A comparative analysis of the nativity wealth gap.
Bauer, Thomas K. ; Cobb-Clark, Deborah A. ; Hildebrand, Vincent A. 等
I. INTRODUCTION
In many countries worldwide, substantial numbers of immigrants are
at (or near) retirement age. In the United States and Germany, for
example, immigrants are as likely as the native born to be over the age
of 55 yr, while in Australia the foreign-born population is currently
aging more rapidly than the native-born population (Bauer et al., 2007).
We know very little about immigrants' capacity to provide for
themselves in old age, although there are reasons to believe that both
wealth levels and portfolio allocations depend on nativity (Amuedo-Dorantes and Pozo, 2002; Cobb-Clark and Hildebrand, 2006a,
2006b). Given this, it is important to know more about the way in which
immigrants accumulate wealth and achieve economic security.
This paper addresses this issue by analyzing the wealth gap between
immigrant and native households in Australia, Germany, and the United
States. We take advantage of recent data collections in each of these
countries using a unique opportunity to assess the wealth position of
immigrant households. Differences in survey design and coverage will
prevent us from directly comparing wealth levels across countries.
However, following studies of cross-national differences in the gender
wage gap (Blau and Kahn, 1992), we will assess how the nativity wealth
gap differs across countries. We are particularly interested in
addressing the following questions: First, how does household net worth
vary by nativity status, region of origin, and immigration cohort?
Second, how important are factors such as income differentials,
disparities in educational attainment, and demographic characteristics
in producing these wealth gaps? Finally, what can we say about the role
that the institutional setting might play in generating any nativity
wealth gap?
These are important questions given that wealth provides the
resources necessary to maintain living standards in retirement or in
times of economic hardship. Moreover, wealth is a fundamental component
of overall economic wellbeing that directly influences immigrants'
ability to integrate into host country society. Wealthier families have
more political influence and live in neighborhoods with better schools,
enhanced health facilities, and less crime (Altonji and Doraszelski,
2005; Gittleman and Wolff, 2004). At the same time, immigrants"
economic wellbeing is almost certainly linked to the institutional
setting, making it difficult to generalize research findings across
national boundaries.
Australia, Germany, and the United States provide an interesting
case study because they span the spectrum from a traditional immigration
country accepting mainly permanent, skilled immigrants (Australia) to a
country with a long history of accepting only temporary, predominantly unskilled workers (Germany). The results of our comparative analysis are
necessarily inferential, but nonetheless provide some important insights
into the ways that the institutional framework might influence the
relative wealth position of immigrant families.
Our results indicate that in Germany and the United States wealth
differentials are largely the result of disparity in the educational
attainment and demographic composition of the native and immigrant
populations, while income differentials are less important in
understanding the nativity wealth gap. In contrast, the relatively small
wealth gap between Australian- and foreign-born households cannot be
explained by either the distribution of income or immigrants'
characteristics. Indeed immigrants would have a wealth advantage if they
accumulated wealth in the same way as the Australian born. On balance,
our results point to substantial cross-national disparity in the
economic well-being of immigrant and native families that is largely
consistent with domestic labor markets and the selection policies used
to shape the nature of the immigration flow.
II. THE INSTITUTIONAL SETTING
The comparative nature of our analysis provides an opportunity to
shed light on the ways in which a country's institutional setting
might lead to a nativity gap in wealth levels. Conceptually, variation
in household wealth stems from differences in inherited wealth, rates of
return, or savings behavior, which in turn is a function of both income
and consumption patterns. Given this, there are several ways in which
both the wealth levels and portfolio choices of immigrants may differ
from those of their native-born counterparts. In this section, we
briefly discuss some of the institutional differences in the three
countries that may affect the incentives to accumulate wealth (see Bauer
et al., 2007 for details).
Migration policy is designed to select the individuals, from the
set of potential migrants, to be allowed to enter. Selection processes
typically result in differences between natives and immigrants in terms
of observable and unobservable characteristics generating disparity in
wealth levels. As the nature of these selection processes varies across
receiving countries, it is likely that the nativity wealth gap will
differ across receiving countries as well. The United States, for
example, places more weight on the reunification of families, while
Australia mainly accepts economic migrants using numerical testing to
evaluate skills (Birrell, 1990). In contrast, immigration to Germany has
historically been predominantly temporary, although since 1973 German
immigration policy has been almost exclusively based on family
reunification and humanitarian considerations (Bauer et al., 2005:
Schmidt and Zimmermann, 1992). This policy environment results in
immigrant populations that are different in terms of their skills and
settlement intentions. Immigrants to Australia are largely skilled and
permanent. Immigrants to Germany, on the other hand, are mainly
unskilled and, at least notionally, temporary. We expect these
differences to contribute to producing a nativity wealth gap that is
likely to be larger in Germany than in Australia.
Wealth accumulation is closely related to households' long-run
earnings potential and hence to labor market institutions. The complex
minimum wage system provides immigrants with some degree of earnings
protection when they first enter the Australian labor market, while the
greater flexibility of the U.S. labor market speeds the earnings
adjustment of new arrivals. At the same time, the German labor market is
considered to be quite rigid in comparison. Both labor costs and
unemployment in Germany are relatively high and Germany follows a
stronger redistribution policy through higher tax rates (OECD, 2006).
Empirical research on the earnings assimilation of immigrants is
consistent with these institutional differences. Immigrants to Australia
have a lower earnings disadvantage upon entry as well as a slower rate
of earnings assimilation than immigrants to the United States (Chiswick
and Miller, 1985; Miller and Neo, 2003), while in Germany, immigrants
face an even higher relative earnings disadvantage at arrival and
experience no significant earnings assimilation over time (Bauer et al.,
2005). These differences lead us to expect that the nativity wealth gap
would be lower in Australia and the United States than in Germany, where
immigrants remain at an earnings disadvantage.
Institutional barriers associated with ethnicity, nativity, legal
status, or language skills may limit migrants' access to financial
markets and hinder the purchase of certain assets like housing. Opening
a bank account, for example, may be difficult for some immigrants due to
a general lack of documentation, while other new arrivals may simply
find it difficult to establish a credit rating (Osili and Paulson,
2004). Furthermore, legal systems and laws of land tenure, governmental
housing and tax policy, and lenders' margins--all of which vary
substantially between countries--combine to produce an institutional
setting that may be more or less favorable to home ownership (Proxenos,
2002). The inability--or unwillingness--to fully participate in the host
country's financial markets together with various constraints in
the housing market are almost certainly related to the common finding
that immigrant households have significantly lower home ownership rates
than comparable native households (Borjas, 2002; Clark et al., 1997;
Coulson, 1999; Myers and Lee, 1996; Painter et al., 2003; Sinning,
2006).
Limited access to social welfare programs may also affect wealth
accumulation (Gruber and Yelowitz, 1997; Shamsuddin and DeVoretz, 1998).
In 1996, for example, the United States restricted noncitizens'
access to the welfare system (Fix and Passel, 2002), while Australia has
subsequently extended the period that new immigrants must wait before
becoming eligible for income support (Chiswick and Miller, 2006). In
contrast, the German social assistance system does not differentiate
between natives and immigrants although refugees and asylum seekers
receive lower benefits until their asylum status is recognized (Nielsen,
2004). Despite these restrictions, immigrants have on average higher
dependency rates on social welfare in all three countries considered in
this paper. Note however, that these differences become very small or
even disappear when controlling for socioeconomic characteristics. While
the probability of receiving social welfare falls with duration of
residence in Germany, it seems to increase in the United States. (1)
Finally, differences in pension systems are also of particular
interest as these institutional differences affect the incentives to
accumulate wealth. The Australian pension system has two components: a
means-tested, aged pension financed through general tax revenue, and
private pension plans financed by mandatory worker and employer
contributions (Gong et al., 2006). The U.S. Social Security program is
also financed through general tax revenue; however, unlike the
Australian aged pension, benefits are paid under uniform formulas based
on recipients' previous earnings histories without regard to wealth
or nonwage income. Germany's pay-as-you-go state pension system,
which is financed by contributions from employers and employees, is
similar to U.S. Social Security in that it is not means tested. Other
retirement savings vehicles are relatively unimportant in Germany.
Comparing average pension wealth, that is, the present value of a future
stream of pension payments can be a useful way of summarizing the effect
of the various institutional aspects of specific pension systems (OECD,
2005). These calculations indicate that the net pension wealth for
individuals with average earnings is highest in Germany ($262,000)
followed by Australia ($189,000) and the United States ($183,000). (2)
We can only speculate about how these cross-country differences in
welfare and pension systems affect the nativity wealth gap. Restricted
access to the welfare system and the relatively low replacement rate of
the U.S. pension system may give immigrants an incentive to accumulate
personal wealth to cope with any financial difficulties and provide for
their own retirement. Thus, we might expect the nativity wealth gap to
be smaller in the United States than in Australia or Germany.
III. DATA AND DESCRIPTIVE ANALYSIS
A. Data Sources
This paper uses data from three nationally representative panels
that all provide information about migration histories and comparable
measures of household wealth. The Australian data come from the
Household, Income and Labour Dynamics in Australia (HILDA) Survey, which
is a longitudinal survey of Australian households (Wooden et al., 2002).
Our analysis exploits the 2002 release of HILDA (Wave 2) encompassing
approximately 13,000 individual respondents living in more than 7,000
households. Wave 2 included a special module on household wealth making
the HILDA Survey the only Australian microdata source that allows
questions regarding nativity and the distribution of wealth to be
examined.
