Click for next page ( 59


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 58
58

OCR for page 58
Who Goes Without Health Insurance? Who Is Most Likely to Be Uninsured? This chapter provides a portrait of the uninsured, to support the Committee's future reports about the consequences of uninsurance. Here, the Committee reviews and summarizes the published literature about what socioeconomic, de- mographic, and geographic characteristics describe the uninsured both collectively and as members of groups in the general population that are more likely than average to be uninsured.] Supporting methodological information and data tables provide the numbers behind the general statements in the text and can be found in Appendixes C and D. All estimates are for persons uninsured during calendar year 1999 (the most recent available Current Population Survey data), unless otherwise indicated. The large number and variety of Americans who are uninsured underscore the Committee's conclusion that the voluntary, employment-based approach to insurance coverage in the United States functions less like a system and more like a sieve. There are many ways to slip through the holes. People of all ages, levels of education, and in all states may be uninsured, although socioeconomic and geo- 1There is a wealth of information about the characteristics of uninsured persons, families, and populations. In addition to the public surveys and databases conducted and maintained by federal agencies such as the Bureau of the Census and the Department of Health and Human Services, surveys and studies of insurance coverage and uninsured persons are supported by the Employee Benefit Research Institute, the Commonwealth Fund, the Kaiser Commission on Medicaid and the Uninsured, the Urban Institute's Assessing the New Federalism project, and The Robert Wood Johnson Foundation through the Community Tracking Study conducted by the Center for Studying Health System Change. 59

OCR for page 58
60 CO VERA GE MA TTERS: INSURANCE AND HEALTH CARE graphic factors that affect coverage are highly correlated. One's chances of being uninsured increase if one works in an occupation or in an employment sector where employers are less likely to offer a health benefit, if one is self-employed or works for a small private-sector firm, or if one has too low an income to afford coverage. The final section of this chapter includes estimates of the relative importance of key social, economic, demographic, and geographic characteristics to one's likelihood of being uninsured, based on a new multivariate analysis of published data. Most of the studies that the Committee reviewed are based on two-way (bivariate) analyses of a characteristic and the probability that an individual with that characteristic will be uninsured. Throughout the chapter, the Committee addresses two questions together: 1. What are the characteristics of the uninsured population? 2. Who is most likely to be uninsured? The distribution of socioeconomic, demographic, and geographic differences in the general population under age 65 affects how these characteristics are distrib- uted among the uninsured population because the relative size of specific popula- tion groups affects their representation among the uninsured. For example, the uninsured rate for the urban population is the same as that of the rural population, although four out of five uninsured people live in urban areas, reflecting the predominance of urban populations nationally. HOW SOCIAL AND ECONOMIC FACTORS AFFECT COVERAGE Full-time, full-year employment offers families the best chances of acquiring and keeping health insurance, as does an annual income of at least a moderate level (greater than 200 percent of the federal poverty level ~FPL]~. Insured status correlates highly with many aspects of employment, including work status, income level, educational attainment, occupation, and employer character- istics such as firm size and employment sector. Work Status Eight out of every ten uninsured people are members of families with at least one wage earner, and six out of every ten uninsured people are wage earners themselves. Nonetheless, members of families without wage earners are much more likely to be uninsured than members of families with wage earners. Families with at least one full-time, full-year worker are more than twice as likely to have health insurance coverage, compared to families whose wage earners work part-time Hess than 35 hours per week), as contingent labor (e.g., on a

