Citation
A cohort study of individual retirement savings behavior

Material Information

Title:
A cohort study of individual retirement savings behavior
Creator:
Wilson, Vickie Merriweather
Publication Date:
Language:
English
Physical Description:
124 leaves : ; 28 cm

Subjects

Subjects / Keywords:
Saving and investment -- United States ( lcsh )
Baby boom generation -- Planning -- United States ( lcsh )
Retirement -- Planning -- United States ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaves 117-124).
General Note:
School of Public Affairs
Statement of Responsibility:
by Vickie Merriweather Wilson.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
40326547 ( OCLC )
ocm40326547
Classification:
LD1190.P86 1998d .W55 ( lcc )

Full Text
A COHORT STUDY OF INDIVIDUAL RETIREMENT
SAVING BEHAVIOR
by
Vickie Merriweather Wilson
B.A., Beloit College, 1972
M.S.M., Purdue University, 1975
A thesis submitted to the
University of Colorado at Denver
in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
Public Administration
1998


This thesis for the Doctor of Philosophy
degree by
Vickie Merriweather Wilson
has been approved
by
j Marjorie B. Lewis
Minnette A. Bumpus

Date


Wilson, Vickie Wilson (Ph.D., Public Administration)
A Cohort Study of Individual Retirement Saving Behavior
Thesis directed by Assistant Professor Marjorie B. Lewis
ABSTRACT
Seventy-six million Americans, born between 1946 and 1964, will
form the next wave of retirees. These boomers, theoretically, are within the
peak years of saving for retirement. Yet, they continue to maintain low
individual saving rates.
This research focuses on the paradox of boomer retirement saving
behavior within a framework of cohort analysis. Cohort analysis theorizes
that current boomer saving behavior is the result of values and attitudes
shaped early, during the generational groups shared youthful experiences.
A study of IRA participants who are boomers, was conducted to
examine the relationships between retirement saving behavior and the
variables that measure boomer attitudes and values towards retirement
saving. Results identified core values that impact boomer saving behavior,
including self-sufficiency, certainty in the future, goal orientation and a self-
perception as a saver. Findings also indicated that neither IRA tax incentive
policy nor stock market activity, had a major influence on saving behavior.
The findings suggest that the development of effective retirement
saving policy requires a recognition of the significance of personal values
and that cohort theory provides a useful framework for identifying these
variables.
This abstract accurately represents the content of the candidates thesis,
recommend its publication.
Signed
Marjorie.'IB. Lewis
m


DEDICATION
This thesis is dedicated to my first teacher and most enduring role model,
my mother, Mrs. Viola F. Hill.


ACKNOWLEDGMENTS
It is both the journey and the destination that has made this one of
my life's most important adventures.
Let me express my appreciation to my committee chair, Dr. Marjorie
B. Lewis, for her tireless efforts to keep me focused and positive, as well as
to my other committee members: Dr. Richard J. Stillman, President John C.
Buechner, and Dr. Minnette A. Bumpus, for their expert guidance. I would
like to extend a special word of thanks to Mary L. Mohr, Esq., who served as
an outside committee member and has been both my mentor and friend.
A great many sacrifices were made so that I could reach this goal,
many of them made by the loves of my life: Myron, Alyse and Alex. I thank
each one of them for sharing the load and keeping me going, when I wanted
to stop.
My sisters, Myrna and Leah and my mother, Viola Hill, were always
there to encourage and inspire me and I will be forever indebted to my
statistical adviser and brother-in-law, Dr. Tyrone Powell.
I want to recognize the institutions that have supported me: First
Trust Corporation, for helping finance my education and supplying my data
set, the Graduate School of Public Affairs for enabling me to find my public
policy voice and UCDs Graduate Research Opportunities Program
(GROP) for the grant that assured the completion of my research.
I would also like to thank family, friends, co-workers and colleagues,
including those at First Trust, within the Bahai community and SABL, who
have kindly listened to my woes, yet, firmly encouraged me to press on.
I have prayed for the strength and the will to persevere many times
over the last four years, and I thank the Lord for giving me the support
system that made all the difference.


CONTENTS
Figures..............................................viii
Tables .............................................. ix
CHAPTER
1. INTRODUCTION...................................1
Research Problem...............................2
Purpose of the Study...........................3
Rationale for the Study .......................4
Focus and Contribution ........................4
Research Approach..............................5
2. LITERATURE REVIEW..............................7
Development of Saving Theory...................7
Retirement Saving Policy......................30
3. METHODOLOGY...................................38
Model.........................................41
Data Collection............................46
Data Analysis..............................46
Summary ...................................47
4. PRESENTATION OF RESULTS.......................48
VI


Response to Research Questions..............52
Additional Descriptive Data ................84
5. CONCLUSIONS.................................89
Significance and Limitations ...............93
Public Policy Implications..................94
Recommendations.............................96
APPENDIX
A. MEASUREMENTS OF SAVING .....................99
B. INTERVIEW GUIDE............................100
C. RETIREMENT SAVINGS SURVEY AND
COVER LETTER...............................103
D. ADDITIONAL SURVEY FINDINGS.................109
BIBLIOGRAPHY ..........................................117
vii


FIGURES
Figure
2.1 Age Specific Rates of Saving......................................19
2.2 U.S. Personal Saving Rates .......................................24
2.3 Historical View of Total IRA Contributions........................34
4.1 Principal Component Extraction....................................79
viii


TABLES
Table
2.1 Individual Retirement Account (IRA) Deductions...............33
3.1 List of Independent Variables ...............................44
4.1 Respondent Description.......................................48
4.2 Marital Status...............................................48
4.3 Household Income.............................................49
4.4 Age of Respondents ..........................................49
4.5 Years of Education ..........................................50
4.6 Number of Children ..........................................50
4.7 Years Remaining on Mortgage .................................51
4.8 Occupation...................................................51
4.9 Correlation Between Total IRA Contributions and Total
Retirement Savings .........................................53
4.10 Will Have Sufficient Retirement Income By Birth Year Range .... 56
4.11 Will Have Sufficient Retirement Income By
Household Income............................................56
4.12 Savings Will Be A Majority of Retirement
Income By Birth Year Range..................................57
4.13 Savings Will Be A Majority of Retirement
Income By Household Income..................................58
IX


4.14 Saving For Retirement is An Important Financial
Priority By Birth Year Range ................................58
4.15 Saving For Retirement Is An Important Financial
Priority By Household Income.................................59
4.16 Encouraged to Save As A Child.................................59
4.17 Saving Patterns As A Child By Birth Year Range ................60
4.18 Saving Patterns As A Child By Household Income................ 61
4.19 Saving Patterns Today By Birth Year Range......................62
4.20 Saving Patterns Today By Household Income ................... 62
4.21 Stock Market Influence By Birth Year Range.....................63
4.22 Stock Market Influence By Household Income ....................64
4.23 IRA Legislation Influence By Birth Year Range..................65
4.24 IRA Legislation Influence By Household Income..................65
4.25 Relationship Using One-Way ANOVA Test
Between SAV-Total Retirement Savings
and SUF-Level of Optimism....................................69
4.26 Relationship Using One-Way ANOVA Tests
Between SAV-Total Retirement Savings
and TOT-Total Estimated 1996 Savings ........................70
4.27 Relationship Using One-Way ANOVA Tests
Between SAV-Total Retirement Savings
and BEF-Levels of Intended Retirement Savings ...............71
4.28 Relationship Using One-Way ANOVA Tests
Between SAV-Total Retirement Savings
and CHT-Patterns of Saving Today.............................71
x


4.29 Relationship Using One-Way ANOVA Tests
Between SAV-Total Retirement Savings
and INC-Levels of Household Income...........................72
4.30 Relationship Using One-Way ANOVA Tests
Between SAV-Total Retirement Savings
and EDU-Number of Years of Schooling.........................73
4.31 Relationship Using ANOVA Tests Between
SAV-Total Retirement Savings,
SUF-Income Will be Sufficient,
MAJ-Majority of Income Will Be Savings
and PRI-Saving As Financial Priority.........................74
4.32 Relationship Using Two-Way ANOVA Tests Between
SAV-Total Retirement Savings
CHT-Saving Preferences Today
and CHA-Saving Preferences As A Child .......................75
4.33 Relationship Using Two-Way ANOVA Tests Between
SAV-Total Retirement Savings, IRA-IRA Influence
and STK-Stock Market Influence................................76
4.34 Correlation Analysis...........................................78
4.35 Factor Analysis ...............................................80
4.36 Varimax and Oblimin Rotation...................................82
4.37 Reasons For Not Saving For Retirement..........................84
4.38 Reasons For Saving For Retirement..............................85
4.39 1996 Retirement Saving ........................................86
4.40 Expected Savings Before Retirement ............................87
4.41 Work Years Before Retirement ..................................87
XI


CHAPTER 1
INTRODUCTION
There are 76 million Americans who were born between 1946 and
1964, and as the baby boomer cohort, they represent the next wave of
retirees. Most of these boomers expect to retire comfortably at or before
age sixty-five. Yet, the facts indicate that few will actually achieve their
financial goals.
The normal retirement age" of sixty-five assumes individuals will
practice a saving habit throughout their lifetime, beginning slowly in their
early years, peaking during their forties and fifties, and finally, accumulating
sufficient funds to retire. However, boomers accumulation rates are falling
consistently short. Their rates of saving are at about one-third the level
necessary for them to retire comfortably (Karpel, 1995).
This saving behavior is not adequately explained by existing
theoretical frameworks and policymakers have yet to effectively respond to
the challenges expected to result from the continuing gap between actual
boomer savings and their future retirement income requirements. This
research describes the current nature of saving behavior among the
boomer cohort, examines existing theory used to describe saving behavior
1


and introduces a theoretical framework that has the potential of identifying
predictors of boomers saving behavior.
Research Problem
In economic theorys Life Cycle Hypothesis (LCH) (Modigliani and
Brumberg, 1954), age and income are identified as key determinants of
saving behavior. Whether individuals are in their household formation
years when saving rates are low, their peak saving years when saving
rates reach their highest, or in their retirement years when saving rates
decline, age and income are assumed to be major determinants of saving
rates.
LCH was descriptive of saving behavior up until the 1970's, when
economists first began to notice the declines in the personal saving rates
among individual Americans. By the 1980's, it was apparent that while baby
boomers were entering their peak saving years, they continued to maintain
low individual saving rates. The increasing disparity between actual saving
behavior and the behavior described by economic theory, led to the
exploration of other frameworks of analysis. In that tradition, this research
examines, within a cohort context, the paradox of boomer saving behavior
by identifying their attitudes towards saving and measuring the impact on
2


saving behavior.
Purpose of the Study
The purpose of this study is to consider the saving behavior of
individuals within the boomer cohort group. Boomers are expected to have
a tremendous impact on future U.S. savings and retirement policy, but
unlike their parents, grandparents and great grandparents, their saving
behavior is not adequately described by current economic theory. This
research looks beyond economics to consider other theoretical constructs
including psychology and sociology, that offer explanatory promise. By
examining boomers as a distinct age cohort, we also have the opportunity to
identify additional variables that might explain their saving behavior.
Underlying this approach is a framework of cohort theory as explored
in the work of Dychtwald and Flower (1989), Braungart and Braungart
(1986) and Massey (1981). These authors suggest that to understand
human behavior, one must also understand the generational group with
which one identifies and the values and belief systems shared. They
indicate that these values are shaped by events that occur among the
members of the generational group during their formative years, through
age 20, and have a profound effect on their later behavior. The research is
3


designed to extend cohort analysis into an inquiry of boomer behavior, by
exploring how shared values might impact saving.
Rationale for the Study
The rationale for this study is closely aligned with the position of
ethicist, Mackie (1977) who states that there are no objective values.
Mackie suggests that values are not part of the fabric of the world, but
rather are simple attitudes and policies with regard to conduct (21). This
research considers saving as an objective value that is subject to shifts in
attitudes and belief systems and is not assumed to be morally good or
manifested as a positive value. The focus is on how the boomer cohort
group behaves relative to thrift: one of societys long-held values.
Focus and Contribution
This research advances the understanding of behavior in the area of
saving and introduces another theoretical basis for retirement saving
policies. The study also examines current retirement saving behavior
among boomers within the context of the Individual Retirement Account
(IRA), a product of 1970's and 1980's U.S. tax policy, designed to
encourage individuals to commit to long-term saving.
4


