1. Hate front ends
or
2. Should never be let within a million miles of them
And for such a trivial saving I wouldn't want to sacrifice the coherent codebase, code reuse and inhouse knowledge of how it actually works. I would guess that the hidden support costs must be significantly higher too.
As for the "Front End" being fun I'd point out that the developers I'm working with are Information Systems people more than Computer Science people. They enjoy working with the actual users, planning features and then implementing those features. What the outsourcing does is allow them to say "I need this string parsed in this way so I can make the rest of this feature work" and just have it done for them rather than having to do the grunt work of writing the relatively easy but time consuming code that parses the string.
Essentially what I do is spec out the project very specifically and have local people write the front end. Then we outsource the backend and have interns use the locally built frontend to thoroughly test the backend delivered from the outsourced company.
This isn’t dramatically cheaper since you have to spec the thing locally and pay interns to test it while still paying the outsourced developers (it only saves about $2.65 an hour by my calculations). But it does have advantages in that it allows us to be more flexible with our schedule (we can employ more or less outsourced workers as needed) and it does make for a happier development environment (allowing local developers to focus on the relatively more fun front end development and less on the nuts and bolts backend stuff)
So in the end I feel it’s successful though again not so much in the financial sense.