In coming years the nation will face important decisions about how to. DataWatch

Transcription

1 DataWatch Long-Term Care Insurance And Medicaid by Marc A. Cohen, Nanda Kumar, and Stanley S. Wallack Abstract: Depending on length-of-stay, somewhere between 29 percent and 38 percent of long-term care insurance purchasers who use nursing homes would qualify for Medicaid payments if they did not own a policy. This is equivalent to between 13 percent and 17 percent of all policyholders. Owning such a policy would, however, reduce spend-down rates among policyholders by 39 percent. Thus, in the presence of long-term care insurance, only 8 to 10 percent of all policyholders would receive Medicaid. Medicaid would spend between $6,492 and $14,179 (in 1990 dollars) on nursing home entrants with long-term care insurance policies-$14,437 to $29,698 if entrants did not have insurance. The ultimate impact on Medicaid expenditures is reduced, however, because many policyholders voluntarily let their policies lapse before entering nursing homes. If policies paid reduced benefits for voluntary lapses, then Medicaid could reap more significant fiscal savings. In coming years the nation will face important decisions about how to provide and pay for the long-term care services needed by older Americans. In 1988 about seven million older people needed long-term care; within seven years that number is expected to increase to nine million. 1 There are now two primary sources of payment for long-term care: the elderly and their families (who pay about half of the roughly $60 billion spent on long-term care), and Medicaid, which pays another 45 percent of costs. 2 While most Medicaid spending is for acute care, Medicaid is also the major source of public funding for long-term institutional care. 3 By 1991 Medicaid expenditures on all institutional care approached $29 billion 4 Given the fact that the elderly and their families finance half of all long-term care costs, it is not surprising that a growing number are seeking alternative means to pay for care. One such method is to acquire long-term care insurance. By December 1991 more than 2.4 million such policies had been sold, many to persons ages sixty to seventy-five. 5 Still, less than 2 percent of total long-term care costs are now paid for by private insurance. Given the rapid growth in state Medicaid costs, state policymakers also are seeking ways to cut spending. In fiscal year 1992 states approved more Marc Cohen is a research associate at the Center for Health and Long-Term Care Research in Waltham, Massachusetts, and is a researcher at the JDC-Brookdale Institute in Jerusalem. Nanda Kumar also is a research associate at the center. Stanley Wallack is director of the Center for Health Policy at Brandeis University in Waltham and is chairman of Lifeplans, Inc., also in Waltham.

2 128 HEALTH AFFAIRS Fall 1994 than $16 billion in tax increases-the largest increase in history. In the same year Medicaid expenditures were on average 6 percent over budget 6 Some policymakers are looking at long-term care insurance as one way to help solve Medicaid s fiscal woes. They view the growth in sales of long term care insurance as leading to a reduction in state Medicaid expenditures. The underlying assumption is that if persons have long-term care policies, then they will not need to rely on Medicaid. This is because it is assumed that the number of persons who will become eligible for Medicaid as a result of having used up their income and assets to pay for care-the spend-down process-will decline. Thus, states like Connecticut and New York, for example, have initiated programs to encourage the purchase of long-term care insurance. To date, however, there is little empirical evidence to support the claim that widespread ownership of long-term care insurance policies will reduce Medicaid expenditures. In fact, what little research has been done in this area suggests that long-term care insurance will have no appreciable effect on Medicaid expenditures. 7 Study purpose. The purpose of this DataWatch is to estimate the extent to which the acquisition of long-term care insurance policies affects the use of Medicaid to pay for long-term care services and leads to reductions in total Medicaid expenditures. We base our empirical analysis on a study of 8,400 purchasers of long-term care insurance. We answer the following four questions: (1) What impact does the possession of a long-term care insurance policy have on the probability of spending down to Medicaid eligibility levels? (2) What impact does the possession of such a policy have on total Medicaid expenditures? (3) What are the demographic characteristics of long-term care insurance purchasers who spend down to Medicaid eligibility levels? (4) What are the long-term care insurance policy designs of persons who spend down to Medicaid eligibility levels? Sample. The data set comprises individual purchasers of long-term care insurance age fifty-five and over. Six companies that participated in the study, and together represented 4.5 percent of total individual long-term care insurance sales in 1990, randomly selected a sample of policyholders who had purchased policies in late 1990 and early Of the 13,800 persons sent mail surveys, 60 percent (about 8,400) returned them. Companies were requested to provide policy design information on the long-term care policy chosen by each person in the sample. The information included choice of benefit duration; daily benefit amount; deductible period; whether home care, inflation protection, or nonforfeiture provisions were included in the policy; and premium levels. This information was then linked by an identification code to each of the returned surveys. The profile of purchasers has been described in detail elsewhere. 8 In short, purchasers are relatively young, with an average age of sixty-eight.

