if the pads are extended, won't the holes in the stencils be extended, thus depositing even more paste onto the board.
Tuning the size of the solder paste apertures is a fine art. If a board is going off to be assembled by a competent shop, I'll leave the solder paste stencil exactly at pad sizes, and it's for them to tweak the holes (I'll tell them this, to avoid the possibility of a completely insane stencil being made!). They'll do things like shift them a bit depending on which way the squeegee is running over that panel, break up large areas into smaller sections, depending on their squeegee pressure and paste type. They'll cut the stencil trapezoidally sometimes, and turn pads into bowties if they like, and for difficult stuff (wide variations in paste requirements, for instance paste-in-hole and small SMT devices on the same board), they can run stepped stencils and / or do multiple stencil runs with different stencils. It's all incredibly process specific. And then there's the insanity that is package-on-package...
In short: If you're pushing the technology and want the best possible yields, this stuff is hard. If you're knocking out a few boards by hand, and can afford to inspect and fix up any problems, it's much simpler, just make the stencil pretty much pad sized. Big pads will let the squeegee bend down and scoop out some paste, so you don't get massive overpasting. Just try really, really hard to get the stencil registration right, or you'll be fighting it from then on.
I've been playing with my solder paste printer this week - got an intern in writing Python to convert pick&place files into usable plot files. So far, so good, but I'm not sure I'll be able to get individual spots onto 0.4mm QFN pads, they're evil (they're pushing the limits of what I can stencil, too).
Aaaand relax. This week just happens to be a 'wrestle with solderpaste' week...