Religion and Wealth Keister+2003

Religion and Wealth: The Role of Religious Affiliation and Participation in Early Adult Asset
Accumulation
Author(s): Lisa A. Keister
Source: Social Forces, Vol. 82, No. 1 (Sep., 2003), pp. 175-207
Published by: Oxford University Press
Stable URL: http://www.jstor.org/stable/3598143
Accessed: 22-08-2014 03:31 UTC
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.
Oxford University Press is collaborating with JSTOR to digitize, preserve and extend access to Social Forces.
http://www.jstor.org
This content downloaded from 130.132.173.204 on Fri, 22 Aug 2014 03:31:44 UTC
All use subject to JSTOR Terms and Conditions
Religionand Wealth:The Role of Religious

Affiliationand Participationin EarlyAdult
AssetAccumulation*
LISAA. KEISTER,Ohio State University
Abstract
Researchershave documented extreme inequalities in wealth ownership, but the
processes that create these inequalities are not well understood. One important
contributingfactor that attracts little attention is religion. This study explores the
relationship between religious participation, religious affiliation, and patterns of
wealth accumulation. I argue that religion affects wealth ownership indirectly by
shaping demographicbehaviors.I also argue that religion directlyinfluenceswealth
accumulation by identifying valuable goals, by providing a set of competenciesthat
direct strategies of action, and by contributing to social contacts that provide
information and opportunities that can enhance wealth ownership. The findings
suggest that Jews enjoy tremendousgains in wealth ownership, while conservative
Protestantsaccumulate relativelylittle wealth. In contrast,mainline Protestantsand
Catholics are indistinguishablefrom each other and from the general population.
The results demonstrate the importance of family processes in shaping wealth
accumulation, and they underscorethe importance of culture in shaping economic
behavior and ultimately in creating social inequality.
Better the little that the righteous have than the wealth of many wicked; for
the power of the wicked will be broken, but the Lord upholds the righteous
(Psalm 37:16-17-NIV).
Lazyhands make a man poor, but diligenthands bring wealth (Proverbs10:4NIV).
* I gratefullyacknowledgecommentsfrom Mariko Chang,JoshDubrow,Penny Edgell,William
Form,RichardHamilton, KieranHealy, KatherineMeyer, Clayton Peoples, VincentRoscigno,
ToddSteen, Brad Wilcox,and participants in the Princeton UniversityCultureand Inequality
Workshop.I appreciatefunding from the American Sociological Association'sFundsfor the
Advancementof the Discipline and the Collegeof Social and BehavioralSciencesat the Ohio
State University.Direct correspondenceto Lisa A. Keister,Department of Sociology,The Ohio
State University,300 BrickerHall, 190 North Oval Mall, Columbus,OH 43210.
? The Universityof NorthCarolinaPress
SocialForces,September2003,82(1):175-207
176 / SocialForces 82:1, September2003

Wealth inequality has grown considerably in recent decades. While basic facts
about the distribution of wealth are well known, the processes that create
wealth inequality are still unclear. Wealth, or net worth, is the value of a person's
assets less the person's debts. Between the 1960s and the 1990s, total household
wealth grew from $8 trillion to nearly $24 trillion (Keister 2000b).1 Between
1989 and 1998, median household net worth increased more than 20%, and
the number of billionaires in the Forbes 400 rose from 85 to 267 (Kennickell
2000). During that time, the proportion of wealth owned by the top 1 percent
of families increased from about 33% to more than 38%. Meanwhile the share
owned by those in the lower 90% declined from 33% to 30% of the total (Wolff
1998). The implications of this severe and growing inequality are apparent
when the advantages of wealth ownership are considered. Wealth provides
current use value (as in the ownership of a home), generates more wealth when
it is invested, provides a buffer during financial emergencies, and can be passed
to future generations. Moreover, wealth may increase political influence,
educational and occupational opportunities, and social advantages for both
current and future generations. Although wealth ownership is clearly
concentrated, the processes that generate this inequality are only vaguely
understood.
One potentially important contributing factor that has received relatively
little attention is family-level religious affiliation and participation that shape
saving and investment behavior. Religion can be among the most significant
defining traits of a family, but previous research on wealth ownership has not
moved beyond relatively casual references to these influences in understanding
wealth accumulation and inequality. Religion is likely to affect asset
accumulation indirectlybecause it shapes many of the processes that determine
family wealth. A rich tradition of research demonstrates clear religious
differences in childrearing, marital stability, divorce, and fertility (Alwin 1986;
Ellison, Bartkowski & Segal 1996; Lehrer 1996b; Sherkat & Ellison 1999) and
other outcomes such as earnings, education, and female employment rates
(Darnell & Sherkat 1997; Lehrer 1999; Wuthnow & Scott 1997). Religion is
also likely to affect wealth ownership directly because it shapes values and
priorities, contributes to the set of competencies from which action is
constructed, and may provide important social contacts. Research on wealth
occasionally references the potential importance of religion, but these studies
focus almost exclusively on the role of income, investment behavior, and
inheritance without systematically exploring the relative importance of religion
(Keister & Moller 2000; Menchik & Jiankoplos 1998; Spilerman 2000).
My aim is to fill this gap by exploring the relationship between religious
affiliation, religious participation, and early adult wealth accumulation in the
U.S. I draw on the rich body of literature on religion and economic behavior
to develop a set of arguments regarding the effect of religious affiliation and
participation in childhood and adulthood on early adult wealth ownership. I
ReligionandWealth/177
focus on wealthaccumulationin young adulthoodbecauseit is duringthis time
that people establish the savings and investment patterns that continue
throughout their lives and because saving behaviorduring the early working
years creates the financial basis for later wealth ownership. I propose that
religious differences in fertility, female work behavior, and educational
attainmentand the strategiesof action learnedduring childhood both suggest
a wealth advantageto being Jewishand a disadvantageof being a conservative
Protestant.I propose that wealth ownershipis highest among Jewsand lowest
among conservativeProtestants,with mainlineProtestantsand Catholicsin the
middle of the distribution.A relativelyhigh propensityamong Jewsto invest
in financialassets, ratherthan real assets such as housing, and relativelylow
educational attainment among conservativeProtestantspropel these groups
to opposite tails of the wealth distribution. Convergencebetween mainline
Protestants and Catholics on a number of demographic and attainment
measures make these groups indistinguishable from each other. I use the
NLS-Yto model these processesand to test these proposalsempirically.
Religion and Wealth Ownership

Religion indirectlyinfluences adult wealth ownership through its effect on
behaviorsand practices,such as fertility,that shapeassetaccumulation.Previous
research on the effect of religious affiliation and religious participation on
economic behaviorand attainmenthas largelyfocusedon these indirecteffects.
Variationsin fertilityacrossfaithsand even denominationswithin a singlefaith,
for example, are important determinantsof family resourcesand ultimately
children's life attainment. It is likely, then, that people who are raised in
religionswhere fertilityis relativelylow are going to accumulatemore wealth
over their lives. Similarly,religiousdifferencesin attitudestowardeducational
attainment and returns to education are important determinantsof wealth
ownership.Thus those from religiousbackgroundsthat encourageeducational
advancementarelikelyto have an advantagein wealthaccumulationover those
who are affiliatedwith a religion that either does not incorporateideas about
education or that is either skeptical of or overtly hostile toward secular
education. Similar arguments can be made about the relationship between
religion and other family practices such as parental work behavior, union
formation,and other criticaldeterminantsof wealth accumulation.These are
certainlyimportantdeterminantsof economic behavior,and because of these
differences,there are likely to be important differencesin wealth ownership
among people from differentreligiousbackgrounds.Yeta more comprehensive
model of the relationshipbetween religion and wealth accumulationneeds to
also systematicallyaccount for the direct effect that religion, as an element of
culture, has on wealth ownership.

Religion can also shape action directly both by defining the end values
toward which behavior is oriented and by providing a tool kit that people draw
on to construct "strategies of action." One of the primary goals of churches is
to distinguish desirable behavior from undesirable behavior and to delineate
and even impose sanctions for noncompliance. A traditional view of culture
suggests that religious affiliation and participation in religious ceremonies
expose people to rituals, symbols, beliefs, and expectations that identify
worthwhile objectives. In addition, Swidler (1986) argues that religious habits
and practices transmitted by parents or during religious services during youth
shape the set of competencies from which strategies of action are constructed.
Similarly, exposure to religious ideals and views in adulthood may define the
repertoire of capacities from which actions are formulated. From this
perspective, strategies organize life and make particular choices and habits both
sensible and useful. A similar frame is Bourdieu's (1977) notion that cultural
patterns provide a structure against which individuals formulate and
implement strategies or habits. The ideas are quite similar and both imply the
same outcome for wealth accumulation. That is, people draw on the tools they
learn from religion to develop consistent strategies for dealing with problems
and for making decisions such as savings, investment, and consumption
decisions. In terms of wealth accumulation, for example, the frequent recourse
to prayer and trust in God among conservative Protestants may reduce their
inclination to invest.
In addition to shaping strategies, religious affiliation and participation affect
wealth accumulation directly by providing social contacts that provide
information, assistance, and referralsto those who can provide these important
things. Like intergenerational influences in other domains, savings behavior
reflects parental asset ownership and the asset ownership of others to whom
people are exposed during childhood (Chiteji & Stafford 2000). Knowledge
about the importance of saving, the avenues available for saving, and saving
strategies is at least partly gained through exposure to the savings behavior of
others. Wealth accumulation depends on having information about a number
of financial instruments and their features. Because information barriers make
it largely impossible for people to gather this information individually,
information from social contacts is often used instead. Social contacts may also
provide direct assistance in the form of capital transfers such as capital for
starting a business, making an initial financial investment, or making a down
payment on a home. The information that contacts have will vary by religion
and their ability to assist more directly will vary as well.
In the following sections, I investigate how values, repertoires, and contacts
vary by religious group, focusing on Jews, conservative Protestants, mainline
Protestants, and Roman Catholics. I discuss the relationship between religion
and total net worth, asset allocation, and financial trajectories in each section.
Decisions about asset allocation can have important implications for patterns
of wealth accumulation.Investingin high-risk,high-returnfinancialassets,for
example,as opposed to relativelyconservativeinstruments,such as certificates
of deposit, can have dramaticeffectson total wealth accumulatedover the life
course (Keister2000a).Likewise,the timing and orderingof financialdecisions
can shape wealth accumulationin importantways. Beginningto save in early
adulthood can have significantadvantagesover postponing saving until later.
Becausethere is a degreeof path dependencebuilt into saving and consumption decisions, people tend to follow paths through their lives, what I call financial trajectories,that influence wealth they accumulateover time. For example, a traditionaltrajectorymight involve first buying a house then investing in financialassets only later in life.
JEWS
A host of unique demographictraitsthat lead to high levels of attainmentfor

