P. 1
Gottschalk & Smeeding

Gottschalk & Smeeding

|Views: 40|Likes:
Publicado porVaishali Parekh

More info:

Published by: Vaishali Parekh on Oct 08, 2010
Copyright:Attribution Non-commercial

Availability:

Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less

10/08/2010

pdf

text

original

American Economic Association

Cvoss-NalionaI Conpavisons oJ Eavnincs and Incone InequaIilv
AulIov|s)· Felev OollscIaII and TinolIv M. Sneedinc
Souvce· JouvnaI oJ Econonic Lilevaluve, VoI. 35, No. 2 |Jun., 1997), pp. 633-687
FuIIisIed Iv· American Economic Association
SlaIIe UBL· http://www.jstor.org/stable/2729789
Accessed· 04/10/2010 21·43
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=aea.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
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.
American Economic Association is collaborating with JSTOR to digitize, preserve and extend access to Journal
of Economic Literature.
http://www.jstor.org
Journal of Economic Literature
Vol. XXXV (June 1997), pp. 633-687
Cross-National Comparisons of
Earnings and Income Inequalit
PETER GOTTSCHALK
Boston College
and
TIMOTHY M. SMEEDING
Syracuse University
We are grateful for the assistance and encouragement given by our colleagues Anthony B.
Atkinson and Lee Rainwater as well as the large number of country experts who provided
useful comments on earlier drafts and participants at seminars at Princeton, Rand, Stanford,
and U. C. Berkeley. Support for this project was provided for Smeeding by the Russell Sage
Foundation and by NSF #SBR-9511521. Helpful comments were received from several ref-
erees, Anthony Atkinson, Anders
Bj5rklund,
Gary Burtless, Sheldon Danziger, John Fitzger-
ald, Johan Fritzell, Joop Hartog, Markus Jdntti, Stephen Jenkins, Robert Lerman, John Myles,
Robert Plotnick, James P. Smith, and Peter Saunders. Esther Gray, Katherin Ross, and Ann
Wicks provided excellent assistance with the manuscript. The authors assume responsibility
for all errors of commission.
I. Introduction
JNTEREST IN THE distribution of earn-
ings and the distribution of household
income was largely a parochial backwater
of economic research in the United
States until the early 1980s. This lack of
interest reflected the view that both the
functional and personal distributions of
income in the United States showed lit-
tle change between the end of the 1940s
and the mid-1970s.1 This led Robert
Lampman (1971, p. 47) to remark that
the stability of the income distributions
was "remarkable in view of the great
changes which have occurred in eco-
nomic structure and in income and
wealth levels." He further noted that
predictions of increasing concentration
of wealth "have been proved completely
wrong by the American experience."
Taking a somewhat more laid-back per-
spective, Henry Aaron (1978, p. 17)
noted that tracking changes in the distri-
bution of income in the United States
"was like watching the grass grow."
The lack of interest in distributional
issues in the United States in general,
and cross-national comparisons in par-
ticular, changed for several reasons in
the early 1980s. First, the view that the
shape of the income distribution was one
of the great constants of economics came
into question by a series of studies, re-
viewed in Frank Levy and Richard Mur-
nane (1992), that showed rising inequal-
ity of labor market income in the United
States and a smaller set of studies that
showed that these changes in the earn-
ings distribution were being translated
into greater inequality in the distribution
of total family income.
1
In contrast, British researchers such as Atkin-
son (1970) and Dutch researchers such as Jan Pen
(1971) and their predecessors made key contribu-
tions to both inequality theory and measurement
during the 1970s.
633
634 Journal of Economic Literature, Vol. XXXV (June 1997)
Second, it became considerably easier
to perform cross-national comparisons of
income distributions. Before 1980 such
comparisons were of a rough and ready
nature and did not withstand close scru-
tiny.2 Yet international comparisons of
income distribution can provide impor-
tant benchmarks of how one nation dif-
fers from or is similar to other nations.
In so doing, they can provide useful in-
formation, just as do cross-national com-
parisons of rates of economic growth,
savings, inflation, and unemployment.
Fortunately, cross-sectional micro data
became publicly available for a variety of
rich OECD countries.3
This opened several avenues for re-
search, primarily by allowing greater
richness in cross-national comparisons.
While cross-national comparisons of av-
erage income had been widely used to
measure differences in standard of living
across countries, these comparisons had
focused on the typical or average family.
Data about the relative standard of living
of persons elsewhere in the distribution
could now provide a more complete pic-
ture of cross-national differences. These
new data also contributed to the litera-
ture on trends in inequality. Researchers
were not only able to address the factual
