4
Race, Socioeconomic Status, and Health in Late Life

James P. Smith and Raynard S. Kington

Introduction

For over a hundred years, medical and social scientists have studied differences in health status among racial groups in the United States. The resulting literature has focused on comparisons between the health of white Americans and that of African Americans, a reflection of the historical and continued prominence of the debates over the status of African Americans in this society. In response to the growth of other racial and ethnic groups, comparisons have been broadened in recent years to include a growing literature addressing the relative health status of Hispanic, Asian, and Native American populations, but the literature remains dominated by black-white comparisons.

In the last 20 years, scientific inquiry has shifted from describing gross health disparities between the races to explaining the underlying factors that account for these differences. Understanding these underlying causes requires disentangling the complex web of factors connecting the nexus between race, socioeconomic status, and health. The more recent literature that has described this nexus has typically posed the research question as, ''How much of the racial difference in health is directly accounted for by differences in socioeconomic status between populations?"

This paper has two interrelated goals. First, it examines racial and ethnic disparities in health outcomes among older Americans using two important new data sets: the Health and Retirement Survey (HRS) and the Asset and Health Dynamics Among the Oldest Old (AHEAD). Second, our research attempts to shed light on the central issue of the underlying causes of the strong relationship



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 106
--> 4 Race, Socioeconomic Status, and Health in Late Life James P. Smith and Raynard S. Kington Introduction For over a hundred years, medical and social scientists have studied differences in health status among racial groups in the United States. The resulting literature has focused on comparisons between the health of white Americans and that of African Americans, a reflection of the historical and continued prominence of the debates over the status of African Americans in this society. In response to the growth of other racial and ethnic groups, comparisons have been broadened in recent years to include a growing literature addressing the relative health status of Hispanic, Asian, and Native American populations, but the literature remains dominated by black-white comparisons. In the last 20 years, scientific inquiry has shifted from describing gross health disparities between the races to explaining the underlying factors that account for these differences. Understanding these underlying causes requires disentangling the complex web of factors connecting the nexus between race, socioeconomic status, and health. The more recent literature that has described this nexus has typically posed the research question as, ''How much of the racial difference in health is directly accounted for by differences in socioeconomic status between populations?" This paper has two interrelated goals. First, it examines racial and ethnic disparities in health outcomes among older Americans using two important new data sets: the Health and Retirement Survey (HRS) and the Asset and Health Dynamics Among the Oldest Old (AHEAD). Second, our research attempts to shed light on the central issue of the underlying causes of the strong relationship

OCR for page 106
--> between socioeconomic status and health outcomes. The rest of this paper is divided into seven main sections. The first sketches the implications of the principal economic model that has been used to analyze health outcomes. The second section presents a brief review of the existing empirical literature on the relation of racial health disparities to socioeconomic status. Using HRS and AHEAD, the third section describes racial differences in a variety of health outcomes. A brief summary of the income and wealth/health gradients obtained from these data is provided in the fourth section. Using these same data, the fifth section highlights both racial and ethnic differences in health risks. The sixth and major section of the paper summarizes a series of empirical models of self-assessed health status. In particular, these models focus on understanding the reasons underlying the strong correlation between income and health and on the implications of that correlation for racial and ethnic health disparities. The final section presents conclusions. The Theory Of Health Production And Its Relevance To Socioeconomic Status-Race-Health Research Most of the research addressing the relationship of socioeconomic status, race, and health has been grounded in a theoretical framework based in sociology. In this framework, social class or socioeconomic status is a way of ranking relative position in a society based on class, status, and power (Liberatos et al., 1988). Only relatively recently have there been significant efforts to explain the well-known differences in health across socioeconomic groups explicitly based on the economic model of health, especially to noneconomists (Selden, 1993; DaVanzo and Gertler, 1990; Dardanoni and Wagstaff, 1987; Wagstaff, 1986; Muurinen and Le Grand, 1985). Rarely have these analyses been extended to address the relationships among socioeconomic status, health, and race. The standard economic model of health is based on a few key principles, largely developed by Grossman (1972). In the economic model, health is considered to be a commodity or "good" that can be viewed as a durable capital stock that produces a flow of services over time, depreciates, and can be increased with investment. Each individual begins life with a genetic health endowment. Choices made over the lifetime, such as the use of preventive medical services or smoking, can decrease or increase the health capital stock, but there are diminishing returns to investment in health. This capital can also be affected by random events that are not under the control of individuals. There are a few important and distinct relationships that form the core of this model. First, there is the relationship between various inputs and the stock or commodity "health" (Ht). The inputs might include one's genetic or background endowment (Go), health promoting activities and other behaviors such as smoking (Bt), use of medical care (MCt), a vector of family education levels (ED), and