IRA contribution and total retirement savings (in dollar amounts) are
measured, as indicators of saving behavior. Data on attitudes towards
retirement saving are also gathered. Both empirical and attitudinal data are
used to expand the knowledge of saving behavior among baby boomers
and to identify their values.
By expanding the discussion of retirement saving beyond existing
economic, sociological or psychological theory, by reframing the paradox of
boomer saving within a cohort analysis framework and by identifying key
variables of retirement saving decision-making, new policy perspectives may
emerge.
Research Approach
The research is designed to introduce and measure new saving
variables and test their statistical relationships. The following are research
questions that are addressed:
Question 1: What, if any, is the correlation between IRA
contributions and total retirement savings estimates?
Question 2: What, if any, are the relationships between the
variables identified with cohort theory and those
variables traditionally identified with saving theory?
5


Question 3: What, if any, are the statistically significant differences
between the variables that have been identified as
describing boomer saving behavior?
Question 4: What, if any, are the factors that can be used to
describe boomer saving behavior?
Survey research is employed to gather data and identify pertinent
variables. This approach includes the use of self-administered mail
surveys. Based on the research questions, there are three major areas of
data collection:
Cohort Theory values, preferences and influences
Traditional Saving Theory age and income
Saving Behavior total retirement savings estimates
In addition to the survey research, another measure of saving
behavior is collected using secondary data from an existing database:
Saving Behavior IRA contributions
The data analysis tests the models variables for correlations,
statistical significance, as well as identifies and measures factors. Finally,
key findings are summarized and public policy implications are shared,
along with recommendations for future research.
6


CHAPTER 2
LITERATURE REVIEW
A review of current saving theory requires consideration of the
theoretical constructs of the past. Economics remains the most dominant
paradigm within which to explain saving behavior and is commonly
employed in the development of saving policy. However, over the last two
decades, the insufficient explanatory power of economic theory and the
resulting ineffectiveness of saving policy, has led to an effort to create a
more expansive theoretical framework.
Chapter 2 reviews the historical development of saving theory, within
economic, psychological and sociological constructs. Cohort analysis is
then introduced as a theoretical response to the saving paradox that exists
between the saving theory and boomer behavior. Current retirement saving
policy is also considered and the Individual Retirement Account is examined
as one of the most prominent examples of that policy.
Development of Savina Theory
Over the last half of this century, considerable attention has been
given to consumption and saving theory, according to Smyth (1993),
7


however, much remains to be learned. He states that, "not only do we not
have agreement on the saving function, our models have failed to provide
us with an adequate explanation of saving behavior" (89). With that
statement, Smyth articulates the problem that has preoccupied saving
theorists since the Great Depression.
In the 1930's, widespread unemployment and personal financial ruin
created an environment in which new economic theories began to redefine
the role of saving. While parsimony had historically been considered a
virtue closely tied to the Protestant work ethic, the Great Depression
changed the perspective of many Americans. Keynes (1936), in The
General Theory of Employment, Interest and Money, argued that during
depressions, high saving rates depressed aggregate demand, thereby
contributing to economic stagnation. Described as the "paradox of thrift",
Keynes also asserted that low rates of saving actually promoted aggregate
investment by stimulating demand.
The most important feature of Keynes' theory of consumer behavior
was the thesis that the extent of saving, at least in the short run, is more or
less completely determined by the size of real personal income. According
to Keynes, the sum of consumption and saving is income:
8


The fundamental psychological law, upon which we are
entitled to depend with great confidence....from our knowledge
of human nature and from the detailed facts of experience, is
that men are disposed, as a rule and on the average, to
increase their consumption as their income increases, but not
by as much as the increase in their income. (1936, 97)
Based on Keynesian theory, consumers and consumption are
assigned a positive role, and perhaps an even more important role in the
economy, than savers and saving. Within this framework, saving decisions
are assumed to be made by a process of balancing the income, saving and
consumption equation.
This theory of rational saving behavior provided not only the
underlying basis for Keynesianism, but also for what Gapinski (1993)
describes as the "wealth theoretic school." Among these theorists was
Freidman (1957), who based his saving theory on the distinction between
permanent income (discounted expected flow of income smoothed out) and
transitory (chance or unplanned) income. Within this framework
consumption and saving behavior were determined by a rational allocation
of permanent income or life resources.
Built on the theories of Freidman and other rational theorists, the Life
Cycle Hypothesis (LCH) (Modigliani and Brumberg, 1954) is probably one of
todays most widely accepted economic theories of saving. The LCH model
9


indicates that individuals formulate long-range financial plans based on a
rational decision-making process. According to the LCH model:
Individual consumption reflects the preferred allocation of
available life resources to consumption over life, while saving
and dissaving perform the function of bridging the gap
between the life-cycle of income and the desired path of
consumption, taking into account their uncertainty. In this
view, individual saving arises from hump saving, a transitory
accumulation of wealth destined for later expenditure. (1993,
249)
Modigliani and Brumberg (1954) suggest that individuals and
households formulate long-range financial plans, for three possible reasons:
1. Retirement
2. Anticipation of future reductions in income
3. Anticipation of future increases in expenditures.
This implies that there is a target level of wealth at each stage of life
and that saving is primarily a function of income and the age distribution of
the population. Modigliani and Brumberg (1954) indicate that throughout
the aging process, saving can be expected to reflect each stage of life.
Individuals in their twenties to mid-thirties are in their household formation
years, a period in which major expenditures for housing, appliances and
automobiles depress saving. As they enter their late thirties and forties, or
peak saving years", the model suggests that major purchases and financial
10


obligations have already been met. By their late fifties, individuals enter the
retirement period, and at that time, saving is again depressed.
Relying on LCH theory, Kauffmann (1993) defines saving as "the
household's decision to forgo consumption at a particular time to be able to
consume at a future date" (35). According to her, although households
may prefer a smooth consumption path, they are more often confronted with
an income and expenditure stream that are not smooth, thus the "hump-
saving" model. This model reflects saving levels that change by age group,
levels are low for the young, rise and peak during the middle years, then
become lower again among the old.
Researchers, Bryan and Farrell (1994) used the LCH model to
hypothesize that the U.S. was about to experience "an incredible surge of
savings". According to them, boomers were to soon enter the period in their
life cycle in which they were to save at dramatically higher rates. Their
incomes were to increase, liabilities decrease and they were to enter their
peak years for financial asset accumulation over the next twenty years.
Bryan and Farrell (1954) hypothesized that based on the theory of LCH and
the age trends of the American population, rational theory required that
within a relatively short period, the saving crisis would demographically
correct itself. However, recent figures indicate that no "surge of savings"
11


has materialized.
Rational economic choice theory provided a framework for the
modem understanding of saving behavior. This was a construct developed
by mathematician von Neumann and economist Morgenstern, who in 1944
wrote, Theory of Games and Economic Behavior. Their work introduced a
theory of decision-making according to the principle of maximizing
expected utility. Dawes (1988) explains the theory as:
Consider, the example, a choice between two gambles:
(a) with probability .20 win $45, otherwise nothing
(b) with probability .25 win $30, otherwise nothing
The expected value of each is equal to the probability of
winning multiplied by the amount to be won. Thus, the
expected value of gamble (a) is .20 X $45, or $9, while that of
gamble (b) is .25 x $30, or $7.50. People need not, however,
prefer gamble (a) simply because its expected value is higher.
Depending upon their circumstances they may find $30 to
have more than 4/5ths the utility of $45, in which case they
would -according the theory- choose gamble (b).
For example, an individual may be out of money at the end of
a week and simply desire to have enough money to eat until
the following Monday. In that situation, the individual may find
the difference in utility between $30 and $35 to be negligible
compared to the difference between 1/4th and a 1 /5th chance.
(1988,11)
According to Dawes (1988), an individual who prefers (a) to (b)
violates the von Neumann and Morgenstern (1944) principle of choosing
according to utility, since according to their theory, rationality is a necessary
12


condition for decision-making. Dawes (1988) defines a rational choice as
one meeting three criteria:
1. Based on the decision makers current assets.
2. Based on the possible consequences of the choice.
3. When these consequences are uncertain, their likelihood is
evaluated without violating the basic rules of probability theory.
However, researchers including Dawes (1988), have questioned
whether it is possible to be rational and make choices on a somewhat
different basis, so long as that person bases the choice on the amount to be
won (possible consequences) and its probability. Psychologists and
behavioralists have been quick to point out that individuals tend to violate
the principle of maximizing expected utility all the time, and are often
patently irrational.
It has been suggested by Bernheim (1991) that perhaps rational
economic theory needs to take into account the role that demographic
variables, personal experience or cultural perspective play in determining
saving behavior. Bernheim finds rational economic saving theory to be
limited and incomplete and makes the case for the inclusion of additional
variables to supplement LCH's explanations of saving behavior:
13


Someone who has lived through the Great Depression may
well be less optimistic, and hence more conservative, than
someone who has only learned about it in school. First hand
experience may change the way information is framed and
may alter the way it is processed. This is not to say that either
the optimist or the pessimist is irrational. On the contrary, the
point is that expectations arise in part from something beyond
an objective notion of rationality. Thus, even if the life-cycle
hypothesis is valid, the psychological determinants of
expectations may be more important than the economic and
demographic ones discussed. (1991, 70)
Proponents of the psychological theory of saving reject the economic
notion that most individuals rationally formulate complex, long-range
financial plans. They suggest that individuals employ a primarily subjective
approach to the decision-making process, deliberately avoiding temptation,
restricting future choices and exercising self-control by mentally separating
their resources into a number of different accounts, some which are
psychologically easier to spend than others.
Psychologists, Kahneman and Tversky (1982) describe their theory
of saving as a process of "mental accounting":
Imagine yourself on your way to a Broadway play with a pair of
tickets for which you have paid $40. On entering the theater
you discover that you have lost the tickets. Would you pay
$40 for another pair of tickets? Now imagine you are on your
way to the same play without having bought the tickets. On
entering the theater you realize that you have lost $40 in cash.
Would you now buy tickets to the play? (1982,160)
According to them, the two situations are identical, in both cases the
14


individual is $40 poorer than they were earlier. However, most people when
presented with the two situations interpret them differently, because they
tend to assign the two losses into different mental accounts. The loss of
$40 cash is mentally posted to "general resources" and therefore, is seen to
have little impact on the decision to purchase the tickets. But the $40 loss
of tickets is posted mentally to the "theater account" and is perceived as
doubling the cost of the ticket.
Parallels can be drawn in describing the way individuals assign
wealth. Rather than being defined by a rational long-term scheme,
individuals may proceed through a mental exercise of assigning and
reassigning, depositing and withdrawing, based on the perceptions of how
they have defined their wealth. The two researchers cite theories in the
psychology of preferences that describe discrepancies between subjective
and objective concepts of decisions. They indicate that theories of rational
choice do not consider the difficulty people have in "maintaining a
comprehensive view of consequences, and that their susceptibility to the
vagaries of framing are examples of such impediments" (72). Perhaps this
explains why boomers continue to postpone saving, even in the face of dire
consequences.
Shefrin and Thaler (1992) used Kahneman and Tversky's research
15


to modify the LCH model and to bridge the rational and psychological
approaches to saving behavior. They incorporated three behavioral
features into the LCH analysis:
1. Self-control the tradeoff between immediate gratification and
long-run benefits, creating a conflict that is not present in a
choice between one option and another.
2. Mental Accounting the process of assigning income and
saving to various accounts.
3. Framing the way in which choices of saving and consuming
are presented.
Self-control is a behavioral term intended to describe the inherent
conflict between the rational and the emotional aspects of saving. Within
the context of self-control, they identify three key elements-internal conflict,
temptation and will power. The researchers imply that the process of
mental accounting can be employed by establishing mental rules that
facilitate the division of wealth into three components: current assets, future
income and current spendable income. For instance, when a saving
account or pension plan is bounded by rules concerning withdrawal and
consumption, it becomes relatively simple for an individual to frame the
wealth components for long-term saving. The policy implications of framing
16


may point to the need to develop saving strategies for boomers that
recognize the importance of the psychological decision-making process that
determines saving, and to develop mechanisms that accommodates it.
Psychological models have helped expand saving theory beyond a
rational economic framework, and to focus on the subjective decision-
making process. They also led to the development of sociological theory to
explain saving behavior. Olander and Seipel (1970) are psychologists who
developed a sociological model that focused on the "sociodemographic
features" of saving behavior. They indicate that these variables must be
considered, because to a large extent they characterize the social
environment and "thus the stimulus situation in which the individual finds
himself (51).
According to Olander and Seipel (1970), the perception of the
environment and of the social groups to which an individual belongs, as well
as age, education and family, can well play a crucial role in the decision-
making process. Their research also confirms economic theory that
suggests that demographic variables such as occupation, education, age
and life cycle are strongly correlated with each other and are closely
associated with the income and wealth variables.
Because Olander and Seipel (1970) conducted their research among
17


both American and Swedish households, they were, perhaps, more aware
of the impact that the subjective feeling of belonging to a particular social or
cultural group might have on saving behavior. They postulate a tendency
for an individual to compare his or her standard of living and consumption
with that of those who are regarded as a reference group, so that the saving
habits of this group are determined to be the meaningful standard of
comparison.
Bernheim (1991) extends Olander and Seipel's (1970) theory to draw
further implications. He suggests that depending on how strong the group
ties are, individuals may use others in their reference group as the standard
for their saving habits. Habitual rules of saving, such as contributing a
percent of earnings, are thereby, supported by social values. Bernheim
(1991) supports this assertion by describing the saving behavior of the
Japanese during the 1970s.
Bernheim (1991) explains that right after World War II, the Japanese
saving rate was dismally low, much lower than the U.S. Having been
defeated and demoralized by the war, the Japanese government realized
that higher saving rates were needed to rebuild and restore confidence in
the economy. The government undertook a comprehensive redevelopment
program that emphasized saving as a virtue and an act of national solidarity.
18


