Changing roles of parental economic resources in children's educational attainment.
KEY WORDS: assets and debt; cohort; education; income; inequality
Equal educational opportunity is considered a key indicator of a society's fairness. It is a deep-rooted belief in U.S. culture that every child should have an equal opportunity to receive the best possible education. Educational opportunity is of particular importance to social workers, whose mission is "to enhance human well-being and help meet the basic human needs of all people, with particular attention to the needs and empowerment of people who are vulnerable, oppressed, and living in poverty" (NASW, 2008, p. 1). Although it is generally recognized that education is a crucial path toward future well-being, it is not clear whether equal educational opportunity exists in the United States.
The existing literature indicates that parents' economic resources are strong predictors of educational attainment: Children from high-income and wealthy families are likely to achieve better educational outcomes than are economically disadvantaged children (Conley, 2001; Ellwood & Kane, 2000; Mare, 1981). Accordingly, measuring the association between parents' economic resources and children's educational attainment over time should provide a way to gauge the United States' progress toward the ideal of equal educational opportunity.
This study investigated changes in the role of parents' economic resources over time by comparing two cohorts of children from the Panel Study of Income Dynamics (PSID) data. Given that household wealth is not perfectly correlated with income (Wolff, 1990), we paid special attention to parents' assets and used diverse measures of economic resources: income, net worth, liquid assets, and home ownership.
Recent decades have witnessed socioeconomic changes that may have affected educational opportunities. First, the cost of college education has risen rapidly, whereas financial aid has shifted more toward non-need-based aid (Kane, 2004). These changes are expected to exacerbate the burden of low-income children in financing college education. Second, the value of a college education has increased continuously since the late 1970s, as reflected in growing earning gaps between workers with and without college degrees (Acemoglu, 2002). The increased value of education will heighten the impact of economic resources if parents with low resources are unable to increase investment in their children's education. It is also plausible that it would motivate low-resource parents to shift their resources toward children's education (Nam, 2004). Third, income and wealth inequalities have widened in the past few decades (Neckerman & Torche, 2007), and geographic concentration of wealth and poverty has intensified (Massey, 1996). Higher levels of income inequality may have contributed to growing educational gaps, improving high-income children's educational outcomes while lowering those of low-income children (Mayer, 2001). Geographic concentrations of wealth and poverty are also likely to exacerbate inequality in school resources given that public schools receive a large share of their funding from local property taxes (Card & Payne, 2002; Conley, 2001). Finally, access to the consumer credit market has widened, especially among moderate-income households. The percentage of households with credit card debt has increased, whereas the gaps between desired and actual levels of borrowing have declined since the 1980s (Lyons, 2003). Increased credit market accessibility may have dual impacts. On the one hand, it may help low-resource children to continue their education when their families' income temporarily declines; on the other hand, it may impose financial constraints in the long term because parents with consumer debt tend to be less willing or able to borrow for children's education.
A small number of recent studies have empirically tested whether the relationship between parents' income and educational attainment has changed over time. Morgan and Kim (2006) and Ellwood and Kane (2000) concluded that the effect of parents' income on college enrollment has remained stable. Belley and Lochner (2007) showed a substantial increase in the effect of parents' income on college attendance but little change in regard to high school graduation. Discrepancies in existing empirical studies justify further investigation on the topic.
The existing literature contains few investigations of the role of assets and access to the credit market. Parents' assets and debt, however, are likely to affect children's educational attainment. Liquid assets (wealth that can be converted into cash easily, such as money in bank accounts and stocks) may reduce households' need to borrow for children's education and may prevent children from dropping out of school by providing economic resources for needed consumption at times of economic difficulty. Home ownership may facilitate borrowing for education by providing collateral to lenders while affecting the quality of educational experience, which is highly correlated with pupil-teacher ratio and a teacher's level of experience (Rouse & Barrow, 2006).Access to the credit market may improve educational outcomes by providing economic means to get through economically difficult times. Debt, however, may be a sign of financial vulnerability, because households with high levels of debt may have trouble getting additional credit for their children's education (Gruber, 2001; Nam & Huang, 2009).
Previous empirical studies show that parents' assets significantly increase children's educational attainment (Conley, 2001; Nam & Huang, 2009; Zhan & Sherraden, 2003). Using a PSID sample of children ages eight to 19 in 1984 (N = 1,126), Conley (2001) showed a strong association between parents' assets and college attendance, noting that doubling parents' net worth is estimated to increase children's probability of attending college by 8.3%. From a sample of PSID children ages 15 to 17 in 1994 (N = 402), Nam and Huang (2009) showed that children from negative-liquid-asset households (those whose debts exceed their savings) are more likely to graduate from high school but less likely to graduate from college than their counterparts from zero-liquid-asset households.
