|
MarketWatchThe Cost Of Medicaid Annuities
Medicaid annuities are annuities that long-term care recipients use to shelter assets, thereby qualifying them early for Medicaid eligibility. As such, these annuities have the potential to increase Medicaid costs. This study estimates the cost of annuities to the Medicaid program. From a sample of Medicaid applications in five states, we found the rate at which annuities were used and simulated their cost to Medicaid. We estimated that in 2004, Medicaid annuities cost Medicaid about $197 million, which represented a small proportion of Medicaid’s almost $50 billion cost for nursing home care.
THE PERCEPTION EXISTS that the shifting of assets by many people in long-term care (LTC) transfers costs to the taxpayer—through the Medicaid program—that rightly should be borne by institutionalized beneficiaries or their family members.1 One such tool for shifting assets is a Medicaid annuity. The elderly purchase such annuities to reduce the amount of assets that are countable when determining their Medicaid eligibility, thereby helping them qualify earlier for Medicaid coverage of nursing home care. How annuities work. Annuities convert assets such as savings accounts, certificates of deposit, and stocks that are countable under Medicaid eligibility rules into assets that are not countable. Annuities generate fixed monthly or annual income payments and often are used to provide sheltered income to a spouse who is living in the community. Even if no spouse exists, an annuity can slow the rate at which assets are spent down during a nursing home stay and help preserve an estate. Not all annuities qualify as noncountable assets, however. To qualify, the annuity must be "actuarially sound"; that is, it must be paid out within the life expectancy of the annuitant, and it must be "irrevocable and not assignable" (that is, it cannot be redeemed or sold). Annuities sold as Medicaid annuities are generally fixed, single-premium immediate annuity (SPIA) instruments, as opposed to the more common variable annuities, which offer a range of investment options that can include stocks and bonds. The return on variable annuities is not guaranteed and depends on the value of the underlying investment option. Variable annuities often form an important part of a person’s retirement planning and would not constitute a Medicaid annuity for sheltering assets from the Medicaid program. Some states have imposed further restrictions on Medicaid annuities—such as limiting the purchase of annuities with balloon payments at the end of the annuity period—but no uniformity exists.2 Thus, annuities can be used to shelter assets in most states, but with variable levels of restrictiveness. Previous estimates of prevalence and costs. Estimates of the prevalence of Medicaid annuities, and their cost to Medicaid, are difficult to find. The U.S. Government Accountability Office (GAO) recently examined asset transfers by the elderly. The study did not focus on annuities but found that about 22 percent of elderly households reported transferring an average of $3,000 cash in 2002.3 It also found that in 2002 the median household income was $24,200, and median nonhousing resources were $51,500. We discuss below the difference between the GAO’s findings and the average household wealth for annuity holders in our study. The most recent estimate comes from a 2002 survey of forty state Medicaid directors.4 The authors first determined the number of private-pay patients in each of the responding states and assumed that 10 percent of these patients had spouses living in the community and that all couples had annuities. The cost of the annuity to Medicaid was assumed to be the annual Medicaid rate, net of the average portion paid by the patient. Multiplying the average annuity cost by the estimated number of private-pay patients with a spouse in the community led to an estimate of $637 million in excess resources that were sheltered by commercial annuities. This, in turn, translated into $1.16 billion in all states. We believe that this number is too high, for a number of reasons. As we show here, the authors overestimated the number of households with annuities; should have excluded states that do not allow Medicaid annuities; and also overestimated the average cost to Medicaid of an annuity because they counted the entire Medicaid payment, excluding only average patient payments.