question of whether inequality grew in
other countries but also to get further
hints as to possible causes. For example,
if countries with binding trade barriers
experienced smaller increases in inequal-
ity then this would be consistent with
the view that increased foreign competi-
tion was at least partially responsible for
the increase in inequality. Likewise,
cross-country comparisons of changes in
industrial structure or unionization
would at least provide some stylized facts
that might inform the debate on the
causes of the increase in inequality.
The third factor contributing to the in-
creased interest in distribution issues
stems not from the positive interest in
understanding the causes of change but
the normative issues coming out of the
debate in the public policy arena over
the "fairness issue." The distributive ef-
fects of changes in government policies,
which had always been a key policy issue
in European, Nordic, and British Com-
monwealth countries, have become an
increasingly important policy issue in the
United States.
In this article, we further develop
Levy and Murnane's (1992) review of
changes in earnings inequality in the
United States in two directions. First, we
expand our review to other major in-
dustrialized countries, largely OECD na-
tions. Second, we broaden the focus
from earnings to household income. As
we will show, the increases in the disper-
sion of both individual earnings and total
household income in the United States
were larger than in almost all other
countries. However, the United States
was not the only country to experience
an increase in inequality during the
1980s and early 1990s. While most coun-
tries experienced at least modest in-
creases in earnings and market income
(income before direct taxes and public
income transfers) inequality, these were
largely offset by changes in other sources
of income, producing a more modest in-
crease in the inequality of disposable in-
comes in most nations.
We review not only what we know
about what has happened to earnings but
also why experiences differed across
countries. While causal explanations are
2
See, for example, Michael Sawyer (1976) and
the strong negative response to Sawyer by Jean
Begue (1976).
3 Database projects such as the Luxembourg In-
come Study LIS), described in Smeeding, Mi-
chael O'Higgins, and Rainwater (1990), and re-
lated efforts to make longitudinal household panel
data comparable, for example, the United States-
German comparative pane project described in
Gert Wagner, Richard Burkhauser, and Friederike
Behringer (1993), have facilitated cross-national
comparisons of inequality.
Gottschalk and Smeeding: Cross-National Income Inequality 635
never easy to pin down, the issues are
fairly well defined. Labor economics
provides a rich theoretical framework
that has been applied with some success,
at least in a partial equilibrium setting,
to explain changes in the structure of
wages.
The expansion from individual earn-
ings to household disposable income,
however, raises a whole host of analytical
as well as measurement issues.4 Eco-
nomic and demographic decisions within
the household are endogenous and so
complex that empirical research is far
from being able to sort out the linkages.
The problem of endogeneity would arise
even in the context of a single country.
However, the problem of endogeneity is
further aggravated by the expansion to
the international context. Social and po-
litical institutions that may affect how
other household members and govern-
ment taxes and transfers respond to
changes in market conditions differ con-
siderably across countries. As a result,
the responses of household disposable
income to changes in the structure of
wages, interest rates, or other prices will
vary across countries. Given the lack of
any unifying theoretical structure to ana-
lyze household income, we will largely
limit ourselves to presenting the basic
facts that any theory would have to ex-
plain. However, there is a strong need
for a better theoretical structure in
which to understand these outcomes.
We begin our review by laying out a
set of stylized facts both for the United
States and for other nations. We present
summaries for both the level and trend
in earnings and income inequality. In
Section II we briefly turn to conceptual
and comparability issues. This is fol-
lowed in Sections III and IV by an over-
view of what we know about changes in
earnings inequality in a variety of coun-
tries and the causes for these changes.
Section V turns to family income in-
equality to answer the same questions.
While we cover a wide range of topics,
not everything under the rubric of
changes in inequality is addressed. We
are concerned with highly developed
countries, almost exclusively the OECD
nations, and do not address inequality in
developing nations or in the transition
countries of Eastern Europe and the for-
mer U.S.S.R.5 Issues related to wealth
inequality, consumption and expenditure
inequality, the tradeoff between equality
and efficiency, social choice theory, and
the theoretical and empirical literature