Jewsarelikelyto increasewealthaccumulationindirectly(Chiswick1986, 1993;
DellaPergola1980;Wuthnow 1999). The diasporahypothesissuggeststhat for
historical reasons, Jewish family and community traditions have arisen that
encouragedevelopment of human capital as opposed to more fixed varieties
of physicalcapital (Brenner& Kiefer1981). Moreover,both fertilityrates and
rates of female employment when children are young are relatively low in
Jewishfamilies (Chiswick1986;DellaPergola1980). Togetherthese createhigh
levels of home investment in child quality and ultimately lead to high
educationalattainmentand relativelyhigh returnsto educationalinvestments
(Chiswick 1988; Lehrer1999;Wilder 1996;Wilder & Walters1998). Because
fertilityrates are low, the dilution of materialresourcesthat are transferredin
the form of inheritanceis low and strainson resourcesin the adult family are
minimal.High ratesof homogamyamong Jewsalso suggeststhat the influence
of these demographictraits is likely to be enhancedby marriageto a person
with a similarpropensityfor attainment(Kalmijn1991;Lehrer1998;Lazerwitz
1995;Thornton 1985). Generallyhigh levels of attainmentsuggestthat wealth
accumulation among Jews is likely to exceed that of other groups by a
considerablemargin.
Yetthere is also likely to be a direct effect of Jewishreligiousaffiliationon
wealth accumulation.In Jewishculture, accumulationis seen as an indicator
of success, and Jews have been shown to have a highly positive cultural
orientationto education and occupationalstatus (Stryker1981). Ratherthan
having an orientationtowardthe afterlifeand downplayingthis world, Jewish
families encourage this-worldly pursuits including actual accumulation of
wealth and other activities that lead to wealth accumulation such as highincome careersand investing.Expectationsfor childreninclude successin this
world, and the use of success-orientedstrategiesin Jewishfamilies creates a
tool kit for childrenthat includesthe skills necessaryto accumulateresources.
180/ SocialForces82:1,September
2003
The diasporahypothesissuggestsa relativelyhigh propensityfor Jewsto invest
in financial assets and a relativelylow propensity to invest in real, or fixed,
assets.2 Because financial assets tend to be higher-risk, higher-return
investments,the returnto investmentsis likelyto be higherin Jewishfamilies,
leading to greateroverallwealth. The pervasivenessof financialinvestments
adds particularskills to the repertoireof skills developed in Jewishfamilies,
which suggeststhat Jewishchildrenare more likelyto make an earlytransition
to the ownershipof financialassets, injecting their financialtrajectorieswith
high-returnassetsthat will haveextendedtime to accumulate.BecauseJudaism
is also an ethnicity,the Jewishcultural repertoireis likely to be more salient
than the repertoiresestablishedin other faiths,and the direct Jewisheffect on
wealth accumulationis likely to be even strongerthan it might otherwisebe.
Opportunities to build relevant social capital are also relativelyhigh in
Jewishfamilies.Socialconnectionsdevelopedthroughschools and universities
can provide information about investment strategies, actual investment
opportunities, and access to capital for investing (Sherkat& Ellison 1999).
Familycontacts and contacts in the local Jewishcommunity can also provide
information, access to investments,and support that make investing feasible
at all, particularlyfor young people. For each of these reasons, I expect that
people who grew up Jewishor who are Jewishas adultswill accumulatemore
wealth as adults than those who are not Jewish.Jewswill inherit more wealth
becauseof the intergenerationaltransmissionof both wealthand the behaviors
that increasewealth accumulation.Jewswill accumulatemore wealth even at
similar levels of inheritance than non-Jews. Similarly, while educational
attainment, fertility, female employment patterns, and other demographic
differenceswill increasewealth accumulationfor Jews,there will be a direct
effect of Jewish religious affiliation reflectingthe importance of the cultural
repertoirethat is transferredintergenerationallyin Jewishfamilies.
CONSERVATIVEPROTESTANTS
In direct contrastto Jewishfamilies,wealth accumulationamong conservative

Protestantsis likely to be relativelylow. In recent decades, Americans have
generallybecome more acceptingof egalitariangender roles, divorce, smaller
families,childlessness,and other nontraditionalfamilybehaviors.Conservative
Protestantshave also become more acceptingof these behaviors,but change
in this group has been less pronounced and slower than among nonconservatives. As a result, conservative Protestants have become more
traditional than others (Lehrer1999; Smith 1998; Wilcox 1998). Traditional
attitudes have translatedinto relativelyhigh fertility rates that are likely to
decrease wealth accumulation by diluting both material and nonmaterial
resources during childhood, reducing inheritance, and making saving more
difficult in adulthood (Lehrer 1999; Thornton 1985). As in Jewish families,
female employment when children are young is low among conservative
Protestants (Thornton 1985). While this suggests that there is potential for
greaterinvestmentsin children in conservativeProtestantfamilies, the effect
is likely offset by relativelyhigh fertilityrates (Lehrer1999).
Reduced achievement among conservative Protestants is evident in low
levels of educationalattainmentin these families.In addition to high fertility,
hostility toward formal education and the scientific method resulting from
literal Bible interpretation reduce educational advancement dramatically
(Lehrer1999). Becausethe aim of science is the pursuit of truth ratherthan
the blind acceptanceof God's word, seculareducation contests the beliefs of
conservativeProtestants(Darnell & Sherkat1997). Parentalexpectationsfor
educationalattainmentare thus low among conservativeProtestants,children
in these families are more likely to be schooled at home, and parentsare less
likely to save for education (Darnell& Sherkat1997;Lehrer1999). Moreover,
while conservativeProtestantmothers and Jewishmothersboth exit the labor
force to take care of their young children, the typical level of education of a
conservativeProtestantmother is much lower than that of a Jewishmother.
As a result,the stay-at-homemother who is a conservativeProtestantis likely
to havelittle effecton her children'seducationaldevelopment.Foreach of these
reasons,childrenfrom conservativeProtestantfamiliesachievelower levels of
education, and the rate of return to education is lower for individualsraised
in conservativeProtestantfamiliesas well. Becauseeducation is an important
predictor of wealth accumulation, lower educational attainment is likely to
reduce wealth accumulation.
While indirectfactorsare important,there is also likelyto be a directeffect
of a conservative Protestant culture on the accumulation of wealth. The
strategies of action that become part of the repertoire of conservative
Protestants, however, are unlikely to include skills necessary for asset
accumulation. Traditional gender role attitudes and corresponding family.
division of labor reducesfemaleemploymentout of the home, which, in turn,
reducessaving and contributesto the creationof patternsthat do not include
high savings(Sherkat& Ellison 1999). When savingis not common, strategies
for investing naturallydo not develop and social capital that might provide
either information or financial backing is not present. Literal Bible
interpretationcan also lead to the conclusionthat wealthaccumulationshould
be avoided, and a steadfastdevotion to tithing exacerbatesthis. Conservative
Protestants"arenot averseto worldlypursuits.However,they are admonished
to avoid choices that might endangertheir souls" (Darnell & Sherkat1997).
Few American religions discourage hard work, saving, or investment. Yet
religious groups also seldom promote the idea that God favors the rich over
the poor (Wuthnow & Scott 1997), and conservative Protestant religious
doctrine includes more messagesof this sort than other doctrines.Like Jews,
conservativeProtestantsare unlikely to marry people of other religions and
2003
thus unlikely to expand their repertoireof skills and strategiesby marrying
someone with a differenttool kit (Kalmijn1991;Thornton 1985). The result
in the case of conservativeProtestants,however,is that they are unlikely to
increasetheir propensityto accumulatewealth.
As these patternscumulate across generations,they are likely to result in
lower rates of inheritance and lower overall wealth, even where there is an
inheritance.Traditionalvalues increasethe likelihoodthat if savingis possible,
funds will be channeled into homeownership rather than into financial
investments, although relatively low overall saving suggests that total
investmentsin housing are likely to be low for those who are able to become
homeowners.Thesepatternsare likelyto createfinancialtrajectoriesthat never
include savings or investment or that include only transitions to
homeownershipwith a low subsequentlikelihood of other types of saving.
AND CATHOLICS
MAINLINEPROTESTANTS
MainlineProtestantsand RomanCatholicswere at one time distinctfrom each

other and from the generalpopulationin the U.S.Thatdistinctiveness,however,
has diminished in recent years. Fertilitypatternsare instructive.Prior to the
1970s, Catholicfertilityrateswere relativelyhigh, exceedingrates both in the
general population and among mainline Protestants(Lehrer1996a;Jones &
Westoff 1979). During the baby boom, fertility increases were actually
disproportionatelyhigh among Catholics,but Catholic fertilityalso declined
more precipitously following the baby boom. In recent years, Catholic and
mainline Protestant fertility rates are comparable (Jones & Westoff 1979;
Lehrer 1996a). Convergence has also been documented in educational
attainment (Lehrer 1999; Sherkat & Ellison 1999), female labor force
participation and time allocation (Lehrer 1996b), union formation (Lehrer
1998;Sander1995;Sherkat& Ellison 1999),and separationand divorce(Lehrer
& Chiswick 1993). This convergence suggests that the processes, such as
educationalattainment and fertility,that indirectlyaffect the wealth of Jews
and conservativeProtestantsare likely to have little effect on the wealth of
mainline Protestantsand Catholics.
Becausethe convergenceof these two groupswith each other and with the
rest of the population is relativelyrecent, there is likely to be some residual
effect of the distinctivenessof prior generations.In particular,those who were
raised as mainline Protestants are likely to inherit more wealth on average
becausetheirparentswerepartof a religiousgroupthat was more affluentthan
it is today.Affiliationwith a mainlineProtestantchurchas an adultis not likely
to have the same effect because of the diminishing distinctiveness.Likewise,
the growing similarity between these two groups and their diminishing
distinctivenessin generalsuggeststhat there is likely to be little other effect of
affiliationwith either religious group on patternsof wealth ownership.
RELIGIOUSPARTICIPATION
Valuesand the strategiesthat people draw on as they make decisions are also
acquiredduring religious servicesand ceremonies.Lenski (1961) emphasized
the communal aspect of religion and Wuthnow (1999) later provided a
number of examples of the mechanism by which the communal nature of
religion shapes behavior.Wuthnowemphasizedthat participationin religious
ceremonies instills an understanding of the importance of social relations,
provides moral instruction, improves understanding of doctrine about the
correctway to live, and providesyouth with practicalskills and role models.
The communalaffirmationof valuesand strategiesby a religiousgroup,a group
that the participanttypicallyperceivedas worthy of emulating,is a powerful
mechanismfor instillingidealsand shapinghabits.When people face problems
and need to make decisions, they are likely to draw on both the skills they
learnedfrom their parentsand those they learnedduringreligiousceremonies
and other religious activities. Because religious doctrine seldom discourages
savingand nearlyalwaysencouragescorrectand conventionalliving,those who
attend services are likely to behave in ways that lead to saving and wealth
accumulation.There is evidence, for example that high school students who
attend religious services and activities devote more time to schoolwork, cut
classes less often, and are more likely to graduate than those who do not
(Sherkat& Ellison 1999). Similarlywe should see a directrelationshipbetween
religiousparticipationand other positive behaviors,and the result is likely to
be greaterwealth accumulation.
Likewise, attendance at religious services builds social networks. Social
capitalcultivatesvalues and norms that encouragepositivebehaviors,promote
the circulation of information, and encourage long-term investment in
relationships(Sherkat& Ellison1999).Not all socialnetworkswill increaseasset
accumulation,but religiousparticipationdoes improvethe possibilityof having
contactswho can provideinformation,capital,or other supportthat might lead
to wealth ownership. Thus, I anticipate that religious participation during
childhood and adulthood will increase wealth accumulation, even for those
who participateonly occasionally.
Data, Measures, and Methods

