| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
| ||||||||||||||||||||||||||||||||
RESEARCH ARTICLE |
1 Department of Consumer Science, Ohio State University, Columbus.
2 Department of Family Studies, University of Kentucky, Lexington.
Address correspondence to Dr. Jinkook Lee, Department of Consumer Science, Ohio State University, 1787 Neil Avenue, Columbus, OH 43210. E-mail: lee.42{at}osu.edu
 |
Abstract |
|---|
 TOP Abstract ModelMethodsResultsDiscussionAppendixReferences |
|---|
Methods. A model deriving from life-cycle theory is proposed and estimated using Waves 1 and 2 of the Asset and Health Dynamics of the Oldest Old (AHEAD) data set.
Results. Both new health events and existing health conditions significantly influence wealth depletion of elders, but their impacts differ across marital status. Whereas an occurrence of new health events brought wealth depletion of elders in married households, having existing chronic health conditions was associated with wealth depletion of elders in single households.
Discussion. Poor health, both a new health event and existing chronic conditions, leads to considerable wealth depletion in elderly individuals. Considering the significant impacts of health on wealth, the public needs to be better informed of potential health events in later life and the associated financial burden. Additional health insurance plays an important role in preventing elders from financial hardship.
In 1999, 7 million elders, about 20% of the United States population over age 65, suffered from chronic disability (National Academy on an Aging Society, 1999
). Chronic disability creates costs not only in the form of the suffering a person experiences but also in the form of financial hardships. Currently, the annual health care expenditure for elders is $300 billion, which is projected to increase by 25% by 2030 (Center for Disease Control and Prevention, 1999). The significant and rising costs of medical care raise the question of how elders deal with these costs and to what extent the costs haunt them as burdensome out-of-pocket expenses.
Health insurance is a principal source of financing elders' health care expenses (Hurd & McGarry, 1997). Almost all elders (98%) have Medicare, which is the public health insurance program that mostly covers hospital expenses, although an increasingly growing share of Medicare went to physician expenses, home health, and nursing facility expenses in the 1990s. However, Medicare covers only about 45% of elders' medical expenses, so that a significant proportion of elders (70%) buy additional private insurance or depend on Medicaid, if qualified (11%).
On the other hand, the level of income for elderly households decreases to below 68% of the income level at the age of 55 (Anderson & Hussey, 1999), with a median annual income of only $17,000. Because of both limited insurance coverage and reduced income, out-of-pocket medical expenses consume 22% to 52% of elderly household income (Maxwell, Moon, & Segal, 2000), resulting in financial burdens. With health care costs rising, such burdens are expected only to increase.
Wealth is an alternative source to insufficient income and low health insurance coverage to pay for health care expenses. Life-cycle theory posits that individuals save during their working years and dissave during retirement for consumption, bequests, and emergencies (Mirer, 1979). Insufficient income and lack of sufficient health insurance coverage can accelerate the depletion of elders' lifetime savings. Coping with the gradual deterioration of health as well as sudden health events potentially puts elders' financial security at risk and creates significant stress (Ferraro & Su, 1999).
Although there has been extensive health economics literature, examining the relation between health and wealth, this literature has been focused on the demand for health care services across different levels of health and wealth (Bengt, 1998; Dardenoni & Wagstaff, 1990; Murrinen, 1982; Picone, Uribe, & Wilson, 1998), and/or saving motives (Hubbard, Skinner, & Zeldes, 1994; Lusardi, 1998). Whereas this literature has provided a framework to examine the impacts of wealth on health through health care demand models, it has not proposed direct implications for the investigation of the impacts of health on wealth.
On the other hand, a handful of studies have investigated the impacts of health on wealth. Smith (1997, 1999) found that new health events in later life result in wealth depletion through out-of-pocket medical expenses. Feinstein and Ho (2000) reported that sudden changes in family structure and health status increase the likelihood of using up assets. Wu (2001) found that changes in a spouse's health condition affect the financial security of the other spouse, and the effects are greater when health changes happen to wife rather than to husband.
