• Managed care and hospital bargaining

    Somehow Vivian Wu’s paper Managed Care’s Price Bargaining with Hospitals escaped my notice until recently (it was published in 2009). It’s loaded with content of interest, which I’ll run through below. First, to give you a better sense of the focus of the paper, here’s the abstract:

    Research has shown that managed care (MC) slowed the rate of growth in health care spending in the 1990s, primarily via lower unit prices paid. However, the mechanism of MC’s price bargaining has not been well studied. This article uses a unique panel dataset with actual hospital prices in Massachusetts between 1994 and 2000 to examine the sources of MC’s bargaining power. I find two significant determinants of price discounts. First, plans with large memberships are able to extract volume discounts across hospitals. Second, health plans that are more successful at channeling patients can extract greater discounts. Patient channeling can add to the volume discount that plans negotiate.

    This is rather intuitive, but quantifying the discounts associated with MC plan size, and ability to channel is hard. These factors are endogenous, requiring great care to estimate causal effects. (Wu uses plausible instruments, the descriptions of which are beyond the scope of a blog post.)

    First Wu reviews the theoretical and empirical literature hospital-plan bargaining as it pertains to plan size and channeling ability (related to its ability to substitute, the notion of elasticity, and its formation of exclusionary networks). She notes that the theoretical models in this area are “limited and incomplete.” She also discusses the role of excess provider capacity, though as you can tell from the abstract, she does not find that to be a significant factor.

    Using data from Massachusetts over years 1994-2000, Wu finds “that MC plans have 26–52% lower payments for hospital services compared with payments in an FFS plan.” Descriptively, it seems that even small plans can achieve large discounts, suggesting that plan size isn’t the only or even the most important factor in hospital price. However, both effects seem rather small:

    A one-standard-deviation increase in payer size leads to a 1-percentage-point discount. The channeling effect is slightly larger. A one-standard-deviation increase in channeling index one enables a MC plan to negotiate the price down by 2-percentage points. These estimated price effects appear to be quite small, given the descriptive evidence claiming that MC may have brought down the growth in health care expenditures to zero or below in the mid-1990s (Glied, 2003). Nonetheless, the actual impact could be larger, given that these estimates are averages from a mature period of MC (1994–1998) as well as from a MC backlash period (1998–2000). […]

    My results suggest that an increase in plan size alone does not directly translate into more bargaining power and, thus, lower hospital prices. However, a boost in the size of a health plan in the insurance market may increase the plan’s market power, which can lead to higher premiums for consumers. Given the findings that health insurance markets may not be very competitive (Dafny, 2008; Robinson, 2004;Wholey et al., 1995), regulators need to be cautious about insurer mergers.

    So, the question remains, how do MC plans achieve large savings if size and channeling don’t explain very much of it? Actually, as Wu made clearer to me in an e-mail exchange, there is a risk selection issue in comparing MC discounts relative to FFS. Maybe MC gets lower prices due to selection? To avoid this problem, the model Wu estimated does not include FFS. It is just a model of MC prices. So long as risk selection doesn’t vary much across MC plans, it’s a fine model. But it isn’t a model of discounts relative to FFS.

    In closing, Wu notes that her results pertain most directly to Massachusetts and may not apply to other hospital markets.

    Share
    Comments closed
     
    • Austin
      There is the:
      1) Evidence she presents
      2) “BIologic plausibility”
      3) What economic models tell us

      Granted, its Mass, like you cite, however, could the customers buying the products explain trends, ie, large vs small market (not just MCO size), MCOs at risk vs as TPAs? Other confounding variables?

      Another theory, as she mentions–the “on avg” possibility looking at early 90s. As markets rebooted with MC penetration, effects of size and channeling marked with reduction in prices. As steady state achieved however, discounting moderated. Plausible, but that should be apparent with regression, no?

      I am curious as her finding might relate to recent HA paper and CA markets. Does this paradigm fit—consistent with latest data?

      Thanks
      Brad

      • @Brad F – Not going into everything you wrote. Just saying, she controls for lots of stuff. In general, economists do. One would have to look at the paper to decide if her controls are adequate. I read it and was satisfied, as were the referees of JHE, a top journal.

    • I would think that MCOs could negotiate lower prices but my experience yesterday is inexplicable. A friend (age 30) recently had a routine upper GI endoscopy at UC Davis medical center. He is covered by Anthem PPO insurance. Anthem sent a statement which said that they paid $3300 (80%) of the $4100 bill for the endoscopy suite (not physician services). Medicare rates for this service (the only publicly available rates) are $600.
      Did Anthem really pay $3300 for this service? Where is the cost savings?

    • @Mark

      That’s the free-market system “we’ve already tried” you’re experiencing.

    • @Mark- That is a larger differential than I am used to seeing, but they are often quite large. Those huge differential are what make me think that Medicare does not act as a price support as many claim. I suspect it is the other way around, with Medicare paying just enough so that they can get docs to see Medicare patients. Maybe we need to work more on the cos aspects of private care.

      Steve

    • This is a good article on the channeling effect, but I have a question about this conclusion: “However,my estimated elasticity effect is much smaller than Sorensen’s… Therefore, Sorensen’s result is likely to have significantly overstated the channeling effect.”

      Another possibility: Wu’s data set was a large national employer in MA (probably Boston) from 1994-2000. Partners was formed in 1994. I would not be surprised if Boston post-1994 had a lower elasticity effect than most other hospital markets, due to the strong market power of this provider.