The data used to study the nativity wealth gap in Germany have been
extracted from the German Socioeconomic Panel (GSOEP). The GSOEP is a
representative longitudinal survey that started in 1984 and includes
German and immigrant households. In 2005, approximately 22,000 persons
in nearly 12,000 households were sampled. Here we use data from the 2002
wave. (3) As less than 5% of the foreign-born population lives in East
Germany (Statistisches Bundesamt, 2006), our analysis focuses on
households residing in West Germany. Immigrants are defined as
foreign-born individuals who immigrated to Germany after 1948 (including
foreign-born persons with German citizenship).
Data for the United States come from the 2001 panel of the Survey
of Income and Program Participation (SIPP). The SIPP is the only U.S.
microdata set containing information about migration histories and
wealth for a broad cross-section of the immigrant population (Cobb-Clark
and Hildebrand, 2006a). The 2001 SIPP panel is a short rotating panel
that is representative of the civilian, noninstitutionalized population
living in the United States. Approximately 35,000 households were
interviewed in the first wave of the 2001 panel. Each household in the
sample was interviewed at 4-mo intervals over a period of roughly 3 yr
beginning in February 2001. Information on immigrants is obtained from
the Wave 2 migration history module and household wealth data are drawn
from the Wave 3 module on assets and liabilities collected between
October 2001 and January 2002. (4)
To facilitate comparisons across countries, we restrict the three
estimation samples to native- and foreign-born, couple-headed households
in which the reference person is between 25 and 75 yr old. Native-born
households include all coupled-headed households in which both partners
were born in Australia, Germany, or the United States, respectively. A
couple-headed foreign-born household is a household in which at least
one partner was born outside of the respective country. This definition
allows an explicit consideration of "mixed households" in
which one partner is native born and the other is foreign born as well
as "immigrant-only" households. In the case of mixed
households, the foreign-born partner is taken to be the reference
person, regardless of whether this person was reported to be the head or
the spouse of his or her respective household. In the case of
native-born and immigrant-only couples, the reference person is the
person who is designated as head of the household. Excluding all
observations with missing values on one or more of the variables of
interest results in samples of 2,235 native-born and 1,271 foreign-born
households in Australia, 2,344 native-born and 691 foreign-born
households in Germany, and 10,283 native-born and 1,671 foreign-born
households in the United States.
B. The Level of Net Worth
Our measures of total household net worth are derived from wealth
components that are either estimated at the household level or directly
measured at the individual level and then aggregated to the household
level. An overview of the specific components of each country's net
worth measure is provided in Table A1. As the scope of financial
investments differs across countries, each survey asks about a somewhat
different range of asset types. Moreover, while information about many
detailed forms of country-specific financial investments was gathered in
both Australia and the United States, only a single broadly defined
measure of the value of financial assets is observed for Germany.
Fortunately, the resulting measure of financial wealth for Germany is
comparable to those derived from both the Australian and U.S. data.
Unlike financial wealth, the remaining components of our net worth
measures (business equity, housing equity, vehicles equity, and total
value of debt) are largely the same across all three countries. (5)
To assess the quality of HILDA's wealth information, Heady (2003) compares HILDA's estimate of aggregate net worth to national
statistics compiled by the Reserve Bank of Australia (RBA). This
comparison of aggregate values suggests that the overall household net
worth derived from HILDA data is fairly accurate. Although the aggregate
value of total debt derived from HILDA is 20% lower than the
corresponding RBA figure, HILDA's overall estimate of household net
worth is only 6.5% lower than the RBA estimate.
For Germany, we use a revised version of the 2002 wealth module of
the GSOEP that accounts for measurement errors and was made available in
2007. Frick et al. (2007) provide an extensive description of editing
and imputation procedures that were applied to obtain the revised wealth
information. They also compare the wealth aggregates of the GSOEP with
corresponding information from official national statistics (national
accounts and data from the German central bank). On balance, their
findings suggest that real estate--the largest wealth component--is
captured well by the wealth module of the GSOEP 2002. At the same time,
the coverage of more heterogeneous and diversified wealth holdings (such
as financial assets) is much lower. Overall, the aggregated total net
wealth in the GSOEP is about 80% of the corresponding macroinformation.
In general, SIPP data are not usually thought of as the best source
of information for studying U.S. wealth. The Survey of Consumer Finance
(SCF) provides a more comprehensive measure of wealth than do
alternative data sources that measure the upper tail of the wealth
distribution particularly poorly (Juster and Kuester, 1991; Juster et
al., 1999; Wolff, 1998). Unfortunately, SCF data do not identify
immigrants making it unsuitable for our purposes. In an effort to
understand the limitations of SIPP data, the Social Security
Administration (2003) compares SIPP wealth data to wealth levels in the
SCF and the Panel Study of Income Dynamics (PSID). The results reveal
that the SIPP estimate of median net worth is only about two-thirds that
of the SCF and 74% that of the PSID. As estimates of liabilities are
essentially the same across surveys, much of the underestimate of net
worth results from assets being underestimated. Underestimation of
assets in the upper tail of the wealth distribution accounts for about
72% of the difference in the SIPP and SCF estimates of asset levels,
while asset categories not measured in the SIPP account for an
additional 13%. Only a very small proportion (at most 15%) of the
disparity in assets is attributable to underestimation of assets among
the nonelderly (Social Security Administration 2003). Thus, we minimize
the disadvantages of using SIPP data to assess wealth holdings in the
United States by eliminating very wealthy households from our sample and
focusing on median (rather than mean) wealth.
In addition to cross-national differences in the extent to which
our data sources measure overall net worth, there are also important
differences in the way in which pension wealth is measured in each
survey. Information about pension entitlements in the GSOEP is limited
to the cash surrender value of life insurance policies, private pensions
plans, and building savings accounts. These assets are included as a
component of financial wealth in our measure of net worth for Germany.
In contrast to the GSOEP, HILDA also collects data about
employer-provided pension wealth. In the United States, the total value
of funds in Thrift/401K accounts is available in the assets and
liabilities module of the SIPP, while the value of pension benefits in
defined contribution plans could in principle be extracted from the
SIPP's pension module (Wave 7). However, as in most wealth surveys,
the value of assets held in defined-benefit pensions and social security
entitlements are not available in the SIPP. Consequently, to make our
measures of net worth as comparable as possible across countries, we do
not include the available information about employer-provided pensions
(HILDA), Thrift/401K (SIPP) accounts, or defined contribution pension
wealth (SIPP) in our respective measures of net worth for Australia and
the United States.
Given these differences in measurement, direct comparisons of the
level of net worth across countries is not possible. This is not
particularly problematic given our research interests. More importantly,
we have no evidence that the merits of our net worth measure within a
country differ substantially between native-born and foreign-born
populations. Consequently, following the literature on international
comparisons of the gender earnings gap (Blau and Kahn, 1992), our
analysis concentrates on relative rather than absolute wealth measures.
In particular, since wealth levels are not directly comparable across
countries, we focus instead on comparing the proportion of the nativity
wealth gap that is attributable to different (explained or unexplained)
factors. To facilitate such a comparison, we convert our measures of net
worth into U.S. dollars using purchasing power parity (PPPs) conversion
factors provided by the OECD for the year 2002. (6)
Information about the level of wealth held by native-born and
foreign-born households in each country is presented in Tables 1-3.
Overall, native-born households have higher net worth than immigrant
households, though the nativity wealth gap is considerably smaller in
Australia than in Germany or the United States. Specifically, the net
worth of immigrants to Australia is approximately 95% ($282,640) of the
average wealth level of their Australian-born counterparts. Immigrant
households in the United States hold about 68% ($111,868) of the net
worth of native households, while the corresponding share in Germany is
below 40% ($86,744). Moreover, there is very little difference in the
proportion of Australian-born (96.4%) and immigrant households (95.9%)
with positive net worth. In the United States, about 89% of the
native-born and 84% of the foreign-born households report positive
wealth levels. In contrast, while the net worth of 79% of the native
households in Germany is positive, the corresponding proportion among
foreign-born households is only 46%. The lower propensity to hold
positive net worth is only a partial explanation for the overall
nativity wealth gap in Germany and the United States however.
Conditional on having positive net worth, immigrant households still
hold only 56% ($154,291) of the wealth of native households in Germany
and 73% ($136,797) in the United States. In contrast, those immigrant
households with positive net worth hold approximately 96% ($295,515) of
the wealth level of Australian-born households.
Figures 1-3 depict the unconditional nativity wealth gap along with
a 95% confidence interval over the entire wealth distribution for all
three countries. (7) These figures show that in all three countries
native-born households have a wealth advantage over immigrant households
at almost all points of the wealth distribution. In Australia, however,
this wealth gap appears not to be statistically significant in most
parts of the wealth distribution. In contrast, the nativity wealth gap
appears to be increasing in wealth levels in Germany and the United
States. At the 25th percentile, for example, the wealth gap is smallest
in Australia with immigrants holding $11,152 less than natives (Figure
1), followed by the United States with a nativity gap of $15,036 (Figure
3) at the 75th percentile and Germany with a nativity gap of $37,956
(Figure 2). The wealth gap remains smallest in Australia ($12,220)
compared to gaps of $72,519 in the United States and $223,334 in
Germany.
Tables 1-3 further describe the relevant socioeconomic and
demographic characteristics of native and immigrant households. The
cross-country disparities in the characteristics of immigrants and
natives are largely as expected given differences in immigration
policies and institutional settings. In all three countries, immigrant
households have a lower income on average than native-born households.