Saving Kale (%)
The government mobilized every facet of Japanese society in support of
saving: local community groups, religious groups, schools and employers.
The government also supported saving efforts through tax incentives.
It was this sustained, all-encompassing: economic, psychological and
sociological assault, that resulted in the 21 percent averages that were
experienced during the 1970s (Figure 2.1).
FIGURE 2.1
AGE-SPECIFIC RATES OF SAVING FOR
U.S. AND JAPANESE HOUSEHOLDS
U S 1972-73 -*-Japan, 1974 j
Less than 2*3 25-34 35-44 45-54 55-64 54 3, Up
Age of Famly Head
Source: Hayashi, Fumio (1986). Why Is Japans Saving Rate So
Apparently High? NBER Macroeonomics Annual. Cambridge,
MA: MIT Press.
19


But it was difficult to maintain such an all-out effort and as the
emphasis on saving lessened, Japanese saving rates began to decline,
falling to approximately 12 percent by the 1990s.
If boomers use themselves as the reference group" with which to
determine their saving decisions, it suggests that perhaps cultural
explanations may be useful in describing their behavior. Cohort group
theory, based the generational theories of Dychtwald and Flower (1989),
Braungart and Braungart (1986) and Massey (1981) suggest that to
understand human behavior, one must also understand the age group with
which one identifies and the values and belief systems shared.
Dychtwald and Flower (1989) describes what he calls the "cohort
effect", in which each generation or "cohort" tends to form its own unique
point of view with regard to key preferences, including social values during
the first 15 to 20 years of life, when basic adult beliefs and goals are being
developed and tested.
These patterns of generational values and preferences are
illustrative of what sociologists call the "cohort effect", a useful
system for understanding the various likes and dislikes of each
unique generation group as it migrates from youth to old
age"....For instance, today's 65+ consumers were deeply
influenced in their youth by the terribly hard and financially
frightening times of the depression. Their point of view is
"Save, save, save....The next generation of elders was
influenced by the great prosperity that followed World War II,
their point of view, therefore is a blend: Save some, spend
20


some (Dychtwald and Flower, 1989, 278).
According to Dychtwald and Flower (1989), the baby boomer cohort
shares very different attitudes towards saving and spending. Having been
fully immersed in the affluent decades, "they have made a complete about-
face from the traditional ethic of saving, self-sacrifice, and delayed
gratification. They have learned to spend their money fast and to borrow
rather than save" (277). Boomers retirement saving patterns appear to be
yet another example of this attitude.
Massey (1981) also provides descriptions of the cohort groups shifts
in attitudes. He describes the following four age clusters":
1. Traditionalists born during the early part of the century, the
experience of the Great Depression caused them to be very
concerned about the future and financial security.
2. In-Between (Silent) Generation they were shaped by the war
effort, and later, the fear of nuclear war fueled a sense of live
for now. These values reflected a shift away from the
conservatism of the previous generation.
3. Rebels (Baby Boomers) values were shaped amidst a wave
of affluence and were influenced by parents who felt that my
kids are going to have it better than I did. This generation was
21


characterized as overindulged" and was programmed to
rebel, rejecting core values and questioning accepted norms.
4. Synthesis Generation -the children of boomers, they tend to
be more conservative and are worried about their futures.
Having been shaped by the economic slowdown of the 1970's
and 1980's, these individuals feel that they may not be able to
do financially as well as their parents.
In response to shifts in generational influences, the cohort groups
have also exhibited changes in behavior. Dychtwald and Flower (1989)
identify examples of boomer behaviors that distinguish them from other
cohorts, including:
1. Delaying marriage, or not getting married at all
Among American women, the average age for first marriages
has risen to 23 years old, its highest point in the nation's
history and among men it 26 years old, the highest since
1900 (98).
2. Postponing having children, or remaining childless
The present number of children American women would have
in their life hit 1.8, less than half the rate of their mothers, and
27 percent lower than the rate of 1970the lowest rate ever
22


recorded in America (11).
3. Divorcing and remarrying at a higher rate
Currently, more than 50 percent of all marriages end in
divorce, a percentage that reflects a 43 percent increase since
1970. Eighty percent of divorced people remarry (98).
4. Changing jobs more often
The average American now changes jobs every three years
(98).
5. Having a lower ratio of children to parents
In 1900 there were 13.6 adults between the ages of 18 and
64 for every person 65 years and older. By 1990, the ratio
dropped to 4.8 to 1. This trend will leave fewer elderly in the
care of their children and more in the care of others (239).
The traditional life cycle model does not adequately describe these
boomer behaviors, and although shifts in social values appear to have had a
profound effect on saving behavior, U.S. policies developed to address the
saving habits of Americans have largely ignored these changes and
therefore, have not been effective in resolving the nations saving crisis.
The severity of the crisis is illustrated in Figure 2.2, U.S. personal
saving rates fell to a low of 3.8 percent in 1997 (Bureau of Economic
23


Analysis, 1988). Saving rates for American households were lower than
during any comparable period in U.S. history (Magnusson, 1994). And
although this trend was clearly not confined to the U.S., as Bemheim (1991)
points out, "the U S. does have the dubious distinction of leading the way"
(5).
FIGURE 2.2
U.S. PERSONAL SAVING RATES
20%
0.0% ------------------------------------------------------------
1970 1975 1980 1985 1990 1995
Year
Source: Bureau of Economic Analysis (1998). National Income and
Product Accounts. U.S. Department of Commerce.
It is within this national context that individual savings for retirement
must be considered. Retirement savings has been metaphorically
24


described by Karpel (1995) as one leg of a "three-legged stool":
The first leg is Social Security and government welfare programs for
the elderly; the second, employer- or labor union-provided pensions;
and the third, direct individual savings (58).
This is particularly descriptive in a time in which future Social Security
payments are in doubt, the number of privately-funded pensions is shrinking
and individual saving rates are at an all-time low. As Karpel (1995) puts it:
The wrap-up on the "three-legged stool" of Social Security,
pensions, and savings: Leg number one is going to be partially
amputated, leg number two has osteoporosis, and leg number
three is one-third long enough to reach the ground (58).
Because individual personal savings is the critical bridge between
insufficient Social Security and vanishing pension benefits. It is alarming to
learn just how wide the savings gap is for those Americans anticipating
retirement.
The gap is best described in the results of a number of studies that
have examined savings behavior. For example, Peterson (1996), in his
book Will America Grow Up Before It Grows Old?, reports that, according to
the Federal Reserve Board, 43 percent of U.S. families spent more than
their income and only 30 percent accumulated assets for long-term savings.
Karpel (1995) describes a 1993 Merrill Lynch-sponsored survey that
considered the saving behavior of Americans. Among boomers, the
findings were that 74 percent of the females and 57 of males were saving
25


nothing for retirement and 15 percent of all boomers said they were saving
nothing for any purpose. Karpel (1995) also mentions a 1993 Yankelovich
survey of nearly 1,500 men and women between the age of thirty and fifty
with household incomes of $30,000 or more found that while 81 percent
were confident that they would have enough income to retire, their median
savings were only at $2,600 per year.
Karpel (1995) highlights a 1993 survey conducted by the Employers
Council on Flexible Compensation, that found that the median amount
employees are saving for retirement was $720 a year. The highest amounts
saved were among forty-five to sixty-four year olds, who averaged $2,408
per year. According to the survey, it was most commonly reported among
employees that they were not saving anything towards retirement (Karpel,
1995, 53).
Bernheim (1993), in Saving for Prosperity: The U.S. Saving Rate,
puts the research results into perspective by comparing calculated saving
targets to the amounts that Merrill Lynch survey participants had managed
to save. He found that based on the savings targets calculated by his
model, that on average, participants had saved only one-third of what they
would need to retire at the age of sixty-five. Interestingly, it appears that
the more the survey participants were earning, the worse their rate of
26


saving.
As Bernheim (1993) describes it, couples with total earnings of
$40,000 to $60,000 who have pension coverage have saved an average of
$12,950 only 40% of what they should have accumulated by age thirty-
five and forty-five and without pensions the same couples have only 25
percent. Couples earning $60,000 to $100,000 with pensions have an
average accumulation of $17,100 or 34 percent of the amount that should
have been accumulated and without pensions, these couples have about 23
percent (Karpel, 1995, 51-53). Based on Bernheim's (1993) analysis,
couples who earn a total of $75,000 before taxes would have to begin
saving between 8.9 percent and 13.0 percent of their after-tax income to
be on track to accumulate enough to retire at sixty-five, assuming there
were no cuts in Social Security.
While these recent studies confirm that national saving rates are at
an all-time low, and that individual retirement saving rates fall considerably
short of their target, they offer little to identify the variables that determine
current saving behavior. An insufficient theoretical framework has resulted
in policy programs that have not effectively responded to declining saving
rates.
Saving behavior, particularly among baby boomers, is defined by
27


policymakers in terms of values. For instance, during January 1995, the
U.S. House of Representatives Ways and Means Committee held hearings
on the American Dream Saving Account, a proposal to reinstate a fully tax
deductible Individual Retirement Account (IRA). During the course of the
hearings, there was general agreement that while the IRA was an important
policy response to saving rate declines, there was a more fundamental
issue that must be addressed.
This was described by Representative Jim McCrery of Louisiana as
the need to create a culture of saving among Americans. McCrery
explained that to increase saving rates by altering tax incentives was simply
not enough, Americans must change their saving behavior through changes
in values and attitudes.
McCrery's comments harken back to a time when parsimony or
thriftiness was a highly regarded virtue, reflecting sacrifice, prudence,
responsibility and discipline. Longman (1996) reminds us of this period in
his book, The Return of Thrift, when he quotes Samuel Smiles, an 1870's
moralist:
It is the savings of the individual which compose the wealth -
in other words, the well-being of every nation...On the other
hand, it is the wastefulness of individuals that occasions the
impoverishment of states. So it is that every thrifty person
may be regarded as a public benefactor, and every thriftless
person as a public enemy (23).
28


Longman (1996) suggests that thrift was not only a virtue in itself, but
was a requisite of civilization and nation building" (23). Yankelovich (1994)
confirms that saving was once a "core value", or a traditional norm (50).
While it was recognized that the saving habit came easier to some than to
others, this was a habit everyone was encouraged to establish, regardless
of age, income or social standing. Saving was both a social value and a
means to a better end. The more one saved, the more one was able to
improve his or her financial position and attain personal goals. The act of
saving was supported and reinforced by the family, the school, the
community, the local financial institutions and was reflected in spiritual
values, embedded within the Protestant work ethic.
Interestingly, saving for retirement is a somewhat recent
phenomena, since "mass retirement" or what Longman (1996) calls "the
assumption that the average person will be able to stop working and turn to
a life of leisure" is a relatively new tradition (31). He dates the widespread
expectation of retirement to 1940, the year Social Security participants
began receiving their first benefit payments. Yankelovich (1994) supports
this, asserting that retirement expectations have also been a direct result of
the "affluence affect", referring to the dramatic affect that affluence has had
on the values of Americans since the end of World War II (18).
29


