They point out that since people often underutilize care (e.g. don’t always take their pills) and their underuse is pretty common for treatments that would help their health substantially (rather than only not taking pills that aren’t that helpful), then lowering co-pays can effectively increase the amount of good treatment they are getting. In other words, low or even negative co-pays act as Pigouvian subsidy that corrects for behavioral reasons why people don’t take enough of their medicine, even when it really helps.
They then point to evidence that lowering co-pays can improve health outcomes materially, suggesting that the marginal treatments (or pills) that people get when copays are low aren’t necessarily less valuable than those treatments that they get when co-pays are high. In short, taking behavioral hazards into account may suggest that health insurance co-pays may be too high. The balance of these behavioral issues vs. the standard moral hazard stories is ultimately an empirical matter. Either way, the idea is quite interesting and counteracts conventional thinking about optimal health insurance coverage.