OCR for page 58
TRIO GOES liVITHOUT HEALTH INSURANCE? 6 seasonal or temporary basis, as employees of contractors, self-employed), or in which there is no wage earner (Copeland et al., 1999; HofFman and Pohl, 2000; Thorpe and Florence, 1999) (Figures 3.1 and 3.2~. The availability of health insurance in the workplace is the most important factor in determining whether wage earners and their families are insured. Yet more than half of the uninsured under age 65 years are members of families with one full-time, full-year worker. Fully 82 percent of uninsured persons are members of families with at least one wage earner (HofFman and Pohl, 2000~. As discussed in Chapter 2, the rate of employment-based coverage has de- clined since the late 1970s, with wage earners in long-term, full-time positions ("core jobs") more likely to be insured than persons recently employed or work- ing less than full-time ("peripheral jobs") (Farber and Levy, 2000~. Contingent workers are less likely than full-time, permanent workers to be offered employ- ment-based coverage and less likely to take up or enroll in an offered plan, although they may receive insurance through a spouse (Buchmueller, 1996-1997; Copeland et al., 1999~. Given the relatively small proportion of contingent work- ers in the labor market (an estimated 10 percent in 1995) and the uniform decline in coverage rates among employment sectors during the 1980s, the net decline in employment-based coverage appears to have been driven by changes other than an increasing number of contingent workers (Long and Rodgers, 1995; Copeland et al., 1999~. However, ongoing and future labor force changes may have more of an adverse impact on employment-based coverage rates. One study has predicted that greater numbers of part-time workers may cause the employment-based Estimated 42.1 Million Uninsured People = 100.0% Families with 1 or More Part-Time Workers 11.6% Families with No Workers 17.6% ...... _l ~ Families with at Least 2 Full-Time Workers ~ OR ~ 55.1% 1~ Families with 1 Full-Time Worker Families with No Workers 82.4% Families with Workers FIGURE 3.1 Distribution of uninsured population under age 65, by work status of self or family breadwinner, 1999. NOTE: Numbers may not add to 100.0 percent due to rounding. SOURCE: Hoffman and Pohl, 2000.

OCR for page 58
62 35 - ~_ ~ 30- c' ~ 25- tL ~ 20- i~ 15- 10- . _ ~ 5 - O - CO VERA GE MA TTERS: INSURANCE AND HEALTH CARE 32.5 31.9 17.9 9.1 17.5 Family with No Wage-~rning Workers 1 or More Part- llme Workers Only 1 Full-llme 2 Full-llme General Worker Workers Population Under Age65 FIGURE 3.2 Probability of being uninsured for the population under age 65, by work status of self or of primary wage earner, 1999. SOURCE: Hoffman and Pohl, 2000. coverage rate to decline by 1 to 7 percentage points by the year 2008 (Acs and Blumberg, 2001~. Most uninsured wage earners are lower income (earning less than 200 percent of FPL) or moderate income (between 200 and 400 percent of FPL) (Budetti et al., 1999; Fronstin, 2000d; O'Brien and Feder, 1998~. Members of lower-income wage-earning families are more likely to lack coverage than are members of moderate- and higher-income families. This is a function of both the reduced likelihood that lower-waged jobs offer health benefits and the relatively costly premium, compared to income, paid by lower-income families to purchase em- ployment-based coverage (Gabel et al., 1999; O'Brien and Feder, 1999~. Income and Poverty Two-thirds of all uninsured persons are members of lower-income families (earning less than 200 percent of FPL). One-third of all mem- bers of lower-income families are uninsured. There are uninsured people at all income levels, although members offamilies earning less than 200 percent of FPL are twice as likely to be uninsured as are members of the general population under age 65 (Fronstin, 2000d).2 Translating 2 In this discussion the term "family" is used to describe both a kinship and an economic relation- ship (e.g., a single adult is considered to be a one-person family). Family income levels are defined as follows:

OCR for page 58
TRIO GOES liVITHOUT HEALTH INSURANCE? Estimated 42.1 Million Uninsured People = 100.0% At Least 300% FPL 1 15.1% 200% - 299% FPL <100% FPL 36.5% 28.7% 100%- 199% FPL (for 1999, 200% of FPL for Family of 4 = $33,400) 63 Under $10 000 $50,000 and Overt _ 19 20io $40-000- $49,999 7.9% 12.6% $30,000- $39~999 16.1 % $20,000-$29,999 21 .3% $1 o,ooo $1 9,999 FIGURE 3.3 Distribution of uninsured population under age 65, by family income level, 1999. NOTE: Numbers may not add to 100.0 percent due to rounding. SOURCE: Fronstin, 2000d. percentages of the FPL into dollars allows for a more vivid comparison: almost two of ten uninsured persons are members of families that earn less than $10,000 per year (Fronstin, 2000d). A family of four must have an income greater than 400 percent of FPL (for 1999, $66,800) to have less than a one in ten chance of being uninsured (Figures 3.3 and 3.4) (Custer and Ketsche, 2000b). Higher income does not necessarily mean a lower uninsured rate. Eligibility for most public insurance (means-tested categorical programs) is restricted to spe- cific categories of low- and lower-income persons. Many members of lower- income families are not eligible for public insurance, yet they are not offered nor can they afford to buy employment-based or individual health insurance. In addition, the number and relative sizes of salaries that make up a family's income may determine whether employment-based health insurance is offered at all. A family having a single wage earner with a salary of $50,000 is more likely to have Low income: an annual income of less than 100 percent of the FPL, which is established on a yearly basis for different types of family groups that comprise a given household, for example, one adult, or one adult and two children; lower income: an annual income less than 200 percent of FPL; and moderate income: an annual income between 200 and 400 percent of FPL for a given family group. Table C. 1 (Appendix C) lists incomes at the FPL and multiples of the FPL for individuals and families of different sizes. In 1999, 200 percent of the FPL for one person was an annual income of $16,480, for a family of two, $22,120, and for a family of 3, $27,760.