on inequality measurement are largely
excluded. Other pertinent issues, such as
the burgeoning literature on growth and
inequality, the dynamics of income, and
intergenerational mobility are also not
covered. Finally, due to constraints of
space and time, the literature on cross-
national comparisons of low incomes or
poverty is also excluded.6
A. Stylized Facts
The growing interest in national and
cross-national differences in earnings
and income inequality has produced a
wide range of recent comparative studies
of the level and trend in inequality along
with dozens of studies and reports on in-
dividual countries. Our summary of the
stylized facts emerging from these stud-
ies is as follows:
4
Disposable household income includes all
sources of income received by all household mem-
bers, including income transfers from govern-
ments and other parties, net of income and payroll
taxes.
5
See Klaus Deinenger and Kenneth Squire
(1995) on income inequality in developing and de-
veloped countries and Atkinson and John Mickle-
wright (1992) on inequality in Eastern and Central
Europe. See Smeeding and Gottschalk (1996) for
comparisons which include the OECD nations, se-
lected Eastern European nations, and Taiwan.
6
Interested readers should consult Michael
Forster (1993), McKinley Blackburn (1994), At-
kinson, Rainwater, and Smeeding (1995a), Rain-
water and Smeeding (1995).
636 Journal of Economic Literature, Vol. XXXV (June 1997)
1. Earnings.
Levels
(1) At any given time there are wide
differences across modern coun-
tries in the level of earnings in-
equality for both men and women.
(2) Nations with centralized wage bar-
gaining (e.g., Sweden, Germany)
have greater equality than nations
with less centralized bargaining
(e.g., the United States and Can-
ada).
Trends
(1) Almost all industrial economies ex-
perienced some increase in wage
inequality among prime aged males
during the 1980s (Germany and It-
aly are the exceptions).
(2) But large differences in trends also
exist across countries, with earn-
ings inequality increasing most in
the United States and the United
Kingdom and least in Nordic coun-
tries.
(3) The increasing demand for more
skilled workers, coupled with dif-
ferences across countries in the
growth in supply of skilled work-
ers, explains a large part of differ-
ences in trends in returns to educa-
tion and experience.
(4) Institutional constraints on wages
also seem to matter. The rise in the
relative unemployment rates of the
least skilled in some, but not all,
countries with centralized wage
setting institutions suggests that
constraints were at least partially
responsible for limiting the rise in
inequality.
B. Disposable Income
Levels
(1) There is substantial diversity in
the inequality of household dispos-
able income across major OECD
nations, with the greatest inequal-
ity in the United States and the
least inequality in Nordic and
Northern European countries.
(2) Post-tax and transfer disposable
money income is more equally dis-
tributed than market income in all
OECD nations, and there is a no-
ticeable correlation between public
cash income transfer expenditures
and disposable income inequality.
(3) Even after adjusting for real in-
come differences across countries
(using purchasing power parity),
low income United States citizens
have real living standards below
those found in most other rich
OECD countries.
Trends
(1) Increases in household income in-
equality were more muted than
were changes in earnings inequal-
ity in most nations. Still, increased
earnings inequality among men
was probably the most important
factor in explaining rising income
inequality.
(2) Income inequality increased in
most, but not all, OECD nations
during the 1980s and early 1990s.
Trends in inequality were not
closely associated with levels of in-
equality. Some nations with low
levels of inequality experienced
some of the largest increases.
(3) Reductions in social welfare
spending for the non-aged and re-
gressive changes in the structure of
income taxes for some countries
during the 1980s account for only a
small part of the trend in post-tax
and transfer inequality in most na-
tions.
(4) Married women's labor force par-
ticipation rates, hours, and wages
increased substantially in almost all
countries during the 1980s. The
Gottschalk and Smeeding: Cross-National Income Inequality 637
positive correlation between hus-
bands' and wives' earnings also in-
creased moderately, thus tending
to increase income inequality.
II. Comparability and Data Quality
In this section we address the mea-
surement problems raised when making
comparisons of earnings and income dis-
tributions across countries. The main
source of United States income data is
the March Supplement to the Current
Population Survey (CPS), in effect an in-
come supplement to a labor force sur-
vey. Other countries similar to the
United States have annual or periodic
surveys of consumer finances or income
(Canada and Australia). Other nations
use specific income surveys or have ex-
tensive surveys of expenditures with de-