Morgan and Kim's (2006) study is, to the best of our knowledge, the only study that investigated changes in the association between parents' assets and children's educational attainment. By comparing two cohorts (1986 and 1996) consisting of youths ages 17 to 21 from the Survey of Income and Program Participation (SIPP) data (N = 6,894), they argued that the effects of parents' net worth and home equity on college attendance have remained stable. Invaluable as it is, Morgan and Kim's (2006) study has limitations. The study did not include liquid assets in its analysis, one of the major financial sources for education. In addition, they used data from the SIPE which is not an ideal data set for studying associations between parents' economic resources and children's education (Ratcliffe et al., 2008).A substantial proportion of their young adult sample was unable to provide information on their parents' income and assets, so Morgan and Kim imputed these values using regression with gender, race, years of education, monthly income, and homeownership. This method, however, is prone to measurement errors. Finally, they used a monthly measure of family income, although previous studies have recommended using at least three years' worth of data to account for income fluctuation (Solon, 1992).
Data and Sample Selection
This study used data from the PSID. The PSID is a longitudinal survey of a nationally representative sample and has collected information on individual and household characteristics since 1968 (Hill, 1992). The PSID collected wealth data every five years between 1984 and 1999 and every two years thereafter. The reliability and validity of the PSID's household income and wealth were tested in previous research (Hill, 1992; Ratcliffe et al., 2008). The PSID is considered an ideal data set for studying relationships between parents' resources and children's adult outcomes because it follows the children of the original sample families after they leave their parents' houses (Hill, 1992; Solon, 1992). In addition to the PSID, the present study also used the Bureau of Labor Statistics' state unemployment rate data.
The sample consisted of two cohorts: (1) children who were 15 to 17 years old in 1984 ('84 cohort) and (2) children who were 15 to 17 years old in 1994 ('94 cohort).We included only white and black children in our sample, because the PSID does not have enough cases for other racial and ethnic groups. The final sample with the net worth variable consisted of 798 individuals (390 from the '84 cohort and 408 from the '94 cohort) after we deleted 50 cases with missing values in control variables. Among the 50 deleted cases, four cases had a missing value for state unemployment rate, three cases had a missing value for head-of-household's education variable, and four cases had a missing value for birth order to mother. These cases were not significantly different from those included in analyses in terms of educational attainment and parents' economic resources. The analysis sample for the liquid assets variable was slightly smaller due to 17 additional cases with missing values for this variable (n = 781--379 and 402 from each cohort).
The dependent variable was educational attainment at age 26, except for those who were 16 years old in 1994. For this group, we used educational outcomes observed at age 27 because the PSID did not conduct a survey in 2004 when they turned 26. The youngest group in our sample (15 years old in 1994) turned 26 in 2005 (the latest data available at the time of this study). Similar to previous studies (Conley, 2001; Ellwood & Kane, 2000; Morgan & Kim, 2006), we categorized educational outcomes into two dichotomous variables: (1) high school graduation (assigning 1 to those with 12 or more years of schooling or a GED and 0 to others) and (2) college enrollment (assigning 1 for at least one year of college and 0 for others).We did not analyze college graduation, because the sample size for this analysis was too small (n = 161) for the earlier cohort.
The major independent variables were economic resources during childhood. The parents' income variable was created by averaging three years of family income (1982 to 1984 for the '84 cohort and 1992 to 1994 for the '94 cohort) to account for income fluctuation (Solon, 1992). Household assets were measured in 1984 for the earlier cohort and in 1994 for the later cohort. This study used three types of household assets: liquid assets, home ownership, and net worth. Liquid assets was defined as the amount of assets easily converted into cash; it is created by subtracting the amount of unsecured debt (the sum of credit card charges, student loans, medical and legal bills, and loans from relatives) from a household's financial assets (the sum of the amount of money in saving and checking accounts, plus the total value of stocks, mutual funds, and investment trusts). We used a dichotomous measure of home ownership (with 1 for those owning a home and 0 otherwise). Net worth was defined as the total amount of wealth of a household (the sum of values of all assets less net of all liabilities): the sum of liquid assets, home equity, other real estate equity, vehicle equity, business or farm assets, and other assets. Income and assets measures were inflation-adjusted to the 2006 dollar using the Consumer Price Index.
