Your idea of "Z-buffers" for 2d graphics is actually a layering system. Most 2D engines just draw things in order back to front, completely foregoing the need for a z-buffer and making things like alpha blending much easier to pull off. A 2D layering system will not help you with shadows and lighting. We're using a full 3D depth buffer setup because the way The Sims' graphics are drawn demands it, despite also having the problem of alpha blending and ordering, which is not solved in our engine, as you can see from numerous artifacts. Generally (meaning, in simple cases), you will not need this.