Comparing the Retirement Savings of the Baby Boomers and Other Cohorts
by Sharon A. DeVaney and Sophia T. Chiremba

Originally Posted: January 24, 2005
Revision Posted: March 16, 2005

Introduction and Purpose

The baby boomers are nearing retirement or, in some cases, already have retired.¹ The baby boom cohort, which consists of persons born between 1946 and 1964, has presented challenges for the U.S. economy since its inception.² First, in the 1950s and early 1960s, more schools were needed to accommodate large numbers of baby boomer students. Later, in the late 1960s and early 1970s, as many boomers reached young adulthood, they contributed to housing shortages and increased competition for jobs. Now, as the baby boomers begin to retire, the next challenge is having enough resources for them to live comfortably during their retirement.

Opinions vary about the economic well-being of the baby boomers and how they will fare in their retirement. Scott A. Bass, for example, argues that most baby boomers are healthier, better educated, and wealthier than previous generations.³ At the same time, however, in a national study of bankruptcy conducted in 1991, Teresa A. Sullivan and her coauthors showed that half of the individuals who filed for bankruptcy protection were baby boomers.⁴ In another study, Sophie M. Korczyk noted that 8 in 10 baby boomers expect to work, at least part time, after they "retire," but only a third expect to scale back their lifestyle during their retirement years.⁵ These conflicting observations motivated this comparison of the retirement savings behavior of baby boomers with that of other cohorts.

Two theories--the life-cycle savings hypothesis⁶ and the theory of planned behavior⁷--provide a framework for examining retirement savings behavior. The life-cycle savings hypothesis assumes that a household attempts to maintain a consistent level of consumption over the lifetime of its members. To do so, many households borrow when its members are younger and their earnings are lower, and then save in anticipation of retirement when its members are in midlife and their earnings are higher. Most households reduce their savings during retirement.

The theory of planned behavior suggests that individuals are more likely to behave in a manner consistent with their intentions when they have control over the factors involved. The theory is a modification of the theory of reasoned action, which assumes that individuals form attitudes based on subjective norms. The attitudes, subjective norms, and perceived control held by a person influence how likely that person is to perform a particular behavior. Karl-Erik Warneryd argues that previous experience with savings behavior should be considered when the theory of planned behavior is applied to savings behavior.⁸ In other words, if an individual has been involved in savings behavior--such as buying a home or making regular allocations to a savings account or retirement plan--he or she will be more likely to save.

Based on these theories, the empirical model for this study is as follows: Retirement savings behavior is a function of attitude, subjective norms, perceived control, and past experience. Retirement savings behavior is measured by (a) whether or not the household has a retirement account, and (b) the amount of retirement savings.

Background for the Study

Retirement income traditionally has been viewed as a "three-legged stool" consisting of Social Security, employer pensions, and private savings. Many believe that the legs have weakened in recent years, however, and that retirees will increasingly require a fourth leg, earnings, to supplement their income during retirement. When workers were asked in the 1995 Retirement Confidence Survey which source of retirement income would be most important to them, they responded as follows: 27.5 percent said "personal savings and investments," 26.1 percent said "employer pensions" (such as a defined benefit plan), 24.7 percent said "my own contribution to a plan at work" (such as a defined contribution plan), 15.1 percent said "Social Security," and 6.6 percent had other responses.⁹ Older respondents were significantly more likely to answer that their most important source of retirement income would be either a defined benefit pension or Social Security benefits, while younger respondents were significantly more likely to answer that it would be either a defined contribution plan or their personal savings and investment.

Previous research has shown that age, education, income, and risk tolerance are positively related to retirement savings behavior.¹⁰ Older individuals, persons who have more education, those with more income, and those who have increased tolerance for risk, are likely to hold larger amounts of retirement savings. Individuals who are married and those who are white are likely to have more retirement savings than unmarried persons and members of nonwhite ethnic or racial groups. Some gender differences exist as well: According to the 1996 Retirement Confidence survey, men were more likely than women to feel in control of accumulating money for retirement (42 percent compared with 28 percent).¹¹ Also, men were more likely than women to calculate the amount of money needed for retirement (39 percent compared with 25 percent).