Following previous studies (Conley, 2001; Morgan & Kim, 2006), we used the log of family income, net worth, and liquid assets in multivariate analyses to address their skewed distribution and potential nonlinear relationships with dependent variables; we assigned a value of 1 to cases with negative or zero values before we converted them into log form to prevent missing values. In separate analyses, we used a categorical measure of liquid assets: negative (less than zero), zero, modest ($1 to $3,000), and high (larger than $3,000). Analysis results using additional categorical variables with different cutoff points ($1,000, $2,000, or $10,000) produced substantively similar results to those reported in this article (results are available from the authors).
We created demographics, household characteristics, and environmental variables as control variables. Time-invariant variables (household heads race, education, and age; child's gender, age, and stepchild indicator) were created on the basis of the base-year information (1984 or 1994). Household head's race was used because the PSID does not collect race information on other household members. The birth order to mother variable was collected in 1993. Family size and the number of children were constructed with three-year averages of childhood observations (1982 to 1984 or 1992 to 1994). The female-headed family variable assigns 1 for children who lived in this type of family at any time during the observation period and 0 otherwise. Two environmental variables were created: (1) the southern origin variable (with 1 assigned to an individual who lived in a southern state in the base year and 0 otherwise) and (2) the state unemployment rate variable (the three-year average during the childhood observation period).
This study used probit regressions because the dependent variables were dichotomous. We used two dichotomous dependent variables instead of a continuous measure (for example, the number of years of education), following Mare (1981). Because overall educational attainment has increased over time, it was important to separate changes in educational distribution (structural mobility) from changes in the association between family backgrounds and educational attainment (circulation mobility) (Mare, 1981). For this reason, we estimated the effects of family background on the probability of moving into a given grade level (for example, college attendance) on the condition that an individual had completed the previous level (for example, high school graduation). That is to say, we include only high school graduates in our analysis of college attendance. In this way, our college attendance analyses were not affected by increased high school graduation rates (increased structural mobility rate) between the cohorts.
We ran two separate sets of probit regressions for each cohort. Comparison of the coefficients between the two cohorts indicates whether and how the association between parents' economic resources and children's educational attainment changed. We also ran Chow tests to see whether differences between the two cohorts were statistically significant, as summarized in equation 1 (Greene, 2003)-
Y = [alpha] + [[beta].sub.1] X + [[gamma].sub.1]C + [[gamma].sub.1]C * X [epsilon] (1)
--where Y = probability of high school graduation or of college attendance, X = a vector of independent variables, and C = cohort (0 if '84 cohort and 1 if '94 cohort).
A Chow test uses a pooled sample of both cohorts. In equation 1, the coefficients of independent variables ([[beta].sub.1]) indicate the associations between these variables and educational outcomes among the '84 cohort. The coefficients and standard errors of these variables were the same as those produced by a separate regression analysis with the '84 cohort. The coefficient of cohort variable ([[gamma].sub.1]) captured an overall difference in educational attainment between the two cohorts. That is to say, the cohort variable controlled for unobserved factors not included in our analyses, such as macroeconomic changes. The coefficient of the interaction of cohort and an independent variable ([[gamma].sub.2]) was equivalent to a difference in coefficients between the two sets of cohort analyses, indicating whether intercohort changes were statistically significant. For example, we can say that the association between liquid assets and educational outcome became significantly stronger when the coefficient of the interaction term of this variable was significantly positive (Greene, 2003).
Because children from the same family were not independently observed (Greene, 2003), our multivariate analyses adjusted for standard errors by clustering multiple children into the family unit. In both descriptive and multivariate analyses, we weighed the data with the last observed individual weight variable (Hill, 1992). We ran two sets of analyses, one with net worth and the other with liquid assets and home ownership. We ran additional analyses to check the robustness of our findings: a model using an alternative measure of parents' education (higher level of parents' education for two-parent families), models using different types of housing assets (value of home and home equity), and analyses with logit regressions. Results from these analyses were substantially the same as those reported in this article (results are available from the authors).
Sample characteristics are reported in Table 1. We compared our sample with households headed by 35- to 44-year-olds and 45- to 54-year-olds in U.S. Census Bureau reports because 89% of our sample lived in households headed by these age groups. Mean family incomes in our sample fell between average incomes of households headed by these two age groups in the U.S. Census Bureau (2008) report ($61,713 and $66,545 for 1984 and $69,703 and $79,367 for 1994). The homeownership rates in our sample were also comparable to the rates in the U.S. Census Bureau (2009a) report (68.92% and 76.46% in 1984 and 65.51% and 75.15% in 1994). The distributions of wealth demonstrated increasing inequality, as documented in previous studies (Carney & Gale, 2001): Gaps between bottom and top quartiles became wider among the '94 cohort for both net worth and liquid assets. Consistent with previous studies (Lyons, 2003), the percentage of households with negative liquid assets increased from 24% to 28%.