I used the NationalLongitudinalSurveyof Youth1979 cohort (NLS-Y)to test
these ideas. The NLS-Yis a nationallyrepresentativelongitudinalsurveythat
was administered 18 times between 1979 and 1998 by the Bureau of Labor
Statistics(BLS).The initial NLS-Ysample included 12,686 individualsage 14
to 22 in 1979 (i.e., born between 1957 and 1964). Nearly 10,000 of the
respondents were interviewedthrough 1998. An extensive battery of wealth

questions was added to the NLS-Y in 1985 when the youngest respondents were
20 years old. I used data from 1985 through 1998 to estimate pooled crosssection time-series models of wealth ownership. I also drew on earlier surveys
to gather information about the respondents' family backgrounds. Wealth
questions were not asked in 1991, and the BLS began conducting the NLS-Y
every other year starting in 1994. Thus, I used wealth data for 4,950 respondents
at time points (1985-1990, 1992, 1993, 1995, 1996, and 1998).3
The NLS-Y is ideal for answering questions about family background and
adult wealth because it combines broad longitudinal coverage of a large sample
with detailed information about wealth holdings, family background, life
transitions, and adult status. In each survey year beginning in 1985, respondents
reported whether they owned a comprehensive list of assets and debts and the
value of each asset or debt if they owned it. Other sources of survey data on
wealth ownership provide more wealthy individuals, those who own most assets.
The Survey of Consumer Finances, for example, is a panel data set that
oversamples high-income households to more accurately estimate wealth
distribution (Kennickell, Starr-McCluer & Sunden 1997; Wolff 1995). Because
the NLS-Y does not oversample wealthy households, it may underestimate
wealth concentration (Juster & Kuester 1991; Juster, Smith & Stafford 1999).
However, my objective is to estimate longitudinal patterns rather than crosssectional levels of inequality, and the NLS-Y data are consistent longitudinally
with estimates from similar surveys and other data sources (Keister & Moller
2000). Moreover, the NLS-Y is suited to estimating long-term family processes
because it contains detailed information about family structure and processes
during childhood, life transitions, and adult behaviors and status (Sandefur &
Wells 1999).
MEASURES
I used four dependent variables in the analyses. First, I modeled the value of
total net worth in the respondent's adult family. Net worth is the value of total
assets less the value of total liabilities. The financial assets included stocks and
bonds; cash accounts such as checking accounts; trust accounts; individual
retirement accounts; 401K plans; and certificates of deposit. The real assets
included the primary residence or home; a business, farm, or investment real
estate; a car; and other possessions. The debts included mortgages on the
primary residence; debt on businesses, farms, or investment real estate; debt
on automobiles; and other debt. I used the CPI to adjust all asset and debt
values to 2000 dollars. Second, in order to explore the degree to which the
relationship between religion and wealth accumulation reflects the religious
practices and wealth of prior generations, I modeled the likelihood that the
respondent ever received an inheritance. This dichotomous indicator includes
inheritances received both during the 1985-98 period when the NLS asked
specificquestionsabout wealthand inheritancesreceivedpriorto 1985.4Third,
I modeled the likelihood that the respondentowned a home. I modeled both
receivingan inheritanceand homeownershipas dichotomous indicatorsthat
vary yearly from 1985-1998. Fourth, I modeled the total value of the
respondent's adult financial assets to indicate the degree to which the
respondentwas invested in nontangibleassets.
I included a seriesof dichotomousvariablesto indicatereligiousaffiliation
in childhood and adulthood. I included separate indicators for Jewish,
conservativeProtestant,mainlineProtestant,and Roman Catholic.I based my
classificationof Protestantson categoriesused by Lehrer(1999) and Lehrer
and Chiswick(1993). MainlineProtestantsinclude Episcopalians,Methodists,
Presbyterians,Lutherans,Unitarians,and severalother ecumenicalbodies. The
conservativeProtestantgroup includes Baptists,Jehovah'sWitnesses,SeventhDay Adventists,ChristianScientists,and numerousother conservativegroups.5
The omitted categoryis those who had no religiousaffiliation.6I includedthree
dummy variables based on respondent self-reports to indicate church
attendancein childhood and adulthood.
I controlledfor various individualand family attributesthat are relatedto
wealth ownership, including financial resources. To capture the level of
nonwealth financial resources available in the respondent's household, I
controlled for the total household income (logged) in the currentyear and a
dummy variable indicating whether the respondent received income from
entrepreneurialactivities.I included a dummy variableindicatingwhetherthe
person ever inherited and a continuous variable indicating the amount
inherited by year (logged) for 1985-98 in some models.7 I included several
familybackgroundindicatorsto control for other characteristicsof the family
of origin that affect adult wealth ownership. Net family income in 1978
(logged) controls for the family'sresources.I also included a dummy variable
indicating that the respondent had not provided information about family
income in 1978 to control for patternsthat might be common to those with
missing values on this key variable (Sandefur& Wells 1999). I also included
two measures of parents' education: separate dummy variables indicating
whether the respondent'sfather and mother had completed college or more
education.8I included a dummy variableindicatingwhether the respondent
was born in the northeasternU.S. to control for region of birth. The single
Northeastmeasurecontrolsfor the possibilitythat a respondentwas born into
a wealthy family but escaped identification as wealthy by the other family
resource measures.9
Previousliteraturesuggeststhat resourcesare diluted in largefamilies,and
each additionalsiblingdiminishesadult attainment(Downey 1995). I used the
total number of siblingsthe respondentever had, reportedin 1998,to indicate
family size in childhood.10I used dummy variablesto indicate whether the
respondent'sparents both worked full-time (more than 35 hours per week)
in 1978 to control for whether nonmaterialparentalresourceswere diluted

becauseboth parentswere absent. Familydisruptionmay also reducethe time
parents have available to nurture children (Mechanic & Hansell 1989). I
included three dummy variablesgiving a snapshot of the respondent'sfamily
structureat age 14 and a dummy variableindicatingwhetherthe respondent
lived with both biological parentsuntil age 18.
I also controlled for severalindividual and adult family traits. I included
two dummyvariablesin most models to indicatewhetherthe personwas black
or Hispanic, as opposed to white. I indicated age in number of years in the
current year and also controlled for the square of age. I included a dummy
variableindicatingwhether the respondentwas male. I also included a series
of education dummy variables indicating whether the respondent had
completed high school, some college, a bachelor's degree, or an advanced
degree. The omitted category for education is respondents who had not
completed high school. I included a dummy variableindicatingwhether the
respondent was married (as opposed to all other marital statuses) in the
current year." I also included a dummy variable indicating whether the
respondent was ever divorced to capture the cumulative effects of marital
disruption. I included a dummy variableindicating whether the respondent
ever had a biological child and a continuous variableindicatingthe number
of children born to the respondent for those who had children. Finally, I
includeda variablethat was an interactionbetweenever havingbeen divorced
and ever having childrento indicate the magnificationof the negativeeffects
of divorce involving children.
I also included a continuous variableindicatingthe number of weeks the
respondent'sspouse workedin the previousyear.In includingindicatorsof the
respondent's spouse's contribution to the household economy, I hope to
capturethe effects of added financialresourcesthat came into the household
through either the spouse's work or independent wealth. Finally,I included
four dummy variablesindicatingregion of residenceto capturevariationsin
economic conditions and opportunities.A single indicatorof urban residence
captures urban-rural differences in wealth ownership.12This variable uses
census data to indicatewhetherthe county of residencehad a centralcore city
and adjacent,closely settled areawith a total combined population of 50,000
or more. A set of three dummy variablesindicatesregion of residencein the
currentyear,including residencein north central states, southern states, and
westernstates,versus those living in northeasternstates.
Table1 containsmeansand standarddeviationsfor the exogenousvariables.
Consistentwith estimatesfrom the GSS (Sherkat& Ellison 1999), 1% of the
sample was raised Jewish and about 30% each were raised as conservative
Protestants, mainline Protestants, and Catholics. Religious affiliation in
childhoodand adulthoodarehighlycorrelated:.82 for Jews,.66 for conservative
Protestants,.65 for mainlineProtestants,and .85 for Catholics.Also consistent
Religion and Wealth / 187

TABLE 1: Means and Standard Deviations for Exogenous Variables, NLS-Y,
1979-1998
Mean (S.D.)
Childhoodreligion
Jewish
Conservative
Protestant
MainlineProtestant
Catholic
(.45)
(.47)
Adultreligion
Jewish
Conservative
Protestant
MainlineProtestant
Catholic
.01
.29
.23
.33
(.10)
(.46)
(.42)
(.47)
(.44)
(.40)
(.47)
Adultreligiousparticipation
Some
.29
.20
Occasionally
.23
Frequently
(.45)
(.40)
(.42)
.01
(.12)
.31
(.46)
.29
.33
Childhoodreligiousparticipation
.27
Some
.20
Occasionally
.33
Frequently
Mean (S.D.)
Familybackground
Familyincome in 1978a
$14,560($15,294)
.16 (.33)
fromcollege
Father
graduated
from
Mother
graduated college .09 (.29)
.19
Bornin theNortheast
(.39)
2.9
Numberof siblings
(2.3)
Bothparentsworkedfull-time .30
(.46)
.39
Motherworkedfull-time
(.48)
.77
Fatherworkedfull-time
(.41)
.08
(.28)
Stepparentfamily
.13
(.33)
Single-parentfamily
.04
Otherfamilystructure
(.20)
until18
.66
Livedwithbothparents
(.47)
Individualandfamilytraits
Black
.12
.07
.95
Bornin theU.S.
.49
Male
b
.37
school
graduate
High
.18
Some college
.12
College graduate
.08
Advanceddegree
Married
.53
.22
Everdivorced
Numberof childrenborn 1.7
Hispanic
(.33)
(.26)
(.20)
(.50)
(.48)
(.39)
(.33)
(.28)
(.50)
(.42)
(1.3)
(N= 6,109)
a
to 2000dollarsusingtheCPI:mean= $38,144
Incomein 1978dollars.Converted
(S.D.= $40,065).
bEducationrefersto highestlevelcompleted.
with other estimates, 27% of the sample attended some religious services in
childhood, 20% occasionally attended, and 33% attended frequently.
METHODS
I used generalized least-squares regression to model net worth, the value of

financial assets, and home value. In the models for home value, I also control
for the predicted probability of homeownership to correct for potential
selection bias. I used logistic regression to model the likelihood that the
respondent received an inheritance or owned a home. I regressed each of the
dependent variables on religious affiliation and participation in childhood and
in adulthood. Because the correlation between childhood and adult religious