Again, this disparity is smallest for Australia. Moreover, immigrants in
Australia also have fewer children under the age of 18, are on average
older, and have more education than natives. In Germany and the United
States, these patterns are reversed. There are also differences in the
distribution of the immigrant population across entry cohorts and
sending regions. In Australia and Germany, the majority of the immigrant
population arrived either before 1974 or after 1984, whereas in the
United States almost half of the immigrant population captured in the
SIPP arrived after 1984. Immigrants to Australia primarily stem from
Europe and Asia, while in Germany most immigrant households originate in Central/Eastern Europe or Turkey and immigrants to the United States
come predominantly from Asia and Mexico.
C. Determinants of Net Worth
To understand how wealth levels vary with household
characteristics, it is first useful to analyze the determinants of
household net worth. As the distribution of net worth is usually skewed to the right, linear regression models are typically estimated using a
log transformation of wealth to obtain a log-normally distributed
dependent variable (Shamsuddin and DeVoretz, 1998). A log transformation
is not appropriate for households with zero or negative net worth,
however. Consequently, we estimate a quantile regression model to
analyze the determinants of household net worth at the median of the
distribution.
Specifically, we estimate the following cross-sectional quantile
regression models of the determinants of net worth ([W.sub.hc]) for
native and immigrant households h residing in country c,
(1) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]
where q reflects a specific percentile of the distribution,
[X.sub.hc] contains information about income (i.e., current net
household income), (8) education (in years), and demographic
characteristics (number of children younger than 18, age and age
squared). Additionally, [I.sub.hc] is an indicator variable that equals
1 for both mixed and immigrant-only couples and 0 otherwise. This
specification allows us to compare the total immigrant population
(including mixed couples in which only one partner is foreign born) with
the native-born population. We are also interested in variation in
wealth within the immigrant population, however. Consequently, the model
also includes vectors of exhaustive, mutually exclusive indicators of
family type, entry cohort, and region of origin. Specifically, immigrant
couples are classified into one of the three possible family types:
mixed couple with a native-born head, mixed couple with a foreign-born
head, and immigrant-only couple. [M.sub.hc] is a vector of three
indicator variables capturing these family types, while [D.sub.hc] and
[R.sub.hc] are vectors of indicator variables, capturing arrival cohort
and region of origin, respectively. Finally, [[beta].sup.q.sub.c] is a
vector of the model parameters to be estimated and
[[epsilon].sup.q.sub.hc] is an error term with the usual properties.
[FIGURE 1 OMITTED]
[FIGURE 2 OMITTED]
[FIGURE 3 OMITTED]
Inclusion of indicator variables that are both exhaustive and
mutually exclusive facilitates interpretation of our results. We achieve
identification of the overall constant by restricting the estimated
coefficients to sum to zero, that is we restrict [[SIGMA].sub.m]
[[beta].sup.q.sub.3cm] = 0, [[SIGMA].sub.n] [[beta].sup.q.sub.4cn] = 0,
and [[SIGMA].sub.p] [[beta].sup.q.sub.5cp] = 0 whereto n, and p are the
numbers of family types, immigration cohorts, and regions of origin,
respectively. Consequently, the parameter [[beta].sup.q.sub.2c] can be
interpreted as the overall difference in the wealth level of native-born
and immigrant couples (i.e., the overall nativity wealth gap)
conditional on characteristics, while [[beta].sup.q.sub.3c],
[[beta].sup.q.sub.4c], and [[beta].sup.q.sub.5c] capture deviations
across family types, entry cohorts, and regions of origin, respectively,
from this overall nativity wealth gap.
Finally, we would also like to know whether the relationship
between wealth and household characteristics differs by nativity.
Consequently, we also estimate
[W.sub.hc] = [[gamma].sup.q.sub.0c] + [X.sub.hc]
[[gamma].sup.q.sub.1c] + [I.sub.hc] [[Y.sub.hc][[gamma].sup.q.sub.Yc] +
[Z.sub.hc][[gamma].sup.q.sub.Zc] + [[eta].sup.q.sub.hc],
where [Z.sub.hc] is a vector that includes net income, education,
the number of children in the household younger than 18 yr of age, and a
dummy variable indicating whether the reference person was younger than
18 yr old at migration. Although the estimated coefficients are
restricted in the same way as in Equation (1), the additional
interaction terms in Equation (2) alter the interpretation of the
estimated coefficient on the immigrant dummy variable ([I.sub.hc]) so
that it cannot be sensibly interpreted on its own and therefore will not
be directly comparable to that estimated in Equation (1). (9) Still,
this specification has the advantage of allowing us to assess the extent
to which the pattern of wealth accumulation across income and education
categories might differ between immigrant and native-born couples.
Estimation results at the median (q = 0.5) are presented in Table
4. The results of Model 1 indicate that--controlling for
households' characteristics--the median net worth of immigrant
households is significantly lower than the median net worth of
native-born households in both Australia and Germany, whereas the
conditional nativity wealth gap is positive in the United States. The
relative wealth position of immigrants depends on their family type,
however. The median net worth of immigrant-only households (in which
both partners are foreign born) is substantially lower than the median
net worth of immigrant couples as a whole in Australia, leaving mixed
households with more wealth, regardless of whether the foreign-born
partner is the head or the spouse of his or her respective household.
Interestingly, while mixed couples in which the head is foreign born
hold more wealth than other immigrant households in Germany, mixed
households with native-born heads are wealthier than other immigrant
households in the United States. The results also suggest that more
established immigrants to Australia and the United States are wealthier
than recent immigrants. In contrast, immigrants' relative wealth is
unrelated to their immigration cohort in Germany. Finally, there are
significant differences in the wealth disadvantage faced by immigrants
arriving from different source countries or at different ages.
Estimation of Equation (2) reveals that the determinants of wealth
differ for native- and foreign-born couples in each country. For
example, immigrants to Australia accumulate more wealth for each dollar
of household net income than do native-born Australians, though
immigrants to the United States accumulate less. On the other hand, the
wealth return to education is lower for immigrants in both the United
States and Australia. In Germany, neither the education nor the income
interaction is significant. Together these results highlight the large
differences in the relative wealth position of different groups of
immigrants in different receiving countries.
IV. DECOMPOSITION OF THE WEALTH GAP
A. Decomposition Method
We investigate the source of the nativity wealth gap using the
semiparametric decomposition method proposed by DiNardo et al. (1996).
This approach allows us to assess the relative impact of various sets of
explanatory factors on differences in the wealth distribution between
native and immigrant households (Cobb-Clark and Hildebrand, 2006c). In
particular, we partition the determinants of wealth into three main
factors: income (y); educational attainment (e); and household
demographic composition (z). We can then write the wealth distribution
of group j as follows:
(2) [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]
where I is our indicator of immigrant status and j = (0, 1).
Equation (2) comprises four conditional densities: the conditional
wealth distribution f given the full set of wealth determinants and
immigrant status I, the conditional income distribution ([f.sub.y|e,z])
given education, household demographic composition, and immigrant
status, the conditional education distribution [f.sub.e|z] given
demographics and immigrant status and finally the distribution [f.sub.z]
of demographic characteristics conditional on immigration status.
Following DiNardo et al. (1996), the wealth gap between immigrant
and native households can then be decomposed into three separate
components explained by disparities in conditional income distributions,
educational backgrounds, and demographic characteristics, respectively,
and a fourth "unexplained" component arising from differences
in the conditional (on y, e, and z) wealth distributions of immigrants
and natives. This decomposition is achieved by constructing a series of
counterfactual wealth distributions in which the conditional income,
education, and demographic distributions among native-born households (j
= 0) are selectively replaced with the distributions faced by their
foreign-born counterparts (j = 1). (l0) These counterfactual
distributions are then used to isolate the effect of various wealth
determinants on the nativity wealth gap. (11)
B. Decomposition Results
To understand the source of the disparity in the wealth levels of
natives and immigrants, we decompose the nativity wealth gap into three
separate vectors of wealth determinants: (1) a quartic in income; (2)
educational attainment, which includes years of education for the
household head and the spouse; and (3) household composition, which
includes an indicator variable for children less than age 18 yr in the
household, as well as a cubic in the age of both partners. (12)
Moreover, since the relative wealth position of mixed households differs
substantially across countries (Table 4), we present separate
decomposition results of two samples with and without mixed households
for the three countries. The results of the decomposition analysis
including mixed households are reported in Table 5. Table 6 contains the
corresponding estimates of a sample without mixed households.
One advantage of the DiNardo et al. (1996) approach is that by
estimating the entire counterfactual wealth distribution it is possible
to decompose differences in summary statistics for these wealth
distributions. In what follows, we consider two alternative statistics
that are useful in describing the differences in the wealth holdings of
natives and immigrants: first, the nativity wealth gap at various
percentiles of the distribution and second, the disparity in wealth
dispersion as reflected in the gap between the 75th-25th percentiles of
the wealth distribution. Our results are obtained by calculating each of
the relevant counterfactuals and then taking the simple average of these
statistics over all of the possible decomposition sequences.
Bootstrapping methods using a normal approximation with 1,000
replications are used to calculate standard errors.
