Options for Defining Medicare Advantage Regions: An Assessment of Tradeoffs




  1. Analysis
  2. Report from Market Participants
  3. Considerations Related to Competitive Dynamics
  4. Competitive Dynamics Scenarios
  5. Pros and Cons of Alternative Numbers and Types of Market Regions

The Medicare Prescription Drug, Improvement and Modernization Act of 2003 (MMA) (P.L. 108-173) was designed to achieve five goals related to private Medicare Advantage (MA) health plans and the Medicare program: (1) maximize the number of beneficiaries with access to MA plans, (2) encourage MA plans to enter areas not served by MA plans today, especially rural areas, (3) promote vigorous competition among MA plans in all markets, (4) expand the range of private plan types in MA, and (5) reduce long-term growth in program outlays. A key policy decision that will affect the achievement of these goals involves selecting the number of market regions into which the nation will be divided and the geographic boundaries of those regions. While leaving this decision to the Secretary of the Department of Health and Human Services (HHS), the MMA limited the secretary's discretion somewhat by mandating that the number of regions be no less than 10 and no greater than 50.

To encourage health plans to serve Medicare beneficiaries with regional PPO products as opposed to as a local plan, the MMA included special and temporary incentives as well as a two-year moratorium on plans initiating new local PPOs. The legislation also requires competitive bidding by MA plans, and sets out three important rules to govern the bidding process: (1) regional MA plans must bid one price for the entire region (local plans are allowed to bid one price for each county); (2) the government payment will be set equal to a benchmark premium (adjusted for relative beneficiary risk), so beneficiaries will have to pay  in addition to their usual Part B premium  the difference between the bid and the benchmark if the bid exceeds the benchmark; (3) benchmarks are determined differently for regional and local plans, and will generally not be equal for the two types of plans even though they may compete against each other in some counties. These different benchmarks could very profoundly affect market dynamics and program outcomes.

To explore the implications of the secretary's decision in light of these rules and the central goals of the MMA, the Office of the Assistant Secretary of Planning and Evaluation (ASPE) of HHS contracted with the Center for Studying Health System Change (HSC) to conduct the analysis documented in this report. The purpose of the report, therefore, is not to make specific recommendations to ASPE about which strategy it should use in drawing MA market regions. Rather, we sought to identify and analyze tradeoffs inherent in alternative strategies. In some sense, the valuations of these tradeoffs are not economic judgments, but political ones, which are best left to policymakers, not policy analysts. We can, however, inform the policy process by articulating what some of the key implications of these judgments are, relative to the very clear goals of the MMA.

This summary reviews HSC's approach to the analysis, the dynamics of plan competition that should be considered in drawing the MA market regions to achieve MMA goals, the scenarios that might come about as a result of these dynamics, and the pros and cons of some alternative numbers and types of market regions.


HSC's analysis of the implications of regional boundary choices comprises four tasks. We held discussions with a range of health plan and hospital executives who have experience with commercial insurance products and Medicare products. We read the MMA conference report and relevant statutory language in detail, along with some official commentary and analysis in the public domain. We analyzed the geographic variance in fee-for-service (FFS) Medicare costs and in Medicare+Choice (M+C) enrollment patterns, using the Centers for Medicare & Medicaid Services (CMS) and Dartmouth Atlas data and tools. And we explored how plans would be likely to behave in the face of different policy choices. That investigation was based on economic theory, policy analysis experience, and the findings from our discussions with market participants.

The economic framework we used to evaluate plan decisions and market dynamics centers on the decision to enter an MMA regional market in which local cost structures differ from one another. Two conditions are necessary (but not sufficient) for entry: (1) the plan must expect the long-run average bid price (P) to exceed long-run marginal operating costs (MC), and (2) the difference between P and the weighted average MC, however diverse MC might be within the region, must exceed the amortized expected cost of entry. The difference between P and MC will be determined by the extent of competition for beneficiaries, the bidding and payment rules established by Medicare, the provider contracts that a plan can negotiate, and the degree of adverse or favorable risk selection that is not accounted for in the risk adjuster for plan payments. For plans, entry costs are likely to be high and uncertainty about all of these factors is likely to be great under the conditions spelled out in the MMA. In discussions with market participants, we sought empirical evidence on how high these costs might be and why and on how much uncertainty exists.

