Competition and the Cost of Medicare’s Prescription Drug Program

The Medicare Prescription Drug, Improvement, and Modernization Act of 2003 (generally referred to as the Medicare Modernization Act, or MMA) substantially expanded the federal Medicare program by creating the prescription drug benefit known as Part D. In fiscal year 2013, Medicare Part D covered 39 million people. The federal government spent $59 billion net of premiums on Part D in that year; after accounting for certain payments from states under the program, the net federal cost was $50 billion, which represented 10 percent of net federal spending for Medicare. A combination of broader trends in the prescription drug market and lower-than-expected enrollment in Part D has contributed to much lower spending for the program—about 50 percent lower in 2013—than CBO projected when the MMA became law in 2003.

Most beneficiaries of Part D choose among private drug plans to receive their coverage; others have employment-based coverage subsidized by Medicare Part D. The competitive design of Part D enables it to adapt flexibly to changing conditions, because plan sponsors (private insurance firms, each of which may offer several different plans) have ongoing incentives to develop new ways to control drug spending so as to minimize their costs, keep premiums low, and attract enrollees. Using the first few years of data from the Part D program, CBO found that spending was lower in years when, and in areas of the country where, more plan sponsors competed for beneficiaries. CBO's analysis suggests that competition between plan sponsors in Part D could be strengthened further, and costs lowered further, through certain changes in the rules of the program, although such changes could have disadvantages as well.

Other government programs use different approaches to deliver prescription drug benefits and hold down the costs of those benefits. In particular, the joint federal-state Medicaid program does not rely on competition between plan sponsors to constrain drug costs; instead, the program's chief cost management tool is statutory rebates that are applied to market-based prices. CBO found that Medicaid pays lower prices than Medicare, on average, for the mix of prescription drugs purchased by Medicare enrollees, primarily because the rebates that the law requires on brand-name drugs under Medicaid are larger than the ones that plan sponsors negotiate with manufacturers under Part D. If policymakers implemented Medicaid's statutory rebates for Part D beneficiaries with low income, but otherwise left Part D unchanged, federal costs would be reduced substantially in the short term. However, firms would respond by charging higher prices before rebates for new drugs (thereby probably offsetting a substantial portion of the savings for the federal government over the longer term) and by curtailing drug innovation.

Why Has the Part D Program Cost Less Than Anticipated?

Broad national trends in the prescription drug market have contributed significantly to the lower-than-expected spending for Part D. Many health care analysts, including those at CBO, expected in 2003 that growth in national drug spending would slow from the rapid rates observed in the late 1990s and early 2000s, but the magnitude of the slowdown that occurred surprised many observers. Drug spending per person for the country as a whole increased by only 2 percent per year, on average, between 2007 and 2010, compared with average growth of 13 percent per year between 1999 and 2003, the five-year period before enactment of the MMA (see the figure below). Drug spending per person in Part D also increased by 2 percent per year, on average, between 2007 and 2010. The greater-than-expected slowdown that began after 2003 caused national drug spending in 2012 to be about 40 percent less than the amount predicted by analysts at the Centers for Medicare & Medicaid Services in 2003.

Two developments accounted for much of the slowdown in growth of national drug spending per person:

• Many existing brand-name drugs lost their patent protection and faced new competition from generic substitutes, which have the same active ingredients as their brand-name counterparts but are much less expensive. Between 2007 and 2010, the share of prescriptions filled with generic drugs increased from 67 percent to 78 percent nationwide (and from 63 percent to 73 percent in Part D).
• New brand-name drugs (which tend to be more expensive than older brand-name therapies) were introduced at a slower rate than in the late 1990s.

Spending per beneficiary in Part D has been lower than CBO projected in part because of those developments affecting nationwide drug spending.

In addition to spending per beneficiary, enrollment in Part D has been smaller than CBO initially projected – by about 12 percent in 2012. CBO initially projected the share of eligible people who would enroll in Part D on the basis of enrollment in similar government health care programs—in particular, Part B of Medicare, which is a voluntary program that primarily covers physicians' and outpatient services for the same population that is eligible for Part D. CBO adjusted that share downward slightly to account for potential enrollees who would have prescription drug coverage through another source, among other factors. But the share of Medicare beneficiaries enrolling in Part D has been substantially lower. One contributing factor is probably that beneficiaries need to make an active effort to enroll in Part D—unlike Part B, in which beneficiaries are usually enrolled by default and must take steps to opt out.

Taken together, the unexpected slowdown in national drug spending per person and smaller-than-expected enrollment in Part D can account for nearly all of the difference between CBO's original estimate and actual Part D spending. CBO's original estimate incorporated an expectation that elements of the program's design that were intended to foster price competition between private plans would help to limit costs per beneficiary. Because other factors have affected costs per beneficiary, determining whether the competitive design of the program has been more or less effective than CBO originally anticipated is not feasible.

How Has Competition Between Plan Sponsors Affected Part D Spending?

Medicare Part D was designed to foster competition between plan sponsors to constrain drug spending. In assessing the impact of competition, CBO found that a larger number of plan sponsors in a region was associated with lower bids, on average, for the group of plans analyzed.

