Mal keine Fragen, sondern eine Erkenntnis:

Der XML-PARSE konvertiert nicht automatisch. IBM schreibt hierzu:

When you parse ASCII XML documents, the document fragments passed to the processing procedure in special register XML-TEXT are encoded in ASCII. Because ILE COBOL operations such as move and comparison rely on EBCDIC encoding or on national characters for proper operation, you must convert the document fragments before using them. To do this when the XML document is in a COBOL program, first convert from the ASCII CCSID of the XML document to national characters using the MOVE statement. Then, if necessary, convert the result from national characters to EBCDIC using the MOVE statement.

Man muss also intern immer erst für jedes gelesene XML-Statement eine Zeichenkonvertierung durchführen, bevor man die Informationen weiterverarbeiten kann.