Report from Market Participants

We held discussions with health plan executives who have launched managed care products and developed provider networks, as well as with hospital contracting executives, in a variety of geographic areas. We also spoke with national experts who are familiar with the MMA and with broader strategic calculations that health plans would have to make in deciding whether to enter a regional market. Key findings from these discussions follow.

  • Most observers expect MA regional boundaries to strongly influence the number and types of plans that choose to participate as well as their prospects for success. In general, the individuals we spoke with believe that creating smaller and more numerous regions will increase the likelihood that more plans will participate.
  • Plan executives believe that expanding networks beyond natural markets will reduce their negotiating leverage and diminish their ability to manage care. Expanded networks thus lead to higher costs of care and higher-priced insurance products. Plans would therefore be less likely to market aggressively in areas outside their natural boundaries.
  • Plans have little experience selling products for uniform, blended rates/prices across broad geographic areas because commercial product prices typically reflect both regional differences in cost of care and group member experience.
  • Precedents and models of plans serving members across multi-state markets exist, but each differs in important respects from the regional MA plan envisioned in the MMA.
  • Competitive dynamics between local and regional MA plans are difficult to predict, in part because payment methods will be based on one benchmark rate for local plans and another for regional plans and because the mechanism by which varying local costs are incorporated into regional benchmarks has not yet been established. Local plans are concerned about maintaining a level playing field among all plans.
  • The role that Part D and private drug plans will play in influencing the appeal of regional PPOs, particularly to Medicare supplemental policyholders, is unclear.
  • Most observers are seriously concerned that the timetable for implementing the regional plan strategy is unrealistic given the time and degree of difficulty associated with assembling multi-state networks. Those we spoke with fear that an unreasonable schedule will result in only a few, and possibly ill-suited, bidders.

Considerations Related to Competitive Dynamics

Competitive dynamics are driven by two differences between regional and local MA plans. First, regional plans must offer benefits at the same premium in a much larger geographic area. CMS will adjust payments for a regional plan to reflect variations in local payment rates within each region. At this point, CMS is considering a range of alternatives to do this. In one, only differences in input prices would be reflected; in another, differences in spending in Medicare fee for service (FFS) would be reflected. Second, the costs incurred by a regional plan to provide the benefits are likely to vary much more throughout this uniformly priced area than costs for local plans vary within a county.

Costs vary substantially within of a region because provider prices, enrollee service use patterns, and plan ability to manage care vary with both the market power held by local providers and historical patterns of care delivery. The greatest source of variation in Medicare FFS expenditures within regions appears to be use patterns, not the administered prices. But plans expect to face very different provider payment rates within MA regions as they develop provider networks. In areas where they have few enrollees or that have little provider competition, plans expect to pay high rates. For regional plans, creating a competitive network in rural areas may be particularly challenging, since the local providers' market power is virtually unchecked. Plans' lack of experience in either provider contracting or marketing to Medicare enrollees in sub-areas of a large region adds to their uncertainty by clouding their ability to predict costs and determine a bid.

No plan currently offers a single MA plan that covers a full region as envisioned in the MMA, so any plan would have to do some things differently to offer a regional product. Plans' experience will affect their ability and willingness to enter the regional market quickly. For example, Blue Cross Blue Shield plans, whose territories in most cases are statewide, typically have more extensive provider networks throughout a state than most commercial plans. If regions conform to 50 state boundaries, those commercial insurers that operate nationally or in many states might perceive themselves to be at a disadvantage in each state relative to the state's Blue plan and therefore be less likely to offer regional MA plans. On the other hand, if regional boundaries encompass a number of states or do not follow state boundaries at all, national or regional commercial insurers would have the advantage on the basis of their experience in operating in multiple states and since Blue plans would have to form joint ventures to cover a multi-state region.