Descriptive statistics demonstrate improvement in educational attainment between the two cohorts. The percentage of high school graduates increased from 85% to 92%, and the percentage of individuals who attended college grew from 47% to 56%. These findings are consistent with existing statistics: The high school graduation rate increased from 85% to 88% between 1995 and 2005 among young adults (Laird, DeBell, Kienzl, & Chapman, 2007), and college attendance rates increased from 47% to 53% (authors' own calculation using U.S. Census Bureau [2009b] data). Larger increases in our study may be attributable to immigration population: Our sample does not include immigrants because of the data limitations in the PSID. The proportion of immigrants has grown rapidly, and their educational attainment is lower than that of the native population (Heckman & LaFontaine, 2007).
Multivariate Analysis Results
Continuous Measures of Economic Resources. Multivariate analysis results in Table 2 show the importance of parents' assets for high school graduation. Parents' income did not have a significant association with the child's probability of finishing high school for either cohort, but parents' assets did. The amount of net worth significantly increased chances of high school graduation among the '84 cohort, and a higher level of liquid assets did the same among the later cohort. Home ownership, however, had no significant associations for either cohort. These results are consistent with those of Zhan and Sherraden (2003), who used a sample (12 to 18 years old during the years 1992 to 1995) similar to this study's '94 cohort. Chow test results show that the associations between parents' economic resources and high school graduation did not significantly differ between the cohorts.
Because marginal effects vary with the value of independent variables in nonlinear analyses (for example,probit regressions), getting predicted probabilities is a recommended method for estimating magnitudes of the impact of independent variables on an outcome measure (Powers & Xie, 1999). Predicted probabilities of graduating high school by parents' net worth and liquid assets ranging from $0 to $30,000 are demonstrated in Figure 1. These figures indicate small intercohort changes in the associations between parents' assets and high school graduation: Intercohort differences were two percentage points or less, except at the very low levels of net worth and liquid assets.
Analysis results on college attendance in Table 2 indicate a positive relationship with income among the '94 cohort: The greater the parents' income, the more likely the child was to attend college. Both net worth and liquid assets had significant associations only among the '94 cohort, whereas homeownership had significant associations only among the '84 cohort (see Table 2). Of the different types of economic resource measures, only the amount of liquid assets showed a significant change between the two cohorts. The association between liquid assets and college attendance was strengthened at a statistically significant level, suggesting that the role of parents' liquid assets increased between the two cohorts.
Predicted probabilities shown in Figure 2 clearly indicate little change in the effect of parents' net worth on college attendance but do indicate an increase in the role of liquid assets between the two cohorts. An increase in net worth from zero to $30,000 improved a child's chance of attending college by 20 percentage points for both cohorts. The same level of change in liquid assets, however, raised college attendance probabilities by 17 percentage points among the '94 cohort but had little impact among the '84 cohort.
[FIGURE 1 OMITTED]
[FIGURE 2 OMITTED]
We ran additional analyses to test whether increased college costs explain the changed role of liquid assets, using average tuition of state universities and community colleges in the state of residence from the National Center for Education Statistics (2005). These analyses produced substantially identical results to those reported in Table 2,implying that changes in tuition were not a driving factor behind increased effects of parents' liquid assets.
Results on race are of particular interest. Among the '84 cohort, black students were more likely to graduate high school and to attend college than were white students, when other demographic characteristics and family backgrounds were controlled for. Although the average level of education was lower among black students, this appears to have been due to disadvantages in their family backgrounds. Among the '84 cohort, black students tended to attain a higher level of education than did white students with the same socioeconomic background. Among the '94 cohort, there were no significant differences in educational attainment between black and white students. Chow tests show that an intercohort change in race effects is statistically significant at the .01 level for high school graduation. These findings, along with similar results in Belley and Lochner (2007), suggest that the positive effects of being black on education have disappeared in recent years.
Other control variables did not show significant changes between the two cohorts in terms of their associations with children's educational attainment (results are not reported in Table 2 but are available from the authors). One exception is the unemployment rate, which showed a significant increase in its association with high school graduation at the .10 level in model 2.
