As with most things space, it all comes down to tradeoffs. The most efficient rocket is one that is purely expendable and has no mass that does not contribute to getting the payload towards orbit.
If the aim is to reuse the rocket, you need a mechanism to achieve control in the upper atmosphere, a method to control descent rate and a method to achieve a landing that does not destroy the hardware (or just tough hardware). And do all of this without increasing the risk of the launch itself failing due to hardware for the return.
Wings are a partial option, they work well for achieving a controllable descent rate, but they do not do much useful at high altitude and need additional supporting hardware (wheels/runway/airbags etc) to achieve a soft touchdown and a fair bit of flight control smarts.
Using the existing rocket engine is less efficient than using wings, but it is something that is already there on the rocket and gives you a system with extra capability in expendable mode.
So the final decision involves a lot of trade offs that often have much to do with seemingly minor details and less on perfection of a single aspect. For example SpaceX has ambitions on the moon and mars, and both of those require mastering powered descent rather than wings.
SpaceX decision making
Wings during ascent
Plans to have wings on SRBs