According to the Employee Benefit Research Institute (EBRI), the most likely participant in a pension plan is white, a male, high-earning, aged 45 to 54, and working for a public sector employer.¹² When pension plan participants were grouped by psychological characteristics, the "successful planners," who represented 21 percent of workers in the study, participated in voluntary retirement plans at much higher rates than the "live-for-today" workers, who represented 14 percent of workers. The psychological characteristics may result partly from being in a cohort.

A cohort is a group of people who share similar experiences and events. As a result, the members of a particular cohort are likely to share certain attitudes and consumer behavior.¹³ In this study, the following cohorts are examined (with the range of years in which cohort members were born shown in parentheses): the Swing cohort (1928-45), older Baby Boomers (1946-54),¹⁴ younger Baby Boomers (1955-64), and Generations X and Y (1965-87). The Swing cohort is sometimes known as the "silent generation." Members of this cohort came of age after World War II, and many were parents of baby boomers. They are often thought to be more frugal and adaptive than other cohorts.¹⁵ Older baby boomers are sometimes described as being idealistic and individualistic, while younger baby boomers have been described as emphasizing personal fulfillment.¹⁶ Members of Generation X are sometimes viewed as being skeptical consumers, while members of Generation Y have been described as having grown up with modern technology, especially computers.¹⁷ (See table 1.)

Hypotheses

Based on the theory of life-cycle savings and previous research on savings behavior, the following hypothesis was developed for this study:

• H1: Households whose heads are older (such as the Swing cohort and the Older Boomers) will be more likely to hold retirement accounts and to have larger amounts saved for retirement than households with younger heads.

Based on the theory of planned behavior and previous research on savings behavior, the following hypotheses also were developed:

• H2: Heads of households with a greater tolerance for risk, who save regularly, and those who have a longer planning horizon will be more likely to hold retirement accounts and to have larger amounts saved for retirement.



• H3: Households whose heads have more education and that have greater household income will be more likely to hold retirement accounts and to have larger amounts saved for retirement than households whose heads have less education and households with less income.



• H4: Households with children and those with self-employed heads will be less likely than households without children and those without self-employed heads to hold retirement accounts and will have smaller amounts saved for retirement.



• H5: Households that spend less than their income will be more likely to hold retirement accounts and to have larger amounts saved for retirement than households that spend the same as or more than their income.



• H6: Households in which the heads are homeowners and those that have larger amounts of assets--both financial and nonfinancial--will be more likely to hold retirement accounts and to have larger amounts saved for retirement than renters or those with smaller amounts of financial and nonfinancial assets.

Data and Sample

The data used in this study were drawn from the 2001 Survey of Consumer Finances (SCF), which was sponsored by the Federal Reserve Board of Governors and conducted by the National Organization for Research (NORC) at the University of Chicago.¹⁸ The 2001 SCF was collected using computer-assisted personal interviewing. The SCF was based on a dual-frame sample design. One set of the survey cases was selected from a standard multistage area-probability design. The other set of the survey cases (the high-income sample) was selected as a list sample from statistical records derived from tax data by the Statistics of Income Division of the Internal Revenue Service.

The sample for the study consists of households in which either the head of household or the spouse was not retired and was less than or equal to 70 years of age. The study does not include anyone who might have been turning 70½ or those who already had retired because they may have begun to withdraw funds from their retirement savings accounts. These criteria for inclusion reflect the assumption that if a respondent or spouse is still working, they are more likely to be saving for retirement. Also, individuals must begin to withdraw from qualified retirement plans on April 1 of the year in which they turn 70½ years old or when they retire, whichever comes later. These criteria reduce the sample size from 4,442 households to 3,428 households. A weight variable was used to provide descriptive statistics that are representative of the entire U.S. population.