tailed income components sections (e.g.,
The Netherlands, United Kingdom, Is-
rael). In a few nations (Sweden, Finland,
Norway) survey respondents give the sta-
tistical office permission to go directly to
government records to measure incomes
and report only demographic informa-
tion to the survey takers. Thus, the type
and purpose of surveys used for interna-
tional comparisons vary widely by coun-
try.7
A. Income Definitions
Ideally income would be measured on
a post-tax and transfer basis consistent
with the Haig-Simons income concept of
real consumption plus (or minus) change
in net worth.8 Income would include
both cash and noncash components,
would be adjusted for economies of scale
in consumption using an appropriate
equivalence scale, and would cover the
period over which families can smooth
consumption by lending or borrowing.
For families that are not credit con-
strained this might require measures of
lifetime post-tax and transfer income ad-
justed for family size. At the other ex-
treme, the relevant measure of income
might be a few pay periods for families
who do not have sufficient assets to
smooth consumption and cannot borrow
against future income. Unfortunately, al-
most all of the existing data sets, includ-
ing the CPS in the United States, mea-
sure income on a yearly basis. This is
certainly too long an accounting period
for families that are severely credit con-
strained, and too short for families that
can smooth consumption over multiple
years. While the problem raises impor-
tant conceptual issues, the existing evi-
dence shows that rankings of countries
with respect to income inequality are ro-
bust with respect to changes in the ac-
counting period (Rolf Aaberge et al.
1995; Richard Burkhauser and John Pou-
pore 1997).
Surveys may also differ in the income
sources they include as earnings. For ex-
ample, unemployment insurance and/or
sick pay are included as a transfer in
most countries but are included in earn-
ings in a few (e.g., Sweden, France). Al-
most all nations include vacation pay
("13th month earnings") and salary bo-
nuses in their measures of earned in-
come. Self-employment income, which
differs nation by nation in quality of data
reported and in its economic impor-
tance, may also be included in earnings.
There is even greater diversity in the
decision of what to include under total
household income. Cross-national com-
parisons of income inequality have fo-
cused primarily on the distribution of
disposable money income after direct
taxes (income and employee payroll) and
7
For discussion of the problems of comparabil-
ity across countries and for additional information
on survey differences, particularly for those sur-
veys from the Luxembourg Income Study, see At-
kinson, Rainwater, and Smeeding (1995a, espe-
cially Chs. 2 and 3 and Appendices 1
through
5
8
This broad definition of income is an attempt
to get closer to the distribution of lifetime utility.
638 Journal
of
Economic
Literature,
Vol. XXXV
(June
1997)
after transfer payments.9 While this defi-
nition of post-tax and transfer disposable
income is broad, it falls considerably
short of the Haig-Simons comprehensive
definition, typically excluding much of
capital gains, imputed rents, home pro-
duction, and income in kind. In general,
no account is taken of leisure, indirect
taxes, or of the benefits from public
spending other than cash transfers.
Further comparability issues are raised
by definitions of income sharing units
and the unit of analysis. Survey-based re-
search on income inequality sometimes
focuses attention on the household as
the unit of income sharing and as the
unit of analysis; other times the unit of
analysis is the individuals within the
household. And definitions of income
sharing units themselves may differ
across nations.10
B. Adjustment for Household Size and
Composition: Equivalence Scales
Most studies of income inequality ad-
just income to take account of differ-
ences in material needs for families of
different sizes. Equivalence scales are
designed to accomplish this adjustment
by taking into account those household
characteristics deemed to affect econo-
mies of scale and economies of scope as
reflected by differences in household
size and composition. Total household
income is divided by the number of
equivalent adults in order to arrive at a
measure of household "equivalent" in-
come.
Brigitte Buhmann et al. (1988) first
proposed a succinct parametric approxi-
mation to equivalence scales which sum-
marized the wide range of scales in use:
Adjusted Income =
Disposable Income/SizeE
The equivalence elasticity, E, varies
between 0 and 1; the larger is E, the
smaller are the economies of scale as-
sumed by the equivalence scale. The
various studies reviewed in this survey
make use of equivalence scales ranging
from E = 0 (no adjustment) to E = 1
(per capita income which ignores econo-
mies of scale). Between these extremes,
the range of possible values is rather
evenly covered.
These adjustments for family size can
have a large effect on the level of mea-
