Differences in net worth between elderly black people and elderly white people.
|Article Type:||Statistical Data Included|
Aged (Economic aspects)
Finance (Demographic aspects)
Ozawa, Martha N.
|Publication:||Name: Social Work Research Publisher: National Association of Social Workers Audience: Academic; Trade Format: Magazine/Journal Subject: Sociology and social work Copyright: COPYRIGHT 2000 National Association of Social Workers ISSN: 1070-5309|
|Issue:||Date: June, 2000 Source Volume: 24 Source Issue: 2|
|Geographic:||Geographic Scope: United States Geographic Code: 1USA United States|
The economic well-being of elderly people is determined partly by
how many assets they have. This article presents the findings of a study
on the differences in the level of net worth held by white elderly
people and black elderly people and the correlates of net worth of these
two groups. The study found an enormous difference in net worth of these
groups, even after other variables were held constant. Regression
results indicated that whereas both lifetime earnings and human capital
variables are significant predictors of the net worth of white men,
these variables have little or no bearing on the net worth of black men.
Separate analyses of data for white women and black women found that
human capital variables are significant predictors of the net worth of
white women, but variable lifetime earnings is the significant predictor
of the net worth for black women.
Key words: economic security; net worth; old age; race
With the 21st century rapidly approaching, the public stance on the economic well-being of elderly people is changing. The philosophy behind the Social Security Act of 1935 was to help the nation share the risk of losing earnings because of old age and to pool the resources to ensure income security in old age for everyone. But the financial viability of social security began to be questioned in the 1970s, and since then, the idea of the privatization of social security has gained currency. Indeed, the centerpiece of one of the three reform proposals--the Personal Security Accounts (PSA) plan--made by the Advisory Council on Social Security, 1994-1996 (1997) is the privatization of a large part of social security. The spread of individual retirement accounts (IRAs), 401(k) plans, and Keogh accounts compounded the development of individual responsibility to ensure individuals' income security in old age.
The combination of the growing financial difficulty in funding social security benefits on a pay-as-you-go basis for an increasing number of elderly people and the spread of IRAs is forcing many countries to adopt the full or partial privatization of social security. Australia and some Latin American countries already have some forms of IRAs instead of traditional social security systems. Sweden allows individuals to shift part of their payroll taxes into private accounts (Feldstein, 1997).
As the traditional form of social security is minimized, the economic well-being of elderly people will depend increasingly on what they have in the form of private pensions and private wealth. Indeed, even now, income stratification among the elderly population is determined not by the amount of social security benefits, but by the amount of income from assets and from private pensions or annuities. Government data indicate that the highest fifth of aged individuals (individuals or couples) draw 24.4 percent of their total income from asset income, compared with only 2.7 percent for the lowest fifth. Likewise, the top fifth received 10.5 percent of their total income from private pensions or annuities, compared with 1.7 percent received by the bottom fifth (Glad, 1996).
The present study focused on the net worth of white and black elderly people. Financial assets, such as stocks and bonds, can generate income (Ozawa, 1997). The ownership of a home provides in-kind income, in the form of rent-free housing. Furthermore, the holding of a large net worth, whether liquid or not, generates an intangible sense of economic well-being. Thus, the investigation of net worth of white elderly people versus black elderly people provides vital information to assess these two groups' economic security in old age.
The specific questions the study addressed were these:
* What is the level of net worth held by white and black people 10 years after retirement?
* Controlling for other variables, what are the racial differences in net worth held 10 years after retirement?
* What are the differential effects of lifetime earnings and human capital variables (education, occupation, and labor force attachment) on the net worth of white men and black men?
* What are the differential effects of lifetime earnings and human capita] variables (education, occupation, and labor force attachment) on the net worth of white women and black women?
CONCEPTUAL FRAMEWORK AND REVIEW OF THE LITERATURE
The purpose of this study was to estimate the net effect of race on net worth in old age and to investigate whether lifetime earnings and human capital variables have a differential effect on the level of net worth of elderly white people and black people. Thus, the independent variables used in the study were race, lifetime earnings, and human capital--education, occupation, and labor force attachment--that is, attributes that contribute to people's performance in the economy. Whether an individual fares well in the economy depends on the level of education, type of job, and degree of attachment to the labor force (Cain, 1966, 1986).
In this study, both lifetime earnings and human capital variables were included as independent variables. Because lifetime earnings are determined partly by human capital variables, which creates the problem of endogenuity, these two sets of variables should not be included in the same regression model. Therefore, we developed two models of regression--one including lifetime earnings and the other including human capital variables.
In addition to these independent variables, other demographic variables that are known to affect the dependent variable were included as controls: gender, marital status, age, and the number of children raised. Number of children was included because raising children incurs extra living expenses, which results in fewer opportunities to accumulate assets (Smith, 1989).
To summarize, the framework was as follows:
* dependent variable: net worth in old age
* independent variables: race, lifetime earnings, and human capital variables--education, occupation and labor force attachment
* control variables: age, gender, marital status, and number of children raised
The literature supports the inclusion of the variables just listed. Smith (1997), using the Asset and Health Dynamics among the Oldest Old (AHEAD), found that the median net worth of black elderly households was $17,000, compared with $90,000 for their white counterparts. The distribution of financial assets was just as skewed: The median of financial assets of white elderly households was $10,500, but that of black elderly households was zero. Smith (1995) also found a similar pattern of inequality in net worth among households of nonelderly people approaching retirement. Eargle (1992) noted a great variation in wealth holding by race, with black people having only a small fraction of the wealth of white people.