OCR for page 58
64 Ct . At Ct Lo ~ ~ (D O O U) Ct ~ lo TIC O 0D a) a) a) a) lo ~ o O O U. O U. O U. O U. O U. O CO CO C\l C\l ~ ~ (lUeD)ed) aleH pe~nsu!un A_ lo lo CO CO 11 o o Lo Cal o C' . _ IL m o Cal o 5- ~o, 5- Cal _4 0 0 ,= 0 0 ~4 .0 Cal All ~ o . . . ~ O ~ U)

OCR for page 58
TRIO GOES liVITHOUT HEALTH INSURANCE? 65 access to health insurance than is a family with two wage earners, each of whom earns an annual salary of $25,000.3 Federal policies have expanded the income eligibility standards for public insurance for specific categories of lower-income persons. However, only some states have adopted these higher standards. Few states offer enrollment to children in families earning more than 200 percent of FPL or to adults in families earning at least 100 percent of FPL. For this reason, persons from families with earnings between 100 and 199 percent of FPL are almost as likely to go without coverage as are people from families whose earnings are below 100 percent of FPL (Hoffman and Schlobohm, 2000~. Expansions of the Medicaid program from the mid-1980s through the mid-1990s and the introduction of the State Children's Insurance Programs (SCHIP) helped reduce the proportion of lower-income persons with- out health insurance from an estimated 38 percent in 1987 to an estimated 32 percent in 1999. These expansions left unchanged the principle that nondisabled persons ages 18-64 may be eligible for Medicaid only when they are parents living in households with children. Educational Attainment More than one-quarter of all uninsured adults have not earned a high school diploma. Almost four of every ten adults who have not graduated from high school are uninsured. Employment-based health insurance coverage is associated increasingly with the presence of a college degree (Figures 3.5 and 3.6) (Gabel, 1999~. In addition to being positively related to income, the attainment of a college degree is associated with employment in certain sectors and types of jobs that are more likely than others to include a health insurance benefit. Also, the worker's educational level has a small effect on the take-up rate of insurance offers. The decline in employ- ment-based coverage between 1977 and 1998 almost entirely affected primary wage earners who had not graduated from college and their dependents (Gabel, 1999~. Compared with a relatively steady 80 percent employment-based coverage rate for college graduates, high school graduates experienced a 5 percentage point decline in employment-based coverage (from 68 percent to 63 percent insured) between 1977 and 1998. Primary wage earners who did not complete high school experienced an 18 percentage point decline in employment-based coverage rate (from 52 percent to 34 percent) during this same time period. Years of education serve to protect against the loss of insurance for holders of core jobs but not for contingent or recently hired workers (Farber and Levy, 2000~. This pattern is influenced by both labor force and employer characteristics- for example an employer's willingness to offer a health benefit with affordable premiums or a subsidy. Almost seven out of ten workers (69 percent) without a 3This may reflect greater subsidy for higher-waged positions (Morrisey, 1993; Blumberg, 1999).

OCR for page 58
66 CO VERA GE MA TTERS: INSURANCE AND HEALTH CARE Estimated 31.3 Million Uninsured Adults = 100.0% At Least a College Degree 12 Some College Education 24.9% High School Diploma 36.1% Less Than High School 26.8% FIGURE 3.5 Distribution of uninsured adults (ages 19-64 years), by level of educational attainment, 1999. NOTE: Numbers may not add to 100.0 percent due to rounding. SOURCE: Hoffman and Pohl, 2000. 45, 40: 35: _ - 0' 30- c,, 25- ~ 20- .- 15- 10 - 5~ O: 39.0 21.2 16.4 9.1 . 19.1 Did Not Graduate High School Some College College All Adults High School Graduate Graduate (Ages 19-64 Years) FIGURE 3.6 Probability of being uninsured for adults (ages 19-64 years), by level of educational attainment, 1999. SOURCE: Hoffman and Pohl, 2000.