sured inequality within and across na-
tions.11 However, using different equiva-
lence scales preserves the general rank
order of countries, albeit at different lev-
els of inequality. Inequality rankings at a
point in time are fairly robust to choice
of equivalence scales (Atkinson, Rain-
water, and Smeeding 1995a, Figure 4.1).
Due to lack of a long time series of com-
9
Direct taxes are most often estimated from tax
imputation models rather than official tax records.
For example the after-tax data for Australia, Ger-
many, New Zealand, and the United States in the
Luxembourg Income Study are obtained using a
tax imputation model at the household level to es-
timate direct taxes. Italy, Belgium, and Luxem-
bourg surveys report only after-tax income; Swe-
den, Finland, and Norway use official records of
taxes paid.
10
While most nations aggregate income across
all members of a househol, a ew use a more nar-
row definition, for example: all related persons liv-
ing together or a famify (e.g., Canada); or even
more narrowly related persons according to in-
come tax
regulations
(e.g., Sweden).
11
See Fiona Coulter, Frank Cowell, and
Stephen Jenkins (1992) and Buhmann et al.
(1988). An important and non-obvious lesson from
these papers is that the relationship between in-
equality measures and elasticities is non-mono-
tonic. Most studies of cross-national distribution
make no adjustments for differences in incomes
within households, assuming that income is
equally shared by all members of the household.
Jenkins (1994), however, shows that the estimates
of overall household income inequality derived
from three different methods of estimating within
household inequality are very different from those
derived using the conventional, equal sharing
within the household assumption. The literature
has moved beyond the one parameter equivalence
scale used here to two parameter scales which in-
clude adjustments for types of individuals (e.g., by
age) as well as for family size. See Jenkins and
Cowell (1994).
Gottschalk and Smeeding: Cross-National Income Inequality 639
parative data, the literature cannot de-
termine if choice of equivalence scale af-
fects trends in measured inequality
across countries. However, evidence for
differences in trends within the United
States indicates that choice of equiva-
lence scale may affect the level of mea-
sured inequality but not its trend (Lynn
Karoly and Burtless 1995).
C. Noncash Benefits and Taxes
The disposable money income mea-
sures used in most studies include only
public cash and near cash benefits (food
stamps and other similar benefits de-
nominated in cash). Hence, one might
suspect differences across countries de-
pending on a nation's preferences for
cash versus noncash transfers. A similar
type of difference may occur if countries
rely on employers to provide some types
of benefits (e.g., health insurance for
workers in the United States, and occu-
pational pensions in many nations), while
governments provide others (e.g., health
insurance and more substantial social re-
tirement pensions in most other nations),
or if demographic composition of nations
are very different.12
Including noncash benefits in esti-
mates of the level and trend of income
inequality also requires the valuation of
these benefits. While several national
studies of noncash benefits have assessed
their impact on the income distribution
as measured by the cost of benefits to
the supplier, the literature has made lit-
tle progress in arriving at a true Hicksian
equivalent variation measure of their
cash equivalent value to households
(U.S. Department of Commerce, Bureau
of the Census 1982; Barbara Wolfe and
Robert Moffitt 1991).
In-kind benefits also tend to be a
small share of total social transfers rela-
tive to cash benefits in nations with small
shares of GDP spent on cash benefits.
Because high cash benefit nations tend
to also be high in-kind benefit nations,
the limited evidence indicates that the
exclusion of noncash benefits does not
have a large impact on the income in-
equality rankings of countries.13
Most studies of income distribution
employ either a measure of all sources of
money income prior to the deduction of
all taxes ("gross income") or a measure
that subtracts "direct taxes"-income
and employee payroll taxes-to arrive at
disposable income. In general, studies
do not count personal property or wealth
taxes as direct taxes. Because of differen-
tial reliance on employer and employee
social security contributions across na-
tions, and because of the differential mix
of personal, business, earnings, income,
property, and goods and services (ex-
penditure, VAT) taxes across OECD
nations, the manner in which taxes are
collected may affect the results of cross-
national comparative analyses. In fact,
we know of only one study that has in-
cluded the full burden of direct and in-
direct taxes in cross-national studies of
income distribution.'4
D. Data Quality Comparison
with National Accounts
One common criticism of earnings and
income distribution data derived from
