Incorporating Data on Assets into Measures of Financial Burdens of Health1

Jessica S. Banthin, Congressional Budget Office
and
Didem Bernard, Agency for Healthcare Research and Quality

INTRODUCTION

In assessments of out-of-pocket burdens for health care, annual income is used to measure the available resources.2 This approach is consistent with poverty measurement, which is also based on gross annual income as reported in the Current Population Survey of households. Assets, however, are counted only to the extent that asset income, such as interest and dividends, is included in the measure of total money income. Although asset holdings may be difficult to measure well in household surveys, it is likely that asset holdings serve as an important financial resource for families confronted by a temporary loss of income resulting from, for example, a spell of unemployment. Similarly, assets are likely to serve as an important financial resource for families with high out-of-pocket medical expenses, especially in the case of unexpected medical expenses.

If asset holdings are generally correlated with income, then ignoring assets in measures of poverty or out-of-pocket burdens for health care may not result in biased or misleading comparisons between population subgroups. That is, if one does not believe certain groups have systemati-

___________________________________

1 The views expressed in this paper are those of the authors and no official endorsement by the Congressional Budget Office, the Agency for Healthcare Research and Quality, or the Department of Health and Human Services is intended or should be inferred, and does not necessarily reflect the views or conclusions of the National Research Council, the Institute of Medicine, the study panel, or the sponsor.

2 Some studies have computed annual disposable income net of taxes, whereas other studies have used gross annual income as reported by survey respondents.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 267
Incorporating Data on Assets into Measures of Financial Burdens of Health1 Jessica S. Banthin, Congressional Budget Office and Didem Bernard, Agency for Healthcare Research and Quality INTRODUCTION In assessments of out-of-pocket burdens for health care, annual income is used to measure the available resources.2 This approach is consistent with poverty measurement, which is also based on gross annual income as reported in the Current Population Survey of households. Assets, however, are counted only to the extent that asset income, such as interest and divi- dends, is included in the measure of total money income. Although asset holdings may be difficult to measure well in household surveys, it is likely that asset holdings serve as an important financial resource for families confronted by a temporary loss of income resulting from, for example, a spell of unemployment. Similarly, assets are likely to serve as an important financial resource for families with high out-of-pocket medical expenses, especially in the case of unexpected medical expenses. If asset holdings are generally correlated with income, then ignoring assets in measures of poverty or out-of-pocket burdens for health care may not result in biased or misleading comparisons between population subgroups. That is, if one does not believe certain groups have systemati- 1  The views expressed in this paper are those of the authors and no official endorsement by the Congressional Budget Office, the Agency for Healthcare Research and Quality, or the Department of Health and Human Services is intended or should be inferred, and does not necessarily reflect the views or conclusions of the National Research Council, the Institute of Medicine, the study panel, or the sponsor. 2  Some studies have computed annual disposable income net of taxes, whereas other studies have used gross annual income as reported by survey respondents. 267

