Content uploaded by Annamaria Lusardi
Author content
All content in this area was uploaded by Annamaria Lusardi on Feb 16, 2015
Content may be subject to copyright.
1
Financial Education and the Saving Behavior of African-American and
Hispanic Households
Annamaria Lusardi
1
Department of Economics
Dartmouth College
Hanover, NH 03755
Tel: (603) 646-2099
September 2005
1
This research project was supported by the U.S. Department of Labor, Employee Benefits Security
Administration. I would like to thank Al Gustman for help, suggestions, and comments. Justin Blesy and
Mark Christman provided excellent research assistance. Any errors are my responsibility. The opinions
expressed in this paper are those of the author and do not reflect the opinions of the Employee Benefits
Security Administration of the Department of Labor.
2
OBJECTIVES AND POLICY RELEVANCE
Hispanic and African-American minorities are, and will continue to be, major
factors in the U.S. economy. According to the latest U.S. Census Bureau estimates,
Hispanics are already the largest minority group in the country. By the year 2050, they
are projected to make up nearly one quarter of the U.S. population, a two-fold increase
from their current size. African-Americans will remain the second largest minority group
and are projected to increase to approximately 15% of the U.S. population in the coming
years.
Because of their substantial and growing presence, their saving and investment
behavior is cause for national concern. For example, these households are not only less
likely to hold high-return assets, such as stocks and business equity, but they do not even
hold basic assets, such as checking accounts. While several initiatives have been taken by
the government and employers to reduce discrimination and foster savings and financial
security, it is not clear how effective these programs are in reaching this segment of the
population.
Though studies of the effects of financial education on saving and investment
behavior have gained presence over recent years, few have focused on minority groups.
The saving behavior of minorities is rather different than the rest of the population. As I
argue in this proposal, information and planning costs can be particularly high among
minorities and a major obstacle to accumulating wealth and investing in high-return
assets. Understanding the saving and investment behavior of African-Americans and
Hispanics is critically important for devising and implementing policies that can be
effective in shaping the behavior of families where saving is most scarce.
3
PREVIOUS WORKS
Previous studies examining the financial position of households have highlighted
the fact that the wealth holdings of African-Americans and Hispanics are very low
(Hurst, Luoh and Stafford (1998)). Smith (1995) and Lusardi (1999, 2000) further
emphasize that many African-Americans and Hispanics arrive at retirement with little
wealth. Close to one quarter of African-Americans and Hispanics approach retirement
with less than $1,000 in total net worth (excluding pensions and Social Security) and
have nothing in financial net worth (which includes savings and checking accounts, CDs
and other short-term securities, bonds, stocks, IRAs and other assets).
Other studies, which have examined portfolio choice or specific assets (such as
housing, stocks, IRAs and 401(k)s) have further documented that African-Americans and
Hispanics do not hold many of the assets commonly present in household portfolios. For
example, Haliassos and Bertaut (1995) find that minorities are much less likely to hold
stocks than White households, and this remains the case even after accounting for a large
set of household and industry characteristics, income, and wealth. Similarly, Charles and
Hurst (2002) find that African-Americans are much less likely to own a home or apply
for a mortgage.
There are many reasons for this heterogeneity in wealth accumulation. For
example, African-Americans and Hispanics often have low education, low income, and
may have been hit by many negative shocks. African-Americans and Hispanics also have
lower financial literacy than Whites (Hogarth and Hilgerth (2002)), which is correlated
with poor saving and investment behavior (Hilgert, Hogarth and Beverly (2003) and
Hogarth and Hilgerth (2002)). They may also display differences in preferences like, for
4
example, a high degree of impatience (Lawrance (1991)) or high risk-aversion (Barsky,
Kimball, Juster and Shapiro (1997)). In addition, they may expect government programs
to support them in the future or rely on a network of relatives and friends. Minorities
may generally feel mistrust of investing in equity markets (Mabry (1999)), or perceive
discrimination and self-select away from formal financial institutions where they feel
discriminated (Swire (1995), Longhofer and Peters (2005)). They are also likely to face
means-tested programs that discourage asset possession (Hubbard, Skinner and Zeldes
(1995)).
This long but partial list of reasons why minorities do not save highlights the
difficulties of studying this topic and the necessity of a rich data set to address this topic.
In this project I propose yet another reason why African-Americans and Hispanics do not
save. I argue that high information and learning costs prevent these families from
accumulating wealth and securing a comfortable retirement. In other works (Lusardi
(2000, 2003b)), I find that these costs are important for the general population and,
particularly, for those with low levels of education (Lusardi (1999, 2002)). In this work, I
concentrate on those groups where savings are most scarce. By reducing these costs,
financial education programs may become an effective remedy for poor wealth
accumulation and naïve portfolio choice.
As shown in much of the existing literature, measuring the effects of financial
education on wealth and portfolio choice has proven to be a difficult task. The difficulty
stems mostly from the fact that attending education programs is largely voluntary. It is
therefore possible, perhaps likely, that those who attend seminars are more likely to have
an interest in them because, for example, they have large wealth holdings. Thus, it may
5
be wealth that affects retirement seminars rather than the other way around. Similarly,
attending retirement seminars could simply proxy for individual characteristics such as
patience and diligence, which are also likely to affect wealth accumulation. Moreover, as
reported by Bernheim and Garrett (2003), retirement education is often remedial and thus
offered in firms where workers do very little savings. Thus, empirical studies may be
bound to find a negative rather than a positive effect of retirement seminars. Very few
data sets have enough information to allow researchers to sort these effects out.
Consequently, empirical results about the effects of retirement seminars have been rather
mixed.
2
In my work, I use rich sources of data complemented by more recent information
on minorities to study the saving behavior and portfolio choice of African-American and
Hispanic households. In the following section, I provide a detailed overview of the data
from the 1992 wave of the Health and Retirement Study (HRS) and the 2002 National
Survey of Latinos (NSL), depicting the divergent financial behavior of African-
Americans and Hispanics. I then examine the extent of this heterogeneity in basic
financial practices by analyzing the effects of race, ceteris paribus, on checking account
ownership. Finally, I analyze the effect of employer provided seminars on saving
behavior and stock ownership among Whites and minorities.
2
See among others, McCarthy and Turner (1996), Bernheim (1995, 1998), Bayer, Bernheim and Scholz
(1996), Clark and Schieber (1998), Muller (2000); Clark and D’Ambrosio (2002), Clark, D’Ambrosio,
McDermed and Sawant (2003), and Bernheim and Garrett (2003).
6
DATA AND DESCRIPTIVE RESULTS
In order to examine the effects of financial education on minority households, I
require a dataset with descriptive demographic information and financial variables. I use
data from the 1992 wave of the HRS to illustrate the financial position of households
whose head is close to retirement when, according to the life-cycle model, households
should have amassed the most wealth. Problems of adequate sample size combined with
poor asset measurement have made research on the wealth position of minorities very
difficult, especially when focusing on a limited age-span, such as pre-retirement years.
Fortunately, the HRS oversamples African-Americans and Hispanics, offering a large and
accurate set of data about these families. Specifically, this survey covers a sample of U.S.
households whose respondents were born between 1931 and 1941 and provides detailed
information on wealth and the retirement process with a focus on health, labor markets,
and economic and psycho-social factors. Questions about wealth are asked to the
financially knowledgeable person in the household.
3
In constructing the sample from the HRS, I delete respondents who are younger
than 40 or older than 65. While the sample includes households where one member is
already partially or fully retired, the age range is such that households should be mostly at
the peak or close to the peak of their wealth accumulation. I will consider only three
racial groups: Whites, African-Americans and Hispanics and focus mostly on the latter
two.
4
3
For a thorough examination of the HRS, the quality of the data, and comparisons with other data sets, see
Juster and Smith (1997) and Smith (1995).
4
The group which is deleted is rather small and includes Native Americans, Asians or Pacific Islanders,
and other races.
7
Table 1 reports the distribution of total net worth across race. Net worth is defined
as the sum of checking and savings accounts, certificates of deposits and Treasury bills,
bonds, stocks, other financial assets, IRAs and Keoghs, housing equity, other real estate,
business equity, vehicles and subtracting all debt. As shown in the table, differences in
wealth holdings across racial groups are very large. For example, looking at medians,
wealth holdings of Whites are four to five times larger than the wealth holdings of
African-Americans and Hispanics. Similar patterns emerge when looking at means. Note
that a quarter of African-Americans and a quarter of Hispanics have minuscule amounts
of wealth. The scenario does not change when considering a different measure of wealth
holdings that excludes home and other real estates, vehicles, and business equity.
African-American and Hispanic households hold very little in terms of financial wealth
(Lusardi (1999)).
One of the reasons why wealth differs so much across racial groups is that
African-Americans and Hispanics display different educational attainment than Whites
(Table 2). While a small proportion (1.5%) of Whites has an elementary education only,
one third of Hispanics have an elementary education. Similarly, more than 50% of
Hispanics and close to 40% of African-Americans do not have a high school degree,
while only 15% of Whites do not have a high school degree. On the side of high
education attainment, while more than 20% of Whites have a college degree or higher
degree, only a small fraction (10%) of African-Americans and of Hispanics (6%) have
college or higher degrees.
The importance of education should not be understated as many studies have
shown there is a strong correlation between wealth and education, even after controlling
8
for permanent income and other demographic characteristics that can account for
preferences or economic circumstances (Browning and Lusardi (1996)). It is still unclear
why households whose head has high education accumulate much more than households
whose head has low education. However, many studies show that more highly educated
respondents are more likely to invest in high-return or tax-favored assets such as stocks
or IRAs (Haliassos and Bertaut (1995), Vissing-Jorgenson (2002), Venti and Wise
(2001)).