These studies have brought much needed research attention to the investigation of the effects of health on wealth, but they are bounded by the following limitations. First, previous studies did not investigate the potential differences between new health events and existing health conditions. These might have different impacts on wealth depletion: new health events may cause an initial shock. After the initial shock, the financial effects of a chronic health condition may decrease and then rise again with the severity of the symptoms. Second, previous studies used descriptive (Smith, 1999) or ad hoc models (Feinstein & Ho, 2000; Wu, 2001) without deriving an explicit model from theories. Third, previous studies did not pay much attention to the impact of health on wealth across marital status. Because the levels of income, wealth, and health status differ between single and married households, the impacts of health on wealth may also differ. Addressing these issues, the following three research questions are investigated in this study: (a) What is the extent of elders' wealth depletion associated with health changes? (b) What are the differential impacts of new health events and existing health conditions? and (c) What are the differences in the impacts of health on wealth depletion across marital status?
|
Model |
|---|
|
TOPAbstractModelMethodsResultsDiscussionAppendixReferences |
|---|
), and the principal sources of wealth depletion are consumption and bequest. Several recent studies (Feinstein & Ho, 2000; Smith, 1999; Wu, 2001) have noted expenses involved with health care as another source of wealth depletion. In this study, we expand this literature.
The life-cycle theory or permanent income theory of consumer behavior postulates that consumption or savings in any given time period is dependent on lifetime or permanent income (Bhalla, 1980). Friedman (1957)'s formulation of permanent income theory is expressed by the following three equations at any given time period:
|
|
From the above equations, Bhalla (1980) derived a savings equation given by
|
|
Estimation Model
A main interest of this article is to examine whether declining health (both existing conditions and new health events) encourages wealth depletion in elders. Suppose there are two time periods. The total wealth at the end of the first time period is given by
|
|
Substituting Equation 4 to 5 gives
|
|
Likewise, we can obtain the total wealth at the end of the second time period:
|
|
We are interested in total wealth change during the two time periods. Given that permanent incomes and factors influencing k in each period are constant, subtracting Equation 6 from 7 gives
|
|
W2 is W2 - W1, Y2 is Yt2 - Yt1, C2 is Ct2 - Ct1, and W1 is W1 - W0.
A considerable portion of consumption in later life is allocated for health care services and transfer to children in addition to general spending such as food, housing or clothing, which is not largely changed. Also, by definition, transitory consumption is not planned but accidental in any period. Therefore, the consumption changes, C2, can be assumed to be determined by health care services consumption (M2) and transfer to children (T2) between this time period.
|
|
Health economics literature (Bengt, 1998; Feldstein, 1998; Picone, Uribe, & Wilson, 1998; Wagstaff, 1986) explains the determinants of health care services by a set of physician and patient factors, which include incidence of illness or need for care, demographics, and economic factors. A behavioral model of health care services utilization (Anderson & Newman, 1973) explains that an individual's use of health care services is determined by societal determinants, health services system, and individual determinants, which consist of predisposing, enabling, and illness level. Although the changes and differences in physician factors, societal determinants, and health care services system are important predictors, the scope of this study is limited to the individual differences within a given social and health care services environment. Focusing on individual determinants of health care services utilization, health care services consumption (M2) can be replaced by individual determinants of illness level, existing health condition (H1) and new health events (H2), health insurance (HI), and demographic characteristics (D), which are predisposing and enabling factors. Therefore, we can express this model as an estimable econometric model like equation:
|
|
is an error term. We deleted interest rate (r), assuming it is almost the same across individual elders. On the basis of this model, the impacts of changes in health on wealth depletion of elders are estimated. |
Methods |
|---|
|
TOPAbstractModelMethodsResultsDiscussionAppendixReferences |
|---|
. The AHEAD data provide in-depth information about the economic status of households, including assets and income, as well as comprehensive information about different aspects of the health status of individuals, including the occurrence of chronic conditions and new health events. Moreover, the longitudinal nature of the data set allows us to estimate the impact of changes in health status on the financial security of elders.
To examine the impacts of the changes in individual health conditions on wealth depletion, only the respondents who participated in both Waves 1 and 2 are included in the analysis. There are 6,952 respondents that met this criterion. In comparing those who participated in both waves with those who participated in only one wave, a systematic difference is noted in Table 1. Specifically, the prevalence of existing health conditions is significantly higher among the respondents who participated in Wave 1 only. The main reason for their not participating in Wave 2 was death (9.86%). Only a small number of respondents participated in only Wave 2, and most of these were added because they married a respondent in the original sample or because they failed to respond to Wave 1 although initial contact was made. By limiting the sample to those who participated in both waves, our findings may underestimate the impacts of health on wealth depletion, because health care expenditures tend to increase near death.
|
; see Appendix, Note 1). This procedure can, however, lead to significant response errors and thereby to incorrect wealth and income data, which are important variables in this study. Following Cao's (2000) suggestion, we used only those records that were imputed based on closed brackets. Thus, 544 respondents whose wealth data were imputed based on open brackets or without brackets and 850 respondents whose income data were imputed based on open brackets or without brackets were excluded from further analysis. After eliminating the observations with missing values on key variables, such as wealth, 5,388 respondents were kept for further analysis. Among them, 2,370 respondents were single, and the remaining 3,018 were married.