3 D ATAWATCH 129 Only 17 percent of purchasers are over age seventy-five, compared with 25 percent of the general population. Also, purchasers reporting income and asset levels are wealthier than their counterparts in the general population, even though a sizable proportion are middle income: About three in ten purchasers have annual household incomes of less than $20,000, and about 25 percent have liquid (nonhousing) assets of less than $30, Finally, purchasers are more likely to be female, married, and college-educated than are their counterparts in the general population. Study Methods And Findings To determine the impact possessing a long-term care policy has on Medicaid eligibility and expenditures, we made several estimates: (1) the projected long-term care costs of policyholders; (2) the probability of spending down to Medicaid eligibility; and (3) the proportion of long-term care costs paid by Medicaid and by insurance. Estimating long-term care costs. To estimate policyholders future long-term care costs, we combined information from previous studies and relied on a number of national data sources. In brief, based on age and gender, policyholders were randomly flagged as entering a nursing home or not. The random assignment was based on two studies that evaluated the lifetime risk of nursing home use by age and sex.10 In total, about 44 percent of the sample, or 3,050 persons for whom complete policy information was available, were flagged as entering a nursing home. 11 To calculate the financial liability associated with a nursing home entry, we randomly assigned persons flagged as entering a nursing home to a length-of-stay category based on gender. Because there is not yet a clear consensus on the average length-of-stay of an admissions cohort to a nursing home, we relied on the findings of two studies that estimated the total amount of time nursing home entrants spend in a nursing home. 12 One study, by Peter Kemper and colleagues, reported a much higher percentage of long-stay nursing home patients than the other study, by Denise Spence and Joshua Wiener, reported. In fact, the Kemper study shows that 21 percent of nursing home entrants have stays of longer than five years, compared with about 10 percent in the Spence/ Wiener study. In the following analyses, each distribution is used to demonstrate the sensitivity of results to length-of-stay estimates. 13 An important issue is the extent to which the utilization data derived from the general population reflect likely use among an insured population. On the one hand, the presence of insurance should lead to higher utilization (moral hazard). Also, a common assumption is that insurers typically face adverse selection and hence higher-than-average utilization. However,

4 130 HEALTH AFFAIRS Fall 1994 evidence suggests that this may not be the case for long-term care insurance. First, the risk of nursing home use increases considerably with age.1 4 This implies that the impact of insurance company underwriting is likely to be greatest at older ages. Yet data show that the vast majority of long-term care insurance purchasers (83 percent) are below age seventy-four, and most (58 percent) are even below age seventy. 15 This means that most persons are purchasing policies five to fifteen years before they face major risk of entering a nursing home. Because most long-term care costs are not attributable to those under age seventy-five, the presence of underwriting probably eliminates most adverse selection at younger ages and minimizes it at older ages. Regarding moral hazard, data garnered from long-term care insurers indicate that the average length-of-stay of insured clients is nearly equal to or lower than what is observed in the general population. 16 Thus, we do not assume adverse selection or moral hazard. Long-term care spending has risen dramatically, not only because of the growing population in need of services, but also because of price inflation. All estimates are described in 1990 dollars. Based on trends of the past decade, we assume that after adjustment for general inflation, nursing home costs will increase at an average rate of 1.8 percent and that real incomes of persons over age sixty-five will decline by roughly 1.5 percent per year. 17 For each purchaser flagged as a future nursing home entrant, we estimated the number of years it will take before nursing home entry occurs. Based on the analysis of nursing home length-of-stay patterns completed by Spence and Wiener, we randomly distributed individual nursing home entrants across possible entry ages such that the overall age distribution for nursing home entry mirrors that found by Spence and Wiener. 18 Finally, based on findings from the 1985 National Nursing Home Survey, we assumed that although most persons in the sample are married, at the time of nursing home entry most will be single. Probability of spending down. Medicaid eligibility for nursing home care is determined in part by state policy. Single persons are now eligible for Medicaid immediately upon admission to a nursing home if their assets (not counting a home) are less than $2,000. The comparable figure for a married person is now typically $14,148. Spouses are also allowed to keep income above 133 percent of the federal poverty level for a couple about $1,000 per month in 1991 and, at state option, up to $1,662 per month. When assets exceed the permitted amount, then the person must pay for care privately, presumably using both current income and assets, until assets are reduced to the eligibility level. In all cases, Medicaid-eligible nursing home residents are expected to contribute most of their income toward the cost of care, except for a small personal-needs allowance, about $600 per year. Medicaid then pays the difference between the resident s contribution and