There is also evidence that retirement saving patterns differ from
other saving behaviors. Although personal saving rates were higher
(ranging between 6.6 percent to 9.3 percent) during the 1970's, current
data does not support the idea that todays retirees were effective savers for
retirement. Data from the EBRI Databook on Employee Benefits
(McDonnell, et al., 1997) on the older populations income indicates that for
those now ages 65 years and over, in 1996 the median total income was
only $11,553 and the mean total income was $17,128 (58). Among their
sources of income in 1995, the elderly listed Social Security (43 percent),
pensions and annuities (19 percent), earnings (18 percent) and income and
assets (18 percent). The 1996 median income generated investments,
including interest, dividends, rents, royalties, estates and trusts amounted to
only $1,216 and the mean income was $3,057 (57).
This suggests that perhaps because of the widespread availability of
Social Security and pension payments, saving for retirement was not as
much of a priority for yesterday's workers as it will be for baby boomers.
Retirement Saving Policy
An example of retirement saving policy is the Individual Retirement
Account (IRA). It was originally proposed as a tax incentive to encourage
30


hardworking Americans to save towards their own retirement. Twenty-five
years ago, when President Richard Nixon first proposed the legislation, it
was an attempt to tap into old-fashioned "core values". Schultz (1992) cites
Nixon's message to Congress on December 8, 1971:
Self-reliance, prudence, and independence are qualities which
our government should work to encourage among our people.
These are also qualities which are involved when a person
chooses to invest in a retirement savings plan, setting aside
money today so that he will have a greater security tomorrow.
In this respect pension plans are a direct expression of some
of the best elements in the American character. Public policy
should be designed to reward and reinforce these qualities.
(224)
Enacted in 1974 and liberalized in 1981, the IRA legislation allowed
every U.S. worker to save and invest in his or her own pension plan. The
following were the retirement plan's primary features:
Retirement savings account could be established at banks, savings
and loans, credit unions, brokerage firms, mutual fund companies
and other types of financial institutions as pension plans for the
benefit of an individual saver. The saver would have discretion over
how the savings would be invested.
IRA participants could save up to $2,000 of their earned income and
deduct the amount from their annual taxable income. Non-working
31


spouses could save and deduct up to $250 each year. All income
and capital gains generated within the IRA would be tax sheltered.
IRA participants could withdraw all or a portion of their savings
without penalty after age 59 and a half, and would be taxed at the
participants current income tax rate. Early withdrawals would result
in a 10 percent penalty and be taxed at the current income tax rate.
The IRA was considered a highly innovative response to increased
retirement savings during its heyday (1981-1986), and was particularly
popular with upper and middle income workers. But the IRA was not
designed for everyone; as a tax incentive, in fact, the highest rate of
participation among American workers ever achieved was, according to the
EBRI, only about 15.9 percent in 1985 (Table 2.1). By 1985, 16.2 million
Americans had established an IRA, but participation was primarily among
the middle and upper income (McDonnell, et al, 1997,132).
In 1986, the IRA legislation fell victim to new tax initiatives, when
Congress "raided the IRA program to fund the Tax Reform Act".
(Magnusson, 1994, 106). Eligibility for tax deductibility was now
determined by income level and by whether or not the worker and/or spouse
32


participated in an existing pension plan. Contributions declined sharply.
Only 3.7 percent of all taxpayers contributed to an IRA in 1994, a 73 percent
drop from the high in 1985.
TABLE 2.1
INDIVIDUAL RETIREMENT ACCOUNT (IRA) DEDUCTIONS
Federal Income Tax Returns
1975-1994
Number of reporting
units claiming an Total Amount
IRA deduction percentage claimed
Tax Year (millions) of all returns ($ billions)
1975 1.2 1.5% $ 1.4
1980 2.6 2.7 3.4
1981 3.4 3.6 4.8
1982 12.0 12.6 28.3
1983 13.6 14.1 32.1
1984 15.2 15.3 35.4
1985 16.2 15.9 38.2
1986 15.5 15.1 37.8
1987 7.3 6.8 14.1
1988 6.4 5.8 11.9
1989 5.8 5.2 10.8
1990 5.2 4.6 9.9
1991 4.7 4.1 9.0
1992 4.5 4.0 8.7
1993 4.4 3.8 8.5
1994 4.3 3.7 8.4
Source: McDonnell, Ken, Paul Fronstin, Kelly Olsen, Pamela Ostuw,
Jack VanDerhei, and Paul Yakoboski (1997). EBRI Databook
on Employee Benefits, Fourth Edition. Washington, D.C.:
Employee Benefit Research Institute
33


More importantly, even among those households that did not have a
pension or had sufficiently low income and continued to be eligible to take
the full deduction contributions fell by almost 26 percent from 1987 to 1996
(Figure 2.3).
FIGURE 2.3
HISTORICAL VIEW OF TOTAL IRA CONTRIBUTIONS
S160 -
SQILUCNS
$12 0
S2Q ^
* J
1933 1934 1995 199E
Deductible
Non Deductible
0 Total
Source: Norquist, Jennifer M. (1997). New IRS Statistics." IRA
Reporter. Brainerd, MN: Universal Pensions. August.
The Taxpayer Relief Act was introduced in 1997, as an attempt to
make the IRA a more attractive savings vehicle. The Act introduced the
Roth IRA, which offers additional features:
34


expands non-working spouses ability to make tax-deductible
contributions
increases income eligibility limits for tax-deductible
Contributions
expands penalty-free distribution options
allows earnings to grow tax-free
For those participants who maintain their contributions in their IRA for
five years or more, distributions can be penalty and tax free. Penalty-free
distributions are also available in the event of first-time home purchases,
higher education or catastrophic illness. Industry experts are uncertain as
to whether these improvements in the features of the IRA, will increase
retirement saving rates, however, there are also other reasons for the
decline in IRA participation.
For instance, IRAs often compete for the same dollars as salary
reduction plans, such as the 401 (k), and in those plans the contributions are
fully tax deductible and often offer an employer match, making it more
attractive for employees to contribute. Participation in salary reduction
plans are more convenient, being offered through the work place and using
automatic payroll deductions. Employers tend to promote salary reduction
plans directly to employees, as well as, to provide saving and investment
35


education.
The number of 401 (k) plans has increased by 900 percent since
1984, from 17,303 to 154,527 and now include 52 percent of all active
private pension plan participants (McDonnell, et al., 116). These trends
indicate that individuals who once participated in traditional defined benefit
plans, now are more likely to be in salary reduction plans such as 401 (k)s.
However, economists Poterba, Venti, and Wise (1996), question, in
light of personal saving rate declines, whether contributions to either the IRA
or 401 (k) contributions are actually net additions to saving. Engen, Gale
and Scholz (1996), also consider how much, if any, of the overall
contributions to existing saving incentives have raised saving. The
economists contend that the additional saving in those areas of personal
finance have been offset by significant spending or borrowing from other
areas. According to them, the phenomena, called the substitution effect,
describes how individuals' retirement saving through tax-favored plans is
offset to some extent by decreased saving or increased indebtedness.
Even during the period in which increased individual retirement
saving was defined as a major policy objective, and IRAs were pronounced
a roaring success, U.S. saving rates declined (Nocera, 1994). According to
some researchers, the IRA tax incentives, instead of encouraging all
36


Americans to increase saving, may actually have encouraged the wealthy
and middle-class to shift their savings from traditional savings accounts to
IRAs (Canterberry, 1993) and (Nocera, 1994). They claim that rather than
stemming saving rate declines, the IRA's popularity simply precipitated a
major shift in private capital, evidenced by fund transfers away from the
banks and savings and loans into the mutual fund and equity markets.
At its best, most would agree that the IRA experience has been
mixed. The results support the notion that traditional saving policy as
defined, in this instance, by tax incentives, has not been sufficient to spur
retirement saving. This suggests that there may be key determinants of
saving behavior that have not been addressed by the IRA.
37


CHAPTER 3
METHODOLOGY
Traditionally, saving behavior has been examined within a rational
economic framework, in which income, consumption, age and the
propensity for long-term rational planning were key variables. More
recently, this framework has been broadened to accommodate
psychological variables including the propensity to assign wealth, to
exercise self-control and to frame choices between saving and not saving.
In addition, sociological variables have also been introduced, such as the
propensity to identify with social values and compare oneself with a
particular reference group.
This dissertation introduces a cultural construct, described as cohort
group analysis. The approach draws upon the theories of Dychtwald and
Flower (1989), Braungart and Braungart (1986) and Massey (1981) who
suggest that to understand human behavior, one must understand the age
group with which one identifies and the values and attitudes shared. The
research is designed to expand the understanding of the impact that baby
boomers values and attitudes have on saving behavior.
While not providing comprehensive answers to the critical questions
38


concerning saving behavior, this study does introduce new variables, offers
insights into key relationships and identifies potential methods for further
research. The following are questions that are addressed by the research:
Question 1: What, if any, is the correlation between IRA
Question 2: contributions and total retirement savings estimates? What, if any, are the relationships between the variables identified with cohort theory and those variables traditionally identified with saving theory?
Question 3: What, if any, are the statistically significant differences between the variables that have been identified as describing boomer saving behavior?
Question 4: What, if any, are the factors that can be used to describe boomer saving behavior?
Survey research was employed to gather data and identify pertinent
variables. This design included the use of self-administered mail surveys.
Such an approach is particularly useful for collecting original data on
attitudes of a large population (Babbie, 1994). Implementation of the
design results in descriptive, as well as quantifiable data.
Based on the research questions, the following data was collected:
Cohort Theory values, preferences and influences
39


Economic Saving Theory age and income
In addition, two sets of data were collected as a proxy for retirement
saving behavior:
Saving Behavior total IRA contributions
total retirement saving estimates
Total IRA contributions were collected using secondary data from an
existing database and total retirement saving estimates were identified from
the survey questionnaire. By comparing each set of data, an analysis was
conducted to examine relationships and to identify the portion of IRA
contributions that was within total retirement savings.
The primary data was gathered from a sample taken from a
population of IRA accounts held at First Trust Corporation of Denver,
Colorado. First Trust is the largest independent trust company in the
United States and administers approximately 300,000 IRA accounts, with
participants throughout the United States.
Through an automated process in which the computer program
selected of every nth account, 2,000 IRA participants who were born
between 1946 and 1964 were identified. The secondary data was collected
from those respondents who returned completed survey questionnaires.
Once received, the surveys were used to identify respondents total IRA
40


contribution levels.
The sample was not chosen to be representative of the general U.S.
population, but was determined to be representative of a segment of the
population who had exhibited retirement saving behavior. From previously
collected First Trust data, it was inferred that this population had a higher
than average household incomes, saved at higher than average levels and
tended to be better informed regarding retirement saving issues, than the
general population.
Model
To gather the data, a model based on cohort theory was developed
to compare and contrast the significance of key indicators of saving
behavior. According to the model, there is a relationship between boomer
retirement saving behaviors (Y = dependent variable) and variables that
predict preferences, values and influences towards savings (X independent
variables):
The survey questionnaire was developed to capture the independent
variables. A group of First Trust employees (6), who were also baby
boomers, were selected by the author to participate in series of personal
interviews. This group sample, while not randomly representative, was
41


selected to include both men and women, black and white, single, married
and divorced, with and without children, college and high school-educated,
ranging in age from 32 to 48 years old and with household incomes ranging
from $26,000 to $100,000. The one-on-one interviews presented broad
questions intended to elicit discussions of values, preferences and key
influences with regard to retirement savings (see Appendix B for the list of
questions). Based on interview results, questions were identified that were
to be tested.
By using the results of the series of personal interviews, a
questionnaire was designed to gather factual and attitudinal data (see
Appendix C). An ordinal scale was used to measure levels of optimism
about achieving financial security, levels of conviction concerning personal
priorities and levels of certainty of the importance of saving:
SUF = Level of optimism of saving sufficiency
PRI = Level of conviction of saving as a financial priority
MAJ = Level of certainty of importance of saving for retirement
The influences of IRA tax policy and stock market changes were
also measured:
STK = Level of influence of stock market
IRA = Level of influence of IRA tax legislation
42