Table 3a reports the distribution of assets across education and race while Table
3b reports the distribution of net worth across education and race. Within each racial
group, investment in financial assets (such as bonds, stocks, and IRAs) increases sharply
with education. Note, however, that percentages are very different across racial groups.
While 13% of Whites with less than high school education invest in stocks, only 2% of
African-Americans and 2% of Hispanics with less than high school education invest in
stocks.
Low education minorities are not only less likely to invest in high-return assets,
but they do not even invest in basic financial assets, such as checking and savings
accounts. A large majority of these families are “unbanked;” only 34% of Hispanics and
38% African-Americans with elementary education have a checking and saving account
and the percentages increase little when we consider minorities with a high school
degree. This suggests that these families may find difficulties not only in accumulating
savings, but also in obtaining a loan to buy a house or start a business, as they do not
have any relationships with banks or a track record of financial transactions.
9
The distribution of wealth across race and education confirms the strong
correlation between wealth and education (Table 3b). Minorities with low education
attainment have stunningly low amounts of wealth. As many as one quarter of African-
Americans and Hispanics with less than high school education have simply no wealth. In
every racial group, it is clear that education is a strong predictor for wealth. Wealth
increases strongly as we move to higher education. However, even within high education
groups, there are large differences in wealth holdings between Whites and African-
Americans and Hispanics.
Why do African-Americans and Hispanics arrive close to retirement with so little
wealth? There are several explanations for this pattern. For example, Lusardi, Cossa, and
Krupka (2001) examine the wealth holding of a younger cohort of households from the
National Longitudinal Survey of Youth. They find that African-Americans and Hispanics
display little wealth holdings also when young. Thus, lack of savings may be a persistent
feature of African-American and Hispanic patterns of wealth accumulation, perhaps
brought on by persistent discriminatory pressures or the intergenerational transmission of
financial practices (Keister (2004)). Not only do these families start with little savings,
but they also shun away from high-return assets. Even though stock market prices
increased sharply during the 1990s, many young African-American and Hispanic families
did not benefit from the performance of the stock market; only 6-7% of young African-
Americans and Hispanics hold stocks (Lusardi, Cossa and Krupka (2001)).
The data I have used so far refer to 1992 only and one may argue that a lot
changed over the past decade. Throughout the 1990s, there was an explosion of products
and programs for financial planning. At the turn of the century, unprecedented amounts
10
of wealth were created and destroyed during what is referred to as “the Internet Bubble.”
The government instituted several programs to foster financial education and employers
increasingly offered retirement seminars to their workers. The Community Reinvestment
Act (CRA) was also reinvigorated in the late 1990’s to help markets better reach low-
income minorities (Barr (2005)). After such an economically tumultuous period, it is not
unreasonable to assume that the financial motivations and behavior of minorities may
have changed.
To analyze the financial behavior of African-Americans and Hispanics in a more
contemporary setting, I have considered data from the Pew Hispanic Center/Kaiser
Family Foundation 2002 National Survey of Latinos (NSL-2002 for brevity).
5
This
survey over-samples Hispanics and provides information on a host of issues including
finances, employment, nationality, trust in government, religion, healthcare, values,
education, and discrimination among Hispanics, African-Americans and Whites. A total
of 4,213 respondents are interviewed, and, of these, 2,929 respondents are Hispanics,
1,008 are Whites and the rest are African-Americans. Statistical weights are provided in
the dataset to correct for the oversampling of Hispanics and the relative undersampling of
African-Americans and Whites. I start by considering the total sample and will later
restrict to older respondents to be better able to compare results with the HRS.
When considering the total sample of respondents in the NSL-2002, which
includes a large share of young people (younger than 40), I find that the educational
attainment of Hispanics is still low. While only 3% of Whites and 4% of African-
Americans have only an elementary education, close to 20% of Hispanics have
elementary education only. Even in 2002, more than 43% of Hispanics have less than
5
See information on this survey on-line at http://pewhispanic.org/reports/report.php?ReportID=15
11
high school education. Hispanics continue to earn substantially less than Whites.
Interestingly, a very high proportion of Hispanics (11%) state they do not know how
much is their income. Even though the proportion of Hispanics holding checking
accounts is 64.8% in 2002, it is still much lower than the proportion of checking accounts
held by Whites (94.7%) and also lower than the proportion of checking accounts held by
African-Americans (76.2%). This is such a striking feature of this population that, in the
next section, I will examine it in more detail. Most importantly, I study whether the lack
of checking accounts can simply be explained by educational attainment and economic
characteristics of African-Americans and Hispanics or there are also other explanations
that are driving this behavior. This analysis will, in turn, inform the evaluation of the
effects of financial education programs on savings and portfolio choice.
EMPIRICAL FACTS OF CHECKING ACCOUNT OWNERSHIP
Many minorities do not own a checking account. This is particularly troublesome
because a checking account is widely considered as the most rudimentary form of
financial market integration. As Hogarth, Anguelov and Lee (2004) wrote:
Conventional wisdom holds that having a bank account is a first step toward
building a financial identity, which leads to further access to financial products
and service and then to advances in family well-being, stability, and security, and
finally to community security and economic development…[a checking account
is] the basic transaction account for most U.S. families.
Though I consider checking accounts as an indicator of a rudimentary level of financial
literacy, there may be unobserved motivations for minorities not to own them. First,
12
minorities may not gain the same advantages as Whites from owning a checking account.
For example, vendors may discriminate against checks from minorities, or minorities
may be less likely to use features like online bill paying or direct deposit. Furthermore,
banks may be difficult to get to, and therefore entail a higher cost of going to one, though
Welborn (2002) finds that bank proximity is not a major determinant of owning a
checking account. Moreover, Hogarth and O’Donnell (1999) find that a significant
number of respondents (20%) who do not have checking account “do not like dealing
with banks.”
In this section, I consider the determinants of having checking accounts and
examine whether, after accounting for preferences and economic circumstances, African-
Americans and Hispanics are still less likely to own checking accounts than Whites. In
Table 4, I report the estimates of probit regressions where I account for some basic
demographic characteristics such as age, gender and marital status. Furthermore, I
account for the education of the financial respondent and for the education of the family
of origin (whether mother or father has high school education), to account for family
background and the intergenerational transmission of financial practices. I also account
for income, wealth and wealth squared, and whether the financial respondent is retired. In
the first 2 columns of the table, I simply include dummies for being African-American
and Hispanic. In the last two columns I account for the country of origin of Hispanic
families. To proxy for the fact that households do not have accounts because they do not
have easy access to banks, I have also constructed an indicator for the availability of
banks across states (bank density hereafter).
6
I have interacted the number of banks per
6
I have obtained authorization from the HRS to use state indicators, which are not available to the general
public.
13
capita in the state with the number of banks per square kilometer in the state.
7
This
indicator takes into account that some states are large but not highly populated. As
expected, the indicator ranks highest for the District of Columbia, New Jersey,
Connecticut and lowest for Alaska, Wyoming and Montana.
As reported in Table 4, education remains an important predictor for owning a
checking account. Even after accounting for wealth and income, having high education
attainment (high school) increases the probability of owning a checking account by 15
percentage points with respect to those with very low education attainment (elementary
education). As mentioned before, more than half of Hispanics and many African-
Americans have less than high school education and this can explains the low prevalence
of checking accounts among this segment of the population. Not only does the education
of financial respondent matter, but also the education of the family of origin is important.
Having a mother or father with a high school degree increases the chances of having
checking accounts by 2.8 percentage points. Income and net worth matter and the effect
of net worth is non-linear. The results do not change when using a measure of financial
wealth or a simple dummy for whether the household has $1500 or less in financial
wealth (which corresponds roughly to the minimum amount of wealth necessary to avoid
bank fees). Even though the variable measuring bank density is not significant, it affects
the estimates of the dummy for being Hispanic.
Even after accounting for demographic characteristics, education, income and
wealth, African-Americans and Hispanics are much less likely to have checking accounts
than Whites. The dummies for African-Americans and Hispanics are not only highly
7
The information about banks is taken from the Federal Deposit Insurance Corporation and the population
statistics are from the 1990 Census.
14
significant, but they are also a powerful predictor for owning a checking account. This
confirms the earlier finding of Caskey and Peterson (1994) using data from the 1980s and
of Hogarth and Hilgert (2002) using data from the 1990s. Of course, it is not
straightforward to interpret what this dummy really captures. One explanation for the
low prevalence of checking accounts among African-Americans and Hispanics is
ethnic/racial discrimination. For example, in their famous “Boston Fed Study,” Munnel
et al (1996) found evidence of discrimination in mortgage lending. Bostic (1996)
revisited their results and found that the minorities most affected are those applicants on
the margin of approval. Recently, the Federal Reserve released a study on the latest
Home Mortgage Disclosure Act HMDA data, suggesting discrimination may still be
prevalent in mortgage lending (Avery, Canner, and Cook (2005)). Furthermore,
Blanchflower, Levine, and Zimmerman (1998) find evidence of discrimination extending
to loans for minority-owned small businesses. Perception of this discrimination may, in
turn, cause minorities to be distrustful of banks, which is important because lack of trust
is a commonly reported reason why minorities, those with low education, and the poor
choose not to hold bank accounts (Barr (2004)).
To further understand the reasons why so few minorities have checking accounts
with the information available in the HRS, I consider the country of origin of Hispanics.
