As I've said before, if you plan to write a game, don't write an engine. There are lots of good engines out there, most of 'em relatively cheap.
If you're working on a platform for which there's currently no decent middleware available, then you might be able to build a case for it. If, though, you're writing an engine for a well-established platform, it's most likely a waste of time and was started based on bad advice from books or colleagues.
If you're currently working on your engine, you should probably abandon it, and the sooner the better.
You'll thank me later.