John, you're the engineer with the scholarly background, so feel free to comment upon my thoughts.
Manifold vac is basically a measure (of sorts) of the restriction of air flow to the air pump (engine). Manifold vac is (crudely stated) inversely proportionate to throttle opening with respect to time.
When one is in gear churning some revs and letting the engine brake, manifold vac can actually exceed the idle reading (the engine, or air pump, is pumping, but the throttle is closed). Now along that line of thinking about the dynamics, if there is an exhaust restriction, there is a back-up of exhaust gasses. This starts to simply bog the motor down (one sign of a clogged exhaust is that under WOT, the car bogs or even dies, but if one backs way off the gas, where the exhaust gasses are not bottlenecking, the car runs alright). As the motor bogs, not as much air is needed, so even if the throttle-blade position doesnt change, not as much air is moving across the throttle blade, and the manifold vac reading decreases.
I know that was hard to follow. That is how I have thought about it in any case.