OCR for page 58
TRIO GOES liVITHOUT HEALTH INSURANCE? 67 high school diploma are offered employment-based coverage. The take-up rate for this group of workers is 82 percent, only modestly below the overall worker take-up rate of 85 percent. Of those with high school diplomas or less education who are offered and decline coverage however, more than a third (36 percent) remain uninsured, twice the rate of residual uninsurance as that of more highly educated workers who decline workplace coverage. Less educated workers who decline coverage are less likely to gain coverage through a spouse than are more educated workers who decline coverage (Custer and Ketsche, 2000b). The de- cline in employment-based coverage of primary wage earners with less education may be attributed to the expense of coverage, whereas workers with higher educational attainment may have other options for coverage. Job Characteristics There are greater numbers of uninsured blue-collar workers than uninsured white-collar workers. Members of families with a primary wage earner who is blue collar are more likely to be uninsured than are members of families with a white-collar worker. In addition to the job characteristics related to work status, the occupation of a family's primary wage earner influences the likelihood that members will be uninsured (McDonnell and Fronstin, 1999~. Uninsured rates vary dramatically with regard to occupation: while almost half of all wage-earners working in private households (maids and domestic laborers) are uninsured, less than 10 percent of all wage earners in professional jobs are uninsured (McDonnell and Fronstin, 1999~. Employer Characteristics Wage-earners in smaller-sized firms, in lower-waged firms, in non- unionized firms, and in nonmanufacturing employment sectors are more likely to go without coverage. Over the past decade, the overall increase in the number of uninsured persons in working families has reflected a variety of dynamics related to employment- based coverage. There are greater numbers of uninsured workers employed by smaller firms (fewer than 25 employees), compared to larger firms, more em- ployed in predominantly lower-waged or nonunionized firms, and more em- ployed in sales ("wholesale and retail trade") compared to the profile of employers and industries that dominated the U.S. economy in the years after World War II when our present employment-based coverage arrangements became established (Gabel, 1999; Starr, 1982~. Economies of scale for employers in purchasing health benefit plans that is, lower costs per person for larger groups mean that firm size plays a key role in influencing the availability of employment-based coverage (Fronstin, 2000d;

OCR for page 58
68 CO VERA GE MA TTERS: INSURANCE AND HEALTH CARE Estimated 24.2 Million Uninsured Wage Earners = 100.0% Public-sector Employees 6.1% _ Businesses with Fewer Than 1 0 Employees Self-employed _ ~ Undo/ 20.2% Businesses with 1,000 or More Employees 3.3% Businesses with 500- 999 Employees Businesses with 100-499 Employees 1 2.3% Businesses with 10-24 Employees 10.6% Businesses with 25-99 Employees FIGURE 3.7 Distribution of uninsured wage earners (ages 18-64 years), by size of employer's firm, 1999. NOTE: Numbers may not add to 100.0 percent due to rounding. SOURCE: Fronstin, 2000d. Fronstin and Helman, 2000~.4 Of the six out of every ten uninsured persons who are wage earners, 46 percent are self-employed or work for private-sector firms with fewer than 25 employees; the uninsured rate for this subgroup is 28 percent (Figures 3.7 and 3.8) (Fronstin, 2000d). Firms that have at least 100 employees account for more than one-third of all uninsured workers, reflecting the fact that over 30 percent of the workforce is employed by larger firms. The uninsured rate for wage earners in medium- and larger-sized firms ranges between 12 and 16 percent (Fronstin, 2000d). Greater numbers of uninsured workers and dependents exist where the work- ers are employed by lower-waged, compared to higher-waged firms, and by nonunion firms, compared to union firms (Fronstin, 2000d; Gabel et al., 1999; McDonnell and Fronstin, 1999~. The chance of being uninsured is substantially greater for workers in small- to medium-sized firms (fewer than 200 employees) than for workers in larger firms (Kaiser-HRET, 2000~. When more than one- third of these smaller firms' workers are considered lower-waged (earning less than $20,000 annually), the coverage rate by employers drops to about half (35 percent) of the coverage rate of comparable firms (85 percent) where less than one-third of workers are lower-waged (Kaiser-HRET, 2000~. Since lower-waged workers are more likely to work for smaller firms, this contributes to a sizable disparity be- tween the 43 percent offer rate for lower-waged workers (defined as earning $7 or 4The administrative costs of coverage per capita decrease with the increasing size of the employer's group.