In this paper we present the results of a study conducted in 2005 for the Centers for Medicare and Medicaid Services (CMS) that examined the use of annuities by Medicaid LTC beneficiaries in five states.5 We provide a new estimate of the cost of annuities to Medicaid based on the applications of those who applied for Medicaid and, in particular, those who had purchased Medicaid annuities within the "look-back" period (generally thirty-six months from the time of their application to Medicaid). We base our estimates on the data contained in these case files and on a simple model we developed that takes into account the specific resource and income constraints imposed by federal and state rules and policies for Medicaid eligibility. From the data collected in the five study states, we extrapolated to other states across the country to derive an estimate of the total cost to Medicaid. Categorizing states. Because policies on annuities differ from state to state, we categorized states on the basis of the "restrictiveness" of their annuity policies. Although we tried to rely on objective measures for categorizing states, we also interviewed state Medicaid officials and county eligibility workers in the five states to better understand why differences in the level of restrictiveness existed. It was clear that all officials were concerned with spousal impoverishment and wanted to avoid inhibiting those in need from becoming eligible for Medicaid. Most of those we spoke with believed that the federal and state rules in place were sufficient to prevent spousal impoverishment and that annuities often went beyond the rules by allowing some of the elderly to take advantage of annuities to avoid paying for their own care. Part of the officials’ perception of whether further restrictions were warranted seemed to be based on specific annuity cases they were familiar with.6
We identified four restrictiveness categories. The groupings are partly subjective but do serve as a useful starting point. Group 1 states did not allow the use of annuities to qualify for Medicaid, Group 2 states restricted or limited their use, Group 3 states had few restrictions, and Group 4 states had few restrictions and an elder law association and financial planning industry that actively promoted the use of annuities (Exhibit 1
Rates of annuity use. Because of limitations in the project’s scope and budget, we collected data from two counties in Arizona, Maryland, Nebraska, and Pennsylvania and from three counties in Missouri. We identified two of the most populous counties in each state and conducted site visits to review Medicaid eligibility applications in each. The existence of an annuity could be determined from an entry on the application, and there was usually a photocopy of the financial instrument (or equivalent) attached as an appendix. We also recorded the number of Medicaid applications that we screened, to calculate the ratio of the number of Medicaid annuitants to the number screened, which we used to represent the rate of annuity use in the state.7 Although we expected to find as many as sixty annuities in each state, we found far fewer in four of the states. This was attributable in part to the unexpectedly low rate at which annuities appeared in the Medicaid applications (conventional wisdom suggested a rate of perhaps 5–10 percent), as well as the time required to collect these data. In the final state sampled, Arizona, we attempted to compensate for the lack of data by oversampling the number of application files.
Price of annuities.
The average price of an annuity was highest in Maryland (in part because of one very large annuity) and Missouri (Exhibit 2
Demographics of annuity holders. Demographic data of beneficiaries with and without annuities (not shown) supports what we expected: that beneficiaries with spouses are much more likely than those without to purchase annuities. The reason, as we show shortly, is the protection offered to the community-dwelling spouse in terms of income belonging solely to that spouse, which is ignored when Medicaid eligibility is being determined for the institutionalized spouse.
State Medicaid parameters.
The Medicaid eligibility application forms that we reviewed contained information on the annuitant’s assets, income, annuity cost (purchase price), monthly (or in a few cases, annual) payment from the annuity, and marital status. Using these data, along with information on the average monthly private nursing facility payment, the average monthly Medicaid payment, and the specific state Medicaid eligibility parameters (Exhibit 3
The average private monthly payment and average Medicaid monthly payment were based on the actual private and Medicaid expenditures, respectively, for nursing home care. The asset spend-down limit is the maximum level of assets that the applicant could have and still qualify for Medicaid. The community spouse asset allowance indicates the amount of jointly held financial assets that would not be subject to spend-down regulations because they had been allocated to the applicant’s community spouse. The minimum income allowance is the minimum amount of income that a community-dwelling spouse is allowed to claim (and can be supplemented if necessary), and the personal needs allowance is the amount of income that the applicant could retain for incidental purchases while in the nursing facility. Finally, the divestiture penalty represents the amount of divested assets required to eliminate one month of Medicaid eligibility that could be imposed if the applicant divested assets during the thirty-six months prior to entering a nursing home. Lengths-of-stay. In the simulation, we used a length-of-stay of 2.5 years or 30 months for Medicaid patients. Because the case files represent "open" (ongoing) stays in a nursing home, we could not measure lengths-of-stay, so we relied on this overall average, supported by three empirical studies.8
Calculating the cost of an annuity to Medicaid.