TABLE Al:
Substitution Costs for Optimal Matching

Savings Checking
Savings
Checking
Home
Bonds
Stocks
-2
-
--
Home
Bonds
2
2
2.5
2.5
2
Stocks
3
3
3
3
affiliationand participationis relativelyhigh, I report separateestimates for

the effect of childhood and adult religion.
I also used optimal matching to explore financial trajectories.Optimal
matching, a method designed to identify common patternsor trajectories,is
based on the notion that we can measure how similar two sequences are by
determininghow difficultit is to transformone into the other (Abbott& Hrycak
1990). Optimal matching has most commonly been used to identify and
understand individual career patterns, but the method is equally suited to
catalogingsequencesin saving behavior.I used optimal matchingto identify
common patternsin the assets respondentsowned over time. I included five
assets with various degrees of risk associated with them: savings accounts,
checkingaccounts,a home, bonds, and stocks.If a personneverowned an asset,
the portfolio would be represented as 00000. If the person then opened a
savings account but purchased no other assets, the portfolio would be
representedas 10000.
Optimal matching would indicate that it would take one substitution,
substitutinga 1 for a 0 in the first column, to make the sequencesidentical.If
each change (insertion, deletion, or substitution) "costs"the same, a simple
count of the number of changes would indicate the complexity of the
transformation;however,some transformationsare inherentlymore difficult
than others.Underlyingthis strategyis the notion that ownershipof relatively
high-riskassets early in life can propel net worth forwardin ways that more
conservativeinvestmentstrategies,or no investmentat all, cannot. If religion
shapesinvestmentpatterns,it is possiblethat this accountsfor the relationship
betweenreligionand wealthownership.The substitutioncosts I used are given
in TableAl.
Using this strategy,I identifiedthree dominantpatternsin asset ownership
over the life course.Many respondentsremainedpermanentlyasset poor; that
is, they never owned an asset and always had a portfolio coded 00000. A
traditionalsequence, and one that emerged as quite common, is a sequence
that involves an early transition to cash accounts and then homeownership.
Those who followed this sequence typically acquired a checking or savings
account (or both) during late adolescence,eventuallybought a home as their

TABLE 2: Childhood Religious Affiliation and Adult Wealth Ownership
All
Conservative Mainline
Protestants Protestants
Jewish
Respondents
Roman
Catholic
Wealth
1998networth,meana
1998networth,mediana
Homeowner
Stockowner
$134,500
$372,300
($280,518) ($497,800)
$48,201
.65
(.48)
.24
(.43)
$150,890
.73
(.45)
.56
(.50)
$93,000 $152,000
($232,400) ($296,000)
$148,470
($230,700)
$26,200
.61
(.49)
.16
(.36)
$62,870
.69
(.46)
.28
(.45)
$61,560
.66
(.48)
.27
(.44)
.31
(.46)
.48
(.50)
.41
(.49)
Inheritance
Proportionwho everinherited .40
(.49)
Amount inheritedb
$1,779
($21,872)
.67
(.47)
$4,713
($17,686)
$1,161
($19,680)
$2,277
($24,950)
$1,422
($16,685)
a Converted
to 2000dollarsusingtheCPI.
bMean
between1988and1998.
yearlyinheritance
first major investment, and may have eventually invested in stocks and bonds.
The third common sequence that emerged was an early transition to financial
wealth. Those who followed this type of pattern bought financial assets, stocks
and bonds, early in life. Most of these people also had checking and savings
accounts and some eventually owned homes as well, but the dominant feature
of this group was early entry into ownership of relatively high-risk assets. I
discuss the distribution of people across these groups and the implications of
each sequence for wealth accumulation below.
Results
NET WORTH
My analyses provide clear evidence that being raised Jewish and practicing
Judaism as an adult are associated with tremendous gains in wealth. Table 2
reports unadjusted estimates of adult wealth by religious affiliation in
childhood.13 Mean 1998 net worth for the full sample was $134,500. For those
raised in Jewish families, the mean was $372,300, more than twice the mean
for the full sample. Median net worth, a more unbiased estimate of wealth given
the skewness of the wealth distribution, was $48,200 for the full sample. For
those raised in Jewish families, the median was more than three times larger
at $150,890. The proportion of Jews who have ever inherited is also much
higher than the proportion in the full sample. Of those raised in Jewish families,
.67 had everreceivedan inheritancecomparedto .40 in the full sample.Among

those who had inherited at some time, the averageinheritancefor Jewswas
$4,713, while the average was less than $2,000 for the full sample. The
descriptivestatisticssuggestthat asset allocationmay,indeed,accountfor some
of this difference.The table includes estimates of the proportion of families
who own homes and stocks.Of those raisedin Jewishfamilies,56%ownedstocks
in 1998, while only 24% of those in the full sample were stock owners. The
differencein homeownershipis much less dramatic.Of the Jewishrespondents,
73%owned theirown homes, while 65%of the full samplewerehomeowners.'4
The unadjusteddescriptivestatisticsin Table2 also provideinitial evidence
that conservativeProtestantsare relativelydeficient in wealth, while mainline
Protestantsand Catholicsare virtuallyindistinguishablefrom each other and
from the full sample. The mean net worth for those raised as conservative
Protestantsis $93,000,and the median for this group is only $26,200,or about
halfthe overallaverage.In addition,a relativelysmallproportionof conservative
Protestantsown homes, ownedstock,or everinheritedanywealth.Amongthose
conservativeProtestantswho did inherit,the size of their inheritancewas also
small relativeto the overallsample.In contrast,the estimatesfor each of these
wealth measures for mainline Protestantsand Roman Catholics was nearly
identical to each other and to the estimates for the overall sample. One
exception is the amount of inheritancethat mainline Protestantsreceive.The
mean inheritancefor the full sample is $1,779, while the mean for mainline
Protestantsis $2,277. The differencein these unadjustedestimatessuggeststhat
prior generationsof mainlineProtestantswere distinctfinancially,even though
more recent generationshave experienceda convergencewith the rest of the
population.
The patterns that are apparentin the descriptivestatistics are upheld in
the multivariateanalyses.Model 1 of Table3 reportsgeneralizedleast-squares
estimates of net worth with the indicators of inheritance omitted from the
equation.15I did not include inheritance in model 1 to demonstrate the
relationshipbetween religion in childhood without consideringthe role that
assetsacquiredfrom prior generationshave on adult wealth.As predicted,the
relationshipbetween being raised in a Jewish family and adult net worth is
significantlygreaterthan 0. In fact, the coefficientestimatefor Jewishreligious
affiliationin childhood is more than 6.6 times greaterthan its standarderror.
Model 2 of Table 3 also controls for whether the respondent received an
inheritance and the amount of the inheritance. Both inheritance indicators
are significantly different from 0 and positively related to adult net worth.
More importantly,even controllingfor inheritance,there is still a very strong,
positive relationshipbetween being raisedJewishand adult net worth. These
estimatesimply that being raisedJewishis associatedwith enormous gains in
wealth, even controllingfor financialresources,familybackground,and other
important individual and family predictors of wealth. Similarly,the results
presentedas models 3 and 4 suggestthat practicingJudaismas an adult is also
verystronglyassociatedwith net worth,even at comparablelevelsof inheritance
and other family resources.
While it is difficultto distinguishthe effectsof childhoodreligionfrom adult
religion, the results in Table3 are somewhat instructive.It is not possible to
include both the childhood and adult indicatorsof religious affiliationin the
same models because the correlationbetween childhood and adult religion is
extremely high and the independent effects are nearly impossible to
differentiate.Moreover,including childhood indicators as opposed to adult
indicatorsmakeslittle differencein the adjustedR2.Comparingthe magnitude
of the religiousaffiliationcoefficientto the coefficientestimatefor inheritance,
included in models 2 and 4, suggests that the relative magnitude of adult
affiliationis greaterthan the magnitudeof childhood affiliation.The ratio of
the coefficientestimatefor the variableindicatingchildhood Jewishaffiliation
to the dummy variable indicating receipt of an inheritance is 3.3 (88.68 /
26.32). In contrast, the ratio between the coefficient estimate for the adult
Jewishaffiliationvariableto the dummy indicating receipt of an inheritance
is 5.3 (137.94/ 26.08). This suggeststhat the relativestrengthof the relationship
between adult affiliation with Judaism and net worth is stronger than the
relativestrengthof the relationshipbetweenchildhood affiliationwith Judaism
and net worth.While this is not conclusiveand while the resultssuggesta very
strong relationship between childhood processes and adult outcomes, the
comparisonof the relativestrengthsof the coefficientssuggestsa premiumfor
affiliationin adulthood.To extendthis a bit furtherstill, the implicationof this
finding is that both childhood and adult family processes contribute to the
repertoireof skills that shape adult wealth ownership,but there may well be
an importantpremium for the adult social contactsthat are availableto those
who are practicingJewsas adults.
The results presentedin models 1 and 2 of Table 3 also provide support
for my other arguments.First,there is evidence that conservativeProtestants
have significantlyless wealth than those who are not raisedin a religion,even
when inheritanceis controlled.Second,the resultsin all four models in Table3
providesupportfor my predictionthat affiliationwith mainlineProtestantand
Catholicchurcheswill have no significantrelationshipwith wealth ownership.
Third, the results presented in each of the models included in Table 3 also
conform to my prediction that church attendance, in both childhood and
adulthood, is significantlypositively related to adult wealth. An interesting
patternthat emergesin models 3 and 4 of Table3 is that there is no significant
associationbetween affiliationwith a conservativeProtestantfaith as an adult
and adultnet worth.This resultimpliesthat the negativeimpactof conservative
Protestantismis entirelya product of childhood processes,suggestingthat the
effect of conservativeProtestantideals on an individual'srepertoireof skills is
strongestwhen the influence begins in childhood. As I argued in the case of