OCR for page 58
TRIO GOES liVITHOUT HEALTH INSURANCE? 89 nic groups collectively described as Asian American and Pacific Is- lander, reflecting the particular group's distinctive social, economic, and demographic characteristics and members' status as immigrants, refu- gees, or U.S.-born citizens (Brown et al, 2000a, Hoffman and Pohl, 2000~. Rates for employment-based health insurance coverage vary considerably, with lower rates for Koreans and Vietnamese (and uninsured rates correspondingly high, greater than 30 percent) and higher rates for Japanese and families with residency extending over multiple generations (Carrasquillo et al., 2000~. Gener- ally, for Asian Americans and Pacific Islanders the rates of public insurance (Med- icaid) are lower than those for other racial and ethnic groups, except for Southeast Asians, whose refugee status allows them to obtain public insurance coverage (Brown et al., 2000a). Gender More men than women are uninsured, and men are more likely than women to be uninsured. Gender disparities in insurance coverage reflect the different experiences of men and women in the workplace and with public policies. There are more uninsured men (ages 18 through 64 years) than women, although women have a lower rate of employment-based coverage (Fronstin, 2000d). More women, on average, are eligible for public insurance because of their lower average income level and the greater likelihood that they may qualify for Medicaid during preg- nancy or as the parent of infants and young children (Short, 1998~. Most adults with Medicaid are women in lower-income families, for the most part pregnant women or the mothers of young children (Wyn et al., 2001~. While fewer women than men go without coverage entirely, the greater number of women with individual insurance coverage and the higher number of women covered by public insurance are cause for concern, because such coverage tends to be un- stable, thus creating more opportunities for gaps in coverage. (Miles and Parker, 1997; Fronstin, 2000d). Both income and marital status are important influences on the likelihood that wage-earning women will be uninsured (Buchmueller, 1996-1997; Short, 1998~. Single women are more likely to be offered employment-based health insurance than are single men (an offer rate of 78 compared to 72 percent), whereas married women are somewhat less likely than married men to be offered employment-based coverage. Lower take-up rates among married women wage earners, compared to married male wage earners (63 percent versus 72 percent) are a consequence of the greater likelihood that married women are insured as depen- dents on their spouse's health insurance policy (Buchmueller, 1996-1997, based on 1993 Current Population Survey data).

OCR for page 58
So CO VERA GE MA TTERS: INSURANCE AND HEALTH CARE HOW GEOGRAPHIC DIFFERENCES AFFECT COVERAGE The decentralized labor and health services markets in the United States, and the distinct public policies in each state and locality, together create unique contexts for the patterns already described for individuals and popu- lation groups. Differences among states with respect to population characteristics, industrial economic base, eligibility for public insurance, and relative purchasing power of family incomes shape the geographic disparities in insurance coverage rates (Marsteller et al., 1998; Rowland et al., 1998; Brown et al., 2000b; Cunningham and Ginsburg, 2001~. Region and State The South and the West, the most populous regions, are home to the greatest numbers of uninsured persons (an estimated 17 million and 12 million, respectively). Residents of these regions are more likely than average to be uninsured. The pattern is similar for persons at all income levels: Southerners and West- erners are more likely to be uninsured than are those who live in the North and Midwest (Figures 3.23, 3.24, and 3.25) (Fronstin, 2000d; Mills, 2000~. Uninsured residents of California and Texas comprise more than one-quarter of the total number of uninsured persons, an estimated 12 million people (Hoffman and Pohl, 2000~. New York and Florida are the third and fourth most populous states, respectively; their uninsured residents account for almost one-fifth of uninsured persons nationally. The remaining 47 jurisdictions (including the District of Co- lumbia) are each estimated to contribute less than 4 percent of the total number of uninsured persons nationally. There is much to be learned about what influences regional variation in uninsurance rates. A multivariate analysis of 60 communities across the United States, whose uninsured rates ranged from 5 to 29 percent, found that "population characteristics, employment, and unexplained or unmeasured geographic varia- tions account for most of the differences" (Cunningham and Ginsburg, 2001~. About one-third of the variation in uninsured rates is attributable to a combination of differences in racial and ethnic group composition (18 percent) and a combina- tion of income and education (14 percent). About one-quarter of the difference is explained by employers' characteristics (21 percent) and employment rates (6 percent). Only about 13 percent of the difference among uninsured rates is ex- plained by differences in Medicaid eligibility guidelines among the states. Urban and Rural Areas Reflecting the predominantly urban concentration of the U.S. popu- lation, most uninsured persons live in urban areas. Rural and urban residents, however, are about equally likely to be uninsured.

