CDATA - strange behaviour by Xerces

I have this peace of code:

Document doc = db.newDocument();
Document docMessage = db.parse(a_message);
Element msg = (Element)doc.importNode(docMessage.getDocumentElement(), true);
Element data = doc.createElementNS("", "Data");
data.appendChild(msg);

well, the problem is that original XML looks like this (before parsing with 
XercesJ):

<msgData><![CDATA[<CC015A></CC015A>]]></msgData>

and after parsing with XercesJ and saving into file:

<msgData>&lt;CC015A&gt;&lt;/CC015A&gt;</msgData>

Why the difference? I more like first syntax. Can I achive it by using XercesJ?

Thank you in advance,
Milan

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com