As reported in Table 4, Hispanics can be divided into Mexicans, Puerto Ricans, Cubans,
South and Central Americans, and other countries. Interestingly, both Mexicans and, in
particular, South Americans are much less likely to have checking accounts than other
families. Similarly, Hispanic respondents of foreign origin may be less familiar and
confident on the working of U.S. financial markets and institutions. This finding points
15
to the fact that information and learning costs can be particular important for this group
of the population. Moreover, it indicates that there may be significant heterogeneity
within Hispanics, implying that “Hispanic” may be too broad of a term.
In order to examine the determinants of having a checking account in a more
contemporary setting, I have constructed a similar set of variables in the NSL-2002 as
those reported in Table 5a. Age, gender, marital status, regions, education, and
retirement status closely match the variables in the HRS. Income is only reported in
brackets in the 2002-NLS (plus a group of respondents who report they do not know their
income) and I grouped the income variables into less than $30,000, between $30,000 and
$50,000, and greater than $50,000. As a proxy for net worth, I use home ownership since
this generally encompasses most of low income individuals’ net worth. Finally, in the
NLS-2002, I can distinguish whether Hispanics of different countries of origin were born
abroad or in the US.
Looking at the estimates for the total sample in Table 5a, I find that, even after
accounting for many demographic characteristics and proxies for income and wealth,
African-Americans and Hispanics are still much less likely to hold checking accounts
than Whites. African-Americans are 7 percentage points less likely to have checking
accounts and Hispanics are 10 percentage points less likely to have checking accounts
than Whites. Thus, race continues to be a strong predictor for having checking accounts
even in 2002.
The persistence of racial inequality in checking account ownership may very well
be attributable, at least in part, to discrimination which may present itself as racial
animosity, statistical or perceived discrimination, or as a learned behavior from past
16
generations and childhood family conditions (Barr (2005), Swire (1995), Welborn (2002)
and Keister (2004)). It is very difficult to disentangle the importance of all these factors,
but there is some information about discrimination in the NLS-2002. Respondents are
asked whether “in the past five years have you, a family member or a close friend
experienced discrimination because of your racial or ethnic background?.” This is a
general question about discrimination which does not refer to bank lending only. The
results are staggering. More than 50% of African-American respondents in the sample
responded yes to the above question, while more than 32% of Hispanics responded
affirmatively. The percentage for Whites was instead 16%. I have added a dummy in the
regressions for those that reported they have experienced discrimination and found that
the dummies for race remain negative and statistically significant indicating that they
proxy for more than just experienced or perceived discrimination (Table 5a).
8
Another important finding in Table 5a is that foreign-born respondents are less
likely to have checking accounts than respondents born in the U.S., suggesting that
familiarity with financial institutions in the U.S. may matter. The NSL-2002 provides
information on primary language and trust in the US government, which I also include in
my regressions in addition to dummies for living in rural areas or in suburban areas
(where the density of banks should be lower). These variables do not change the values
of the estimates of interest and, for brevity, these regressions are not reported.
To compare results with the HRS, in Table 5b, I exclude respondents younger
than 40. The sample decreases substantially but I still have sizable fractions of Hispanics
and Whites in the sample, while the number of African-Americans becomes very small.
8
I have also interacted the dummy for discrimination with the dummies for African-Americans and
Hispanics, but the interaction terms are not statistically significant. For brevity, estimates are not reported.
17
Even after a 10-year period, Hispanic households are still “unbanked” and much less
likely than Whites to have checking accounts. Hispanics which are not born in the US,
and especially those born in Mexico, are disproportionately less likely to have checking
accounts. Thus, the road to assimilation is still a long one and many Hispanic do not yet
fully access and utilize the banking system in the U.S.
The lack of participation to financial market is damaging insofar as it inhibits
gaining experience in financial matters. This may result in and contribute to low levels of
financial literacy and knowledge, which are correlated with poor financial behavior
(Hilgert, Hogarth and Beverly (2003)). Hogarth, Anguelov and Lee (2004) and Caskey
(2002) recommend financial education as a method to improve financial behavior for
low-income and minority households. The HRS provides information on financial
education and I now turn to examine the effects of these programs among different racial
groups.
THE EFFECT OF FINANCIAL EDUCATION ON SAVINGS AND PORTOLIO
CHOICE OF MINORITIES
1) SAVINGS
To remedy lack of savings among workers, many employers have started offering
some form of financial education in the workplace. By improving financial literacy,
seminars should reduce information costs and foster savings. While several studies have
found a positive correlation between attending a retirement seminar and private wealth or
contributions to pension funds, it is not clear what this correlation means (see Berheim
and Garrett (2003) and the references therein).
18
The empirical strategy for estimating the effect of financial education is based on
the following specification:
W
i
/Y = α
0
+ α
1
African-Americans
i
+ α
2
Hispanics
i
+ α
3
Fin Educ
i
+ Z
i
β + u
i
where W/Y is a measure of household wealth normalized by permanent income. African-
Americans and Hispanics are dummies for race. The Z vector includes demographic
characteristics. The controls are included to capture potential differences in preferences
and economic circumstances across households and the hump-shaped profile of wealth
over the life cycle. Fin Educ is a dummy for financial education as will be explained
below. If financial literacy and planning costs matter, we expect this variable to have a
positive effect on savings.
There are two innovative features for assessing the effects of financial education
on savings using data from the HRS. First, I can account for a rich set of controls for
wealth, which are not present in other data (Z). Second, I can look not only at wealth
accumulation but also at pension wealth and portfolio choice (W). This is where lack of
information and financial literacy is potentially more relevant and binding.
A criticism often raised in the empirical work on savings is that researchers use a
very restrictive version of the life-cycle model and it becomes, perhaps, too easy to find
evidence against the predictions of standard saving models. One of the advantages of
using the HRS is that it provides a rich set of information on individual respondents. This
allows researchers to examine many of the reasons for household behavior towards
savings. I provide below the set of controls (Z) I will use in the empirical regressions,
starting with the more original ones that are not present in other data sets:
19
A. Controls
a.1) Expectations about the future: Savings are inherently related to the future. Thus, it
is critically important to account for expectations about the future. In the empirical
regressions, I account for the probabilities that home prices will increase more than
the increase in the general price level and that Social Security will become less
generous in the future. Since these are two of the major components of total
household total accumulation, leaving this information out may lead to regressions
that have little explanatory power. I will also include the probability of living up to
75 since expected longevity is clearly a predictor for wealth. Other studies have
shown that this variable well match the mortality tables.
9
a.2) Other motives to save. Households may save not only to offset the decline in income
at retirement but also to provide for the extended family or to leave bequests. I use
the subjective expectations of giving major financial help to family members in the
next 10 years to account for support of the extended family. In addition I account
for the desire to leave a bequest by using the information on whether respondents
are likely to leave a bequest to their children.
a.3) Past economic circumstances: In addition to future events, respondents in the HRS
are asked to provide information on past economic circumstance such as past
shocks. I account for these shocks by adding dummies for whether respondents
have been unemployed in the past and whether they faced any episodes that made it
difficult to meet financial needs. I also account for positive shocks and add a
9
An excellent examination of subjective probabilities in the HRS is provided in Hurd and McGarry (1995).
20
dummy for whether respondents have received inheritances.
10
These positive and
negative shocks are another important explanation for the wide differences in
wealth holdings that we observe empirically.
a.4) Pension wealth: Using the HRS, it is possible to calculate pension wealth from the
self-reported pension information.
11
Thus, in my work, I can rely on an extended
measure of household resources when examining saving behavior.
a.5) Preferences: Another not yet well explored dimension along which households can
differ is preferences. While it is very hard to measure individual preferences, it is
also the case that parameters, such as the coefficient of risk aversion or the rate of
time preference, play a pivotal role in many models of intertemporal optimization.
There is a way to infer this information in the HRS, and therefore to account for
variation in preferences when explaining household wealth holdings. In particular, I
use the analysis provided in Barsky, Kimball, Juster, and Shapiro (1997) on
willingness to take gambles to construct proxies for the coefficient of risk aversion.
I also use data on planning horizons (medium and long horizons corresponding to
horizons of 5 years and horizons longer than 5 years) to proxy for the rate of time
preference and/or individual heterogeneity. Demographic variables that are related
to the rate of time preference, such as education, race, and country of origin, are
also included in the empirical estimation (Lawrance (1991)).
a.6) Permanent income. To construct a measure of permanent income, I regress total
household income on a set of demographics and firm characteristics. I use age and
10
Since the focus is on saving behavior of minorities, I did not consider transfers from relatives or from
insurance settlements as these transfers are rather rare among minorities.
11
For a detailed explanation of the construction of the pension data, see Venti and Wise (2001).
21
age squared, sex, race, marital status, regions of residence, education and
occupation dummies individually and interacted with age. I also use dummies for
whether the respondent works in a small firm (fewer than 20 employees), whether
the respondent belongs to a union, and whether he/she works part time. In addition,
I use dummies for whether income will go up or down in the following year.
12
a.7) Checking accounts. As I mentioned above, some households do not hold basic assets
and do not participate to financial markets. I account for this important fact by
adding a dummy for whether the family has checking accounts.
a.7) Business ownership. As my previous study shows, business owners are different
than other households in both preferences and motives to save (Hurst and Lusardi
(2004)). In the regressions, I always add a dummy for owning a business.
a.8) Additional controls. I have also added a variable for bank density across states and
for whether the financial respondent is already retired.
B. Financial Education
b.1) Seminars offered by employers: The HRS reports information on whether
respondents (or spouse) have ever attended a retirement seminar and asks who offered the
seminar. I have defined a dummy that takes the value one if respondents have indicated
they attended a seminar offered by the employer. This is the critical variable for my
empirical work (Fin Educ).
