I'm always amazed at how stupid the security situation is in these cases. Banks, telecoms services, etc. do actually call up and try to 'take me through security', and when I say "tell me something you know about me first so I know you're who you say you are", the best they can usually manage is "well, uh, you bank with [Bank]". It just perfectly trains us to fall for scams.
Now, it should be supported, but I don't want the folks on the front lines guessing (or figuring out on their own) what sorts of mathematical games are safe. Erring on the side of caution is the right approach for CSRs.
So given banks have nothing to lose by scams, I suppose that explains why they just don't care about the fact they're training users to ignore them. The bank just does whatever's easiest for it, which in this case is just to call the customer.