MB Isn't talking about using bank angle to increase descent rate (although it probably would to a degree). He is saying that using angle of bank to positively g-load the rotor-head while descending keeps you from the zero-g regime while still descending.
Lift vector goes sideways, smaller vertical component -> greater acceleration downward for the same (positive) g-loading. So, I shouldn't have said descent rate, downward acceleration is more accurate.
At least that's how I understand it. Semantics, of course. :icon_wink