In their study on net worth of families headed by people ages 24 to 34, Blau and Graham (1990) indicated that the following variables explain why the net worth of black people is so small: lower current income and permanent income (the expected level of income, given the economic and social backgrounds of the respondent). The study also found that these income variables, although significant, have a relatively weaker influence on black people's net worth than on white people's. Moreover, it demonstrated that even assuming that black people had the same level of incomes and the same degree of adverse locational and demographic characteristics, a large portion of the gap in wealth--78 percent--would remain. In the same vein, Oliver and Shapiro (1989) showed that given the same income level, the net worth of black families is only a fraction of that of white families. For example, the median net worth of black families with annual incomes of $45,000 to $59,999 was only 42 percent of that of their white counterparts.
On the basis of their research findings, Blau and Graham (1990) speculated that there were two major reasons for the great disparity in net worth of black and white families. First, the amount of inherited wealth and the amount of intergenerational transfers when parents are still alive are significantly smaller among black families than among white families. Second, black families tend to have more liquid assets (such as bank checking accounts and savings accounts) than other types of assets (such as business assets and home equity) that have higher rates of return. Therefore, black families find it more difficult to accumulate assets.
Lifetime earnings are expected to affect net worth in old age. People with low earnings are unable to set aside money to accumulate net worth. Lifetime earnings is the most appropriate indicator of what economists call "permanent income" (see Blau & Graham, 1990), which measures the enduring level of income of one's life. Ozawa, Lum, and Tseng (1999) indicated that lifetime earnings were significantly related to the level of net worth among people ages 73 and older, particularly among men. Smith (1997) found a strong correlation between family income and net worth among those ages 70 and over. Several studies demonstrated that the distribution of net worth is considerably skewed in favor of people with high incomes. Smith (1997) found that the net worth of the top 10 percent of household income distribution was $384,000; of the median, $7,800; and of the lowest 10 percent, $150. Eargle's (1992) study showed that the median net worth of the bottom quintile of elderly people was only $25,200, of which $21,700 represented home equity, but the median net worth of the top quintile was $343,000, of which home equity was $175,000. Del Bene and Vaughn (1992) reported that contingency assets (that is, assets minus home equity) were small among low-income people; only 16 percent of poor people had contingency assets in excess of $5,000.
The literature indicates that education has a positive influence on net worth. Better education is expected to provide people with skills in and knowledge of saving and investing. Ozawa et al. (1999) found that when other variables were controlled, the net worth of people with at least some college education was 435 percent greater than the net worth of those with an elementary school education. Smith (1997) noted a similar relationship between education and net worth.
Occupation also is expected to make a difference in net worth in old age. Ozawa and Law (1993) found, for example, that although physicians have higher lifetime earnings, their annual income in old age was lower than that of lawyers, implying that physicians did not do well in saving and investing while they practiced medicine. For another example, professors in many private universities receive free tuition benefits for their children's education, effectively freeing up their financial resources for accumulating assets. Furthermore, the degree of labor force attachment is expected to make a difference in net worth in old age. Strong attachment to the labor force means that earnings are steady throughout a person's working life, providing a stable lifestyle, which is conducive to setting aside part of earnings for savings and investment.
The literature also indicates that there is a great disparity in the net worth of men and women. In her study of net worth among people ages 73 and older, Ozawa (1997) found that in 1992, the median net worth of men was $130,823, compared with $79,318 for women ($146,301 versus $88,702 in 1996 dollars). Because there is a great difference in the amount of net worth that men and women hold, it is important to disaggregate the data by gender when studying racial differences in net worth.
Marital status is expected to make a difference in net worth in old age. For example, divorce results in the division of net worth previously held jointly by spouses. Widowhood also may cause a decline in net worth because of the extraordinary expenditures incurred by the death of a spouse. People who never marry may be at a disadvantage in accumulating net worth as well, because they do not benefit from the economy of scale in consumption expenditures that married couples generally enjoy. A multiple regression analysis by Ozawa et al. (1999) of the net worth of elderly people ages 73 and older indicated that the net worth of nonmarried individuals was significantly smaller than that of married people and that divorced people had the least net worth.
Ozawa et al. (1999) also found that age was positively related to net worth among men and women when other variables were taken into account. At the descriptive level, however, Radner (1993) stated that the median net worth of the old-old (ages 80 and over) was less than that of the young-old (ages 65 to 69). The difference stems from the fact that Rader used a univariate variable whereas Ozawa et al. used a multivariate variable.
The source of data for the study was the New Beneficiary Data System (NBDS), which allowed us to merge the data from the 1982 New Beneficiary Survey (NBS) and the 1991 New Beneficiary Followup (NBF). The NBS had a sample of 18,136 individuals, representing about 2 million people who had become new beneficiaries of Old-Age, Survivors, and Disability Insurance (OASDI) or Medicare or both between mid-1980 and mid-1981; the response rate was 85.9 percent. The NBF interviewed 12,128 surviving NBS respondents and 1,834 surviving spouses of NBS respondents who had died; the response rate was 87.5 percent (Maxfield, 1985; Social Security Administration, 1994).
