hello,
I'm trying to use entropy balancing to match my treatment and control firms. My data spans over 2 years. There are exactly the same number of firms each year, i.e., each firm has an observation in each year, with a total of 2636 firm-year observations. There are 633 treatment firms and 2003 control firms. I do not use PMS to match because I want to preserve all the 2636 observations. My understanding is that if I use entropy balancing, each observation will get a weight (_webal), however, that is not the case when I run -ebalance-, there are 1171 observations with _webal == .
I wonder why this is the case, and my code is as follows
$xx are a set of covariants used in matching, they are also the control variables used in my ols regression; treat indicates whether the firm is a treatment firm.
After -ebalance-, my regression is as follows
Please kindly help me how I can do -ebalance- to preserve all the observations. Thanks a lot for any help!
I'm trying to use entropy balancing to match my treatment and control firms. My data spans over 2 years. There are exactly the same number of firms each year, i.e., each firm has an observation in each year, with a total of 2636 firm-year observations. There are 633 treatment firms and 2003 control firms. I do not use PMS to match because I want to preserve all the 2636 observations. My understanding is that if I use entropy balancing, each observation will get a weight (_webal), however, that is not the case when I run -ebalance-, there are 1171 observations with _webal == .
I wonder why this is the case, and my code is as follows
Code:
ebalance treat $xx ,target(3)
After -ebalance-, my regression is as follows
Code:
reghdfe y i.treat##i.post $xx i.industry i.year [aweight=_webal], noabsorb
