You're understanding is correct. As to why it's configured that way, there are "two schools" on bandwidth allocations.
One schools provides the more bandwidth to more important traffic, insuring that traffic doesn't lack for bandwidth.
The second school presumes if more important is only allocated the bandwidth it needs, so less important traffic might have a much larger allocation, to reflect the bandwidth not needed by the more important traffic. (What you have is likely the second school approach.)
Of course, much also depends on your SLA needs and the capability of the equipment. With the good QoS features, you can sort of merge both schools (but often not on switches which have weaker QoS features).
For example, you might have something like (on a router):
policy-map Sample
class Important
bandwidth percent 99
fair-queue
shape average percent 10
class Not_Important
bandwidth percent 1
In the above, the Important class has the bulk of the bandwidth allocation, also insuring a high priority for dequeuing, but it's also limited to only using 10% of the bandwidth. I.e. some misbehaving application won't "starve" the Not_Important traffic (and further, the FQ in the class will keep one or perhaps a few such "bad" flows" from "starving" other flows in that class.