Given that not every household reports the information described above, I have to
perform additional exclusions on my sample. First, a few respondents do not report
information on subjective future probabilities or pensions so I deleted these observations
from the sample. Since the distribution of the ratio of savings to permanent income is so
12
This is the specification used in Lusardi (2003b) and Lusardi (2004).
22
wide, I trim it and exclude the top and bottom 1%. I also exclude those business owners
with large amount of business equity (business equity greater than $2,000,000). The total
sample has 6,297 observations while the samples of African-Americans and Hispanics
have 1,183 and 530 observations respectively.
13
In the empirical work reported below, I consider the patterns of accumulation in
the total sample and across racial groups to examine the major differences between
African-American and Hispanic families and other families. Estimates are reported in
Table 6.
There are several important results to discuss. First, even after accounting for a
large set of economic and demographic characteristics and a large set of determinants of
wealth, African-American and Hispanic families still hold less wealth than White
families. In particular, African-Americans, hold much lower amounts of wealth than
Whites, ceteris paribus. Thus, there are other reasons not accounted for in the model and
the empirical specifications for why minorities save less than other families. Moreover,
several of the variables that affect savings among Whites have little or no effect for
African-Americans and Hispanics. For example, receiving inheritances has a strong effect
on savings for White families but little or no effect on African-Americans and Hispanics,
who are less likely to receive inheritances. Help in the other direction, i.e., expectations
of giving major financial help to family members in the next 10 years, affects savings of
the Whites but not savings of African-Americans and Hispanics. Having a checking
account or owning a business has an effect in the total sample and among Whites, but the
effect is particularly strong and large among African-Americans and Hispanics. Indeed,
13
See also the data appendix for more detail on the final sample and some descriptive statistics of the
variables used in the empirical work.
23
those minorities that do not have a checking account do not have much else (their total
wealth is very low).
The effect of seminars is positive and statistically significant in the total sample
and for White families only. To fully assess the effects of seminars, however, we need to
rely on a different estimation strategy. If, as suggested by Bernheim and Garrett (2003),
seminars are remedial and offered by firms where workers most need it, we should be
more likely to find an affect at the bottom of the wealth distribution and among those
with low education. Since low wealth and low education workers have usually very little
wealth, it is hard to argue that the causality goes the other way.
In Table 7, I report quartile estimates and examine the effects of seminars across
three quartiles of the wealth distribution. In Table 8, I report median estimates for low
and high education groups.
14
For brevity, only the estimates of retirement seminars are
reported. Consistent with the facts that retirement seminars are remedial, there is an effect
of retirement seminars for African-Americans, but only in the first quartile of the wealth
distribution. Given that the families at the bottom of the wealth distribution save so little,
this is still a remarkable effect. Similarly, the effect of seminars is positive and significant
among both Whites and African-Americans, but only for those with low education. The
effect is not significant for Hispanics, but very few low education and low wealth
Hispanics have ever attended a seminar. Note that, as reported in the Data Appendix,
only 5% of Hispanics ever attended retirement seminars while 13% of Whites and 12% of
African-Americans attended retirement seminars.
14
Given that the sample of African-Americans and Hispanics is small, I cannot do quantile estimation
among education group and I consider medians only.
24
Total net worth is a partial measure of accumulation because many families also
have pension wealth. In Table 9, I report estimates in the total sample and across quartiles
when considering an enlarged measure of wealth that includes, in addition to total net
worth, the self-reported value of pensions. Similar to the findings of Lusardi (2003b),
retirement seminars affect this measure of retirement wealth across the whole
distribution. However, the effect is present only for Whites and African-Americans but
not for Hispanics. These results overall confirm the previous findings, but should be used
with caution. As reported by Gustman and Steinmeier (2004), workers do not seem well
informed about their pensions. Only half of respondents with linked pension data to their
employer correctly identify their pension plan (whether it is Defined Benefits, Defined
Contributions or a mix of the two) and fewer than half identify, within one year, dates of
eligibility for early and normal retirement benefits. Earlier papers had also suggested that
workers are less than fully informed about their pensions (Mitchell (1988) and Gustman
and Steinmeier (1989)). Information about Social Security seems also scanty. Only 43%
of respondents in the HRS even ventured a guess about their expected Social Security
benefits and many respondents knew little about the rules governing Social Security.
Given these findings and the fact that Social Security wealth is highly illiquid and one
cannot borrow against it, I have not considered a measure of total retirement savings that
includes Social Security wealth in addition to pensions.
To put estimates in perspective, I have examined the effects of retirement
seminars across other relevant determinants of wealth. For African-Americans in the first
quartile of the net worth to permanent income ratio distribution, attending a seminar has
as large an effect as having received inheritances, holding a checking account, or having
25
a long planning horizon. When looking at those with low education, seminars have a
similar effect as having very good health or not having been unemployed in the past.
Given the inherent difficulties or costs of changing these other variables, retirement
seminars may represent a viable alternative to stimulate savings.
Other studies, such as Garman (1998) and the references therein, have argued that
financial education increases workers’ productivity and reduce absenteeism to deal with
personal financial matters. The value to employers of these benefits of financial
education is estimated at around $400, a figure easily above the costs of providing
financial education to each worker. While these studies are often qualitative and based on
small samples, they represent some further evidence in support of financial education
programs.
2) PORTFOLIO CHOICE
Portfolio choice can reveal a great deal about household behavior, and it is here
that we may be able to detect the effectiveness (or lack of effectiveness) of financial
education. Even though stocks have outperformed bonds historically, only a relatively
small fraction of households invest in stocks. In fact, an important puzzle is why so few
households hold stocks.
15
Additionally, many household portfolios seem rather
unsophisticated (Lusardi (1999)). If much effort has to be exerted to obtain information
about complex investment assets, such as stocks, agents facing high costs or displaying
little financial literacy will be less likely to invest in those assets.
15
See the discussion in Haliassos and Bertaut (1995) and Vissing-Jorgensen (2002).
26
The dependent variable in my regressions is now a dummy variable for whether
households hold stocks. As for the previous regressions on household savings, I have
considered a large set of controls that can proxy for both household resources and
preferences that can explain stock-ownership.
16
Though not reported, I have accounted
also for age, gender, marital status, number of children, and health status. In addition, I
have added an indicator of bank density across states.
The estimates I obtain for the total sample are consistent with other work on stock
ownership.
17
For example, households with higher wealth and permanent income are
more likely to invest in stocks (Vissing-Jorgensen (2002)). As reported before, total net
worth is often so small for many households that they are unlikely to invest it in stocks.
Respondents with pensions are more likely to invest in stocks. Furthermore, respondents
with pensions usually have to choose how to invest their pension assets, and studies such
as Weisbenner (2002) suggest that this may also affect the allocation of their non-pension
assets. Consistent with the estimates of Heaton and Lucas (2000), households who own a
business and, thus, face high income risk are less likely to invest in stocks.
The most relevant result, however, is that even after controlling for many factors
that can explain stock ownership, African-Americans and Hispanics are much less likely
to invest in stocks than Whites. African-Americans are 13 percentage points less likely to
hold stocks, while Hispanics are 10 percentage points less likely to hold stocks than
Whites. This result could be indicative of many reasons, including pervasive mistrust
16
The variables measuring planning horizons are now divided into medium or long horizon (5 years or
longer) and short horizons (few years). The reference group is respondents with horizons of 1 year or less. I
add these variables to be consistent with the existing work on portfolio choice (see, for example, Halissos
and Bertaut (1995)).
17
See Haliassos and Bertaut (1995), Heaton and Lucas (2000), and Vissing-Jorgensen (2002).
27
between minorities and equity markets. Moreover, financial education is only significant
for Whites, being small and insignificant for African-Americans and Hispanics. Results
remain the same when examining sub-groups of the population, i.e., those with lower
wealth (lower than the median) and lower education (high school or less). So few low-
wealth and low-education minorities hold stocks that almost no variable has explanatory
power among these sub-groups.
When looking at the estimates within African-Americans and Hispanics only
(Table 10, last 2 columns), I find that most of the conventional variables that can explain
stock-ownership among Whites, such as education and risk aversion, have little or no
predictive power among minorities. Notably, country of birth is a predictor for stock-
ownership among minorities. Those born in the U.S. are 2-3 percentage points more
likely to hold stocks. Overall, for minorities, permanent income, wealth and having a
pension and a checking account are the most important predictors for stock-ownership.
DISCUSSION
The estimates above provide some, albeit limited, evidence that retirement
seminars can have an effect in stimulating savings among groups of the population,
specifically African-Americans, where savings are scarce. To fully evaluate these
estimates, one should keep in mind at least three issues about the empirical work
described above and the work on retirement seminars in general. First, no information is
provided in the HRS about when the seminar was attended. If respondents attended the
seminars recently, the effect may not have shown up in wealth yet. Second, no
information is provided on the number or content of seminars. If respondents attended
28
one seminar only, we should not be surprised about finding small effects. One of the
lessons we have learned from the literature on savings is that there is large heterogeneity
in saving behavior. One session and “one-size-fits-all” education programs may do little
to stimulate saving and may itself be a major disincentive to attend a financial education
program. Third, the variable may be measured with error if respondents do not recall well
what they did in the past. This also leads to a downward bias in the estimates.
There is also another important feature of the empirical work. Attending a
retirement seminar is clearly a decision variable rather than an exogenous variable. There
are two ways around the potential endogeneity of financial programs: perform instrument
variables estimation or run randomized experiments. Lusardi (2003b) undertook the first
strategy and used the densities of large firms across states as instrument for the
availability (rather than the use) of seminars in addition to the age differences between
respondents and their older siblings to proxy for planning costs. I found that the effect of
seminars is larger than the estimates reported in Tables 6 and 9. The estimates reported
above may again be considered a lower bound on the effects of seminars.
