mIRC is UTF-8 encoding $chr(173) when it sends it to the server, because it falls out of the ASCII range (up to 127). That's what mIRC does now that it is a Unicode application. The problem is that there is no way to know if the original nickname (or channel) mIRC received was UTF-8 encoded or not, which gave birth to the following discussion: http://forums.mirc.com/ubbthreads.php?ubb=showflat&Number=223751&page=1#Post223751
In the meantime, you can use /raw -n to bypass this behavior when sending the ban command to the server. Good luck, though, knowing when a user's nick (or a channel) is UTF-8 encoded in the instance of the server, and when it's not.