Staggered Difference in Difference

[Tesfaye Salarin]

Dear Stata Users,
I am new on the platform and also on Stata, so I am looking for someone to kindly help me with an issue. Basically, I have a panel dataset from 1998 to 2015 on Italian municipalities' public finances (i.e. cost of general administration for the municipality "i" for each year, though I happen to have some missing values. I collected data from the Ministry of Interior). Then I created two dummy variables: one called "financial distress" and the other one called "financial restructuring plan", respectively indicating whether a local entity ever encountered financial distress or if it ever activated a long-term restructuring plan. Financial Distress=1 in the year when a municipality entered financial distress onwards, and 0 otherwise. The same goes for the other dummy variable, being financial restructuring plans=1 the year when a local entity activated a restructuring plan onwards. Finally, I created another dummy called "treated" taking value=1 if either "financial distress" or "financial restructuring plan" = 1. My goal is to estimate the impact of financial distress and of the activation of restructuring plans on local public finances, but I don't know how to build the model considering that the dummy variables don't start all at the same period (ex. Rome; cost of general administration=100; financial distress=1 from 2005 onwards; financial restructuring plan = 0 in every year; treated=1 from 2005 onwards etc.). I am quite sure that the right model to adopt is a staggered DiD but as I am 'new' in this field I would really appreciate some help. I know it sounds easy to many of you, but I am just kindly looking for some help for my research.

Thank you in advance and I hope you all have a nice day!
Tesfaye

[Justin Niakamal]

Welcome to Statalist. I suggest searching this forum for generalized difference in differences. In particular, the posts by Clyde Schechter .

[Tesfaye Salarin]

* Example generated by -dataex-. To install: ssc install dataex
clear
input int anno str28(regione provincia) str49 nomeente double codiceente float costo_tot_dir_amm_gen byte(dissesti riequilibri treated)
1998 "SICILIA" "CATANIA" "ACI SANT'ANTONIO" 5190210050         . 0 . 1
1999 "SICILIA" "CATANIA" "ACI SANT'ANTONIO" 5190210050   1592229 0 . 1
2000 "SICILIA" "CATANIA" "ACI SANT'ANTONIO" 5190210050         . 0 . 1
2001 "SICILIA" "CATANIA" "ACI SANT'ANTONIO" 5190210050   3388569 0 . 1
2002 "SICILIA" "CATANIA" "ACI SANT'ANTONIO" 5190210050   2126342 0 . 1
2003 "SICILIA" "CATANIA" "ACI SANT'ANTONIO" 5190210050   2076850 0 . 1
2004 "SICILIA" "CATANIA" "ACI SANT'ANTONIO" 5190210050         . 0 . 1
2005 "SICILIA" "CATANIA" "ACI SANT'ANTONIO" 5190210050         . 0 . 1
2006 "SICILIA" "CATANIA" "ACI SANT'ANTONIO" 5190210050   2368812 0 . 1
2007 "SICILIA" "CATANIA" "ACI SANT'ANTONIO" 5190210050   2198225 0 . 1
2008 "SICILIA" "CATANIA" "ACI SANT'ANTONIO" 5190210050   2178940 0 . 1
2009 "SICILIA" "CATANIA" "ACI SANT'ANTONIO" 5190210050   2369071 0 . 1
2010 "SICILIA" "CATANIA" "ACI SANT'ANTONIO" 5190210050   2020367 0 . 1
2011 "SICILIA" "CATANIA" "ACI SANT'ANTONIO" 5190210050         0 0 . 1
2012 "SICILIA" "CATANIA" "ACI SANT'ANTONIO" 5190210050 1297625.3 0 . 1
2013 "SICILIA" "CATANIA" "ACI SANT'ANTONIO" 5190210050         0 1 . 1
2014 "SICILIA" "CATANIA" "ACI SANT'ANTONIO" 5190210050         0 1 . 1
2015 "SICILIA" "CATANIA" "ACI SANT'ANTONIO" 5190210050 1297625.3 1 . 1
1998 "SICILIA" "CATANIA" "ADRANO"           5190210060         . . 0 1
1999 "SICILIA" "CATANIA" "ADRANO"           5190210060   4629682 . 0 1
2000 "SICILIA" "CATANIA" "ADRANO"           5190210060      1297 . 0 1
2001 "SICILIA" "CATANIA" "ADRANO"           5190210060   1464242 . 0 1
2002 "SICILIA" "CATANIA" "ADRANO"           5190210060    894307 . 0 1
2003 "SICILIA" "CATANIA" "ADRANO"           5190210060    760113 . 0 1
2004 "SICILIA" "CATANIA" "ADRANO"           5190210060         . . 0 1
2005 "SICILIA" "CATANIA" "ADRANO"           5190210060         . . 0 1
2006 "SICILIA" "CATANIA" "ADRANO"           5190210060         0 . 0 1
2007 "SICILIA" "CATANIA" "ADRANO"           5190210060         0 . 0 1
2008 "SICILIA" "CATANIA" "ADRANO"           5190210060         . . 0 1
2009 "SICILIA" "CATANIA" "ADRANO"           5190210060         0 . 0 1
2010 "SICILIA" "CATANIA" "ADRANO"           5190210060         0 . 0 1
2011 "SICILIA" "CATANIA" "ADRANO"           5190210060         0 . 0 1
2012 "SICILIA" "CATANIA" "ADRANO"           5190210060         0 . 0 1
2013 "SICILIA" "CATANIA" "ADRANO"           5190210060         0 . 0 1
2014 "SICILIA" "CATANIA" "ADRANO"           5190210060         0 . 0 1
2015 "SICILIA" "CATANIA" "ADRANO"           5190210060         0 . 0 1
end

[Tesfaye Salarin]

Justin Niakamal Thank you very much for your help! I have been looking for similar posts the whole day (I am in Italy so here it is almost evening), but I would really appreciate if you could propose me any suggestion on how to deal with my issue. Again, very much appreciated from you"

[Justin Niakamal]

You can start with a TWFE with dummies for your treatment variable(s) as a baseline and move into some of the more cutting-edge tools that apply to staggered adoption. There's an excellent repository here

https://github.com/asjadnaqvi/DiD