PSM for binary dependent variable (panel data)

[Muqaddas Khalid]

I have panel data of 2028 obseravtions (156 companies for 13 years). My dependent variable is binary (distressed being 0, non-distressed being 1). I got 438 observations out 2028 as distressed for which I need a matching sample of 438 non-distressed observations from remaining 1590 observations.

My question is how can I match the distressed observations with those of non-distressed observations. Can I use Propensity Score Matching (PSM) method? If yes, is it possible to apply PSM on dependent variable? If yes, how? I am stuck in matching these observations. I need your kind help.

Regards,
Muqaddas Khalid.

[Muqaddas Khalid]

I am giving the following command in Stata but not getting the desired results (its giving '438 distressed and 219 non-distressed' instead of '438 distressed and 438 non-distressed'). Kindly correct me if my command has some problem.

logit distress1 age roe size leverage liquidity
predict pscore
drop if pscore==.
bys years ind_dummy distress1: g exp=_n if distress1
bys years ind_dummy : egen m=max(exp)
drop if m==.
expandcl m if distress1!=1,cl(years stockid) gen(g)
bys years stockid: replace exp=_n if distress1!=1
bys years ind_dummy exp: egen r=sum(pscore*distress1)
g q=abs((pscore-r)/r) if distress1!=1
bys years ind_dummy exp: egen u=min(q) if !distress1
drop if u!=q&!distress1
egen mat=group(years ind_dummy exp)
bysort mat:gen fre1=_n
drop if fre1==1
drop exp-fre1

where age roe size leverage liquidity are the control variables and distress1 is the binary dependent variable of the study (distressed being 0, non-distressed being 1).

Much help needed. Thanks in advance.