The only reason I can think of is it might be used to identify the encoding as UTF-8. Could be useful for parsers when the encoding type is not specified, say in the XML declaration, or when there is a mismatch between the BOM and the declaration? Other than that, it's unecessary.