OCR for page 267
268 MEDICAL CARE ECONOMIC RISK cally higher levels of assets than other groups with similar income, then the current approach measures poverty or health care burdens consistently across policy-relevant subgroups. It is worth examining these assumptions, however, with respect to the elderly, who, simply by virtue of age, have had more time to accumulate assets than younger families with the same in- come.3 If the elderly do have systematically higher levels of assets, all other things equal, then income-based measures of financial deprivation may be misleading. This issue may be particularly relevant to the measurement of medical care burdens, because health care expenses due to illness and dis- ability are widely recognized as one of the major financial risks of old age for which to save. The question then becomes one of how to incorporate assets into an income-based measure. A growing literature has examined out-of-pocket expenditures for medical care as a function of income. The literature typically defines one or more thresholds, such as 10 and 20 percent of family income, so that the distribution of the population according to the thresholds can be reported. As explained elsewhere, this approach reduces bias due to reporting error in income and provides an intuitive measure of the risk of incurring high medi- cal burdens (Banthin and Bernard, 2006). This threshold approach mirrors the method used in measuring poverty, which is also based on thresholds. As far as we are aware, however, the literature has always analyzed medical care financial burdens and risks separately for the elderly and nonelderly subpopulations. There are two main reasons for this distinction. First, the two groups differ in their primary sources of insurance coverage. Thus, the reasons for and the policy implications of high out-of-pocket medical care burdens also differ by subpopulation. Because almost all persons ages 65 and over are covered by Medicare, the policy implications of high burdens among the elderly center on the Medicare program. Indi- viduals under age 65, in contrast, are covered primarily by employment-­ sponsored insurance, individually purchased policies, and Medicaid. Many are uninsured. The policy implications of high burdens among the non- elderly are related to the functioning of private insurance markets. A second and equally important reason for analyzing the two groups separately is methodological. Because elderly and nonelderly individuals and families spend very different proportions of their income on health care, it is difficult to define a single threshold for both age groups. The elderly and nonelderly devote different proportions of their family income to health care because both parts of the equation—their expected health care needs and their resources to meet those needs—are very differ- ent. What is an appropriate threshold for a nonelderly individual or family 3  The self-employed population is another group that may have systematically higher levels of assets. It is treated in the annex.

OCR for page 267
INCORPORATING DATA ON ASSETS 269 compared with an elderly individual or family? If the thresholds are not the same, is there a way to develop a consistent threshold for elderly and nonelderly families that recognizes their very different resources and needs regarding medical care as a part of their overall household budget? In the work that follows, we show that the distributions of burdens for elderly and nonelderly families are quite different. We then investigate the distribu- tion of assets in both groups. Finally, we compute burdens using different thresholds and different measures of resources to define high burdens. One approach incorporates 5 percent of total net assets into the resources avail- able to elderly families to pay medical expenses as a simplified method for drawing down assets in retirement. We do not make this adjustment for nonelderly families, because they are expected to be saving for the future rather than drawing down on current savings. PRIOR LITERATURE ON HEALTH CARE BURDENS In a previous study, we estimated changes in annual financial out-of- pocket burdens for medical care, for the population under age 65 (Banthin and Bernard, 2006). Our key estimate of total financial burden included out-of-pocket expenditures for health care services plus out-of-pocket ex- penditures for premiums as a function of family income. High financial bur- dens were defined using thresholds of 10 and 20 percent of family income. In another paper we applied a threshold of 5 percent of income to non- elderly families living below 200 percent of poverty (Banthin, Cunningham, and Bernard, 2008). Other studies apply 5 and 10 percent thresholds to indicate high burdens among nonelderly individuals and families (Schoen et al., 2011). Under the Patient Protection and Affordable Care Act, there are premium and cost-sharing subsidies broadly consistent with these thresh- olds that apply to the low-income population under age 65. Researchers often take a broader approach in analyzing out-of-pocket medical care spending in the elderly population. Although some papers have looked at annual burdens for medical care, another vein of research has focused on the amount of money needed to pay for medical care, in- cluding long-term care, over a lifetime. Among papers that do examine an- nual burdens, Selden and Banthin assessed changes in annual out-of-pocket burdens for medical care for the elderly between 1987 and 1996 and ap- plied thresholds of 20 and 40 percent of after-tax family income to indicate individuals living in families with high burdens (Selden and Banthin, 2003). The methodological challenge of comparing the elderly and nonelderly arises because the two subpopulations differ in terms of health care spend- ing as a function of income. Setting a common threshold for both groups against which to assess financial burden or risk is difficult. The reasons for their differences are worth reviewing. The nonelderly population is a

