* Ths file clean the data 
* and transform some dollar nominal variables
* into real 1990 dollars
*******************************************;
options compress=no;
libname in 'D:/cm/';
libname on 'D:/enrico/imported/data';


data in.extract; merge in.merged1 in.errors; by ppn;
*CPI;
if yr=82 then va  = va*1.445;    *output;
if yr=82 then sw  = sw*1.445;    *labor;
if yr=82 then lc  = lc*1.445; 
if yr=82 then ww  = ww*1.445; 
if yr=82 then ow  = ow*1.445; 
if yr=82 then vlc = vlc*1.445;
if yr=82 then le  = le*1.445;
if yr=82 then bae = bae*1.445;   *capital;
if yr=82 then mae = mae*1.445;
if yr=82 then nb  = nb*1.445;    *investment;
if yr=82 then nm  = nm*1.445;
if yr=82 then ne  = ne*1.445;
if yr=82 then ub  = ub*1.445;
if yr=82 then um  = um*1.445;
if yr=82 then ue  = ue*1.445;
if yr=82 then tce = tce*1.445;
if yr=82 then nm  = nm*1.445;
if yr=82 then nmc = nmc*1.445;
if yr=82 then nmo = nmo*1.445;
if yr=82 then nmt = nmt*1.445;


* MULTI-ESTABLISHMENTS;
if substr(id,1,1)='0' then MU =0; else MU=1;
label MU = 'multi-establishment dummy';

* FIRM ID (see LRD manual);
if substr(id,1,1)='0' then FID=ID;
else FID=substr(ID,1,6)||'0000';
label FID ='FID=ID if MU=0; FID=1234560000 if MU=1';


keep 
ppn MU FID smsa80 fst yr ind 
medu medu80 medu2 medu280 mcoll mcoll80 mcoll2 mcoll280
te pw ph
va sw ww ow lc le vlc             
bae mae 
nb nm ne ub um ue tce nmc nmo nmt 
miss82 miss92 err1 err2 rural82 rural92;                      

proc means;
run;