Of course 1 operation is less overhead than >1 operations. 
I would do this:

FsrcCompIn IF   E           K DISK    PREFIX('SRC.') 
F                                     EXTDESC('SRCCOMP') EXTFILE('SRCCOMP')
FsrcCompOutO  A E             Disk    PREFIX('OUT.') 
F                                     EXTDESC('SRCCOMP') EXTFILE('SRCCOMP')

D SRC       E DS                      EXTNAME(SRCCOMP) QUALIFIED INZ
D OUT       E DS                      EXTNAME(SRCCOMP) QUALIFIED INZ

 /free
       read srcCompIn SRC;
       DOW NOT %EOF();
            eval-corr  out = src;
            write srcCompOut OUT;
            read srcCompIn SRC;
       enddo;


Alternatively, this might work, if it does, it would be even better:

FsrcCompIn IF   E           K DISK    PREFIX('SRC.') 
F                                     EXTDESC('SRCCOMP') EXTFILE('SRCCOMP')
FsrcCompOutO    E             DISK    PREFIX('OUT.') 
F                                     EXTDESC('SRCCOMP') 

D SRC       E DS                      EXTNAME(SRCCOMP) QUALIFIED INZ

 /free
       read srcCompIn SRC;
       DOW NOT %EOF();
            write srcCompOut SRC;
            read srcCompIn SRC;
       enddo;

Scenario: Create data from a source company database to one or more target 
companies.

In above situation when reading data from source and writing back to file for 
target with a number of fields is it better to set fields one by one or use an 
externally described data structure and move all fields in one operation. 
Would there be any performance improvement in one method over the other? 

SF