Categorical Measure of Liquid Assets. Results with a categorical measure of liquid assets are reported in Table 3. Results show an increased effect of liquid assets on high school graduation. A difference between high and zero liquid asset was statistically significant only among the '94 cohort, and an intercohort difference was significant at the .01 level, suggesting that the advantage of having a high level of liquid assets increased between the cohorts. Results on negative liquid assets are of particular interest: Its coefficient was not significant among the '84 cohort but was significantly positive among the '94 cohort. That is to say, children from negative-asset households had the same chance of graduating high school as those from zero-asset households among the earlier cohort; among the latter cohort, children from negative-asset households had a better chance of graduating than did those from zero-asset households. An intercohort difference in negative liquid assets was significant at the .05 level. For a typical case (described in the notes of Figures 1 and 2), the probability of high school graduation for a child from a negative-asset family was about 10 percentage points higher than that for a child from a zero-asset family among the '94 cohort (96% versus 85%) but two percentage points lower among the '84 cohort (96% versus 98%).
As shown in Table 3, none of the liquid assets categories was significantly different on college attendance among the '84 cohort, whereas a high level of liquid assets had a significantly positive coefficient among the '94 cohort. However, a Chow test indicates that the intercohort difference was not significant at the .10 level (p = .117).
This study, by comparing two cohorts from PSID data, examined whether relationships between parents' economic resources and children's educational attainment have changed over time. We paid special attention to parental assets, in contrast to most existing studies that have focused solely on income (Belley & Lochner, 2007; Ellwood & Kane, 2000).
This study's findings confirm the role of parents' assets in children's educational attainment. Multivariate analyses and predicted probabilities, after controlling for parents' characteristics, show that children with a high level of net worth or liquid assets are much more likely to graduate high school and attend college than are those without assets.
This article shows that the roles of liquid assets changed between the two cohorts. The effects of the amount of liquid assets on college attendance significantly increased as shown in Table 2 and Figure 2. The analyses with a categorical measure also show that changes in the coefficient sizes of negative and high liquid assets were significant. Income, net worth, and home ownership did not show significant changes between the cohorts.
Of particular interest are changes in the role of negative liquid assets in high school graduation: Chance of high school graduation among children from negative-liquid-asset families significantly increased. This shift may reflect expanded access to the consumer credit market among the '94 cohort. Along with findings in Lyons (2003), our data show that the percentage of negative-liquid-asset households increased between the two cohorts, particularly among those at moderate income levels. Although the overall percentage of households with negative liquid assets increased by 4% between the cohorts, the percentage rose from 25% to 34% among those at the second lowest quartile of income distribution. These results suggest that increased access to the credit market among moderate-income households may have encouraged households to meet their consumption needs with unsecured debt. Access to the credit market, therefore, may have prevented children from dropping out of high school during times of economic strain. It is important to note that this explanation is hypothetical and has not been tested empirically.
Findings on race are also of interest. In the '84 cohort, black students were more likely to graduate high school and to attend college than were their white counterparts when parents' socioeconomic status was equivalent. This was not true among the '94 cohort, suggesting that black students' advantage over their white counterparts in educational attainment disappeared in later years. Increased tuition and decreased need-based financial aid may explain this intercohort change. This change may be attributed to increased gaps in the quality of education between urban and suburban areas because racial compositions differ between the two areas. Weakened affirmative action in college admission may also have contributed to this change. This finding calls for further research and policy intervention on racial disparities in education.
Our study has limitations. First, we were unable to test whether the effects of parents' assets found in this study were caused by unobserved parental characteristics or environmental factors. For instance, parents with a strong future orientation often save more and encourage children to obtain higher levels of education than do others. The PSID does not collect information on parents' attitudes, which prevents us from further investigating this issue. Second, this study was unable to control fully for environmental factors other than state unemployment rates and southern origin due to data limitations. For example, we could not fully control for macroeconomic conditions, although the two observation periods were at contrasting economic cycles (a recession in 1984 and an economic expansion in 1994). Therefore, we cannot tell whether shifts in the relationship between parents' liquid assets and educational attainment observed in this study were affected by discrete macroeconomic conditions. Third, data limitations prohibited us from examining what caused the shifts found in this study. Although we tested the role of increased college tuition, we could not assess whether an increase in the college premium in the labor market, growing economic inequality or expanded access to the credit market contributed to the enhanced role of liquid assets in college attendance. Last, we could not examine the long-term effects of parents' economic resources, because we were unable to study college graduation due to small sample size.
This article has the following implications for future research, practice, and policy development. Findings indicate the need to include assets and debts as well as income when studying the impact of parents' economic resources. As illustrated in this study, different types of parents' resources have distinct associations with children's education. Our findings suggest that policymakers may consider savings incentives and credit market accessibility as ways to narrow educational gaps. Universal child development accounts (CDAs) may be a promising policy alternative. For instance, the Maine Scholarship Grant Program provides a $500 grant to every child born in the state to be invested into a college savings plan. A statewide experiment using universal CDAs was initiated in Oklahoma in 2007 (Clancy & Sherraden, 2003). Furthermore, shifts in education financial aid policies from need-based to merit-based aid should be reevaluated. Results also call for policy interventions for racial equality. As reported earlier, black students' relative advantage over white students with comparable family backgrounds in education disappeared in later years. Considering deeply rooted racial inequality, this finding is alarming. Policymakers should address this issue by strengthening affirmative action, expanding need-based financial aid for low-income racial and ethnic minority children, and improving educational environments in urban school districts where racial and ethnic minority students are overrepresented.