Each summer, every Part D plan submits a bid that reflects the total amount it would be willing to accept to offer Part D coverage for a Medicare beneficiary of average health for the following year. Once the bids from all plans have been submitted, the government determines the amount it will pay toward the benefit for the average beneficiary. The premium for each plan depends on the difference between a plan's bid and the government's payment. Plans with lower costs can submit lower bids and thus offer lower premiums and attract more beneficiaries. Other features of a plan, such as its cost-sharing provisions and the specific drugs it covers, also influence a plan's attractiveness to potential enrollees.

CBO analyzed bids for "basic stand-alone" Part D plans between 2006 and 2010 and found that plans in regions with more plan sponsors tended to have lower bids and premiums than those in regions with fewer sponsors. (Basic stand-alone Part D plans, which accounted for about half of total Part D enrollment over that period, offer a standard level of prescription drug coverage; CBO excluded from its analysis of plan bids stand-alone plans that offer more generous drug benefits, employment-based plans, and plans that combine drug coverage with coverage for other medical benefits, such as hospitalization and physicians' services.) Between 2006 and 2007, an average of 6 new plan sponsors joined the market in each of the 34 Part D regions that together cover the United States, contributing to lower bids and lower government spending. However, between 2007 and 2010, the average total number of plan sponsors per region fell by 4 (from 22 to 18), because more sponsors exited the market or merged with other sponsors than entered the market; that decrease in competition is associated with higher bids and higher government spending.

As Part D is currently structured, two features of the program could be changed to encourage plan sponsors to submit lower bids for their plans. First, in the component of Part D that serves low-income beneficiaries, the government usually pays the full amount of a plan's bid up to a threshold, regardless of whether other plans bid lower. Second, low-income beneficiaries enrolled in plans whose bid rises above the threshold are automatically reassigned in equal proportions to plans with bids below the threshold (unless a beneficiary has actively signed up for a particular plan). Both of those features encourage plans to set their bids close to (though below) the threshold.

Has Growth in Payments to Part D Plans Been in Line With Growth in Drug Spending?

The payments to plans by the government and beneficiaries for the basic Part D benefit increased more rapidly between 2007 and 2010 than did spending for drugs by those plans. Specifically, the payments to stand-alone plans for the basic benefit grew by 3.3 percent per year per beneficiary, on average, whereas plans' spending per beneficiary on drugs for the basic benefit grew by an average of 2.8 percent per year.

The difference between those growth rates represents an increase in the sum of plans' administrative costs and profits over the 2007–2010 period. Drawing firm conclusions about the cause of that increase is difficult, in part because of the short time frame of the analysis and a lack of information about whether the initial amounts of administrative costs and profits were unusually low. Nonetheless, some increase in the sum of administrative costs and profits could be explained by the reduction in the number of plan sponsors between 2007 and 2010.

How Do Prices for Drugs Differ Between Part D and Medicaid?

For the drug classes representing the great majority of drug spending by Part D beneficiaries, CBO found that Medicaid's average price for drugs was between 27 percent and 38 percent lower than Part D's average price in 2010 after controlling for differences in health conditions between beneficiaries of the programs. (Prices are measured net of rebates.) CBO expects that the difference in average prices will narrow over time as drug manufacturers respond to new rules that increased Medicaid's rebates beginning in 2010 but that Medicaid's average price will remain at least 20 percent to 30 percent lower than Part D's average price after controlling for differences in health conditions.

The difference in average drug prices between Part D and Medicaid in 2010 occurred primarily because Medicaid's statutory rebates on brand-name drugs were generally much larger than the rebates on those drugs negotiated by plan sponsors in Part D. Rates of generic drug use were similar in the two programs, so the higher prices paid by Part D for brand-name drugs were not offset by significantly greater use of generic drugs. However, the higher prices for given brand-name drugs were offset in part by greater use in Part D of lower-priced drugs within therapeutic classes (groups of drugs that are intended to treat common sets of medical conditions and that typically have similar modes of action). If the different patterns of use within therapeutic classes stemmed entirely from differences in the structure of the programs, then the lower end of those ranges—27 percent in 2010 and at least 20 percent in the future—reflects the relative effectiveness of those program structures in containing drug prices. If, instead, the different patterns of use within therapeutic classes stemmed entirely from differences in health conditions between beneficiaries of the programs, then the higher end of those ranges—38 percent in 2010 and at least 30 percent in the future—reflects the relative effectiveness of those program structures in containing drug prices.

Some policymakers have proposed applying Medicaid’s statutory rebates to drug purchases made by Part D beneficiaries who receive low-income subsidies (while retaining the existing structure of Part D in other respects). CBO expects that adopting such a policy would lower the average cost of brand-name drugs in Part D and thus reduce the federal government’s costs over the first decade after the policy was adopted. But a substantial portion of those savings would probably erode over time because drug manufacturers would counter the larger rebates by raising the prices for new brand-name drugs. In addition, that policy would reduce the incentive for firms to develop new drugs.