We first determined the patient’s countable assets and then simulated spending them down according to the average private monthly nursing facility payment, net of the patient’s available income. Once the patient had spent down, Medicaid was assumed to cover the remainder of the thirty-month stay (net of any divestiture penalty and the patient’s available income). Calculating the cost of long-term care to the institutionalized person and the Medicaid program is usually fairly straightforward, but it is more complicated when a community-dwelling spouse is present and when the couple has substantial assets, including annuities. In contrast to the average (nonhousing) wealth found by the GAO, which averaged about $51,000 for all elderly families, the applicants with annuities in our study generally had substantial assets. For example, the average Maryland annuitant had about $62,000 in other countable assets (not counting the annuity, which in Exhibit 2
Therefore, we developed a simple model that took account of the Medicaid parameters as well as the appropriate resource and income constraints for the institutionalized spouse in each state.9 The implications of our model are that Medicaid only pays above and beyond what is available after the spending down of assets and after any contributions from the institutionalized spouse. Therefore, we refer to the base case as our interpretation of the case file, where the family unit (an individual or married couple) has purchased an annuity, most likely for the community-dwelling spouse or, if there isn’t a spouse, for the patient or any children. We then calculated the counterfactual Medicaid costs, where the annuity premium was included as part of the countable assets, and any annuity income paid to the patient was omitted. If the annuity income had been paid to the community-dwelling spouse, this income was also omitted. The difference between these two estimates is then our best estimate of the cost effect of the annuity on Medicaid, which we denote in Exhibits 4
A second estimate of the cost to Medicaid, one that represents an upper-bound effect, is the purchase price of the annuity itself. In other words, if the Medicaid recipient purchased a $100,000 annuity, the purchase reduces the total countable assets of the family unit by $100,000, which could have been used to pay for LTC. One advantage of this simple measure is that state and county Medicaid officials often cite it as their perceived effect of an annuity on program costs. However, large annuities would not reduce spend-down dollar for dollar, given the average Medicaid payment rate and the thirty-month stay we assumed in our calculations. For example, if the Medicaid payment rate were $4,000 per month and the annuity purchase price were $100,000, the upper-bound cost effect would be $100,000 (the lesser of $100,000 and $120,000).
Costs in each study state. Exhibit 4  shows the results of the extrapolation of our estimates of the individual costs of Medicaid annuities in each state to the overall state average. Exhibit 2 showed that Arizona and Missouri had the largest annuity incidence rates of the five states and that these two states, together with Maryland, generally had most of the large annuities. We calculated an average effect on Medicaid of about $31,000 and $28,000 in Arizona and Missouri, respectively. Maryland, with a lower incidence rate than Arizona and Missouri, appears to have a larger average Medicaid cost effect—about $54,000. When we compared these values with what we call the upper-bound effect (based on the lower of the purchase price of the annuity or the amount spent for nursing home care), we observed that our estimate was generally only half of the calculated upper-bound estimate. This indicates, we believe, that the simple purchase price or nursing home costs overstate the cost to Medicaid and that it is necessary to develop a method to account for the relevant state Medicaid parameters and asset and income constraints.
Costs for all states.
Next, we determined the total number of Medicaid nursing facility patients in the study states and then extrapolated their annuity costs to all other states in their respective groups.10 These data represented Medicaid patients in 2001 and were trended forward to 2003 by using the growth in Medicaid patients that was observed in the Online Survey, Certification, and Reporting (OSCAR) system.11 Under the assumption that these numbers represent the total Medicaid patients in the state during 2003 (and our best estimate for 2004), the expected number of annuitants among them can be estimated by multiplying this number by the annuity rate for that state (from Exhibit 2 Our final calculation was to determine the total Medicaid costs attributable to annuities across the states by multiplying the average annuity costs by the estimated number of Medicaid annuitants. We felt that given the financial challenges facing state Medicaid programs, it was important to be conservative in determining the effects on total Medicaid costs and better to overstate the costs of annuities than to underestimate them. We already pointed out one estimate of the costs to the program—about $1.16 billion. We wanted to examine a range of estimates based on what we observed in the data we collected. For that reason, even though we included Nebraska in the Group 3 states originally, we based our estimates for Group 3, which contains the largest number of states of the four categories, only on Maryland when extrapolating to other states.