The NBS collected information on demographic characteristics, employment history, marital status, health status, work history, current income, and assets. The NBF collected, in addition, information on employment history during the 10 years after the NBS, changes in marital status, and their economic effects. Furthermore, the NBF data file included administrative data on annual earnings from 1951 through 1991, as well as other data, such as the Primary Insurance Amount, Medicare expenditures, and Supplemental Security Income applications and denials.
For the purposes of our study, we selected individuals who started receiving OASDI benefits between mid-1980 and mid-1981 as retired workers, wives, divorced wives, widows, and surviving divorced wives and individuals who started receiving Medicare coverage without receiving cash retirement benefits (Medicare-only beneficiaries) during the same period and survived long enough to be interviewed in the NBF, which collected data between November 1990 and July 1992. We excluded all people who started receiving social security benefits as widows or surviving divorced wives before age 62. We further selected those who were either white or black. Thus, the sample was 8,352, of whom 686 were black and 7,666 were white.
Dependent Variable: Net Worth
The dependent variable was the level of net worth held in 1992 by the respondent and his or her spouse, if married. Net worth--the total assets minus debts--consists of money market accounts, certificates of deposit, savings or credit union accounts, checking accounts, bonds, stocks and mutual funds, IRA/ Keogh accounts, home equity (market value of home minus mortgage), business equity, professional practice equity, farm equity, and rental property/vacation home/commercial property/land.
Race. The value of 1 was assigned to black respondents, and the value of zero, to white respondents.
Lifetime Earnings. To develop this variable, earnings in covered employment in each year from 1951 to 1991 were indexed to the 1992 consumer price index and then summed.
Education. Education was measured by dummy variables for the groups of respondents who had some high school education, completed high school, and had at least some college education. Those who had an elementary school education were assigned to the reference group.
Occupation. Occupation meant the occupation in the job held the longest. It was measured by dummy variables for the groups of respondents whose longest jobs were managerial, technical, precision production, operative, and farming and for those who held no jobs. Those whose longest jobs were service jobs were assigned to the reference group.
Labor Force Attachment. This variable was measured by two variables--quarters of covered employment from 1937 through 1977 and the number of years with earnings from 1982 through 1991. The former measured the degree of labor force attachment before the initial receipt of social security or Medicare benefits, and the latter measured the degree of labor force attachment thereafter. That the data for the quarters of covered employment ended in 1977 and did not include the quarters for 1978 through 1981 constituted a limitation of this study.
Marital Status. Marital status was measured by dummy variables for respondents who were (1) widowed, (2) separated or divorced, and (3) never married. Those who were married were assigned to the reference group.
Age, Gender, and Number of Children. Age was the age at the time of the NBF, which was the year of the initial receipt of social security benefits or Medicare, plus 11. The value of 1 was assigned to men, and the value of zero, to women. The variable for the number of children was self-explanatory.
The following variables were measured at the time of the NBS: gender, race, education, and occupation. The rest of the variables were measured at the time of the NBF.
We used the weight variable that was developed by the Social Security Administration to generate descriptive statistics on all variables. This procedure was needed to adjust for the sampling, poststratification, and nonresponse biases in the NBS and NBF data sets. The unit of analysis was the individual. In performing OLS regression analysis, net worth and lifetime earnings were logged because the distribution of these variables was skewed. To deal with the endogenuity problem mentioned earlier, we developed model 1 and model 2 for regression analysis-one including lifetime earnings and the other including variables related to human capital.
Characteristics of Respondents. The sample comprised 92.6 percent white respondents and 7.4 percent black respondents (Table 1). The proportion of females was larger among black people (55.4 percent) than among white people (51.0 percent). A smaller percentage (46.1 percent) of black people than of white people (65.1 percent) were married.
TABLE 1--Characteristics of Elderly Respondents (from the NBS and NBF)
NOTES: NBS = New Beneficiary Survey and NBF = New Beneficiary Follow-up (Social Security Administration). Percentages do not necessarily add up to 100 due to rounding.
Educational achievement differed as well, with 13 percent of black people having at least some college education, compared with 31.2 percent of white people. Similarly, the types of occupations that black people and white people had in their longest jobs differed greatly. For example, whereas 24.9 percent of white respondents had managerial jobs, only 10 percent of black respondents did. The mean number of children raised by black respondents was 3.2, compared with 2.5 raised by white respondents.
The median amount of lifetime earnings of black people was $223,573, compared with $490,169 for white people. The median number of quarters of covered employment from 1937 to 1977 was 76 for black people, compared with 95 for white people. The median number of years with earnings from 1982 to 1991 was zero for both black and white people; the mean number of years with earnings was 2.2 and 2.3, respectively.
Level of Net Worth. The median net worth of black people was considerably smaller than that of white people: $16,091 versus $110,839 (Table 2). The gender differential in net worth was greater among white people than among black people on the basis of either mean or median figures.
TABLE 2--Net Worth of Elderly White People and Black People, 1992 (in 1992 dollars)
It is important to clarify the meaning of regression coefficients in multiple regression analyses in which the dependent variable is transformed into a natural log and independent variables take the form of dummy variables. Most researchers interpret multiple regression coefficients in this situation simply as a percentage difference. A coefficient of 0.363, for example, is interpreted as "36.6 percent greater." However, Halvorsen and Palmquiest (1980) warned that the traditional way of interpreting regression coefficients is wrong. They argued that
C = ln(1 + g) [e.sup.c] = 1 + g
g = [e.sup.c] - 1
where C = regression coefficient and g = relative effect.