Social workers can play pivotal roles in promoting equal educational opportunity. With the knowledge produced by this study, social workers may educate economically disadvantaged children and their parents about the increasing importance of saving. Social workers, especially school social workers, may encourage low-income students to make a financial plan while helping them to find financial supports for their education. Community social workers may work with local financial institutions in developing financial products for low-income families while educating members of their communities on available savings programs, such as the Savers' Credit and Individual Development Accounts. With an aim of promoting asset accumulation among low-income populations, social workers as a whole may advocate and support legislative initiatives under consideration, including the America Saving for Personal Investment, Retirement, and Education (ASPIRE) Act (H.R. 4682; S. 3577), and Young Savers accounts.
Acemoglu, D. (2002). Technical change, inequality, and the labor market. Journal of Economic Literature, 40(1), 7-72.
America Saving for Personal Investment, Retirement, and Education (ASPIRE) Act, H.R. 4682, 111th Cong. (2009).
America Saving for Personal Investment, Retirement, and Education (ASPIRE) Act, S. 3577, 111th Cong. (2009).
Belley, P., & Lochner, L. (2007). The changing role of family income and ability in determining educational achievement. Journal of Human Capital, 1, 37-89.
Card, D., & Payne, A. (2002). School finance reform, the distribution of school spending, and the distribution of student test scores. Journal of Public Economics, 83, 49-82.
Carney, S., & Gale, W. (2001). Asset accumulation among low-income households. In T. M. Shapiro & E. N. Wolff (Eds.), Assets for the poor: The benefits of spreading asset ownership (pp. 165-205). New York: Russell Sage Foundation.
Clancy, M., & Sherraden, M. (2003). The potential for inclusion in 529 savings plans: Report on a survey of states (CSD Report 03-25). St. Louis: Washington University, Center for Social Development.
Conley, D. (2001). Capital for college: Parental assets and postsecondary schooling. Sociology of Education, 74, 59-72.
Ellwood, D.T., & Kane, T.J. (2000).Who is getting a college education? Family background and the growing gaps in enrollment. In S. Danziger & J. Waldfogel (Eds.), Securing the future: Investing in children from birth to college (pp. 283-324). New York: Russell Sage Foundation.
Greene, W. H. (2003). Econometric analysis (5th ed.). Upper Saddle River, NJ: Prentice-Hall.
Gruber, J. (2001).The wealth of the unemployed. Industrial and Labor Relations Review, 55, 79-94.
Heckman, J., & LaFontaine, P.A. (2007). Web appendix: The American high school graduation rate: Trends and levels. Retrieved front http://jenni.uchicago. edu/estimating_hsgraduation/Ch1_Data_ Appendix_2008-10-23a_smt.pdf
Hill, M. S. (1992). The panel study of income dynamics:A user's guide. Newbury Park, CA: Sage Publications.
Kane, T.J. (2004). College-going and inequality. In K. M. Neckerman (Ed.), Social inequality (pp. 319-354). New York: Russell Sage Foundation.
Laird, J., DeBell, M., Kienzl, G., & Chapman, C. (2007). Dropout rates in the United States: 2005 (No. NCES 2007-059).Washington, DC: National Center for Education Statistics.
Lyons, A. C. (2003). How credit access has changed over time for U.S. households. Journal of Consumer Affairs, 37, 231-255.
Mare, R. D. (1981). Change and stability in educational stratification. American Sociological Review, 46, 72-87.
Massey, D. S. (1996).The age of extremes: Concentrated affluence and poverty in the twenty-first century. Demography, 33, 395-428.
Mayer, S. E. (2001). How did the increase in economic inequality between 1970 and 1990 affect children's educational attainment? American Journal of Sociology, 107, 1-32.
Morgan, S. L., & Kim, Y. (2006). Inequality of conditions and intergenerational mobility: Changing patterns of educational attainment in the United States. In S. L.
Morgan, D. B. Grusky, & G. S. Fields (Eds.), Mobility and inequality (pp. 165-194). Stanford, CA: Stanford University Press.
Nam, Y. (2004). Is America becoming more equal for children? Changes in the intergenerational transmission of low- and high-income status. Social Science Research, 33, 187-205.
