Contexts
Element text needs & and < escaped (and > so that “]]>” can never appear). Attribute values also need the surrounding quote escaped, and tabs and line breaks written as character references, because XML normalises literal ones to spaces. CDATA cannot contain “]]>”, so it is split across two sections.
Unescaping decodes &, <, >, ", ' and numeric references exactly once: &lt; becomes <, not “<”. Unknown names such as (an HTML entity, not an XML one) are left as written and listed.
Questions
Why is not decoded?
XML predefines only five entities. exists in HTML or in a DTD; without that declaration it is an error in XML. Use   instead.
Is the unescaped result shown as HTML?
No. Output is always plain text in a text box, so decoded markup is never interpreted by the page.