OCR for page 267
270 MEDICAL CARE ECONOMIC RISK working-age population, and many live in families with children. Working- age families tend to be larger in size and have higher incomes than retired families, although their official poverty rates are higher compared with the elderly.4 Working-age families have more competing demands on their resources. For example, they typically incur work-related and childrearing expenses and spend more on transportation compared with older families. In addition, working-age families should be saving from current income for future retirement or to invest in the education of their children. On average, their health needs are lower than those of the elderly. In contrast, most individuals ages 65 and over are retired or close to retirement, and few in this age group are still raising children. Upon retire- ment, individuals and families typically begin drawing down on their assets rather than continuing to save. In addition, the need for medical care grows as people age. The implication of these different consumption and saving patterns is that elderly and nonelderly families would be expected to devote different shares of family income to out-of-pocket medical care and health insurance premiums. DATA AND METHODS Using the Medical Expenditure Panel Survey The Medical Expenditure Panel Survey (MEPS) includes detailed infor- mation on medical expenditures by source of payment, including out-of- pocket payments. Information is also collected on out-of-pocket premiums, income, assets, and other individual and household data. Although data on income and expenditures support annual estimates, the information on assets is collected only once per panel, at the end of the last round of data collection. In this paper, we pool three panels together in order to increase sample sizes for the elderly and self-employed and to support analyses of the distribution of assets across different poverty groups. In MEPS, a new panel is started each year. Panels 10, 11, and 12 started in 2005, 2006, and 2007, respectively. Because asset information is collected in the second year of the panel, all measures of assets and income are adjusted for inflation to bring them to 2008 using the Consumer Price Index for Urban Areas. Although the MEPS asset variables are not currently available on pub- lic use files, they are available to any researcher to use in the Agency for Healthcare Research and Quality Data Center. The asset section of MEPS collects information on financial and nonfinancial assets. Information on debt is also collected. Financial assets include checking and savings ac- 4  As of 2010, 9.0 percent of persons ages 65 or older lived in poverty compared with 22.0 percent of children and 13.7 percent of nonelderly adults.

OCR for page 267
INCORPORATING DATA ON ASSETS 271 counts, money market funds, stocks, bonds, mutual funds, certificates of deposit, and individual retirement accounts. The measure of net worth used in this study is the net value of nonfinancial assets, including residential property, other real estate, business equity, and transportation vehicles, as well as financial assets minus all debt. We have published other papers us- ing these variables and have compared MEPS national estimates of various definitions of wealth to estimates from the Survey of Income and Program Participation (SIPP) and the Survey of Consumer Finances (SCF) (Bernard, Banthin, and Encinosa, 2009). The MEPS asset data compare well to asset information collected in SIPP; however, both surveys appear to underreport wealth holdings compared with information collected in the SCF. The value of the MEPS asset variables for this study is in the consistency of informa- tion collected across the entire population, including those both under and over age 65, along with information on income and out-of-pocket medical expenses. Regardless of underreporting, there is no indication of bias by age. For this analysis, the family is defined as the health insurance eligibility unit (HIEU), which consists of all individuals related by blood or marriage that would typically be eligible for a family policy under most private insur- ance plans. Families with half or more of their members age 65 and over are designated as elderly families. The rest are designated as nonelderly families for purposes of examining family-level assets. Construction of Measures of Out-of-Pocket Burden For this analysis, we rely on the same approach we have refined in sev- eral previous papers to calculate financial out-of-pocket burden for medi- cal care. We define out-of-pocket burden for medical care as a family-level concept in the same way that poverty is a family-level concept, because in both measures it is assumed that family-level resources are shared among individual family members. Thus, we sum out-of-pocket expenditures on health care services and premiums across all members of the family to define the numerator. Gross reported family income is used to define the denominator. The measure of income is not adjusted for taxes. In some variations of our estimates, we add 5 percent of total net as- sets to annual income in the denominator. We do this for elderly families, because they are expected to be drawing down their assets in retirement. We do not make this adjustment for nonelderly families because they are expected to be saving for their future retirement. We chose 5 percent of total net assets as the draw down percentage because this is very close to what some financial planners advise. We then compute the share of family income used to cover all medical expenses and report statistics at the family level. The resulting distribution