Preferences toward saving or not saving were measured and compared:
CHA = Patterns of saving as a child
CHT = Patterns of saving as an adult
Respondents were also asked to describe their retirement saving
behavior and to identify the reasoning behind their decisions to save or not
to save. Relationship and media influences were also explored, as was the
impact of preferences on saving behavior. Expectations were discussed to
identify what respondents planned to have saved at retirement, what income
sources they thought would then be available to them and the number of
years that they anticipated working. Demographic questions were used to
confirm birth years and to identify income, total retirement savings, marital,
family and occupational data (Table 3.1).
Data was collected using close-ended questions to facilitate
comparisons among respondents. Additional questions were designed to
gather descriptive and demographic data associated with saving behavior.
A pretest was conducted on a small, sample (25) of computer
selected (every nth account) respondents, to reduce error and improve the
instrument's readability. Fixed responses were used to provide clarity and
to increase the response rate. The "other" response option was provided, to
avoid constraining responses.
43


TABLE 3.1
LIST OF INDEPENDENT VARIABLES
Values, attitudes and judgements towards saving:
SUF level of optimism that income will be sufficient at or before retirement age
MAJ levels of certainty that majority of retirement income will come from savings
PRI levels of conviction that saving for retirement is an important financial priority
Description of reasoning used to decide not to save for retirement:
NAV determination that funds are unavailable
OLD intention to save when older
PER expectation of sufficient company pension
SSI expectation of sufficient Social Security
SFD expectation of already having sufficient savings
INH expectation of inheritance
WRK- expectation of not retiring
Description of reasoning used to decide to save for retirement:
AVA determination that funds for saving are available
FLF expectation of having to plan for financial future
TAX expectation of benefiting from saving tax incentives
SEL recognition of responsibility for own retirement as self-employed
SSN reduced expectation of Social Security
PNN reduced expectation of pension
Expectation of retirement income sources
SST Social Security
IRA- Individual Retirement Account
FOI 401 (k)
FOB 403(b)
ANN annuity
BNK bank/credit union savings account
BRK brokerage/mutual fund account
CPN company pension
STP stock option
RER real estate
INH inheritance
PER percent of estimated 1996 income saved for retirement
TOT total of estimated 1996 income saved for retirement
BEF levels of retirement savings intended to be saved
Selected types of saving accounts:
FOO- 401 (k)
FBB-403(b)
IRR- Individual Retirement Account
ANT annuity
SAA bank/credit union savings account
BMF brokerage/mutual fund account
44


Table 3.1. (Cont.)
Preferred savings vehicles:
CSH- check/cash
PAY -payroll deduction
EFT -electronic transfer
Preferences for Saving
CHA-pattems of saving as a child
CHT-pattems of saving as an adult
Influences on Saving Behavior
STK -levels of influence of stock market on saving
IRA levels of influence of IRA tax legislation on saving
ENC-encouragement of saving behavior as a child
PTS -parental influence
SPS -spousal/significant other influence
OTH-other family members influence
FRD- friend(s)/colleagues(s) influence
SUP- supervisor/employer influence
SPA -spiritual advisor influence
INV investment/tax advisor influence
BKS -influence of books
TVR -influence of television/radio
ADS -influence of advertisement/television commercials
NEW-influence of newspaper/magazine/newsletter
EPI influence of employer provided information
ITS influence of investment/tax services
ONL -influence of on-line services
SEM- influence of seminars/courses
Retirement Expectations
YRS -number of expected years until retirement
COM-estimate of total retirement money saved
Demographic Data:
SEX -sex of respondent
MAR-marital status
BIR year of birth
BIS year of birth of spouse
KID number of children within three age ranges
INC levels of household income
MTG-number of years remaining on mortgage loan
EDU-number of years of schooling
EMP-changed employers within last five years
OCC-occupation
An ordinal scale was designed to measure variables including levels of optimism about
achieving financial security, of conviction to personal priorities, of certainty of the importance
of saving and to identify elements of saving decision-making.
45


Data Collection
In late October of 1997, 2,000 questionnaires were mailed to
randomly selected IRA participant with accounts at First Trust Corporation.
The one page two-sided questionnaire was sent along with a cover letter
(Appendix C), printed on First Trust letterhead, and a postage paid return
envelope. Each questionnaire was coded so that upon its return, it could be
identified with an IRA account. The letter, signed by the author, Vickie
Wilson (an employee of First Trust), stated the purpose of the study and
indicated that it was jointly supported by First Trust and the University of
Colorado at Denver. It was hoped that this approach would lend the study
credibility and encourage higher participation. Survey recipients were also
requested to mail the completed survey in the provided envelope within
three weeks.
Data Analysis
Over a period of six weeks. 282 completed surveys (14.1 percent)
were returned. Because First Trust IRA account owners are regularly
surveyed about products, service and pricing, this was considered a very
good" response rate, within the corporation. It might have been possible to
increase participation, if a monetary incentive had been offered to
46


respondents, however, the resources of the researcher were limited.
Instead, participants were offered a copy of the completed survey results,
and over 82 percent indicated an interest in receiving the report.
The Statistical Package for the Social Sciences, more commonly
known as SPSS (SPSS for MS Windows Release 6.0), was the personal
computer software program used in the analysis of all study data.
Summary
This study postulates that based on cohort theory, there is a
relationship between baby boomers retirement saving behavior and their
values, preferences and influences. Using cohort theory, a model was
developed. A survey research instrument was designed to identify and
measure key variables.
To examine retirement saving behavior, a data base of IRA
participants was selected and a mail survey was sent to 2,000 individuals.
These were individuals who had saved for retirement through their IRA. IRA
contributions were used as a measure of their saving behavior. In addition,
respondents provided estimates of their total retirement savings to get a
more complete picture. A total of 282 completed surveys were returned and
SPSS software was used to analyze the data.
47


CHAPTER 4
PRESENTATION OF RESULTS
Of the 2,000 questionnaires mailed to First Trust Corporation IRA
participants born between 1946 and 1964, there were 282 responses (14.1
percent). The group breakdown by sex is presented in Table 4.1:
TABLE 4.1
RESPONDENT DESCRIPTION
Sex Number of Respondents Percentage of Total
Male 166 58.9
Female 116 41.1
Total 282 100.0
More than 72 percent of the respondents were married, and almost
80 percent had household incomes of $50,000 or more (Tables 4.2 and
4.3).
TABLE 4.2
MARITAL STATUS
Status Number of Responses Percentage of Total
Mamed 204 72.3
Single 49 17.4
Divorced 26 9.2
Widowed 3 1.1
Total 282 100.0
48


TABLE 4.3
HOUSEHOLD INCOME
Income Range Number of Responses Percentage of Total
Under $25,000
$25,001 $50,000
$50,001 -$100,000
Over $100,000
10
46
121
99
3.6
16.7
43.8
35.9
Total
276
100.0
Data was also gathered on respondents birth years, dividing them
between senior boomers (80.5 percent), born between 1946 to 1955, and
junior boomers (19.5 percent), born between 1956, and 1964 (Table 4.4).
This age division was introduced in pension research on boomers by Woods
(1994). The majority of respondents (71.4 percent) were well educated, with
more than 16 years of schooling (Table 4.5).
TABLE 4.4
AGE OF RESPONDENTS
Birth Year
Number of Responses Percentage of Total
1946-1955
(Senior Boomers)
1956- 1964
(Junior Boomers)
223
54
80.5
19.5
Total
277
100.0
49


TABLE 4.5
YEARS OF EDUCATION
Education Range Number of Responses Percentage of Total
Less than 12 Yrs
12-15 years
16 years
More than 16 yrs
3
77
65
135
1.1
27.5
23.2
48.2
Total
280
100.0
To gather information on potential future financial obligations,
respondents were asked to select the age ranges of their children and the
number of years remaining on their mortgage. Tables 4.6 and 4.7 show
that a majority (85.2 percent) of the respondents had children who might
represent higher education obligations, and 42.5 percent indicated that they
had long-term (16 years or more) mortgage obligations.
TABLE 4.6
NUMBER OF CHILDREN
Age Range
Number of Responses Percentage of Total
12 yrs old and under
13-21 yrs old
Over 21 yrs old
248
103
61
60.2
25.0
14.8
Total
412
100.0
50


TABLE 4.7
YEARS REMAINING ON MORTGAGE
Mortgage Years Number of Responses Percentage of Total
0 years 39 15.0
1 -15 yrs 110 42.5
16-25 yrs 73 28.2
more than 26 yrs 37 14.3
Total 259 100.0
To identify respondents occupations, responses were grouped under
categories including: management (19.7 percent), technical (16.3 percent),
professional (15.9 percent), sales and marketing (11.3 percent), and
education (8.4 percent) (Table 4.8).
TABLE 4.8
OCCUPATION
Category Number of Responses Percentage of Total
Management 47 19.7
Technical 39 16.3
Professional 38 15.9
Sales 27 11.3
Education 20 8.4
Clerical 19 7.9
Self-Employed 15 6.3
Health 11 4.6
Social Service 5 2.1
Other 18 7.5
Total 239 100.0
A descriptive profile of the average respondent could be as follows:
51


A married, college-educated male, with a management or
technical job, at least one school age child, an annual
household income in excess of $50,000, who has less than 15
years remaining on his mortgage.
Response to Research Questions
The four research questions presented in Chapter 3 were designed
to identify and measure key variables and to examine their relationships.
The results were expected to expand the understanding of how values and
attitudes impact boomer saving behavior. The presentation of the following
data, responds to each of the questions, while discussing the overall
relevance of the findings.
Question 1. What, if any, is the correlation between IRA contributions and
total retirement savings estimates?
Two sets of continuous data were collected, both of which measured
retirement saving behavior: IRA retirement saving contributions and total
retirement saving estimates.
Table 4.9 reports a correlation coefficient of R=.337, which indicates
the strength of the bivariate linear relationship. The results show that there
was found to be a moderate positive relationship between the two
52


measurements of retirement saving.
TABLE 4.9
CORRELATION BETWEEN IRA CONTRIBUTIONS AND
TOTAL RETIREMENT SAVINGS
IRA TOTAL
IRA 1.000 .337
(262) (214) P=.000
TOTAL .337 1.000
(214) P=.000 (215)
(Coefficient/Cases) / 2-tailed significance)
Variable Cases Mean Std Dev
IRA 262 22725.824 46175.535
TOT 215 112358.140 139153.573
By calculating R2=.114, the correlation of determination indicates that
for the linear relationship, IRA contributions explains about 11.4 percent of
the sample variance of total retirement savings and the remaining 88.6
percent of the variance is unexplained or residual. The reported P-value
equals .000 is statistically significant at ^ = .01.
The results of this analysis suggest that while there is a linear
relationship between the two sets of data, it is not a perfect fit. Intuitively,
since the sample consisted of IRA accounts, it could be expected that IRA
53


savings would be at least a portion of the total retirement savings picture.
The data reports that the mean for IRA contributions was $22,726 and for
total retirement savings was $112,358.
This suggests that for some respondents the IRA was their total
savings and for others, only a portion. If that is the case, the total retirement
savings estimates are a better indicator of retirement savings than the IRA
contributions. The total retirement savings estimates could be expected to
include not just IRA contributions, but also saving, brokerage and mutual
fund accounts, as well as salary reduction plans, such as the 401 (k) and
403(b) accounts. Based on these results, the rest of the data analysis will
employ total retirement savings as the variable that best measures
retirement saving behavior.
Question 2. What, if any, are the relationships between the variables
identified with cohort theory and those variables traditionally
identified with saving theory?
Variables identified with cohort theory included those measuring
levels of optimism about achieving financial security, levels of conviction
concerning personal priorities and levels of certainty of the importance of
saving.
SUF = Level of optimism of saving sufficiency
54