18
Another
approach to evaluating the effects of financial education programs is to run experiments,
where a randomly chosen group of participants is exposed to financial education and their
behavior is then compared to an otherwise similar group which was not exposed to the
program (control group). This is the approach taken by Duflo and Saez (2003). A random
group of non-faculty employees at a large university were given financial incentives to
participate to a benefit fair. Participation to pensions and pension contributions of this
group were then compared to those who were not induced to participate. According to the
18
This approach cannot be used in this paper because the samples of African-Americans and Hispanics are
small and instruments do not have enough predictive power in the first stage regression.
29
authors (Duflo and Saez (2003 and 2004)), the effects of this program are mixed and
overall pretty small. Attending the benefit fair induced more employees to participate to
pensions but the increase in contributions was negligible.
These types of experiments have the same, perhaps even more severe, problems
as the empirical estimates reported above. First, if financial illiteracy is widespread and
individuals know very little about financial matters, attending a benefit fair is unlikely to
affect behavior. Moreover, as mentioned previously, a one-time exposure to financial
education may do little to affect savings. This is not because financial education is
ineffective but because the “cure is not adequate for the disease.”
To best evaluate the effects of seminars, we need a good understanding of the
obstacles people face when planning for retirement. Designing financial programs and
evaluating those programs is intimately intertwined with understanding the determinants
of savings. This argument is particularly important for savings among African-Americans
and Hispanics. These families exhibit distinctly different patterns of accumulation than
White families. Given how resilient low savings are among these groups of the
population (they save little both when young as well as when old), financial education
programs should be designed for and targeted to these segments of the population. One
feature that has emerged throughout the paper, when considering closely African-
Americans and Hispanics, is that they neither invest in high-return assets nor hold basic
assets such as checking accounts. This may reflect a basic lack of information and
experience in the working of financial markets. Thus, for these families, financial
education programs may more effective if they were better able to address very basic
financial knowledge and needs.
30
CONCLUSIONS
This paper examines whether retirement seminars help explain the wide
differences in retirement accumulation that we observe across older households and
across race. The estimates presented in this paper show that seminars have some effect on
savings, particularly for those at the bottom of the wealth distribution, and those with low
education. However, only African-Americans are affected by financial education while
the behavior of Hispanics seems largely unaffected by these programs. Financial
education does not affect the portfolio choice of minorities, adding to the puzzle why so
many African-Americans and Hispanics do not hold stocks or even basic assets such as
checking accounts.
The behavior of minorities may be influenced by complex issues such as
discrimination and cultural preference. Unfortunately, even though the HRS is rather rich
in providing information and over-sampling minorities, it does not contain data that
would allow me to control for such influences. Regardless, these findings suggest that
education programs offered by the government or employers have to focus on basic
financial planning strategies and, in the case of minorities, be more targeted to their
specific needs. One finding that emerges throughout the paper is that the financial
behavior of African-Americans and Hispanics is very different than the behavior of
Whites. Thus, to be effective, financial education programs should be tailored to these
groups of the population and address lack of financial knowledge and experience in
dealing with financial markets. Moreover, while the provision of information and the
reduction of planning costs could play an important role in improving the financial
security of many U.S. households, it should be recalled that only a small number of
31
workers currently attends retirement seminars. Consequently, many remain untouched by
employers’ efforts to provide financial education. This fact represents an important topic
for future research and policy intervention.
32
DATA APPENDIX
The data used in this paper are from the first wave of the Health and Retirement
Study (HRS) in 1992. The HRS is a representative sample of individuals born in the year
1931-1941 (approximately 51-61 at interview), but African-Americans, Hispanics, and
Floridians were over-sampled. The individual deemed most knowledgeable about the
family’s assets, debts, and retirement planning was asked questions on housing, wealth,
and income.
An important innovation of the HRS is the use of bracketing or unfolding
techniques to reduce the size of the missing data problem in the measurement of financial
variables. It is well known that missing data represent a major problem in survey
measurements of household wealth. In the HRS, respondents who reported they did not
know or refused to provide an estimate of the size of a net worth component were asked
to report the value in a set of brackets. Smith (1995) and Juster and Smith (1997) report
an evaluation of these techniques and a detailed description of their advantages in
improving the accuracy of information about household wealth.
To construct the final sample, I deleted the respondents who do not report
information on the variables used in the empirical estimation. I also deleted races other
than Whites, African-Americans and Hispanics. Since the distribution of the ratio of total
net worth to permanent income is so wide, I trim the distribution and exclude the top and
bottom 1%. I also delete respondents with large amounts of business equity (2,000,000 or
more). The following table reports simple statistics of the variables used in the empirical
estimation.
33
Table A1: Descriptive Statistics
Whites
Mean (s.d.)
African-Americans
Mean (s.d.)
Hispanics
Mean (s.d.)
Net Worth/ permanent income 3.72 (4.09) 1.84 (2.88) 2.40 (3.56)
Net worth + pension / perm inc 5.33 (4.84) 3.25 (3.99) 3.26 (4.33)
Stock Ownership 0.34 (0.47) 0.09 (0.28) 0.07 (0.27)
Attended Retirement Seminars 0.13 (0.34) 0.12 (0.32) 0.05 (0.22)
Has checking account 0.89 (0.31) 0.58 (0.49) 0.50 (0.50)
Has a business 0.18 (0.38) 0.07 (0.25) 0.09 (0.29)
Permanent Income/1000 50.91 (20.9) 34.00 (20.8) 30.78 (18.2)
Age 55.43 (4.26) 55.13 (4.07) 54.80 (4.06)
Number of children 3.05 (1.91) 3.58 (2.54) 3.92 (2.68)
Married 0.71 (0.45) 0.44 (0.50) 0.64 (0.48)
Male 0.53 (0.50) 0.39 (0.49) 0.48 (0.50)
US born 0.96 (0.19) 0.95 (0.20) 0.50 (0.50)
Excellent health 0.26 (0.44) 0.12 (0.32) 0.18 (0.39)
Very good health 0.31 (0.46) 0.23 (0.42) 0.16 (0.37)
Good health 0.26 (0.44) 0.32 (0.46) 0.29 (0.45)
Past unemployment 0.30 (0.46) 0.29 (0.45) 0.39 (0.49)
Past shocks 0.34 (0.47) 0.27 (0.45) 0.29 (0.45)
Received inheritances 0.24 (0.43) 0.04 (0.21) 0.05 (0.22)
High risk aversion 0.63 (0.48) 0.66 (0.47) 0.58 (0.49)
Moderate risk aversion 0.13 (0.33) 0.11 (0.31) 0.09 (0.29)
Medium risk aversion 0.11 (0.31) 0.09 (0.29) 0.14 (0.35)
Expectations to live to 75 0.64 (0.29) 0.65 (0.32) 0.55 (0.34)
Expectations of house prices 0.46 (0.28) 0.56 (0.33) 0.58 (0.31)
Expectations about SS 0.61 (0.29) 0.52 (0.33) 0.53 (0.35)
Expectations to give finan. Help 0.39 (0.31) 0.43 (0.35) 0.42 (0.35)
Medium horizon 0.31 (0.46) 0.22 (0.41) 0.17 (0.37)
Long horizon 0.10 (0.29) 0.06 (0.24) 0.04 (0.19)
Parents are still alive 0.67 (0.47) 0.59 (0.49) 0.65 (0.48)
Expect to leave bequest 0.41 (0.49) 0.45 (0.50) 0.39 (0.49)
Has pension 0.43 (0.49) 0.38 (0.48) 0.27 (0.44)
Can rely on help from family 0.42 (0.49) 0.38 (0.48) 0.38 (0.48)
West region 0.19 (0.39) 0.10 (0.30) 0.42 (0.49)
Midwest region 0.27 (0.44) 0.20 (0.40) 0.06 (0.24)
Northeast region 0.22 (0.42) 0.22 (0.41) 0.14 (0.35)
# of observations 4,584 1,183 530
34
REFERENCES
Avery, Robert, Glenn Canner, and Robert Cook (2005), “New Information Reported
Under HMDA and its Application in Fair Lending Enforcement,” Federal Reserve
Bulletin, September.
Barr, Michael (2004), “Banking the Poor,” Yale Journal on Regulation 21(121), pp. 123-
238.
Barr, Michael (2005), “Modes of Credit Market Regulation,” in N. Retsinas and E.
Belsky, (eds.), Building Assets, Building Wealth: Creating Wealth in Low-Income
Communities, Washington, DC: Brookings Institution.
Barsky, Robert, Miles Kimball, Thomas Juster and Mathhew Shapiro (1997), “Preference
Parameters and Behavioral Heterogeneity: An Experimental Approach in the Health and
Retirement Study,” Quarterly Journal of Economics, 62, pp. 537-579.
Bayer, Patrick, Douglas Bernheim, and J. Karl Scholz (1996), “The Effects of Financial
Education in the Workplace: Evidence from a Survey of Employers,” NBER Working
Paper n. 5655.
Bernheim, Douglas (1995), “Do Households Appreciate Their Financial Vulnerabilities?
An Analysis of Actions, Perceptions, and Public Policy,” Tax Policy and Economic
Growth, Washington, DC: American Council for Capital Formation, pp. 1-30.
Bernheim, Douglas (1998), “Financial Illiteracy, Education and Retirement Saving,” in
Olivia Mitchell and Sylvester Schieber (eds.), Living with Defined Contribution
Pensions, Philadelphia: University of Pennsylvania Press, pp. 38-68.