OCR for page 267
272 MEDICAL CARE ECONOMIC RISK is highly skewed, with a long tail of families that spend high proportions of income on medical care. After presenting the distribution, we analyze the risk of high burdens by counting families with burdens that exceed a certain threshold (e.g., 5, 10, or 20 percent). This approach provides an intuitive measure of the risk of incurring high burdens. We do not truncate the distribution of spending as a function of income as some other studies have done. It is theoretically possible for some families to spend more than 100 percent of income on medical care in a given year. RESULTS Table A-1 shows the percentile distribution of out-of-pocket burdens for the elderly and nonelderly to illustrate the differences between the two groups. Overall, the median burden for elderly families was 10.7 percent compared with just 2.9 for nonelderly families. As expected, younger fami- lies devoted a much smaller share of family income to medical care. Among those living in poverty, the elderly spent about 13.5 percent of family in- come on medical care, whereas the nonelderly spent about 2.7 percent of family income. These differences between elderly and nonelderly populations were similar across poverty groups at the median and increased at higher points in the distribution. At the 75th percentile, the burden for the elderly was about three times higher than the burden for the nonelderly at 20.7 percent versus 7.0 percent of family income. If we were to use the 75th percentile to suggest a cutoff point as the basis for measuring high burdens, then the thresholds would be quite different for the two age groups. Table A-2 presents the distribution of total net assets by family age group. This measure includes the net value of all financial and nonfinancial assets.5 In the overall section of the table, at the median, the elderly reported $146,000 in family net wealth, and the nonelderly reported $20,000. Thus, at the median, elderly families have about 7 times as much net wealth as do nonelderly families. These large disparities in net assets can be seen along all points of the distribution. Overall, at the 20th percentile, elderly families reported about $5,000 in net assets compared with zero reported by nonelderly families. At the 90th percentile, elderly families reported about $797,000 in net assets compared with $433,000 held by nonelderly families. Table A-2 also presents the distribution of assets by poverty status. Among families living below poverty, elderly families reported more than 5  Not shown are tables that examined the distribution of financial assets and retirement a ­ ssets. We chose to focus on total net assets, because this measure conveys the large differences between the two age groups.

OCR for page 267
INCORPORATING DATA ON ASSETS 273 TABLE A-1 Percentiles (PCTs) of Out-of-Pocket Total Burden, Elderly and Nonelderly Health Insurance Eligibility Units, Pooled Panels 10-12 (2008 dollars) Poverty N PCT50 PCT75 PCT90 Overall          Elderly 3,970 10.65 20.67 37.64   (0.25) (0.60) (1.35)  Nonelderly 17,513 2.93 7.01 16.00     (0.05) (0.11) (0.41) <100%          Elderly 685 13.47 57.96 *     (2.14) (20.06)   Nonelderly 3,260 2.74 19.09 *     (0.31) (1.51)   100-199%          Elderly 1,134 16.53 27.93 43.07     (0.56) (1.27) (2.01)  Nonelderly 3,849 2.58 9.04 20.36     (0.17) (0.32) (0.98) 200-399%          Elderly 1,069 13.19 20.34 31.67     (0.45) (0.77) (1.77)  Nonelderly 5,190 3.74 8.05 14.78     (0.12) (0.21) (0.46) 400%+          Elderly 1,082 6.44 10.53 17.21     (0.21) (0.34) (0.83)  Nonelderly 5,214 2.60 5.01 8.50     (0.05) (0.10) (0.20) NOTE: Standard errors are in parentheses.   *Sample size is too small to make reliable estimates. SOURCE: Medical Expenditure Panel Survey—Household Component, Panels 10-12. $20,000 in net wealth at the median, and nonelderly families reported zero. Among low-income families (with family income between 100 and 199 percent of poverty), the median net assets for an elderly family was about 33 times as much as that for a nonelderly family ($77,000 versus $2,300). Table A-3 presents four measures of burden. In the column labeled BURD10, we show the percentage of families who were spending 10 per- cent or more of family income on medical care. Overall, about 52 percent of elderly and 17 percent of nonelderly families had high burdens according to this threshold. In the next column, BURD20, we show that about 26 percent of elderly and 7 percent of nonelderly families spent 20 percent or more on medical care. In the final two columns, we use the same thresholds of 10 and 20 percent of family income, but we adjust the family income