Measurement of Variables

Dependent variables. There are two dependent variables: whether or not the head of household or spouse holds one or more retirement accounts (1 = yes, 0 = no), and the total amount in the retirement accounts held by the household. Retirement accounts are defined by the 2001 SCF to include thrift accounts (defined contribution plans), future pensions (defined benefit plans), Individual Retirement Accounts (IRAs), and Keogh accounts.

Holding at least one retirement account is examined using logistic regression.¹⁹ Logistic regression is an appropriate method when the dependent variable is dichotomous. The total amount held in the household retirement accounts is examined using tobit regression.²⁰ Tobit regression is the appropriate method when the dependent variable is equal to zero for some proportion of the observations, but their corresponding predictor variables are known.

Independent variables. The groups of independent variables include attitude, subjective norms, perceived control, and past experience. Attitude is measured by risk tolerance in making saving and investment decisions, being a saver, and planning horizon. In the SCF, risk tolerance was measured by the question, "Which of the following statements comes closest to the amount of financial risk that you are willing to take when you save or make investments?" The responses were "take substantial financial risks," "take above average financial risks," "take average financial risks," and "not willing to take any financial risk." Each response was coded as a dichotomous variable and the response "not willing to take any financial risk" was used as the reference category.

Being a saver was defined as having at least one positive response to the question, "Which of the following statements comes closest to describing you and your spouse’s (or partner’s) saving habits?" The possible responses were "save regularly by putting money aside each month," "spend regular income and save other income," "save income of one family member and spend the other," "save whatever is left over at the end of the month," or "not saving." The reference category was the "not saving" response.

Planning horizon was measured by the question, "In planning your family’s saving and spending, which of the following time periods is most important to you?" The possible responses were "the next few months," "next year," "next few years," "next 5-10 years," or "longer than 10 years." Each response was coded as a dichotomous variable with "next few months" used as the reference category.

Subjective norms were measured by age, race, and marital status. The term subjective norm is defined as individuals choosing to behave in a way that is typical of a certain group such as those of their same age, race, or marital status. Age was measured by the parameters of the four cohorts: the Swing cohort, Older Baby Boomers, Younger Baby Boomers, and Generation X and Y. The Older Baby Boomers were the reference group. Race was coded as 1 if the head of household was white and zero if the head of household was nonwhite. Similarly, regarding marital status, respondents who were married were coded 1, and those who were never married, divorced or separated, or widowed were coded zero.

Perceived control was measured by years of education, presence of children aged 18 years old or younger, household income, and self-employment. The term perceived control is defined as the factors over which the individual has control, such as the level of education that they attain or the number of children that they have. Education and income were continuous variables. Households that had children aged 18 years or younger were coded 1, while those without children in that age range were coded zero. Self-employment was coded 1 if the respondent was self-employed and zero if not.

Past experience in regard to savings behavior was measured by home ownership, financial assets, nonfinancial assets, and a question about the relationship of spending to income. Financial assets included all types of financial investments except retirement funds. Nonfinancial assets included the value of vehicles, residences, investment real estate, and equity in businesses. Spending relative to income was measured by the question, "Over the past year, would you say that your spending exceeded your family’s income, that it was about the same as your [family’s] income, or that you spent less than your [family’s] income?" Each response was coded as a dichotomous variable. Spending equal to income was the reference category. (See table 2 for the coding of variables.)

Results

Descriptive statistics. There was at least one retirement account in 57 percent of the households. The average or mean amount in the retirement accounts was $49,944, but the standard deviation was $174,193, suggesting that the dollar amount held in retirement accounts varies widely by individual households. The median amount held in retirement accounts--$2,000--provides another indication of the wide variation in the amounts held by households. (See table 3.)