PRI = Level of conviction of saving as a financial priority
MAJ = Level of certainty of importance of savings in retirement
Preferences toward saving as a child and as an adult were also variables:
CHA = Patterns of saving as a child
CHT = Patterns of saving as an adult
Levels of influences of IRA tax policy and stock market changes on
saving behavior were also included as variables:
STK = Level of influence of stock market
IRA = Level of influence of IRA tax legislation
Responses to the questions designed to measure cohort variables
were cross tabulated against two variables traditionally associated with
saving theory: age (Birth Year) and income (Household Income). The
questions were designed on an ordinal scale.
SUF: I feel that I will have sufficient income to retire at or
before age 65.
Respondents were asked to select responses based on a scale of
strongly agree", agree", disagree, strongly disagree or unsure. Tables
4.10 and 4.11 show that 53.3 percent of the respondents felt that they would
have sufficient retirement income. There appears to be little difference in
responses among, senior and junior boomers. However, boomers with
55


lower household incomes ($50,000 or less) were far more likely (64.0
percent) to disagree or strongly disagree that they would have sufficient
income at retirement. This suggests that while age may not be a key factor,
income is important in determining respondents level of optimism in
achieving retirement goals.
TABLE 4.10
WILL HAVE SUFFICIENT RETIREMENT INCOME
BY BIRTH YEAR RANGE
Senior Boomers Junior Boomers All Boomers
Scale 1946-55 % 1956-64 % Total %
Strongly agree 43 21.1 12 23.5 55 21.6
Agree 65 31.8 16 31.4 81 31.7
Disagree 43 21.1 14 27.5 57 22.4
Strongly disagree 53 26.0 9 17.6 62 24.3
Total 204 100.0 51 100.0 255 100.0
TABLE 4.11
WILL HAVE SUFFICIENT RETIREMENT INCOME
BY HOUSEHOLD INCOME
Scale Under $25K $25K $50K i -* cn o o Over $100K Total %
Strongly agree 3 6 18 28 55 21.6
Agree 0 9 34 38 81 31.7
Disagree 0 14 26 17 57 22.4
Strongly disagree 5 13 33 11 62 24.3
Total 8 42 111 94 255 100.0
56


MAJ: I believe that the majority of my retirement income will
come from savings.
More than 71 percent of the respondents believed that savings will
make up the bulk of their retirement income. Younger boomers (84.9
percent) agreed more often than older ones (67.9 percent) (Table 4.12) and
household income was also a determinant (Table 4.13) Among those with
higher incomes (over $50,000), 74.8 percent agreed with the statement,
while 56.0 percent of the lower income respondents agreed. This suggests
that respondents with lower incomes do not expect to have the level of
savings needed to make up the majority of their income.
TABLE 4.12
SAVINGS WILL BE A MAJORITY OF RETIREMENT INCOME
BY BIRTH YEAR RANGE
Senior Boomers Junior Boomers All Boomers
Scale 1946- 55 % 1956- 64 % Total %
Strongly agree 77 35.8 29 54.7 106 39.6
Agree 69 32.1 16 30.2 85 31.7
Disagree 34 15.8 7 13.2 41 15.3
Strongly disagree 35 16.3 1 1.9 36 13.4
Total 215 100.0 53 100.0 268 100.0
57


TABLE 4.13
SAVINGS WILL BE A MAJORITY OF RETIREMENT INCOME
BY HOUSEHOLD INCOME
Scale Under $^oK $25 K -$50K $50K -$100K Over $100K Total %
Strongly agree 3 10 41 52 106 39.6
Agree 4 11 39 31 85 31.7
Disagree 0 11 21 8 41 15.3
Strongly disagree 1 10 17 9 36 13.4
Total 8 42 118 100 268 100.0
PRI: Saving for retirement is an important financial priority.
An overwhelming 94.2 percent of respondents agreed that saving for
retirement was a financial priority (Tables 4.14 and 4.15), and the
agreement was across all income and age levels.
TABLE 4.14
SAVINGS FOR RETIREMENT IS AN IMPORTANT FINANCIAL PRIORITY
BY BIRTH YEAR RANGE
Senior Boomers Junior Boomers All Boomers
Scale 1946 55 . % 1956-64 % Total %
Strongly agree 155 70.5 43 78.1 198 72.0
Agree 52 23.6 9 16.4 61 22.2
Disagree 12 5.5 3 5.5 15 5.5
Strongly disagree 1 0.4 0 0.0 1 0.3
Total 220 100.0 55 100.0 275 100.0
58


TABLE 4.15
SAVINGS FOR RETIREMENT IS AN IMPORTANT FINANCIAL PRIORITY
BY HOUSEHOLD INCOME
Under $25K $50K Over
Scale S2! 5K - $50K -S100K S100K Total %
Strongly agree 8 30 85 75 198 39.6
Agree 3 11 29 18 61 31.7
Disagree 0 5 4 6 15 15.3
Strongly disagree 0 0 1 0 1 13.4
Total 11 46 119 99 275 100.0
Questions were also posed to identify when saving preferences were
formed, beginning with the following: When you were a child (before age
18), were you encouraged to save? Respondents were asked to answer
yes" or no. A sizeable majority (79.6 percent) of respondents were
encouraged to save as children (Table 4.16).
TABLE 4.16
ENCOURAGED TO SAVE AS A CHILD
Response Number of Responses Percentage of Total
Yes 207 79.6
No 53 20.4
Total 260 100.0
Two follow-up questions were then asked:
CHA: When you were a child, how would you have
59


characterized your saving patterns?
Respondents were asked to select among an ordinal scale of four
responses: Save frequently, Save occasionally, Seldom save and I
dont know. Tables 4.17 and 4.18 indicate that among senior boomers,
72.0 percent indicated that they saved frequently or occasionally as a child,
as opposed to 75.9 percent of the junior boomers. Higher income (over
$50,000) respondents tended to have saved more often. About 76 percent
saved frequently or occasionally as a child, as opposed to 60 percent
among lower income respondents. This suggests that perhaps preferences
to save developed early and among those who now have higher incomes.
TABLE 4.17
SAVINGS PATTERNS AS A CHILD
BY BIRTH YEAR RANGE
Senior Boomers Junior Boomers All Boomers
Scale 1946 -55 % 1956-64 % Total %
Frequently 89 40.1 24 44.4 113 40.9
Occasionally 71 31.9 17 31.5 88 32.0
Seldom 51 23.0 12 22.2 63 22.8
Dont Know 11 5.0 1 1.9 12 4.3
Total 222 100.0 54 100.0 276 100.0
60


TABLE 4.18
SAVINGS PATTERNS AS A CHILD
BY HOUSEHOLD INCOME
Scale Under $25K $25K $50K $50K $100K Over $100K Total %
Frequently 4 18 49 42 113 40.9
Occasionally 0 11 40 37 88 32.0
Seldom 2 13 30 18 63 22.8
Dont Know 4 3 2 3 12 4.3
Total 10 45 121 100 276 100.0
CHT: How would you characterize your savings patterns
today?
Respondents were again asked to select among an ordinal scale of
four responses: Save frequently", Save occasionally", Seldom save and
I dont know. Table 4.19 indicate that the age of respondents did not
appear to be a key determinant. The results suggests that both senior
boomers (91.9 percent) and junior boomers (98.2 percent) indicated that
they saved frequently or occasionally today. There was more of a
difference among income groups. Higher income respondents (over
$50,000) tended to save more often (Table 14.20). Over 95 percent saved
frequently or occasionally today, as opposed to 83.6 percent of lower
income respondents. This suggests that preferences, while identified
61


across age ranges, tends to be more frequent among higher incomes.
TABLE 4.19
SAVINGS PATTERNS TODAY
BY BIRTH YEAR RANGE
Senior Boomers Junior Boomers All Boomers
Scale 1946-55 % 1956 64 % Total %
Frequently 154 69.4 40 74.1 194 70.3
Occasionally 50 22.5 13 24.1 63 22.8
Seldom 18 8.1 1 1.9 19 6.9
Dont Know 0 0.0 0 0.0 0 0.0
Total 222 100.0 54 100.0 276 100.0
TABLE 4.20
SAVINGS PATTERNS TODAY
BY HOUSEHOLD INCOME
Under $25K $50K Over
Scale $25K - $50K -S100K $100K Total %
Frequently 4 20 89 81 194 70.3
Occasionally 4 18 25 16 63 22.8
Seldom 2 7 8 2 19 6.9
Dont Know 0 0 0 0 0 0.0
Total 10 45 122 99 276 100.0
Questions designed to identify influences on saving behavior were
also cross tabulated against traditional saving variables: age and income.
STK: Recent growth in the stock market has influenced me to
62


save.
Respondents were asked to select a response based on a scale of
strongly agree, agree, disagree or strongly disagree. Tables 4.21
and 4.22 indicate that the majority of both senior (67.5 percent) and junior
(61.5 percent) boomers, disagreed or strongly disagreed that the stock
market had influenced them to save, and lower income respondents (under
$50,000) disagreed with the statement (65.3 percent) almost as much as
higher income respondents (66.5 percent). These results suggest that stock
market fluctuations did not influence saving behavior for the majority of
respondents.
TABLE 4.21
STOCK MARKET INFLUENCE
BY BIRTH YEAR RANGE
Senior Boomers Junior Boomers All Boomers
Scale 1946 -55 % 1956-64 % Total %
Strongly agree 28 13.2 7 13.5 35 13.2
Agree 41 19.3 13 25.0 54 20.5
Disagree 92 43.4 21 40.4 113 42.8
Strongly disagree 51 24.1 11 21.1 62 23.5
Total 212 100.0 52 100.0 264 100.0
63


TABLE 4.22
STOCK MARKET INFLUENCE
BY HOUSEHOLD INCOME
Scale Under $25K $25K $50K $50K -$100K Over S100K Total %
Strongly agree 3 8 12 12 35 13.2
Agree 0 6 24 24 54 20.5
Disagree 1 20 49 43 113 42.8
Strongly disagree 2 9 33 18 62 23.5
Total 6 43 118 97 264 100.0
IRA: Recent favorable changes in IRA tax legislation have
influenced me to save.
Respondents were again asked to select a response based on an
ordinal scale of strongly agree", agree, disagree", strongly disagree or
unsure". Tables 4.23 and 4.24 indicate similar results as with the stock
market question. The majority of both senior and junior boomers (under
$50,000) disagreed. Among older income respondents, 70.8 percent
disagreed or strongly disagreed, as did 66.0 percent of younger
respondents. The majority of both lower (70.2 percent) and higher (69.8
percent) income respondents disagreed that IRA tax legislation influenced
their saving behavior. These findings support those who suggest that
existing IRA tax policy does little to influence respondents saving behavior.
64


TABLE 4.23
IRA LEGISLATION INFLUENCE
BY BIRTH YEAR RANGE
Senior Boomers Junior Boomers Ail Boomers
Scale 1946 -55 % 1956-64 % Total %
Strongly agree 21 10.1 6 11.4 27 10.3
Agree 40 19.1 12 22.6 52 20.0
Disagree 80 38.3 21 39.6 101 38.4
Strongly disagree 68 32.5 14 26.4 82 31.3
Total 209 100.0 53 100.0 262 100.0
TABLE 4.24
IRA LEGISLATION INFLUENCE
BY HOUSEHOLD INCOME
Scale Under $25K $25K $50K $50K -$100K Over $100K Total %
Strongly agree 1 4 12 10 27 10.3
Agree 2 7 21 22 52 20.0
Disagree 2 21 46 32 101 38.4
Strongly disagree 2 8 38 34 82 31.3
Total 7 40 117 98 262 100.0
Based on the results of the cross tabulation of cohort theory variables
against traditional saving theory variables, the findings are summarized for
the research question #2:
What, if any, are the relationships between the variables identified
with cohort theory and those variables traditionally identified with
65


saving theory?
1. There appears to be a positive relationship between
household income levels (INC) and levels of optimism in
achieving retirement goals (SUF).
2. A relationship between household income (INC) and levels of
certainty that the majority of retirement income will come from
savings (MAJ) was indicated. Lower income respondents did
not appear to be as certain as higher income respondents that
they would attain the level of savings that would be necessary
to make up the majority of their income.
3. Frequent saving patterns as a child (CHA) appear to have
developed more often among those who have higher incomes.
4. Frequent saving patterns today (CHT), while occurring across
age ranges, tend to appear more often among those with
higher incomes.
5. Neither IRA tax legislation changes nor stock market
fluctuations appear to have influenced saving behavior for the
majority of respondents, regardless of age or income.
Question #3: What, if any, are the statistically significant differences
between the variables that have been identified as describing
66


boomer saving behavior?
As a first step, analysis of variance (ANOVA) was applied to the
variables identified as describing boomer saving behavior. ANOVA refers to
statistical methods used to evaluate the effectiveness of differences
between groups. The technique compares the size of between-group"
differences with the size of within-group differences due to individual
variability.
Retirement saving behavior (Y) was defined as Total Retirement
Savings (SAV), which was gathered as continuous data. Using this data,
one-way (one variable) ANOVA tests were conducted. The value of Total
Retirement Savings (Y) was then tested against the following independent
variables (X):
SUF levels of optimism that income will be sufficient at or before
retirement age
MAJ levels of certainty that a majority of retirement income will
come from savings:
PRI levels of conviction that saving for retirement is an important
financial priority
ENC patterns of encouragement of saving behavior as a child
CHA patterns of saving as a child
67