The combination of a single premium bid and geographic variation in the costs of providing services to enrollees will create strong incentives for plans to attract enrollees from relatively lower-cost areas and to avoid enrollees from higher-cost areas. Plans have a number of tools at their disposal to draw enrollees disproportionately from lower-cost areas. One is to build their local MA products to serve the higher-cost areas that way. Another tool is selective marketing. Since marketing involves radio and television, newspapers, or appearances before local organizations, plans have several avenues to reach the sub-areas that they most value. Regulation could address this, but to be effective, it would probably have to be detailed and costly.

Competitive Dynamics Scenarios

Our economic analysis informed by the discussions with executives suggests that most of the competitive dynamics will be driven by the differences in benchmarks for local and regional plans. Under the MMA, the benchmark for local plans will be the MA payment rate in the county. But the benchmark for regional plans will be a blend of the average plan bids for the region and the average MA payment rate over the region.

The competitive dynamics will depend greatly on whether and how CMS adjusts for variation in local payment rates within a region. The MMA gives the secretary discretion on this issue. If the regional plan payment is uniform  corresponding to a single region-wide bid (for example, the average of county-level MA payment rates, weighted by Medicare beneficiaries), or if the geographic adjustments are relatively small  such as adjusting only for input prices  regional MA plans will have incentives to market in the low-cost areas and avoid high-cost areas. In contrast, if adjusters reflect variation in FFS Medicare spending, MA plans will have the same incentives as M+C plans had to avoid the low-cost areas. This would mean that the Congress's goals of giving beneficiaries in rural areas more choices of health plans would not be realized.

We outline three scenarios to describe the range of competitive dynamics that could occur with different types of boundaries for regional MA products. Most of the discussion assumes that the plan payment rates are uniform throughout the region or that geographic adjustments for variation in local payment rates are turn out to be less extensive than full reflection of Medicare FFS spending.

Scenario A. Regional MA plans dominate throughout the region. The basis for this scenario is that local plans today are concentrated in markets with the highest Medicare FFS costs, such as Miami and Los Angeles. Regional plans will cover lower-cost areas as well and would be able to offer products at lower average cost. This would permit them to offer lower prices to beneficiaries in higher cost areas than those charged by local plans and undercut them. Regional plans would market aggressively in lower-cost areas but would still serve beneficiaries in the high-cost areas, and over time, their share of the high-cost markets would grow.

If this scenario came to pass, it would likely take considerable time to play out because local plans are not only already established, but they also have a substantial number of enrollees, many of whom are both happy with their coverage and unlikely to switch immediately to a regional plan even if it is offering a more attractive package.

This scenario would be more likely  at least early on  with smaller regions that conformed mostly to state boundaries. Smaller regions would make it easier for plans to offer MA regional products, especially Blue plans, so there would be more regional MA activity. It is possible that the initial advantages of the Blue plans would deter national and regional plans from offering regional MA products  especially since they can continue offering local MA products. To the extent that this is the case, over time, larger multi-state regions could result in more competitors in each region.

If adjustment for local payment rates followed variations in FFS spending, Scenario A would be very unlikely to come to pass. Regional plans would not have any advantages over local plans in high cost areas, so that local MA plans would dominate in the areas that they now are serving. Few regional plans would probably be formed because to do so would involve offering products in areas that they have to date chosen not to operate under the similar terms. Plans would be accepting a lower benchmark to serve those areas that have to date been neglected by local plans because the payment rates are too low. If regional MA plans did in fact form, they would not have the advantage over local plans in high-cost areas.

Scenario B. Local MA plans dominate throughout the region. Under this scenario, the problems with the potential regional boundaries not corresponding to what plan executives refer to as natural markets would dissuade plans from offering regional MA products. The more geographic boundaries depart from natural markets, the more likely would be this scenario. Regions that have large metropolitan statistical areas (MSAs) at their core correspond most closely to natural markets and thus would attract the most regional MA activity. Regions that split MSAs or are vary large would be least attractive to regional MA plans.