Attitudinal variables. About 34 percent of households preferred not to take any risk when saving or investing, while 39 percent would take average risk, and 27 percent would take above average or substantial risk. Three-fourths of the sample reported that they saved using at least one of the saving methods described in the survey. The responses for time preference suggested that the households were more likely to prefer a longer period such as 5 to 10 years.

Subjective norms. Of the four cohorts, the Swing cohort was the smallest, representing just 14 percent of households. One reason is that the members of this cohort are older, by definition, and some of the respondents had already retired and thus were not included in the sample. Seventy-three percent of the household heads in the sample were white, and 53 percent were married.

Perceived control. On average, respondents had completed about 13½ years of education. Forty-six percent of the households had children aged 18 years or younger living in the home. Thirteen percent of the heads of household were self-employed. The average or mean household income of the sample was $72,673; the median household income was $44,000.

Past experience. Fifty-six percent of the households were homeowners. The average amount of financial assets was $101,518, and the average amount of nonfinancial assets was $250,590. Forty-four percent claimed that they had spent less than their income in the previous year, while 19 percent indicated that they had spent more than their income, and 37 percent reported that they had spent an amount equal to their income.

Analysis

Prior to conducting the regression analyses, Chi-square tests were used to determine if there was a relationship among the categorical variables and the cohorts, and Analysis of Variance (ANOVA) was used to compare the means of the continuous variables by cohorts.²¹ However, Chi-square tests and Analysis of Variance consider the effect of only one variable at a time while the regression analyses consider the effect of all of the variables simultaneously. Table 4 shows the results of the Chi-square tests, and table 5 shows the ANOVA results.

Older Baby Boomers were a little more likely to have a retirement account than the other cohorts. The heads of households in the Older Boomer cohort and in the Swing cohort were more likely to hold average or above average risk tolerance than younger Boomers and Generation X and Y. Older Boomers also were the most likely cohort to report that they are savers. Both Older and Younger Boomers had a high preference for the longest time horizon for saving. Generation X and Y were most likely to prefer the shortest time horizon which was the next few months to a year.

The Swing Cohort was more likely than the other cohorts to have a head of household who is white. The Younger Boomers were most likely to have children 18 years or younger living at home. Heads of households in the Swing cohort were more likely to be self-employed. Members of Generation X and Y and the Swing cohort were more likely to report that they spend less than their total income. Three-fourths of both the Older Boomers and the Swing cohort were homeowners, while only 30 percent of Generation X and Y were homeowners.

The ANOVA results revealed that all of the cohorts differed significantly on the total amount in their retirement accounts. Households in the Swing cohort had the largest amount in retirement savings, but it was only $14,000 more, on average, than the amount of the Older Boomers’ retirement savings. Households in the Older Boomer cohort had twice as much in their retirement accounts, on average, as the Younger Boomers. Older Boomers had attained the highest level of education. Household income was highest for the Older Boomers and lowest for the Generation X and Y cohort. The amount of financial assets was highest for the Swing cohort and lowest for the Generation X and Y cohort. Nonfinancial assets were also highest for the Swing cohort and lowest for the Generation X and Y cohort.

Logistic regression results for likelihood of having an account. When all of the factors were considered simultaneously using logistic regression, the following factors affecting the likelihood of holding a retirement account were statistically significant: risk tolerance, being a saver, planning horizon, age cohort, race, marital status, education, self-employment, spending behavior, and being a homeowner. (See table 6.) Heads of households with more tolerance for risk were more likely to hold a retirement account, as were those who reported that they saved. Those who preferred longer planning horizons (5 to 10 years and more than 10 years) were more likely to hold a retirement account. The Generation X and Y cohort were less likely to hold retirement accounts than the Older Boomer cohort.

