Estimating the Budgetary Effects of Legislation Involving the Center for Medicare & Medicaid Innovation

Posted by
Paul Masi
Tom Bradley
July 30, 2015

As discussed in yesterday’s post, this week we are featuring responses to some of the questions that we received from Members of Congress after hearings last month before the House and Senate Budget Committees. The posts highlight topics that have generated substantial interest among lawmakers and Congressional staff: Today’s will focus on the Center for Medicare & Medicaid Innovation (CMMI, which was created by the Affordable Care Act). You can view the complete set of questions and CBO’s answers from the June 3 hearing and June 17 hearing on our website.

Question: CMMI has broad authority to test different innovative approaches regarding the delivery of and payment for health care in the Medicare and Medicaid programs. How does CBO estimate CMMI’s effects on federal spending under current law? How does CBO estimate the budgetary effects of legislation that would affect CMMI’s current or future activities?

Answer: CMMI was established to test innovative approaches regarding the delivery of and payment for health care, primarily in the Medicare and Medicaid programs. Importantly, the Department of Health and Human Services (HHS) also was given broad authority to expand approaches that reduce spending and to terminate approaches that do not. Because CBO expects that some projects will prove successful and will expand, the agency estimates that CMMI will generate savings that exceed the costs of conducting the tests and trials. However, CBO cannot predict which projects will work. Over the 2016–2025 period, those net savings will total $27 billion, CBO estimates. Projects that CMMI has initiated are still in their early stages and so have not yet yielded noticeable savings; that result is broadly consistent with CBO’s current estimates.

CBO examines any legislative proposals that seek to enact approaches similar to ones that CMMI is testing, to determine whether HHS would do something different under the proposal from what it would do under current law. That analysis involves some judgment, which CBO must apply on a case-by-case basis. To the extent that legislative proposals overlap with initiatives that CMMI is undertaking (or is expected to undertake), the potential for additional savings is reduced. And if those proposals would delay implementation of promising initiatives or limit CMMI’s flexibility, they could increase federal costs. When such overlaps may arise, CBO tries to work closely with staff from the relevant committees as those proposals are being developed, both to understand the proposals and to explain the analysis. If those proposals are made public and contained in legislation that a committee approves or that comes to a vote in the House or Senate, CBO endeavors to make its budgetary analysis and the basis for that analysis public as soon as possible.

Given the substantial interest in issues related to CMMI, the discussion below further describes the context in which CMMI operates and how CBO estimates the budgetary effects of CMMI itself and of legislative proposals that may overlap with initiatives being tested through CMMI.

Background on Medicare Demonstrations and CMMI

Before CMMI was created, the federal government had long used demonstration projects to test new policies for Medicare. Those demonstrations were either initiated by HHS, acting under its statutory authority, or mandated by legislation. Demonstrations that HHS initiated were funded through its research budget, which has varied greatly, whereas funding for demonstrations mandated by legislation often came from that legislation. HHS initiated most demonstrations through the early 1980s, but legislation mandated most demonstrations that came later. For example, one recent study found that about 60 percent of the Medicare demonstrations that were planned or under way as of January 2008 were legislatively mandated.

Over the years, Medicare has conducted many demonstrations of new payment and service delivery models to determine whether they improve the quality of care, reduce program spending, or both. According to CBO’s analysis, the results have been mixed: Although most demonstrations have not reduced program spending, a few have. In the past, legislation mandating demonstrations typically did not give HHS the flexibility to modify projects on the basis of early experience, expand projects that succeeded, or terminate projects that did not succeed. In addition, the Medicare Payment Advisory Commission found that funding constraints have limited HHS’s ability to develop and test new models.

In light of that history, the legislation that created CMMI included several key provisions that influence CBO’s estimates of CMMI’s budgetary effects:

  • It provided dedicated resources for CMMI ($10 billion for fiscal years 2011–2019 and an additional $10 billion for each subsequent decade).
  • It created a mechanism to solicit, screen, and develop ideas for new models to be tested that is much broader and more rigorous than the development process that existed under prior law.
  • It gave priority to designing demonstration projects that could be scientifically evaluated using appropriate research methods, including requirements for sample sizes large enough to allow statistical analysis.
  • It gave the Secretary of HHS broad authority to modify and refine the models being tested midstream.
  • It created an incentive for the Secretary to end unsuccessful models by supplying a finite amount of funds to develop and test models ($10 billion every 10 years).
  • Finally—and most important for CBO’s conclusion that CMMI would reduce federal spending over a 10-year period—the legislation authorized the Secretary to expand models that proved successful.

Although CMMI has demonstrations under way or in development that are testing a diverse set of payment and delivery models, little information on potential savings is available. Designing, implementing, and evaluating demonstrations requires several years of effort. CBO expects that detailed assessments of some models will become available over the next few years. To date, detailed findings are available for only one model, which did not yield net savings. But those preliminary findings, covering only the first year of operation, may not indicate future results.

CBO’s Baseline Projections for CMMI

CMMI received initial funding of $10 billion in 2010 to identify, develop, test, and evaluate models through 2019. It will receive another appropriation of $10 billion for each subsequent decade beginning in 2020. Spending for CMMI has increased as the program develops new models (with models that start being tested in the same year commonly referred to as a cohort). CBO projects that CMMI’s spending will reach a steady state of slightly more than $1 billion per year in 2017, with total spending of about $11 billion over the 2016–2025 period.