Following this equation and applying the coefficient of 0.363, the relative effect g is 0.438, which can be interpreted as "43.8 percent greater." Thus, in the data analysis that follows, we used the procedure specified by Halvorsen and Palmquiest (1980). We calculated the values of the relative effect g (expressed in percentage) for all coefficients for dummy variables in the OLS regression analyses involving the natural log of net worth as the dependent variable. (This transformation is not required for independent variables that are interval.) We listed relative effects in percentage terms after coefficients.
OLS Regression Analysis of Net Worth for All Respondents. A regression analysis for all the respondents was performed to estimate the racial difference in net worth, controlling for other variables. In model 1 (Table 3), which includes lifetime earnings as an independent variable, the net worth of black people was found to be 95.4 percent smaller (p [is less than] .001) than that of white people, when other variables were controlled. In model 2, which includes variables related to human capital instead of lifetime earnings, the racial difference in net worth was -89.8 percent (p [is less than] .001). Under either model, it seems clear that the black-white difference in net worth is significant.
TABLE 3--OLS Multiple Regression Analysis of Net Worth (Log): All Respondents
NOTE: Reference groups are in parentheses.
(*) p < .05. (**) p < .01. (***) p < .001.
Lifetime earnings were strongly and positively related to net worth in old age (p [is less than] .01). This finding indicates that those who had earned more during their working lives did have larger net worth in their old age.
Model 2 (Table 3) shows the coefficients of human capital variables. Education exerted a strong, positive effect on net worth in old age. For example, the net worth of those who had at least some college education was 305.2 percent greater (p [is less than] .001) than that of those with an elementary school education. Also, the type of job was strongly related to net worth. The net worth of those who had managerial jobs in their longest-held jobs was 189.7 percent greater (p .001) than the net worth of those who had service jobs. Caution is in order with regard to the coefficient of "never worked," which was larger than that for "services." This anomalous relationship was due to the fact that many women in the sample had never worked (8.6 percent; data not shown), some of whom might have been married to men with a sizable net worth. In separate data analyses, we found that the net worth of women who had never worked was significantly higher than that of those who had service jobs, but the net worth of men who had never worked was as low as the net worth of those who had service jobs (data not shown). The variable quarters of covered employment from 1937 to 1977 was significantly and positively related to net worth (p [is less than] .001); so was the number of years with earnings from 1982 to 1991 (p [is less than] .001).
OLS Regression Analyses of Net Worth of White Men Compared with Black Men. There were amazingly different regression results for black men and white men (Tables 4 and 5).
TABLE 4--OLS Multiple Regression Analysis of Net Worth (Log): White Men
NOTE: Reference groups are in parentheses.
(*) p < .05. (**) p < .01. (***) p < .001.
TABLE 5--OLS Multiple Regression Analysis of Net Worth (Log): Black Men
NOTE: Reference groups are in parentheses.
(*) p < .05. (**) p < .01. (***) p < .001.
Model 1 (Tables 4 and 5) shows that when other variables were controlled, both lifetime earnings and human capital variables (education, occupation, quarters of covered employment from 1937 to 1977, and years with earnings from 1982 to 1991) exerted a positive and significant effect on the net worth of white men, but for black men, only quarters of covered employment from 1937 to 1977 (p[is less than].05) had bearing on net worth in old age.
Although the number of children raised was just a control variable, its differential coefficients for white men and black men are interesting. For white men, the more children raised, the smaller the net worth (p [is less than] .001), but for black men, there was no such relationship.
These regression results indicate that for black men, the level of education, the type of jobs they had during their working lives, and how much they earned in their lifetime had no effect on how much net worth they eventually had in old age. Only quarters of covered employment from 1937 to 1977--one of the indicators of labor force attachment--was related to net worth (p[is less than].05). In contrast, the level of net worth for white men clearly was determined by education, occupation, labor force attachment, and lifetime earnings. The levels and the significance of these variables were all high (p [is less than] .001), and the directions of these variables were in the predicted directions.
OLS Regression Analyses of Net Worth of White Women Compared with Black Women. We found a strong contrast in the regression results for white women and black women as well. With the exception of years with earnings from 1982 to 1991, all human capital variables had a significant effect on the net worth of white women; but the effect of these variables on the net worth of black women was much weaker (Tables 6 and 7). In particular, for black women, only the following variables were significant: high school education (p [is less than] .05), college education (p [is less than] .001), technical job (p [is less than] .05), and precision-production job (p [is less than] .05). It is noteworthy, however, that college education made an enormous difference in net worth of black women: These women's net worth was 1,023.9 percent greater (p [is less than] .001) than the net worth of black women with an elementary school education. On the other hand, lifetime earnings exerted a strong and significant effect on the net worth of black women (p [is less than] .01), but not on the net worth of white women. This difference was an interesting contrast.
TABLE 6-OLS Multiple Regression Analysis of Net Worth (Log): White Women
NOTE: Reference groups are in parentheses.
(*) p < .05. (**) p < .01. (***) p < .001.
TABLE 7--OLS Multiple Regression Analysis of Net Worth (Log): Black Women
NOTE: Reference groups are in parentheses.
(*) p < .05. (**) p < .01. (***) p < .001.
