PSM The balancing Property is not satisfied

[Diogo Carvalho]

Hi everyone,

I have been working on a dataset regarding the impact of environmental subsidies on green innovation and when I performed the PSM on the model that I have, it stated that the balancing property was not satisfied. This is an academic work for my MSc and my professor does not want us to solve the problem by reducing the number of covariates in the model. Thus, I need to solve the problem with the current model that I have, without having to change the current covariates. This is what I have done:

*Treatment variable: dum_envi

global treatment dum_envi
global ylist pan
global xlist roalag age debt roa2 debt2
global breps 5

describe $treatment $ylist $xlist
summarize $treatment $ylist $xlist
bysort $treatment: summarize $ylist $xlist

*Regression with a dummy variable for treatment (t-test)*
xi: xtreg $ylist $treatment

*Regression with a dummy variable for treatment controlling for x*
xi: xtreg $ylist $treatment $xlist
outreg2 using reg_with_dummy, word

*Application of the PSM*
*Propensity score matching with common support*
pscore2 $treatment $xlist, pscore(myscore) blockid(myblock) comsup


Can anyone help, please?

Note: I cannot use the entropy balancing solution either (requested by the professor as well)

[Felix Bittmann]

I recommend using kmatch instead, as it allows a much more nuanced assessment of the situation, see https://www.stata.com/meeting/uk17/slides/uk17_Jann.pdf