Exhibit 5
Medicaid payments for nursing home care costs totaled about $46 billion, which was almost half of total nursing home costs, in 2004.12 Therefore, the $197 million in Medicaid costs attributable to Medicaid annuities represents only about 0.4 percent of the annual Medicaid budget for nursing home care. This may be surprising to the many observers who have perhaps evaluated the cost of annuities to Medicaid by the amount of assets placed in annuities without recognizing that not all of these assets translate into Medicaid costs. The average premium payment was rarely more than about two times the estimated average cost to Medicaid calculated above, so even by that standard, the cost of Medicaid annuities is still small. One factor that reduces Medicaid annuities’ effectiveness in sheltering income is the minimum income allowance for Medicaid recipients with spouses living in the community. This allowance sometimes has the effect of transferring income from the Medicaid recipient to the spouse. The purchase of a Medicaid annuity, however, could reduce this existing way of sheltering income, because the spouse would have income from the annuity instead that could be used to achieve this minimum income level. Thus, from the perspective of sheltering assets, Medicaid annuities might be less effective for those whose spouses have little income of their own. Our estimate is much smaller than the previous $1.16 billion annual cost estimate—indeed, roughly one-sixth as large. We believe that our methodology more accurately captures the rate at which annuities are used in the entire Medicaid population and the cost of these annuities to Medicaid. Our estimate represents our "best guess" of the levels of the various parameters. We conducted sensitivity analyses using alternative estimates of (1) the stock of Medicaid patients in the system in 2004, (2) the percentage of the annuity premium that was assumed to represent the cost of the annuity to Medicaid, (3) annuitants whose annuities generated Medicaid costs, and (4) what constituted a Medicaid annuity. None of these factors persuaded us to depart from our conclusion that the cost of annuities to Medicaid was very small, relative to the size of the total Medicaid budget. IN THE CURRENT political environment, many perceive that Medicaid is rife with abuse and that much could be trimmed from its budget if loopholes such as Medicaid annuities were closed.13 The GAO report points out that the president’s fiscal year 2006 budget proposed saving an estimated $1.5 billion over five years by tightening existing rules on asset transfers, although the proposed changes would have little effect on the use of Medicaid annuities. Many states hope that federal rules allowing such instruments will be tightened or that current restrictions imposed in other states, which are often under litigation, will be upheld in the courts before they impose further restrictions. Finally, most of the Medicaid officials with whom we spoke felt that restrictions on asset transfers that did not take account of all assets, including annuities, would likely fail, as financial experts and attorneys would simply devise other instruments for transferring assets. Whatever may be done, our study suggests that closing the Medicaid annuity loophole alone would produce little in savings.
Robert Levy (levyr{at}cna.org) is a senior economist at the CNA Corporation in Alexandria, Virginia. John Nyman is a professor, Health Services Research and Policy, in the School of Public Health, University of Minnesota (UM), in Minneapolis. Mary Gabay is president of Gabay Consulting LLC in Silver Spring, Maryland. William Riley is an associate professor, and Roger Feldman is a professor, Health Services Research and Policy, at the UM School of Public Health. The authors express their appreciation to the state and county Medicaid officials in the five states—Maryland, Pennsylvania, Missouri, Nebraska, and Arizona—that participated in their study and provided them access to the information without which this work could not have been done.
| |||||||||||||||||||||||||||||||||||||||||||||
Top
Study Data And Methods
Study Results