We find that native-born households are wealthier than foreign-born
households (see the second column of Table 5) throughout the entire
wealth distribution. At the same time, the magnitude of the nativity
wealth gap differs substantially across countries. Immigrants to
Australia have approximately $9,000 less wealth at the median than
native-born Australians--a gap that is not significantly different from
0. In the United States, the median nativity wealth gap is almost five
times higher than that in Australia (approximately $44,000), while in
Germany the median wealth gap exceeds $158,000. In contrast to
Australia, where the largest wealth gaps appear to be at the tails of
the wealth distribution, the nativity wealth gap is increasing with
higher wealth levels in the United States and Germany reaching more than
$72,000 in the United States and as much as $194,000 in Germany at the
75th percentile. These results point to substantial disparity in the
economic wellbeing of immigrant and native families. Moreover, this
disparity varies dramatically across receiving countries and is
consistent with crossnational differences in the policies used to select
immigrants. In particular, German selection policy has resulted in an
immigrant population that is largely unskilled and, at least notionally,
temporary, while immigrants to Australia are both skilled and permanent.
Given this, it is not surprising that the nativity wealth gap is much
larger in Germany than in Australia.
Interestingly, income differentials explain relatively little of
this nativity wealth gap in the United States and Germany. In Germany,
the income component of the decomposition is small in magnitude,
explaining between 10% of the raw wealth gap at the 75th percentile and
the median and 28% at the 25th percentile. The income component is even
smaller in the United States. Income differentials appear to account for
a much larger share of the wealth gap between immigrant and native
households in Australia. For example, at the median, almost $7,000 of
the overall $9,000 nativity wealth gap in Australia (79%) stems from the
fact that foreign- and Australian-born households have different
incomes. However, it is important to note that the wealth gap itself is
much smaller in Australia.
Overall, these results suggest that--conditional on the
household's educational attainment and demographic
characteristics--income disparities do not themselves lead to a
substantial higher nativity wealth gap. This result is particularly
surprising in light of the empirical evidence demonstrating the large
nativity earnings gap and the often slow earnings assimilation of
immigrants in these countries. At the same time, economic theory
suggests that it is permanent rather than current income that is most
closely related to consumption, savings decisions, and ultimately wealth
accumulation. Our data unfortunately do not provide us with a sensible
permanent income measure and our results are consistent with the
theoretically weaker relationship between current income and wealth.
(13) Any effects of our inability to control for disparities in
permanent income will be captured in the unexplained component of the
nativity wealth gap.
In all three countries, differences in educational attainment
account for as much or more of the nativity wealth as the conditional
income functions. This is particularly important as educational
qualifications are often one of the key factors underlying a
nation's immigration selection process. Again, Australia stands out
relative to the United States and Germany as immigrants have more
education on average than native-born Australians. Our decomposition
analysis indicates that this educational advantage contributes to
reducing the nativity wealth gap in Australia. In other words, if
immigrants to Australia had the same educational qualifications as the
native-born population, we would expect that immigrants' wealth
disadvantage at the median would be approximately $14,000 higher than we
observe it to be. On the other hand, the relative educational
disadvantage of immigrant households in the United States and Germany
accounts for a substantial portion of their wealth disadvantage.
Specifically, at the median, educational differences account for about
13% of the nativity wealth gap in Germany. In the United States,
disparities in conditional education functions explain an increasingly
large proportion of the nativity wealth gap as one moves up the wealth
distribution with more than half of the gap attributable to education
differences at the 75th percentile. Thus, selection policies that
ultimately shape the nature of the immigration flow are likely to have a
large effect on the relative wealth holdings of the immigrant
population.
A household's demographic characteristics-in particular
partners' ages and numbers of children--also explain a large
portion of the nativity wealth gap in Australia and the United States.
In Australia, immigrant households are older and less likely to have
children than native-born households. Thus, their relative demographic
characteristics, like their relative educational advantage, contribute
to reducing the overall nativity wealth gap by approximately $17,000 at
the median. In the United States, however, 32% of the median wealth gap
is explained by households' demographic characteristics. In
Germany, the estimated contributions of the demographic factor to the
overall wealth gap is insignificant over the entire distribution
explaining only between 3 and 13% of the observed wealth gap. To the
extent that this simply reflects differences in households'
position in the life cycle, the gap in wealth resulting from demographic
composition may not be particularly worrying. At the same time,
immigrants' fertility patterns can differ widely across regions of
origin suggesting that the national-origin mix of the immigrant
population is likely to be related to the magnitude of the nativity
wealth gap.
We assess the robustness of our findings by replicating the
analysis dropping mixed immigrant households (Table 6). We find that the
unconditional wealth gap is larger in all three countries indicating
that the relative wealth position of immigrant-only households is more
problematic than that of mixed immigrant households. At the median, the
wealth gap between immigrant-only couples reaches $24,000 in Australia,
$165,000 in Germany, and $54,000 in the United States. As before, the
income component explains relatively little in the United States and
Germany. Educational attainment again appears to be an important factor
in explaining wealth differentials in all three countries, whereas
households' demographic characteristics contribute to explaining
the nativity wealth gap in Australia and the United States.
Overall, these results indicate that wealth differentials are
largely the result of disparity in the education level and demographic
composition of the native and immigrant populations, while income
differentials play a role that is relatively minor. Interestingly,
Australia differs from the United States and Germany in that disparity
in educational attainment and demographic characteristics serve to
narrow rather than widen the nativity wealth gap. If immigrants to
Australia accumulated wealth in the same way as the native-born
population, we would expect that they would have a median wealth
advantage of approximately $32,000 (Table 5). Instead, we observe a
median wealth gap of more than $8,600 implying that immigrant couples in
Australia accumulate substantially less wealth given their
characteristics than do similar native-born couples. Thus, the
relatively small wealth gap in Australia obscures the fact that
immigrant households, especially mixed households, are relatively
advantaged in their characteristics. Unfortunately, they do not
translate their educational and demographic advantage into higher wealth
levels. In Germany, as much as 74% of the observed median nativity
wealth gap is unexplained by immigrants' income, education, and
demographic characteristics, while in the United States at most 28% of
the gap cannot be explained by the fact that immigrants to the United
States have lower income, less education, and are on average younger
with larger families.
V. CONCLUSIONS
Given the increasing numbers of individuals worldwide living
outside their country of birth, it is imperative that researchers move
beyond a narrow focus on the economic assimilation of immigrant workers
to also consider the economic well-being of immigrant families and the
economic integration of immigrant communities more broadly. Many
traditional immigrant-receiving countries are facing aging immigrant
populations as large numbers of immigrants reach retirement age. The
extent to which these immigrants have accumulated sufficient wealth to
provide for themselves in old age is, however, largely unknown. This is
problematic as there are many reasons to believe that nativity status
may affect wealth holdings and asset portfolios. This paper uses recent
data collections that link wealth holdings to migration histories to
analyze the relative wealth position of immigrant and native households
in Australia, Germany, and the United States.
Our results indicate that native-born households are wealthier than
immigrant households across the entire wealth distribution. At the
median, immigrant households in Australia hold approximately $9,000 less
wealth than native households. This nativity wealth gap is substantially
higher in both Germany and the United States reaching approximately
$44,000 in the United States and more than $158,000 in Germany at the
median. Using the semiparametric decomposition method proposed by
DiNardo et al. (1996), we demonstrate that income disparities between
native- and foreign-born households are not the primary explanation of
these wealth gaps. Rather, wealth differentials are largely the result
of differences in the educational attainment and household composition
of native and immigrant households. Once these differences are taken
into account, the wealth gap becomes considerably smaller in the United
States. In contrast, the relatively small wealth gap between
Australian-born and foreign-born households becomes larger once
differentials in household characteristics are considered, indicating
that immigrants to Australia do not translate their relative educational
and demographic advantage into a wealth advantage. In particular, we
would expect immigrant households to have a wealth advantage if they
accumulated wealth in the same way as the Australian born. In Germany,
most of the wealth gap between immigrant and native households cannot be
explained by the fact that immigrant households have lower income, less
education, and are on average younger with larger families.
On balance, the substantial cross-national disparity in the
economic well-being of immigrant and native families is largely
consistent with the selection policies used to shape the skills and
settlement intentions of the immigration population. These policies
ensure, for example, that immigrants to Australia are generally both
permanent and relatively skilled, while immigrants to Germany are mainly
unskilled and notionally temporary. It is not surprising then that we
observe the nativity wealth gap to be larger in Germany than in
Australia. These wealth differentials are most likely reinforced by
labor market differences that lead immigrants to earn higher relative
wages at entry in Australia and enjoy faster earnings adjustment in the
United States, but remain at an earnings disadvantage in Germany. At the
same time, given the large empirical literature documenting these
crossnational differences in the relative labor market outcomes of
immigrants, we might have expected a larger role for income disparities
in producing these nativity wealth gaps.
A number of issues remain important topics for future research.
Unfortunately, given the nature of our data, we have been unable to
directly assess how cross-national differences in welfare and pension
systems affect the nativity wealth gap. Moreover, we can only speculate
about how institutional barriers associated with ethnicity, nativity,
legal status, or language skills might limit immigrants' ability to
accumulate financial or housing wealth. Making progress in these areas
is likely to be largely dependent on the development of data sets that
allow detailed comparisons to be made across a number of institutional
settings. At the same time, a deeper understanding of the role of
immigration policy in shaping the nativity wealth gap is likely to come
from a comparison of the wealth levels and asset portfolios of
immigrants selected on the basis of productive skills or family
relationships rather than out of humanitarian concerns. Such a
comparison is likely to be particularly important in the case of
Australia as our results suggest that the nativity wealth gap is largely
driven by behavioral differences in the accumulation of wealth.