Although determining the costs of operating CMMI is relatively straightforward, estimating how its operations affect federal spending is more complex. CMMI’s effects on the federal budget depend not only on the amount of spending for model development and evaluation but also on the amount of any savings or additional costs that the models being tested generate. As with CBO’s assessment of prior demonstrations, the agency expects that only a few models in any given cohort will reduce program spending. However, CBO cannot predict which models will succeed, and CMMI has not operated long enough to determine its overall track record. Therefore, CBO generally bases its projections on the following judgments, which reflect both the provisions governing CMMI and experience with HHS demonstrations:

  • On average, models that succeed will operate for four to seven years before HHS decides whether to expand them.
  • HHS will expand models that succeed, and expanded models will yield federal savings over time.
  • On average, models that do not succeed will operate for two to five years before HHS cancels them.

For most models that succeed and some that do not, those time frames include a two-year period in which to collect data, conduct an independent evaluation, and obtain actuarial certification of the results.

The upshot of those judgments is that—for a given cohort of models—CBO projects that federal costs will increase in the initial years. But those net costs will start to decline in year three and then will change to net savings as HHS expands successful models and cancels unsuccessful models. Across cohorts, the net budgetary effect of CMMI’s overall operations thus depends on how many operating cohorts are in their initial (cost-increasing) stages versus their later (cost-reducing) stages and on how many completed cohorts have resulted in cost-reducing expansions of successful models.

In total, CBO projects that CMMI will increase federal spending by about $0.2 billion in 2015, that the program will start generating net savings in 2017, and that the net savings in 2025 will amount to about $6 billion. Over the 2016–2025 period, CMMI’s operations are projected to reduce spending by about $27 billion, on net. Those net savings reflect the projected amount by which savings generated by successful programs—including those HHS will have expanded—will exceed the sum of additional costs from unsuccessful programs and CMMI’s spending to develop and evaluate models. Specifically, CBO estimates that savings totaling $38 billion over the 2016–2025 period will more than offset CMMI’s spending of about $11 billion on model development and evaluation over that period.

CBO monitors the entire CMMI process, from the collection of ideas for new models through the testing, refinement, and evaluation phases of models selected for testing. As the CMMI program matures, CBO also will monitor expansion decisions and the implementation of those decisions, updating projections to account for those decisions.

Estimates for Proposals That May Overlap With CMMI Models

CBO analyzes proposals for new demonstrations on a case-by-case basis. A crucial factor in the agency’s analysis is the nature and extent of any overlap with models that CMMI is testing or is likely to test under current law—and thus whether the proposal would yield results that differ from current law. In making that assessment, CBO accounts for how much interest CMMI has shown in similar models, either in terms of the tools it uses (for example, management of prescriptions before a hospital discharge) or the opportunities for savings that it targets (such as avoiding hospital readmissions). CBO looks at both tools and targets because multiple tools could target the same problem, but the resulting savings can often be realized only once. For example, although many approaches may seek to reduce hospital readmissions, a given readmission can be avoided only once.

CBO might estimate reduced federal spending for some legislative proposals even though they overlap with CMMI’s efforts. Indeed, CBO probably would estimate savings for an overlapping proposal under the following circumstances:

  • If the proposal would implement a model that HHS has decided not to expand, even though the evaluation showed that the model reduced spending and the Chief Actuary has certified that expanding the model would also generate savings;
  • If the proposal would speed up implementation of a model that is expected to generate savings (for example, CBO estimated savings for a provision that would have broadened a demonstration involving certain ambulance services that CMMI had already initiated in selected states); or
  • If HHS has decided to expand a successful model that would be implemented on a voluntary basis, but a proposal would make the implementation mandatory (for example, CBO estimated in 2013 that implementing bundled payments and reducing Medicare’s payments for those bundles by a specified percentage nationwide could generate federal savings, even though CMMI had already begun testing such approaches on a voluntary basis).

At the same time, CBO has often advised committee staff that proposals to implement models similar to ones being developed and tested through CMMI would increase federal spending. In general, that conclusion stems from one or more of the following considerations:

  • The proposal includes a statutory requirement that would delay implementation of a model under development and therefore delay realized savings if that model proved successful.
  • The proposal would limit the flexibility that HHS has to design and refine a model, thereby decreasing the likelihood that the model would succeed and reducing the expected savings if the model does succeed.
  • The proposal would permit or require expansion of a model without requiring that the model meet the cost criteria (including actuarial certification) in current law for expansion under CMMI. Such a proposal would make expansion of a cost-increasing model more likely and thus would increase spending in relation to CBO’s baseline—because under current law, the agency estimates that CMMI will terminate cost-increasing models.

CBO always seeks to consult with the staff of the relevant committees as they are developing their proposals to better understand the proposals, discuss any evidence related to their likely effects, and explain relevant analysis that the agency has undertaken.

Paul Masi is an analyst in CBO’s Budget Analysis Division. Tom Bradley is Chief of the Health Systems and Medicare Cost Estimate Unit in the Budget Analysis Division.