Again, it is interesting to note that in both models 1 and 2, the number of children had a negative effect on the net worth of white women (p [is less than] .001 in both models), but for black women, this variable was significant only in model 1 (p [is less than] .01).
The findings illustrate the great differences in the situations of white people and black people. The difference in the net worth of white men and black men is particularly vivid. Lifetime earnings and human capital variables made little or no difference in the amount of net worth that black men had in old age (Table 5). In contrast, the coefficients for white men were all significant and in the expected directions (Table 4). What white men did in their economic lives while they were young was all related to the amount of net worth they had in old age.
DISCUSSION AND IMPLICATIONS
This study revealed an enormous difference in the amount of net worth of white people and black people in old age. Even when other variables were held constant, the net worth of black people was about 90 percent smaller (-95.4 percent in model 1 and -89.8 percent in model 2) than the net worth of white people (Table 3). Furthermore, the explanatory power of the independent variables under investigation is quite different for white people and black people. What is the most striking is that for white men, lifetime earnings and human capital variables were related to net worth as expected, but for black men, there was little or no relationship between the dependent variable and these independent variables, except the degree of labor force attachment from 1937 to 1977.
Another interesting finding was that although the lifetime earnings of white men did make a difference in their eventual net worth in old age, this variable did not make a difference in the net worth of white women. But for black people, the opposite was true: This variable had a positive effect on black women's net worth, but not on black men's.
From these findings, we can say that the net worth of white men is an end result of the amount of their human capital and how much they earned in their lifetimes. Albeit to a much lesser degree, such an observation applies to black women as well. For white women, what really counts is human capital. For black men, neither human capital variables nor lifetime earnings make much difference.
Caution should be exercised in interpreting the findings, for two reasons. First, the relatively weak coefficients of the independent variables for black people, in part, may be due to the relatively small sample of black people. Even taking this phenomenon into account, however, it seems clear that the black-white differences in the level of net worth and in the determinants of net worth remain.
Second, the proportion of variance explained by the regression models in the study was relatively small. For example, the [R.sup.2]s for models 1 and 2 were only 0.115 and 0.167, respectively (Table 5). This means that these regression models explained only a fraction of the variance in the dependent variables. It is possible that the inclusion of other types of variables may increase the predictive power of the regression models.
Third, the racial difference in net worth that was found in this study may not be present among black elderly people and white elderly people in general, because this study focused on elderly people ages 73 and older. However, a study by Ozawa (1997), which used both cross-sectional and longitudinal data from the NFDS, found that the same degree of racial difference was present among these elderly people 10 years earlier--that is, when they were ages 63 and older. A subsequent study by Ozawa et al. (1999), which used the two-stage regression technique, supported Ozawa's (1997) findings.
With these cautionary notes in mind, you should ask: Why do black people have so little net worth in old age, and why are the economic activities in their working lives not related to the amount of net worth they have in old age? The answer may be a culmination of many events that black people undergo throughout their lives and from generation to generation. As Blau and Graham (1990) argued, the black people in this study might have had little inheritance from their parents or financial help from their parents to buy homes. Also, even if they had financial resources, they might have had them in liquid forms, such as savings accounts at banks. (People need to invest in stocks and bonds instead to have high rates of return, although stocks and bonds are riskier forms of investment.) On top of all this, the doors to investment institutions might have been closed to black people because of racial discrimination in the business practices of these institutions (Jackman & Jackman, 1980; Kain & Quigley, 1972). Thus, the composition of their assets--if they had assets--might have been different from that of white people, resulting in lower rates of returns.
Profound policy implications can be drawn from the findings of this study. Because black people's assets are so small, they will have little or no income from assets. As a result, social security benefits will constitute a major part of their retirement income. Thus, the PSA plan, if adopted, may have an adverse effect on black people.
The PSA plan attempts to transform the current OASDI into a two-tier system, with the first tier providing a flat-amount benefit (equivalent to 65 percent of the poverty line) and the second tier consisting of large-scale individualized retirement accounts, financed by 5 percentage points of the OASDI employee payroll tax.
Because black workers, on average, receive considerably lower wages than do white workers (U.S. Department of Commerce, 1998), 5 percent of the taxable payroll that they would put in their individual retirement accounts would be small in absolute terms. With this in mind, suppose black people have the same likelihood of investing in the equity market, allocate the same proportion of their financial resources to the equity market, and obtain the same rate of return in their equity investment as do white people. Even under the best circumstances, as assumed here, a vital policy question remains: Would the proportionately same return be good enough for black people from the social policy point of view?
The OASDI, as designed, purports to provide low-wage earners with disproportionately higher rates of return than with high-wage earners. This policy reflects the social adequacy principle embedded in the benefit formula. But under the privatized system of social security, the benefits would be determined by individuals' investment behavior. Under such a scheme, black people would be expected to have a lower rate of return on their investment-let alone the proportionately same rate of return, because they tend not to invest their money in the equity market, which is known to yield a higher rate of return than do other types of investments (Ozawa & Lum, 1999).
The crucial vehicle that may mitigate the anticipated disadvantage of black people would be the first-tier flat-amount of social security benefits that the PSA plan incorporates. As mentioned earlier, the first-tier flat-amount benefits would be equal to 65 percent of the poverty line, which is further adjusted by the years of work in covered employment--with 35 years being considered full, lifetime work. Thus, the issue is whether the flat-amount benefits would offset the anticipated poor investment outcomes among black workers so that the combination of first-tier and second-tier benefits would provide proportionately higher rates of return for black people. Unfortunately, this flat-amount benefit, which is about $410 in today's dollars, may be inadequate to achieve this goal. Although this is an empirical question that needs to be investigated further, the prognosis seems poor.
