I was at the Academy Health National Health Policy Conference this week and it was wonderful. I learned a lot, I nerded out a lot, I saw a bunch of people that I mainly see on Twitter. I also am in the middle of a really good rethink on network choice structures.
Katherine Ho of Columbia made a very good, economist lens point that I want to paraphrase and then extend on.
Individual market products which are not just the ACA but also Medicaid Managed Care, Medicare and Medicare Advantage have a single decision maker and either a single individual reacting to the network quality or at most a single family group. If there is a sufficient network as defined by the single decision maker, the person screaming about a bad network is the decision maker. There is heterogeneity in defining what a “good” network looks like. For a lot of people a “good” network can be a very small network. Janet Weiner of Penn made the point that for most prospectively healthy people with ex ante good risk a “good network” can be what ever network that has their PCP and a local urgent care in network and everything else is almost irrelevant. This is the space for individual insurance market narrow networks that can lead to significant price per unit reductions.
On the other hand, group insurance products for large firms have the HR department being the decision maker. They face a budget constraint and a “keep most employees reasonably happy” sub-constraint. If there is a geographic dispersion of employees, knocking out one hospital in Town X and making all employees travel 15 miles to Town Y’s much cheaper and equally good hospital will produce a lot of screaming from employees. Smaller employer groups may be facing a much stronger budget constraint than large groups but this dynamic makes forcing large group insurers into narrower networks much harder. And if narrower networks are less plausible, then the negotiating leverage shifts to the providers which means higher rates for the providers.
I have no idea how to systemically measure or define a “good” network as the summary statistics wave away a lot of detail but this is forcing me to think really hard right now.