OCR for page 106
--> environmental factors (Et) such as the air pollution level. This relationship is described as the health production function: Health changes over the life course, and the trajectory of these changes, are the result of a number of factors. In its most simple form, health in time period t, Ht, is the result of the stock of health in the period time period t - 1, Ht-1, depreciation over the previous time period, and investments to improve health in the previous time period. This production function, which summarizes the transformation of these inputs into health outputs, is typically governed by biological considerations. Health is produced by a number of different inputs, including a wide variety of purchased medical inputs, the adoption of good personal health behaviors (exercise), and the avoidance of bad ones (smoking, excessive drinking). These inputs, such as the demand for medical care, are "derived" demands: not valued directly but valued only because of their impact on health. Because the purchase of these inputs or the adoption of these health-related behaviors is a choice individuals or families can make, they are, in the parlance of economics, "endogenous" variables. In addition to purchased inputs and health behaviors, the stock of health may enter into the health production function. To put it simply, individuals in better health may be more able to translate other inputs into more productive health investments. Therefore, today's investments are influenced by today's health status and produce tomorrow's health status. Education may enter this production function because it may affect the way individuals can transform inputs into good health. For example, more educated households may choose more qualified doctors, be more aware of the harmful health effects of behaviors such as smoking or environmental risks, or be better able to provide self-care to prevent illness or to mitigate its more harmful effects. Since some family members may be more adept at performing these functions than others, a vector of education levels of all family members is included in the production function. Family background or genetic endowments (Go), which are typically unobserved by the researcher, have played an important role in contemporary research on this topic. Rosenzweig and Schultz (1983) have argued that the existence of these unobserved background factors, which can often be traced to early childhood, may seriously bias estimates of this production function. For example, a person who has been generally sickly throughout life may require more medical care. If we do not control for this persistent unhealthiness, a regression of current health on medical services will understate the efficacy of medical care.1 1   We do not deal with the important issue of family background effects in this paper. For a recent evaluation of the importance of family background on latter life health outcomes, see Smith and Kington (1996).

OCR for page 106
--> Another fundamental insight of the household production approach is that health is a stock. The current inputs and behaviors chosen are investments that produce increments to the stock of health. If these increments are affected by current inputs and current behaviors, today's stock of health is determined by the entire history of current and past inputs and behaviors. A corollary implication is that additional current economic resources are unlikely to have a quantitatively large impact on the current stock of health, especially in the age groups that are the focus of this research. Additional economic resources may increase health care utilization or induce good health behaviors, but these sorts of behavioral changes may be slow to be adopted. Even if these behaviors were altered instantaneously, they can have a direct impact only on health investments and not on health capital. A second relationship describes the process underlying the behavioral choices that affect health. These choices are guided by a utility function (U) measured at the individual or household level. Health (Ht) is one commodity in the function, and Xt represents all other commodities that go into this utility function, U: Individuals or households maximize lifetime utility subject to a lifetime budget constraint. Thus, total expenditures across all periods on health- and non-health-related activities must not exceed total lifetime financial resources, Y; P is a vector of prices for non-health-related activities, PH is a vector of health-related prices, and HX is a vector of health-related activities: Health is desired for two different but related purposes. Health has both consumption benefits (i.e., the benefit of feeling good) and production benefits (i.e., allowing one to engage in activities that produce income). Under utility maximization, individuals will invest in health until the gain in benefits from more health equals associated costs in terms of time or money. Equation 3 highlights another central insight of this model. The budget constraint that limits household choices is a lifetime budget summarizing the discounted sum of lifetime income and current asset income. In general, households are not limited solely by their current-period resources. Financial resources in any period consist of the earnings Wt of all household members, retirement-related income Rt, government transfers Tt, and asset income At. Over the lifetime, these resources are spent on medical services and other desired commodities.2 2   There is a similar constraint in time devoted to various activities. In order to maintain focus on the essential points, this equation is not discussed in the text.

OCR for page 106
--> An important consideration is that to different degrees and in different ways, each of these income sources may be affected by the stock of health. For example, earnings (Wt) in each period are a function of an individual's human capital and a set of local labor market demand and supply conditions (dt). In this formulation, human capital is broadly defined to include health (Ht) and other forms of skills (Kt), including those formed in school and acquired during on-the-job training. Most directly, healthier people can work longer hours in any given week and more weeks during a year, which leads to higher earnings. Similarly, poor health may trigger the receipt of means-tested government transfer income, inducing a causation from health to income. Equations 3 and 4 illustrate our central point that health enters into the model in two ways, producing a two-way causation between health and income. We have already seen that people desire good health as an outcome and that higher income enables them to purchase more of it. Current health also affects a person's ability to earn in some quite fundamental ways. As we have stated, healthier people can work longer hours in any given week and more weeks during a year. This "labor supply" effect leads to higher earnings. Similarly, healthier people may have more incentive to invest in other forms of human capital and therefore command higher wages in the labor market. While good health may facilitate the receipt of some income sources (earnings), it may discourage the receipt of others (transfers). Most of the applied health literature has emphasized the first pathway from income to health, but we will present evidence in this paper that the reverse pathway from health to income cannot be ignored. Another relationship that flows from this approach is a series of derived demand functions for each input into the health production function. These input demand functions have as arguments all the input prices and the underlying determinants of the level of health demand, including household income and tastes. For example, the demand for medical care is a function of its own price (Pmc), the price of other inputs (Po), education of each family member (ED), household resources (Y) and tastes (T). As with all goods, an increase in the price of medical care will reduce the demand for it; however, as the price of other inputs change, the demand for medical care will increase or decline depending on whether these other goods are substitutes for medical care or complements to it. Education enters into these demand functions in part because it may affect the efficiency with which households can transform inputs into good health. Finally, household income acts as a

