The welfare impact of Medicare HMOs.

We estimate the welfare associated with the Medicare HMO program, now known as Medicare+Choice (M+C). We find that the creation of the M+C program resulted in approximately $18.7 billion in consumer surplus and $52 billion in profits from 1993 to 2000 (in 2000 dollars). This program most likely generated significant net social welfare. However, we find that consumer surplus is geographically unevenly distributed. Prescription drug coverage accounts for approximately 45% of the estimated consumer surplus for 2000. Consumer surplus increases in the number of plans in a county, and most of the increase in welfare is due to increased premium competition.

1. Introduction

* In 1982, Congress passed the Tax Equity and Fiscal Responsibility Act (TEFRA), which mandated the provision of managed care plan options to Medicare beneficiaries. Under the statute, the Health Care Financing Administration (HCFA), now called the Centers for Medicare and Medicaid Services (CMS), was directed to contract with health maintenance organizations (HMOs) to provide a managed care option to Medicare enrollees. Under Medicare+Choice (M+C), the current name for the program, Medicare enrollees can forgo the traditional, fee-for-service (FFS) Medicare program and enroll in a qualified HMO. The HMO agrees to provide health insurance coverage for all FFS, Medicare-covered services (Parts A and B) for the enrollee in exchange for a per-capita payment from CMS. In addition, HMOs are free to offer more generous benefits than available in the standard, FFS program. The rationale underlying TEFRA is that HMOs may be more efficient at providing care, thereby reducing federal Medicare expenditures. Medicare enrollees may also benefit from the option to enroll in competing HMOs that would probably offer a more generous benefit package than available in the traditional FFS program.

We exploit the institutional properties of the Medicare HMO market and use a market-level, plan dataset to estimate the welfare that Medicare beneficiaries receive from the M+C program. We use methods suggested by Berry (1994) to estimate the preferences Medicare beneficiaries have over differentiated HMOs. In this framework, utility is posited to be a function of the plan characteristics (both observed and unobserved by the researcher) and plan premium.

Our estimates indicate that the Medicare HMO program added approximately $18.7 billion in consumer surplus and $52.0 billion in HMO profits from 1993 to 2000) That is, the federal government's creation of an alternative health insurance market to Medicare's FFS insurance resulted in substantial welfare gains. In 2000, the average Medicare enrollee earned $113 in consumer surplus because the Medicare+Choice program was available. This translates into $62 in consumer surplus per M+C enrollee per month. To calculate the net social surplus created by M+C, we must compare the benefits of the program to its costs. The cost of the program to the federal government is the cost of M+C enrollees if they were to remain in the FFS system relative to its outlays for those enrollees in the M+C sector. We estimate that the program increased welfare if the cost of enrolling the typical Medicare HMO enrollee in the Medicare FFS program is not less than 71% of the average payment made to Medicare HMOs. A conservative estimate of the cost of insuring the typical M+C enrollee in the FFS system is 80% of the average FFS expenditure. Even under this conservative estimate, the total net welfare from the M+C program is substantial: $24.8 billion.

Our estimates imply that prescription drug coverage plays an important role in generating consumer surplus. In 2000, we estimate that drug benefits account for 45% of consumer surplus. This translates into an implied annual average value of drug benefits of $445 for those who enrolled in an M+C plan that offered drug benefits. In light of the recent debate over whether and how to add prescription drug benefits to Medicare, this finding has significant policy implications. Our results indicate that M+C plans can deliver valued prescription drug benefits, and thus policies that promote M+C plans as the vehicle to deliver prescription drug coverage to Medicare enrollees can generate significant welfare.