TABLE 3: Generalized Least-Squares Parameter Estimates for Models of
Adult Net Worth, 1985-1998
ChildhoodReligion
Model 1
Model 2
Religiousaffiliation
Jewish
ConservativeProtestant
79.86***
(14.10)
-12.48**
MainlineProtestant
(6.17)
-6.10
RomanCatholic
(6.09)
-5.76
(6.26)
Churchattendance
Some
88.68***
(17.53)
-17.00**
(7.94)
AdultReligion
Model3
Model4
119.50***
(14.21)
-4.04
-12.31
(4.66)
5.18
137.94***
(17.72)
-1.51
(6.02)
6.44
(7.81)
(4.73)
(6.09)
-6.46
(8.01)
6.14
10.59
(4.73)
(6.08)
12.67**
11.41**
12.28**
Occasionally
9.64**
(4.00)
21.39***
(5.18)
27.07***
(3.72)
12.21**
13.59**
Frequently
(4.34)
9.00**
(5.62)
12.15**
(4.07)
(5.22)
Financialresources
Income(log)
income
Entrepreneurial
Everreceivedan inheritance
.00***
(.00)
52.73***
(4.01)
-26.32***
.00***
(.00)
53.41**
(5.54)
-26.08***
Familybackground
Familyincome in 1978(log)
Familyincome (1978)
not reported
Father'seducation
-4.00***
(.51)
.00***
(.00)
52.72***
(.00)
53.40***
(4.01)
(5.53)
.84*
.50
.78*
.43
(.40)
5.59
(.51)
2.67
(.40)
5.35
(.51)
2.29
(6.29)
19.00***
(5.38)
19.29**
(6.60)
Number of siblings
-3.47***
Bothparentsworkedfull-time
(.66)
-2.77
Bornin the Northeast
(5.37)
(.51)
(4.91)
21.00***
(4.23)
13.92**
(5.18)
5.17
(4.80)
Mother'seducation
.00***
(5.44)
12.44**
(3.74)
(3.74)
Amountof inheritance
(4.20)
10.69**
(4.16)
(4.83)
3.35
(6.11)
(4.90)
20.98***
(6.28)
19.21***
(4.22)
(5.37)
13.31**
18.84**
(6.60)
(5.19)
3.96
2.35
-3.60***
(4.77)
-3.45***
(6.07)
-3.52***
(.84)
-2.36
(.66)
-2.40
-1.80
(.84)
(2.93)
(3.74)
(2.93)
(3.74)
Familystructureat age 14
Stepparentfamily
-2.81
-2.51
-2.45
-2.39
(7.40)
8.14
(5.73)
Single-parentfamily
(5.73)
7.17
8.09
(7.40)
9.02
(5.16)
(6.65)
Other
-4.73
.20
(5.17)
-3.19
(6.65)
1.83
(7.35)
3.17
(9.48)
7.98
(7.36)
2.64
(9.48)
(3.95)
(5.12)
(3.95)
(5.12)
Livedwithbothparentsuntil18
7.34
Religion and Wealth I 193

TABLE 3: Generalized Least Squares Parameter Estimates for Models of
Adult Net Worth, 1985-1998 (Cont'd)
ChildhoodReligion
Model I
Model2
Individualtraits
Black
AdultReligion
Model 3
Model4
-28.22***
-26.66***
-27.12***
-25.66***
Hispanic
(4.91)
-22.79***
(6.18)
-20.45***
(4.90)
-23.97***
(6.17)
-20.82***
Bornin the U.S.
(6.30)
-16.91**
(7.93)
-15.67**
(6.25)
-17.06**
(7.86)
-16.77*
Age
(7.44)
-9.94***
-12.18***
(7.43)
-10.0*
(9.51)
-12.31***
Age squared
(2.89)
.28***
(9.52)
(4.45)
.31***
(2.89)
.28***
(4.45)
.31***`
(.05)
3.93
(.07)
6.43
(.05)
4.62
(.07)
7.18
High school graduate
(2.73)
.07
(3.53)
-2.12
(2.73)
-.17
(3.53)
-2.40
Somecollege
(4.04)
15.92***
(5.28)
16.32***
(4.04)
15.75***
(5.28)
16.36***
Collegegraduate
(4.63)
36.50***
(6.05)
33.45***
(4.63)
34.71***
(6.05)
31.57***
Advanceddegree
(5.43)
41.28***
(7.08)
34.34***
(5.45)
39.82***
(7.12)
32.74***
(6.04)
(7.91)
(6.06)
(7.95)
40.12***
41.89***
40.06***
41.75***
Everdivorced
(2.51)
21.67***
(3.35)
26.16***
(2.51)
21.79***
(3.35)
26.36***
Everhadchildren
(4.66)
-5.41
(6.29)
-.81
(4.66)
-6.00
(6.29)
-1.42
Numberof childrenborn
(3.78)
3.25*
(4.99)
3.19*
(3.78)
3.25*
(4.99)
3.18*
(1.46)
-10.99***
(1.88)
-11.10*
(1.46)
-1 1.04***
-1 1.22***
(2.47)
.28**
(3.21)
.32**
(.06)
(.08)
Male
Familyin adulthood
Married
DivorcedX everhadchildren
Weeksspouseworked
(2.47)
.28***
(.06)
Residence
Urban
(3.22)
.33***
(.08)
(1.88)
-2.73
-4.18
-2.79
-4.13
North Central
(2.54)
-3.69
(3.54)
-8.44
(2.54)
-3.48
(3.53)
-8.16
South
(4.97)
-.11
(6.40)
-4.11
(4.98)
.71
(6.40)
-3.58
West
(4.64)
6.87
(5.99)
1.72
(4.65)
7.72
(6.00)
2.92
(5.14)
.33
(6.65)
.34
(5.14)
.32
(6.63)
.34
AdjustedR2
N
4,799
4,799
4,799
4,799
Notes:Standarderrorsarein parentheses.Samplesize (n) reflectsthe numberof respondentsincludedin each

year;thesemodels included52,789 (or 4,799 X 11years)observations.
*p<.05
**p<.Ol
***p<.001

TABLE4:
Logistic ParameterEstimatesfor Inheritanceand Homeownership, 1985-1998

EverInherited
Childhood
Adult
Religion
Religion
Jewish
(.12)
.27**
(.12)
-.1 1**
-.26***
MainlineProtestant
(.05)
.15*
(.05)
(.04)
-.13***
RomanCatholic
Churchattendance
Some
.54***
.05
(.04)
-.09
(.05)
(.04)
.23***
.26***
Own a Home
Childhood
Adult
Religion
Religion
-.41**
(.14)
.19**
(.07)
.11
(.06)
.19*
(.07)
-.06
-.37**
(.14)
.10*
(.05)
.01
(.05)
.15*
(.05)
.19***
(.04)
Occasionally
(.03)
.07**
(.03)
.17***
(.04)
.08
(.03)
Frequently
(.03)
.09**
(.05)
.05
(.04)
.18***
(.03)
(.03)
(.04)
(.04)
Financialresources
Income(log)
income
Entrepreneurial
.09***
.00***
.00**
.00***
.00***
(.00)
.38***
(.00)
.38***
(.00)
.33***
(.00)
.32***
(.05)
(.05)
(.07)
.08***
(.07)
.07***
(.03)
(.03)
(.01)
(.01)
Familybackground
.28***
.00
.00
.02***
.02***
.00
.00
(.00)
.25***
(.00)
.26***
(.00)
(.00)
-.01
-.02
(.04)
.48***
(.04)
.48***
(.05)
.01
(.05)
.00
Mother'seducation
(.03)
.35***
(.03)
.36***
(.04)
-.04
(.04)
-.04
(.04)
.18***
(.04)
.21**
(.05)
.29***
(.05)
.28***
Number of siblings
(.04)
-.04***
(.04)
-.04***
(.05)
-.05***
(.05)
-.05***
(.01)
(.01)
-.06**
-.06**
(.02)
(.02)
(.01)
.06*
(.03)
(.01)
Familyincome(1978)
notreported
Father'seducation
Stepparentfamily
Single-parentfamily
-.09*
-.08
(.04)
(.04)
.01
.00
.06*
(.03)
.02
.04
(.06)
(.06)
.01
(.06)
.30***
.01
(.06)
-.04
(.08)
.21***
(.06)
-.02
(.08)
.20***
(.03)
(.04)
(.04)
I - I
Other
(.04)
-.25***
(.04)
-.24***
(.06)
.30***
(.03)

TABLE4:
Logistic ParameterEstimatesfor Inheritanceand Homeownership, 1985-1998

(Cont'd)
EverInherited
Adult
Childhood
Religion
Religion
Individual traits
Black
Own a Home
Childhood
Adult
Religion
Religion
-.81'**
-85***
-.96***
-97"""
Hispanic
(.04)
-.84***
(.04)
-.85***
(.05)
-.47***
(.05)
-.50"""
Bornin the U.S.
(.05)
-.04
(.05)
.00
Age
(.06)
.02
Age2
(.03)
.00
Male
(.00)
-.11***
(.00)
-.10***
(.02)
.39***
(.02)
(.03)
Somecollege
(.03)
.75***
(.04)
1.13***
(.04)
1.47***
(.05)
(.04)
1.14***
(.04)
1.48**"
(.05)
Collegegraduate
Advanceddegree
Familyin adulthood
Married
.00
(.07)
(.07)
(.06)
-.10
(.08)
-.10
(.08)
.01
(.03)
(.05)
(.05)
-.01"""
-.01***
(.00)
-.10'*
(.03)
(.00)
-.10*
(.03)
.00
.39***
.77***
.01
(.02)
.49"""
.49"""
.40***
.39"""
(.04)
.43***
(.04)
.40"""
(.05)
.51'**
(.05)
.48"""
(.06)
(.06)
.27***
.31'**
(.07)
(.07)
1.85***
(.03)
.10
(.05)
1.84***
(.03)
.11
(.05)
Everdivorced
(.02)
.05
Everhad children
(.04)
.07*
(.04)
.07
(.03)
-.08***
(.03)
-.08***
(.04)
.03*
(.04)
.03
(.01)
-.05
(.01)
-.05
(.02)
(.02)
-.13**
(.03)
Weeksspouseworked
(.02)
.00**
(.02)
-.13**
(.03)
(.00)
(.00)
Residence
Urban
.18'**
North Central
South
West
(.02)
-.11*
(.04)
-.12
(.04)
.12*
(.04)
Likelihoodratio
N
7,417***
4,877
.05
.00***
.29***
.00**
.28"""
.00**"
(.00)
(.00)
-.26***
-.27"""
(.02)
(.03)
-.07
.61**
(.06)
(.03)
.60***
(.06)
.19'**
(.04)
-.09*
(.04)
.14'*
(.04)
7,391***
4,877
.52***
.52"""
(.05)
(.06)
.12*
(.06)
10,463***
4,877
.13*
(.06)
10,501**"
4,877
Notes: Standarderrors are in parentheses. Sample size (N) reflectsthe number of respondents included
in each year; these models included 53,647 (or 4,877 X 11 years) observations.
* < .05
** p < .01
***p <.001
p
196 / Social Forces 82:1, September 2003