OCR for page 106
--> scaler expanding the demand for health and thereby increasing the demand for the inputs used to produce good health. Whereas in most applications, the arguments in these input demand functions are taken to be exogenous, an important exception in our case is household income, which we have already argued has an important feedback relation from health to income. The final equation in this system is the reduced-form demand function for health. The purpose of the reduced form is to solve out for the endogenous variables in the system by expressing health as a function of all the exogenous variables. Equation 6 expresses current health as a function of all input prices and total household income. While this is a frequent expression of a reduced form, there are two issues with equation 6 that raise a concern. The first issue results from the inclusion of lagged health in the function. Last period's health is determined by last period's set of prices, so that current health is more correctly a function of all past prices. Secondly, there may exist important feedback relations from health to income. This second possibility is the central focus of this paper. A reformulation of equation 6 highlights the empirical difficulties in uncovering the relation between socioeconomic status and health. Sequentially solving current health can be expressed as a function of all past prices and past incomes. This argument implies that equation 6 can be solved sequentially as where the ~ indicates a time series vector of values. Even this simple formulation highlights the extreme demands placed on data, especially cross-sectional surveys. To monitor the evolution of health outcomes over the life cycle, we would ideally like to know the entire lifetime sequence of health stocks, health behaviors, prices, and components of incomes and wealth. Although eventually the longitudinal nature of HRS and AHEAD will be an important step in that direction for an older population, such data do not currently exist. Consistent with the limitations imposed by current data, our aim here is a more modest but important step in the direction of understanding the reasons for the demographic and economic correlates of health outcomes at older ages. This step rests on the distinction between contemporaneous (current period) feedbacks from health to economic status and health behaviors and the full lifetime sequence of such feedback relationships. The full lifetime sequence of interactions between health and socioeconomic status is beyond the scope of our inquiry in this paper, and we concentrate instead on informing the nature of the possible contemporaneous feedbacks.

OCR for page 106
--> The Socioeconomic Status-Health Gradient In the theoretical framework outlined above, socioeconomic status may affect health and health-related behaviors in many ways. At the most basic level, income and wealth determine the budget constraint: those who are poor have fewer resources to devote to health. As a result, they may purchase fewer medical services or be less able to afford medical insurance. From this view, health is no different than any other commodity such as housing, food, or entertainment—the more well-to-do consume more. If this were all there is, knowing the socioeconomic status-health gradient would require only the estimation of the wealth elasticity of health. In the computation of that elasticity, economic status should be defined as broadly as possible to include all income sources of all household members and wealth. That is not all there is, largely owing to the real possibility of reverse causality or simultaneity bias. A large amount of the literature that has addressed the relationship between health and socioeconomic status has been based on cross-sectional data. Thus, these data have not allowed a simple but important question to be addressed: To what extent does low socioeconomic status lead to poor health rather than poor health's leading to low socioeconomic status? The ambiguity of the association between contemporaneous health and socioeconomic status is most obvious when income is the measure of socioeconomic status, and a classic example of the analytic issue illustrates the problem. There are two plausible explanations for the relationship between contemporaneously measured low income and poor health. First, low income may lead to poor health by, for example, limiting the use of preventive health care services as predicted in the economic model of health production. An alternate plausible explanation, however, is that poor health may lead to lower socioeconomic status by, for example, limiting an individual's ability to work or the wage he or she can earn.3 The statistical conditions for identification of the causal pathways are relatively easy to state and quite difficult to implement. For example, to statistically identify the pathway from health to income requires having exogenous variables that affect income only through their effect on health (that is, these variables have no direct effect on income). In this case, the health-income correlation induced by this variable only reflects a causal pathway from health to income. To use one illustration, a lower price of health care can directly affect health status through increased utilization of health care. Because health status is altered, there may also be subsequent alterations in household income. However, this lower price of health care should not have any direct impact on household income (outside of its 3   The analogous statistical conceptualization of this problem is called simultaneity bias. Namely, estimation of relationships using standard statistical techniques may be biased if explanatory variables can be a consequence as well as a cause of the dependent variable such as health (Garber, 1989).

OCR for page 106
--> influence on health care). In this case, variation in the price of health care can be used to identify the causal pathway from health to income.4 Unfortunately, our current data sources do not contain the type of statistical variation that would allow us to formally identify the causal pathways. Consistent with the limitations imposed by current data, our aim here involves a more modest but important step toward understanding the health-socioeconomic status nexus. This step rests first on a distinction between contemporaneous (current period) feedbacks from health to economic status and health behaviors and the full lifetime sequence of such feedback relationships. While our cross-sectional data imply that unraveling the full lifetime interrelated sequences is beyond our scope, our rich array of economic, demographic, health behavior and outcome data allows us to make progress on the contemporaneous relationships. Our research strategy begins with a separation of household income into its important components. We argue a priori that some of these income components largely reflect causation from health to income. After these contaminated components are separated out, it is more likely that the other income components will reflect a pathway from income to health. At a minimum, this empirical strategy can serve as an important diagnostic device about the relative importance of the two pathways that connect income and health. For example, both HRS and AHEAD allow us to separate income into its distinct components. Some of these income components are strongly affected by contemporaneous feedbacks from health to economic status. Past the retirement age, other income components are largely free of these feedbacks so that, at a minimum, we are able to mitigate the contemporaneous feedbacks from health to income. There are a number of other possible sources of bias that complicate the estimation of the effect of socioeconomic status on health. For example, financial status may also determine where one lives, which may be related to a range of exogenous factors from the quality of health providers to exposure to air pollution and toxic waste to public expenditures on prevention of communicable diseases. Although environmental factors are often considered to be exogenous, in fact residence may also be a choice partly determined by such factors as regional health risks (Preston and Taubman, 1994). Financial status may also affect one's choices for such activities as smoking or exercise by determining opportunity sets for the trade-offs between alternative utility-increasing or utility-decreasing activities and the associated increases or decreases in health risk (Muurinen and Le Grand, 1985). For example, individuals clearly derive some benefit from smoking. A person with limited alternative resources to satisfy such needs may be more willing to accept the health risks associated with smoking. Uncertainty may also play a role in explaining differences in investments in health across socioeconomic groups. Early models of 4   A completely symmetrical argument exists for the identification of the pathway from income to health.