White heads of households and those who are married were more likely to hold a retirement account than households with a nonwhite head or those in which the head is not married. Homeowners were more likely to hold retirement accounts than renters. Those who reported that they spent less than their income were more likely to hold a retirement account than those who said that they spent an amount equal to their income. Self-employed heads of households were less likely to hold a retirement account than those who were wage and salary workers. Self-employed persons are often less likely to save for retirement in tax-deferred savings options because they prefer to have access to their assets for their business. Some business owners intend to sell their businesses in order to retire.²²

Tobit regression results for amount saved in retirement accounts. The final step in the analysis was to examine the relationship between the independent variables and the amount of retirement savings. The amount in retirement savings was larger for those who were willing to take risk, those who saved, those with planning horizons of 5 to 10 years, those who were married household heads, and those with more education. Compared with the Older Boomer cohort, the Generation X and Y cohort and the Younger Boomers had less in retirement savings and the Swing cohort had more in retirement savings. Those who reported that they spent less than their income had more in retirement savings than those who reported that they spent an amount equal to their income, and homeowners had more in retirement savings than renters. The amount in financial assets and the amount in nonfinancial assets were both positively related to the amount in retirement accounts.

Discussion and Implications

The results of the Chi-square test showed that each succeeding cohort was more likely than the previous cohort to hold some type of retirement account. Generation X and Y were least likely to hold a retirement account, while the Swing cohort was most likely to hold a retirement account (45 percent compared with 73 percent). However, the logistic regression showed that when all of the factors were considered simultaneously, there was no statistical difference between the Younger and Older Boomers and the Swing cohort on the likelihood of holding a retirement account. Nevertheless, the results of the tobit regression showed that the amount of retirement savings was significantly different for the cohorts. The Younger Boomers and Generation X and Y had smaller amounts saved for retirement than the Older Boomers, but the Swing cohort had more saved than the Older Boomers. Hence, the life-cycle hypothesis that household savings tends to increase with age was supported.

The findings also support the theory of planned behavior. Retirement savings behavior was shown to be influenced by attitude, subjective norms, perceived control, and past experience. Increased tolerance for risk when saving or investing, reporting being a saver, being married, more education, being a homeowner, and reporting spending less than income were significantly related to both dependent variables--holding a retirement account and the amount saved for retirement.

This study examined the retirement savings behavior of baby boomers compared with that of other age cohorts. It found that obtaining more education, being more willing to accept risk, and enhancing past savings behavior were among the factors that were most influential in having a larger amount saved for retirement. In this study, income was marginally significant (p = .0805) in predicting the amount saved for retirement. This shows that, as hypothesized, income was positively related to the amount of retirement savings. Future studies might examine income in more detail, such as by using income quartiles. Another possibility for future studies would be to look at attitudes toward the use of credit and their influence on retirement savings behavior.

Sharon A. DeVaney, Ph.D.
Professor of Family and Consumer Economics, Department of Consumer Sciences and Retailing, Purdue University.

Sophia T. Chiremba
Ph.D. candidate, Department of Consumer Sciences and Retailing, Purdue University.

NOTE: This is a revised version of an article that originally was published in January 2005. This version corrects table 4. The heading for the second row of table 4 was corrected; it now says, "Not holding an account." In addition, the values in the second column ("Generations X and Y") of the eighth and ninth rows ("Saver" and "Not a saver") have been corrected.

Notes

1 AARP, Baby Boomers Envision Retirement II: Survey of Baby Boomers’ Expectations for Retirement (Washington, DC, AARP, May 2004); available on the Internet at http://research.aarp.org/econ/boomers_envision.pdf.

2 Cheryl Russell, The Master Trend: How the Baby Boom Generation is Remaking America (New York, Plenum Press, 1993).

3 Scott A. Bass, "Emergence of the Third Age: Toward a Productive Aging Society," in Francis G. Caro, Robert Morris, and Jill R. Norton, eds., Advancing Aging Policy as the 21st Century Begins (New York, Haworth Press, 2000).

4 Teresa A. Sullivan, Elizabeth Warren, and J. Lawrence Westbrook, The Fragile Middle Class: Americans in Debt. (New Haven, CT, Yale University Press, 2000).