Advisory Council on Social Security, 1994-1996. (1997). Report of the 1994-1996 Advisory Council on Social Security: Vol. 1. Findings and recommendations. Washington, DC: Author.
Blau, R. D., & Graham, J. W. (1990). Black-white differences in wealth and asset composition. Quarterly Journal of Economics, 105, 321-339.
Cain, G. G. (1966). Married women in the labor force: An economic analysis. Chicago: University of Chicago Press.
Cain, G. G. (1986). The economic analysis of labor market discrimination: A survey. In O. Ashenfelter & R. Layard (Eds.), Handbook of labor economics (Vol. 1, pp. 694-785). New York: Elsevier Science.
Del Bene, L., & Vaughn, D. R. (1992). Income, assets and health insurance: Economic resources for meeting acute health care needs of the aged. Social Security Bulletin, 55(1), 3-25.
Eargle, J. (1992). Household wealth and asset ownership: 1988 (Current Population Reports, Series P-70, No. 22). Washington, DC: U.S. Department of Commerce, Bureau of the Census.
Feldstein, M. S. (1997). The case for privatization. Foreign Affairs, 46, 24-38.
Glad, S. (1996). Income of the population 55 or older, 1994. Washington, DC: Social Security Administration.
Halvorsen, R., & Palmquiest, R. (1980). The interpretation of dummy variables in semilogarithmic equations. American Economic Review, 70, 474-475.
Jackman, M. R., & Jackman, R. W. (1980). Racial inequalities in home ownership. Social Forces, 63, 1221-1234.
Kain, J. E, & Quigley, J. M. (1972). Housing market discrimination, home ownership, and savings behavior. American Economic Review, 72, 263-277.
Maxfield, L. D. (1985). Income of new retired workers by age at first benefit receipt: Findings from the New Beneficiary Survey. Social Security Bulletin, 48(7), 7-26.
Oliver, M. L., & Shapiro, T. M. (1989). Race and wealth. Review of Black Political Economy, 17(4), 5-25.
Ozawa, M. N. (1997). Income and net worth of the elderly in the United States. In International Longevity Center (Ed.), Economic conditions of the elderly and the safety net (pp. 170-260). Tokyo: Shinkodo Press.
Ozawa, M. N., & Law, S. W. (1993). Earnings history of social workers: A comparison to other professional groups. Social Work, 38, 542-551.
Ozawa, M. N., & Lum, Y. S. (1999, November 20). Taking risks in investing in the equity market: Racial and ethnic differences. Paper presented at the 52nd Annual Scientific Meeting of the Gerontological Society of America, San Francisco.
Ozawa, M. N., Lum, Y. S., & Tseng, H. Y. (1999). Net worth at retirement and 10 years later. In S. S. Nagel (Ed.), The substance of public policy (pp. 245-275). New York: Nova Science.
Radner, D. B. (1993). Economic well-being of the old old: Family unit income and household wealth. Social Security Bulletin, 56(1), 3-19.
Smith, J. P. (1989). Women, mothers, and work. In M. N. Ozawa (Ed.), Women's life cycle and economic insecurity: Problems and proposals (pp. 42-70). New York: Greenwood Press.
Smith, J. P. (1995). Racial and ethnic differences in wealth using the HRS. Journal of Human Services, 30, S 158S183.
Smith, J. P. (1997). Wealth inequality among older Americans. Journal of Gerontology 52B (Special issue), 74-81.
Social Security Administration. (1994). The New Beneficiary Survey data system: User's manual. Washington, DC: Author.
U.S. Department of Commerce. (1998). Statistical abstract of the United States, 1998. Washington, DC: U.S. Government Printing Office.
Original manuscript received January 5, 1999
Final revision received November 3, 1999
Accepted December 23, 1999
Martha N. Ozawa, PhD, is Bettie Bofinger Brown Professor of Social Policy, George Warren Brown School of Social Work, Washington University, St. Louis, MO 63130-4899; e-mail: email@example.com. Huan-yui Tseng, PhD, is assistant professor, Department of Social Work, Shih-Chien University, Taipei, Taiwan.