OCR for page 58

OCR for page 58
92 ~ U) U) Go US ~ ~ ~ ~ o Z ~ ~ ~ o I Q no it\_:' In . _ 'id I l Cal Lo Cal sly o ~4 Cal o sol to, _4 o ~ o o ~ ^ I= o o 0 ~ Cal Cal o ~ o .. ~ O ~ U)

OCR for page 58
93 - ~ CD Ct o ~ Aeon (I, ~ ~ U3o ~ .O o U. a: i ,~. Ct ._ _ ma: ~ Z ~ l o o o o o o o o CD ~ ~ CO (~uea~ad) a6~aAo~ lo salad 0 0 0 N ~ Ct In 53 o In Ct Ct .2 In C' . _ 53 In z . o so so Cal o Cal sly . . Ed o O ._4 sly ~ Cal O ~ O Lo ~ Cal sly ~ O ~ O O Cal ~ o o sly 0, ~ ~ O C~ s~ o c~ ~ c~ ~ .0 c~ ~ ~ o . ~ c~ ~ c~ o ~ ~ ~ ~^ s~ 0 ~ ~ ~ - o ~ ~ u) 0 ~ ~ o ~ c~ s~ cd .. ~ ~ o ) o o o

OCR for page 58
94 CO VERA GE MA TTERS: INSURANCE AND HEALTH CARE More than four times as many uninsured persons live in urban as in rural areas, yet rural and urban residents have about the same chance of being uninsured (Figure 3.25~.9 As with other types of geographic comparisons, these general at- tributes mask underlying differences in local economies, health services infrastruc- ture, public policies, and population characteristics that distinguish urban from rural areas (Hartley et al., 1994; Ormond et al., 2001~. Although uninsured resi- dents of rural areas are fewer in number, their presence is no less a concern. Similar to the overall trend in the 1990s, the number and proportion of uninsured among rural residents has increased (Pol, 2000~. Rural and urban areas differ in the mix of sources of coverage for their residents, with a higher private coverage rate in urban (71 percent) than in rural (68 percent) areas and a higher public coverage rate in rural areas (14 percent) compared to urban (11 percent) areas (Rhoades and Chu, 2000~. The difficulties that small businesses face in purchasing affordable health insurance policies for their employees account for much of the disparity in coverage between rural and urban wage earners (Coburn et al., 1998; Mueller et al., 1998; Pol, 2000~. In addition, rural uninsured workers are more likely to be employed by lower-waged firms, to work on a contingent basis, and to work in particular employment sectors (e.g., agriculture) with lower-than-average coverage rates. Even though there are greater numbers of lower-income uninsured persons among urban than among rural residents, rural uninsured workers are even more likely than their urban counterparts to earn relatively lower wages and to be members of lower-income families. For urban areas, uninsured rates vary not only with differing population densities but also with the socioeconomic status of residents and with the presence of sizable immigrant communities (Brown et al., 2000b). The uninsured rates for the 85 largest metropolitan statistical areas (MSAs) range from 7 percent (Akron, Ohio, and Harrisburg, Pennsylvania) to 37 percent (E1 Paso, Texas) and employ- ment-based coverage rates vary between 84 percent (Milwaukee, Wisconsin) and 49 percent (E1 Paso, Texas) (Brown et al., 2000b, based on 1997 data). Compared to the national average uninsured rate, 27 of these urban areas have significantly lower rates, while 12 have significantly higher rates. 9Definitions of urban and rural are not uniform. Differences in definitions and in survey methods may give differing estimates of the numbers of uninsured persons. The current Population survey cPs) does not include a single variable to distinguish urban from rural areas. For the latter half of the 1990s, cPs data give higher uninsured rate estimates for urban com- pared with rural areas, while Medical Expenditure Panel survey data give higher uninsured rate estimates for rural areas (Pol, 2000). There are two common and distinct ways to distinguish between urban and rural areas, one devised by the Offfice of Management and Budget and the other used by the cPs (Ricketts et al., 1999). In addition, there are coding schemes that differentiate among metropolitan statistical areas (MSAs), areas adjacent to MSAs, and areas that are not adjacent to MSAs (rural areas).

