As Patrick said, loss aversion is present on the scales small enough for the DMU to not matter. Slightly more mathematically, when, given the utility function x->U(x), the gain vs loss utility ratio for the same change in the argument is small: Δx << U’(x)/U”(x). It does not invalidate the author’s point though, that there exists a phenomenon that is better described as the utility hysteresis: one ends up with less utility after gaining 2x and then losing x than after just gaining x.
As Patrick said, loss aversion is present on the scales small enough for the DMU to not matter. Slightly more mathematically, when, given the utility function x->U(x), the gain vs loss utility ratio for the same change in the argument is small: Δx << U’(x)/U”(x). It does not invalidate the author’s point though, that there exists a phenomenon that is better described as the utility hysteresis: one ends up with less utility after gaining 2x and then losing x than after just gaining x.