Percentage All White Characteristic (N = 8,352) (N = 7,666) Race White 92.6 -- Black 7.4 -- Gender Male 48.7 49.0 Female 51.3 51.0 Race White 92.6 -- Black 7.4 -- Marital status Married 63.8 65.1 Widowed 25.8 25.0 Separated/divorced 6.4 5.9 Never married 4.0 4.0 Education Elementary school 21.0 18.7 Some high school 18.9 18.5 High school graduate 30.3 31.6 At least some college 29.8 31.2 Occupation Managerial 23.8 24.9 Technical 28.6 30.2 Service 12.6 10.3 Operative 15.4 14.7 Farmer 3.6 3.5 Precision production 11.9 12.2 Never worked 4.1 4.2 Mean Age 73.8 73.8 Number of children raised 2.6 2.5 Covered earnings, 1951-91 $440,402 $451,263 Education 11.5 11.7 Quarters of covered employment, 1937-77 85.0 85.7 Years with earnings, 1982-91 2.3 2.3 Median Age 75.0 75.0 Number of children raised, 1982 2.0 2.0 Lifetime earnings, 1951-91 $457,138 $490,169 Education 12.0 12.0 Quarters of covered employment, 1937-77 93.0 95.0 Years with earnings, 1982-91 0.0 0.0 Percentage Black Characteristic (N = 686) Race White -- Black -- Gender Male 44.6 Female 55.4 Race White -- Black -- Marital status Married 46.1 Widowed 35.5 Separated/divorced 13.2 Never married 4.6 Education Elementary school 49.1 Some high school 23.7 High school graduate 14.2 At least some college 13.0 Occupation Managerial 10.0 Technical 9.7 Service 40.1 Operative 23.8 Farmer 4.7 Precision production 8.4 Never worked 3.3 Mean Age 73.7 Number of children raised 3.2 Covered earnings, 1951-91 $300,634 Education 8.8 Quarters of covered employment, 1937-77 74.9 Years with earnings, 1982-91 2.2 Median Age 74.0 Number of children raised, 1982 2.0 Lifetime earnings, 1951-91 $223,573 Education 9.0 Quarters of covered employment, 1937-77 76.0 Years with earnings, 1982-91 0.0
Net Worth Respondent Categories Mean Median All 218,325 100,899 White 232,175 110,839 Men 300,024 144,214 Women 166,306 85,911 Black 40,091 16,091 Men 48,748 18,785 Women 32,665 15,452
Model 1 Relative Variable Coefficient Effect (%) t value Intercept 3.576(***) 3.986 Black -3.079(***) -95.4 -29.711 Male 0.189(**) 20.8 2.762 Age 0.107(***) 11.3 8.829 Marital status Widow/widower -1.120(***) -67.4 -15.812 Separated/divorced -2.712(***) -93.4 -25.272 Never married -1.927(***) -85.4 -13.438 (Married) Number of children raised -0.150(***) -10.131 Lifetime earnings, 0.031(**) 3.205 1951-91 (Log) Education Some high school High school graduate At least some college (Elementary school) Occupation Managerial Technical Precision production Operative Farmers Never worked (Services) Quarters of covered employment, 1937-77 Years with earnings, 1982-91 N 8,352 [R.sup.2] 0.221(***) F 296.142 Model 2 Relative Variable Coefficient Effect (%) t value Intercept 5.626(***) 6.124 Black -2.286(***) -89.8 -21.581 Male 0.007 0.7 0.094 Age 0.054(***) 4.344 Marital status Widow/widower -0.962(***) -61.8 -13.668 Separated/divorced -2.506(***) -91.8 -23.750 Never married -1.715(***) -82.0 -12.318 (Married) Number of children raised -0.093(***) -6.304 Lifetime earnings, 1951-91 (Log) Education Some high school 0.574(***) 77.5 6.332 High school graduate 1.067(***) 190.7 12.299 At least some college 1.399(***) 305.2 14.784 (Elementary school) Occupation Managerial 1.064(***) 189.7 9.640 Technical 0.900(***) 146.0 8.938 Precision production 0.782(***) 118.5 6.477 Operative 0.326(**) 38.6 2.960 Farmers 1.279(***) 259.3 7.569 Never worked 0.721(***) 105.7 4.106 (Services) Quarters of covered 0.005(***) 6.051 employment, 1937-77 Years with earnings, 0.032(***) 3.786 1982-91 N 7,890 [R.sup.2] 0.289(***) F 177.492
Model 1 Relative Variable Coefficient Effect (%) t value Intercept -3.115(**) -2.778 Age 0.157(***) 11.022 Marital status Widow/widower -0.680(***) -49.3 -6.827 Separated/divorced -2.377(***) -90.7 -15.440 Never married -1.514(***) -78.0 -7.623 (Married) Number of children raised -0.109(***) -6.135 Lifetime earnings, 0.258(***) 8.319 1951-91 (Log) Education Some high school High school graduate At least some college (Elementary school) Occupation Managerial Technical Precision production Operative Farmers Never worked (Services) Quarters of covered employment, 1937-77 Years with earnings, 1982-91 N 3,875 [R.sup.2] 0.126(**) F 92.758 Model 2 Relative Variable Coefficient Effect (%) t value Intercept 2.359(*) 2.211 Age 0.097(***) 6.822 Marital status Widow/widower -0.607(***) -45.5 -6.389 Separated/divorced -2.238(***) -89.3 -15.294 Never married -1.111(***) -67.1 -5.766 (Married) Number of children raised -0.079(***) -4.658 Lifetime earnings, 1951-91 (Log) Education Some high school 0.401(***) 49.3 3.853 High school graduate 0.763(***) 114.5 7.999 At least some college 1.056(***) 187.6 10.409 (Elementary school) Occupation Managerial 1.376(***) 296.1 9.408 Technical 1.109(***) 203.2 7.572 Precision production 0.991(***) 169.3 6.831 Operative 0.662(***) 93.8 4.378 Farmers 1.596(***) 393.1 8.95 Never worked 0.707 102.8 0.371 (Services) Quarters of covered 0.004(***) 4.871 employment, 1937-77 Years with earnings, 0.037(***) 4.12 1982-91 N 3,859 [R.sup.2] 0.206(***) F 62.240
