I'm trying to do an Instrumental Variable (IV) estimation for an endogenous variable within an interaction term. Furthermore, I use a three-dimensional panel dataset. My dataset includes 4864 supplier-customer relationships. Each row in my dataset contains the id of the supplier (id_supplier), the id of the customer (id_customer), the sustainability value of the supplier (value_supplier), the sustainability value of the customer (value_customer), the exogenous sustainability pressure of the customer (iv) a variable that indicates if the relationship was active in that year (rel_active) and the year (year) of the observation. Moreover, control variables for the supplier (control_supplier) and for the customer (control_customer) are included. In some rows, the control variables are missing. Note that value_supplier, value_customer, rel_active and iv are all binary. Furthermore, note that some suppliers and customers have multiple supplier-customer relationships and occur multiple times in the dataset. Below you can find an example dataset. I try to estimate b in the following equation:
value_supplier = a + b*rel_act*value_customer + g*control_supplier + l*control_customer + e (1)
In order to do this, I use the following code:
In the previous setting, value_customer is expected to be endogenous. Hence, I want to instrument it with iv . The first stage estimation would look like this:
value_customer = a1 + b1*iv + l1*control_customer + e1 (2)
The second stage would again be the same as equation (1).
I tried the following:
However, this instruments the full interaction term, while I only want to instrument a specific part of the interaction term. More specifically, I want to instrument value_customer with iv and multiply this estimate with rel_act. Furthermore, I only want to use the control variable for the customer (control_customer) in the first stage, while in the second stage I want to include both the supplier and customer control variables. Can someone help me out to do this in Stata?
This is the first time I make use of this platform so please let me know if anything is not clear. Thank you in advance for your time.
I'm trying to do an Instrumental Variable (IV) estimation for an endogenous variable within an interaction term. Furthermore, I use a three-dimensional panel dataset. My dataset includes 4864 supplier-customer relationships. Each row in my dataset contains the id of the supplier (id_supplier), the id of the customer (id_customer), the sustainability value of the supplier (value_supplier), the sustainability value of the customer (value_customer), the exogenous sustainability pressure of the customer (iv) a variable that indicates if the relationship was active in that year (rel_active) and the year (year) of the observation. Moreover, control variables for the supplier (control_supplier) and for the customer (control_customer) are included. In some rows, the control variables are missing. Note that value_supplier, value_customer, rel_active and iv are all binary. Furthermore, note that some suppliers and customers have multiple supplier-customer relationships and occur multiple times in the dataset. Below you can find an example dataset. I try to estimate b in the following equation:
value_supplier = a + b*rel_act*value_customer + g*control_supplier + l*control_customer + e (1)
In order to do this, I use the following code:
Code:
reghdfe value_supplier c.value_customer#c.rel_act control_supplier control_customer, absorb(year id_customer id_supplier) cluster(id_customer id_supplier)
value_customer = a1 + b1*iv + l1*control_customer + e1 (2)
The second stage would again be the same as equation (1).
I tried the following:
Code:
ivreghdfe value_supplier (c.value_customer#c.rel_act = c.iv#c.rel_act) control_supplier control_customer, absorb(year id_customer id_supplier) cluster(id_customer id_supplier)
This is the first time I make use of this platform so please let me know if anything is not clear. Thank you in advance for your time.