OCR for page 267
274 MEDICAL CARE ECONOMIC RISK TABLE A-2 Distribution of Total Net Health Insurance Eligibility Unit- Level Assets by Family Age for Pooled Panel 10-12 Health Insurance Eligibility Units <100% 100-199% 200-399% 400%+ Percentiles Overall Poverty Poverty Poverty Poverty Nonelderly Health Insurance Eligibility Units 10 –318 –4,113 –3,871 –961 –8   (203) (908) (799) (513) (6) 20 0 –43 –40 0 10,295   (0) (11) (11) (0) (1,092) 30 1,500 –28 –15 2,134 37,024   (204) (8) (5) (302) (2,909) 40 6,408 –14 501 6,500 79,230   (446) (4) (188) (546) (3,962) 50 20,151 0 2,341 15,518 133,838   (1,296) (38) (261) (1,092) (6,266) 60 53,843 1,023 5,515 34,967 207,507   (2,564) (134) (490) (2,523) (7,831) 70 111,069 3,115 14,137 72,021 307,964   (3,866) (296) (1,447) (4,094) (10,214) 80 210,245 7,840 39,498 129,015 460,008   (6,779) (1,148) (3,010) (4,730) (13,658) 90 432,096 45,923 105,752 242,669 788,162   (13,090) (5,943) (6,426) (12,301) (27,702) 95 729,088 119,340 198,769 404,244 1,236,414   (26,304) (10,434) (19,051) (23,215) (52,333) N 17,513 3,260 3,849 5,190 5,214 Elderly Health Insurance Eligibility Units 10 –8 –63 –18 17 7,788   (3) (18) (6) (193) (4,843) 20 4,997 –13 1,063 6,622 102,772   (1,005) (41) (435) (2,323) (14,466) 30 37,356 779 9,641 32,122 187,887   (4,433) (313) (2,342) (7,436) (13,612) 40 88,161 3,313 40,222 89,808 265,312   (6,051) (2,698) (5,447) (9,805) (11,498) 50 146,334 20,686 77,301 136,472 355,370   (6,515) (6,095) (6,073) (7,924) (21,152) 60 215,083 51,848 111,150 190,027 469,780   (8,927) (9,929) (6,733) (10,334) (27,918) 70 298,604 101,682 159,509 252,973 640,134   (9,729) (13,996) (12,221) (11,908) (32,898) 80 450,609 185,478 234,056 348,177 959,475   (17,681) (24,155) (12,678) (18,702) (47,838) 90 796,624 302,952 376,518 546,099 1,432,970   (38,581) (17,667) (30,868) (39,165) (64,339) 95 1,226,427 423,549 522,502 807,366 2,128,943   (66,078) (60,816) (39,722) (58,823) (143,601) N 3,970 685 1,134 1,069 1,082 NOTE: Standard errors are in parentheses. SOURCE: Medical Expenditure Panel Survey—Household Component, Panels 10-12.