OCR for page 106
--> investment suggested that the poor may invest less in prevention, because greater risk aversion among the poor may push them away from relatively riskier investments in health capital (Dardanoni and Wagstaff, 1987). However, more recent extensions of the models suggest that there may be a countervailing incentive for the poor to invest in health because they are less able to afford losses in income because of ill health (Selden, 1993). Measurement of Socioeconomic Status In studies of the impact of socioeconomic status on health, education, occupation, and financial resources have typically been used as proxies. How each is defined may affect analyses of the race-socioeconomic status-health nexus in late life. If these variables are imprecisely measured, incorrect conclusions may be drawn about the relationships among the variables (Garber, 1989). In this section, we briefly discuss some of the major issues that arise with each proxy. Education Education is an important explanatory variable in both economic and sociological-based empirical models of the socioeconomic status-health relationship. As is demonstrated in the theoretical model outlined above, education may affect health status through a number of channels. First, schooling is an important determinant of economic status. Individuals with more schooling in general have significantly higher lifetime wealth than those with less schooling. In addition, schooling may alter the efficiency of health production—that is, the efficiency of the process by which the various inputs are transformed into health. For example, better educated individuals may have more information at their disposal about the effect of nutrition on health and may thus make healthier choices in eating habits. An often-cited advantage of education as a proxy for socioeconomic status is that decisions about education are usually completed by early adulthood. This temporal ordering is taken to imply that schooling is free of reverse causality (i.e., not a result of poor health). While there is some truth to this argument, it applies only for health conditions that are unanticipated at the time schooling decisions are made. If poor health conditions are known, they will generally influence investments in schooling since future work effort will be lower owing to poor health. Socioeconomic status may also be related to so-called third factors affecting health investments (e.g., preventive activities). For example, education may be related to one's willingness to invest in health now by giving up something else in order to have an improvement in health in the future (Fuchs, 1982). The imprecise measurement of education may be especially relevant when race is added to the relationship.5 In the United States, simple counts of years of

OCR for page 106
--> educational attainment present problems because historically there have been large differences in the content and quality of education between races. These differences are probably greatest among the current generation of elderly, many of whom were born into a rigidly segregated society with large racial differences in public investment in education (Smith, 1984). Financial Resources The measurement of financial resources presents even more challenges. Most studies in the health literature have measured financial resources with some form of individual or household/family income in the year of or the year before a cross-sectional survey, sometimes called contemporaneous income. There are several potential problems with the use of contemporaneous income as a measure of financial resources. First, income in a single year may not adequately measure the financial resources available to an individual over the lifetime in which decisions affecting health are made. This timing issue is different from the reverse causation question and may be even more important in late life. It may be especially important in assessing the comparative health status of currently older blacks versus whites because of the large changes in the relative income of blacks versus whites that have occurred over their lifetimes. Second, income may not be the best measure of economic resources among older individuals, especially those who are retired. Instead, wealth may be a far better proxy for their command over economic resources. Income is typically lower after retirement than before. In the extreme, an older person may be worth a million dollars and simply live off this principal with no income. Wealth captures an important dimension of financial resources because it may be a indicator of long-run income. The distinction between income and wealth may be especially critical for understanding racial health disparities because racial differences in wealth are even greater than in income.6 Some studies have simplified the measurement of financial resources by translating contemporaneous income into a dichotomous variable indicating whether a household's income is above or below the federally defined poverty line. This is in general a mistake because it substitutes a political concept for a scientific construct. One potential problem with this practice is that income effects are known to extend across the spectrum of income even into high-income 5   Education is generally measured in years of education attained. The effects of education on such outcomes as income are typically not linear. Analogously, flexibility in the form of the effect of education on health should be permitted. For example, attainment could be expressed in categories such as high school graduate, college graduate, and so forth. 6   Finally, regional differences in costs of living (Liberatos et al., 1988) also have an impact on the significance of income. To the extent to which there are racial differences in geographic distribution, failure to control for geographic location may bias estimates of the racial differences in the relationships among race, socioeconomic status, and health.

OCR for page 106
--> ranges. Dichotomizing income may lead to incorrect conclusions about the relationships among race, health, and income. For example, blacks are much more likely to have income near the poverty level. Thus, a multivariate analysis that simply includes a poverty dummy may result in a still significant race dummy effect even in the absence of a real racial difference because the range of upper level income and the corresponding variation in health in the wealthier white population are not accounted for. Occupation A final measure of socioeconomic status that has been commonly used, especially by sociologists, is occupation.7 One reason for this is that occupation is arguably a better measure of long-run economic status than current income is. Whatever the traditional merit of this argument, it is certainly mitigated when data sets contain measures of wealth and long-run time series on income. There are some other problems that may affect the interpretation of occupation. In comparisons across racial and ethnic groups with different status rankings of occupations, a particular occupation may translate into a different status in a community depending on the racial composition of the community. Second, broad groupings of occupation may not capture significant variation within occupational categories. Third, there is controversy over how one measures occupation in late life. For example, for many important occupational health exposures (e.g., the relationship between asbestos and lung cancers), the temporal relationship between the exposure and the outcome is distant. Thus, for studies of racial differences in the relationship between health and occupation in late life, it is especially unclear which occupation is most appropriate. Occupational categories may also present added problems in the context of the socioeconomic status-race-health analysis because there may be differences between races in exposures and treatment of persons in the same occupational category. The Race Connection The Role of Race Typically, race per se has not been explicitly analyzed in the context of an economic model of health production. The reasons for the failure to explicitly incorporate race into this framework remain unclear. In one standard textbook of 7   Although occupation has historically been an important measure of socioeconomic status for sociological studies, it has been relatively less important in the economic literature on health. Occupation is related to a number of important factors in health in late life. First, occupation is related to occupational injuries and exposures that have an impact on health. Second, occupation is related to both education and income.

