I am writing this post in order to receive some help with the verification of parallel trends assumption. The analysis I am conducting concerns the effect of the implementation of a CSR disclosure policy in Europe on Agency Costs of firms. I got data from 2014 to 2019, the majority of countries apply the CSR policy in 2017 except for Spain that applies it in 2018, Greece and Estonia which apply it in 2016.
This is a sample of data, I created a dummy variable for pre-post treatment period (i.e. DummyT), a dummy variable for treated observations (i.e. CSR) anche the interaction term between the two (i.e. InteractionDID).
Code:
* Example generated by -dataex-. To install: ssc install dataex clear input str109 CompanyName str11 CountryofIncorporation int YEAR float(AgencyCost_Assets dummyT CSR InteractionDID) "Agrana Beteiligungs AG" "Austria" 2015 1.0359905 0 0 0 "Agrana Beteiligungs AG" "Austria" 2016 1.1045318 0 0 0 "Agrana Beteiligungs AG" "Austria" 2017 1.0321829 1 1 1 "Agrana Beteiligungs AG" "Austria" 2018 1.089074 1 1 1 "Agrana Beteiligungs AG" "Austria" 2019 1.0224495 1 1 1 "AMAG Austria Metall AG" "Austria" 2014 .7561717 0 0 0 "AMAG Austria Metall AG" "Austria" 2015 .8270431 0 0 0 "AMAG Austria Metall AG" "Austria" 2016 .6521036 0 0 0 "AMAG Austria Metall AG" "Austria" 2017 .7376047 1 1 1 "AMAG Austria Metall AG" "Austria" 2018 .7055686 1 1 1 "AMAG Austria Metall AG" "Austria" 2019 .709861 1 1 1 "ams AG" "Austria" 2014 .48626965 0 0 0 "ams AG" "Austria" 2015 .5093262 0 0 0 "ams AG" "Austria" 2016 .3864685 0 0 0 "ams AG" "Austria" 2017 .3261792 1 1 1 "ams AG" "Austria" 2018 .3979056 1 1 1 "ams AG" "Austria" 2019 .4252402 1 1 1 "Andritz AG" "Austria" 2014 .9773194 0 0 0 "Andritz AG" "Austria" 2015 1.1037016 0 0 0 "Andritz AG" "Austria" 2016 .97425 0 0 0 "Andritz AG" "Austria" 2017 .9399393 1 1 1 "Andritz AG" "Austria" 2018 .871771 1 1 1 "Andritz AG" "Austria" 2019 .9225614 1 1 1 "AT & S Austria Technologie & Systemtechnik AG" "Austria" 2014 .643964 0 0 0 "AT & S Austria Technologie & Systemtechnik AG" "Austria" 2015 .54636586 0 0 0 "AT & S Austria Technologie & Systemtechnik AG" "Austria" 2016 .5673439 0 0 0 "AT & S Austria Technologie & Systemtechnik AG" "Austria" 2017 .5672092 1 1 1 "AT & S Austria Technologie & Systemtechnik AG" "Austria" 2018 .6480774 1 1 1 "AT & S Austria Technologie & Systemtechnik AG" "Austria" 2019 .5761894 1 1 1 "Athos Immobilien AG" "Austria" 2016 .09226921 0 0 0 "Athos Immobilien AG" "Austria" 2017 .09088276 1 0 0 "Athos Immobilien AG" "Austria" 2018 .09001458 1 0 0 "Athos Immobilien AG" "Austria" 2019 .08367804 1 0 0 "Burgenland Holding AG" "Austria" 2014 0 0 0 0 "Burgenland Holding AG" "Austria" 2015 0 0 0 0 "Burgenland Holding AG" "Austria" 2016 0 0 0 0 "Burgenland Holding AG" "Austria" 2017 0 1 0 0 "Burgenland Holding AG" "Austria" 2018 .00012560148 1 0 0 "Burgenland Holding AG" "Austria" 2019 0 1 0 0 "CA Immobilien Anlagen AG" "Austria" 2014 .0550363 0 0 0 "CA Immobilien Anlagen AG" "Austria" 2015 .05296383 0 0 0 "CA Immobilien Anlagen AG" "Austria" 2016 .05246661 0 0 0 "CA Immobilien Anlagen AG" "Austria" 2017 .05108768 1 0 0 "CA Immobilien Anlagen AG" "Austria" 2018 .0486253 1 0 0 "CA Immobilien Anlagen AG" "Austria" 2019 .04771639 