Code:
* Example generated by -dataex-. For more info, type help dataex clear input float(id_supplier id_customer year value_supplier value_customer rel_act iv control_supplier control_customer) 297 1 2011 1 0 1 1 11.639135 7.345152 297 1 2012 1 0 1 1 11.665071 7.538813 297 1 2013 1 0 0 1 11.688667 7.533293 297 1 2014 1 0 0 1 11.745806 7.658811 297 1 2015 1 0 0 1 11.674466 7.69298 297 1 2016 1 0 0 1 11.612725 7.67313 297 1 2017 1 0 0 1 11.673938 7.732457 297 1 2018 1 0 0 1 11.738913 8.246067 297 1 2019 1 0 0 1 11.72304 8.196299 297 1 2020 1 1 0 1 11.932858 8.214087 297 2 2011 1 0 0 1 11.639135 11.523262 297 2 2012 1 0 1 1 11.665071 11.6708 297 2 2013 1 1 1 1 11.688667 11.78367 297 2 2014 1 1 1 1 11.745806 11.70608 297 2 2015 1 1 1 1 11.674466 11.693303 297 2 2016 1 1 0 1 11.612725 11.680945 297 2 2017 1 1 0 1 11.673938 11.773896 297 2 2018 1 1 0 1 11.738913 11.82932 297 2 2019 1 1 0 1 11.72304 11.852294 297 2 2020 1 1 0 1 11.932858 11.936676 628 2 2011 0 0 0 1 5.331902 11.523262 628 2 2012 0 0 0 1 5.630122 11.6708 628 2 2013 0 1 0 1 6.251455 11.78367 628 2 2014 0 1 0 1 6.396355 11.70608 628 2 2015 0 1 0 1 6.677984 11.693303 628 2 2016 0 1 1 1 6.789804 11.680945 628 2 2017 0 1 0 1 7.197271 11.773896 628 2 2018 0 1 0 1 7.409732 11.82932 628 2 2019 0 1 0 1 7.487259 11.852294 628 2 2020 0 1 0 1 . 11.936676 664 2 2011 1 0 0 1 9.45995 11.523262 664 2 2012 1 0 1 1 9.663421 11.6708 664 2 2013 1 1 1 1 9.721091 11.78367 664 2 2014 1 1 1 1 9.733117 11.70608 664 2 2015 1 1 1 1 9.794256 11.693303 664 2 2016 1 1 0 1 9.812799 11.680945 664 2 2017 1 1 0 1 9.933483 11.773896 664 2 2018 1 1 0 1 10.029675 11.82932 664 2 2019 1 1 0 1 10.104348 11.852294 664 2 2020 1 1 0 1 10.301924 11.936676 223 3 2011 1 0 0 1 8.384896 10.60933 223 3 2012 1 0 0 1 8.661746 10.72179 223 3 2013 1 0 0 1 8.756004 10.64137 223 3 2014 1 1 0 1 8.836883 10.606857 223 3 2015 1 1 1 1 8.813504 10.570445 223 3 2016 1 1 1 1 8.897864 10.51461 223 3 2017 1 1 0 1 8.834514 10.494575 223 3 2018 1 1 0 1 8.834119 10.40765 223 3 2019 1 1 0 1 8.870854 10.389642 223 3 2020 1 1 0 1 8.924748 10.42371 228 3 2011 0 0 0 1 8.402884 10.60933 228 3 2012 0 0 0 1 8.438716 10.72179 228 3 2013 0 0 1 1 8.478651 10.64137 228 3 2014 0 1 1 1 8.524513 10.606857 228 3 2015 0 1 1 1 8.5107765 10.570445 228 3 2016 0 1 0 1 8.53775 10.51461 228 3 2017 0 1 0 1 8.457214 10.494575 228 3 2018 0 1 0 1 8.4991255 10.40765 228 3 2019 0 1 0 1 8.437344 10.389642 228 3 2020 0 1 0 1 8.500991 10.42371 236 3 2011 0 0 0 1 5.210557 10.60933 236 3 2012 0 0 0 1 5.354584 10.72179 236 3 2013 0 0 0 1 5.498945 10.64137 236 3 2014 0 1 0 1 5.635346 10.606857 236 3 2015 0 1 0 1 5.721793 10.570445 236 3 2016 0 1 1 1 5.711586 10.51461 236 3 2017 0 1 1 1 5.754479 10.494575 236 3 2018 0 1 0 1 5.899917 10.40765 236 3 2019 0 1 0 1 6.074744 10.389642 236 3 2020 0 1 0 1 6.106804 10.42371 310 3 2011 0 0 0 1 8.637799 10.60933 310 3 2012 0 0 1 1 8.727941 10.72179 310 3 2013 0 0 1 1 8.763412 10.64137 310 3 2014 0 1 0 1 8.914007 10.606857 310 3 2015 0 1 0 1 8.9985075 10.570445 310 3 2016 0 1 0 1 9.028938 10.51461 310 3 2017 0 1 0 1 9.084494 10.494575 310 3 2018 1 1 0 1 9.126818 10.40765 310 3 2019 1 1 0 1 9.155863 10.389642 310 3 2020 1 1 0 1 9.188789 10.42371 399 3 2011 0 0 0 1 7.840037 10.60933 399 3 2012 0 0 0 1 7.96419 10.72179 399 3 2013 0 0 0 1 8.284226 10.64137 399 3 2014 0 1 0 1 8.388246 10.606857 399 3 2015 0 1 0 1 8.4487 10.570445 399 3 2016 0 1 0 1 8.541574 10.51461 399 3 2017 0 1 0 1 8.5806 10.494575 399 3 2018 0 1 0 1 9.058843 10.40765 399 3 2019 0 1 0 1 9.161801 10.389642 399 3 2020 1 1 0 1 9.236495 10.42371 450 3 2011 0 0 0 1 9.201338 10.60933 450 3 2012 0 0 0 1 9.295307 10.72179 450 3 2013 0 0 0 1 9.3210125 10.64137 450 3 2014 0 1 0 1 9.43675 10.606857 450 3 2015 0 1 0 1 9.493589 10.570445 450 3 2016 0 1 0 1 9.416951 10.51461 450 3 2017 0 1 0 1 9.397379 10.494575 450 3 2018 0 1 1 1 9.647943 10.40765 450 3 2019 1 1 1 1 9.63692 10.389642 450 3 2020 1 1 0 1 9.774329 10.42371 end