Expanding our focus on first-generation immigrants to also consider the
wealth holdings of second-generation immigrants is an important first
step in understanding the extent to which the nativity wealth gaps we
observe are driven by the relatively higher investments that immigrants
make in the human capital of their children. Finally, future research
should explore the extent to which nativity wealth gaps arise out of
differentials in the consumption and saving patterns or bequest behavior
of native-born and immigrant households.
ABBREVIATIONS
GSOEP: German Socioeconomic Panel
HILDA: Household, Income and Labour Dynamics in Australia
PSID: Panel Study of Income Dynamics
RBA: Reserve Bank of Australia (RBA)
SCF: Survey of Consumer Finance
SIPP: Survey of Income and Program Participation
APPENDIX
TABLE A1
Components of Net Worth
HILDA
Total interest earning assets in banks
+ Total interest earning assets in other institutions
+ Total stocks and mutual funds
+ Total other investments (a)
- Total value of unsecured debt
= A. Net financial wealth
Business equity
= B. Business equity
Housing and real estate equity
= C. Housing equity
Total value of vehicles (b)
= D. Vehicles equity
SOEP
Value of financial assets
+ Equity in other assets (c)
- Value of financial debts
= A. Net financial wealth
Business equity
= B. Business equity
Housing and real estate equity
(including vehicles)
= C. Housing equity
(see C.)
= D. Vehicles equity
SIPP
Total interest earning assets in banks
+ Total interest earnings in other institutions
+ Equity in stocks and mutual funds
+ Equity in other assets (d)
+ Equity in IRA and KEOGH accounts
- Total value of unsecured debt
= A. Net financial wealth
Business equity
= B. Business equity
Housing and real estate equity
= C. Housing equity
Vehicles equity
= D. Vehicles equity
A+B+C+D
= Net worth
(a) Life insurances + trust funds + collectibles.
(b) Car loan is included in the total value of unsecured debt
(see A) which also includes other loans, higher purchase, and
overdraft.
(c) Life insurances + equity in tangible assets.
(d) Including mortgages held from sale of real estate.
REFERENCES
Altonji, J. G., and U. Doraszelski. "The Role of Permanent
Income and Demographics in Black/White Differences in Wealth."
Journal of Human Resources, 40, 2005, 1-30.
Amuedo-Dorantes, C., and S. Pozo. "Precautionary Savings by
Young Immigrants and Young Natives." Southern Economic Journal, 69,
2002, 48-71.
Barsky, R., J. Bound, K. C. Kerwin, and J. P. Lupton.
"Accounting for the Black-White Wealth Gap: A Nonparametric Approach." Journal of the American Statistical Association, 97,
2002, 663-73.
Bauer, T. K., D.A. Cobb-Clark, V. A. Hildebrand, and M. G. Sinning.
"A Comparative Analysis of the Nativity Wealth Gap." Ruhr
Economic Papers No. 6, 2007, 1-52.
Bauer, T. K., B. Dietz, K. F. Zimmermann, and E. Zwintz.
"European Migration: What Do We Know?" in Migration and
Development, edited by K. F. Zimmermann. Oxford, UK: Oxford University
Press, 2005.
Birrell, R. The Chains that Bind: Family Reunion Migration to
Australia in the 1980s. Canberra, Australia: Australian National
Printing Service, 1990.
Blau, F. D. "The Use of Transfer Payments by Immigrants."
Industrial and Labor Relations Review, 37, 1984, 222-39.
Blau, F. D., and J. W. Graham. "Black-White Differences in
Wealth and Asset Composition." Quarterly Journal of Economics, 105,
1990, 321-39.
Blau, F. D., and L. M. Kahn. "The Gender Earnings Gap:
Learning from International Comparisons." American Economic Review,
82, 1992, 533-38.
Borjas, G. J. "Homeownership in the Immigrant
Population." Journal of Urban Economics, 52, 2002, 448-76.
Borjas, G. J., and L. Hilton. "Immigrants and the Welfare
State: Immigrant Participation in Means-Tested Entitlement Programs." Quarterly Journal of Economics, 11 l, 1996, 575-602.
Borjas, G. J., and S. J. Trejo. "Immigrant Participation in
the Welfare System." Industrial and Labor Relations Review, 44,
1991, 195-211.
Chiswick, B. R., and P. W. Miller. "Immigration Generation and
Income in Australia." Economic Record, 61, 1985, 540-53.
--. "Immigration to Australia During the 1990s: Institutional
and Labor Market Influences," in Public Policy and Immigrant
Settlement, edited by D. A. Cobb-Clark and S. Khoo. Surrey, UK: Edward
Elgar Publishing Ltd, 2006.
Clark, W. A.V., M.C. Deurloo, and F. M. Dieleman. "Entry to
Home-ownership in Germany: Some Comparisons with the United
States." Urban Studies, 34, 1997, 7-19.
Cobb-Clark, D. A., and V. A. Hildebrand. "The Wealth and Asset
Holdings of U.S.-Born and Foreign-Born Households: Evidence from SIPP
Data." Review of Income and Wealth, 52, 2006a, 17-42.
--. "The Portfolio Choices of Hispanic Couples*" Social
Science Quarterly, 87, 2006b, 1344-63.
--. "The Wealth of Mexican Americans*" Journal of Human
Resources, 41, 2006c, 841-68.
Coulson, N. E. "Why are Hispanic- and Asian-American
Homeownership Rates so Low? Immigration and Other Factors." Journal
of Urban Economics, 45, 1999, 209-27.
DiNardo, J., N. M. Fortin, and T. Lemieux. "Labor Market
Institutions and the Distribution of Wages, 19731992: A Semiparametric
Approach." Econometrica, 64, 1996, 1001-44.
Fertig, M., and C. M. Schmidt. "First- and Second-Generation
Migrants in Germany--What Do We Know and What Do People Think?" in
Migration Policy and the Economy: International Experiences, edited by
R. Rotte and P. Stein. Munchen, Germany: Hans-Seidel-Stiftung, 2002,
179-218, Studies and Comments No. 1.
Fix, M., and J. Passel. "The Scope and Impact of Welfare
Reform's Immigrant Provisions." Urban Institute, Discussion
Paper, 2002.
Frick, J. R., M.M. Grabka, and J. Marcus. Editing and Multiple
Imputation of Item-Non-Response in the 2002 Wealth Module of the German
Socio-Economic Panel (SOEP). DIW Data Documentation 18. Berlin: DIW
Berlin, 2007.
Gibson, J., T. Le, and S. Stillman. "What Explains the Wealth
Gap Between Immigrants and the New Zealand Born?" New Zealand
Economic Papers, 41(2), 2007, 21-52.
Gittleman, M., and E. N. Wolff. "Racial Differences in
Patterns of Wealth Accumulation." Journal of Human Resources, 39,
2004, 193-227.
Gong, X., C. Ryan, and S. Whelan. "Senior Australians in the
Workforce." SPEAR Centre Report to the Department of Family,
Communities Services, and Indigenous Affairs, 2006.
Gruber, J., and A. Yelowitz. "Public Health Insurance and
Private Savings." NBER Working Paper No. W6041, 1997.
Haisken-DeNew, J. P. and M. Hahn. "PanelWhiz: A Menu-Driven
Stata/SE Interface for Accessing Panel Data." Mimeo, 2006. Accessed
4 January 2009. http://www.panelwbiz.eu.
Heady, B. "Income and Wealth--Facilitating Multiple Approaches
to Measurement and Permitting Different Levels of Aggregation."
HILDA Project Discussion Paper Series, 3/03, 2003.
Juster, F. T., and K. A. Kuester. "Differences in the
Measurement of Wealth, Wealth Inequality, and Wealth Composition
Obtained from Alternative U.S. Surveys." Review of Income and
Wealth, 37, 1991, 33-62.
Juster, F. T., J. P. Smith, and F. Stafford. "The Measurement
and Structure of Household Wealth." Labour Economics, 6, 1999,
253-75.
Maani, S. A. "Immigrants and the Use Government Transfer
Payments." Australian Economic Review, 26(4), 1993, 65-76.
Marks, G. N., B. Heady, and M. Wooden. "Household Wealth in
Australia: Its Components, Distribution and Correlates." Journal of
Sociology, 41, 2005, 47-68.
Miller, P., and L. Neo. "Labour Market Flexibility and
Immigrant Adjustment.'" Economic Record, 79, 2003, 336-56.
Myers, D., and S. W. Lee. "immigration Cohorts and Residential
Overcrowding in Southern California." Demography, 33, 1996, 51-65.
Nielsen, N. "Social Transfers to Immigrants in Germany and
Denmark," in Migrants, Work and the Welfare State, edited by T.
Tranaes and K. F. Zimmermann. Odense, Denmark: University Press of
Southern Denmark, 2004, 245-84.
OECD. Pensions at a Glance: Public Policies Across OECD Countries.
Paris, France: Organization for Economic Co-operation and Development
(OECD), 2005.
--. International Migration Outlook, Annual Report. Paris, France:
Organization for Economic Co-operation and Development (OECD), SOPEMI,
2006.
Osili, U. O., and A. Paulson. "Prospects for Immigrant-Native
Wealth Assimilation: Evidence from Financial Market Participation."
Federal Reserve Bank of Chicago, Working Paper, 04-18, 2004.
Painter, G., L. Yang, and Z. Yu. "Heterogeneity in Asian
American Homeownership: The Impact of Household Endowments and Immigrant
Status." Urban Studies, 40, 2003, 505-30.