TABLE5: GeneralizedLeast-SquaresParameterEstimatesfor Models of Wealth, 1985-1998
FinancialAssets
Childhood
Adult
Religion
Religion
Jewish
191.36***
-22.20
(18.75)
-21.81
(18.29)
-9.56
(18.75)
297.40***
(40.26)
13.71
(14.37)
16.72
(14.45)
18.71
(14.38)
13.49
(12.31)
17.38
(13.37)
6.76
(12.44)
2.47
(11.34)
-15.21
(12.78)
-6.33
(12.63)
0.00***
(0.00)
36.60*
(19.07)
8.73
(9.06)
12.00***
(1.73)
(.00)
35.95*
(19.00)
9.52
(9.03)
11.58***
(1.72)
(40.07)
MainlineProtestant
RomanCatholic
Churchattendance
Some
Occasionally
Frequently
Financialresources
Income (log)
income
Entrepreneurial
Familybackground
.00"**
0.43
0.33
(1.35)
-2.82
(1.35)
-3.57
(15.86)
29.75*
(12.50)
30.46
(15.58)
-17.31
(14.40)
-3.43**
(1.04)
-6.14
(8.71)
Familyincome(1978)
notreported
Father'seducation
(15.90)
28.5 1
Mother'seducation
(12.56)
31.92*
(15.60)
-17.06
Number of siblings
(14.55)
-3.91**
(1.05)
-6.89
(8.74)
Home Value
Childhood
Adult
Religion
Religion
37.27***
(8.15)
-15.19"**
(3.81)
-11.46***
(3.68)
(9.26)
-16.35**"
(4.30)
-13.29**"
.49
(4.16)
-.84
(3.80)
(4.30)
2.93
3.68
(2.26)
3.12
(2.55)
.10
(2.31)
(2.57)
3.38
(2.87)
1.13
(2.62)
.00***
0.00***
(.00)
.28
(3.56)
8.02***
(1.61)
1.60***
(.32)
1.56
(3.54)
8.60***
(1.80)
1.09***
(.30)
2.14***
(.21)
25.38***
(2.60)
14.38***
(2.23)
10.55***
(2.82)
12.39***
(3.32)
2.14**"
(.24)
23.98***
(2.97)
15.87***
(2.55)
11.05**"
(3.23)
15.07**"
(3.64)
(.00)
-.58
-.59
(.45)
2.21
(1.60)
(.49)
1.94
(1.82)
-6.07
-7.55
-5.90
(3.36)
-4.54
(3.82)
-4.78
Stepparentfamily
16.99
Single-parentfamily
(17.82)
-6.46
Other
(15.68)
-3.99
(15.62)
-3.02
(3.20)
-9.05
(22.33)
(22.29)
12.74
(12.16)
11.90
(12.14)
(4.70)
-2.73
36.85***
14.88
(17.78)
(2.46)
(3.62)
-11.91
(5.29)
-2.61
(2.75)

TABLE5: GeneralizedLeast-SquaresParameterEstimatesfor Models of Wealth, 1985-1998
FinancialAssets
Adult
Childhood
Individualtraits
Black
Hispanic
Bornin the U.S.
Religion
Religion
Religion
Religion
-5.46***
-5.76***
-1.10***
-1.20***
(.60)
-3.36
(.89)
-1.57
(4.66)
-29.83***
(5.16)
-34.71***
(4.23)
-2.36
(4.81)
2.70
(4.64)
.10
(4.68)
.02
(.06)
1.05
(.06)
1.06
(1.64)
-5.06
(1.82)
-8.44
(3.59)
(3.86)
.66
(1.87)
-5.27
(18.71)
.72
(22.30)
Age
Age2
Male
Somecollege
Collegegraduate
Advanceddegree
Familyin adulthood
Married
Everdivorced
Everhad children
Weeksspouseworked
-52.34
(68.01)
.75
(.91)
12.49
(8.32)
1.22
(14.74)
7.87
(16.38)
15.50
(18.56)
46.39
(20.39)
-10.63
(13.07)
9.58
(18.69)
1.37
(14.00)
-1.43
(4.71)
-2.82
(7.44)
-.15
(.19)
Residence
Urban
North Central
South
West
(1.75)
-1.04
(18.46)
-1.96
(22.22)
-52.22
(67.91)
.75
(0.91)
13.82
(8.29)
2.34
(14.73)
11.51
(16.34)
17.34
(18.62)
48.68
(20.46)
-9.30
(13.03)
9.26
(18.63)
2.24
(13.96)
-1.67
(4.70)
-2.82
(7.42)
-.16
(.19)
-2.13
(3.91)
20.31***
(4.31)
25.55***
(4.13)
-18.24
(11.70)
9.36**
(3.23)
-5.25
(2.93)
-13.54
(11.79)
12.45***
(3.59)
-3.41
(3.16)
.68
(1.81)
.19
(.03)
(.04)
-3.43*
16.27***
(8.55)
(1.99)
9.15
(16.02)
9.87
(15.22)
-6.50
(16.87)
9.99
(15.98)
8.23
(15.19)
.37
4,913
(4.54)
(.93)
-4.62**
(1.97)
.22
-1.71
-4.44
(4.64)
34.25***
(.84)
(8.58)
(16.76)
(4.20)
25.54***
1.01
-1.90
-43.15***
(4.62)
-42.16***
(4.30)
9.84***
(3.60)
13.59***
(2.11)
-40.11***
(4.88)
-39.46***
(4.54)
13.59***
(4.00)
.39
84.28***
(29.27)
.31
75.68***
(29.49)
.33
4,913
4,149
4,149
Lambda
-2 Loglikelihood
N
Home Value
Adult
Childhood
Notes:Standarderrorsarein parentheses.Samplesize (N) reflectsthe numberof respondentsincludedin each

year;thesemodelsincluded54,043 (or 4,913 X 11years)observations.Lambdais the predictedprobabilityof
homeownership. * p < .05 **p < .01 ** p < .001.
198 / SocialForces 82:1,September2003
Jewishreligious affiliation,it also is possible that the differencebetween the

effects of childhood and adult religious affiliation is a result of the social
contacts that are only available to adults who are affiliated with particular
faiths.In contrastto Jewishcongregations,the pool of contactsin a conservative
Protestantchurch, on average,affordsfew opportunitiesto increasewealth.
To what extent does inheritance account for these patterns? It is not
surprising that the results in Table 3 suggest that inheritance is positively
associatedwith net worth. But to what extent are there religiousdifferencesin
inheritance?Table 4 includes coefficient estimates from logistic regression
models predicting the likelihood that the respondent ever received an
inheritance as a function of childhood and adult religious affiliation and
participation.Consistentwith my predictions,those who were raisedin Jewish
familiesor who wereJewishas adultswere significantlymore likelyto have ever
receivedan inheritance.Likewise,affiliationwith a conservativeProtestantfaith
either in childhood or in adulthood was negativelyassociatedwith receiving
an inheritance.The estimated effects for affiliationwith mainline Protestant
and Catholic churches,however,revealan interestingdifference.There is no
associationbetweenbeing Catholicand inheriting.In contrast,those who were
raisedin mainline Protestantchurcheswere more likely to inheritwhile those
who were affiliatedwith mainlineProtestantchurchesas adultswere less likely
to inherit than those who were not affiliated with a church. The positive
relationshipbetween childhood affiliationwith a mainline Protestantchurch
and inheritanceis suggestiveof the loss of socioeconomic distinctivenessof
mainline Protestantismin the U.S. That is, the prior generation of mainline
Protestantsappearfrom these data to have been significantlywealthierthan
those who were not affiliatedwith a religion.Parentsfrom that generationwere
thus better able to leave an inheritancefor their children.The absence of an
associationbetween Catholicismand inheritancesuggeststhat Catholicshave
had to save and invest more as adults to attain wealth levels that appear,on
average,to be equivalentwith mainline Protestants.
AND PATHS
PORTFOLIOS
What other factors might account for the relationshipbetween religion and
wealth ownership? Asset allocation, or portfolio behavior, is certainly an
important contributingfactor.Asset allocation refersto decisions about how
to savemoneythat a familyis ableto saveor invest.The most simpledistinctions
in asset allocation are between real assets and financial assets, and decisions
within each of these categoriesvary in the degreeto which they are risky,with
riskier assets typically creating higher returns. Table4 also includes logistic
regression coefficient estimates for models predicting the likelihood of
homeownership. Purchasing a home is traditionally one of the first major
investmentsAmericansmake, and homeownership,while it can be lucrative,
Religion and Wealth/ 199

TABLE 6: Percent of People Following Three Typical Trajectories by
Religious Affiliation in Childhood
Trajectory
All
Jewish ConservativeMainlineCatholic
asset-poor
Permanently
Earlytransitionto cashandhome
Earlytransitionto financialwealth
.04
.17
.02
.01
.35
.33
.15
.03
.00
.09
.22
.07
.07
.20
.04
is a relatively low-risk investment. Consistent with the diaspora hypothesis, the