Measures
The dependent variable of this study is wealth depletion. Although the estimation model is derived as a continuous variable of wealth changes, this study focuses on wealth depletion rather than wealth changes. Wealth depletion is defined as a binary variable, indicating whether or not a household depleted more than 10% of its wealth between Waves 1 and 2. The cutoff of more than 10% depletion is used to identify wealth depletion that is greater than a simple market fluctuation. Because the 10% cutoff is rather arbitrary, other cutoffs are also examined to ensure the stability of the findings. Net wealth is equal to the total value of all assets minus total debts (see Appendix, Note 2). All dollar figures for income and net wealth are converted into 1998 dollars, using the current methods version of the Consumer Price Index for all urban consumers.
Health status is measured using the following two proxy variables: incidence of new health events and existing health conditions. AHEAD data surveyed eight types of chronic conditions that are most prevalent among elders: high blood pressure, diabetes, cancer, lung disease, heart condition, stroke, psychiatric problems, and arthritis. Whereas a variety of ways of measuring health status are available, such as a simple count of all health conditions, a set of dummy variables for each condition, and separate counts of serious and nonserious conditions, each measure has pros and cons (Ferraro & Wilmoth, 2000). The incidence of new health events in our data is so small that it does not allow us to use dummy variables for each condition nor separate counts of serious and nonserious conditions. Thus, to avoid the problem of inadequate statistical power associated with this variable, we used a binary variable of having or not having one of eight types of chronic conditions in a baseline time period (Wave 1) and whether or not diagnosed with one of these chronic disease during the second time period (Wave 2).
The variable of income changes is obtained by estimating a model in which income in each period is regressed on age, marital status, race/ethnicity, residence, and education. Predicted values in this regression can be considered as a proxy of permanent income, and residuals can be a proxy of transitory income in the theoretical model (Bhalla, 1980). Then, income changes are represented by differences between residuals in the second period and the first period.
As the health insurance variable, a binary variable of having or not having additional health insurance, such as Medigap or private insurance, is used. The descriptions of other independent variables are presented in Table 2.
|
, 1999) and Smith and Kington (1997a, 1997b) reported that at older ages changes in economic resources have little additional impact on health. We tested the endogeneity of health status variables, using the Hausman Test with two instrumental variables, smoking and drinking. The results indicate that health status variables are not endogenous (see Appendix, Note 3). As Equation 9 suggests, the independent variables include income changes, existing health conditions, new health events, health insurance, demographics, wealth transfers, and wealth changes in previous time period. The only demographic variable included is living arrangement, because other demographic characteristics, such as age, education, race/ethnicity, and residence were used to estimate permanent income. To examine the potential differences in the impacts of health on wealth across marital status, we conducted separate estimations for single and married households. For single households, only the impacts of the respondents' health statuses are examined, whereas for married households, the impacts of health statuses of both husband and wife are examined.
|
Results |
|---|
|
TOPAbstractModelMethodsResultsDiscussionAppendixReferences |
|---|
|
To further illustrate the probability changes caused by these variables, predicted probability was computed using the mean values of all other independent variables. Thus, for the discrete change of significant binary variables, such as new health events (from no occurrence [=0] to the occurrence [=1]) and health insurance (from no additional health insurance [=0] to the holding of additional health insurance [=1]),
|
|
The estimated probability changes are presented in Table 4. The probability of wealth depletion was 0.61 for the married respondents without new health events. Once the respondents develop new chronic conditions, the probability of wealth depletion went up to 0.67. That is, the new health events increased the probability of wealth depletion by 6%. On the other hand, having additional health insurance decreased the probability of wealth depletion by 5%.
|
On the other hand, the income changes between the waves are found to be negatively associated with wealth depletion, suggesting that income increase decreases the probability of wealth depletion. The marginal effect of the income change on wealth depletion can be estimated as follows:
|
|
|
Discussion |
|---|
|
TOPAbstractModelMethodsResultsDiscussionAppendixReferences |
|---|
Among the married households, only new health events of husbands are significantly associated with the household's wealth depletion; the impacts of new health events of wives are not statistically significant. This finding is contrary to that of Wu (2001). Using a sample of 51- to 61-year-olds, Wu found that only the wife's health events led to declines in household wealth, whereas the husband's new health events did not significantly affect household wealth. The difference in age group between the samples suggests that households may react differently to health events across different stages of their life, or that different age cohorts may react to health events differently. However, further study is needed to clarify these differences.