Proxenos, S. "Homeownership Rates: AGlobal Perspective."
Housing Finance International, 17(2), 2002, 3-7.
Riphahn, R. "Immigrant Participation in the German Welfare
Program." FinanzArchiv, 55(2), 1998, 163-85.
Schmidt, C. M., and K. F. Zimmermann. "Migration Pressure in
Germany: Past and Future," in Migration and Development, edited by
K. F. Zimmermann. Berlin, Germany: Springer, 1992, 207-36.
Shamsuddin, A. F. M., and D. J. DeVoretz. "Wealth Accumulation
of Canadian and Foreign-Born Households in Canada." Review of
Income and Wealth, 44, 1998, 515-33.
Sinning, M. "Home-Ownership and Economic Performance of
Immigrants in Germany." RWI Discussion Papers No. 45, 2006.
Social Security Administration. "Survey Estimates of Wealth: A
Comparative Analysis and Review of the Survey of Income and Program
Participation." Social Security Bulletin, 65(1), 2003, 47-68.
Statistisches Bundesamt. Statistisches Jahrbuch 2006. Wiesbaden:
Statistisches Bundesamt (Federal Statistical Office), 2006.
Wolff, E. N. "Recent Trends in the Size Distribution of
Household Wealth.'" Journal of Economic Perspectives, 12(3),
1998, 131-50.
Wooden, M., S. Freidin, and N. Watson. "The Household, Income
and Labour Dynamics in Australia (HILDA) Survey: Wave 1."
Australian Economic Review, 35 2002, 339-48.
(1.) See, for example, Blau (1984), Borjas and Trejo (1991), and
Borjas and Hilton (1996) for the United States, Riphahn (1998) and
Fertig and Schmidt (2002) for Germany, and Maani (1993) for Australia.
(2.) These numbers represent estimated pension entitlements based
on preretirement earnings.
(3.) The data used in this paper were extracted from the GSOEP
Database provided by the DIW Berlin (http://www.diw.de/GSOEP) using the
Add-On package PanelWhiz v1.0 (October 2006) for Stata(R). PanelWhiz was
written by Dr. John P. Haisken-DeNew (john@panelwhiz.eu). The PanelWhiz
generated DO file to retrieve the GSOEP data used here and any Panelwhiz
Plugins are available upon request. Any data or computational errors in
this paper are our own. Haisken-DeNew and Hahn (2006) describe PanelWhiz
in detail.
(4.) All calculations are weighted using the weights available in
the respective data sets.
(5.) Marks et al. (2005) provide a comprehensive and detailed
description of the household wealth module of HILDA. The components of
net worth in the GSOEP are described in Frick et al., (2007). The Social
Security Administration (2003) provides a description of the way wealth
is measured in the SIPP.
(6.) The conversion factors are 1.340 for Australia, 0.959 for
Germany, and 1.000 for the United States. See
http:/www.oecd.org/std/ppp/.
(7.) Quantile regressions of our wealth measure on a constant term
and a dummy variable for immigrants have been used to obtain these
figures.
(8.) We also estimated an alternative specification including a
quadratic in household net income. The quadratic term was insignificant
in all cases and was dropped.
(9.) Specifically, the estimated coefficient on the immigrant dummy in Model 2 captures the wealth gap between native-born couples and
immigrant couples that have no income, no education, no children younger
than 18, and in which the reference person migrated as an adult.
(10.) The net worth distribution of immigrants is considerably
narrower than that of natives in all three countries. Therefore,
reweighting the immigrant wealth distribution would involve
extrapolating the immigrant conditional wealth distribution beyond the
income range actually observed in the data. For that reason, we have
chosen to create our counterfactual distributions by reweighting the
wealth distribution of native households (Barsky et al. 2002).
(11.) The proportion of the wealth gap attributable to each of the
explanatory factors will depend on the sequence (or order) in which we
consider them (DiNardo et al., 1996). As we have no particular
preference for one sequence over another, we will calculate each in turn
and present results based on the simple average across all possible
sequences (Cobb-Clark and Hildebrand, 2006c). Gibson et al. (2007) adopt
a similar approach for New Zealand.
(12.) To compare the decomposition results to the estimates of the
quantile regression, we also estimated a more parsimonious model
including only household net income, education, an indicator variable
for children less than 18, and a quadratic function in age. The
decomposition results from this alternative specification do not differ
substantially from those presented in the paper and are available upon
request.
(13.) More specifically, our analysis controls for current,
after-tax household income because none of our data sets provide a
measure of permanent income. As a robustness check, we proxied permanent
income using a measure of predicted income following Blau and Graham
(1990). We found that using predicted income did not affect the results
of the decomposition analysis substantially. However, using predicted
income makes identifying the effects of education and demographic
characteristics more challenging. Consequently, we report only results
based on current income.
THOMAS K. BAUER, DEBORAH A. COBB-CLARK, VINCENT A. HILDEBRAND and
MATHIAS G. SINNING **
* This paper uses confidentialized unit record file data from the
HILDA survey. The HILDA Project was initiated and is funded by the
Department of Families, Community Services, and Indigenous Affairs
(FaCSIA) and is managed by the Melbourne Institute of Applied Economic
and Social Research (MIAESR). The findings and views reported in this
paper, however, are those of the authors and should not be attributed to
FaCSIA or MIAESR.
** The authors are grateful for support from the Social and
Economic Dimensions of an Aging Population Research Program, a SSHRC Grant (#410-2007-1477), and funding from an Australian Research Council
Discovery Grant (DP0666158). The authors would also like to thank Juan
Bar6n, Markus Grabka, David Mare, the participants of the Immigration
Workshop 2006 at the Australian National University, and two anonymous
referees for very helpful comments.
Bauer: Professor, Rheinisch-Westfalisches Institut fur
Wirtschaftsforschung (RWI) Essen, Hohenzollernstr. 1-3, 45128 Essen,
Germany. Phone 49-201-8149264. Fax 49-201-8149284, E-mail
bauer@rwi-essen.de
Cobb-Clark: Professor, Research School of Social Sciences,
Australian National University, HC Coombs Building, Canberra ACT 0200,
Australia. Phone 61-2-6125-3267, Fax 61-2-6125-0182, E-mail deborah.
cobb-clark@anu.edu.au
Hildebrand: Professor, Glendon College, York University, 2275
Bayview Avenue, Toronto, Ontario, Canada M4N 3M6. Phone 416-736-2100,
Fax 416-487-6852, E-mail vincent @econ.yorku.ca
Sinning: Research Fellow, Research School of Social Sciences,
Australian
National University, HC Coombs Building, Canberra ACT 0200,
Australia. Phone 61-2-6125-2216, Fax 61-2-6125-0182, E-mail mathias.
sinning@anu.edu.au
doi: 10.1111/j.1465-7295.2009.00221.x
TABLE 1
Descriptive Statistics, Australia (a)
Natives
Variable SD N
Net worth
Mean net worth ($) 295,078 396,724 2,235
Median net worth ($) 181,567 128,257 2,235
Mean net worth ($) if >0 306,876 399,323 2,141
% of HH with net worth >0 0.964 0.187 2,235
Mean net worth by age group ($)
25-34 154,352 301,899 522
35-44 224,573 275,803 624
45-54 338,094 360,300 502
55-64 506,342 618,015 326
65-75 339,686 326,420 261
Mean components of net worth ($)
Net financial wealth 57,473 165.596 2,235
Business equity 36,269 186,344 2,235
Housing equity 181,743 235,880 2,235
Vehicles equity 19,592 27,068 2,235
Explanatory variables
Household net income (b)) ($) 48,102 40,894 2,235
Age (years) 46.7 13.2 2,235
Number of kids <18 0.9 1.1 2,235
Years of education 11.9 2.4 2,235
Place of residence (%)
New South Wales 0.306 0.461 2,235
Victoria 0.248 0.432 2,235
Queensland 0.230 0.421 2,235
South Australia 0.079 0.271 2,235
Western Australia 0.087 0.282 2,235
Tasmania 0.032 0.176 2,235
Northern Territories 0.004 0.065 2,235
Australian Capital Territory 0.013 0.114 2,235
Immigration cohort (%)
< 1965
1965-1974
1975-1984
> 1984
Place of origin (%)
English speaking
Europe
Asia
Other
Immigrants
Variable SD N
Net worth
Mean net worth ($) 282,640 383,011 1,271
Median net worth ($) 180,074 130,811 1,271
Mean net worth ($) if >0 295,515 385,696 1,219
% of HH with net worth >0 0.959 0.197 1,271
Mean net worth by age group ($)
25-34 112,602 208,093 197
35-44 231,226 320,537 345
45-54 334,413 441,129 334
55-64 376,203 404,528 237
65-75 338,712 413,850 158
Mean components of net worth ($)
Net financial wealth 49,598 157,331 1,271
Business equity 24,552 133,759 1,271
Housing equity 192,703 243,617 1,271
Vehicles equity 15,786 26,773 1,271
Explanatory variables
Household net income (b)) ($) 47,563 45,468 1,271
Age (years) 48.4 12.5 1,271
Number of kids <18 0.8 1.0 1,271
Years of education 12.1 2.6 1,271
Place of residence (%)
New South Wales 0.327 0.469 1,271
Victoria 0.264 0.441 1,271
Queensland 0.152 0.359 1,271
South Australia 0.082 0.274 1,271
Western Australia 0.125 0.331 1,271
Tasmania 0.016 0.126 1,271
Northern Territories 0.010 0.099 1,271
Australian Capital Territory 0.023 0.152 1,271
Immigration cohort (%)
< 1965 0.265 0.442 1,271
1965-1974 0.238 0.426 1,271
1975-1984 0.148 0.356 1,271
> 1984 0.348 0.476 1,271
Place of origin (%)
English speaking 0.471 0.499 1,271
Europe 0.241 0.428 1,271
Asia 0.192 0.394 1,271
Other 0.096 0.295 1,271
(a) Amounts in $US purchasing power parities.