results in Table 4 suggest that Jews are less likely than other families to purchase
a home. Conservative Protestants and Catholics are somewhat more likely to
become homeowners, although the association between these faiths and
homeownership is only moderately different from 0. The slightly increased
propensity for conservative Protestants and Catholics to own homes reflects
the somewhat more traditional values espoused by these faiths. In general, the
distinctiveness of Catholics is fading. However, Catholics do demonstrate a
slightly increased propensity to become homeowners, perhaps reflecting a
residual effect of Catholic tendencies to have larger families.
What is perhaps more telling of their eventual wealth is the degree to which
a family is invested in a particular asset or set of assets. For instance, buying a
single share of stock will seldom affect overall wealth in a noticeable way, while
buying hundreds of thousands of dollars in stocks will certainly shape overall
net worth. Table 5 includes generalized least-squares coefficient estimates for
models of the total value of financial assets (e.g., stocks, bonds, savings accounts,
checking accounts) and the total value of the home for those who are
homeowners. The results in this table demonstrate that Jews own substantially
more financial assets than all other families. The magnitude of the relationship
between the indicator of Jewish religious affiliation and financial assets suggests
that financial assets do account for much of the Jewish wealth advantage. This
finding is consistent with the diaspora hypothesis and with the argument that
social capital affords Jews advantages in investing that are not available to nonJews. The particularly large (relatively to other variables in both models)
relationship between the indicator of adult Jewish affiliation and financial
assets again suggests that while childhood process are important, there is a
particular advantage to connections and other benefits of adult affiliation for
Jews. The results in Table 5 also demonstrate that when they become
homeowners, Jews own much more valuable homes than those of other faiths.
Protestants, both conservative and mainline, buy houses that are less valuable
on average than the houses of those with no affiliation. In both models of home
value, I control for selection bias by including the predicted value of
homeownership. The negative effect of Protestantism certainly accounts for
some of the overall wealth disadvantage associated with being Protestant, but
2003
the effectis likelyovershadowedby the Jewishadvantagein acquiringfinancial
assets.
The pathspeople take duringtheirfinanciallives can also affectadultwealth
in criticalways. For instance,saving earlyin life can disproportionatelyaffect
adult wealth because of compounding.Naturally,early saving in high return
financialinstrumentscan have an even more noticeableimpact.An important
part of the financialrepertoirethat childrenlearn, and that can be associated
with the family'sreligiouspreferences,is a propensityto begin savingearlyor
to save in particularways. Table 6 includes simple descriptivestatistics that
emerge from optimal matching. Using optimal matching on savings in five
particularlycommon assetsat variouslevels of risk (savingsaccounts,checking
accounts, homeownership, bonds, and stocks), I identified three common
financialtrajectories.The firsttrajectory,permanentlyasset-poor,includesthose
who never own any of the five assets.The second trajectory,earlytransitionto
cash and home, includes all trajectoriesthat reflect some ownershipof these
assetswith a tendencyto startby owning the low-riskassets(savings,checking)
in early adulthood, moving to homeownership, and then perhaps acquiring
riskierfinancialassets (stocks, bonds) later in life. The third trajectory,early
transitionto financialwealth, includes all trajectoriesthat again include some
ownership of these assets but that include relativelyearly transitions to the
ownershipof financialassets (stocks,bonds).16
The results presentedin Table6 indicate that only 4% of the full sample
remainedpermanentlyasset-poor throughout their lives. Of those who were
raisedas conservativeProtestants,15%remainedasset-poorthroughouttheir
adult lives. In contrast, only 1% of those who were raised as Jews remained
asset-poor.'7Earlytransitionto cash and home is perhapsthe most traditional
of these financialtrajectories,and 17%of the full sample followed a path that
could be classifiedunder this heading.Of those raisedas mainlineProtestants,
22% followed this path. Similarly, of those raised as Catholics, 20% were
relativelytraditionalin their overall financialdecision making. Perhapsmost
instructiveis the enormouslydivergentproportionof Jewsto follow the most
high-risk,high-returntrajectory,the earlytransitionto financialwealth.In the
full sample,only 2% of respondentsfollowedthis path. Among those raisedas
Jews,however,a full 33% took a financialpath that could be classifiedin this
high-riskcategory.Not surprising,none of the conservativeProtestantsfollowed
this final path and only 7% and 4% of the mainlineProtestantsand Catholics,
respectively,followedthis path.18Theseresultsimplythat the repertoireof skills
and decision-makingabilitieslearnedin childhood may very well set a course
of action that ultimatelytranslatesinto high wealth.
TABLE6: Percentof People FollowingThree TypicalTrajectoriesby
ReligiousAffiliationin Childhood
ConservativeMainlineCatholic
Trajectory
All
Jewish
assetpoor
Permanently
Earlytransitionto cash
andhome
Earlytransitionto
financialwealth
.04
.01
.15
.09
.07
.17
.35
.03
.22
.20
.02
.33
.00
.07
.04
Discussion
Recentresearchon wealthownershiphas identifiedextremeinequalitiesin the
distributionof asset ownership,but this literaturehas only begun to identify
the processesthat account for wealth inequality(Conley 1999; Keister2000a;
Oliver & Shapiro 1995). In this article, I have argued that religion is an
importantdeterminantof wealth ownership,and I have empiricallyidentified
importantpatternsin the relationshipbetweenreligionand wealth that isolate
the mechanisms underlyingthese relationships.I have argued that religious
affiliationin childhood and adulthood can shape action indirectlyby altering
fertility and marriagebehavior, educational attainment,work behavior, and
other behaviors and processes that influence wealth ownership. I have also
argued, however, that religion is an important element of culture. As such,
religion directly affects wealth accumulation by defining the goals people
identify as important, by creating a repertoire of skills and knowledge that
people draw on when making decisions, and by determining the nature of
people'ssocial contacts.When they are exposedto religiousceremonies,rituals,
and values,people develop a set of competenciesand habitsthat they drawon
in makingdecisionsaboutconsumption,saving,and investment.Affiliationwith
a religious group also creates social capital that may improve understanding
of saving and investing and may actuallyprovide investmentopportunities.
I identifieddistinctpatternsin the relationshipbetweenreligiousaffiliation
and wealth ownership.Those who were raisedJewishor practicedJudaismas
adultsowned considerablymore wealththan non-Jews.I showed that Jewsare
more likelyto receivean inheritance,but I also show that the wealthadvantage
enjoyed by Jews does not fade when inherited wealth is controlled. I
demonstratedthat Jewsown more high-risk,high-returnfinancialassetsthan
non-Jews and that Jews are less likely to own a house. Both findings are
consistent with the diasporahypothesisthat arguesthat, for historic reasons,
Jews have a preferencefor human capital and other types of capital that are
transportableratherthan fixed. I also found evidenceto supportmy argument
that Jews are more likely to follow a financial trajectorythat involves early
investment in financial assets such as stocks and bonds and a relativelylate
2003
transition to the ownership of fixed assets. Following this type of trajectory
facilitiesearly capitalaccumulationthat can compound considerablyover the
life course and be used as collateral for other investments, in the direct
acquisition of assets such as a home, and as buffer that can reduce the risk
inherantin risky investments.
It is also noteworthy that I identified a distinct negative relationshipbetween affiliationwith a conservativeProtestantchurch and wealth ownership.
In direct contrastto Jews, conservativeProtestantsowned less overallwealth
and fewerfinancialassets.They were somewhatmore likelyto own a home a reflection of their relativelytraditionalvalues - although the homes they
owned were significantlyless valuable than the averagehome. Conservative
Protestantswere more likelyon averageto remainasset-poorthroughouttheir
lives, and when they did acquireassets,they made a somewhatearlytransition
to homeownershipand only seldom acquiredfinancialassets.Consistentwith
my expectations, I also found no significant difference between the wealth
ownership of mainline Protestantsand Catholics, and affiliationwith either
of these churches neither increasednor impeded wealth accumulation.
It is importantto note that in emphasizingthe relationshipbetweenreligion
and wealthownership,I do not intend to reducethe complexprocessof wealth
accumulationor wealth inequalityto a single set of inputs. In other research,
I have documented that wealth ownership is associated with a number of
factors, including individual and family processes such as family structure,
marital behavior, and union separation and aggregate processes such as
demographic trends, market fluctuations, and policy shifts (Keister2000b;
Keister & Moller 2000). The results that I discuss in this article, however,
highlightan importantpart of the picturethat has been neglectedpreviously.
Understandingthat religion is relatedin criticalways to wealth accumulation,
net of its indirect effects on other demographicbehaviors,casts light on the
importanceof family processesthat shape the way people behave and, in this
case, the way they accumulateassets. My results also identify the importance
of longitudinalpatternsof savingbehaviorin shapinga person'slifetimewealth
ownership.The person who startslife without the knowledgeor skills to save
or the understandingof how to save startsat a distinctdisadvantage.Likewise,
understanding that social contacts can facilitate this process suggests that
providingopportunitiesfor those without these criticalcontacts to gain both
the information and opportunities the social relations may supply could
enhance efforts to increaseequality.
My results suggest that efforts to improve understanding of different
methods of savingas well as effortsto remove structuraland policy barriersto
investmentmight lessen wealthinequality.As Caskey(1994) noted, our society
devotes very few resourcesto monitoring and regulatingfringe banking.We
devote considerable resources to protecting consumers in the middle- and

upper-income brackets, but almost nothing to protect low-income consumers
of financial services. Not only would additional protection be desirable, but the
evidence reported here suggests that if we provided opportunities and
incentives to low-income, low-wealth households to save and to invest in more
long-term, sounder financial instruments, we could go a long way toward
reducing wealth inequality. Eliminating wealth inequality, however, is not as
easy. Persistent inequality in wealth ownership is deeply embedded in
inequality in earnings, education, and other behaviors and processes that
indirectly affect wealth ownership. Not until these disparities are eliminated
will current differences in asset ownership be reduced to tolerable levels.
Notes
1. All dollarvalues are 2000 dollars,convertedusing the ConsumerPrice Index.
2. The diasporahypothesisdoes not implythattoday'sadultJewsin Americahavea sense
that their physicalassetsmight be taken,ratherthe hypothesissuggeststhat familyand
communitytraditionshavedevelopedthatplacea relativelyhighvalueon humancapital
and financialassets.
3. Samplesizes in the tablesare slightlysmalleras a resultof missingvalues.I found no
significantwealthor religiondifferencesbetweenthoseincludedin my sample(N = 3,054)
and the full sample.I use all 11 yearsto take advantageof the changesin respondents'
wealthover time. Manyrespondents'wealthchangednoticeablyduringthis period,and
using all the years takes this into account. Requiringvalid data for all years may be
restrictive,but sensitivitytests indicatedthat the results for the key test variablesare
robust.
4. Modelingthe value of the inheritancereceivedproducedsimilarresults.
of the denominations,including
5. I experimentedwith a verywide rangeof classifications
categorizing those who are evangelical as conservative regardless of the standard
classificationandincludingrelativelysmallsectsandrelativelydistinctsectsof conservative
Protestantsin their own categoriesand in variouscombinationsof categories.Acrossa
multitudeof specifications,the resultsdid not changein a substantiveway.Therewas
no discernabledifferencebetween across denominationswithin the main categoriesI
include (e.g., conservativeProtestant).Includinga separateindicatorfor Mormonsdid
not change the results.
6. The omittedcategoryalsoincludesthe smallnumberof respondentswho wereaffiliated
with religions such as Taoism that are not explicitly modeled. Including a separate
indicatorfor other religionsand omitting these respondentsentirelydid not affectthe
results.
7. Unfortunatelythere is no indicatorof familywealth during childhood.Inheritance,
however,reflectsthe relativesize of the family'sassets fairlyaccurately.

8. Including a measure of number of years of education for the parents was less
informative than controlling for various levels of education completed. Because
completionof less than a collegeeducationwas not significantlyrelatedto adultwealth,
I includedonly one dummyvariable.
9. Descriptiveanalysessupportthe popularperceptionthat a disproportionatenumber
of the old rich were born in the Northeast.
10. Decomposing the total number of siblings in various ways (e.g., number older,
numberyounger)did not improvethe fit of any of the models.Likewise,controllingfor
position among siblingsand relatedmeasures(e.g., oldest,youngest,middle,spacingof
siblings, gender of siblings) did not improve the explanatorypower of the models. I
include stepsiblingsand half-siblingswho lived in the home. Removingnonbiological
siblingsfromthe analysisand includingonly siblingswho livedin the home for extended
periodsdid not substantivelyaffectthe results.
11. Demographicstudiesshow distinctivepatternsof familyformationand structureby
religion.To capturethe potentialeffectsof this on wealthaccumulation,I experimented
with various detailed maritalstatus controls (e.g., controls for those who were never
married,divorceandtimingof divorce,widowhoodandtimingof widowhood,andvarious
combinations of marital status changes). The controls I included reflect the model
specificationthat is most consistentwith prior researchand with patternsin the data.
12. Controllingfor residencein New YorkCity,other specificlocations,housing price
variations,and other regionalindicatorsdid not affectthe results.
13. I do not includewealthestimatesby adult religiousaffiliationbecausealthoughthe
estimatesnaturallydifferfromthose reportedin Table2, the patternsarealmostidentical.
14. The proportionof respondentsin the NLS-Ywho owned their own homes in 1998
was somewhatlowerthan otherestimatesof homeownershipbecausethe NLS-Ysample
is slightlyyoungerthan a nationallyrepresentedcross section of the U.S. Longitudinal
increasesin homeownership,however,in the NLS-Yareconsistentwith other estimates,
suggestingthat the underlyingpatternsin the NLS-Yare consistentwith patternsthat
are apparentin the overallpopulation.
15. The effectof the religiousfactorswas much strongerbefore I addedthe controls.I
do not include the "gross"resultsbecause the fully specified models reflect the true,
underlyingprocessmore accurately.
16. Eachtrajectorythat I have groupedunder these names includesmultiplepaths that
optimalmatchingidentifiedas relativelysimilar.These arenot the only trajectoriesthat
emerged,but they are the most common paths followedby those in the sample.
17. Reportingthese resultsas a function of adult religiousaffiliationproducessimilar
patterns.
18. It is possibleto conductmultivariateanalyseswith the pathsas the dependentvariable
defined as either dichotomousor multinomial.I have conductedthese analyses,and I
can makethe resultsavailableupon request.I do not reportthem hereto conservespace
and because the resultsdo not add substantiallyto the resultsthat are evident in the
simple descriptivestatistic.