Bernheim Douglas and Daniel Garrett (2003), “The Effects of Financial Education in the
Workplace: Evidence from a Survey of Households,” Journal of Public Economics
, 87,
pp. 1487-1519.
Blanchflower, David and Andrew Oswald (1998), “What Makes an Entrepreneur?,”
Journal of Labor Economics, 16, pp. 26-60.
Blanchflower, David, Phillip Levine, and David Zimmerman (1998), “Discrimination in
the Small Business Credit Market,” NBER Working Paper n. 6840.
Bostic, Raphael (1996), “The Role of Race in Mortgage Lending: Revisiting the Boston
Fed Study,” Working Paper, Federal Reserve Board of Governors.
Browning. Martin and Annamaria Lusardi (1996), “Household Saving: Micro Facts and
Micro Theories,” Journal of Economic Literature, 34, pp. 1797-1855.
35
Caskey, John (2002), “Bringing Unbanked Households Into the Banking System,”
Capital Xchange Journal, 2002.
Caskey John and Andrew Peterson (1994), “Who Has a Bank Account and Who Doesn’t:
1977 and 1989,” Eastern Economic Journal, 20, pp. 61-73.
Charles, Kerwin and Erik Hurst (2002), “The Transition to Home Ownership and the
Black/White Wealth Gap,” Review of Economics and Statistics, 84, pp. 281-297.
Charles, Kerwin and Erik Hurst (2003), “The Correlation of Wealth Across Generations,”
Journal of Political Economy, 111, pp. 1155-1182.
Clark, Robert, Madeleine D’Ambrosio (2002), “Saving for Retirement: The Role of
Financial Education,” TIAA-CREF Institute Working Paper n. 4-070102-A.
Clark, Robert, Madeleine D’Ambrosio, Ann McDermed, Kshama Sawant (2003),
“Financial Education and Retirement Saving,” TIAA-CREF Institute Working Paper 11-
020103.
Clark, Robert and Sylvester Schieber (1998), “Factors Affecting Participation Rates and
Contribution Levels in 401(k) Plans,” in Olivia Mitchell and Sylvester Schieber (eds.),
Living with Defined Contribution Pensions, Philadelphia: University of Pennsylvania
Press, pp.69-97.
Duflo, Esther and Emmanuel Saez (2003), “The Role of Information and Social
Interactions in Retirement Plan Decisions: Evidence from a Randomized Experiment,”
Quarterly Journal of Economics, 118, pp. 815-842.
Duflo, Esther and Emmanuel Saez (2004), “Implications of Pension Plan Features,
Information, and Social Interactions for Retirement Saving Decisions,”in Olivia Mitchell
and Stephen Utkus (eds.), Pension Design and Structure: New Lessons from Behavioral
Finance, Oxford: Oxford University Press, pp. 137-153.
Evans, David and Linda Leighton (1989), “Some Empirical Aspects of
Entrepreneurship,” American Economic Review, 79, pp. 519-535.
Fairlie, Robert (1999), “The Absence of the African-American Owned Business: An
Analysis of the Dynamics of Self-employment, Journal of Labor Economics
, 17, pp. 80-
108.
Ferguson, Michael and Stephen Peters (1995), “What Constitutes Evidence of
Discrimination in Lending?” Journal of Finance, 50(2), pp. 739-748.
Garman, Thomas (1998), “The Business Case for Financial Education,” Personal
Finances and Worker Productivity, 2, pp. 81-93.
36
Gentry, William and Glenn Hubbard (2004), “Entrepreneurship and Household
Saving,”Advances in Economic Analysis & Policy, 4, article 8.
Gustman, Alan, Olivia Mitchell, Andrew Samwick and Thomas Steinmeier (1999),
“Pension and Social Security Wealth in the Health and Retirement Study,” in James
Smith and Robert Willis (eds.), Wealth, Work and Health. Ann Arbor: University of
Michigan Press.
Gustman, Alan and Tom Steinmeier (1989), “An Analysis of Pension Benefits Formulas,
Pensions Wealth and Incentives from Pensions, Research in Labor Economics, 10, 53-
106.
Gustman, Alan and Tom Steinmeier (2004), “What People Don’t Know about their
Pensions and Social Security,” in William Gale, John Shoven and Mark Warshawsky
(eds), Private Pensions and Public Policies,” Washington, DC: Brookings Institution, pp.
57-125.
Haliassos Michael and Carol and Bertaut (1995), “Why Do So Few Hold Stocks?”
Economic Journal, 105, pp. 1110-1129.
Heaton, John and Deborah Lucas (2000), “Portfolio Choice and Asset Prices: The
Importance of Entrepreneurial Risk,” Journal of Finance, 55, pp 1163-1197.
Hilgert, Marianne, Jeanne Hogarth, and Sondra Beverly (2003), "Household Financial
Management: The Connection between Knowledge and Behavior," Federal Reserve
Bulletin.
Hogarth, Jeanne, Chris Anguelov, and Jinkook Lee (2004), “Why Don’t Households
Have a Checking Account?” The Journal of Consumer Affairs, 38(1).
Hogarth, Jeanne and Marianne Hilgert (2002), "Financial Knowledge, Experience and
Learning Preferences: Preliminary Results from a New Survey on Financial Literacy,"
Consumer Interest Annual, 48.
Hogarth, Jeanne and Kevin O’Donnell (1999), “Banking Relationship of Lower-Income
Families and the Government Trend Toward Electronic Payment,” Federal Reserve
Bulletin, 85, pp. 459-473.
Holtz-Eakin, Douglas, David Joulfaian and Harvey Rosen (1994b), “Entrepreneurial
Decisions and Liquidity Constraints,” Rand Journal of Economics
, 25, pp. 334-347.
Hubbard, Glenn, Jonathan Skinner and Stephen Zeldes (1995), “Precautionary Saving
and Social Insurance,” Journal of Political Economy
, 103, pp. 360-399.
37
Hurd, Michael and Kathleen McGarry (1995), “Evaluation of the Subjective Probabilities
of Survival in the Health and Retirement Study,” Journal of Human Resources, 30, pp.
268-292.
Hurst, Erik, Ming Ching Luoh and Frank Stafford (1998), “The Wealth Dynamics of
American Families: 1984-1994,” Brookings Papers on Economic Activity, 1.
Hurst, Erik and Annamaria Lusardi (2004), “Liquidity Constraints, Households Wealth
and Entrepreneurship,” Journal of Political Economy
, 112, pp. 319-347.
Juster, Thomas and James Smith (1997), “Improving the Quality of Economic Data:
Lessons from the HRS and AHEAD,” Journal of the American Economic Association
,
92, pp. 1268-1278.
Keister, Lisa (2004), “Race, Family Structure, and Wealth: The effect of Childhood
Family on Adult Asset Ownership,” Sociological Perspectives, 47(2), pp. 161-187.
Lawrance, Emily (1991), “Poverty and the Rate of Time Preference: Evidence from Panel
Data,” Journal of Political Economy, 99, pp. 54-77.
Longhofer, Stanley and Stephen Peters (2005), “Self-Selection and Discrimination in
Credit Markets,” Real Estate Economics, 33, pp.237-268.
Lusardi, Annamaria (1999),"Information, Expectations, and Savings for Retirement," in
Henry Aaron, ed., Behavioral Dimensions of Retirement Economics, Washington, D.C.:
Brookings Institution Press and Russell Sage Foundation, 1999, pp. 81-115.
Lusardi, Annamaria (2000), “Saving for Retirement: The Importance of Planning,”
Research Dialogue, n 66, TIAA-CREF Institute, December.
Lusardi, Annamaria (2002), “Increasing Saving among the Poor: The Role of Financial
Literacy,” Poverty Research News
, Northwestern/University of Chicago Joint Center for
Poverty Research, vol. 6(1), January-February.
Lusardi, Annamaria (2003a), "Planning and Saving for Retirement," Working Paper,
University of Chicago and Dartmouth College.
Lusardi, Annamaria (2003b), “Preparing for Retirement. The Importance of Planning
Costs,” Report for the US Department of Labor, Employee Benefits Security
Administration.
Lusardi, Annamaria (2004), “Saving and the Effectiveness of Financial Education,” in O.
Mitchell and S. Utkus (eds.), “Pension Design and Structure: New Lessons from
Behavioral Finance,” Oxford: Oxford University Press, 2004, pp. 157-184.
Lusardi, Annamaria, Ricardo Cossa and Erin Krupka (2001), “Savings of Young
38
Parents,” Journal of Human Resources, 36, pp. 762-794.
Mabry, Tristan (1999). “Black Investors Shy Away From Stocks,” Wall Street Journal,
May 14, 1999.
McCarthy, David and John Turner (1996), “Financial Sophistication, Saving and Risk
Bearing,” mimeo, U.S. Department of Labor.
Meyer, Bruce (1990), “Why Are There So Few African-American Entrepreneurs?”
NBER Working Paper n. 3537.
Mitchell, Olivia (1988), “Worker Knowledge of Pensions Provisions,” Journal of Labor
Economics, 6, pp. 28-29.
Muller, Leslie (2000), “Investment Choice in Defined Contribution Plans: the Effects of
Retirement Education on Asset Allocation,” Social Security Administration Working
Paper.
Munnel, Alicia, Geoffrey Tootell, Lynn Brown, and Jame McEneayney (1996),
“Mortgage Lending in Boston: Interpreting HMDA Data,” American Economic Review,
86(1), pp. 25-53.
Smith, James (1995), “Racial and Ethnic Differences in Wealth in the Health and
Retirement Study,” Journal of Human Resources, 30, pp. 159-183.
