Hello, this is my first Statalist question, so I'm sorry for making any mistakes.
I have data from firms listed in the EU over the years from 2005-2014. What I want to investigate is whether the adoption of a directive (policy change) has a liqudity effect on the firms. The adoption is made at the country level and thus effects the firm that are listed in the countriy. However, while some countries implement the directive in 2008 other do in 2009 or 2010, thus i have a staggered adoption - these adoption countries serve as my treatment group. Moreover there are countries that had to implement the directive but are not affected by the implementation of the directive, because they already have required what the directive requires, and thus serve as the control group. I thought about having two dummy variables: TREAT: is coded 1 for firms that belong to the treatment group and 0 for the control group. POST: is coded 1 for firm-years after the adoption of the directive on the country level and 0 otherwise. I then thought to implement the following model: (ID is the firm ID and year is the year)
xtset ID year
xtreg liquiditymeasure TREAT POST TREATxPOST i.year, fe
However, I'm not sure whether this procedure is the right one to use. Moreover to estimate the POST dummies I used the adoption years of the country where the firm is listed. However, I'm not sure whether I need to match the control firms with the treatment firms first to be sure that I compare the same years in the pre/post (and if so, I don't know how to do that) or whether this does not matter. Moreover I'm insecure whether I can or should somehow use the staggered adoption -i.e. use the later adoption countries as controls for the earlier adoption countries rather than the countries not affected as control group?
Thanks in advance!
Kind regards,
Dominik
I have data from firms listed in the EU over the years from 2005-2014. What I want to investigate is whether the adoption of a directive (policy change) has a liqudity effect on the firms. The adoption is made at the country level and thus effects the firm that are listed in the countriy. However, while some countries implement the directive in 2008 other do in 2009 or 2010, thus i have a staggered adoption - these adoption countries serve as my treatment group. Moreover there are countries that had to implement the directive but are not affected by the implementation of the directive, because they already have required what the directive requires, and thus serve as the control group. I thought about having two dummy variables: TREAT: is coded 1 for firms that belong to the treatment group and 0 for the control group. POST: is coded 1 for firm-years after the adoption of the directive on the country level and 0 otherwise. I then thought to implement the following model: (ID is the firm ID and year is the year)
xtset ID year
xtreg liquiditymeasure TREAT POST TREATxPOST i.year, fe
However, I'm not sure whether this procedure is the right one to use. Moreover to estimate the POST dummies I used the adoption years of the country where the firm is listed. However, I'm not sure whether I need to match the control firms with the treatment firms first to be sure that I compare the same years in the pre/post (and if so, I don't know how to do that) or whether this does not matter. Moreover I'm insecure whether I can or should somehow use the staggered adoption -i.e. use the later adoption countries as controls for the earlier adoption countries rather than the countries not affected as control group?
Thanks in advance!
Kind regards,
Dominik
Comment