5 D ATAWATCH 131 the established Medicaid rate for the facility. Exhibit 1 shows the distribution of nursing home entrants by payment status and selected demographic characteristics for each of the two length- Exhibit 1 Nursing Home Users, By Medicaid Status And Selected Demographic Characteristics, Without Long-Term Care Insurance Total sample of nursing home residents (N = 3,050) Spence and Wiener Kemper et al. length-of-stay length-of-stay No Medicaid Medicaid No Medicaid Medicaid (n = 2,177) (n = 873) (n = 1,893) (n = 1,157) Age and older Sex Male Female Income Less than $5,000 $5,000- $9,999 $10,000-$14,999 $15,ooo-$19,999 $20,000- $24,999 $25,000-$34,999 $35,000- $49,999 $50,000 or more Total liquid assets Less than $20,000 $20,000- $29,999 $30,000 - $49,999 $50,000- $74,999 $75,000-$99,999 $100,000 or more Total length-of-stay Less than 3 months 3 months to 1 year 1-3 years 3-5 years 1-5 years More than 5 years Sources: D. Spence and J. Wiener, Nursing Home Length-of-Stay Patterns, The Gerontologist 30, no. 1 (1990); and P. Kemper et al., A Lifetime Perspective on Proposals for Financing Nursing Home Care, Inquiry (Winter 1991): a Differences significant at the.001 level.

6 132 HEALTH AFFAIRS Fall 1994 of-stay distributions. Also shown is the relationship between payment status and length-of-stay in a nursing home. These estimates do not take into account benefits from a person s long-term care insurance policy. Thus, results relate to payment status in the absence of insurance. Depending on the length-of-stay distribution, somewhere between 29 percent and 38 percent of long-term care insurance purchasers who use nursing homes would, in the absence of their long-term care policy, receive some Medicaid payment throughout their stay. This is equivalent to about percent of all policyholders. The vast majority of those who would receive Medicaid payments would qualify through spending down, and slightly less than 2 percent would be eligible for Medicaid at entry to a nursing home. This finding is particularly important because it shows that many long term care insurance purchasers likely would have to rely on Medicaid at some point during their nursing home stay. To the extent that possession of a long-term care insurance policy delays or prevents Medicaid eligibility, then real savings to Medicaid can accrue. How do spend-down rates compare to those found in the general population of elderly? Recent studies suggest that percent of nursing home entrants spend down to Medicaid eligibility levels, whereas percent are eligible at the time of entry. In total, percent of all nursing home entrants receive some Medicaid payment throughout their nursing home stay. 19 Clearly, given the income and asset profile of long-term care insurance purchasers, their potential rate of Medicaid use would be less than rates found in the general population. Also, those long-term care insurance purchasers who would qualify for Medicaid payments do so by spending down to eligibility levels, whereas in the general population, most persons qualify at initial entry to a nursing home. Thus, it is reasonable to assume that the amount of money spent by Medicaid per long-term care insurance policyholder would be less than the amount spent on a nonpolicyholder. The probability of having Medicaid as a payment source is inversely related to income and asset levels and positively related to length-of-stay in a nursing home. For example, nearly 70 percent of persons who would spend down to Medicaid have incomes less than $15,000, compared with less than 19 percent of non-medicaid nursing home entrants. Moreover, about 90 percent of potential Medicaid recipients have assets less than $30,000; far fewer non-medicaid nursing home entrants-about 20 percent-have assets below this amount. Finally, those long-term care policyholders who (in the absence of their policy) would qualify for Medicaid are three times as likely to spend more than five years in nursing home care than are policyholders who would not qualify for Medicaid benefits. One of the reasons why persons purchase long-term care insurance is that