Swire, Peter (1995), “The Persistent Problem of Lending Discrimination: A Law and
Economics Analysis,” Texas Law Review, 73.
Venti, Steven and David Wise (2001), “Choice, Chance, and Wealth Dispersion at
Retirement,” in S. Ogura, T. Tachibanaki and D. Wise (eds.), Aging Issues in the United
States and Japan, Chicago: University of Chicago Press.
Vissing-Jorgensen, Annette (2002), “Towards an Explanation of Household Portfolio
Choice Heterogeneity: Nonfinancial Income and Participation Cost Structure.” Mimeo,
Northwestern University.
Weisbenner, Scott (2002), “Do Pension Plans with Participant Investment Choice Teach
Households to Hold More Equity?” Journal of Pension Economics and Finance
,1, pp.
223-248.
Welborn, John (2002), “The Checking Account Puzzle: The Role of Race and Family
History in the Formation of Individual Banking Practices,” Senior Honors Thesis,
Dartmouth College.
39
Table 1
Distribution of Total Net Worth Across Race
Distribution Whites African-Americans Hispanics
10
th
percentile 6,200 -88 0
25
th
percentile 45,942 100 850
Median 125,000 27,500 30,000
75
th
percentile 274,900 82,000 94,000
90
th
percentile 583,400 167,000 213,835
95
th
percentile 1,009,000 271,000 347,500
Mean
(Std.dev.)
275,271
(565,713)
81,470
(243,423)
103,191
(410,946)
# obs 5,116 1,382 682
Note: This table reports the distribution of total net worth across race in the HRS. All
figures are weighted using survey weights.
Table 2
Education Attainment Across Race
Education Level Whites African-Americans Hispanics
Elementary 0.015 0.068 0.327
Less than High School 0.148 0.330 0.247
High School 0.398 0.307 0.213
Some College 0.211 0.183 0.150
College 0.125 0.057 0.040
More than College 0.101 0.053 0.022
# obs 5116 1382 682
Note: This table reports the proportion of respondents in each racial and education groups
in the HRS. All figures are weighted using survey weights.
40
Table 3a
Asset Ownership Across Race and Education
Whites African-Americans Hispanics
Less than
HS
High
School
More
Than HS
Less than
HS
High
school
More than
HS
Less than
HS
High
School
More than
HS
Checking 0.70 0.90 0.94 0.38 0.59 0.77 0.34 0.58 0.77
CDs 0.15 0.31 0.35 0.06 0.14 0.22 0.02 0.15 0.18
Bonds 0.01 0.04 0.14 0.00 0.01 0.02 0.00 0.01 0.03
Stocks 0.13 0.28 0.49 0.02 0.07 0.19 0.02 0.09 0.22
IRAs 0.22 0.45 0.62 0.05 0.12 0.29 0.04 0.19 0.21
Business 0.13 0.17 0.25 0.02 0.05 0.14 0.05 0.05 0.19
Home 0.74 0.84 0.85 0.48 0.60 0.69 0.48 0.61 0.75
Real est. 0.18 0.23 0.34 0.06 0.11 0.22 0.07 0.16 0.21
# obs 855 2,056 2,205 571 429 382 401 145 136
Note: This table reports the percentages of asset ownership across race and education in the HRS. All figures are weighted
using survey weighs.
41
Table3b
The Distribution of Total Net Worth Across Race and Education
Whites
African-Americans Hispanics
Less than
HS
High
School
More
Than HS
Less than
HS
High
school
More
than HS
Less than
HS
High
School
More
than HS
10
th
perc. 0 8,000 22,200 -600 -125 0 0 0 3,000
25
th
perc 7,000 44,000 79,000 0 400 18,000 0 5,000 27,000
Median 51,300 109,500 190,000 5,000 30,000 66,000 14,600 41,800 85,500
75
th
perc. 129,000 214,900 390,500 41,800 81,300 140,500 56,000 107,000 194,025
90
th
perc. 291,500 408,950 830,000 89,000 164,500 273,900 142,000 213,835 350,000
95 perc 506,000 624,500 1,537,000 144,800 233,000 520,000 200,000 347,500 597,100
Mean
(S. dev)
148,383
456,925
205257
421542
386,041
684,224
43,569
224,218
76,452
211,538
137,960
285,799
52,258
127,118
91,740
162,330
252,270
836,100
# obs 855 2,056 2,205 571 429 382 401 145 136
Note: This table reports the distribution of total net worth across race and education in the HRS. All figures are weighted using
survey weighs.
42
Table 4
Who Has Checking Accounts? Probit Regressions
I II
Coeff St. Err Coeff St Err
Age 0.0038** 0.0010 0.0037** 0.0010
Male -0.0124 0.0081 -0.0126 0.0081
Married 0.0470** 0.0112 0.0475** 0.0112
Less than High School 0.0811** 0.0107 0.0819** 0.0107
High school 0.1545** 0.0159 0.1554** 0.0160
Some College 0.1264** 0.0120 0.1267** 0.0121
College 0.1162** 0.0094 0.1164** 0.0095
More than College 0.1126** 0.0101 0.1129** 0.0101
Father or mother has high
school education
0.0285** 0.0085 0.0284** 0.0086
Income/1000 0.0018** 0.0002 0.0018** 0.0002
Net worth/1000 0.0005** 0.00005 0.0005** 0.00005
Net worth squared/1000 -0.0002** 0.00003 -0.0002** 0.00003
Retired 0.0093 0.0130 0.0100 0.0130
African-Americans -0.1450** 0.0152 -0.1447** 0.0152
Hispanics -0.1463** 0.0201
Mexicans -0.1681** 0.0265
Puerto Ricans -0.1278** 0.0490
Cubans -0.1135** 0.0533
South Americans -0.2163** 0.1002
Central Americans -0.0025** 0.0552
Other Hispanics -0.1585** 0.0556
Bank Density -0.00002 0.0003 -0.00006 0.0003
Pseudo R
2
0.292 0.293
# obs 7,053 7,053
Note: This table reports probit estimates of the probabilities of holding checking accounts
using data from the HRS. Marginal effects are reported. All figures we weighted using
survey weights.
43
Table 5a
Who Has Checking Accounts? Evidence from the 2002 National Survey of Latinos
Total Sample Total Sample Born in the US Born Abroad
Coeff St. Err Coeff St. Err Coeff St. Err Coeff St. Err
Age 0.0015** 0.0004 0.0015** 0.0004 0.0015** 0.0004 0.0016** 0.0004
Male 0.0096 0.0106 0.0095 0.0105 0.0089 0.0106 0.0101 0.0107
Married 0.0006 0.0116 0.0002 0.0115 -0.0063 0.0115 0.0023 0.0121
Less Than HS 0.0254 0.0131 0.0243 0.0132 0.0332** 0.0115 0.0245 0.0140
High School 0.0427** 0.0158 0.0422** 0.0157 0.0590** 0.0152 0.0430** 0.0162
GED 0.0348 0.0112 0.0343 0.0121 0.0399* 0.0095 0.0337 0.0138
Some College 0.0827** 0.0158 0.0819** 0.0156 0.0973** 0.0156 0.0827** 0.0156
College 0.0619** 0.0115 0.0610** 0.0156 0.0691** 0.0115 0.0631** 0.0112
More than College 0.0506** 0.0101 0.0496** 0.0101 0.0568** 0.0098 0.0522** 0.0098
Income $30K-$50K 0.0265** 0.0099 0.0267** 0.0098 0.0304** 0.0099 0.0269** 0.0100
Income > $50K 0.0645** 0.0129 0.0649** 0.0129 0.0732** 0.0125 0.0661** 0.0128
Income Don't Know -0.0082 0.0196 -0.0072 0.0194 -0.0200 0.0228 -0.0134 0.0211
Home Owner 0.0347** 0.0135 0.0348** 0.0134 0.0424** 0.0135 0.0359** 0.0137
Retired -0.0193 0.0266 -0.0213 0.0270 -0.0101 0.0245 -0.0217 0.0276
African-Americans
-0.0726** 0.0301 -0.0815** 0.0325 -0.0534** 0.0272 -0.0639** 0.0287
Hispanics
-0.1084** 0.0219 -0.1104** 0.0221
Experienced discrim.
0.0193 0.0114
Mexicans
-0.0538** 0.0200 -0.1331** 0.0343
Puerto Ricans
-0.0588** 0.0363 -0.1219** 0.0448
Cubans
-0.0526* 0.0384 -0.0533** 0.0256
South Americans
-0.0640 0.0632 -0.0287 0.0242
Central Americans
0.0084 0.0229 -0.1142** 0.0411
Other Hispanics
-0.0349 0.0466 -0.1246** 0.0485
Regional Dummies Yes Yes Yes Yes
Pseudo R
2
0.3012 0.3037 0.2788 0.2954
# obs 3,802 3,802 3,802 3,802
Note: See text for detail.