(b) Annual income reported.
TABLE 2
Descriptive Statistics, Germany (a)
Natives
Variable SD N
Net worth
Mean net worth ($) 231,368 290,112 2,344
Median net worth ($) 175,501 139,839 2,344
Mean net worth ($) if >0 273,701 297,906 1,894
% of HH with net worth >0 0.790 0.407 2,344
Mean net worth by age group ($)
25-34 83,955 222,685 242
35-44 184,068 226,901 639
45-54 259,899 336.813 540
55-64 304,479 303,229 493
65-75 252,800 288,409 430
Mean components of net worth ($)
Net financial wealth 51.797 115,704 2,344
Business equity 15,208 104,373 2,344
Housing equity 164,362 208,002 2,344
Explanatory variables
Household net income (b) ($) 34,333 16,652 2,344
Age (years) 51.5 12.9 2,344
Number of kids < 18 0.6 0.9 2,344
Years of education 12.3 2.6 2,344
Place of residence (%)
Berlin 0.023 0.150 2,344
Schleswig-Holstein 0.047 0.211 2,344
Hamburg 0.021 0.142 2,344
Lower Saxony 0.140 0.347 2,344
Bremen 0.011 0.104 2,344
North Rhine-Westphalia 0.266 0.442 2,344
Hesse 0.077 0.266 2,344
Rhineland-Palatinate, Saarland 0.085 0.279 2,344
Baden-Wuerttemberg 0.144 0.351 2,344
Bavaria 0.180 0.384 2,344
Immigration cohort (%)
< 1965
1965-1974
1975-1984
> 1984
Region of origin (%)
OECD member country
Central and Eastern Europe
Turkey
Ex-Yugoslavia
Other
Immigrants
Variable SD N
Net worth
Mean net worth ($) 86,744 230,067 691
Median net worth ($) 7,627 10,830 691
Mean net worth ($) if >0 154,291 181,553 339
% of HH with net worth >0 0.460 0.499 691
Mean net worth by age group ($)
25-34 25,713 146,381 114
35-44 86,655 362,941 189
45-54 125,012 190,062 156
55-64 102,036 151,304 146
65-75 76.034 145,698 86
Mean components of net worth ($)
Net financial wealth 24,643 59,353 691
Business equity 8,354 163,192 691
Housing equity 53,746 122,972 691
Explanatory variables
Household net income (b) ($) 27,724 13,332 691
Age (years) 48.2 12.9 691
Number of kids < 18 1.0 1.2 691
Years of education 11.3 2.7 691
Place of residence (%)
Berlin 0.044 0.206 691
Schleswig-Holstein 0.026 0.159 691
Hamburg 0.009 0.096 691
Lower Saxony 0.093 0.291 691
Bremen 0.009 0.096 691
North Rhine-Westphalia 0.293 0.456 691
Hesse 0.094 0.291 691
Rhineland-Palatinate, Saarland 0.047 0.211 691
Baden-Wuerttemberg 0.230 0.421 691
Bavaria 0.155 0.362 691
Immigration cohort (%)
< 1965 0.109 0.312 691
1965-1974 0.282 0.450 691
1975-1984 0.174 0.379 691
> 1984 0.435 0.496 691
Region of origin (%)
OECD member country 0.262 0.440 691
Central and Eastern Europe 0.284 0.451 691
Turkey 0.228 0.420 691
Ex-Yugoslavia 0.105 0.307 691
Other 0.122 0.327 691
(a) Amounts in $US purchasing power parities.
(b) Annual income reported.
TABLE 3
Descriptive Statistics, United States (a)
Natives
Variable SD N
Net worth
Mean net worth ($) 163,479 268,938 10,283
Median net worth ($) 70,946 66,033 10,283
Mean net worth ($) if >0 186,943 276,553 9,103
% of HH with net worth >0 0.887 0.316 10,283
Mean net worth by age group ($)
25-34 48,108 116,109 1,744
35-44 115,871 205,622 2,780
45-54 173,237 263,908 2,573
55-64 250,801 342,642 1,766
65-75 268,793 329,513 1,420
Mean components of net worth ($)
Net financial wealth 55,871 151,847 10,283
Business equity 14,757 106,119 10,283
Housing equity 87,586 132,806 10,283
Vehicles equity 5,265 8,481 10,283
Explanatory variables
Household net income (b) ($) 17,313 13,843 10,283
Age (years) 48.0 13.0 10,283
Number of kids < 18 0.9 1.1 10,283
Years of education 13.6 2.3 10,283
Place of residence (%)
New England 0.048 0.213 10,283
Middle Atlantic 0.133 0.340 10,283
East North Central 0.173 0.378 10,283
West North Central 0.080 0.271 10,283
South Atlantic 0.191 0.393 10,283
East South Central 0.069 0.254 10,283
West South Central 0.114 0.318 10,283
Mountain 0.050 0.218 10,283
Pacific 0.114 0.318 10,283
Immigration cohort (%)
< 1965
1965-1974
1975-1984
> 1984
Place of origin (%)
Europe (c)
Asia
Mexico
C/S America
Other
Immigrants
Variable SD N
Net worth
Mean net worth ($) 111,868 219,330 1,671
Median net worth ($) 27,577 29,945 1,671
Mean net worth ($) if >0 136,797 231,617 1,382
% of HH with net worth >0 0.835 0.371 1,671
Mean net worth by age group ($)
25-34 40,373 120,635 425
35-44 96,897 214,846 507
45-54 129,388 232,662 397
55-64 194,675 262,981 218
65-75 231,446 276,161 124
Mean components of net worth ($)
Net financial wealth 27,508 99,646 1,671
Business equity 9,733 77,928 1,671
Housing equity 70,908 133,033 1,671
Vehicles equity 3,718 7,453 1,671
Explanatory variables
Household net income (b) ($) 14,610 13,171 1,671
Age (years) 44.1 12.0 1,671
Number of kids < 18 1.2 1.3 1,671
Years of education 12.4 3.9 1,671
Place of residence (%)
New England 0.050 0.217 1,671
Middle Atlantic 0.160 0.366 1,671
East North Central 0.088 0.283 1,671
West North Central 0.020 0.139 1,671
South Atlantic 0.168 0.374 1,671
East South Central 0.012 0.107 1,671
West South Central 0.121 0.327 1,671
Mountain 0.062 0.242 1,671
Pacific 0.312 0.464 1,671
Immigration cohort (%)
< 1965 0.117 0.322 1,671
1965-1974 0.146 0.353 1,671
1975-1984 0.249 0.433 1,671
> 1984 0.488 0.500 1,671
Place of origin (%)
Europe (c) 0.219 0.414 1,671
Asia 0.257 0.437 1,671
Mexico 0.274 0.446 1,671
C/S America 0.177 0.382 1,671
Other 0.073 0.261 1,671
(a) Amounts in $US.
(b) Quarterly income reported.
(c) Includes also Canada and Australia.