Although the benefits from the M+C program are large, they are not evenly distributed across the country. In 2000, most counties, accounting for approximately a quarter of all Medicare beneficiaries, did not have an M+C plan, and very few counties (less than 6%) had four or more competing plans. Our estimates imply that consumer surplus dramatically increases in the number of plans in a county. For example, in 2000, mean consumer surplus in one-plan counties is $.91 per beneficiary per month. In counties with four plans, mean consumer surplus is approximately 12 times greater than in monopoly counties. Since we have estimated structural parameters, we can decompose the sources of the differences in consumer surplus between one-plan and four-plan counties by performing several simulation exercises. We find that most (81%) of the difference in consumer surplus between one-plan and four-plan counties is driven by decreases in plan premiums due to increased competition. The remaining difference in consumer surplus is divided between the greater availability of drug benefits in four-plan counties (8%), increases in non-drug-plan quality (8%), and increases in product diversity (3%).

As a by-product of the welfare calculation, we can estimate Medicare enrollees' semi-elasticity of demand for HMOs and that figure is of independent interest. Our results indicate that demand for M+C plans is inelastic. We estimate the monthly semi-elasticity to be -.009 for a typical HMO--a $1 increase in the premium reduces HMO enrollment by .9%. Conditional on charging a positive premium, the median plan elasticity is -.33.

Not only are the results relevant for understanding the impact of the current Medicare system, they are also pertinent for evaluating Medicare reform proposals. Several commentators have observed that the current Medicare FFS system most likely creates a large welfare loss, and they have proposed replacing the fee-for-service system with a form of managed competition (e.g., Dowd, Feldman, and Christianson, 1996). The efficacy of these reforms will depend largely on their ability to generate meaningful competition between HMOs. Our analysis bears directly on this matter. We find that managed competition can generate significant consumer surplus for the elderly population if there is significant HMO competition. However, generating sufficient HMO competition appears to be difficult. Our results also hint at the potential political difficulty of implementing such reforms. According to our estimates, managed care plans appear to gain disproportionately from M+C participation, and programs that mostly benefit HMOs may be politically unpalatable to the populace.

The rest of the article has is structured as follows. Section 2 discusses the Medicare HMO program. Section 3 sets out our empirical framework, and Section 4 discusses the data. Section 5 presents the results. Section 6 concludes.

2. Medicare HMOs

* Medicare began to permit HMO participation in 1972. The Medicare program's interest in HMOs as an alternative method for delivering care grew during the 1970s, and it began experimenting with HMOs as demonstration projects in the late 1970s and early 1980s. By 1980, however, HMOs had virtually no role in the Medicare program--only one managed care organization had signed a risk contract (Group Health Cooperative of Puget Sound).

In 1982, with the passage of TEFRA, Congress mandated the provision of managed care plan options to Medicare beneficiaries. The statute allows Medicare beneficiaries to enroll in risk or cost contract HMOs. Plans wishing to offer a Medicare risk product sign annual contracts with Medicare's administrative agency, CMS, to provide benefits to individuals who voluntarily enroll with the HMO. In exchange for their participation, the plans receive a payment for each beneficiary they enroll. From 1982 until 1997, Medicare paid HMOs 95% of its projected cost (Parts A and B) to treat a similar enrollee in the FFS program, where "similar" was defined using age, gender, welfare status, institutional status, and location. This payment is called the average adjusted per-capita costs (AAPCC). Under this structure, geographic units are counties, so the CMS payment (the AAPCC) varies by county and over time. In 2000, approximately 39 million individuals were enrolled in one of Medicare's three main eligibility categories: aged (those over 65 years old), disabled, and end-stage renal dialysis. Medicare HMOs receive different payments for enrollees who are in the different Medicare programs.

In Table 1 we present the quartiles of the CMS payment rates in 1993 and 2000 (along with other summary statistics that we discuss later). The spread of the AAPCC in both years is rather large--25% of the 1993 median and 20% of the 2000 median. The determination of the AAPCC (until 1998) was based on a five-year moving average of Medicare's realized cost, with the moving average window starting three years prior to the current year. The payment rate in period t is calculated using Medicare's realized costs in years t - 3 to t - 8. There is a large variation in the AAPCC across counties and over time within a county.