Hello I have a longitudinal panel of two time periods where I consider the death of a hh member between surveys as my treatment, Post would be data taken in round 2.
my model is as follows; (its in the nature of a triple DID)
I intend to find the differentiated impact of the treatment on the wives vs husband of my household( that is why i add the female indictaor)
Is it okay to add a female interaction with time varying covariates such as total household consumption exp (COTOTAL5) and no of children(Nchildren), although they will be the same between husband and wife. I add this because I feel the more number of children will affect the wife differently than the husband.Similarly consumption items( within total consumption expenditure) will be of different use for the husband and wife. Is this alright?
Also I can only add the female indicator as a fixed effect, I cannot add this as
as female gets ommitted due to collinearity
my model is as follows; (its in the nature of a triple DID)
Code:
drop if PERSONid==75 & post==0 //drop dead id's from round 1 gen treated= treated group==1& post==1 gen treatedfemale= treated*female reghdfe y treated treatedfemale , absorb( female HHPBASE SURVEY#female COTOTAL5#female Nchildren#female district_id#c.SURVEY#female) cluster(IDPSU)
Is it okay to add a female interaction with time varying covariates such as total household consumption exp (COTOTAL5) and no of children(Nchildren), although they will be the same between husband and wife. I add this because I feel the more number of children will affect the wife differently than the husband.Similarly consumption items( within total consumption expenditure) will be of different use for the husband and wife. Is this alright?
Also I can only add the female indicator as a fixed effect, I cannot add this as
Code:
reghdfe y treated treatedfemale female, absorb()
