Is it your own site? If so, you've got control over the meta "charset" line and can set it appropriately.
Entities are 100% safe but if you use a lot of them-- say you've got bits of text in other languages, or one short Slavic name ;) --your raw html becomes unreadable. Your call.
There are three kinds of entities. They can be intermixed without harm, except for the added risk of confusing yourself.
-- named html 4 entities like ü There is a finite set of these. My most ancient browser can only read non-ascii characters if they have named entities. (I don't have to use the entity, but it has to exist.) Everyone of course uses < > and & I also use — -- rarely – -- and for practical reasons although I could perfectly well type them.
-- decimal entities in the form &#nnn; These can be used for absolutely anything including vanilla ASCII.
-- hexadecimal entities in the form &#xnnn; Ditto. I use these for a simple mechanical reason: if I get a character I can't type in and have to look it up, my lookup gives the number in hexadecimal form. So conversion would require one more lookup. Otherwise there's no major difference.