OCR for page 106
--> TABLE 4-16 Ordered Probit Analysis of Self-Assessed Health Status Change, HRS and AHEAD Samples   HRS Sample     AHEAD Sample     Co-variate Parameter z Parameter z Parameter z Parameter z Black -.1500 4.42 -.0008 00.02 -.1456 3.21 .0114 0.23 Hispanic -.1686 3.90 -.0060 0.71 -.1626 3.94 .0424 0.61 Female .0436 2.04 .0566 2.29 .0172 0.62 .0431 1.31 Marital status Never married     -.0182 0.25     .1091 1.28 Separated     -.1672 2.12     .0413 0.59 Divorced     -.0884 1.94         Widowed     .0026 0.05     .0457 1.17 Individual education 12-15 years     .1347 4.71     .1533 4.73 16 or more years     .1719 4.25     .1899 3.69 Advanced degree     -.1059 0.67     .0529 0.39 Spousal education 12-15 years     .0912 3.01     .0485 1.15 16 or more years     .0730 1.72     .0252 0.39 Advanced degree     -.2611 1.39     .0471 0.28 Income and wealtha Total income     .00160 5.07     .0007 1.88 Total wealth     .00005 1.23     .0002 2.48 Cohort 1935-1937b (1919-1923)c     .0631 2.09     .3164 6.78 1938+ (1914-1918)     .0839 3.24     .2481 5.24 (1909-1923)             .1759 3.62

OCR for page 106
--> Risk factors Smoking Ever smoked     -.0718 2.26         Current smoker     -.0427 1.01     -.1202 1.66 Cigarettes smoked per day     -.0031 1.92     -.0018 0.48 Drinking < 1 drink per day     .0668 2.69     .1558 4.97 1 or 2     .1134 2.85     .0539 1.02 3 or 4     .0869 1.44     .0912 0.86 5 or more     -.1540 1.73     -.2081 0.83 Light exercise 3 times per week     .5541 13.77         1-2 times per week     .4554 10.35         1-3 times per month     .3989 7.72         less than once a month     .2613 4.90         Vigorous exercise 3 times per week     .3721 10.07         1-2 times per week     .2138 5.37         1-3 times per month     .2276 5.33         less than once a month     .1788 5.82         Occupational hazard Ever exposed     -.0266 0.57         Years exposed     -.0047 3.21        

OCR for page 106
-->   HRS Sample     AHEAD Sample     Co-variate Parameter z Parameter z Parameter z Parameter z Body mass index (BM1) BM1     .0221 2.01     .0753 4.51 BM12     -.0005 3.17     -.0015 4.92 Prior year's health Excellent -1.387 28.2 -1.876 35.2 -1.597 24.5 -1.792 26.4 Very good -1.138 24.3 -1.525 30.7 -1.456 25.0 -1.636 27.2 Good -1.089 23.5 -1.347 28.0 -1.447 25.7 -1.565 27.2 Fair -1.084 22.2 -1.201 24.1 -1.423 24.7 -1.485 25.5 Cut point 1 -2.509   -2.561   -2.202   -1.046   Cut point 2 -1.686   -1.735   -.153   1.048   Cut point 3 -0.755   .710           Cut point 1.409   1.61           a Expressed in thousand dollar units. b Cohort category for HRS sample. c Cohort category for AHEAD sample.

OCR for page 106
--> TABLE 4-17 Ordered Probit Analysis of Health Change with Alternative Measures of Incomea Co-variate Parameter z Parameter z Parameter z HRS Sample (Ages 51-61) Individual Earnings .0018 4.06 .0008 1.58     Weekly wages         .0084 0.86 Retirement income -.0011 0.68 .0085 3.84 .0079 3.56 Welfare income -.0187 2.49 -.0009 0.09 .0029 0.27 Earnings = 0     -.1528 5.35     Weekly wages = 0         -.1584 3.94 Retirement income = 0 .1723 4.23 .1308 3.20     Welfare income = 0     .1048 1.88 .0620 1.11 Spouse Earnings .0021 4.16 .0008 1.40     Weekly wages         -.0134 1.34 Retirement income -.0007 0.36 .0062 2.63 .0089 3.77 Welfare income -.0271 3.27 -.0022 1.94 -.0018 1.57 Earnings = 0     -.1675 5.36     Weekly wages = 0         .1190 2.89 Retirement income = 0 .0704 1.60 .0866 1.96     Welfare income = 0     .0224 0.35 -.0021 0.03 Asset income -.0001 0.19 .0012 1.93 .0007 1.07 Asset income = 0     .0502 2.12 .0404 1.70 Wealth .0001 1.63 .0001 2.42 .0001 2.86 Black -.0027 0.07 -.0127 -0.35 .0111 0.30 Hispanic -.0045 0.10 -.0061 -0.13 .0043 0.09 Female .0643 2.36 .0679 2.47 .0917 3.42 AHEAD Sample (Ages 70 and Over) Individual Earnings .0049 3.01 .0001 0.04     Social Security .0042 0.89 .0065 1.21     Other retirement income .0020 0.91 .0014 0.55     Earnings = 0     -.2675 5.37     Social Security = 0     .1166 1.54     Other retirement income = 0     -.0280 0.81     Spouse Earnings -.0001 0.06 -.0009 0.42     Social Security .0042 0.72 .0096 1.43     Retirement income .0043 1.73 .0028 1.00     Earnings = 0     -.0236 0.39     Social Security = 0     .1917 2.09     Retirement income = 0     -.0491 1.07    

