Hi everyone,
using EHCP micro data, I am trying to predict the rent that households in social housing would have paid on the market. For this purpose, I have a variable describing the tenant_status: 0 if no information on the landlord, 1 if owner, 2 if social rent and 3 if market rent.
Hence, I am running an hedonic regression for private rents only (tenant_status==3) i.e. I am regressing the log-monthly gross rent on a set of characteristics of the dwelling, neighborhood and regions of each country and then predict the rents for those in social market (tenant_status==2) and the owners (tenant_status==1):
However, I suspect selection bias due to segregation in the rental markets and I was thinking to run an Heckman model to check effectively whether the rho estimate is different from 0.
But I have a doubt: in this case, to have censored outcome, have I to omit the " if tenant_status==3" in the previous equation and substitute it with a dummy in the selection equation?
Or which other way do I have to check for selection bias in this case?
Thank you for your precious help!
using EHCP micro data, I am trying to predict the rent that households in social housing would have paid on the market. For this purpose, I have a variable describing the tenant_status: 0 if no information on the landlord, 1 if owner, 2 if social rent and 3 if market rent.
Hence, I am running an hedonic regression for private rents only (tenant_status==3) i.e. I am regressing the log-monthly gross rent on a set of characteristics of the dwelling, neighborhood and regions of each country and then predict the rents for those in social market (tenant_status==2) and the owners (tenant_status==1):
Code:
levelsof country, local(c)
foreach country in `c' {
capture noisily reg log_rent ha006a i.ha009 i.ha010 i.ha011 i.ha012 i.ha013 i.ha015
i.ha016 i.ha021 i.ha022 i.regions if country==`country' & wave==5 & tenant_status==3 [pw=hg004], r
capture noisily predict rent_`country' if country==`country' & inlist(tenant_status, 1,2), xb
}
But I have a doubt: in this case, to have censored outcome, have I to omit the " if tenant_status==3" in the previous equation and substitute it with a dummy in the selection equation?
Or which other way do I have to check for selection bias in this case?
Thank you for your precious help!