7 D ATAWATCH 133 they believe that the long-term care benefits paid by a policy will enable them to gain access to care without having to depend on Medicaid. 20 How well does the purchase of long-term care insurance meet this objective? Exhibit 2 shows the distribution of nursing home entrants by payment source and policy design features. Here, long-term care insurance benefits are taken into account in simulating the distribution of payments among Medicaid, insurance, and out-of-pocket sources. Exhibit 2 Nursing Home Users, By Medicaid Status And Long-Term Care Policy Design Characteristics, With Long-Term Care Insurance Total sample Benefit duration 1 year 2-3 years 4-5 years 6 years Lifetime Spence and Wiener Kempet et al. length-of-stay length-of-stay No Medicaid Medicaid No Medicaid Medicaid (n = 2,537) (n = 513) (n = 2,306) (n = 744) Daily benefit amount Less than $60 $60-$69 $7 0-$ 79 $80-$89 More than $90 Elimination period 0 days 1-30 days days days More than 100 days Inflation protection No Yes Monthly premium $50 or less $50-$74 $75 -$99 $100-$125 More than $125 Sources: D. Spence and J. Wiener, Nursing Home Length-of-Stay Patterns, The Gerontologist 30, no. 1 (1990); and P. Kemper et al., A Lifetime Perspective on Proposals for Financing Nursing Home Care, Inquiry (Winter 1991): a Differences significant at the.001 level.

8 134 HEALTH AFFAIRS Fall 1994 Having a long-term care insurance policy significantly reduces the probability of spending down to Medicaid eligibility levels. In fact, the number of persons spending down declines by percent; in other words, about 39 percent of nursing home entrants who in the absence of their policy would receive Medicaid payments no longer have to depend on Medicaid in the presence of their policy. The policies purchased by persons who continue to spend down to Medicaid eligibility levels are characterized by relatively low benefit durations (less than three years), low daily benefit amounts (less than $60 per day), longer deductible periods (more than ninety days), and no inflation protection. Not surprisingly, the premiums for these policies are also relatively low: More than 60 percent of policyholders who are eligible for Medicaid, notwithstanding long-term care insurance benefits, spend less than $75 per month on a policy. The average premium paid by persons who will not qualify for Medicaid payments is $99 per month, compared with $69 for those who would receive Medicaid even in the presence of an insurance policy. 21 Net savings to Medicaid. The daily Medicaid expenditure for an eligible nursing home user is equal to the gap between the average daily cost of care and the amount that the person can contribute to his or her care from income and unsheltered assets. Total Medicaid expenditures are a function of the total amount of time a person spends in a nursing home after Medicaid begins paying for care. We summarize the distribution of longterm care costs by payment source for each of three groups (Exhibit 3). In this way, it is possible to estimate the net fiscal benefit (savings) to Medicaid of long-term care policy ownership. In 1990 dollars, the expected long-term care costs for nursing home entrants range from $49,000 to $91,000 (depending on length-of-stay distribution). Not surprisingly, those who would continue to spend down even in the presence of a long-term care policy have the longest lengths-of-stay, hence greater expenses; they spend 1.5 to 1.8 times more on long-term care than those who would not spend down to Medicaid eligibility. The costs to Medicaid for the subset of long-term care insurance purchasers who enter nursing homes are between $6,492 and $14,179 (depending on length-of-stay distribution). If these persons did not have long-term care policies, then the average Medicaid costs per nursing home entrant for this cohort of long-term care policyholders would vary between $14,437 and $29,698. Viewed in the aggregate, Medicaid expenditures would be reduced by $7,945 to $15,519 for every nursing home entrant who had a long-term care insurance policy. Thus, long-term care insurance policy ownership halves the expected Medicaid liability per nursing home entrant as insurance dollars replace Medicaid dollars. The reduction in Medicaid expenditures per person results from two