CHT patterns of saving today
STK levels of influence of stock market on saving
IRA levels of influence of IRA tax legislation on saving
SEX sex of respondent
MAR marital status
BIR year of birth
KID number of children within three age ranges
INC levels of household income
MTG number of years remaining on mortgage loan
EDU number of years of schooling
EMP changed employers within last five years
OCC type of occupation
PER percents of estimated 1996 income saved for retirement
TOT totals of estimated 1996 income saved for retirement
BEF levels of retirement savings intended to be saved
YRS number of expected years until retirement
COM estimates of total retirement money saved
When the ANOVA tests were performed significant results did
emerge. The Scheffe procedure (Scheffe, 1970) was employed to identify
statistical significance between groups. The relationships found to be
68


significant are identified below by the independent variable: SUF levels of
optimism that income will be sufficient at or before retirement age.
In Table 4.25, a difference was found in the main effect, SUF, and
the dependent variable, Total Retirement Savings (SAV), univariate _F
(3,196 = 6.312, £<01). A Sheffe test for significance confirmed that those
who strongly agree that they would have sufficient retirement income,
significantly differed in their total retirement savings from those who
disagree and strongly disagree.
TABLE 4.25
RELATIONSHIP USING ONE-WAY ANOVA TESTS BETWEEN
SAV-TOTAL RETIREMENT SAVINGS AND
SUF LEVEL OF OPTIMISM THAT INCOME WILL BE SUFFICIENT
Summary Table
Source Degrees of Freedom F Ratio F Probability
SUF Levels of Optimism 3,196 That Income Will Be Sufficient 6.312 .000
TOT total of estimated 1996 income saved for retirement
In Table 4.26, a difference was found in the main effect TOT and the
dependent variable, Total Retirement Savings (SAV), univariate _F (2,183 =
8.591, p<.01). A Sheffe test for significance confirmed that those who
saved $0 $5,000" and $5,000 $10,000" significantly differed in their
69


total retirement savings from those who saved from $10,000-520,000".
TABLE 4.26
RELATIONSHIP USING ONE-WAY ANOVA TESTS BETWEEN
SAV TOTAL RETIREMENT SAVINGS AND
TOT TOTAL ESTIMATED 1996 SAVINGS
Summary Table
Source Degrees of Freedom F Ratio F Probability
TOT Total Estimated 2,183 8.591 .000
1996 Savings
BEF levels of retirement savings intended to be saved
In Table 4.27, a difference was found in the main effect BEF and the
dependent variable, Total Retirement Savings (SAV), univariate JF (3,184 =
6.524, e<.01). A Sheffe test for significance confirmed that those who
estimated that they intended to save over $1 million" significantly differed
in their total retirement savings from those who estimated 5100,000 -
5500,000.
70


TABLE 4.27
RELATIONSHIP USING ONE-WAY ANOVA TESTS BETWEEN
SAV TOTAL RETIREMENT SAVINGS AND
BEF LEVELS OF INTENDED RETIREMENT SAVINGS
Summary Table
Source Degrees of Freedom F Ratio F Probability
BEF-Levels of 3,184 6.524 .000
Intended Retirement
Savings
CHT-patterns of saving today
In Table 4.28, a difference was found in the main effect CHT and the
dependent variable, Total Retirement Savings (SAV), univariate _F (2,212 =
4.128, p<.05). A Sheffe test for significance confirmed that those who
saved frequently significantly differed in their total retirement savings from
those who seldom saved".
TABLE 4.28
RELATIONSHIP USING ONE-WAY ANOVA TESTS BETWEEN
SAV TOTAL RETIREMENT SAVINGS AND
CHT-PATTERNS OF SAVING TODAY
Summary Table
Source Degrees of Freedom F Ratio F Probability
CHT Saving Patterns 2,212 4.128 .017
Today
71


INC levels of household income
In Table 4.29, a difference was found in the main effect INC and the
dependent variable, Total Retirement Savings (SAV), univariate _F (3,208 =
17.333, £><.01). A Sheffe test for significance confirmed that respondents
with incomes under $25,000" differed significantly from those with incomes
of $25,000 to $50,000", $50,000 to $100,000" and Over $100,000".
TABLE 4.29
RELATIONSHIP USING ONE-WAY ANOVA TESTS BETWEEN
SAV TOTAL RETIREMENT SAVINGS AND
INC LEVELS OF HOUSEHOLD INCOME
Summary Table
Source Degrees of Freedom F Ratio F Probability
INC Levels of Income 3,208 17.333 .000
EDU-number of years of schooling
In Table 4.30, a difference was found in the main effect EDU and the
dependent variable, Total Retirement Savings (SAV) univariate _E (3,211 =
11.385, £<.01). A Sheffe test for significance confirmed that those who had
at least 16 years of education significantly differed in their total retirement
savings from those that had less than 16 years.
72


TABLE 4.30
RELATIONSHIP USING ANOVA TESTS BETWEEN
SAV TOTAL RETIREMENT SAVINGS AND
EDU-NUMBER OF YEARS OF SCHOOLING
Summary Table
Source Degrees of Freedom F Ratio F Probability
EDU Years of Schooling 3,211 11.385 .000
In summary, the one-way ANOVA test results were, as follows:
1. Total retirement savings differed significantly among groups that were
optimistic about having sufficient retirement savings and those who
were not optimistic.
2. Total retirement savings differed significantly among those who had
smaller savings ($10,000 or less) in 1996 and those with larger
amounts.
3. Total retirement savings differed significantly depending on the
amounts respondents estimated that they would need to save.
4. Total retirement savings differed significantly between those making
under $25,000 and the rest of the household income ranges.
5. Total retirement savings differed significantly between those who had
at least a college education (16 years or more of schooling) and
those who did not.
73


In addition to one-way ANOVA techniques, factorial ANOVA
techniques were employed to determine if there were statistically significant
differences between groups of two and three variables. In a three-way
analysis of variance between SUF, MAJ and PRI, a difference was found in
the main effect, SUF and the dependent variable, Total Retirement Savings
(SAV), univariate _F (9,183 = 5.479, e<.01). When the three variables were
examined separately, only SUF tested as statistically significant with a P-
value of .001 (Table 4.31).
TABLE 4.31
RELATIONSHIP USING THREE-WAY ANOVA TESTS BETWEEN
SAV TOTAL RETIREMENT SAVINGS,
SUF INCOME WILL BE SUFFICIENT,
MAJ MAJORITY OF INCOME WILL BE SAVINGS, AND
PRI SAVING AS A FINANCIAL PRIORITY
Summary Table
Source Degrees of Freedom F Ratio F Probability
lain Effects 9,183 3.352 .001
SUF 3 5.479 .001
MAJ 3 1.769 .155
PRI 3 1.763 .156
Two-way ANOVA tests were applied to CHT-saving patterns today
and CHA- saving patterns as a child. As might be expected, there was a
significant relationship between saving patterns today and total retirement
74


savings. In a two-way analysis of variance, a statistically significant
difference was found in the main effect, CHT and the dependent variable,
Total Retirement Savings (SAV), univariate _F ( 2,203) = 4.673 p = .010,
p <.01 (Table 4.32).
TABLE 4.32
RELATIONSHIP USING TWO-WAY ANOVA TESTS BETWEEN
SAV -TOTAL RETIREMENT SAVINGS, CHT- SAVING PREFERENCES
TODAY, AND CHA SAVING PREFERENCES AS A CHILD
Summary Table
Source Degrees of Freedom F Ratio F Probability
Main Effects 5,203 2.177 .058 .
CHA 3 .118 .949
CHT 2 4.673 .010
CHA by CHT 6 .605 .726
In a two-way analysis of variance, (Table 4.33) when STK and
IRA influences were tested, no statistical significant differences were found
in the main effect, STK, IRA the interaction and the dependent variable,
Total Retirement Savings (SAV): univariate _F ( 6,184) = 1.604 p = .148, p
> .05, univariate _E ( 3,184) = 1.395 p = .246, p > .05, univariate _E (
3,184) =.628 p = .598, p > .05, and univariate _F ( 9,184) = .813 p =
.605, p > .05.
75


TABLE 4.33
RELATIONSHIP USING TWO-WAY ANOVA TESTS BETWEEN
SAV -TOTAL RETIREMENT SAVINGS,
IRA IRA SAVINGS INFLUENCE AND
STK STOCK MARKET INFLUENCE
Summary Table
Source Degrees of Freedom F Ratio F Probability
Main Effects 6,184 1.604 .148 .
STK 3 1.395 .246
IRA 3 .628 .598
STK by IRA 9 .813 .605
The fourth research question that is responded to using statistical
techniques is:
Question #4: What, if any, are the factors that can be used to describe
boomer saving behavior?
For this question, factor analysis was used to simplify the process
and to reduce the possibility of multicollinearity. Factor analysis is
particularly useful when there are a number of original variables, allowing
them to be replaced by a few combined variables. The factor analysis
included seven predetermined variables:
Valuative variables:
SUF measured levels of optimism that respondents' will achieve
income sufficiency at or before retirement age
MAJ measured levels of certainty that respondents will receive a
76


majority of retirement income from savings
PRI measured levels of conviction that saving for retirement is an
important financial priority
Preferential variables:
CHA measured patterns of saving preferences as a child
CHT measured patterns of saving preferences as an adult
Influential variables:
STK measured levels of influence of the stock market on saving
IRA measured levels of influence of IRA tax legislation on saving
Cohort saving theory suggests that there may be a linear relationship
between the seven variables, representing 1). preferences, 2). values, 3).
influences and boomer saving behavior:
SAV = SUF + MAJ + PRI + CHA + CHT+ STK + IRA
The results of the correlation analysis provided components for
principle component analysis. An Alpha of .5608 was reported as an
indicator of internal consistency of items (Table 4.34).
77


TABLE 4.34
CORRELATION ANALYSIS
Reliability Analysis Scale Alpha
Mean Std Dev Cases
1. SUFFICT 2.4051 1.0679 237.0
2. MAJORITY 2.0127 1.0556 237.0
3. PRIORITY 1.3418 .6083 237.0
4. CHARIZE 1.8397 .8283 237.0
5. CHARACTE 1 3080 .5543 237.0
6. STOCKMKT 2.7722 .9648 237.0
7. IRATAX 2.9536 .9263 237.0
Correlation Matrix
SUFFICT MAJORITY PRIORITY CHARIZE CHARACTE
SUFFICT 1.0000
MAJORITY .2924 1.0000
PRIORITY .1774 .3364 1.0000
CHARIZE .1600 .0944 .1429 1.0000
CHARACTE .3252 .1816 .2394 1449 1.0000
STOCKMKT .1763 1485 .1910 0760 .0684
IRATAX .0105 . 0439 .1185 0153 .0610
STOCKMKT IRATAX
STOCKMKT 1.0000
IRATAX .4670 1.0000 N of Cases = 237.0
N of Statistics for Mean Variance Std Dev Variables
Scale 14.6329 10.4113 3.2267 7
Item Means Mean Minimum Maximum Range Max/Min Variance
2.0904 1.3080 2.9536 1.6456 2.2581 .4251
Item Variances Mean Minimum Maximum Range Max/Min Variance
.7724 .3073 1.1403 .8330 3.7112 .1117
Inter-item Mean Minimum Maximum Range Max/Min Variance
Correlations .1638 -.0153 .4670 .4822 -30.5941 .0135
Item-total Statistics
Scale Scale Corrected
Mean Vanance Item- Squared Alpha
if Item If Item Total Multiple if Item
Deleted Deleted Correlation Correlation Deleted
SUFFICT 12.2278 7.3292 .3358 .1916 .5015
MAJORITY 12.6203 7.4399 .3225 .1736 .5076
PRIORITY 13.2911 8.7157 .3691 .1726 .5065
CHARIZE 12.7932 8.9020 .1666 .0491 .5636
CHARACTE 13.3249 9.0762 .3077 .1500 5258
STOCKMKT 11.8608 7.5610 .3617 .2655 4896
IRATAX 11.6793 8.4391 .2070 .2302 5532
Reliability Coefficients 7 items Alpha = .5608
Standardized item alpha = .5783
78