1 0 0 "DO & CO AG" "Austria" 2014 1.239242 0 0 0 "DO & CO AG" "Austria" 2015 1.351606 0 0 0 "DO & CO AG" "Austria" 2016 1.459347 0 0 0 "DO & CO AG" "Austria" 2017 1.4699473 1 1 1 "DO & CO AG" "Austria" 2018 1.503071 1 1 1 "DO & CO AG" "Austria" 2019 1.4405873 1 1 1 "EVN AG" "Austria" 2015 .328524 0 0 0 "EVN AG" "Austria" 2016 .3121482 0 0 0 "EVN AG" "Austria" 2017 .3432431 1 1 1 "EVN AG" "Austria" 2018 .26544163 1 1 1 "EVN AG" "Austria" 2019 .26915467 1 1 1 "Fabasoft AG" "Austria" 2014 .9129975 0 0 0 "Fabasoft AG" "Austria" 2015 .978932 0 0 0 "Fabasoft AG" "Austria" 2019 .7771518 1 0 0 "FACC AG" "Austria" 2014 .9614663 0 0 0 "FACC AG" "Austria" 2015 .7363977 0 0 0 "FACC AG" "Austria" 2016 .8726558 0 0 0 "FACC AG" "Austria" 2017 1.0296525 1 1 1 "FACC AG" "Austria" 2018 1.0625663 1 1 1 "FACC AG" "Austria" 2019 .8864281 1 1 1 "Flughafen Wien AG" "Austria" 2014 .3330248 0 0 0 "Flughafen Wien AG" "Austria" 2015 .33176735 0 0 0 "Flughafen Wien AG" "Austria" 2016 .3674309 0 0 0 "Flughafen Wien AG" "Austria" 2017 .3650937 1 1 1 "Flughafen Wien AG" "Austria" 2018 .3705393 1 1 1 "Flughafen Wien AG" "Austria" 2019 .3727733 1 1 1 "Frauenthal Holding AG" "Austria" 2014 1.708831 0 0 0 "Frauenthal Holding AG" "Austria" 2015 1.9760165 0 0 0 "Frauenthal Holding AG" "Austria" 2016 2.1863856 0 0 0 "Frauenthal Holding AG" "Austria" 2017 2.1635895 1 1 1 "Frauenthal Holding AG" "Austria" 2018 2.3392391 1 1 1 "Frauenthal Holding AG" "Austria" 2019 2.1511493 1 1 1 "Frequentis AG" "Austria" 2014 1.5880376 0 0 0 "Frequentis AG" "Austria" 2015 1.591476 0 0 0 "Frequentis AG" "Austria" 2016 1.3725097 0 0 0 "Frequentis AG" "Austria" 2017 1.41069 1 1 1 "Frequentis AG" "Austria" 2018 1.4844916 1 1 1 "Frequentis AG" "Austria" 2019 1.1400046 1 1 1 "Hutter & Schrantz Stahlbau AG" "Austria" 2014 1.2100103 0 0 0 "Hutter & Schrantz Stahlbau AG" "Austria" 2015 1.1128067 0 0 0 "Immofinanz AG" "Austria" 2014 .06504148 0 0 0 "Immofinanz AG" "Austria" 2015 .05618835 0 0 0 "Immofinanz AG" "Austria" 2016 .03097149 0 0 0 "Immofinanz AG" "Austria" 2017 .05350833 1 0 0 "Immofinanz AG" "Austria" 2018 .05717643 1 0 0 "Immofinanz AG" "Austria" 2019 .057424 1 0 0 "Josef Manner & Comp AG" "Austria" 2014 1.5359948 0 0 0 "Josef Manner & Comp AG" "Austria" 2015 1.39001 0 0 0 "Josef Manner & Comp AG" "Austria" 2016 1.3905193 0 0 0 "Josef Manner & Comp AG" "Austria" 2017 1.419019 1 1 1 "Josef Manner & Comp AG" "Austria" 2018 1.4994643 1 1 1 "Josef Manner & Comp AG" "Austria" 2019 1.526918 1 1 1 "Kapsch Trafficcom AG" "Austria" 2014 .8591163 0 0 0 "Kapsch Trafficcom AG" "Austria" 2015 .8851436 0 0 0 "Kapsch Trafficcom AG" "Austria" 2016 1.0241873 0 0 0 end
Following some of the previous posts, I launched the following command on STATA:
preserve
collapse (mean) AgencyCost, by(CSR YEAR)
reshape wide AgencyCost, i(YEAR) j(CSR)
graph twoway connect AgencyCost* YEAR if YEAR < 2017
restore
but I guess I need to change something since not all countries apply the policy in 2017 and because with this formula I only get one trend line (just for non treated countries before 2016).
How can I solve the problem? Which is the right formula I should use?
Thanks a lot for the answers!
0 Response to Parallel trends Graphing
Post a Comment