OCR for page 106
--> Co-variate Parameter z Parameter z Parameter z Asset income .0003 0.75 .0003 0.69     Asset income = 0     -.0184 0.58     Welfare income     -.0194 0.90     Welfare income = 0     .2329 3.17     Wealth .0001 1.81 .0001 1.63     Black .0346 0.71 .0457 0.92     Hispanic .1165 1.61 .1376 1.89     Female .0476 1.22 .0487 1.22     a All continuous income and wealth coefficients expressed in thousands of dollars. dates current health transitions. Fortunately, we are able to do this in the part of the HRS sample that is linked to Social Security records. These records contain the complete history of a respondent's Social Security earnings.28 Two proxies for Social Security wealth are used in this research. The first measures the sum of the household's Social Security earnings up to age 50, the starting age of the HRS sample. The second construct sums household Social Security earnings up to age 40. The advantage of the second measure is that it predates HRS health measurement by at least 10 years, effectively eliminating short-run reverse causality from health to economic status. The empirical estimates summarized in Table 4-18 do suggest that long-term wealth as measured by Social Security earnings affects health trajectories of mature men and women. This effect is present whether we sum past earnings to age 50 or age 40. While both spouses' Social Security earnings have positive effects, the impact of one's own Social Security is larger. While these estimates are less subject to short-run reverse causality problems, they are not immune from long-run problems. If individuals work and earn less as a result of poorer health in the past, causation may still flow from health to economic status. Conclusions This paper has critically explored the role that socioeconomic status plays in explaining racial and ethnic differences in health outcomes of Americans during their middle and old age. Although our results are consistent with other research suggesting an important role for socioeconomic status as a factor accounting for 28   The use of these data creates a number of issues. First, in only two-thirds of all households did the respondent and spouse agree to the Social Security linkage. All estimates in this section are confined to that subsample. Second, Social Security earnings are capped at the limit.

OCR for page 106
--> TABLE 4-18 Ordered Probit Analysis of Health Change with Alternative Measures of Social Security Earnings Histories, HRS Sample (ages 51-61) Characteristic Parameter z Parameter z Parameter z Social Security Earnings to Age 50 Total incomea .0011 2.18 b   b   Wealtha .0001 1.27 .0001 2.11 .0001 2.12 Social Security Earningsa Household .0002 3.95 .0001 2.93     Respondent         .0002 2.98 Spousal         .0001 1.15 Black -.0456 0.92 -.0680 -1.30 -0.673 1.34 Hispanic 0.465 0.68 .0403 0.58 .0398 0.57 Female .0339 1.00 .0324 0.64 .0111 0.26 Social Security Earnings to Age 40 Total income .0012 2.36 b b     Wealth .0001 1.27 .0001 2.14 .0001 2.15 Social Security Earnings Household .0004 4.30 .0003 3.41     Respondent         .0003 3.25 Spousal         .0002 1.69 Black -.0413 0.83 -.0639 -1.30 -.0635 1.26 Hispanic .0533 0.76 .0476 0.68 .0471 0.67 Female .0391 0.76 .0377 1.00 .0207 0.63 a All continuous income and asset coefficients expressed in thousands of dollars. b For last four columns, the model includes all income components and non-receipts of income components. racial and ethnic differences, our results indicate that the relationship among race and ethnicity, socioeconomic status, and health is far more complex than many current analyses recognize. We focus attention on the complexity involved in accounting for economic status as an underlying factor in health status. First, there are two important dimensions of economic status—income and wealth—each with distinct conceptual and empirical associations with health. Second, the association of some common measures of socioeconomic status with health status is highly nonlinear. For example, the association of both income and wealth with self-reported general health status is strongest among the poorest households and is relatively weak among the most affluent members of our society.

OCR for page 106
--> Both of these issues may affect how we account for racial and ethnic differences in health in later life. Finally, there is compelling evidence that the feedbacks from health to current socioeconomic status are quantitatively strong and should not be ignored in empirical investigations. In particular, the entire association between current household income and health among households with a member in his or her fifties appears to reflect causation from health to income rather than from income to health. As new longitudinal data sets with more detailed and varied measures of economic status and health status become available, future research should progress toward a more complete understanding of the pathways linking race and ethnicity, socioeconomic status, and health across the lifespan. Acknowledgment This research was supported by Grant 5PO1-AG08291 awarded by the National Institute on Aging, U.S. Department of Health and Human Services, and by Grant 5P50-HD 12639 awarded by the National Institute of Child Health and Human Development. References Berman, J.R., R. Sickles, P. Taubman, and A. Yazbeck 1991 Black-white mortality inequalities. Journal of Econometrics 50:183-203. Blesch, K., and S. Furner 1993 Health of Older Black Americans. Vital and Health Statistics S2:229-273. Broderick, J.P., T. Brott, T. Tomsick, G. Huster, and R. Miller 1992 The risk of subarachnoid and intracerebral hemorrhages in blacks as compared with whites. New England Journal of Medicine 326:733-736. Caplan, L.S., B.L. Wells, and S. Haynes 1992 Breast cancer screening among older racial/ethnic minorities and whites: Barriers to early detection. Journal of Gerontology 47:101-110. Clark, D.O., and G.L. Maddox 1992 Racial and social correlates of age-related changes in functioning. Journal of Gerontology 47:S222-232. Clark, D.O., G.L. Maddox, and K. Steinhauser 1993 Race, aging, and functional health. Journal of Aging and Health 5:536-553. Cooper, R. 1986 The biological concept of race and its application to public health and epidemiology. Journal of Health Politics, Policy Law 11(1):97-116. Dardanoni, V., and A. Wagstaff 1987 Uncertainty, inequalities in health and the demand for health. Journal of Health Economics 6:283-390. DaVanzo, J., and P. Gertler 1990 Household production of health: A microeconomic perspective on health transitions. An unpublished paper presented at workshop on the Measurement of Health Transition Concepts in London, June 7-9, 1989; N-3014-RC. RAND, Santa Monica, CA. Dowd, J.J., and V.L. Bengtson 1978 Aging in minority populations: An examination of the double jeopardy hypothesis. Journal of Gerontology 33:427-436.