OCR for page 58
TRIO GOES liVITHOUT HEALTH INSURANCE? Estimated 43.8 Million Uninsured People = 100.0% Rural 1 8.5~ Northeast Midwest (Estimated (Estimated 7.2 Million Uninsured) 7.6 Million Uninsured) Rural 8.7% An' Rural 25.1 Rural 24.7, 95 81 .5% U than South (Estimated 17.1 Million Uninsured) West (Estimated 12.0 Million Uninsured) Rural 11.3: U than 91 .3% U than 74.9% U than 75.5% U than 88.7% FIGURE 3.25 Distribution of uninsured population under age 65, by population density (rural or urban), 1998. NOTE: Numbers may not add to 100.0 percent due to rounding. SOURCE: Pol, 2000. In MSAs with higher-than-average uninsured rates, a smaller proportion of people are wage earners and greater proportions of wage earners are employed in smaller firms. In addition, rates of unionization are lower, and greater proportions of wage earners are in employment sectors with relatively high uninsured rates, such as sales. The immigrant status of residents distinguishes MSAs with high uninsured rates from those with low uninsured rates. MSAs with high rates tend to have larger immigrant communities than those with low rates.l Urban areas with high uninsured rates are home to greater proportions of people in lower-income families, and there is greater income inequality among residents (Brown et al., 2000b). Members of lower-income families are even more likely to be uninsured if they live in cities with high uninsured rates than they would be if they lived in cities with low uninsured rates (Brown et al., 2000b). Lower-income residents in urban areas with high uninsurance have a 30 percent employment-based coverage rate, compared with a 50 percent rate for those who 10These areas include Arizona (Phoenix-Mesa, Tucson), California (Los Angeles), Florida (West Palm Beach, Miami, Fort Lauderdale, Tampa), New Jersey Jersey City), New York (New York), and Texas (E1 Paso, Dallas).

OCR for page 58
96 CO VERA GE MA TTERS: INSURANCE AND HEALTH CARE live in areas with low uninsured rates, and this disparity across urban areas remains even when racial and ethnic group identity and citizenship status are taken into account. WHAT INFLUENCES AN UNINSURED RATE THE MOS" If all things were equal, how much of the difference between uninsured rates could be attributed solely to social or economic characteristics or to differences in immigrant status or race and ethnicity? If all states were home to populations with similar characteristics, how much variation among the states in uninsured rates can be attributed to regional and local differences in industrial economies and health services markets or to state policies for public programs? In this report, the discussions so far have been based on two-way compari- sons, for example, between income level and the likelihood of being uninsured. These comparisons give us a general picture of the dynamics of insurance coverage but do not allow us to evaluate or rank the relative importance of one factor independently of all others. By using more sophisticated (multivariate) statistical methods, we can look at the influence of one or more characteristics at a time on the uninsured rate and better understand their distinct influences.l1 For example, both young adults and never married single persons have higher-than-average probabilities of being uninsured. Multivariate methods allow us to isolate the effect of youth from that of having never married. In the multivariate analysis carried out by the Committee, much of the variation in uninsured rates among individuals and among population groups is associated with the following measured characteristics: income, occupation, em- ployment sector and firm size of employer, education, health status, age, gender, race and ethnicity, citizenship status, and geography. However, large and statisti- cally significant differences in uninsured rates remain after this analysis, and the variation in uninsured rates among population groups is not eliminated com- pletely. For example, if Hispanics had the same probability of being uninsured as non-Hispanic whites with similar characteristics (except for ethnicity), the unin- sured rate for Hispanics, which is about 22 percentage points higher than the uninsured rate for non-Hispanic whites, would be predicted to shrink to about a 7 percentage-point difference. The 15 percentage-point difference between the actual and predicted uninsured rates represents about a two-thirds decrease; thus, an estimated two-thirds of the difference in rates can be accounted for by differ- ences in each group's measured socioeconomic, demographic (except for ethni- 1lThe small number of published multivariate statistical analyses in the research literature address more limited sets of questions than those explored by the Committee in its analysis presented here. See Appendix D for information about analysis and data.