OCR for page 267
TABLE A-3 Alternative Income Measures and Percentage with High Burdens (5% of assets added to elderly) for Elderly and Nonelderly Health Insurance Eligibility Units, Pooled Panels 10-12   N TOTINC ADJINC BURD10 BURD20 BURD10ADJ BURD20ADJ Overall                Elderly 3,970 41,592 57,644 52.53 26.06 40.01 16.54     (992) (91,417) (0.94) (0.90) (0.98) (0.72)  Nonelderly 17,513 53,751 53,751 17.02 7.68 17.02 7.68     (682) (682) (0.36) (0.27) (0.36) (0.27) <100% poverty                Elderly 685 6,550 11,962 54.09 43.52 43.48 28.64     (199) (575) (2.50) (2.53) (2.49) (2.41)  Nonelderly 3,260 7,282 7,282 33.20 24.54 33.20 24.54     (141) (141) (1.19) (1.04) (1.19) (1.04) 100-199% poverty                Elderly 1,134 15,435 22,628 70.75 40.72 57.40 26.58     (162) (461) (1.55) (1.88) (1.72) (1.57)  Nonelderly 3,849 20,516 20,516 22.43 10.23 22.43 10.23     (212) (212) (0.87) (0.68) (0.87) (0.68) 200-399% poverty                Elderly 1,069 30,295 41,772 63.51 26.01 48.48 15.37     (418) (730) (1.58) (1.61) (1.76) (1.22)  Nonelderly 5,190 40,140 40,140 18.77 5.90 18.77 5.90     (339) (339) (0.70) (0.42) (0.70) (0.42) 400%+ poverty                Elderly 1,082 85,197 116,230 27.35 7.76 17.00 4.91     (2,050) (2,981) (1.50) (0.89) (1.39) (0.74)  Nonelderly 5,214 96,085 96,085 7.46 2.05 7.46 2.05     (1,013) (1,013) (0.36) (0.22) (0.36) (0.22) NOTES: Standard errors are in parentheses. ADJINC = adjusted income (dollars); BURD10 = percentage of families who were spending 10 percent or more of family income on medical care; BURD20 = percentage of families who were spending 20 percent or more on medical care; BURD10ADJ = percentage of elderly families with high out-of-pocket burdens; BURD20ADJ = percentage of families with high burdens; TOTINC = total adjusted income (dollars). 275 SOURCE: Medical Expenditure Panel Survey—Household Component, Panels 10-12.

OCR for page 267
276 MEDICAL CARE ECONOMIC RISK measure of elderly families to include 5 percent of the value of total net assets. We do not make this adjustment for the nonelderly, because they are supposed to be saving for the future rather than drawing down on their accumulated assets. Overall, the adjustment shifts average income up by about $16,000 for elderly families. In the second-to-last column, under BURD10ADJ, we show that 40 percent of elderly families and 17 percent of nonelderly families had high out-of-pocket burdens according to this measure. In the last column, under BURD20ADJ, we show that about 16.5 percent of elderly families had high burdens according to this measure compared with about 7.7 percent of nonelderly families. Among elderly families living below poverty (as classified by the origi- nal reported income), the adjustment increases average income from $6,550 to almost $12,000 while simultaneously shifting the percentage with medi- cal burdens exceeding 20 percent of family income down from 44 percent to about 29 percent. Similar shifts are seen among the low-income elderly, for whom average income increases by about $7,000 and the percentage with medical burdens exceeding 20 percent of family income shifts down from about 41 to about 27 percent. DISCUSSION The preliminary analyses presented here suggest that further work is needed to develop consistent measures of medical risk that combine the elderly and nonelderly populations. It is clear from the data presented here, however, that ignoring assets in the measurement of economic deprivation has far-reaching implications in comparing the relative status of elderly and nonelderly subpopulations. Drawing down assets or annuitizing wealth is one approach to take in measuring the resources of retired persons. It does not make sense to take this approach in measuring the resources of nonelderly families, which are supposed to be saving for future retirement. It could be argued that nonelderly family income should be reduced in order to account for such saving. Another difference worth mentioning is that working-age families face tax penalties if they use tax-sheltered assets, such as retirement ac- counts, for current health care expenses. It is beyond the scope of this paper to address the issues surrounding home equity as a potential resource for covering health care expenses and how this differs between elderly and nonelderly families. Home equity is a large asset for some families, but it may not be easily liquidated. Furthermore, it is likely that the probability of having paid off a home mortgage differs by age group. Applying different thresholds to different subpopulations by age is another approach to take in developing a consistent measure of medical risk. For example, as mentioned above, using the 75th percentile of the