44
Table 5b
Who Has Checking Accounts? Evidence from the 2002 National Survey of Latinos
(Age >=40)
Total Sample Total Sample Born in US Born Abroad
Coeff St. Err Coeff St. Err Coeff St. Err Coeff St. Err
Age 0.0010* 0.0006 0.0011* 0.0006 0.0013** 0.0006 0.0011* 0.0006
Male 0.0095 0.0107 0.0095 0.0107 0.0102 0.0114 0.0094 0.0111
Married 0.0016 0.0124 0.0019 0.0126 -0.0032 0.0128 0.0019 0.0130
Less Than HS 0.0137 0.0135 0.0138 0.0135 0.0198 0.0124 0.0132 0.0149
High School 0.0319** 0.0151 0.0320** 0.0151 0.0442** 0.0158 0.0332** 0.0158
GED 0.0204 0.0131 0.0206 0.0131 0.0263 0.0106 0.0199 0.0155
Some College 0.0510** 0.0139 0.0510** 0.0139 0.0627** 0.0144 0.0521** 0.0141
College 0.0414** 0.0108 0.0414** 0.0108 0.0485** 0.0114 0.0432** 0.0108
More than College 0.0339** 0.0099 0.0338** 0.0099 0.0405** 0.0102 0.0357** 0.0098
Income $30K-$50K 0.0054 0.0121 0.0054 0.0121 0.0010 0.0121 0.0065 0.0124
Income > $50K 0.0361** 0.0130 0.0361** 0.0130 0.0468** 0.0136 0.0383** 0.0134
Income Don't Know 0.0059 0.0162 0.0062 0.0161 0.0012 0.0204 0.0046 0.0179
Home Owner 0.0304** 0.0171 0.0306** 0.0171 0.0379** 0.0182 0.0333** 0.0177
Retired -0.0142 0.0179 -0.0146 0.0178 -0.0102 0.0182 -0.0151 0.0187
African-Americans
-0.0446* 0.0331 -0.0458* 0.0344 -0.0317* 0.0299 -0.0395* 0.0318
Hispanics
-0.1066** 0.0301 -0.1080** 0.0302
Experienced discrim.
0.0046 0.0166
Mexicans
-0.0559** 0.0302 -0.1256** 0.0504
Puerto Ricans
-0.0444 0.0559 -0.1014** 0.0552
Cubans
-0.0018 0.0397 -0.0272* 0.0209
South Americans
-0.2976** 0.1795 -0.0201 0.0152
Central Americans
-0.0058 0.0545 -0.1935** 0.0856
Other Hispanics
-0.1013** 0.0584
Regional Dummies Yes Yes Yes Yes
Pseudo R
2
0.2817 0.2819 0.2554 0.2752
# obs 1759 1759 1756 1,759
Note: See text for detail.
45
Table 6
The Effects of Retirement Seminars on Savings: Estimates across Race
Total Sample Whites African-Americ. Hispanics
Coeff. Std. err Coeff. Std. err Coeff. Std. err Coeff. Std. err
Constant -13.129 5.480 -20.78 6.663 23.146 9.48 8.901 17.443
Seminars 0.314** 0.140 0.395** 0.165 -0.181 0.253 -0.562 0.678
Race
African-Americans
-1.230** 0.155
Hispanics
-0.463** 0.217
Pos./Neg. Shocks
past unemployment -0.594** 0.099 -0.609** 0.120 -0.331* 0.176 -0.583* 0.309
past shocks -0.596** 0.095 -0.661** 0.115 -0.226 0.174 0.381 0.322
Received inheritances 0.795** 0.113 0.758** 0.127 0.229 0.373 1.178* 0.661
Risk Aversion
high risk aversion -0.171 0.134 -0.144 0.164 -0.524** 0.227 -0.065 0.390
Medium risk aversion -0.221 0.173 -0.295 0.209 0.028 0.312 0.223 0.577
Moderate risk aversion -0.422** 0.179 -0.486** 0.218 -0.786** 0.322 0.293 0.509
Subjective Expectations
expect. live to 75 -0.037 0.164 0.115 0.205 -0.115 0.256 -0.100 0.462
expect. SS more gener. 0.141 0.149 0.122 0.185 0.269 0.237 -0.151 0.416
expect. house price up -0.219 0.153 -0.201 0.189 -0.331 0.240 -0.218 0.466
exp. give help to fam. 0.284** 0.144 0.466** 0.178 -0.064 0.232 -0.381 0.424
Bequests and Help
Bequests 1.824** 0.092 1.933** 0.111 1.042** 0.165 1.292** 0.310
parent alive 0.021 0.101 0.089 0.124 -0.061 0.170 -0.776** 0.322
can rely on help 0.070 0.091 0.160 0.109 -0.340** 0.161 -0.458 0.313
Planning Horizon
Medium horizon -0.192* 0.101 -0.268** 0.119 -0.208 0.194 -0.396 0.404
Long horizon 0.794** 0.160 0.875** 0.187 0.284 0.323 0.083 0.747
Income and pensions
Permanent income -0.032** 0.004 -0.038** 0.005 0.015* 0.009 -0.033* 0.018
Pensions -0.395** 0.107 -0.428** 0.128 0.028 0.191 -0.426 0.372
Checking & business
Has a checking account 1.349** 0.136 1.418** 0.181 1.126** 0.178 1.565** 0.340
Has a business 2.780** 0.128 2.790** 0.148 2.567** 0.321 3.138** 0.528
Bank Density 0.005 0.003 0.008 0.004 -0.009* 0.006 -0.007 0.015
# of observations 6,297 4,584 1,183 530
Adjusted R
2
0.252 0.247 0.189 0.198
Note: See text for detail.
* indicates statistical significance at the 10% level
** indicates statistical significance at the 5% level
46
Table 7
The Effects of Seminars on Savings: Quartile Estimates
Total sample 1
st
quartile Median 3
rd
quartile
Seminars:
Coeff. Std. err Coeff. Std. err Coeff. Std. err Coeff. Std. err
Whites 0.395** 0.165 0.355** 0.069 0.552** 0.105 0.532** 0.223
African-
Americans
-0.181 0.253 0.304** 0.074 0.113 0.080 -0.024 0.326
Hispanics -0.562 0.678 0.207 0.230 -0.038 0.394 -0.265 0.245
Note: This table reports estimates of the effects of attending retirement seminars across quartiles of the
distribution of total net worth.
Table 8
The Effects of Seminars on Savings: Estimates across Education Groups
Median Regressions
Total sample Low education High education
Seminars:
Coeff. Std. err Coeff. Std. err Coeff. Std. err
Whites 0.395** 0.165 0.653** 0.170 0.272 0.174
African-
Americans
-0.181 0.253 0.222* 0.131 -0.007 0.117
Hispanics -0.562 0.678 0.296 0.452 -0.140 0.426
Note: This table reports median estimates of the effects of attending retirement seminars across education.
Low education refers to high school or less, high education refers to more than high school.
47
Table 9
The Effects of Seminars on Total Net Worth + Pensions: Quartile Estimates
Total sample 1
st
quartile Median 3
rd
quartile
Seminars: Coeff. Std. err Coeff. Std. err Coeff. Std. err Coeff. Std. err
Whites 1.766** 0.192 1.253** 0.131 1.703** 0.213 2.441** 0.285
African
Americans
0.719** 0.339 0.686** 0.112 1.305** 0.168 1.268** 0.367
Hispanics 0.511 0.830 0.211 0.141 0.183 0.380 0.397
0.345
Note: This table reports estimates of the effects of attending retirement seminars on the distribution of
total net worth + pensions across race.
48
Table 10
The Effects of Seminars on Stock Ownership Across Race
Total Sample Whites Afr-American Hispanics
Coeff. Std. err Coeff. Std. err Coeff. Std. err Coeff. Std. err
Seminars 0.092** 0.019 0.103** 0.023 0.021 0.016 -0.009 0.013
Race
African-Americans
-0.135** 0.018
Hispanics
-0.106** 0.026
Education
High school 0.110** 0.021 0.125** 0.027 0.022 0.019 0.0009 0.017
Some college 0.191** 0.029 0.211** 0.035 0.041* 0.029 0.002 0.021
College or more 0.239** 0.039 0.283** 0.045 -0.004 0.026 -0.022 0.010
Country of birth
US born 0.024 0.027 -0.019 0.039 0.029* 0.010 0.032** 0.016
Pos./Neg. Shocks
past unemployment -0.018 0.013 -0.024 0.017 -0.002 0.011 -0.0006 0.012
past shocks -0.020 0.013 -0.024 0.016 -0.011 0.010 0.016 0.016
Given inheritances 0.097** 0.015 0.105** 0.018 0.036 0.031 0.022 0.034
Risk Aversion
high risk aversion -0.037** 0.018 -0.057** 0.023 0.004 0.015 0.018 0.018
med. risk aversion -0.039* 0.021 -0.069** 0.027 0.074** 0.040 0.030 0.044
mod. risk aversion 0.003 0.024 -0.010 0.030 0.024 0.028 0.036 0.045
Subjective Expectations
expect. live to 75 0.015 0.023 0.017 0.029 0.006 0.017 0.0007 0.019
expect. SS more gener. 0.016 0.020 0.027 0.026 -0.022 0.015 -0.014 0.018
Planning Horizon
Short 0.038** 0.016 0.041** 0.020 0.009 0.013 0.004 0.014
Medium or long 0.081** 0.016 0.089** 0.020 0.012 0.014 0.021 0.022
Incom. pens & wealth
Permanent income 0.0001 0.0007 -0.0005 0.0009 0.001** 0.0006 0.002** 0.0008
Pensions 0.042** 0.014 0.041** 0.018 0.034** 0.013 0.025* 0.018
Net worth/100,000 0.048** 0.003 0.055** 0.004 0.006** 0.003 0.010** 0.004
Checking & business
Has a checking account 0.123** 0.017 0.141** 0.024 0.025** 0.012 0.027* 0.016
Has a business -0.029* 0.016 -0.042** 0.020 0.020 0.022 0.007 0.020
Bank Density 0.0009** 0.0004 0.001* 0.0005 0.0006* 0.0003 -0.0004 0.0006
# of observations 6,297 4,584 1,183 530
Pseudo R
2
0.213 0.176 0.277 0.335
Note: See text for detail.
* indicates statistical significance at the 10% level
** indicates statistical significance at the 5% level