OCR for page 106
--> Eley, J.W., H.A. Hill, V.W. Chen, D.F. Austin, M.N., Wesley, H.B. Muss, R.S. Greenberg, R.J. Coates, P. Correa, C.K., Redmond, C.P. Hunter, A.A. Herman, R. Kurman, R. Blacklow, S. Shapiro, and B.K. Edwards 1994 Racial differences in survival from breast cancer: Results of the National Cancer Institute black/white cancer survival study. Journal of the American Medical Association 272:947-954. Ewbank, D.C. 1987 History of black mortality and health before 1940. Milbank Quarterly 65:100-128. Feldstein, P.J. 1983 Health Care Economics, 2nd. ed. New York: Wiley Medical Publication. Ferraro, K.F. 1987 Double jeopardy to health for black older adults. Journal of Gerontology 42:528-533. Fuchs, V.R. 1982 Time preference and health: An exploratory study. In Economic Aspects of Health, V.R. Fuchs, ed. Chicago: University of Chicago Press. Furner, S.E. 1993 Special topics. Vital and Health Statistics (29):37-55. Garber, A. 1989 Pursuing the links between socioeconomic factors and health: Critique, policy implications, and directions for future research. In Pathways to Health—The Role of Social Factors, J.P. Bunker, D.S. Gomby, and B.H. Kehrer, eds. Menlo Park, CA: Henry J. Kaiser Family Foundation. Gibson, R.C. 1991 Age-by-race differences in the health and functioning of elderly persons. Journal of Aging Health 3:335-351. Grossman, M. 1972 The Demand for Health: A Theoretical and Empirical Investigation. New York: National Bureau of Economic Research. Guralnik, J.M., and G.A. Kaplan 1989 Predictors of health aging: Prospective evidence from the Alameda County Study. American Journal of Public Health 79:703-708. Guralnik, J.M., A.Z. LaCroix, R.D. Abbott, L.F. Berkman, A. Satterfield, D.A. Evans, and R. Wallace 1993 Maintaining mobility in late life: Demographic characteristics and chronic conditions. American Journal of Epidemiology 137:845-857. Guralnik, J.M., K.C. Land, D. Blazer, G.G. Fillenbaum, and L.G. Branch 1993 Educational status and active life expectancy among older blacks and whites. New England Journal of Medicine 329(2):110-116. Haan, M.N., and G.A. Kaplan 1985 The contribution of socioeconomic position to minority health. Report to the NIH Task Force on Black and Minority Health. Bethesda, MD: National Institutes of Health. Hahn, R.A. 1992 The state of federal health statistics on racial and ethnic groups. Journal of the American Medical Association 267(2):268-271. House, J.S., R.C. Kessler, A.R. Herzog, R.P. Meto, A.M. Kinney, and M.J. Breslow 1990 Age, socioeconomic status, and health, Milbank Quarterly 68:383-411. James, S.A., D.S. Strogatz, S.B. Wing, and D.L. Ramsey 1987 Socioeconomic status, John Henryism, and hypertension in blacks and whites. American Journal of Epidemiology 126:664-673. Jones, C.P., T.A. LaVeist, and M. Lillie-Blanton 1991 "Race" in the epidemiologic literature: An examination of the American Journal of Epidemiology. American Journal of Epidemiology 134(10):1079-1084.