Scenario C. Local MA plans and regional MA plans dominate in different parts of the region. Local MA plans would concentrate in the high-cost counties--like those areas they already serve. Regional plans would draw most of their enrollees from lower-cost areas, such as small metropolitan areas and rural areas. The MMA would, in a sense, be providing regional plans a subsidy to do this, since the contribution of the high-cost counties to the calculation of a benchmark for the region would provide an incentive to enroll beneficiaries living in lower-cost counties who are not currently being served by MA plans. If regional plans had an enrollee mix that drew more heavily from counties with low MA rates than from the overall Medicare population in the region, Medicare would be paying more than it does under current policies. But this overpayment would work toward another of the legislation's goals, which is to provide more of the Medicare beneficiary population with the option to choose a private plan.

Pros and Cons of Alternative Numbers and Types of Market Regions

The overarching goals of the MA portions of the MMA are to promote vigorous private health plan competition and more choices for beneficiaries throughout the country, especially in rural areas. The geographic boundaries of the market regions ultimately chosen by the secretary of HHS will profoundly affect the extent to which these goals are achieved.

Markets form naturally in some areas and not others, and this fact led to elements in the MMA that were intended to promote development of viable markets for MA plans in areas that have historically been served only by the Medicare FFS program. Natural Medicare markets can be expected to form where rates of service use and provider prices are relatively uniform, where effective provider networks for non-Medicare products have already been constructed, where the density of population allows plans to realize important economies of scale, and over areas that are linked economically so that travel and referral patterns (for secondary and tertiary care) are already established. In other words, natural health plan market areas are those in which a single premium is expected to be appropriate throughout the area for any product with a given actuarial value and target enrollee population.

Trying to create a market out of contiguous unnatural areas will be costly in some form, since inducements would then have to be provided to plans to get them to offer products. This effort will require either taxpayer-financed incentives or mandatory service requirements, with the latter likely to produce compensating distortions in plan behavior. In particular, forcing regional plans to offer the Medicare statutory benefit package at one bid price across an entire region with heterogeneous cost patterns will necessarily force plans to offer a blended or average premium, a kind of regional community rate. Thus, in the name of making a regional market, the MMA, in some sense, actually promotes distortions between prices and costs that do not exist in today's commercial markets and probably not at least to the same degree in today's MA markets either, since price and service decisions are still made at the county level. These distortions will most likely end up increasing Medicare outlays, but that may be a price worth paying in the minds of some policymakers if those beneficiaries living in areas not being served by MA today, such as rural areas, end up with more choices than they have today.

Distortions between price and cost send the wrong signals to plans about allocating their resources. For instance, if price is higher than average cost in an area, then plans will be overpaid and inclined to devote substantial resources to marketing and enrollment. If price is lower than cost, then plans will be underpaid and inclined to avoid enrolling beneficiaries where possible. Regional boundaries that minimize these distortions are therefore preferred to those that do not, all other things equal.

We assess three market boundary options that capture the range of feasible alternatives. Two options  10 multi-state regions and 50 state regions  have been discussed often. We developed an alternative designed to minimize the distortions discussed above. Regions would be built around large MSAs, with smaller MSAs and rural counties assigned to the most appropriate large MSA. Under this alternative, the number of regions could be limited to 50  which is specified in the MMA. In evaluating these options, we assessed the extent to which the varying approaches are likely to achieve the following goals: (1) minimizing price-cost distortions, (2) limiting the percentage of a region that is new and unfamiliar to current MA plans, and (3) minimizing the risk to plans when they expand their market to include rural areas.

MSA-Centered Regions. Price-cost distortions could be minimized by forming regions out of areas with homogeneous price and cost patterns, i.e., organized around metropolitan statistical area (MSA) markets that are largely already formed for commercial and current MA products. The second two goals could be reached by limiting the number of beneficiaries in counties outside of the natural market that are attached to each MSA core of the would-be regions. Rural counties would be included in the region built around each MSA, with each county assigned to the nearest MSA, which would reflect both travel distance and existing referral patterns.