References
Abbott,Andrew,and AlexandraHrycak. 1990."MeasuringResemblancein SequenceData:An
Optimal MatchingAnalysisof Musicians'Careers."AmericanJournalof Sociology96:14485.
Alwin, Duane. 1986."Religionand ParentalChildrearingOrientations:Evidencefor a CatholicProtestantConvergence."AmericanJournalof Sociology92:412-20.
Bourdieu,Pierre. 1977. Outlineof a Theoryof Practice.CambridgeUniversityPress.
Brenner,Reuven,and Nicholas M. Kiefer.1981. "The Economics of Diaspora:Discrimination
and OccupationalStructure."EconomicDevelopmentand CulturalChange29:517-33.
Caskey,Richard.1994.FringeBanking:Check-CashingOutlets,Pawnshops,and the Poor.Russell
Sage Foundation.
Chiswick, Barry R. 1986. "LaborSupply and Investment in Child Quality:A Study of Jewish
and Non-JewishWomen."Reviewof Economicsand Statistics68:700-703.
. 1988."Differencesin Educationand EarningsAcrossRacialand EthnicGroups:Tastes,
Discrimination,and Investmentsin Child Quality."QuarterlyJournalof Economics103:57197.
1993."The Skillsand Economic Statusof AmericanJewry:Trendsover the LastHalfCentury."Journalof LaborEconomics11:229-42.
Chiteji,Ngina S., and FrankStafford.2000. "AssetOwnershipacrossGenerations."JeromeLevy
InstituteWorkingPaper.
Conley,Dalton. 1999.BeingBlack,Livingin the Red:Race,Wealth,and SocialPolicyin America.
Universityof CaliforniaPress.
Darnell, Alfred,and Darren E. Sherkat.1997. "The Impact of ProtestantFundamentalismon
EducationalAttainment."AmericanSociologicalReview62:306-15.
DellaPergola,Sergio. 1980."Patternsof AmericanJewishFertility."Demography17:261-73.
Downey, Douglas B. 1995. "When Bigger Is Not Better:Family Size, ParentalResources,and
AmericanSociologicalReview60:746-61.
Children'sEducationalPerformance."
Ellison, Christopher G., John P. Bartkowski, and Michelle L. Segal. 1996. "Conservative
Protestantismand the ParentalUse of CorporalPunishment."SocialForces74:1003-28.
Jones,CharlesF, and EliseF.Westoff.1979."TheEnd of CatholicFertility."
Demography16:20918.
Juster,Thomas F.,and KathleenA. Kuester.1991. "Differencesin the Measurementof Wealth,
Wealth Inequality, and Wealth Composition Obtained from Alternative U.S. Wealth
Surveys."Reviewof Incomeand Wealth37:33-62.
Juster,Thomas F.,JamesP.Smith, and FrankStafford.1999."TheMeasurementand Structure
of Household Wealth."LabourEconomics6:253-75.
Kalmijn,Matthijs.1991."ShiftingBoundaries:Trendsin Religiousand EducationalHomogamy."
Keister,LisaA. 2000a. "Raceand WealthInequality:The Impact of RacialDifferencesin Asset
Ownershipon the Distribution of Household Wealth."SocialScienceResearch29:477-502.
.2000b. Wealthin America.CambridgeUniversityPress.
206 / Social Forces 82:1, September 2003

Keister,LisaA., and StephanieMoller. 2000. "WealthInequalityin the United States."Annual
Reviewof Sociology26:63-81.
Kennickell,Arthur B. 2000. "AnExaminationof Changes in the Distribution of Wealth From
1989-1998."FederalReserveBoardWorkingPaper.
Kennickell,Arthur B., Martha Starr-McCluer,and Annika E. Sunden. 1997. "FamilyFinances
FederalReserveBulletin
in the U.S.:RecentEvidencefrom the Surveyof ConsumerFinances."
(January):1-24.
Lazerwitz,Bernard.1995."Jewish-Christian
Marriagesand Conversions,1971and 1990."Sociology
56:433-43.
of Religion.
Journalof PopulationEconomics
Lehrer,EvelynL. 1996a."ReligionAs a Determinantof Fertility."
9:173-96.
.1996b."TheRoleof the Husband'sReligionon the Economicand DemographicBehavior
of Families."Journalfor the ScientificStudyof Religion35:145-55.
.1998. "ReligiousIntermarriagein the United States:Determinantsand Trends."Social
ScienceResearch27:245-63.
.1999. "ReligionAs a Determinantof EducationalAttainment:An EconomicPerspective."
SocialScienceResearch28:358-79.
Lehrer,EvelynL.and CarmelU. Chiswick.1993."Religionas a Determinantof MaritalStability."
Demography30:385-404.
Lenski,Gerhard.1961. TheReligiousFactor:A SociologicalStudyof Religion'sImpacton Politics,
Economics,and FamilyLife.Doubleday.
Mechanic,David, and StephenHansell. 1989."Divorce,FamilyConflict,and Adolescents'WellBeing."Journalof Healthand SocialBehavior30:105-16.
Menchik, Paul L., and Nancy A. Jiankoplos. 1998. "Economics of Inheritance."Pp. 45-59 in
Inheritanceand Wealthin America,editedby J.RobertK.Millerand S. J.McNamee.Plenum.
Oliver,Melvin 0., and Thomas M. Shapiro. 1995. BlackWealth/WhiteWealth.Routledge.
Sandefur,GaryD., and ThomasWells.1999."DoesFamilyStructureReallyInfluenceEducational
Attainment?"SocialScienceResearch28:331-57.
Sander,William. 1995. The CatholicFamily:Marriage,Children,and Human Capital.Westview
Press.
Sherkat, Darren E., and Christopher G. Ellison. 1999. "Recent Developments and Current
Controversiesin the Sociology of Religion."AnnualReviewof Sociology25:363-94.
Embattledand Thriving.Universityof Chicago
Smith,ChristianS. 1998.AmericanEvangelicalism:
Press.
Spilerman,Seymour. 2000. "Wealthand StratificationProcesses."Annual Review of Sociology
26:497-524.
Stryker,Robin. 1981. "Religio-EthnicEffects on Attainments in the EarlyCareer."American
SociologicalReview46:212-31.
Swidler,Ann. 1986."Culturein Action: Symbols and Strategies."AmericanSociologicalReview
51:273-86.
Thornton, Arland. 1985. "ChangingAttitudes towards Separation and Divorce: Causes and
AmericanJournalof Sociology90:856-72.
Consequences."

AttainmentandExpressions
of JewishIdentification,
Wilder,EstherL. 1996."Socioeconomic
1970 and 1990."Journalforthe ScientificStudyof Religion35:109-27.
1998."EthnicandReligious
of theJewish
Wilder,EstherL.,andWilliamH.Walters.
Components
IncomeAdvantage,
1969-1989."
Inquiry68:426-36.
Sociological
Wilcox, W. Bradford.1998. "ConservativeProtestantChildrearing:Authoritarianor
Authoritative?"
Richer:LatestDataon HouseholdWealth
Wolff,EdwardN. 1995."TheRichGetIncreasingly
inPoliticsandSociety,
vol.5, editedbyR.E.Ratcliff,
Pp.33-68in Research
duringthe 1980s."
MelvinL.Oliver,andThomasM. Shapiro.JAIPress.
Wealth."
.1998."Recent
Trendsin theSizeDistribution
of Household
Journal
ofEconomic
Perspectives12:131-50.
Wuthnow,Robert. 1999. GrowingUp Religious:Christiansand Jewsand theirJourneysof Faith.
BeaconPress.
andEconomicBehavior."
Wuthnow,Robert,andTracyL.Scott.1997."Protestants
Pp.260-95
in New Directionsin AmericanReligiousHistory,edited by Harry S. Sout and D.G. Hart.
OxfordUniversityPress.

Religion and Wealth Keister+2003

Transféré par

Informations du document

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Religion and Wealth Keister+2003

Transféré par

Droits d'auteur :

Formats disponibles

Religion and Wealth: The Role of Religious Affiliation and Participation in Early Adult Asset

Religionand Wealth:The Role of Religious

176 / SocialForces 82:1, September2003

Religion and Wealth Ownership

178 / SocialForces 82:1, September2003

A host of unique demographictraitsthat lead to high levels of attainmentfor

In direct contrastto Jewishfamilies,wealth accumulationamong conservative

MainlineProtestantsand RomanCatholicswere at one time distinctfrom each

Data, Measures, and Methods

184 / SocialForces 82:1, September2003

186 / SocialForces 82:1, September2003

in 1978 to control for whether nonmaterialparentalresourceswere diluted

Religion and Wealth / 187

I used generalized least-squares regression to model net worth, the value of

188 / SocialForces 82:1, September2003

Substitution Costs for Optimal Matching

affiliationand participationis relativelyhigh, I report separateestimates for

Religion and Wealth / 189

190 / SocialForces 82:1, September2003

.67 had everreceivedan inheritancecomparedto .40 in the full sample.Among

192 / SocialForces 82:1, September2003

Bornin the Northeast

Religion and Wealth I 193

Bornin the U.S.

High school graduate

Notes:Standarderrorsarein parentheses.Samplesize (n) reflectsthe numberof respondentsincludedin each

194 / SocialForces 82:1, September2003

Logistic ParameterEstimatesfor Inheritanceand Homeownership, 1985-1998

Bornin the Northeast

Religion and Wealth / 195

Logistic ParameterEstimatesfor Inheritanceand Homeownership, 1985-1998

Bornin the U.S.

High school graduate

196 / Social Forces 82:1, September 2003

Bornin the Northeast

Religion and Wealth / 197

Notes:Standarderrorsarein parentheses.Samplesize (N) reflectsthe numberof respondentsincludedin each

198 / SocialForces 82:1,September2003

Jewishreligious affiliation,it also is possible that the differencebetween the

Religion and Wealth/ 199

is a relatively low-risk investment. Consistent with the diaspora hypothesis, the

Religion and Wealth/ 203

204 / SocialForces 82:1, September2003

Religion and Wealth/ 205

206 / Social Forces 82:1, September 2003

Religion and Wealth/ 207

Vous aimerez peut-être aussi