The results of the principle component extraction are two factors:
Factorl and Factor2 (Figure 4.1).
FIGURE 4.1
PRINCIPLE COMPONENT EXTRACTION
28.9% variance
Eigenvalue: 2.02611
PRI
SUF
MAJ
CHT
STK
IRA
CHA
.63090
.61244
.60830
.55810
.54816
.38257
.35306
19.1% variance
Eigenvalue: 1.33757
.77491
.64185
-.32011
-.30900
According to the factor analysis (Table 4.35), the loadings suggests
that Factorl represents:
Factorl = VAL = PRI + SUF + MAJ + CHA + STK + IRA + CHT
The factor pattern shows that all seven variables: PRI, SUF, MAJ,
CHA, STK, IRA, CHT positively loaded on to Factorl (VAL) with coefficients:
.63090, .61244, .60830, .55810, .54816, .38257, .35306, respectively. The
eigenvalues of the correlation matrix for Factorl was 2.026 and 28.9% of
the variance within the seven variables was explained by this factor.
79


TABLE 4.35
FACTOR ANALYSIS
Correlation Matrix:
SUF MAJ PRI CHA CHT STK
1.00000
.29240 1.00000
.17737 .33640 1.00000
.15997 .09441 .14287 1.00000
.32521 .18159 .23941 .14494 1.00000
.17632 .14846 .19101 .07604 .06840 1.00000
.01052 .04394 .11852 -.01526 .06097
IRA
SUFFICT
MAJORITY
PRIORITY
CHARIZE
CHARACTE
STOCKMKT
IRATAX
Determinant of Correlation Matrix = .4765325
.46698 1.00000
Kaiser-Meyer-Olkin Measure of Sampling Adequacy = .61434
Bartlett Test of Sphericity = 172.58057, Significance = .00000
1-tailed Significance of Correlation Matrix:. is printed for diagonal elements.
Extraction 1 for analysis 1, Principal Components Analysis (PC
Initial Statistics:
Variable Communality Factor Eigenvalue Pet of Var Cum Pet
SUFFICT 1.00000 1 2.02611 28.9 28.9
MAJORITY 1.00000 2 1.33757 19.1 48.1
PRIORITY 1.00000 3 .93005 13.3 61.3
CHARIZE 1.00000 4 .86639 12.4 73.7
CHARACTE 1.00000 5 .77000 11.0 84.7
STOCKMKT 1.00000 6 .58744 8.4 93.1
IRATAX 1.00000 7 .48243 6.9 100.0
PC extracted 2 factors. Factor Matrix:
Factor 1 Factor 2
SUFFICT .61244 -.30900
MAJORITY .60830 -.21644
PRIORITY .63090 -.08435
CHARIZE .35306 -.27056
CHARACTE .55810 -.32011
STOCKMKT .54816 .64185
IRATAX .38257 .77491
Final Statistics
Variable Communality Factor Eigenvalue Pet of Var Cum Pet
SUFFICT .47057 1 2.02611 28.9 28.9
MAJORITY .41688 2 1.33757 19.1 48.1
PRIORITY .40515
CHARIZE .19785
CHARACTE .41394
STOCKMKT .71245
IRATAX .74684
80


Table 4.35 reflects the second factor as follows:
Factor2 = INF = IRA + STK +CHT +SUF. Variables PRI and STK
positively loaded on Factor2 (INF) with coefficients: .77491 and .64185,
respectively and CHT and SUF negatively loaded with coefficients, -.32011
and -.30900, respectively. The eigenvalues of the correlation matrix for
Factor2 (NF) was 1.33757 and 19.1 percent of the variance within the four
variables was explained by this factor. For both factors, the cumulative
variance explained was 48.1 percent.
In Table 4.36, the Varimax and Oblimin rotations yielded similar
results as the factor analysis. The Varimax converged in three iterations
and identified two factors. The variables SUF, CHT, MAJ, PRI and CHT
positively loaded on to Factorl (VAL), with coefficients: .68511, .64338,
.63573, .59003 and .44067, respectively. For Factor2 (INF), IRA and STK
positively loaded on to the factor with coefficients .82898 and .86271,
respectively. The Oblimin rotation extracted two Factors with positive
loadings for Factorl (VAL) on SUF, MAJ, CHT, PRI and CHA. The
coefficients are .68581, .64221, .64155, .60570 and .43517, respectively.
Factor2 (INF) positively loaded on the same two variables as the Varimax
rotation: IRA and STK. The positive loadings resulted in coefficients of
.83669 and .85840, respectively. The Factor Correlation Matrix is .129.
81


TABLE 4.36
VAR I MAX AND OBLIMIN ROTATION
Varimax Rotation
VARIMAX rotation 1 for extraction 1 in analysis 1 Kaiser Normalization.
VARIMAX converged in 3 iterations.
Rotated Factor Matrix:
Factor 1 Factor 2
SUFFICT .68511 .03440
MAJORITY .63573 .11280
PRIORITY .59003 .23878
CHARIZE .44067 -.06050
CHARACTE .64338 -.00213
STOCKMKT .15890 .82898
IRATAX -.05083 .86271
Factor Transformation Matrix:
Factor 1 Factor 2
Factor 1 .86908 .49467
Factor 2 -.49467 .86908
Oblimin Rotation
OBLIMIN rotation 1 for extraction 1 in analysis 1 Kaiser Normalization.
OBLIMIN converged in 4 iterations.
Structure Matrix:
Factor 1 Factor 2
SUFFICT .68581 .07342
MAJORITY .64221 .14887
PRIORITY 60570 .27204
CHARIZE .43517 -.03527
CHARACTE .64155 .03457
STOCKMKT .21824 .83669
IRATAX .01148 .85840
Factor Correlation Matrix:
Factor 1 Factor 2
Factor 1 1.00000
Factor 2 .12884 1.00000
82


VAL valuative determinants of saving behavior
The factor analysis findings support the hypothesis that there are
core values that determine saving behavior. The loadings on Factorl (VAL)
which measures sufficiency, certainty in the future, goal-orientation and the
self-perception of being a saver, reflect values that appear to directly impact
the saving decision-making process. The factor analysis suggests that as a
variable, VAL, could be useful in the development of a predictive model of
boomer retirement saving behavior.
INF Non-influential factors on saving behavior
The loadings on Factor2 (INF) combine the stock market and IRA tax
legislation influence variables. These two variables were identified because
of speculation that each might have a direct impact on boomer retirement
saving behavior, however, the loadings indicate that neither of them are a
major influence on saving behavior. This finding is especially important for
those policy makers who have developed current IRA tax incentives as
major cornerstones of retirement saving policy.
In summary, the factor analysis provided a means of simplifying the
analysis and avoiding multicollinearity. The method extracted relationships
and the results suggest that there may be predictive power in the variables
that measure saving values. The findings do not support assumptions that
83


IRA tax incentives or stock market changes impact saving behavior.
Additional Descriptive Data
In addition to the data applied to determine the relationships between
saving variables, there was more descriptive data generated to provide
insights into the boomers' saving decision-making process. Detailed results
are presented in Appendix D, however, the following is a summary of the
key findings. When asked if they were actively saving for retirement, 238
respondents ( 86.2 percent) indicated yes" and 38 (13.8 percent) indicated
no". Those who responded no were also asked to indicate their reasons
for not saving. The following responses are given in Table 4.37:
TABLE 4.37
REASONS FOR NOT SAVING FOR RETIREMENT
Reasons Given Number of Responses Percentage of Total
No funds available 27 49.1
Do not plan to retire 9 16.4
Expect sufficient pension 5 9.1
Expect sufficient Social Security 4 7.3
Expect inheritance 4 7.3
Intend to save when older 3 5.4
Have sufficient retirement funds 3 5.4
Total 55 100.0
Financial constraints appeared to be the most compelling reason for
84


not saving for retirement. Respondents who did not save, most often
indicated (49.1 percent) that it was because they did not have the funds
available". An indication of a potential trend was reflected among the 16.4
percent who said that they did not save because they did not plan to retire.
Neither Social Security payments nor an expected inheritance appeared to
have a major impact on decision-making.
Among those respondents who were actively saving for retirement,
they most often selected the following reasons for saving** (Table 4.38):
TABLE 4.38
REASONS FOR SAVING FOR RETIREMENT
Reasons Given Number of Responses Percentage of Total
Plan for financial future
Not sufficient Social Security
Tax incentives
Not sufficient pension
Have funds available
Self-employed
Total
222 25.5
213 24.4
137 15.7
124 14.2
118 13.5
58 6.7
872 100.0
Their responses suggest that they value the importance of having
control over their own financial futures. They most often selected must
plan for my financial future, do not expect Social Security to provide
sufficient retirement income", tax incentives and do not expect sufficient
retirement income from my company pension". Responses suggest that
85


these are boomers who ultimately expect to be responsible for their own
retirement income. This is consistent with responses to questions
concerning expected sources of retirement income. Expectations are that
IRAs (20.5 percent), 401 (k)s (13.8 percent), brokerage/ mutual fund
accounts (13.5 percent), Social Security (13.0 percent), savings accounts
(10.7 percent) and company pension plans (10.7 percent) will be the
primary sources of retirement income (Appendix D).
Among those who save, Table 4.39 shows that in 1996 the majority
saved between 1 and 15 percent and saved less than $10,000 of their
household income:
TABLE 4.39
1996 RETIREMENT SAVING
As a Percentage of Household Income
Percentage Range Number of Responses Percentage of Total
1 5% 47 18.7
6-15% 139 55.2
16-25% 50 19.8
Over 26% 16 6.3
Total 252 100.0
As A Dollar Amount of Household Income
Amount Number of Responses Percentage of Total
$0 $10,000 139 55.8
$10,001 -$20,000 68 27.3
Over $20,000 42 16.9
Total 249 100.0
86


When asked how much they intended to save before retirement,
respondents estimates varied widely, however, more than 60 percent
expected to save at least $500,000 for retirement.
TABLE 4.40
EXPECTED SAVINGS BEFORE RETIREMENT
Amount Number of Responses Percentage of Total
Under $100,000 18 7.2
$100,001 -3500,000 82 33.0
$500,001 -$1,000,000 66 26.5
Over $1,000,000 83 33.3
Total 249 100.0
Among the 282 who responded, 53.9 percent were certain that they
planned to retire in fifteen years or less.
TABLE 4.41
WORK YEARS BEFORE RETIREMENT
Responses Number of Responses Percentage of Total
0-15 years 152 53.9
16-30 years 110 39.0
More than 30 years 11 3.9
I do not plan to retire. 9 3.2
Total 282 100.0
Results in Appendix D indicated that the relationships that most often
influence boomer saving behavior are primarily: parents (29.8 percent),
87


spouse (25.9 percent), investment adviser (19.4 percent) and friends ((10.8
percent). The sources of financial information most commonly relied on for
saving decisions are newspapers (26.9 percent), investment/tax services
(21.1 percent) and employer-provided information (16.6%).
In summary, the data confirms that among non-savers, a lack of
funds was the most important reason for not saving for retirement. Among
savers, the need to plan for the future and the expectation that they cannot
depend on Social Security, were their most compelling reasons for saving.
Boomers did generally agree that their primary retirement income will come
from IRAs, 401 (k)s, brokerage accounts and Social Security.
Although most boomers considered themselves savers, the majority
saved less than $10,000 for retirement in 1996. And even though most
respondents planned to retire in fifteen years or less, the research suggests
that based on their future financial obligations, including childrens education
and mortgage payments, many may have to postpone those plans.
88


CHAPTER 5
CONCLUSIONS
A gap exists between current saving rates and expected saving
behavior. Having consistently delayed saving, 76 million Americans are
theoretically, well within their so-called peak saving years", their retirement
years loom ahead, yet saving rates continue to lag.
This research focuses on the population segment most affected by
the saving gap: baby boomers, born between 1946 and 1964. These
individuals have demonstrated attitudes toward saving that are quite
different from previous cohorts, with an emphasis on consumerism, rather
than thrift and instant gratification, rather than a focus on the long-term. A
recognition and understanding of these shifts in attitudes has become
increasingly important as policy makers begin to respond to the needs of the
next wave of retirees.
In response to shifts in values, the purpose of this study is to expand
the theoretical framework of boomer saving behavior beyond rational
economics, to recognize the explanatory power of psychological and
sociological theories, and to introduce a cultural construct. Cohort analysis
89