OCR for page 106
--> Kaplan, G.A., T.E. Seeman, R.D. Cohen, L.P. Knudsen, and J. Guralnik 1987 Mortality among the elderly in the Alameda County study: Behavioral and demographic risk factors. American Journal of Public Health 77:307-312. Kaplan, N.M. 1994 Ethnic aspects of hypertension. Lancet 344:450-452. Keil, J.E., S.E. Sutherland, R.G. Knapp, and H.A. Tyroler 1992 Does equal socioeconomic status in black and white men mean equal risk of mortality? American Journal of Public Health 82:1133-1136. Kellie, S.E., and J.A. Brody 1990 Sex-specific and race-specific hip fracture rates. American Journal of Public Health 80(3):326-328. Kington, R., and J.P. Smith In Socioeconomic status and racial and ethnic differences press in functional status associpress ated with chronic diseases. American Journal of Public Health. Krieger, N. 1990 Social class and the black/white crossover in the age-specific incidence of breast cancer: A study linking census-derived data to population-based registry records. American Journal of Epidemiology 131:804-814. Krieger, N., D.L. Rowley, A.A. Herman, B. Avery, and M.T. Phillips 1993 Racism, sexism, and social class: Implications for studies of health, disease, and well-being. American Journal of Preventive Medicine 9:82-122. Kuczmarski, R.J., K.M. Flegal, S.M. Campbell, and C.L. Johnson 1994 Increasing prevalence of overweight among U.S. adults. Journal of the American Medical Association 272:205-211. LaViest, T.A. 1994 Beyond dummy variables and sample selection: What health services researchers ought to know about race as a variable. Health Services Research 29:1-16. Lewontin, R.C. 1972 The apportionment of human diversity. Evolutionary Biology 6:381-398. Liberatos, P., B.G. Link, and J.L. Kelsey 1988 The measurement of social class in epidemiology. Epidemiology Reviews 10:87-121. Macken, C. L. 1986 A profile of functionally impaired elderly persons living in the community. Health Care Financing Review 7:33-49. Manton, K.G., C.H. Patrick, and K.W. Johnson 1987 Health differentials between blacks and whites: Recent trends in mortality and morbidity. Milbank Quarterly 65:129-199. Massey, D., and N. Denton 1989 Hypersegregation in U.S. metropolitan areas: Black and Hispanic segregation along five dimensions. Demography 26(3):373-441. Menchik, P.L. 1993 Economic status as a determinant of mortality among black and white and older men: Does poverty kill? Population Studies 47:427-436. Mermelstein, R., B. Miller, T. Prohaska, V. Benson, and J.F. Van Nostrand 1993 Measures of health. Vital Health Statistics 3(27):9-21. Mor, V., J. Murphy, and S. Masterson-Allen 1989 Risk of functional decline among well-elders. Journal of Clinical Epidemiology 42:895-904. Mutchler, J.E., and J.A. Burr 1991 Racial differences in health and care service utilization inlater life: The effect of socioeconomic status . Journal of Health and Social Behavior 32:342-356.

OCR for page 106
--> Muurinen, J.M., and J. Le Grand 1985 The economic analysis of inequalities in health. Social Science and Medicine 20:1029-1035. Osborne, N.G., and M.D. Feit 1992 The use of race in medical research. Journal of the American Medical Association 267(2):275-279. Otten, M.W., S.M. Teutsch, D.F. Williamson, and J.S. Marks 1990 The effect of known risk factors on the excess mortality of black adults in the United States. Journal of the American Medical Association 263:845-850. Preston, S.E., and P. Taubman. 1994 Socioeconomic differences in adult mortality and health status. In Demography of Aging, L.G. Martin and S.H. Preston, eds. Washington, DC: National Academy Press. Rogers, R.G. 1992 Living and dying in the U.S.A: Sociodemographic determinants of death among blacks and whites. Demography 29:287-303. Rosenzweig, M.R., and T.P. Schultz 1983 Estimating a household production function: Heterogeneity, the demand for health inputs, and their effects on birth weight. Journal of Political Economy 91(5):723-746. Satariano, W.A. 1986 Race, socioeconomic status and health: A study of age differences in a depressed area. American Journal of Preventive Medicine 2:1-5. Schoenberg, B.S., D.W. Anderson, and A.R. Haerer 1986 Racial differentials in the prevalence of stroke. Archives of Neurology 43:565-568. Selden, T.M. 1993 Uncertainty and health care spending by the poor: The health capital model revisited. Journal of Health Economics 12:109-115. Serjeant, G.R. 1985 Sickle Cell Disease. New York: Oxford University Press. Smith, J.P. 1984 Race and human capital. American Economic Review 74(4):685-698. 1994 Hispanics and the American dream. Unpublished manuscript. Santa Monica, CA: RAND. 1995 Racial and ethnic differences in wealth using the Health and Retirement Study. Journal of Human Resources 30:S158-S183. In press Wealth inequality among older Americans. Journal of Gerontology. Smith, J.P., and R. Kington 1996 Demographic and Economic Correlates of Health in Old Age. Santa Monica, CA: RAND, DRU-1316-NIA. Sobal, J., and A.J. Stunkard 1989 Socioeconomic status and obesity: A review of the literature. Psychology Bulletin 105:260-275. Sommer, A., J.M. Tielsch, J. Katz, H.A. Quigley, J.D. Gottsch, J.C. Javitt, J.F. Martone, R.M. Royall, K.A. Witt, and S. Ezrine 1991 Racial differences in the cause-specific prevalence of blindness in East Baltimore. New England Journal of Medicine 325:1412-1417. Sorlie, P., E. Rogot, R. Anderson, N.J. Johnson, and E. Backlund 1992 Black-white mortality differences by family income. Lancet 340:346-350. Strawbridge, W.J., G.A. Kaplan, T. Camacho, and R.D. Cohen 1992 The dynamics of disability and functional change in an elderly cohort: Results from the Alameda County study. Journal of the American Geriatric Society 40:799-806.

OCR for page 106
--> U.S. Department of Health and Human Services 1990 Healthy People 2000: National Health Promotion and Disease Prevention Objectives. Washington, DC: U.S. Department of Health and Human Services. Wagstaff, A. 1986 The demand for health: Theory and applications. Journal of Epidemiology and Community Health 40:1-11. Waitzman, N.J., and K.R. Smith 1994 The effects of occupational class transitions on hypertension: Racial disparities among working-age men. American Journal of Public Health 84:945-950. Williams, D.R. 1994 The concept of race in Health Services Research: 1966 to 1990. Health Services Research 29(3):261-274. Williams, D.R., R. Lavizzo-Mourey, and R.C. Warren 1984 The concept of race and health and status in America. Paper from the CDC-ATSDR Workshop on the use of Race and Ethnicity in Public Health Surveillance. Public Health Reports 109(1):26-41. Zick, C.D., and K.R. Smith 1991 Marital transitions, poverty, and gender differences in mortality. Journal of Marriage and the Family 53:327-336.