OCR for page 267
INCORPORATING DATA ON ASSETS 277 distribution of out-of-pocket medical burdens from Table A-1 as a guide, one could apply thresholds of 20 percent of income for elderly families and 10 percent (rounded up from 7 percent) of income for nonelderly families to indicate high burdens. Combining these approaches, we show in Table A-3 that about 16.5 percent of elderly families have medical care burdens exceeding 20 percent of adjusted income compared with about 17.0 percent of nonelderly families with medical care burdens exceeding 10 percent of reported income. Based on the very different distribution of burdens, this method is also worth considering. Annex Comparing Self-Employed and Non-Self-Employed Families in Terms of Burdens and Distribution of Assets Another group that may have higher levels of assets relative to other groups with similar levels of income is the self-employed population. In this annex we also investigate the distribution of burdens and assets among non- elderly families in which at least one person is self-employed and compare them with nonelderly families in which no one is self-employed. The same data and methods described above are used to analyze the self-employed. Specifically, we identify self-employed families as nonelderly families with at least one person age 25 or older who reports being self-employed. Non- self-employed families are the rest of nonelderly families. Annex Tables A-1 through A-3 present the same estimates for compar- ing the self-employed with the non-self-employed, restricting the compari- son to those under age 65. The major concern regarding the self-employed is that they may have high burdens that are misleading because of high levels of assets. Unlike the elderly, however, the self-employed as defined in this analysis do not have substantially higher burdens than their non-self- employed counterparts. The median burden for the self-employed was 3.3 compared with 2.9 for the non-self-employed (Annex Table A-1). As expected, self-employed families did report higher net assets across all deciles of the distribution (Annex Table A-2). Although the self-­ mployed e as a group reported higher net assets, they also reported higher levels of average income, as shown in Annex Table A-3 in the third column of fig- ures. Unadjusted income-based measures of financial burden show that the self-employed had higher burdens when using the 10 percent threshold compared with the non-self-employed (20.1 versus 16.6). Using the 20 percent threshold, the two groups are not statistically significantly different in their level of burden (7.6 and 7.7). Although they have higher average incomes, self-employed families may have higher burdens at the 10 percent level than their non-self-employed counterparts because of higher out-of- pocket premium payments. Self-employed families are more likely to buy

OCR for page 267
278 MEDICAL CARE ECONOMIC RISK insurance in the nongroup market, in which the tax treatment of premiums differs from the tax treatment in the employment-based market. There are other differences between the employed and the self-employed populations beyond the scope of this paper (Selden, 2008). In conclusion, there does not appear to be any strong argument for incorporating the assets of self-employed families into their measure of re- sources. It is not clear that any special measurement procedures are needed to account for the health care burdens faced by this group, although a narrower definition of self-employment might reach different conclusions. ANNEX TABLE A-1 Percentiles (PCTs) of Burdens for Self-Employed, Employed Nonelderly Health Insurance Eligibility Units, Pooled Panels 10-12   N PCT50 PCT75 PCT90 Overall          Self-employed 2,069 3.32 8.11 17.15     (0.16) (0.33) (0.80)  Employed 15,444 2.86 6.81 15.69     (0.06) (0.11) (0.45) <100% Poverty          Self-employed 213 2.44 22.64 *     (1.40) (9.40)    Employed 3,047 2.74 19.00 *     (0.32) (1.39)   100-199% Poverty          Self-employed 406 3.34 12.62 27.23     (0.65) (1.60) (3.57)  Employed 3,443 2.49 8.59 19.67     (0.17) (0.34) (0.87) 200-399% Poverty          Self-employed 606 4.23 9.93 17.91     (0.48) (0.64) (1.03)  Employed 4,584 3.69 7.85 14.37     (0.12) (0.21) (0.42) 400%+ Poverty          Self-employed 844 3.03 6.55 11.01     (0.15) (0.32) (0.66)  Employed 4,370 2.51 4.80 8.01     (0.06) (0.10) (0.23) NOTE: Standard errors are in parentheses.   *Sample size is too small to make reliable estimates. SOURCE: Medical Expenditure Panel Survey—Household Component, Panels 10-12.