TABLE 4
Median Quantile Regressions
Australia
Model (1) Model (2)
Age 13248.43 (0.000) 12891.30 (0.000)
Age Squared -67.39 (0.000) -63.66 (0.000)
Household net income 1.89 (0.000) 1.77 (0.000)
Years of education 17077.09 (0.000) 19190.23 (0.000)
Number of kids <18 3773.62 (0.186) 1247.81 (0.698)
Immigrant (including -31833.34 (0.001) 1417.19 (0.965)
mixed households)
Family type:
Immigrant only -86965.01 (0.000) -74349.10 (0.001)
Mixed native-born head 59266.65 (0.000) 49612.77 (0.000)
Mixed foreign-born head 27698.36 (0.034) 24736.33 (0.059)
Interaction terms:
Immigrant x
Year of migration:
<1965 27450.30 (0.005) 25422.91 (0.017)
1965-1974 -3307.11 (0.717) -5441.94 (0.541)
1975-1984 2736.53 (0.812) 2113.47 (0.847)
>1984 -26879.72 (0.003) -22094.45 (0.031)
Age at arrival < 18 1703.06 (0.899)
Household net income 0.55 (0.000)
Years of education -4936.85 (0.058)
Number of kids <18 5518.74 (0.312)
Regions of origin:
English speaking -11031.62 (0.190) -9051.36 (0.267)
Europe 22563.74 (0.032) 15958.98 (0.118)
Asia -11180.45 (0.339) -6151.90 (0.582)
Other -351.66 (0.980) -755.73 (0.957)
OECD
Central and Eastern
Europe
Ex-Yugoslavia
Turkey
Other
Europe
Asia
Mexico
C/S America
Other
Constant -556789.65 (0.000) -566678.60 (0.000)
N 3506 3506
Germany
Model (1) Model (2)
Age 9740.20 (0.000) 9547.83 (0.000)
Age Squared -49.08 (0.040) -48.01 (0.024)
Household net income 4.09 (0.000) 4.33 (0.000)
Years of education 5947.24 (0.000) 5221.81 (0.002)
Number of kids <18 11825.56 (0.010) 2956.56 (0.522)
Immigrant (including -52455.12 (0.000) -74751.96 (0.054)
mixed households)
Family type:
Immigrant only -45803.53 (0.001) -53562.27 (0.000)
Mixed native-born head 3818.24 (0.813) 6062.19 (0.663)
Mixed foreign-born head 41985.28 (0.027) 47500.08 (0.004)
Interaction terms:
Immigrant x
Year of migration:
<1965 -41407.68 (0.056) -35758.32 (0.066)
1965-1974 -4671.82 (0.774) 20211.46 (0.158)
1975-1984 31115.88 (0.067) -1162.33 (0.937)
>1984 14963.63 (0.340) 16709.19 (0.277)
Age at arrival < 18 63902.90 (0.001)
Household net income -0.95 (0.132)
Years of education 1942.15 (0.551)
Number of kids <18 18364.34 (0.026)
Regions of origin:
English speaking
Europe
Asia
Other
OECD 49826.15 (0.012) 47517.87 (0.007)
Central and Eastern -20319.89 (0.209) -1597.10 (0.912)
Europe
Ex-Yugoslavia -15798.63 (0.441) -21956.45 (0.244)
Turkey 4224.55 (0.811) -8164.96 (0.601)
Other -17932.17 (0.382) -15799.36 (0.371)
Europe
Asia
Mexico
C/S America
Other
Constant -408328.90 (0.000) -395139.56 (0.000)
N 3035 3035
United States
Model (1) Model (2)
Age -1128.55 (0.034) -1260.91 (0.011)
Age Squared 52.18 (0.000) 52.96 (0.000)
Household net income 3.47 (0.000) 3.40 (0.000)
Years of education 8160.25 (0.000) 10992.49 (0.000)
Number of kids <18 3972.15 (0.000) 2390.59 (0.008)
Immigrant (including 13449.78 (0.000) 112162.27 (0.000)
mixed households)
Family type:
Immigrant only -4212.21 (0.240) -6542.04 (0.054)
Mixed native-born head 9004.99 (0.027) 12937.45 (0.001)
Mixed foreign-born head -4792.78 (0.312) -6395.40 (0.148)
Interaction terms:
Immigrant x
Year of migration:
<1965 37671.01 (0.000) 35431.17 (0.000)
1965-1974 -11474.28 (0.027) -14741.34 (0.003)
1975-1984 -8954.27 (0.048) -10567.30 (0.014)
>1984 -17242.46 (0.000) -10122.53 (0.000)
Age at arrival < 18 13392.75 (0.040)
Household net income -0.59 (0.003)
Years of education -8213.27 (0.000)
Number of kids <18 5208.64 (0.015)
Regions of origin:
English speaking
Europe
Asia
Other
OECD
Central and Eastern
Europe
Ex-Yugoslavia
Turkey
Other
Europe 11187.34 (0.031) 16886.77 (0.001)
Asia -7824.41 (0.094) 3848.08 (0.390)
Mexico 26635.82 (0.000) 3635.75 (0.458)
C/S America -10583.86 (0.043) -12038.81 (0.014)
Other -19414.88 (0.008) -12331.80 (0.072)
Constant -152335.24 (0.000) -182142.01 (0.000)
N 11954 11954
Note: Amounts in $U.S. Purchasing Power Parities. p-values are given
in parentheses.
TABLE 5
DFL Decomposition--Foreign-Born to Native-Born Households
Raw Gap Income
Australia
25th 18,694.03 [7,232.89] 4,212.19 [1,442.40] (23)
50th 8,638.06 [9,186.04] 6,792.29 [2,292.40] (79)
75th 13,488.81 [19,710.91] 8,832.09 [5,138.49] (65)
P75-P25 -5,205.22 [18,471.24] 4,619.90 [4,464.98]
Germany
25th 29,968.72 [4.286.16] 8,505.98 [2.298.00] (28)
50th 158,870.49 [8,806.65] 15,443.41 [5,129.17] (10)
75th 194,383.94 [23,221.04] 20,051.89 [7,220.72] (10)
P75-P25 164,415.22 [22,841.57] 11,545.92 [6,648.22]
United States
25th 15,063.71 [915.00] 1,668.50 [365.57] (11)
50th 43,658.21 [3,481.53] 1,903.03 [942.52] (4)
75th 72,519.33 [6,833.101 -1,005.85 [2,294.16] (-1)
P75-P25 57,455.62 [6,644.68] -2,674.35 [2,085.90]
Education
Australia
25th -8,295.27 [2,259.05] (-44)
50th -13,778.61 [3,455.44] (-160)
75th -26,347.01 [6,841.01] (-195)
P75-P25 -18,051.74 [5,558.46]
Germany
25th 8,519.22 [2,229.25] (28)
50th 20,313.59 [5,675.89] (13)
75th 24,567.43 [6,535.99] (13)
P75-P25 16,048.21 [5,754.24]
United States
25th 4,937.40 [443.23] (33)
50th 15,533.93 [1,270.00] (36)
75th 39,554.71 [3,026.24] (55)
P75-P25 34,617.31 [2,801.64]
Demography
Australia
25th -13,193.03 [3,112.431 (-71)
50th -16,614.43 [3,953.62] (-192)
75th -25,097.01 [6,709.02] (-186)
P75-P25 -11,903.98 [5,484.60]
Germany
25th 3,912.44 [2,933.16] (13)
50th 11,186.27 [8,357.29] (7)
75th 6,495.37 [7,794.481 (3)
P75-P25 2,582.93 [6,111.83]
United States
25th 5,786.77 [576.41] (38)
50th 13,912.88 [1,475.61] (32)
75th 28,347.50 [2,800.88] (39)
P75-P25 22,560.74 [2,495.42]
Unexplained
Australia
25th 35,970.15 [6,851.34] (192)
50th 32,238.81 [8,675.40] (373)
75th 56,100.75 [19,153.17] (416)
P75-P25 20,130.60 [18,943.12]
Germany
25th 9,031.07 [3,106.69] (30)
50th 111,927.22 [11,821.70] (70)
75th 143,269.24 [21,665.29] (74)
P75-P25 134,238.17 [21,745.02]
United States
25th 2,671.04 [631.30] (18)
50th 12,308.37 [2,929.43] (28)
75th 5,622.96 [5,820.62] (8)
P75-P25 2,951.92 [5,719.05]
Notes: Percentage of total variation explained in parentheses.
Standard errors of explained variation are reported in brackets.
TABLE 6
DFL Decomposition-Foreign-Born to Native-Born Households, Excluding
Mixed Households
Raw Gap Income
Australia
25th 25,485.07 [10,648.40] 9,575.87 [2.307.40] (38)
50th 23,936.57 [12.754.53] 11,947.14 [3,307.21] (50)
75th 42,686.57 [18,785.23] 22,265.80 [6,984.02] (52)
P75-P25 17,201.49 [18.554.62] 12,689.93 [6,123.98]
Germany
25th 29,968.72 [4,321.17] 9,945.32 [2,500.27] (33)
50th 165,079.25 [8,108.23] 18,780.64 [7,744.37] (11)
75th 270,473.83 [18,844.53] 21,347.90 [11,563.63] (8)
P75-P25 240,505.11 [18,432.87] 11,402.57 [11,096.98]
United States
25th 15,813.54 [867.24] 2,513.95 [457.57] (16)
50th 54,148.14 [3,134.151 5,052.84 [1,139.91] (9)
75th 106,758.90 [8.563.86] 5,392.79 [2,793.42] (5)
P75-P25 90,945.36 [8,427.17] 2,878.84 [2.502.46]
Education
Australia
25th -10,286.07 [3,152.49] (-40)
50th -15,677.86 [4,812.43] (-65)
75th -27,510.32 [8,789.15] (-64)
P75-P25 -17,224.25 [6,590.59]
Germany
25th 11,973.27 [2,892.46] (40)
50th 35,846.64 [9,671.87] (22)
75th 40,215.99 [9,525.18] (15)
P75-P25 28,242.72 [8,382.69]
United States
25th 6,199.09 [578.27] (39)
50th 20,067.04 [1,574.68] (37)
75th 52,002.94 [3,771.07] (49)
P75-P25 45,803.85 [3,500.72]
Demography
Australia
25th -20,573.38 [4,701.20] (-81)
50th -25,840.17 [5,305.60] (-108)
75th -40,203.23 [9,229.86] (-94)
P75-P25 -19,629.85 [7,538.26]
Germany
25th 2,977.13 [2,833.99] ()0)
50th 6,705.98 [9,348.61] (4)
75th -391.62 [8,446.70] (-0)
P75-P25 -3,368.75 [6,785.12]
United States
25th 5,352.80 [649.15] (34)
50th 14,216.23 [1,675.41] (26)
75th 28,893.38 [3,199.64] (27)
P75-P25 23,540.58 [2,795.35]
Unexplained
Australia
25th 46,768.66 [11,040.14] (184)
50th 53,507.46 [12,282.42] (224)
75th 88,134.33 [20,757.98] (206)
P75-P25 41,365.67 [21.111.25]
Germany
25th 5,072.99 [2,901.10] (17)
50th 103,745.99 [17.496.33] (63)
75th 209,301.57 [20,305.64] (77)
P75-P25 204,228.57 [19,858.09]
United States
25th 1,747.70 [620.50] (11)
50th 14,812.03 [2,831.15] (27)
75th 20,469.78 [7,617.71] (19)
P75-P25 18,722.09 [7,537.22]
Note: See the notes in Table 5.