Nam, Y., & Huang, J. (2009). Equal opportunity for all? Parental economic resources and children's educational achievement. Children and Youth Services Review, 31, 625-634.
National Association of Social Workers. (2008). Code of ethics of the National Association of Social Workers. Retrieved from http://www.naswdc.org/pubs/code/ code.asp
National Center for Education Statistics. (2005). Average undergraduate tuition and fees and room and board rates charged for full-time students in degree-granting institutions, by type and control of institution: 1964-65 through 2004-05. Retrieved from http://nces.ed.gov/ programs/digest/d05/tables/dt05_312.asp
Neckerman, K. M., & Torche, F. (2007). Inequality: Causes and consequences. Annual Review of Sociology 33, 335-357.
Powers, D. A., & Xie, Y. (1999). Statistical methods for categorical data. San Diego: Academic Press.
Ratcliffe, C., Chen, H., Williams Shanks, T. R., Nam, Y., Schreiner, M., Zhan, M., et al. (2008). Assessing asset data. In M. Sherraden & S.-M. McKernan (Eds.), Asset building and low-income families (pp. 239-270). Washington, DC: Urban Institute Press.
Rouse, C. E., & Barrow, L. (2006). U.S. elementary and secondary schools: Equalizing opportunity or replicating the status quo. Future of Children, 16, 99-123.
Solon, G. (1992). Intergenerational income mobility in the United States. American Economic Review, 82, 393-408.
U.S. Census Bureau. (2008). Historical income tables: Households. Retrieved from http://www.census.gov/ hhes/www/income/histinc/h10ar.html
U.S. Census Bureau. (2009a). Housing inventory estimates by age and family status. Retrieved from http://www. census.gov/hhes/www/housing/hvs/historic/index. html
U.S. Census Bureau. (2009b). Years of school completed by people 25 years and over, by age and sex: Selected years 1940 to 2008 (Table A-l). Retrieved from http:// www.census.gov/population/socdemo/education/ cps2008/tabA-1.xls
Wolff, E. N. (1990).Wealth holdings and poverty status in the US. Review of Income and Wealth, 36, 143-165.
Zhan, M., & Sherraden, M. (2003). Assets, expectations, and children's educational achievement in female-headed households. Social Service Review, 77, 191-211.
Yunju Nam, PhD, is assistant professor, School of Social Work, University at Buffalo, State University of New York, 685 Baldy Hall, Buffalo, NY 14260-1050; e-mail: yunjunam@buffalo. edu. Jin Huang, PhD, is assistant professor, School of Social Work, Saint Louis University. This study was supported in part by the Center for Social Development at Washington University in St. Louis. The authors are grateful to Michael Sherraden and Shanta Pandey for their thoughtful comments on earlier drafts of this article, to Julia Stevens and Eric Lundgren for their editorial help, and to Kate Koch for her research assistance.
Original manuscript received February 11, 2009
Final revision received December 1, 2009
Accepted February 28, 2010
Table 1: Characteristics of Sample, by Cohort 1984 1994 Variable Cohort Cohort Annual family income ($) (a) M (without top and bottom 64,145 74,217.48 5%) (SD) (b) (33,923) (33,846) Bottom quartile 37,839 47,710 Median 57,959 67,784 Top quartile 83,392 97,181 Net worth ($) M (without top and bottom 139,861 150,052 5%) (SD) (b) (162,074) (187,891) Bottom quartile 29,105 23,126 Median 86,055 78,899 Top quartile 188,212 206,769 % with negative net worth 5.49 6.00 % with positive net worth 91.48 90.52 Liquid assets ($) M (without top and bottom 15,671 20,073 5%) (SD) (b) (33,557) (36,037) Bottom quartile 0 0 Median 970 3,401 Top quartile 15,523 27,751 % with negative liquid assets 24.00 28.13 % with positive liquid assets 59.09 58.06 Home owner (%) 75.96 72.75 Black (%) 13.78 13.85 Female (%) (d) 40.09 46.99 Birth order to mother (M) (b) 2.62 1.93 Stepchild (%) 7.96 7.51 Child's age (M) (b) 16.10 15.89 Parent's age (M 43.93 43.55 Parent's education (M) (b) 12.58 13.26 Family size (M) (b) 4.58 4.22 Number of children (M) (d) 2.30 2.13 Female-headed household (%) (d) 17.62 24.58 Southern origin (%) 32.62 33.01 State unemployment rate (M) (b) 9.17 6.78 Sample size 390 408 (a) The means of income, net worth, and liquid assets