OCR for page 267
INCORPORATING DATA ON ASSETS 279 ANNEX TABLE A-2 Distribution of Health Insurance Eligibility Unit- Level Assets by Employment Type for Nonelderly Health Insurance Eligibility Units, Pooled Panels 10-12   Net Assets Employed Self-Employed Percentiles Nonelderly Nonelderly 10 –626 –114   (279) (49) 20 0 5,335   (0) (1,419) 30 831 29,103   (131) (4,146) 40 4,803 71,515   (308) (6,381) 50 13,785 131,849   (950) (7,310) 60 39,459 211,128   (2,278) (14,943) 70 86,891 333,701   (3,774) (19,810) 80 173,412 543,679   (6,339) (29,910) 90 357,911 985,398   (10,413) (54,411) 95 581,271 1,702,469   (18,430) (151,266) N 15,444 2,069 NOTE: Standard errors are in parentheses. SOURCE: Medical Expenditure Panel Survey—Household Component, Panels 10-12.

OCR for page 267
280 MEDICAL CARE ECONOMIC RISK ANNEX TABLE A-3 Percentage with High Burdens for Nonelderly Health Insurance Eligibility Units, Pooled Panels 10-12 (2008 dollars) N PREMOOPX TOTINC BURD10 BURD20 Overall  Self-employed 2,069 3,993 76,803 20.07 7.56 (121) (1,699) (1.05) (0.67)  Employed 15,444 2,196 50,304 16.57 7.69 (36) (680) (0.40) (0.28) <100% Poverty  Self-employed 213 1,562 9,803 33.24 27.26 (253) (633) (4.24) (4.12)  Employed 3,047 822 7,097 33.20 24.34 (65) (143) (1.23) (1.06) 100-199% Poverty  Self-employed 406 2,573 25,383 29.61 15.12 (264) (766) (2.89) (2.37)  Employed 3,443 1,304 19,882 21.50 9.59 (49) (207) (0.91) (0.67) 200-399% Poverty  Self-employed 606 3,574 48,923 24.95 7.03 (204) (1,051) (2.10) (1.23)  Employed 4,584 2,295 38,909 17.91 5.74 (55) (337) (0.74) (0.41) 400%+ Poverty  Self-employed 844 5,051 119,829 12.16 2.63 (192) (2,396) (1.18) (0.62)  Employed 4,370 3,069 91,418 6.54 1.94 (66) (1,073) (0.38) (0.23) NOTES: Standard errors are in parentheses. PREMOOPX includes out-of-pocket expenditures for care and insurance premiums. See Table A-3 Notes for definitions of column headings. SOURCE: Medical Expenditure Panel Survey—Household Component, Panels 10-12. REFERENCES Banthin, J.S., and Bernard, D.M. (2006). Changes in financial burdens for health care: Na- tional estimates for the population younger than 65 Years, 1996 to 2003. Journal of the American Medical Association 296(22):2,712-2,719. Banthin, J., Cunningham, P., and Bernard, D. (2008, January/February). Financial burden of health care. Health Affairs 27(1):185-195. Bernard, D., Banthin, J., and Encinosa, W. (2009, May/June), Wealth, income and the afford- ability of health insurance. Health Affairs 28(3):887-896. Schoen, C., Doty, M., Robertson, R., and Collins, S. (2011, September). Affordable Care Act reforms could reduce the number of underinsured U.S. adults by 70 percent. Health Af- fairs 30(9):1,762-1,771. Selden, T.M. (2008). The effect of tax subsidies on high health care expenditure burdens in the United States. International Journal of Health Care Finance and Economics 8:209-223. Selden, T.M., and Banthin, J.S. (2003). Health care expenditure burdens among elderly adults: 1987 and 1996. Medical Care 41(7 Supp.):iii-13–iii-23.