I handled a similar problem (mainframe ebcdic data with german codepage to UNIX(AIX) with en-US) by creating a macro that contains the translate function for all the German "Umlauts" and applying that to all strings read from the mainframe. After that, a single change to the macro will have an impact in all programs using it, making maintenance quite easy.
Since SAS obviously does not support UTF in the dbms=xls in proc import 8as per your quoted article), I suggest to switch from xls to a more sensible file format. Use CSV or another text-based file format for transfer, and it is that much easier to deal with codepage and multibyte character problems.
... View more