are calculated after excluding cases at the top and bottom 5% of the distribution to remove extreme values. (b) Difference is significant at the .01 level. (c) Difference between the two cohorts is significant at the .10 level. (d) Difference is significant at the .05 level. Table 2: Probit Regressions: Parents' Economic Resources and Educational Achievement High School Graduation Model 2: Model 1: Liquid Assets Net Worth and Home '84 '94 '84 '94 Variable Cohort Cohort Cohort Cohort Income .02 -.13 .00 -.23 (.09) (.20) (.09) (.23) Net worth .06 ** .04 .05 .05 * Liquid assets .03 .06 ** .00 .05 ** Home .19 .09 .47 * .30 Black 1.13 *** -.14 (b) 1.20 *** -.20 (.29) (.37) (.29) (.36) Female .51 ** .42 * .53 ** .46 * (.20) (.22) (.20) (.24) Stepchild -.17 -.74 * -.11 -.67 (.29) (.42) (.32) (.44) Parent's age -.01 .01 -.01 .01 (.02) (.02) (.02) (.02) Parent's education .11 *** .23 *** .11 ** .20 *** (.04) (.07) (.04) (.07) # of children -.05 -.06 -.09 -.O8 (.17) (.30) (.19) (.30) Female-headed -.51 -.41 -.58 * -.35 household (.33) (.34) (.32) (.35) Unemployment Rate -.08 .07 -.10 .10 (a) (.05) (.09) (.06) (.09) N 390 408 379 402 Pseudo [R.sup.2] .17 .21 .17 .21 College Attendance Model 2: Model 1: Liquid Assets Net Worth and Home '84 '94 '84 '94 Variable Cohort Cohort Cohort Cohort Income .22 .42 ** .28 .35 * (.18) (.18) (.19) (.18) Net worth (.03) (.04) (.04) (.03) Liquid assets (.03) (.03) (.02) (.02) Home (.03) (.03) (.03) (.03) Black .56 ** .10 .55 * .12 (.29) (.27) (.29) (.27) Female .17 .13 .13 .12 (.19) (.17) (.20) (.17) Stepchild .03 -.59 -.03 -.59 (.30) (.38) (.32) (.37) Parent's age -.01 .03 * -.01 .03 (.02) (.02) (.02) (.02) Parent's education .25 *** .18 *** .24 *** .18 *** (.04) (.04) (.04) (.05) # of children .11 .41 * .13 .45 * (.17) (.24) (.18) (.25) Female-headed -.20 -.53 * -.16 -.55 * household (.33) (.31) (.34) (.32) Unemployment Rate .03 -.02 .03 .01 (.05) (.07) (.05) (.07) N 324 370 314 367 Pseudo [R.sup.2] .22 .27 .23 .28 Notes: Standard deviations are given in parentheses. The liquid asset measure has 17 cases with missing values. Accordingly, analysis sample sizes with the liquid assets measure are slightly smaller than others. The sample sizes of college attendance analyses are smaller than those of high school graduation analyses because the former include only high school graduates. We report results on control variables whose associations with education outcomes show significant changes between the two cohorts. In addition to variables listed in the table, each model include the following variables: birth order to mother, head's education, number of children in the household, living in female-headed household, and southern origin. (a) Difference between the two cohorts is significant at the .10 level. (b) Difference is significant at the .01 level. * p < .10. ** p < .05. *** p < .01. Table 3: Probit Regressions: Categorical Liquid Assets and Educational Achievement High School Graduation College Enrollment Variable '84 Cohort '94 Cohort '84 Cohort '94 Cohort Negative liquid -.19 .77 *** .40 .46 assets (.28) (.38) (.30) (.30) Modest liquid .23 .16 -.19 .44 assets (.33) (.40) (.34) (.38) High liquid .08 1.45 *** .24 .95 *** assets (.31) (.39) (.31) (.33) Sample size 379 402 314 367 Pseudo [R.sup.2] .17 .26 .24 .29 Notes: Standard deviations are given in parentheses. In addition to variables listed in the table, each model include the following variables: gender, race, birth order to mother dummy variables, stepchild indicator, head of household's age, head of household's education, household size, number of children in the household, living in female-headed household, state unemployment rate, and southern origin. Full results are available from the authors. (a) Difference is significant at the .05 level. (b) Difference is significant at the .01 level. ** p < .05, *** p < .01.
|Printer friendly Cite/link Email Feedback|
|Author:||Nam, Yunju; Huang, Jin|
|Publication:||Social Work Research|
|Date:||Dec 1, 2011|
|Previous Article:||Our best schools of social work: how good are they?|
|Next Article:||Parenting practices among depressed mothers in the child welfare system.|