Sorry, I want to create a new variable that counts backswords from ac_choice=0 by gvkey- so data will look like that:    gvkey fyear ac_choice     before                           a         1990       0                 -3 a          1991       0                  -2 a           1992       0               -1 a          1993        1               0 a           1994        0                  -2 a             1995       0                -1 a             1996           1             0 a              1997        0                . a            1998           0              . b              2000          0                 -2 b              2001           0                  -1 b              2002           1                0 b               2003          0                    -1 b               2004           1                  0 b               2005            0                           . ... View more

Thank you! it is great! I dont have any missing years so I used the last option. It is Awesome.  ... View more

MY ID  variable is sting, with letters- is that a problem?  I do not want to skip zero. I will write again.  My data is  ID YEAR POST A 1990 0 A 1991 0 A 1992 0 A 1993 1 A 1994 1 B 2000 0 B 2001 0 B 2002 1 B 2002 1 B 2003 1   AND I WANT- zero on the first post of the ID: ID YEAR POST B_F A 1990 0         -3 A 1991 0      -2 A 1992 0        -1 A 1993 1       0 A 1994 1        1 B 2000 0        -2 B 2001 0           -1 B 2002 1        0 B 2002 1         1 B 2003 1         2     ... View more

Thanks, but I get this  "ERROR: Invalid DO loop control information, either the INITIAL or TO expression is missing or the BY expression is missing, zero, or invalid."... Any idea why?  ... View more

THANK YOU SO MUCH!! ... View more

I have tried to run this one- but I can't even run it. I don't have an output example. %macro match_by_yr(inset=, outset=, startyear=, endyear=, matches=, calip=, ctrlrestvar=, pscorevar=, treatvar=); proc sql; drop table &outset; quit; %do i=&startyear %to &endyear; %put Year &i; proc sql; /* generate treated dataset for this firm-year */ create table _treat as select unique_id as idT, &pscorevar as pscoreT from &inset where yr = &i and &treatvar = 1; /* generate non-treated dataset for this firm-year */ create table _nontreat as select unique_id as idC, &pscorevar as pscoreC from &inset where yr = &i and &treatvar = 0 and &ctrlrestvar=0; quit; proc sql noprint; select count(*) into :obs_count from _treat; quit; proc sql noprint; select count(*) into :obs_count2 from _nontreat; quit; %if (&obs_count ge 1) and (&obs_count2 ge 1) %then %do; /* match */ %PSMatching(datatreatment= _treat, datacontrol= _nontreat, method= NN, numberofcontrols= &matches, caliper=&calip, replacement= no, out = _matches); *if a firm, either control or treatment, is selected in a year, then it can not be selected again in future years; proc sql; create table &inset as select a.*, b.IdSelectedControl as id_delec from &inset as a left join _matches as b on a.unique_id=b.IdSelectedControl; quit; proc sql; create table &inset as select a.*, b.MatchedToTreatID as id_delet from &inset as a left join _matches as b on a.unique_id=b.MatchedToTreatID; quit; data &inset; set &inset; if not missing(id_delet) or not missing(id_delec) then delete; drop id_delet id_delec; run; /* Start the dataset if First firm-industry */ %if &i = 1 %then %do; data &outset; set _matches; run; %end; /* otherwise append new matches to the running dataset */ %else %do; data &outset; set &outset _matches; %end; %end; /* clean up temp datasets for next iteration */ proc sql; drop table _matches; quit; %end; %mend match_by_yr; ... View more