IDNs, HHI and switching costs

Warning — this is going to be a nerdy post even by my standards —-

Last week at the National Health Policy Conference, I attended the most depressingly enlightening session on the schedule. It was an examination of the concentration of both insurers and providers in the health care industry. There are a dozen powerful slides, but I want to look at one in particular as I think it is understating the story of the lack of competition.

#NHPC18 hhi by insurer and hospital industries

Competitive markets are rare pic.twitter.com/G43ZRnurJh

— David Anderson (@bjdickmayhew) February 6, 2018

This slide looks at the HHI index of insurers and hospitals in several hundred regions.  An industry is considered competitive if the HHI score is under 1,500 points.  It is considered somewhat competitive if the HHI score is under 2,500 points. HHI is calculated as the sum of the square of each entitie’s market share and then that sum is multiplied by 10,000.  A perfectly competitive market like Girl Scout cookies will have an HHI of 0 while a perfect monopoly will have an HHI of 10,000.

The regions that are down and to the left of the red lines are very competitive.  The regions contained within the yellow lines are somewhat competitive.  Every other region has at least the insurer or the hospital market as being very concentrated.

And I think this graph is understating the problem on a pragmatic basis.  Integrated Delivery Networks (IDN) like, my former employer, UPMC, like Geissenger, like Kaiser, like Sutter, like Medstar, are both insurers and hospitals.  Some IDNs are open access to both other hospitals and other insurers.  Some IDNs will have the insurer only contract with their owned hospitals but the hospitals take other insurers, while some IDNs are true walled gardens where the insurer only interacts with the owned hospital and the owned hospital only interacts with the owned insurer.

Let’s talk about the Southwestern Pennsylvania insurance and provider market.  Using traditional HHI, the hospital market  is a concentrated market.  UPMC being a dominant provider while the Highmark BCBS owned Allegheny Health Network (AHN) has half a dozen hospitals under its control south of I-80.  There are a couple of small chains (Excela with 3 in Westmoreland County, Heritage Valley with 2 in Beaver County, Washington PHO owns two hospitals near I-79 south of the city) and a decent number of independents.  UPMC and AHN are the two sources of very high end specialist care while Hopkins and the Cleveland Clinic act as relief valves for the very strange and unusual cases.

The insurer side is a bit more competitive.  UPMC has slightly more marketshare than Highmark.  Medicare Fee for Service is the third biggest payer in the region while the national carriers have never been able to build a large membership base.

Southwestern Pennsylvania is a double-concentrated market.  However I think the simple HHI analysis understates the concentration for most potential buyers.

Near Pittsburgh, the two big IDNs have fairly tight barriers against the other IDN except for Medicare Advantage.  UPMC broad network insurance products include UPMC owned facilities and doctors as well as most of the independents.  AHN owned facilities are out of network.   UPMC narrow network products are built on a core of UPMC owned facilities with some independents as gap-fillers.  AHN has the converse.  The broad network Highmark products have AHN and most of the independents.  The narrow network Highmark products have AHN and some of the independents at the first tier.

For people who get all of their care via the independent hospitals and docs, they see a market that is reflective of the calculated HHI.  If they are happy with their docs, they can switch UPMC from Highmark or vice versa with only the normal friction.

However for people who get their care from one of the two vertically integrated silos, the switching costs are far higher if they want to move their insurance.  To change insurance, they have to change docs.  For people with no medical needs and who barely touch the system, this is not a high cost.  For people with complex care needs, switching systems and thus mothballing all of the tacit knowledge of navigating a system is a massive friction.

A UPMC Health Plan covered life who gets all of their care in Oakland at UPMC hospitals can not easily switch to getting their care on the Northside at AHN Allegheny General.  If they don’t like their insurer, they have the option of switching to a national carrier if their employer offers an expensive national carrier.  If they are covered on Exchange, they have no ability to cheaply switch insurers while holding onto their docs and hospitals.  Medicaid beneficiaries probably have the easiest switch as there are multiple insurers that have overlapping networks in Allegheny County.

Pragmatically, IDNs with significant switching frictions reduce the ability of people to switch out which means the HHI understates the lack of effective competition in a region.  I don’t know what should be done.  I don’t know what can be done.  I don’t know what awesome literature has already answered this question and my next seven follow-up questions, but this slide was bugging me even after I asked a question that got a good answer but I am still groping towards a better thought on the Pittsburgh problem.