9 D ATAWATCH 135 Exhibit 3 Long-Term Care Utilization And Cost Parameters, By Classification Group And Length-Of-Stay Distribution Utilization and cost parameter Total long-term care costs Total length-of-stay Daily nursing home cost Out-of-pocket casts per person Insurance costs per person Classification group All nursing home entrants (n = 3,050) $48,507-$91,186 54%1,013 days $89-$90 $18,768-$35,604 $23,247-$41,403 Those who spend down Those who spend down in absence of a policy in presence of a policy (n = 873 to 1,157) (n = 513 to 744) $85,167-$130, ,428 days $90-$91 $34,725-$51,748 $91,203-$139, ,504 days $92-$93 $23,507-$39,389 $29,100-$42,386 Medicaid costs per person (with insurance) $6, 492-$14, 179 $38,596-$58,125 Medicaid costs per person (without insurance) $14,437-$29, 698 $50,442-$78, 289 Sources: D. Spence and J. Wiener, Nursing Home Length-of-Stay Patterns, The Gerontologist 30, no. 1 (1990); and P. Kemper et al., A Lifetime Perspective on Proposals for Financing Nursing Home Care, Inquiry (Winter 1991): Note: Where a range is indicated, the first number is for the length-of-stay distribution based on the work of Spence and Wiener, and the second is based on the length-of-stay distribution derived by Kemper and colleagues. phenomena. First, in the presence of insurance, fewer people would qualify for Medicaid payments. Second, of those persons who still qualify for Medicaid, the average daily payment is reduced by an amount equal to the daily insurance benefit. Not surprisingly, whereas Medicaid finances about 43 percent of the nursing home care of elders in the general population, among this sample of nursing home users with long-term care insurance policies, Medicaid would pay only 15 percent of nursing home costs. In 1990 real dollars, the average savings that accrue to Medicaid are $3,500-$6,854 per policyholder over the life of the cohort. For every million policyholders who keep policies until they enter nursing homes, Medicaid stands to save $3.5 billion to $6.9 billion (1990 dollars) over the next twenty-five years. Because younger people are buying policies, the cumulative impact of long-term care policies on Medicaid expenditures increases over time; current expenditures are not greatly affected. Sensitivity Analyses Adverse selection and moral hazard, An important assumption underlying our analysis is that long-term care insurance purchasers and persons in the general population face the same lifetime risk of needing nursing home

10 136 HEALTH AFFAIRS Fall 1994 services and ultimately consume similar amounts of services. These assumptions rely on data provided by a number of long-term care insurers, which show that underwriting has protected large sellers against adverse selection and that observed utilization patterns are somewhat lower than what is found in the general population. If, however, there were a great degree of adverse selection and moral hazard, then one would expect overall utilization rates, spend-down rates, and Medicaid expenditures to increase. It is difficult to estimate how this would affect Medicaid savings. If these persons purchased richer benefits, then Medicaid savings would increase; if, however, their benefits were less comprehensive than average, then Medicaid savings would decline. Income elasticity of demand for nursing home care. Research indicates that nursing home entry is related to income status. Korbin Liu and colleagues found that the rate of nursing home entry over a two-year period for persons with incomes above $20,000 is only one-third as great as for persons with incomes below $20, However, once in a nursing home, the relationship between income and length-of-stay is U-shaped: Poor persons and wealthy persons tend to have the longest lengths-of-stay (at least over a two-year period). The distribution of income among long-term care insurance purchasers is significantly different than what is observed in the general elderly population. In 1990 about 36 percent of the general population had incomes in excess of $20,000, compared with about 52 percent of the sample at the time of simulated nursing home entry. Thus, applying the general population s utilization rates to the sample of long-term care insurance purchasers may overstate the proportion of wealthy persons entering nursing homes but may underestimate the amount of time they spend in a nursing home. Insofar as our interest is in that subsample of purchasers who would be eligible for Medicaid, such a bias would not significantly affect results. Of those persons who would qualify for Medicaid in the absence of a policy, 88 percent have incomes below $20,000. For the remaining 12 percent, Medicaid payments are 30 percent less than payments made for persons with incomes of $20,000 or less. The impact of lapses. A key assumption of the analysis is that persons who purchased policies will still have them at the time that they enter a nursing home. In fact, many elders allow their policies to lapse well in advance of entering a nursing home. Lapses can occur for one of three reasons: Persons die; they voluntarily stop paying premiums; or they switch or upgrade policies and begin paying premiums to another company. Lapsing affects the actual number of policies in force, but it also can influence who owns a policy at the time of nursing home entry. If those who lapse do so primarily because of a decline in income, then the amount of

11 D ATAWATCH 137 savings attributed to Medicaid in this analysis will be overestimated. If, on the other hand, wealthier persons drop policies because they believe that they can afford alternatives to nursing home care, then the savings ascribed to Medicaid as a result of long-term care insurance ownership probably will not be affected. Alternatively, lapses may reflect antiselection against companies over the long term, so that those who maintain their policies may ultimately use more services than ascribed to them in this analysis. Personal communications with a number of companies that contributed samples to the study indicate that average annual industry lapse rates vary between 5 percent and 10 percent. For each person in the sample, we estimated the probability of lapsing a policy before entry to a nursing home. We did this for each of three different annual lapse rate assumptions: 5 percent, 7 percent, and 9 percent (Exhibit 4). If the cumulative probability of lapsing a policy is 50 percent or greater at the time of simulated nursing home entry, a person is flagged as lapsing. Lapse rates have a large impact on the probability of spending down and on the projected savings to Medicaid from long-term care policy ownership. For example, when an overall lapse rate of 7 percent is assumed, everything else held constant, the rate of spending down increases by about six percentage points. Also, estimated Medicaid savings are significantly reduced, Exhibit 4 Impact Of Alternative Lapse Assumptions On The Probability Of Spending Down And Impact Of Long-Term Care Policy Ownership On Medicaid Expenditures Alternative lapse rate assumptions Base case 5% lapse 7% lapse 9% lapse Proportion of initial cohort of policyholders who enter nursing homes and spend down to Medicaid eligibility 17%-24% 21%-29% 23%-31% 24%-32% Medicaid savings per nursing home entrant $7,954-$15,579 $4,133-$7,641 $3,186-$5,508 $2,963-$4, 467 Medicaid savings per initial long-term care insurance purchaser $3,500-$6,854 $1,819-$3,362 $1,402-$2,424 $1,304-$1, 965 Average reduction in savings from base case (zero lapse assumption) -5 0 % -6 3 % -6 7 % Sources: D. Spence and J. Wiener, Nursing Home Length-of-Stay Patterns, The Gerontologist 30, no. 1 (1990); and P. Kemper et al., A Lifetime Perspective on Proposals for Financing Nursing Home Care, Inquiry (Winter 1991): Note: Where a range is indicated, the first number is for the length-of-stay distribution based on the work of Spence and Wiener, and the second is based on the length-of-stay distribution derived by Kemper and colleagues.

12 138 HEALTH AFFAIRS Fall 1994 from $7,900-$15,500 to $3,200-$5,500. Thus, if most lapses happen when persons voluntarily drop their policies, and there is no relationship between future service use and the probability of lapsing, then the impact of longterm care insurance on Medicaid expenditures will be greatly reduced. Policy Implications The issue of long-term care financing was addressed in the Clinton administration s health care reform plan and in several proposals now before Congress. Although there is considerable uncertainty about the ultimate shape of health reform, there does appear to be a widespread consensus that both the public and private sectors will play a role in the financing of long-term care, with the private sector continuing to offer products that insure against the risk of nursing home placement. This analysis shows not only that growth in the private market will protect more persons, but also that Medicaid can reap long-term fiscal savings from the proliferation of policies, especially if benefit designs continue to improve and insurers offer reduced benefits for those who voluntarily let their policies lapse. The research and analysis fur this study were supported by a grant from The Robert Wood Johnson Foundation. The data collection effort was supported by a contract from the Health Insurance Association of America. The views expressed in this paper are those of the authors, who are responsible for any errors herein. NOTES 1. U.S. Senate Special Committee on Aging, Aging America: Trends and Perceptions (Washington: U.S. Government Printing Office, 1991). 2. K.R. Levit, M.S. Friedland, and D.R. Waldo, National Health Care Spending Trends: 1988, Health Affairs (Summer 1990): L. Carpenter, Medicaid Eligibility for Persons in Nursing Homes, Health Care Financing Review (Winter 1988): 67-78; and Health Care Financing Administration, National Health Expenditures, 1988, Health Care Financing Review (Summer 1990): B. Burwell, Medicaid Long-Term Care Expenditures for FY 1991 (New York: Syste- Metrics/ McGraw-Hill Healthcare Management Group, 10 January 1992). 5. S. Van Gelder and D. Johnson, Long-Term Care Insurance: A Market Update, Health Insurance Association of America Research Bulletin (Washington: HIAA, 1991). 6. K. Mahoney, Financing Long-Term Care with Limited Resources: Combining the Resources of the Public and Private Sectors, Journal of Aging and Social Policy 4, no. 1/ 2 (Spring/ Summer 1992): A. Rivlin et al., Caring for the Disabled Elderly: Who Will Pay? (Washington: The Brookings Institution, 1988); and T. Rice, K. Thomas, and W. Weissert, The Effect of Owning Private Long-Term Care Insurance Policies on Out-of-Pocket Costs, Health Services Research (February 1991):

13 D ATAWATCH M.A. Cohen, N. Kumar, and S.S. Wallack, Who Buys Long-Term Care Insurance? Health Affairs (Spring 1992): Slightly more than 45 percent of purchasers chose not to reveal their level of income and assets. Those not reporting income tended to be older, more educated, and more likely to be unmarried than persons providing such information. We imputed missing values by using discriminant function analysis to classify individuals into one of eight income and seven asset categories. Independent classification variables included age, sex, marital status, and education level. 10. P. Kemper, B. Spillman, and C. Murtaugh, A Lifetime Perspective on Proposals for Financing Nursing Home Care, Inquiry (Winter 1991): ; and M. Cohen, E. Tell, and S. Wallack, The Lifetime Risks and Costs of Nursing Home Use among the Elderly, Medical Care (24 December 1986): A simplifying assumption, discussed in a subsequent section, is that the age-specific probability of nursing home entry is unrelated to wealth. 12. D. Spence and J. Wiener, Nursing Home Length-of-Stay Patterns: Results from the 1985 National Nursing Home Survey, The Gerontologist 30, no. 1 (1990): 16-20; and Kemper et al., A Lifetime Perspective on Proposals for Financing Nursing Home Care. 13. There is no clear evidence of a linear relationship between income levels and length of stay in a nursing home. K. Liu, K. Manton, and B. Liu, Functional Disability and Long-Term Care (Report submitted to the Health Insurance Association of America, July 1988). Therefore, because income is negatively related to nursing home entry yet positively related to length-of-stay, we make the simplifying assumption that the income elasticity of demand for total nursing home days is zero. 14. E. Hing, Use of Nursing Homes by the Elderly: Preliminary Data from the 1985 National Nursing Home Survey, Advance Data 135 (Washington: National Center for Health Statistics, 14 May 1987). 15. Cohen et al., Who Buys Long-Term Care Insurance? 16. Personal communication to actuaries in four major insurance companies selling longterm care insurance. 17. Rivlin et al., Caring for the Disabled Elderly; U.S. Bureau of the Census, unpublished data from the March 1990 Current Population Survey; and S. Zedlewski and T. McBride, The Changing Profile of the Elderly: Effects on Future Long-Term Care Needs and Financing, The Milbank Quarterly 70, no. 2 (1992): Spence and Wiener, Nursing Home Length-of-Stay Patterns. 19. K. Liu and K. Manton, Nursing Home Length of Stay and Spend Down in Connecticut, , The Gerontologist 31, no. 2 (1991): ; Spence and Wiener, Nursing Home Length-of-Stay Patterns; T. Rice, The Use, Cost, and Economic Burden of Nursing Home Care in 1985, Medical Care 27 (1989): ; K. Liu and K. Manton, The Effect of Nursing Home Use on Medicaid Eligibility, The Gerontologist 29, no. 1(1989): 59-66; L. Gruenberg et al., An Analysis of the Spend-Down Patterns of Individuals Admitted to Nursing Homes in the State of Connecticut (Waltham, Mass.: Long-Term Care Database, 1989),; and L. Schofield, C. Pattee, and K. Liu, Data Set the Stage for Policy Decisions, Business and Health (August 1988): Cohen et al., Who Buys Long-Term Care Insurance? 21. Multivariate logistic analysis indicates that both policy features and demographic characteristics are positively related to the probability of continuing to spend down to Medicaid eligibility even in the presence of a private long-term care insurance policy. These include benefit durations of less than two years; daily benefit amounts lower than $90; elimination periods greater than ninety days; the absence of inflation protection; being married, male, and under age seventy-five; assets below $75,000; income levels below $35,000; and having a nursing home length-of-stay in excess of three years. 22. Liu et al., Functional Disability and Long-Term Care.