OCR for page 58
TRIO GOES liVITHOUT HEALTH INSURANCE? 97 city), health status, and geographic characteristics. The remaining difference be- tween the two uninsured rates reflects unmeasured differences between these two population groups. Differences in family income level account for a substantial portion of the difference in uninsured rates among groups in the general population. According to the Committee's multivariate analysis, the difference in uninsured rates be- tween low-income families Tess than 100 percent of FPL) and families with at least a moderate income (at least 200 percent of FPL) would decrease by one-third if these families resembled one another demographically, geographically, and in terms of health status.l2 The level of educational attainment of a family's primary wage earner has an even larger independent effect. More than 40 percent of the difference in unin- sured rates between families with primary wage earners who have not graduated from high school and families whose primary wage earners have post-college education would be eliminated if these families resembled one another demo- graphically, geographically, and in terms of health status.l3 Immigrant and nativity status have a pronounced influence on differences in uninsured rates among groups. Nearly 60 percent of the difference between uninsured rates for U.S.-born residents and naturalized citizens would disappear if naturalized citizens as a group shared the socioeconomic, demographic, health status, and geographic distribution characteristics of the U.S.-born population.l4 Differences between uninsured rates diminish when multivariate analysis is used to compare the population of long-term residents who are not citizens with persons born in the United States (a 26 percent decrease) and between short-term residents who are not citizens and U.S.-born residents (a 50 percent decrease). Race and ethnicity play a significant role, both independently and together with immigrant and nativity status. If non-Hispanic African Americans as a group had the same measured characteristics as non-Hispanic whites, the difference between the uninsured rates for the two groups would decrease by roughly half.l5 When immigrant status is considered in addition to race and ethnicity, the size of these differences among the groups diminishes but remains significant. Differences in state uninsured rates shrink considerably if variations in the socioeconomic, demographic, and health status characteristics within each state's population are taken into account.l6 Given the limits of any statistical model, one would not expect differences among the states' uninsured rates to disappear completely. One See Appendix D for information about analysis and data. See Appendix D. See Appendix D. See Appendix D. 16The differences among uninsured rates for states reflect differences in individuals' characteristics rather than differences among states considered as a whole. See Appendix D for data and information about methods.

OCR for page 58
98 CO VERA GE MA TTERS: INSURANCE AND HEALTH CARE state with a higher-than-average uninsured rate, California, would have a reversal from a rate 4.9 percent above the national average to rate 1.0 percent below the national average. For states such as Hawaii with a lower-than-average uninsured rate, using multivariate analysis to compare populations results in an even lower- than-average rate (a 66 percent decrease). SUMMARY Who Goes Without Health Insurance? A snapshot of the uninsured population gives us a portrait that reflects the relative size of population groups within the general population under age 65. More than 80 percent of uninsured persons are wage earners or members of working; families, and two-thirds are members of lower-income families (earning less than 2()() percent of FPL). Three-quarters of the uninsured are adults between the ages of 18 and 64, with one-half between the ages of 18 and 34 and one- quarter under the age of 18. Almost 80 percent are U.S.-born citizens, and half are non-Hispanic whites. Most are residents of the South and West, and three- ~ . quarters live in urban areas. Who Is Most Likely to Go Without Coverage? key influence on the probability include work status, family income, educational attainment, selected character~s- tics of a primary wage-earner's employer, and the age of a family's primary wage earner. Marital status and the presence of children each affect the potential oppor- . ~ , , In bivariate analyses, a highly correlated set of socioeconomic factors exerts a ~ ~ ~ that a person will be uninsured. These factors . . . . . . . . turtles tor family members to obtain coverage. (:overage disparities tor ~mm~- grants, for members of racial and ethnic minority groups, and to a lesser extent, for adult women, all reflect the importance of socioeconomic status, as well as the supporting roles played by public policies at the federal, state, and local levels. In addition, uninsured rates vary regionally and across the states. The presence of comparable uninsured rates between urban and rural areas can mask important differences in sources of coverage for rural and urban residents. In addition, a lower-income urban resident's chances of obtaining coverage decline if he or she lives in a city with a higher-than-average uninsured rate rather than in a city with a lower-than-average uninsured rate. Socioeconomic, demographic, and geographic characteristics all have signifi- cant independent effects on the likelihood that one person will be uninsured compared to another. Differences in income, occupation, employment sector and firm size, education, health status, age, gender, race and ethnicity, citizenship status and length of residency, and geography account for much of the variability among people in their likelihood of being uninsured. Disparities in coverage rates persist among population groups, and not all of these differences can be accounted

OCR for page 58
WHO GOES liVITHOUT HEALTH INSURANCE? 99 for by the commonly measured factors that most directly affect the chances of having health insurance. The next and final chapter presents the Committee's analytic plan for tracing out the consequences of uninsurance. This plan will be fulfilled in the five reports that follow this one.