It is a good question. I think one possible answer is that a custom EOS token is essentially a smart contract (sames as an ERC20 is) So the owner of an application that utilizes the custom tokenz would be able to do so, within a fixed state of bandwidth proportional to native EOS held.
Maintaining a low barrier of entry is relative to an abundance of native eos yet application bandwidth availabilty is expected to be independent, entirely of "any token price"
Curious to understand how application bandwidth availability is expected to be independent. My understanding is that they are directly interconnected for now. But I admit I haven't seen much economics by the way of tokens relative to bandwidth, storage, etc.
I'm going on what is in the technical paper. In that application independence is down to usage based on quantity of tokens held. Seems logical that being unaffected by fluctuactions in price is a major benefit. It ties in nicely with the concept of token property, renting and bandwith. Computational capacity allocation on a fractural reserve basis is a good description.
There's an indirect interconnection in that the higher a token market price the more resources may be availble per token yet an application would run within its resource alloction independent of price.