It is not only perfectly possible for there to be a role for early exchange for understanding the origins of currency systems. It is almost surely necessary.
the very purpose of the "barter story" is precisely to explain why barter cannot lead to the kinds of population and social changes we know happened in ancient societies. That something else necessarily had to happen.
Let's look into the actual ethnography.
Humphrey's primary ethnographic research for the paper Graeber cites as definitive was based on her research of the Lhomi of the Arun Valley in northern Nepal near the Tibetan border.
Of the Lohmi, Humphrey writes (1985, pp.54-55): "Before the virtual closure of the Tibetan border by the early 1970's which followed the Chinese invasion, the Lhomi engaged in three kinds of barter." These are: