Hostname: page-component-cd9895bd7-q99xh Total loading time: 0 Render date: 2024-12-26T04:48:43.210Z Has data issue: false hasContentIssue false
  • English
  • Français

Retiree Out-of-Pocket Healthcare Spending: A Study of Consumer Expectations and Policy Implications

Published online by Cambridge University Press:  06 January 2021

Allison K. Hoffman  and
Howell E. Jackson
Allison K. Hoffman
Affiliation:
UCLA School of Law. Contact information: [email protected]
Howell E. Jackson
Affiliation:
Harvard Law School. Contact information: [email protected]
Get access Rights & Permissions [Opens in a new window]

Extract

Even though most American retirees benefit from Medicare coverage, a mounting body of research predicts that many will face large and increasing out-of-pocket expenditures for healthcare costs in retirement and that many already struggle to finance these costs. It is unclear, however, whether the general population understands the likely magnitude of these out-of-pocket expenditures well enough to plan for them effectively. This study is the first comprehensive examination of Americans' expectations regarding their out-of-pocket spending on healthcare in retirement. We surveyed over 1700 near retirees and retirees to assess their expectations regarding their own spending and then compared their responses to experts' estimates. Our main findings are twofold. First, overall expectations of out-of-pocket spending are mixed. While a significant proportion of respondents estimated out-of-pocket costs in retirement at or above expert estimates of what the typical retiree will spend, a disproportionate number estimated their future spending substantially below what experts view as likely. Estimates by members of some demographic subgroups, including women and younger respondents, deviated relatively further from the experts' estimates. Second, respondents consistently misjudged spending uncertainty. In particular, respondents significantly underestimated how much individual health experience and changes in government policy can affect individual out-of-pocket spending. We discuss possible policy responses, including efforts to improve financial planning and ways to reduce unanticipated financial risk through reform of health insurance regulation.

Type
Article
Information
American Journal of Law & Medicine , Volume 39 , Issue 1 , March 2013 , pp. 62 - 133
Copyright
Copyright © American Society of Law, Medicine and Ethics and Boston University 2013

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

References

1 PAUL FRONSTIN ET AL., EMP. BENEFIT RESEARCH INST., NO. 351, ISSUE BRIEF: FUNDING SAVINGS NEEDED FOR HEALTH EXPENSES FOR PERSONS ELIGIBLE FOR MEDICARE 3 (2010), available at http://www.ebri.org/pdf/briefspdf/EBRI_IB_12-2010_No351_Savings3.pdf.

2 Baicker, Katherine & Levy, Helen, The Insurance Value of Medicare, 367 NEW ENG. J. MED. 1773, 1773 (2012)Google Scholar.

3 FRONSTIN ET AL., supra note 1, at 4 (citing variation in Plan F premium amounts across states).

4 JULIETTE CUBANSKI ET AL., HENRY J. KAISER FAMILY FOUND., MEDICARE CHARTBOOK 72 (4th ed. 2010) [hereinafter KFF CHARTBOOK].

5 ELIOT FISHMAN ET AL., COMMONWEALTH FUND, MEDICARE OUT-OF-POCKET COSTS: CAN PRIVATE SAVINGS INCENTIVES SOLVE THE PROBLEM? viii (2008) (discussing low-income retirees, retirees in fair or poor health, or those over eighty-five years old.).

6 RICHARD W. JOHNSON & CORINA MOMMAERTS, URBAN INST., RETIREMENT POLICY PROGRAM, WILL HEALTHCARE COSTS BANKRUPT AGING BOOMERS? 1-2 (2010) (reporting that by 2040 half of adults over sixty-five will spend 19% of income or more on healthcare, up from 10% in 2010); see also, e.g., ALICIA H. MUNNELL ET AL., CTR. FOR RET. RESEARCH BOS. COLL., No. 8-3, HEALTHCARE COSTS DRIVE UP THE NATIONAL RETIREMENT RISK INDEX 4 (2008); Gruber, Jonathan & Levy, Helen, The Evolution of Medical Spending Risk, 23 J. ECON. PERSP. 25, 40 (2009)Google Scholar.

7 JOHNSON & MOMMAERTS, supra note 6, at 1-2.

8 Healthcare spending comprises an increasing share of the Gross Domestic Product (GDP) and of federal budget dollars. CONG. BUDGET OFFICE, HEALTH CARE AND THE BUDGET: ISSUES AND CHALLENGES FOR REFORM 2-3 (2007), available at http://www.cbo.gov/sites/default/files/cbofiles/ftpdocs/82xx/doc8255/06-21-healthcarereform.pdf.

9 Studies report savings shortfalls, healthcare costs as exacerbating retirement risk (defined as a substantial and detrimental decrease in standard of living) or healthcare costs consuming a large portion of household assets. See ALICIA H. MUNNELL ET AL., CTR. FOR RET. RESEARCH BOS. COLL., NO. 9-7, LONG-TERM CARE COSTS AND THE NATIONAL RETIREMENT RISK INDEX 6 (2009) (including long-term care and healthcare in calculations of national retirement risk increased estimates of those at risk from 44% to 61% for the overall population); ALICIA H. MUNNELL ET AL., supra note 6, at 5; JACK VANDERHEI, EMP. BENEFIT RES. INST., NO. 10, NOTES: RETIREMENT SAVINGS SHORTFALLS FOR TODAY's WORKERS 2 (2010) (estimating average retirement savings shortfall of over $47,000 per individual for both basic living expenses and out-of-pocket healthcare costs, not including nursing home and home healthcare costs which, if added, increase the average shortfall by an additional $32,000 for the average man and by $46,000 for the average woman); Amy S. Kelley et al., Out-of-Pocket Spending in the Last Five Years of Life, J. GEN. INTERNAL MED., Sept. 5, 2012, at 4. Some studies link healthcare spending to bankruptcy for seniors in particular. See generally JOHNSON & MOMMAERTS, supra note 6. Others find the same, looking at the entire population. See David Dranove & Michael L. Millenson, Medical Bankruptcy: Myth Versus Fact, 25 HEALTH AFF. W74, W79 (2006) (reporting medical bills as a contributing factor in 17% of bankruptcies); Himmelstein, David U. et al., Medical Bankruptcy in the United States, 2007: Results of a National Study, 122 AM. J. MED. 741, 743 (2009)Google Scholar (finding that medical bills contribute to a majority of bankruptcies). Jennifer Prah Ruger suggests that these methods for measuring financial insecurity and health expenditures underestimate adverse consequences. Ruger, Jennifer Prah, An Alternative Framework for Analyzing Financial Protection in Health, 9 PLOS MED., no. 8, 2012, at 1, 5CrossRefGoogle Scholar.

10 See, e.g., Jonathan Starkey, Financial Literacy: Health Care's Big Bite, NEWS J., Nov. 8, 2012, http://www.delawareonline.com/article/20111023/BUSINESS10/110230328/Financial-Literacy-Health-care-s-big-bite?gcheck=1&nclick_check=1; Paul Sullivan, Planning for Retirement: Don't Forget Health Care Costs, N.Y. TIMES, Oct. 5, 2012, http://www.nytimes.com/2012/10/06/your-money/planning-for-retirement-dont-forget-health-care-costs.html?pagewanted=all.

11 Nardi, Mariacristina De et al., Why Do the Elderly Save? The Role of Medical Expenses, 118 J. POL. ECON. 39, 72–73 (2010)Google Scholar (“[W]e find that out-of-pocket medical expenditures, and the way in which they interact with the consumption floor, go a long way toward explaining the elderly's saving decisions and should be accounted for when considering old-age policy reforms.”).

12 See PAUL FRONSTIN ET AL., EMP. BENEFIT RES. INST., NO. 317, ISSUE BRIEF: SAVINGS NEEDED TO FUND HEALTH INSURANCE AND HEALTH CARE EXPENSES IN RETIREMENT: FINDINGS FROM A SIMULATION MODEL 23 (2008).

13 See BANKERS LIFE & CASUALTY CO., CTR. FOR SECURE RET., RETIREMENT HEALTHCARE FOR MIDDLE-INCOME AMERICANS 18-21 (2012) (showing that middle-age Americans and near retirees are largely unaware of the benefits and coverage available in the Medicare program and what is not covered, including vision, dental, and most long-term care benefits); McCormack, Lauren et al., Health Insurance Literacy of Older Adults, 43 J. CONSUMER AFF. 223, 240 (2009)Google Scholar.

14 See RUTH HELMAN ET AL., EMP. BENEFIT RES. INST., NO. 355, ISSUE BRIEF: THE 2011 RETIREMENT CONFIDENCE SURVEY: CONFIDENCE DROPS TO RECORD LOWS, REFLECTING “THE NEW NORMAL” 10 (2011).

15 Lusardi, Annamaria, Household Saving Behavior: The Role of Financial Literacy, Information, and Financial Education Programs 11–13Google Scholar (Nat’l Bureau of Econ. Research, Working Paper No. 13824, 2008) (describing financial literacy studies that show that lack of information impedes financial planning).

16 E.g., Choi, James J. et al., Small Cues Change Savings Choices 2–3Google Scholar (Nat’l Bureau of Econ. Research, Working Paper No. 17843, 2012) (showing use of education about 401k savings limits leads to increased savings for members of a defined-contribution plan of a large technology company); Goda, Gopi Shah et al., What Will My Account Really Be Worth? An Experiment on Exponential Growth Bias and Retirement Saving 3Google Scholar (Nat’l Bureau of Econ. Research, Working Paper No. 17927, 2012) (“[P]roviding income projections along with general plan information and materials assisting people through the steps of changing contribution rates resulted in a 29 percent higher probability of a change in contributions relative to a control group over a six-month period … and increased their annual contributions by $85 more than the control group … .”).

17 Lusardi, Annamaria & Mitchell, Olivia S., Financial Literacy and Retirement Planning in the United States, 10 J. PENSION ECON. & FIN. 517, 523 (2011)CrossRefGoogle Scholar (“[I]t appears that financial literacy does drive retirement planning”); Annamaria Lusardi & Olivia S. Mitchell, Financial Literacy and Retirement Planning: New Evidence from the Rand American Life Panel 19 (Oct. 2007) (unpublished manuscript) (on file with authors) (“By every measure, and in every sample we have examined, we conclude that financial literacy is a key determinant of retirement planning.”).

18 For example, section 1013(g) of the Dodd-Frank Act for Wall Street Reform and Consumer Protection calls for the creation of an Office of Financial Protection for Older American with a charge that includes research into best practices to educate about long-term savings and planning for retirement and long-term care. Dodd–Frank Wall Street Reform and Consumer Protection Act (Dodd-Frank Act), Pub. L. No. 111-203, 124 Stat. 1376 (2010) (to be codified as amended in scattered sections of U.S.C.). Governmental agencies, including the Department of Health and Human Services, have invested in improving financial and health literacy. ELIZABETH FRENTZEL ET AL., AM. INSTS. FOR RESEARCH, CONSUMER EDUCATION INITIATIVES IN FINANCIAL AND HEALTH LITERACY 3 (2010) (“These challenging times have created an increasing awareness that a lack of financial and health literacy can serve as a major barrier to the well-being of individual families and communities … a number of agencies have attempted to improve financial and health literacy.”).

19 See, e.g., RICHARD H. THALER & CASS R. SUNSTEIN, NUDGE (2008); Beshears, John et al., The Importance of Default Options for Retirement Savings Outcomes: Evidence from the United States 6 (Nat’l Bureau of Econ. Research, Working Paper No. 12009, 2006)CrossRefGoogle Scholar (describing that default enrollment in defined-contribution savings plan increases participation in savings plans); see also John Beshears et al., Public Policy and Savings for Retirement: The “Autosave” Features of the Pension Protection Reform Act of 2006, in BETTER LIVING THROUGH ECONOMICS (John J. Siegfried, ed. 2010) (reviewing economic evidence underlying auto-enrollment rules); On Amir & Orly Lobel, Liberalism and Lifestyle: Informing Regulatory Governance with Behavioural Research, 1 EUROPEAN J. OF RISK REG. 17, 20 (2012) (“[B]ehavioural insights can help policy improve individual decisions-making processes as well as identify limits of the corrective solutions to cognitive failures.”).

20 Most of these responses were near the benchmark estimate ranges, but some were high outliers, as discussed below.

21 In this study, we did not explore the important related question of whether individual expectations correlate to actual savings for those expenses.

22 In analyzing our survey results, we attempted to explore the extent to which respondent estimates correlated with factors that experts have found to be associated with higher retiree healthcare costs and found mixed results. Such demographic factors can predict about 20% to 25% of the variance in spending among members of a population. Newhouse, Joseph P., Reimbursing Health Plans and Health Providers: Efficiency in Production Versus Selection, 34 J. ECON. LITERATURE 1236, 1256 (1996)Google Scholar. Respondent estimates of costs corresponded as experts would predict with some factors, including household income levels (which experts associate with higher expenditures, as discussed below), some financial literacy proxies, and anticipated insurance coverage. In contrast, women in our survey estimated lower lifetime spending than men, contrary to what experts report.

23 Interpretation of this result is ambiguous, as discussed below.

24 FRONSTIN ET AL., supra note 1, at 9 fig.2.

25 Berk, Marc L. & Monheit, Alan C., The Concentration of Health Care Expenditures, Revisited, 20 HEALTH AFF. 9, 9 (2001)Google Scholar.

26 See infra notes 120-23 and accompanying text.

27 CONG. BUDGET OFFICE, THE LONG-TERM BUDGET OUTLOOK 1, 27 (2010) (reporting that from 1975-2008, excess cost growth in Medicare was 2.5%, in Medicaid was 2.0%, in all other forms of health insurance was 1.8%, and overall was 1.9%).

28 This inflation is often called “excess cost growth,” defined as the increase in healthcare spending per person over the growth of GDP per person, adjusted for demographic changes in the population that might affect healthcare spending. See id. at 10.

29 Patient Protection and Affordable Care Act of 2010, Pub. L. No. 111-148, 124 Stat. 119, amended by Health Care and Education Reconciliation Act of 2010, Pub. L. No. 111-152, 124 Stat. 1029 (to be codified as amended primarily in scattered sections of 42 U.S.C.).

30 See infra notes 120-23 and accompanying text.

31 KFF CHARTBOOK, supra note 4, at 70 fig.7.2.

32 See id. at 71 fig.7.3 (Bar graph shows breakdown of out-of-pocket expenditures between premiums and services in 2006).

33 See also id. at 70 (reporting that Medicare finances 48% of total costs of healthcare for Medicare fee-for-service beneficiaries, who will have a lower portion of their costs financed than Medicare Advantage beneficiaries).

34 Remler, Dahlia K. & Glied, Sherry A., What Other Programs Can Teach Us: Increasing Participation in Health Insurance Programs, 93 AM. J. PUB. HEALTH 67, 68 (2003)Google Scholar (reporting that 99% of eligible persons take up Medicare Part A and 95.5% Medicare Part B).

35 KFF CHARTBOOK, supra note 4, at 22. Part A is premium-free if an individual or spouse worked forty or more quarters of Medicare-covered employment where they contributed Medicare payroll taxes. Medicare Part A, EXTENDHEALTH.COM, https://www.extendhealth.com/medicare/part-a (last visited Dec. 14, 2012).

36 KFF CHARTBOOK, supra note 4, at 22.

37 GRETCHEN JACOBSON ET AL., HENRY J. KAISER FAMILY FOUND., THE ROLE OF MEDICARE FOR PEOPLE DUALLY ELIGIBLE FOR MEDICARE AND MEDICAID 9-10 (2011) (describing how Medicaid defrays premiums or cost sharing for Qualified Medicare Beneficiaries (QMBs), who must earn under 100% of the FPL to receive assistance with Medicare premiums and cost-sharing; Specified Low-Income Medicare Beneficiaries (SLMBs), who have incomes between 100% and 120% of the FPL and are eligible for assistance with Medicare Part B premiums; and Qualified Individuals (QIs), who earn between 120% to 135% of the FPL and receive assistance with premiums in limited circumstances. To qualify for any of these programs, a beneficiary must have assets at or below $6880 for an individual or $10,020 for a couple in 2011). For reference, in 2012, the FPL was just over $11,000 for an individual and just over $15,000 for a couple. 2012 HHS Poverty Guidelines, ASPE, http://aspe.hhs.gov/poverty/12poverty.shtml (last updated Feb. 9, 2012).

38 Dorn, Stan & Shang, Boaping, Spurring Enrollment in Medicare Savings Programs Through a Substitute for the Asset Test Focused on Investment Income, 31 HEALTH AFF. 367, 368-70Google Scholar (estimating 3.6 million eligible for the three MSP programs and under a third enrolled in each). The authors explain low enrollment as due in part to the application process, including a “burdensome” asset test and recommend replacing the asset test with an investment income test. Id. at 368-69.

39 What Are the Medicare Premiums and Coinsurance Rates for 2013?, U.S. DEP't OF HEALTH & HUMAN SERVS., http://answers.hhs.gov/questions/3006. (last visited Jan. 13, 2012).

40 Medicare Costs at a Glance, MEDICARE.GOV, http://www.medicare.gov/your-medicare-costs/costs-at-a-glance/costs-at-glance.html (last visited Dec. 9, 2012).

41 Id.

42 Id. There is also significant cost sharing and limited coverage for skilled nursing. Id.

43 KFF CHARTBOOK, supra note 4, at 60. While we use this data on supplemental coverage as a benchmark, it does not perfectly reflect frequency of forms of supplemental coverage among retirees for two reasons. First, it includes non-elderly disabled on Medicare. Second, KFF only listed one form of supplemental coverage for each individual according to the following hierarchy: “1) Medicare Advantage, 2) Medicaid, 3) Employer, 4) Medigap, 5) Other public/private coverage, 6) No supplemental coverage. Individuals with more than one source of coverage were assigned to the category that appears highest in the ordering.” Id. This methodology will underestimate forms of supplemental coverage lower in the hierarchy, such as “other public/private coverage.” Id. As an example, 22% of Medicare Advantage enrollees have an additional form of coverage (10% self-purchased private coverage; 10% employer-sponsored; 1% both self-purchased and employer-sponsored). HENRY J. KAISER FAMILY FOUND., EXAMINING SOURCES OF COVERAGE AMONG MEDICARE BENEFICIARIES: SUPPLEMENTAL INSURANCE, MEDICARE ADVANTAGE, AND PRESCRIPTION DRUG COVERAGE 17 exhibit 3.9 (2008).

44 Those with Medigap supplemental coverage face the greatest total out-of-pocket exposure (even more than those with no supplemental coverage, who are spared premium costs and may consume less care than they would otherwise). Goldman, Dana P. & Zissimopoulos, Julie M., High Out-of-Pocket Care Spending by the Elderly, 22 HEALTH AFF. 194, 198 (2003)Google Scholar. In contrast, those with Medicaid are likely to spend much less out-of-pocket, due to the low premiums and cost-sharing obligations and possibly also due to consumption constraints. Id. at 198-99; see also KFF CHARTBOOK, supra note 4, at 78. Including residential long-term care, Kaiser reports average out-of-pocket spending in 2006 of $5066 for a beneficiary with supplemental Medigap, $4275 with supplemental ESI, $3979 with no supplemental coverage, $3518 with Medicare Advantage, and $2843 with Medicaid. KFF CHARTBOOK, supra note 4, at 78. Another study, based on 2005 MCBS data (prior to Medicare Part D) and also including long-term care spending, reports median spending of $3819 for a beneficiary with supplemental Medigap, $2909 with ESI, $2258 for Medicare Advantage, $1864 with no supplemental coverage, and $490 with Medicaid. TRICIA NEUMAN ET AL., HENRY J. KAISER FAMILY FOUND., REVISITING ‘SKIN IN THE GAME’ AMONG MEDICARE BENEFICIARIES: AN UPDATED ANALYSIS OF THE INCREASED FINANCIAL BURDEN OF HEALTH CARE SPENDING FROM 1997 TO 2005 2 (2009). This variability persists with regard to total lifetime spending. FRONSTIN ET AL., supra note 1, at 9 (estimating median spending of $65,000 for a man with wraparound Medicare coverage, $66,000 for ESI coverage that an employer subsidizes, and $109,000 for unsubsidized ESI coverage).

45 See KFF CHARTBOOK, supra note 4, at 60.

46 See id. at 72 (reporting average premiums of $2000 in 2006).

47 See FRONSTIN ET AL., supra note 12, at 14. The percentage of private-sector employers offering coverage to Medicare-eligible retirees decreased from 21.6% in 1987 to 12.7% in 2005. Id. at 12. Some attribute this decline to a 1990 rule by the Financial Accounting Standards Board that required employers to report retiree health liabilities in annual reports. See id. at 11. Even when employers offer ESI, it has become more expensive and less widely available among retirees. See id. at 14; see also HENRY J. KAISER FAMILY FOUND. & HEWITT, FINDINGS FROM KAISER/HEWITT 2006 SURVEY ON RETIREE HEALTH BENEFITS 19-20 (2006) (listing survey results on ESI changes that affected under sixty-five and over sixty-five retiree health benefits).

48 See KFF CHARTBOOK, supra note 4, at 60.

49 Medicare Advantage Plans, MEDICARE.GOV, http://www.medicare.gov/navigation/medicare-basics/medicare-benefits/part-c.aspx (last visited June 3, 2011). Some pay an additional monthly premium on top of the Part B premium; others plans are “zero premium.” The average premium in 2011 was forty-three dollars, based on the cost of plans with prescription drug coverage. Medicare Advantage providers often receive government rebates, based on plan cost savings over traditional Medicare, which they can use to provide additional services or reduce premiums. Id.

50 MARSHA GOLD ET AL., MEDICARE ADVANTAGE 2012 SPOTLIGHT: PLAN AVAILABILITY AND PREMIUMS 1 (2011).

51 KFF CHARTBOOK, supra note 4, at 60. Medigap policies typically do not cover long-term care, vision, dental, hearing aids, or private nursing care. See id. at 20.

52 See Baicker & Levy, supra note 2, at 1773-74.

53 See Medigap Policy Search, MEDICARE.GOV, http://www.medicare.gov/find-a-plan/questions/medigap-home.aspx (last visited May 31, 2011). Premiums were calculated based on information provided by the search feature on May 31, 2011.

54 AM.'s HEALTH INS. PLANS, MEDIGAP: WHAT YOU NEED TO KNOW 3 (2011), available at http://www.ahip.org/MedigapWhatYouNeedtoKnow/ (reporting 17% of beneficiaries enrolled in Plan C and 45% in Plan F in 2009).

55 See KATHRYN LINEHAN, NAT’L HEALTH POLICY FORUM, NO. 845, RECENT PROPOSALS TO LIMIT MEDIGAP COVERAGE AND MODIFY MEDICARE COST SHARING, NAT’L 5 (Feb. 24, 2012), http://www.nhpf.org/library/issue-briefs/IB845_MedigapandCostSharing_02-24-12.pdf.

56 Id.

57 See JACOBSON ET AL., supra note 37, at 3 (reporting that 21% of Medicare eligibles were dual-eligibles for Medicaid, just over three-quarters of whom are “fully” eligible for Medicaid benefits).

58 For example, a majority of states are required by Federal Medicaid participation rules to provide full Medicaid dual eligibility to those who meet the Supplemental Security Income (SSI) Program income and asset limits, which for an individual is income under 75% of the FPL and assets under $2000. Id. at 8. Some states, known as “209(b) states,” may set lower eligibility levels. Id. Even if not fully eligible for Medicaid, some Medicare beneficiaries are eligible for Medicaid assistance with all or some of their Medicare premiums and cost sharing through MSPs, as discussed above. Id. at 2-3. Most states provide full Medicaid benefits at slightly higher income and asset levels than required or for non-mandatory populations, including the “medically needy,” nursing home residents, or others in community-based long-term care under a waiver program. Id. at 8.

59 Id. The average monthly Part D plan premium is just over forty dollars. JACK HOADLEY ET AL., HENRY J. KAISER FAMILY FOUND., MEDICARE PART D SPOTLIGHT: PART D PLAN AVAILABILITY IN 2011 AND KEY CHANGES SINCE 2006 2 (2010), available at http://www.kff.org/medicare/upload/8107.pdf (estimating 2011 premium weighted by enrollment, based on 2010 enrollment). In 2010, about 60% had a Medicare Part D plan for prescription drugs, nearly 20% had coverage through an ESI retiree plan, and 13% had some other coverage. KFF CHARTBOOK, supra note 4, at 34 fig.3.1.

60 Pub. L. No. 108-173, 117 Stat. 2066 (2003).

61 JACK HOADLEY ET AL., HENRY J. KAISER FAMILY FOUND., ANALYSIS OF MEDICARE PRESCRIPTION DRUG PLANS IN 2012 AND KEY TRENDS SINCE 2006 exhibit 1 (2012) (depicting cost sharing in Part D plans including $4700 “True Out-of-Pocket Spending” or “TROOP,” which triggers catastrophic coverage).

62 KFF CHARTBOOK, supra note 4, at 70. On average, across all forms of supplemental coverage, the costs of premiums tend to comprise between 40% and 60% of total out-of-pocket expenses and cost-sharing and costs of uncovered healthcare make up the rest. Id. at 72.

63 Author's analysis of data in Figure 7.2 in id. at 70 (on file with the authors).

64 See infra note 136.

65 FRONSTIN ET AL., supra note 1, at 9.

66 See, e.g., FRONSTIN ET AL., supra note 1; Hurd, Michael D. & Rohwedder, Susann, The Level and Risk of Out-of Pocket Healthcare Spending (Univ. of Mich. Ret. Research Ctr., Working Paper No. 2009-218, 2009)Google Scholar; Webb, Anthony & Zhivan, Natalia, How Much Is Enough? The Distribution of Lifetime Healthcare Costs (Ctr. for Ret. Research Bos. Coll., Working Paper No. 2010-1, 2010)Google Scholar; see also Goldman & Zissimopoulos, supra note 44, at 194; Fidelity Investments Estimates Health Care Costs for Couples Retiring in 2011 Will Drop to $230K in One-Time Reduction, FIDELTY.COM (Mar. 31, 2011), http://www.fidelity.com/inside-fidelity/individual-investing/2011-rhcce [hereinafter Fidelity Investments Estimates Health Care Costs]; Retirees Face Estimated $240,000 in Medical Costs, FIDELITY VIEWPOINTS (May 16, 2012), https://www.fidelity.com/viewpoints/retirees-medical-expenses.

67 JOHNSON & MOMMAERTS, supra note 6 (using the Urban Institute DYNASIM3 model to simulate insurance coverage and project spending as a function of insurance coverage and 2006 Health and Retirement Survey (HRS) data on insurance coverage and 2006 MEPS data, which only includes community-dwelling individuals, on out-of-pocket costs. They exclude the costs of long-term care and indicate that they use a 2009 intermediate growth rate of 2.8% for medical cost growth, which they say they have based on Medicare Trustees’ projections).

68 Id. at 11.

69 Id.

70 Id. at 13.

71 A Kaiser Family Foundation analysis of 2006 MCBS data, which includes long-term care costs, reports average per capita cost in 2006 of $4241; no medians are available. KFF CHARTBOOK, supra note 4, at 70. Long-term care costs were 19% on average, which means average annual out-of-pocket spending was just over $3400 when excluding long-term care, slightly more than $100 higher than Johnson and Mommaerts’ average ($3278). Id.; JOHNSON & MOMMAERTS, supra note 6, at 11. CMS estimated average annual out-of-pocket spending of $3800 for an individual retired in 2007, again with no medians reported. MUNNELL ET AL., supra note 6, at 3.

72 This estimate is based on comparing EBRI's estimated median spending for a man and a woman with wraparound Medicare coverage from 2009, before PPACA, to their estimate in 2010, after PPACA. FRONSTIN ET AL., supra note 1, at 9 (estimating costs after PPACA); see also PAUL FRONSTIN ET AL., EMP. BENEFIT RES. INST., NO. 6, SAVINGS NEEDED FOR HEALTH EXPENSES IN RETIREMENT: AN EXAMINATION OF PERSONS AGES 55 AND 65 IN 2009 2 (2009), available at http://www.ebri.org/pdf/notespdf/EBRI_Notes_06-June09.HlthSvg-RetFndg1.pdf (estimating costs in 2009, before PPACA).

73 Id. (estimates for retirees with employment-based supplemental coverage vary less, showing a decrease of 3% to 10%).

74 FRONSTIN ET AL., supra note 1, at 9. This benchmark study uses MEPS data, which excludes institutionalized patients (i.e., those in residential nursing home care) who tend to be more expensive, which could make the EBRI estimates lower than they would be if the entire population were considered.

75 Id. Authors don't indicate the figure they are using for excess cost growth, but the 2011 Medicare Trustees report assumed excess cost growth of 1.4% for Medicare Parts A and B and 2.5% for Part D for the first ten years and assumes growth of GDP plus 1% after year seventy-five. Estimates for years ten to twenty-five are based on linear interpolation between year ten and twenty-five. BDS. OF TRS., FED. HOSP. INS. & FED. SUPPLEMENTARY MED. INS. TRUST FUNDS, 2011 ANNUAL REPORT OF THE BOARDS OF TRUSTEES OF THE FEDERAL HOSPITAL INSURANCE AND FEDERAL SUPPLEMENTARY MEDICAL INSURANCE TRUST FUNDS 12 [hereinafter BDS. OF TRUSTEES 2011], available at https://www.cms.gov/ReportsTrustFunds/downloads/tr2011.pdf.

76 Id.

77 FRONSTIN ET AL., supra note 1, at 7.

78 Id. at 9 (median estimate of $109,000 for a man and $146,000 for a woman and seventy-fifth percentile estimate of $165,000 for a man and $192,000 for a woman retiring in 2010 with unsubsidized ESI).

79 See KFF CHARTBOOK, supra note 4, at 72.

80 See supra note 72 and accompanying text.

81 FRONSTIN ET AL., supra note 1, at 7.

82 In 2011, Fidelity actuaries estimated $230,000 lifetime out-of-pocket spending for the average couple saving to achieve 75% certainty of sufficiency (comparable to the above-cited EBRI estimate of $255,000 for a couple). See Fidelity Investments Estimates Health Care Costs, supra note 66; see also Putting a Price on Health, FIDELITY VIEWPOINTS (May 20, 2010), https://guidance.fidelity.com/viewpoints-workplace/putting-a-price-on-health-pr. A 2010 study by Webb & Zhivan estimates $197,000 in 2009 dollars for an average couple (most comparable to the EBRI $158,000 median) with a high school education and free of chronic disease at age 65, excluding long-term care expenses. Webb & Zhivan, supra note 66, at 37. This study uses a 4.2% rate of inflation-adjusted cost growth, based on 1960-2007 experience. Using a lower rate of 3.2%, based on CMS projections from 2007, they calculate an NPV that is 11% lower. This study excludes Medicaid-eligible households, those with long-term care insurance, and those with zero medical expenses and assumes that households are not subject to spending constraints, focusing on those who will finance most out-of-pocket spending on their own. Id. at 4.

83 For example, Hurd & Rohwedder caution that the data in the HRS study is higher than the other two surveys by as much as 50% at the mean. Hurd & Rohwedder, supra note 66, at 17 tbl.6. In contrast, Samuel Marshall, Kathleen McGarry and Jonathan Skinner disagree that the HRS numbers are inflated, even if higher. Marshall, Samuel et al., The Risk of Out-of-Pocket Healthcare Expenditures at End of Life, (Nat’l Bureau of Econ. Research, Working Paper No. 16170, 2010)CrossRefGoogle Scholar. They argue the detailed questions in the HRS elicit data that respondents may omit in other studies. Id. HRS also conducts “exit interviews” with relatives of deceased participants to capture spending in the last year of life and uses “unfolding brackets” to reduce non-response, both of which increase estimates and, perhaps, accuracy. Webb & Zhivan, supra note 66, at 8.

84 HRS is a long-running biennial panel survey that is broader than healthcare and collects data from about 20,000 individuals fifty-one or older. Hurd & Rohwedder, supra note 66, at 4. This survey asks about all categories of out-of-pocket spending, including prescription drugs but focuses less on such inquiries than other studies do. Id.

85 MEPS is a two-year household panel survey of community-dwelling individuals (i.e. excludes nursing home residents), which has a smaller sample of the older population than HRS and thus lower expenditures per person on average. See GARY ENGELHARDT & JONATHAN GRUBER, CTR. FOR RET. RESEARCH BOS. COLL., NO. 11-8, DOES MEDICARE PART D PROTECT THE ELDERLY FROM FINANCIAL RISK? (2011), available at http://crr.bc.edu/wp-content/uploads/2011/06/IB_11-8-508.pdf. MEPS triangulates data from the patient survey with a provider survey. Some believe the data to be better quality. Hurd & Rohwedder, supra note 66, at 4.

86 MCBS is a rotating four-year panel survey of people enrolled in Medicare, who may reside in either community or long-term care facilities, and asks participants to keep health spending diaries to capture data in more detail and more accurately. Id. at 4-5.

87 See KFF CHARTBOOK, supra note 4, at 68-73. Because long-term care takes many forms, some studies do, however, pick up some long-term expenditures for non-institutional patients, including short term nursing home stays, home-based care, or post-acute care, especially when financed by Medicare.

88 Hurd & Rohwedder, supra note 66, at 3.

89 John N. Friedman, Predicting Medicare Cost Growth, in IMPROVING HEALTH CARE COST PROJECTIONS FOR THE MEDICARE POPULATION: SUMMARY OF A WORKSHOP 83, 83 app. A (Gooloo S. Wunderlich rapporteur, 2010), available at http://www.ncbi.nlm.nih.gov/books/NBK52808/pdf/TOC.pdf (describing the different methodologies used to calculate Medicare cost growth and the limitations of each); see also BDS. OF TRS., FED. HOSP. INS. & FED. SUPPLEMENTARY MED. INS. TRUST FUNDS, 2012 ANNUAL REPORT OF THE BOARDS OF TRUSTEES OF THE FEDERAL HOSPITAL INSURANCE AND FEDERAL SUPPLEMENTARY MEDICAL INSURANCE TRUST FUNDS 12-20 (describing the complex methodology used to project Medicare cost growth); Memorandum from John D. Shatto & M. Kent Clemens, Ctrs. for Medicare & Medicaid Servs., Office of the Actuary, on Projected Medicare Expenditures Under Illustrative Scenarios with Alternative Payment Updates to Medicare Providers (May 18, 2012) (on file with authors) (explaining generally the difficulty in deciding on Medicare cost growth projections under PPACA).

90 CONG. BUDGET OFFICE, supra note 27, at 27.

91 See IMPROVING HEALTH CARE COST PROJECTIONS FOR THE MEDICARE POPULATION: SUMMARY OF A WORKSHOP, supra note 89, at 7-8 (describing OACT's methodologies for projecting Medicare expenditures). OACT projects each category of spending for the first ten years into the future, using “demographically-adjusted extrapolations of past cost growth.” Id. OACT then uses a Computable General Equilibrium Model (CGE) to forecast years twenty-five to seventy-five and fill in the median timeframe through linear interpolation between year ten and twenty-five. Id. This method benefits from capturing “endogenous growth reduction” as healthcare consumes more of income, but does not take into account heterogeneity of preferences and assumes that healthcare operates as a standard good, not dealing with issues of moral hazard, adverse selection, or supply-side incentives for use of care. Id. at 7.

92 The model does not specify how such slowdown would occur, but it might occur in theory though increased supply and decreased demand as more of income is taken up by out-of-pocket costs or due to policy changes. Id. at 8. Parties disagreed over the appropriate assumption for long-term medical cost growth in 2012 in light of the policy changes enacted by PPACA. Memorandum from Shatto & Clemens, supra note 89, at 1.

93 Other governmental offices use different methods and, for example, the Congressional Budget Office's (CBO) estimates are higher than the OACT estimates because the only brake on cost growth that CBO assumes is that non-healthcare consumption will not decline. See IMPROVING HEALTH CARE COST PROJECTIONS FOR THE MEDICARE POPULATION: SUMMARY OF A WORKSHOP, supra note 89, at 8.

94 Newhouse, supra note 22.

95 Goldman & Zissimopoulos, supra note 44, at 197 exhibit 2 (reporting spending for low, middle and high income earners with increasing out-of-pocket spending as income increases); Marshall et al., supra note 83, at 4 (finding that spending in the last year of life is greater at higher income quartiles). Low incomes is defined as less than $12,600, middle as $12,600 to $38,860, and high income as above $38,860, all in 1998 dollars); De Nardi et al., supra note 11, at 53 fig.3 (modeling average medical expensive by permanent income quintile from age 74 to 100 and showing increased spending at each income quintile, including nursing home costs).

96 Marshall et al., supra note 83, at 25.

97 Id.

98 Id. at 4.

99 See, e.g., NEUMAN ET AL., supra note 44, at 2 (reporting mean spending of $2761 under 100%, $4001 at 100% to 199%, $4406 from 200% to 300%, and $4997 above 400% of FPL, including long-term care spending).

100 FRONSTIN ET AL., supra note 1.

101 Actuarial Life Table, 2007, SOC. SEC. ADMIN. (Apr. 10, 2012), http://www.ssa.gov/OACT/STATS/table4c6.html. Twenty-five percent of men would live to eighty-seven and women to ninety, and 10% of men would live until ninety-one and women to ninety-five. Id.

102 FRONSTIN ET AL., supra note 1, at 9. These are the estimates for beneficiaries with wraparound Medicare coverage.

103 See, e.g., NEUMAN ET AL., supra note 44, at 2 (reporting a mean spending of $4281 for a woman and $3765 for a man and median spending of $2908 and $2532, all for 2005 and including long-term care). Because this study included long-term care expenditures, it is less useful to us as a benchmark.

104 Hurd & Rohwedder, supra note 66, at 9 (describing persistence of bad health/high spending and good health/low spending as present but not perfect); see NEUMAN ET AL., supra note 44, at 2 (finding health to be an important factor for high annual costs); Webb & Zhivan, supra note 66, at 15 (concluding that “current good health provides only a very limited guarantee of future good health”).

105 KFF CHARTBOOK, supra note 4, at 71 fig.7.3 (including long-term care costs).

106 WEI SUN ET AL., CTR. FOR RET. RESEARCH BOS. COLL., NO. 10-8, DOES STAYING HEALTHY REDUCE YOUR LIFETIME HEALTHCARE COSTS? (2010).

107 Id. at 2 (reporting that in 2009, excluding nursing home care, a household where the husband is age seventy to seventy-four and in good health will spend $6000 on average compared to $7416 for a household with a husband not in good health—defined as having ever been diagnosed with a chronic disease).

108 Id. at 1 (including home health and nursing home costs, but not costs of assisted living facilities or long-term care insurance premiums).

109 See id.

110 Poterba, James M. et al., The Asset Cost of Poor Health 18–19 (Nat’l Bureau of Econ. Research, Working Paper No. 16389, 2010)CrossRefGoogle Scholar.

111 See generally Marshall et al., supra note 83, (finding that “out-of-pocket expenditures are often elusive” and represent a large drain on financial resources, especially for households nearing death). See also Webb & Zhivan, supra note 66, at 7; Seshamani, Meena & Gray, Alastair M., A Longitudinal Study of the Effects of Age and Time to Death on Hospital Costs, 23 J. HEALTH ECON. 217, 230 (2004)Google Scholar (“Average hospital costs increased seven-fold in the last three years of life, compared to a 30% increase from age 65-80.”).

112 Marshall et al., supra note 83, at 2. This study uses data from HRS exit interviews and normalizes spending to a twelve month period. The authors seek to omit outliers that might be erroneous. A large part of this spending, particularly at the high ends of the distribution, is for long-term care, which is beyond the scope of this study.

113 Hartman, Micah et al., U.S. Health Spending by Age, Selected Years Through 2004, 27 HEALTH AFF. W2 (Nov. 2007)CrossRefGoogle Scholar (with respect to total expenditures, insured and out-of-pocket, showing a doubling from cohorts ages sixty-five to seventy-four to ages seventy-five to eighty-four, and a tripling between ages sixty-five to seventy-four and over eighty-five); see also Webb & Zhivan, supra note 66, at 7 (reporting increasing out-of-pocket spending by age). But see Stewart, Susan T., Do Out-of-Pocket Health Expenditures Rise with Age Among Older Americans?, 44 GERONTOLOGIST 48, 50–51 (2004)Google Scholar (reporting generally no increase in out-of-pocket costs when long-term care spending is excluded and certain costs, including hospital costs, decrease).

114 Webb & Zhivan, supra note 66, at 2, 22.

115 Hurd & Rohwedder, supra note 66, at 17. Their mean and median estimates based on MCBS and MEPS data are lower. Id.

116 Berk, Marc L. & Monheit, Alan C., The Concentration of Health Care Expenditures, Revisited, 20 HEALTH AFF. 9 (2001)Google ScholarPubMed.

117 See Reschovsky, James D. et al., Following the Money: Factors Associated with the Cost of Treating High-Cost Medicare Beneficiaries, 46 HEALTH SERVICES RES. 998 (2011)Google Scholar.

118 Newhouse, supra note 22, at 1256.

119 JOHNSON & MOMMAERTS, supra note 6, at 11.

120 Hurd & Rohwedder, supra note 66, at 17 (explaining that based on HRS data for a retiree in the sixty-five to sixty-nine age bracket, they estimate $720 at the median and $21,950 at the ninety-ninth percentile; for those eighty-five and older, spending is $950 at the median to $25,150 at the ninety-ninth percentile).

121 FRONSTIN ET AL., supra note 1, at 9.

122 Id.

123 Id. For a couple turning sixty-five in 2009, one study estimated a doubling of expenditures from $260,000 on average to $570,000 at the ninety-fifth percentile, including nursing home care but excluding the costs of assisted living facilities. Webb & Zhivan, supra note 66, at 20. Excluding all nursing home care, the average and ninety-fifth percentile estimates were $197,000 and $311,000—still an over 50% increase from the mean to the ninety-fifth percentile. Id.

124 BDS. OF TRUSTEES 2011, supra note 75, at 202-04.

125 CONG. BUDGET OFFICE, supra note 27, at 27 (reporting that from 1975-2008, excess cost growth in Medicare was 2.5%, in Medicaid was 2.0%, in all other was 1.8%, and overall was 1.9%).

126 See Vladeck, Bruce C., Fixing Medicare's Physician Payment System, 362 NEW ENG. J. MED. 1955 (2010)Google Scholar.

127 Aaron, Henry, The Independent Payment Advisory Board – Congress's “Good Deed,” 364 NEW ENG. J. MED. 2377 (2011)CrossRefGoogle Scholar.

128 Id. at 2378-79.

129 Editorial, We Thought They Were Worried About Costs, N.Y. TIMES, Mar. 9, 2012, at A30.

130 KFF CHARTBOOK, supra note 4, at 79.

131 H.R. Con. Res. 112, 112th Cong. (2012) (proposing to replace Medicare with a “premium support” program); see also HENRY J. KAISER FAMILY FOUND., PROPOSED CHANGES TO MEDICARE IN THE “PATH TO PROSPERITY” 1 (April 2011) (summarizing terms of Paul Ryan plan).

132 PAUL RYAN, HOUSE BUDGET COMM., THE PATH TO PROSPERITY: A BLUEPRINT FOR AMERICAN RENEWAL 52-55 (2012), available at http://budget.house.gov/fy2013prosperity/. Several plans, including Representative Ryan's and legislation introduced by Senators Lieberman and Coburn in 2011, also propose to increase the Medicare eligibility age to sixty-seven. Id.; see also Rick Unger, The Coburn-Lieberman Medicare Proposal - The Good, the Bad and the Ugly, FORBES (June 29, 2011, 12:46 PM), http://www.forbes.com/sites/rickungar/2011/06/29/the-coburn-lieberman-medicare-proposal-the-good-the-bad-and-the-ugly/.

133 RYAN, supra note 132, at 53.

134 See CONG. BUDGET OFFICE, LONG-TERM ANALYSIS OF A BUDGET PROPOSAL BY CHAIRMAN RYAN (2011), available at http://www.cbo.gov/sites/default/files/cbofiles/ftpdocs/121xx/doc12128/04-05-ryan_letter.pdf.

135 ROBERT BERENSON & JOHN HOLOHAN, ROBERT WOOD JOHNSON FOUND. & URBAN INSTIT., HOW WILL THE PATIENT PROTECTION AND AFFORDABLE CARE ACT AFFECT SENIORS?: TIMELY ANALYSIS OF HEALTH POLICY ISSUES 1-2 (2010).

136 Id. at 2. Part D led to a reduction of $180 in annual out-of-pocket costs for the median participant and $800 at the ninetieth percentile. ENGELHARDT & GRUBER, supra note 85, at 3-4. Prior to PPACA, after just over $3000 in spending, retirees would enter the so-called “donut hole” in coverage where they had to pay 100% of the next $3610 in spending before reaching the “catastrophic coverage” level ($6440 in 2010), after which Medicare and the plan together pay 95% of the costs. A beneficiary would spend $4550 total out-of-pocket on cost-sharing before qualifying for catastrophic coverage. HENRY J. KAISER FAMILY FOUND., MEDICARE PRIMER 7 (2010).

137 PAUL FRONSTIN ET AL., EMP. BENEFIT RES. INST., NO. 8, NOTES: THE IMPACT OF REPEALING PPACA ON SAVINGS NEEDED FOR HEALTH EXPENSES FOR PERSONS ELIGIBLE FOR MEDICARE 3 (2011), available at http://www.ebri.org/pdf/notespdf/Notes.Aug11.PPACA-Final.pdf.

138 See Patient Protection and Affordable Care Act of 2010, Pub. L. No. 111-148, §§ 4104-4108, 124 Stat. 119, 557-64, amended by Health Care and Education Reconciliation Act of 2010, Pub. L. No. 111-152, 124 Stat. 1029 (to be codified as amended in 42 U.S.C. §§ 1395-1396).

139 See BERENSON & HOLAHAN, supra note 135, at 2-4 (discussing PPACA efforts to reduce provider payment rates through the IPAB, Accountable Care Organizations, and other delivery reform policies).

140 Id. at 2.

141 CONG. BUDGET OFFICE, COMPARISON OF PROJECTED ENROLLMENT IN MEDICARE ADVANTAGE PLANS AND SUBSIDIES FOR EXTRA BENEFITS NOT COVERED BY MEDICARE UNDER CURRENT LAW AND UNDER RECONCILIATION LEGISLATION COMBINED WITH H.R. 3590 AS PASSED BY THE SENATE (2010), available at http://www.cbo.gov/sites/default/files/cbofiles/ftpdocs/113xx/doc11379/macomparisons.pdf.

142 See supra note 47.

143 See Patient Protection and Affordable Care Act of 2010, Pub. L. No. 111-148, § 9001(a), 124 Stat. 119, 847, amended by I.R.C. § 49801(b) (West 2010) (to be codified as amended in 26 U.S.C. § 4980I). Other policies could have a similar effect. For example, starting in 2013, the subsidy to employers who offer retiree drug coverage will also be taxed, eliminating an exemption created under the Medicare Modernization Act and costing employers an additional $233 per retiree on average that must be reported as a liability in annual reports. PAUL FRONSTIN, EMP. BENEFIT RES. INST, NO. 338, ISSUE BRIEF: IMPLICATIONS OF HEALTH REFORM FOR RETIREE HEALTH BENEFITS 12 (2010), http://www.ebri.org/pdf/briefspdf/EBRI_IB_01-2010_No338_Ret-Hlth.pdf.

144 See Patient Protection and Affordable Care Act § 9001.

145 Id.

146 In an effort to validate the form of the survey, we circulated preliminary versions to a number of experts and conducted a small pilot survey to ensure that questions were comprehensible and answers appropriate. In light of comments received in this process, we revised the survey and attempted to reduce the complexity of the questions to tenth grade reading comprehension level or lower.

147 For additional information on the Rand American Life Panel, see Panel Composition, RAND AM. LIFE PANEL, https://mmicdata.rand.org/alp/index.php?page=panelcomposition (last updated Sept. 6, 2012).

148 For purposes of the analysis presented here, our survey responses were uniquely weighted to ensure that our cohort samples are representative of the national population for non-institutionalized individuals over the age of forty. Rand weights are generated using an iterative raking algorithm adjusting for gender, education, age, and income. For a general overview of the Rand weights, see Panel Weighting, RAND AM. LIFE PANEL, https://mmicdata.rand.org/alp/index.php?page=weights (last visited Dec. 14, 2012).

149 In our survey, we offered respondents the following guidance about the kinds of costs we were interested in:

“In this survey, we want to find out how much you expect to pay for healthcare in retirement. We are interested in your out-of-pocket costs. Out-of-pocket costs are any expenses that you pay yourself. In addition to any direct payments, these costs include insurance premiums for government programs and other health insurance plans. Out-of-pocket costs also cover deductibles and co-pays. Out-of-pocket costs do not include payments made on your behalf or reimbursed by government programs or other insurance plans. In all cases, we are asking about your own personal healthcare costs in retirement. Do not include healthcare costs of other members of your household. Unless otherwise indicated, please do not include in your estimates the cost of long-term residential health-care services (such as extended stays in nursing homes) or premiums for long-term healthcare insurance. Some questions ask for estimates about costs in the future. Please do not attempt to adjust your estimates to reflect price increases from overall inflation. Just make your estimates using the value of money today.”

150 For survey instructions with regard to excluding long-term care cost, see supra note 149.

151 The survey introduction to these questions read as follows:

“Many different government programs and insurance plans can cover healthcare expenses of retirees. With all these choices, many people are confused which plans and programs will provide them coverage. The next questions ask how likely you think it is that particular government programs and insurance plans will cover at least a portion of your healthcare expenses at some point in retirement. If you are certain that you will be covered, you should click the ruler on 100%. If you are certain that you will not be covered, you should click the ruler at 0%. If you think you may be covered but are not sure, click on the scale on the point on the ruler that best reflects your assessment of the likelihood that you may be covered or type the number reflecting that likelihood in the box below.”

152 We also surveyed respondents about their expectations regarding coverage from Veterans Administration programs but do not report those results in this analysis.

153 The mean response can be interpreted as the expected coverage level across all respondents because we asked each respondent to report the likelihood of personal coverage.

154 While the benchmarks in the literature review might be slightly low, especially with regard to Medigap for reasons discussed above, any underreporting of Medicaid or Medigap coverage would be small at most.

155 Here, and elsewhere throughout this Article, we make comparisons between responses of younger and older cohorts. Where differences are noted, we tested for statistical significance using a Hodges-Lehman non-parametric estimator for the median differences across groups, incorporating probabilistic weighting of data. We used non-parametric tests because in many instances the distributions of responses were skewed. Unless otherwise indicated, the differences were significant at the 99% confidence level.

156 On the other hand, as discussed below, the younger cohorts do not estimate consistently higher levels of out-of-pocket expenses than other cohorts of the sort that one would expect if younger cohorts were consciously anticipating less generous Medicare coverage in the future.

157 For those respondents in Treatment B and C who indicated that they thought there was some probability that they would maintain Medicare coverage at some point in retirement, we asked whether they expect to maintain Part D Prescription Drug Coverage or to participate in Medicare Advantage. Respondents overwhelming reported that they expect to maintain Part D Prescription Drug coverage (quite consistent with the 60% coverage levels reported in the expert literature). Of respondents giving definitive answers, over 75% indicated that they expected to have Part D coverage (611 of 814). Respondents reported greater uncertainty about Medicare Advantage participation, with nearly half of all respondents reporting that they did not know or had not decided about the issue. Those giving a firm answer to the question reported a good deal higher level of Medicare Advantage take-up (281 of 640 or nearly 44%) than the literature review indicates is currently the case (25% of current Medicare beneficiaries). The figures reported in this footnote are not weighted.

158 For purposes of this and similar tables below, we have not attempted to eliminate outliers in the data. See discussion infra note 190.

159 See discussion supra Part II.A.

160 As discussed below, cost estimates among respondents are positively correlated with income levels, and this is also true of total Medicare premium estimates, where the median estimate of respondents in the top income quintile was $250, whereas the median estimate of those in the bottom quintile was $100. These figures are based on a combination of respondents in Treatments B and C.

161 Note also that mean estimates for these monthly costs skew high, pulled up by a handful of respondents who tend to “high-ball” their estimates, perhaps reflecting unwillingness or possibly an inability to respond to our estimation requests.

162 Additional information on our anchoring information is available from the authors.

163 Whether the difference in medians between Treatment B and Treatment C is meaningful for policy purposes is an interesting question. As noted below, Treatment B and Treatment C estimates for total out-of-pocket costs were surprisingly similar both to each other and to the estimates of Treatment A respondents. So however one judges the difference in median estimates about premiums, those differences largely disappear when respondents were asked to estimate overall costs.

164 One alternative benchmark for these purposes is the 2010 benchmark as the closest reflection of current costs, and at several points we refer to that alternative benchmark for illustrative purposes, but it is less appropriate for the typical respondent. Yet another approach would have been to use the 2030 benchmark on the grounds that our average respondent would spend much of retirement in years beyond 2020 with higher costs. We discuss below how using the 2030 benchmark would affect the analysis in certain respects but decided on 2020 as the primary benchmark, both because our benchmarks do not incorporate likely reductions in spending from PPACA (and thus are all arguably high) and also because, as also discussed above, it is possible that our instruction to estimate spending in present day dollars might have dissuaded respondents from considering growth in healthcare costs.

165 Treatment A respondents were asked a single question about average monthly costs during retirement and for these respondents we used that single estimate in Table Four and accompanying figures. Respondents in Treatments B and C were asked to give different monthly estimates for age sixty-five, seventy-five, and eighty-five. Respondents who were sixty-five or older were first asked for their current average monthly estimates and then also asked to estimate average monthly expenses at seventy-five (if they were not yet seventy-five) and eighty-five. For respondents in Treatment B and C, average monthly costs is the average of estimates of all of their monthly estimates.

166 The left hand column of Figure Two A indicates what share of the responses were below the 2020 expert benchmark for the twenty-fifth percentile of expenditures; in the next column the share that fell between the twenty-fifth percentile and the median; in the next column the share between the median and the seventy-fifth percentile; in the next column the share between the seventy-fifth percentile and the ninetieth percentile; and in the final column the share above the ninetieth percentile. In the unlikely event that our respondents’ estimates perfectly matched this expert benchmark, the first three columns of the histogram would equal 75% of respondents and the final two columns would sum to 25% (with 15% in the fourth column and 10% in the fifth).

167 If one were to use the 2010 benchmarks rather than the 2020 benchmarks to make these comparisons, similar, though somewhat less pronounced results would be produced. With the 2010 benchmark, respondents’ median estimate of $200 was only marginally below the benchmark median of $215. Slightly over 32% exceeded the seventy-fifth percentile of the 2010 benchmark with 20.4% above the ninetieth percentile. On the left hand side of the distribution, 41.9% of respondents had estimates below the twenty-fifth percentile. In short, while our survey responses track more closely the 2010 benchmarks, the lower end of the survey distribution is still substantially over-represented in the bottom quartile of the benchmark. Were one to employ the 2030 benchmark, which had a median estimate of $381, only 14.5% of survey responses were above the seventy-fifth percentile and only 6.2% above the ninetieth percentile. Measured against the 2030 benchmarks, more than 60% of respondents had estimates beneath the twenty-fifth percentile.

168 An alternative interpretation is that respondents may have understood our instructions to express answers in terms of current dollars and not to adjust for general inflation as guidance that they should avoid any source of increase in out-of-pocket costs, whether from excess medical care cost growth or the reduction of government insurance programs. In this case, our responses might reflect confusion regarding general inflation versus other economic or cost growth.

169 There is a risk of demand effect, namely that inquiring about last year costs suggests that such costs will be higher. Even if demand effect is occurring here, it is nonetheless interesting that the magnitude of estimates is close to experts’ estimates and that, when prompted, people intuit higher costs in their final year.

170 This expert ratio is calculated based on the estimate in the Marshall study of the median last year of life ($5061) divided by the overall median annual estimate in the Johnson and Mommaerts study ($2583). Marshall et al., supra note 83, at 37; see also JOHNSON & MOMMAERTS, supra note 6, at 11.

171 A comparable increase of 33% can be seen in the median estimates from the Hurd & Rohwedder study of $720 for a sixty-five to sixty-nine year old and $950 for an over eighty-five year old. See Hurd & Rohwedder, supra note 66.

172 The actual question read as follows:

“In planning for retirement, some individuals like to think in terms of how much money they would need to save by the time they turn 65 in order to have enough money to cover out-of-pocket costs in retirement. Imagine that you were asked to give advice to someone with similar preferences and health characteristics as your own. If such a person wanted to have enough money to cover a reasonable estimate of their total out-of-pocket costs for healthcare in retirement, how much do you think they would need to have set aside? Please give your answer in terms of the total amount of dollars needed at age 65.”

173 It is possible—as one reader noted—that respondents could interpret this question to mean how much they would need to have saved to avoid bankruptcy or significant retirement risk, rather than to cover all out-of-pocket costs. In this case, retirees’ estimates would be lower than their expectations of total costs, in which case their expectations of total costs would be even closer to experts’ estimates than we report herein.

174 See generally Brown, Jeffrey R. et al., Framing and Claiming: How Information-Framing Affects Expected Social Security Claiming Behavior (Nat’l Bureau of Econ. Research, Working Paper No. 17018, 2011)CrossRefGoogle Scholar (showing that individuals’ choices on when to claim Social Security benefits, from ages sixty-two to seventy, vary based on how this claiming decision is framed); Jeffrey R. Brown, et al., Do Consumers Know How to Value Annuities? Complexity as a Barrier to Annuitization (June 7, 2012) (unpublished manuscript) (on file with authors) (showing difficulty among survey respondents in valuing annuities).

175 The absence of a twenty-fifth percentile benchmark for lump sum estimates makes it harder to identify the extent to which our lump sum responses gravitate to the left hand side benchmark distributions. However, for our full sample as well as both male and female subsamples, the twenty-fifth percentile response was $10,000, very far below median lump sum benchmarks for either men ($109,000) or women ($156,000). Thus, it appears quite likely that our lump sum responses were also skewed to lower end estimates.

176 One sees similar differentials if survey lump sum responses are measured against the 2010 benchmarks. The 2010 benchmark median for women is $93,000, which is almost three times our actual median female estimate of $30,000, whereas the 2010 benchmark median for men is $65,000, which is actually quite close to the median male estimate of our respondents: $60,000.

177 As noted below, respondents did slightly underestimate the likelihood that they would survive to ages sixty-five and seventy-five, when compared with expert assessments, which could explain one of the reasons why respondents’ lump sum estimates fall a bit further beneath expert estimates than was the case with respondents’ average monthly cost estimates.

178 For the wording of our survey question, see supra note 172.

179 One hypothesis suggested by a workshop participant is these responses reflect an expectation of younger respondents that political forces will not allow out-of-pocket costs to increase above $50,000 in current dollars. Under this view, respondents collectively might have a more accurate view of future out-of-pocket costs than experts focusing primarily on past trends and without accounting for political constraints.

180 Conclusions with respect to younger cohort responses are probably best drawn from a complete review of survey response. We attempt such a summary in the conclusion of this paper.

181 Framing of the questions of life expectancy can have an effect on responses, as reported in a recent study in Payne, John W. et al., Life Expectancy as a Constructed Belief: Evidence of a Live-to or Die-by Framing Effect (Columbia Bus. Sch. Research, Paper No. 12-10, 2012)CrossRefGoogle Scholar. Questions framed as the probability of “living to” a particular age generate higher average estimates than those framed in terms of “dying by” that same age. Id. at 5. When compared with estimates of life expectancy, based on SSA data and adjusted to each respondent's age and gender, the “living to” frame produced subjective estimates closer to the actual estimates than the “dying by” frame at ages sixty-five and seventy-five, equally accurate at eighty-five, and less accurate at ninety-five, when both frames lead to overly optimistic responses. Id. at 9 fig.3a. For purposes of our survey,, we adopted a “living to” frame to survey respondents’ anticipated life expectancy.

182 See Actuarial Life Table, supra note 101. Our median responses for surviving past ninety-five closely match expert views (9%). Respondents were directionally accurate in reporting longer life expectancies for women than for men.

183 As respondents were instructed to estimate future monthly costs in terms of current dollars, a real (as opposed to nominal) discount rate was employed. As a robustness check, we recalculated respondents’ implied lump sums using both a 3.0% and a 0% discount rates. With the 3.0 % discount rate, median response for cohorts range between $35,000 and $45,000 with an overall median of about $10,000 lower than actual lump sum estimates. With a 0% discount rate, the median response for cohorts ranged from $45,000 to $75,000 with an overall median about $8000 higher than actual lump sum estimates. These results crudely suggest that respondents may be using a mental discount rate closer to 1.5%.

184 Even with a 0% discount rate, the median implied lump sum estimate was just under $58,000.

185 The upper range of actual lump sum estimates exceed implied lump sum estimates even when calculated using a 0% discount rate. So, for example, the ninetieth percentile implied lump sum of the forty-five to forty-nine age cohort calculated with a 0% discount rate is roughly $225,000 compared with the $750,000 ninetieth percentile estimate for actual lump sum for that age cohort.

186 We focus here on findings that the regression analysis, discussed below, suggests are significant. Monthly estimates increased with higher levels of self-reported health status of respondents, as reported in Table Eight, but the regression analysis below suggests this effect may simply reflect interactions with income or education level.

187 Wealthier respondents also gave higher estimates of total Medicare premiums. See supra notes 95-99.

188 See supra note 102 and accompanying text.

189 See supra note 103.

190 Responses to questions regarding familiarity with government programs and the level of attention they pay to monthly healthcare costs and other expenses showed little or no effect on estimates. Although familiarity does not appear to be strongly associated with differences in respondent estimates, older respondents reported a much higher degree of familiarity on these dimensions than did younger respondents.

191 We utilized these functional forms to limit the influence of outliers on the analysis. As noted earlier, some respondent estimates—especially with respect to lump sum estimates—seemed unreasonably high, suggesting that perhaps some responses may have been protest bids in response to inherently difficult questions or for some other reason providing implausible answers. In prior sections of our analysis we have relied on median and percentile analysis to limit the impact of these outliers. As compared to traditional OLS regression, quantile regressions serve a similar function. The log form of the fourth model also reduces the influence of outliers as does our trimming of the top 1% of observations, which eliminates survey responses with average monthly costs over roughly $2500.

192 When we segmented the sample into male and female subsamples and re-ran our regressions, the coefficient for the younger cohort variables remained statistically significant for only the male subsample, suggesting that men were driving the higher cost estimates from younger cohorts reported in the text.

193 Trimming of the top 1% of lump sum estimates eliminated responses with estimates in excess of $3.0 million on the fourth model.

194 In other regression runs not reported here, we found the relationship between respondents’ estimates and expectations regarding health insurance coverage were generally consistent with expert estimates of the relationship between supplemental coverage and total out-of-pocket costs. For example, respondents reporting higher expectations of having Medicaid coverage made lower estimates of out of pocket costs, while those expecting Medigap coverage estimated higher spending.

195 See supra note 121 and accompanying text.

196 See supra note 134 and accompanying text. It is possible that our respondents chose a middle option out of the five multiple choice responses. Nonetheless, such random selection would suggest ignorance. Furthermore, answers with regard to individual health experience were skewed more toward the second choice and policy uncertainty toward the third, which confirms less concern with individual health experience than with policy uncertainty.

197 An illustrative question here read as follows: “Research suggests that healthcare expenses in retirement can vary considerably from individual to individual based on differences in the health of individuals and their medical needs. As a result, out-of-pocket costs for some individuals can be much higher than those of the average retiree. How much would you be willing to pay each month for an insurance policy that fully protected you from incurring out-of-pocket costs higher than those of the average retiree, regardless of your own health or medical needs?”

198 The cohort on the eve of retirement, ages sixty to sixty-four, did, however, skew somewhat higher at the top end of the distribution in their willingness to pay for insurance against policy changes and against high individual healthcare costs.

199 As discussed above, a number of reasons could explain why respondents fell somewhat further beneath expert benchmarks for their lump sum estimates than they did for monthly cost estimates. In part, the difference could be a result of respondents somewhat underestimating their likelihood of surviving to age seventy-five. In addition, respondents may have had difficulty in calculating what is essentially the NPV of expected healthcare costs, although, as noted above, our implied lump sum analysis suggests that at least some respondents erred in overestimating the NPV equivalent of their projected monthly costs discounted by self-reported life expectancies. In addition, some respondents may have misinterpreted our question about lump sum estimates. Finally, our benchmark studies are based on somewhat different assumptions and do not perfectly align with each other.

200 See supra note 193.

201 Lusardi, Annamaria, Financial Literacy: An Essential Tool for Informed Consumer Choice? 2 (Nat’l Bureau of Econ. Research, Working Paper No. 14084, 2008)Google Scholar (explaining that financial “[i]lliteracy is widespread among the general population and particularly acute among specific demographic groups, such as women”); Dan Kadlec, Women and Money: Even College Grads Flunk Personal Finance, TIME (June 28, 2012), available at http://moneyland.time.com/2012/06/28/women-and-money-even-college-grads-flunk-personal-finance/ (describing a new study that reports low financial literacy scores for women); Annamaria Lusardi & Olivia Mitchell, Financial Literacy and Retirement Planning: New Evidence from the Rand American Life Panel 10 (Mich. Retirement Research Ctr., Paper No. WP 2007-157, 2007), available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1095869; see also Lusardi, Annamaria et al., Financial Sophistication in Older Population 11 (Nat’l Bureau of Econ. Research, Working Paper No. 17863, 2012)CrossRefGoogle Scholar.

202 For an overview of cognitive biases with respect to risk and uncertainty, see Christine Jolls & Cass R. Sunstein, Debiasing through Law, 35 J. LEGAL STUD. 199, 203-25 (2006). See also Howell E. Jackson, Accounting for Social Security Benefits, in BEHAVIORAL PUBLIC FINANCE 261, 271-75 (Edward J. McCaffery & Joel Slemrod eds., 2006). For an overview of the annuitization puzzle with a helpful review of the literature, see Jeffrey Brown, Understanding the Role of Annuities in Retirement Planning, in OVERCOMING THE SAVINGS SLUMP: HOW TO INCREASE THE EFFECTIVENESS OF FINANCIAL EDUCATION AND SAVINGS PROGRAMS 178-206 (Annamaria Lusardi ed., 2008).

203 De Nardi et al., supra note 11, at 72-73 (describing the strong effect of future healthcare spending needs on savings).

204 Of course, individuals do not need to be primed to save an amount sufficient to cover total retiree healthcare costs. See generally Barbara A. Butrica et al., The Changing Impact of Social Security on Retirements Income in the United States, SOC. SEC. BULLETIN, no. 3, 2003/2004, available at http://www.ssa.gov/policy/docs/ssb/v65n3/v65n3p1.pdf. As a result of social security, most American have some amount of annuity income in retirement and a portion of that income could be used to support monthly costs. Id. Thus, sensible financial planning for retiree healthcare costs might consist of a combination of precautionary savings and budgeted monthly costs.

205 Amir & Lobel, supra note at 19, at 20 (“P]erhaps more than with some policy fields … health policy cannot be simply about directing healthy behavior but must aim for an understanding of how individuals reason and decide.”).

206 See Daniel Schwarcz, Transparently Opaque: Understanding the Lack of Transparency in Insurance Consumer Protection 9, 18-19 (Minnesota Legal Studies Research Paper Series, Paper No. 12-35, 2012), available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2130908.

207 Lusardi, supra note 15, at 20-23 (describing the mostly ineffective results of financial education programs).

208 See Schwarcz, supra note 206, at 18-19.

209 See supra note 19 and accompanying text; see also Lusardi, supra note 15, at 23-26 (discussing studies showing the effectiveness of default programs for savings, including the lauded Save More Tomorrow (SMarT) program devised by Richard Thaler and Shlomo Benartzi that increases default savings rates as employees’ income increases).

210 See Lusardi, supra note 15, at 26-20 (discussing efforts to simplify decisions to save, without using strong defaults).

211 FISHMAN, supra note 5, at ix (reporting the effects incentives programs have on savings for low-income populations).

212 Id.

213 See THALER & SUNSTEIN, supra note 19, at 72 (describing cognitive biases that might produce undesirable results even in the face of perfect knowledge); Korobkin, Russell B. & Ulen, Thomas S., Law and Behavioral Science: Removing the Rationality Assumption from Law and Economics, 88 CALIF. L. REV. 1051, 1084-85 (2000)Google Scholar (discussing studies showing individual biases in decision-making in the face of uncertainty).

214 See Michael Bond, Risk School, 461 NATURE 1189, 1191 (2009) (describing debate over whether people can be taught to understand risk and make well-informed decisions based on it or whether it is more appropriate for regulators to guide consumers to better risk decisions through a “nudge approach”). But see Rachlinski, Jeffrey J., The Uncertain Psychological Case for Paternalism, 97 NW. U. L. REV. 1165, 1206-19 (2003)Google Scholar (describing how understanding the heuristics that people use in decision-making can be the basis for policies to help them make better decisions).

215 See, e.g., On Amir & Orly Lobel, Risk Management for the Future: Age, Risk and Choice Architecture 19-23 (July 9, 2012) (unpublished manuscript), available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2102541 (discussing use of choice architecture to help especially older people make more future-oriented, less risky choices).

216 See Dorn & Shang, supra note 38 (discusses strategies for increasing MSP enrollment).

217 PPACA attempts to address this shortcoming for working-age populations by requiring reporting on the actuarial value of policies that will be sold in the new health exchanges; despite good intentions, actuarial value may be too complicated a measure for most consumers to translate into likely personal spending. See RYAN LORE ET AL., COMMONWEALTH FUND, CHOOSING THE “BEST” PLAN IN A HEALTH INSURANCE EXCHANGE: ACTUARIAL VALUE TELLS ONLY PART OF THE STORY 6 (2012), available at http://www.commonwealthfund.org/∼/media/Files/Publications/Issue%20Brief/2012/Aug/1626_Lore_choosing_best_plan_HIE_actuarial_ib_v2.pdf. Also pursuant to PPACA, the National Association of Insurance Commissioners (NAIC) developed a new form for summaries of coverage. See NAIC, SAMPLE COMPLETED SUMMARY OF COVERAGE (2012), available at http://www.naic.org/documents/committees_b_consumer_information_hhs_dol_submission_1107_soc_populated.pdf.

218 See THALER & SUNSTEIN, supra note 19, at 72 (describing cognitive biases that might produce undesirable results even in the face of perfect knowledge); Korobkin & Ulen, supra note 215, at 1091-92 (describing overconfidence biases).

219 Marsha Gold, An Illustrative Analysis of Medicare Options Compare: What's There and What's Not?, 27 INSIGHT ON ISSUES, April 2009, available at http://assets.aarp.org/rgcenter/health/i27_options.pdf.

220 See Karen Davis et al., In the Literature: Medicare Extra: A Comprehensive Benefit Option for Medicare Beneficiaries, HEALTH AFF. (Oct. 2005), http://content.healthaffairs.org/content/early/2005/11/15/hlthaff.w5.442.short.

221 See generally Thomas Rice et al., The Medicare Catastrophic Coverage Act: A Post-Mortem, 9 HEALTH AFF. 75 (1990) (describing factors that explain the repeal of this Act that attempted to limit the catastrophic costs a retiree might face by imposing additional taxes on higher-income elderly).

222 A study of several Medigap reform proposals that attempt to control the growth in Medicare spending by increasing cost-sharing, with varying limits on total out-of-pocket costs, showed cost reduction for most enrollees but cost increases for about 21%, disproportionately affecting those in fair/poor health and lower-income enrollees. HENRY J. KAISER FAMILY FOUND., MEDIGAP REFORMS: POTENTIAL EFFECTS OF BENEFIT RESTRICTIONS ON MEDICARE SPENDING AND BENEFICIARY COSTS v (2011).

223 See Patient Protection and Affordable Care Act of 2010, Pub. L. No. 111-148, § 3210, 124 Stat. 119, 461, amended by Health Care and Education Reconciliation Act of 2010, Pub. L. No. 111-152, 124 Stat. 1029 (to be codified as amended in scattered sections of 42 U.S.C.).

224 LINEHAN, supra note 55, at 11 (describing proposals from the Simpson-Bowles commission, the Congressional Budget Office, and the Obama Administration); see also Amanda Cassidy, Health Policy Brief: Putting Limits on ‘Medigap, ’ HEALTH AFF. (Sept. 21, 2011), http://www.healthaffairs.org/healthpolicybriefs/brief.php?brief_id=52.