Some MSAs (and surrounding rural counties) would be too small to be viable as regions for MA plans. They may not have enough Medicare beneficiaries to support viable regional plans, especially given the desirability of having more than one plan so as to have competition. Also, attention needs to be paid to the administrative burden for CMS of conducting bidding in a large number of areas.

There are 371 MSAs and over 360 Dartmouth Atlas Health Referral Regions (HRRs)With a threshold of 20,000 beneficiaries, 231 MSAs would qualify. With a threshold of 50,000 beneficiaries, 113 MSAs would qualify. Since the MMA limits the number of regions to 50, this could be the determining factor behind the threshold used to decide how large a Medicare population is required for CMS to decide that the MSA should be the core of a region that also includes smaller MSAs and rural counties.

A problem with having up to 50 MSA-centered regions is that many would include multiple states. This would imply regulation by multiple states. It would also pose difficulty for Blue plans, whose territories do not cross state boundaries. Joint ventures would have to be developed to offer MA regional products. We are told that with the exception of adjacent plans that have merged, these arrangements would not be easy to develop. Nevertheless, 50 MSA regions might conform more to the notion of a natural market than 50 states.

The number of commercial PPOs now operating in a potential region should not be a significant factor in deciding which areas are large enough to be an effective MA region because PPOs are so dominant in commercial markets. A measure of current competitors in the commercial market is not a proxy for the number of plans that will be willing to participate as regional PPOs in the MA program. If plans believe that there are opportunities for sustained profitability as a regional PPO in a given region, then both local commercial PPOs and new entrants may decide to participate. Conversely, if the business does not appear profitable, neither new entrants nor plans currently operating commercial PPOs in the market would be likely to participate. Indeed, in the mid-1990s, observers frequently reported to interviewers from the Community Tracking Study (CTS) that, in certain geographic areas, the Medicare risk contracting business was the most attractive of all health insurance products, drawing insurers into areas in which they had not previously operated on a commercial basis.

Ten CMS Regions. CMS has aggregated the states into 10 administrative regions. Having the MA regions conform to the CMS boundaries involves accepting a large degree of cost heterogeneity in exchange for bringing many rural counties and small MSAs into a region with large MSAs. This approach could engender many plan choices in areas that have not been served to date, such as rural areas, in the long -run, but would also most likely discourage the formation of regional plans in the early yearsbecause the short-term challenges of establishing effective provider networks over large areas are so formidable. The larger degree of heterogeneity in costs would also lead to larger increases in Medicare outlays than for regions that are smaller and more homogeneous.

Fifty State Regions. Administrative simplicity and fairly easy implementation support this option. Multi-state regulatory coordination would not be necessary and many Blue plans would find these regions familiar, although in states with multiple Blue plans, such as New York, joint ventures would likely be required. States vary considerably in the extent to which costs vary by county within their boundaries, but the variation is generally less than that within CMS multi-state regions. The process of creating statewide networks is not trivial, but it is much more feasible than a multi-state network, especially in the short run. The single greatest drawback to this option is that it would convey a large advantage to many Blue plans, which have statewide provider networks for at least some commercial products today. This advantage could deter other plans from trying to challenge the Blues as a regional plan. Thus, the greatest risk here is that long-run competition for rural beneficiaries may not be forthcoming as envisioned by some drafters of the MMA (unless local MA plans are enticed to enter the market because of the new higher payment rates).

In summary, drawing geographic boundaries involves difficult tradeoffs. Boundaries that will attract the most regional PPO products in the early years of the program may foster less competition down the road. But boundaries that look most promising in the long run pose risks that disappointment with the lack of participation could lead to abandonment of the regional MA plan concept in the short run and never getting to the long run. Adjusters for variations in local payment rates involve other tradeoffs. Regional plan payment rates that are uniform or that vary payment rates less than changes in FFS spending will encourage regional MA plans to operate in areas that local plans are not operating today. But with local MA plans in the high-cost areas, this will increase Medicare outlays. On the other hand, full adjustment for differences in FFS spending risks discouraging the development of regional MA plans and discouraging those that do form from investing to serve those areas that have not had Medicare private plan options.


"execsum_50.pdf" (pdf, 147.29Kb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®