Model 1 Relative Variable Coefficient Effect (%) t value Intercept -3.544 -0.429 Age 0.096 0.922 Marital status Widow/widower -1.655(*) -80.9 -2.526 Separated/divorced -3.507(***) -97.0 -5.049 Never married -1.658 -81.0 -1.328 (Married) Number of children raised -0.123 -1.563 Lifetime earnings, 1951-91 0.438 1.799 (Log) Education Some high school High school graduate At least some college (Elementary school) Occupation Managerial Technical Precision production Operative Farmers Never worked (Services) Quarters of covered employment, 1937-77 Years with earnings, 1982-91 N 275 [R.sup.2] 0.115(***) F 5.819 Model 2 Relative Variable Coefficient Effect (%) t value Intercept 9.150 1.127 Age -0.019 -0.177 Marital status Widow/widower -1.319(*) -73.3 -1.997 Separated/divorced -3.378(***) -96.6 -4.735 Never married -1.573 -79.3 -1.263 (Married) Number of children raised -0.073 -0.932 Lifetime earnings, 1951-91 (Log) Education Some high school 0.311 36.5 0.468 High school graduate -0.051 -5.0 -0.065 At least some college 0.864 137.3 0.946 (Elementary school) Occupation Managerial 1.918 580.9 1.826 Technical 0.882 141.5 0.923 Precision production -0.256 -22.6 -0.331 Operative -1.099 -66.7 -1.686 Farmers -1.104 -66.8 -1.108 Never worked 0.000 -- -- (Services) Quarters of covered 0.014(*) 2.18 employment, 1937-77 Years with earnings, -0.048 -0.618 1982-91 N 271 [R.sup.2] 0.167(***) F 3.395
Model 1 Relative Variable Coefficient Effect (%) t value Intercept 4.798(***) 3.625 Age 0.096(***) 5.336 Marital status Widow/widower -1.154(***) -68.5 -12.628 Separated/divorced -2.819(***) -94.0 -19.797 Never married -1.873(***) -84.6 -10.000 (Married) Number of children raised -0.180(***) -7.642 Lifetime earnings, 0.003 0.324 1951-91 (Log) Education Some high school High school graduate At least some college (Elementary school) Occupation Managerial Technical Precision production Operative Farmers Never worked (Services) Quarters of covered employment, 1937-77 Years with earnings, 1982-91 N 3,791 [R.sup.2] 0.122(***) F 87.704 Model 2 Relative Variable Coefficient Effect (%) t value Intercept 5.648(***) 4.055 Age 0.052(**) 2.761 Marital status Widow/widower -1.006(***) -63.4 -10.727 Separated/divorced -2.633(***) -92.8 -18.303 Never married -1.878(***) -84.7 -10.087 (Married) Number of children raised -0.083(***) -3.307 Lifetime earnings, 1951-91 (Log) Education Some high school 0.783(***) 118.8 5.537 High school graduate 1.374(***) 295.3 10.063 At least some college 1.695(***) 444.6 11.213 (Elementary school) Occupation Managerial 0.903(***) 146.6 5.634 Technical 0.812(***) 125.2 6.046 Precision production 0.879(***) 140.7 3.348 Operative 0.329(*) 39.0 2.030 Farmers 1.186(**) 227.4 2.685 Never worked 0.798(***) 122.2 4.053 (Services) Quarters of covered 0.005(***) 3.853 employment, 1937-77 Years with earnings, 0.016 1.111 1982-91 N 3,370 [R.sup.2] 0.203(***) F 53.253
Model 1 Relative Variable Coefficient Effect (%) t value Intercept 19.599(***) 3.838 Age -0.160(*) -2.359 Marital status Widow/widower -2.277(***) -89.7 -4.729 Separated/divorced -2.130(***) -88.1 -3.434 Never married -3.216(***) -96.0 -3.476 (Married) Number of children raised -0.183(**) -2.635 Lifetime earnings, 0.195(**) 3.075 1951-91 (Log) Education Some high school High school graduate At least some college (Elementary school) Occupation Managerial Technical Precision production Operative Farmers Never worked (Services) Quarters of covered employment, 1937-77 Years with earnings, 1982-91 N 411 [R.sup.2] 0.098(***) F 7.321 Model 2 Relative Variable Coefficient Effect (%) t value Intercept 21.036(***) 3.939 Age -0.171(*) -2.419 Marital status Widow/widower -2.247(***) -89.4 -4.594 Separated/divorced -2.376(***) -90.7 -3.760 Never married -3.191(***) -95.9 -3.444 (Married) Number of children raised -0.137 -1.872 Lifetime earnings, 1951-91 (Log) Education Some high school 0.284 32.9 0.571 High school graduate 1.432(*) 318.8 2.429 At least some college 2.419(***) 1,023.9 3.398 (Elementary school) Occupation Managerial 0.134 14.4 0.186 Technical 1.319(*) 274.0 2.013 Precision production 3.951(*) 5,098.3 2.463 Operative 0.125 13.3 0.206 Farmers 0.916 149.8 0.566 Never worked -0.451 -36.3 -0.454 (Services) Quarters of covered 0.006 1.161 employment, 1937-77 Years with earnings, 0.073 1.108 1982-91 N 390 [R.sup.2] 0.169(***) F 4.749
|Gale Copyright:||Copyright 2000 Gale, Cengage Learning. All rights reserved.|