In contrast to married households, we found that existing chronic conditions contributed to the wealth depletion of single households. Previous studies (Smith, 1999; Wu, 2001) did not investigate the effects of existing chronic conditions. The results of this study indicate that single elders have financial burdens when their chronic conditions persist. Some chronic conditions, such as high blood pressure and arthritis, are not life threatening in the short term, but are associated with high medication costs.
For married households, additional health insurance, such as Medigap and private insurance, contributes significantly to keeping elders from depleting wealth. This result supports the idea that health insurance coverage plays an important role in elders' financial security. However, such a positive influence is not found in single households. One possible explanation is that the benefits of holding additional health insurance may be greater for married households than single households.
We did not find any significant effects of wealth transfers and living arrangements for both married and single households in more than 10% of wealth depletion. However, living arrangement was found to significantly influence wealth depletion when the cutoff is 50% or more. One can argue that living arrangement is not a true exogenous variable, considering that elders typically move in with their children only when they become so frail that they cannot manage independent living or when they can no longer support themselves financially. Further investigation is needed to delineate the impact of this variable.
Finally, although there is considerable debate on whether some elders transfer their wealth in order to be eligible for Medicaid, we did not find any significant effects of wealth transfers. However, because only the data during the 2-year time period are used, the possibility of transferring wealth is not adequately captured in this study.
Conclusions and Implications
In this study, we found that new incidences of chronic conditions lead to considerable wealth depletion for elders among married households and that existing chronic conditions lead to considerable wealth depletion for single households. Considering that married households tend to be younger than single households, we can see a pattern of wealth depletion caused by health problems. For married households, an initial health shock depletes household wealth. Then, as the elders get older, they become single and their own health problems continue to deplete their wealth.
The significant effect of additional health insurance among married households suggest its benefits in reducing the shock of new health events, yet such benefits diminish as the health conditions become no longer new. Considering that current health insurance coverage is more comprehensive for hospital costs than pharmaceutical costs, and that the insurance often covers the households, the benefits are therefore greater for married than single households.
Awareness of the financial consequences of new health events and sustained health conditions in the later life has many messages to the public. The possibility of the significant costs of new health events calls for not only preventive health care behavior but also precautionary savings and the purchase of comprehensive health insurance. The public needs to consider these financial consequences of health care costs in later life in calculating their retirement savings goals.
One quarter of the elders who participated in AHEAD were absolutely certain to use up all their savings in 5 years due to medical expenses (data not shown). The elders fear wealth depletion caused by health problems, and their fear is real. Our findings demonstrate this effect of health on wealth depletion. However, our findings are bounded by the following limitations. First, we examined the impact of health on wealth depletion during a 2-year period, without investigating long- or mid-term effects. As elders age, their health status deteriorates, which may incur increasingly greater expenses over time. Alternatively, their demands for health care and associated expenses may not increase, because they may become accustomed to existing chronic conditions. Although we expect even greater impacts of health on wealth depletion for a longer period of time, this needs to be empirically examined.
Second, this study did not directly explain how the demands for health care costs influence wealth depletion in elders. Health care costs, in terms of out-of-pocket medical expenses associated with health service utilization (e.g., the number of doctors' visits or hospital nights stays) are expected to influence the wealth depletion of elders. However, this study was focused on the examination of the effect of existing and new chronic conditions without explicitly examining health care demand or associated out-of-pocket expenses.
|
Appendix |
|---|
|
TOPAbstractModelMethodsResultsDiscussionAppendixReferences |
|---|
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Received for publication April 19, 2002. Accepted for publication October 24, 2002.
|
References |
|---|
|
TOPAbstractModelMethodsResultsDiscussionAppendixReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  L. M. Murray and S. Boyd Protecting Personhood and Achieving Quality of Life for Older Adults With Dementia in the U.S. Health Care System J Aging Health, April 1, 2009; 21(2): 350 - 373. [Abstract] [PDF]
|
|
|
|
 |
|
 H. Kim and J. Lee The impact of comorbidity on wealth changes in later life. J. Gerontol. B. Psychol. Sci. Soc. Sci., November 1, 2006; 61(6): S307 - S314. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
| HOME | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|---|
