In theory you are absolutely right. Excel can be used to import a file and then export it again into a "localized" csv format, but Excel enjoys converting long numbers like EAN codes or personal id (like your SSN) into so called scientific notation and when it sees something like 2/4 it must be April 2nd or whatever. So every import into excel takes a lot of time in that import dialog where I need to say for every column if it's a text or number or what.
In the end, I lose a lot of time in application A or application B. Plus I have a bigger problem with Excel in my country. The Mac version hates our diacritics (šđčćž) in text files. Even some Windows version are strange, so each time I import/export from Excel there is a great chance some, if not all of these characters will get lost.
But the big question of my first post was why is csv designed that way? Whay is separator volatile? Why is decimal separator unstable? What about dates? Shouldn't be better if the separator was always comma, decimal point defines numbers and application should take care of localized presentation of the data.