5 Sophie M. Korczyk, "Baby Boomers Head for Retirement," Journal of Financial Planning, March 2001, pp. 116-23.

6 Albert Ando and Franco Modigliani, "The Life Cycle Hypothesis of Saving: Aggregate Implications and Tests," American Economic Review, March 1963, pp. 55-84.

7 Icek Ajzen, "From intentions to actions: A theory of planned behavior," in Julius Kuhl and Jürgen Beckmann, eds., Action-control: From Cognition to Behavior (Heidelberg, Germany, Springer, 1985) pp. 22-39; and Icek Ajzen, "The Theory of Planned Behavior," in Edwin A. Locke, ed., Organizational Behavior and Human Decision Processes, Volume 50, 1991, pp. 179-211.

8 Karl-Erik Warneryd, The Psychology of Saving: A Study on Economic Psychology, (Cheltenham, UK, Edward Elgar Publishing, 1999).

9 See Sharon A. DeVaney and Ya-ping Su, "Factors Predicting the Most Important Source of Retirement Income," Compensation and Working Conditions, Fall 1997, pp. 25-31.

10 See, for example, Sharon A. DeVaney and T. Catherine Zhang, "A Cohort Analysis of the Amount in Defined Contribution and Individual Retirement Accounts," Financial Counseling and Planning, Volume 12, Issue 1, 2001, pp. 89-102; and Zhan Chen and Sharon A. DeVaney, "What Factors Affect the Household Net Worth of Employees and Business Owners?" Financial Services Review, Volume 11, Number 4, 2002, pp. 381-91.

11 See Sharon A. DeVaney and Ya-ping Su, "Gender Differences in Retirement Planning Knowledge," Personal Finances and Worker Productivity, Volume 1, Issue 1, 1997, pp. 160-71.

12 Rachel Christensen, "Value of Benefits Constant in a Changing World: Findings from the EBRI/MGA Value of Benefits Survey," EBRI Notes, March 2002, pp. 1-3.

13 Geoffrey Meredith and Charles Schewe, "The Power of Cohorts," American Demographics, December 1994, pp. 22-31.

14 This study follows others in separating the baby boomers into an older and a younger cohort. These studies tend to use the end of the Vietnam War as the break point. Although the United States officially ended its involvement in the conflict in 1975, baby boomers born after 1954 were much less likely to have participated in the war or otherwise to have been affected in the same way as those born in 1954 or earlier.

15 Maurice J. Johnson and Evelyn C. Moore, Apparel Product Development, 2nd ed. (Englewood Cliffs, NJ, Prentice Hall, 2001.)

16 William Strauss and Neil Howe, Generations: The History of America’s Future, 1584 to 2069. (New York, William Morrow, 1991); and Walker J. Smith and Ann Clurman, Rocking the Ages (New York, Harper Collins, 1997).

17 Susan Mitchell, Amercan Generations: Who They Are, How They Live, What They Think, 2nd ed. (Ithaca, NY, New Strategist, 1998).

18 Arthur B. Kennickell, Codebook for 2001 Survey of Consumer Finances, 2003, (Washington, DC, Board of Governors of the Federal Reserve System, 2003).

19 For more information on logistic regression techniques, see Paul D. Allison, Logistic Regression: Using the SAS System: Theory and Application. (Cary, NC, SAS Institute and Wiley, 2001.)

20 For more information on tobit regression techniques, see Peter Kennedy, A Guide to Econometrics, 4th ed. (Cambridge, MA, MIT Press, 1998).

21 For more information on the ANOVA technique, see John Neter, William Wasserman, and Michael H. Kutner, Applied Linear Regression Models, 2nd ed. (Homewood, IL, Irwin, 1989).

22 Sharon A. DeVaney, Deanna L. Sharpe, Constance Y. Kratzer, and Ya-ping Su. "Retirement Preparation of the Nonfarm Self-Employed," Financial Counseling and Planning, Vol. 9, No. 1, 1998, pp. 53-59.