Bluematter.

What happens when the insurer market becomes less competitive?

We test whether insurers that experience larger enrollment increases due to Part D negotiate lower drug prices with pharmacies. Overall, we find that 100,000 additional insureds lead to 2.5-percent lower pharmacy prices negotiated by the insurer, and 5-percent reductions in pharmacy profits earned on prescriptions filled by enrollees of that insurer.

The results seem genuine:

Estimated enrollment effects are much larger for drugs with therapeutic substitutes, and virtually zero for branded drugs without therapeutic substitutes.

And there's also this interesting bit:

We also present evidence that most insurer savings are, on the margin, passed on as lower premiums.

The authors also put the results in perspective:

Naturally, the optimal degree of competitiveness faced by manufacturers depends
both on efficient drug pricing, and the provision of sufficient incentives to innovate. Therefore, it is not clear whether policies to increase competition among manufacturers would harm future welfare by more than they enhance current welfare.

It's interesting how discussions on healthcare reform have focussed almost exclusively on the financial costs of health-care, the bulk of which represent transfers from consumers to health providers and big pharma, with very little discussion of deadweight losses beyond overheads, or any serious attempt at quantifying them.

We spend more time discussing how much we spend on health-care than how much we waste. For economic thinkers, this is very out of character.

And another random observation: I may be wrong here, but I feel like the (non-articulated) current concensus in policy discussions is more-or-less that dynamic effects are not important, and that spending has little to do with innovation and improved health outcomes in the future. Going further out on a limb, I think I can trace this change to Robin Hanson's campaign and his popularisation of the RAND study, but of course this is just a gut feeling.