12The mix of cash plus noncash benefits across
OECD nations is, however, more uniform than is
the distribution of cash benefits alone. See
Smeeding et al. (1993) and Peter Whiteford and
Steven Kennedy (1994).
13 Smeeding et al. (1993) find that imputation of
health and education benefits and some housing
benefits had an equalizing impact in all countries,
but did not affect the inequality ranking of coun-
tries. For one estimate of the effect of including
in-kind benefits in income distribution in the
United States, see U.S. Department of Com-
merce, Bureau of the Census (1995a). These esti-
mates indicate that including noncash benefits af-
fects the level but not the trend in inequality since
1979.
14 See Clive Bell and Christoph Rosenberg
(1993). Kenneth Messere (1993) presents aggre-
gate data on the tax mix across countries.
640 Journal of Economic Literature, Vol. XXXV (June 1997)
household surveys is that they are incom-
plete in coverage of income. One way
of determining the size of the under-
reporting is to compare the total income
of different types reported in the house-
hold surveys with external information
drawn from national accounts and coun-
try data registers, which have been ad-
justed to make them comparable to the
microdata sources.
Not all countries have been able to
compare survey data with national
accounts or other external data. Still
the available information indicates that
total income estimates based on the
surveys used for income distribution
studies are about 90 percent of the
comparable national income totals in
six of the eight countries for which
comparison data are available in the Lux-
embourg Income Study (Canada, Fin-
land, Italy, The Netherlands, the United
Kingdom, and the United States). In
two nations (Australia and West Ger-
many) there is an aggregate shortfall
of some 20 percent, but part of the dif-
ference can be explained by the fact
that the totals are not fully compar-
able. Wage and salary income is, how-
ever, generally well reported in all coun-
tries.15
While underreporting may be small
for the most important income sources,
this may be of little comfort for distri-
butional measures. What is relevant is
not only the amount of underreporting
but its distribution. If underreporting
were small but non-random, this would
affect both measures of central tendency
and dispersion.
E. Level versus Trend
Point in time comparisons of the level
of inequality across countries impose
much stronger data requirements than
comparisons of trends in inequality. As
long as differences across countries (in
income measures, importance of income
components, adjustments for income
sharing, quality of income reporting, and
survey data collection practices) are con-
stant across time, these differences will
cancel. As a result, country-specific id-
iosyncrasies would affect levels of in-
equality but not trends in inequality. On
the other hand, if data quality changes
over time, if income components that are
less (or more) well reported increase in
significance over time, or if factors such
as top coding have different impacts over
time, then trends as well as levels will be
affected.
The Luxembourg Income Study (LIS)
data sets were assembled specifically to
overcome many of the problems ad-
dressed in these sections. LIS collects
none of its own data. Rather it takes data
collected mainly by national statistical
agencies and applies consistent measures
and concepts across countries to produce
greater uniformity in cross-national com-
parisons.'6 Access to micro data in LIS
also makes it possible to impose consis-
tency on additional elements such as the
unit of observation, income definition,
and adjustments for differences in family
size. Moreover, it is possible to test sen-
sitivity to alternative choices of units,
definitions, and other measurement is-
sues such 'as top and bottom coding of
income. But while the aim of the LIS
project is to increase the degree of cross-
national comparability, complete cross-
15
See Atkinson, Rainwater, and Smeeding
(1995a, Table 3.2 and Appendix 6). Evidence in
the United States derived from a direct matching
of individual responses to administrative and tax
records (Daniel Radner 1983), indicates that the
problem of property income underreporting is pri-
marily found among upper income households
with heads aged 65 and over, but no evidence on
direct matching is available for other nations.
16
For instance, the earnings and income data
presented in this paper come from the same
source. Other data, such as that found in the In-
ternational Social Survey datafiles, cover earnings
well have but have very limited information on in-
comes.
Gottschalk and Smeeding: Cross-National Income Inequality 641
national uniformity will never be possi-
ble because the country surveys that
form the starting point for LIS vary in
focus and scope, and because certain as-
pects of surveys cannot be adjusted ex
post facto (e.g., a country survey's choice
of a singular unit of income aggrega-
tion).'7
The International Social Survey Pro-
gramme (ISSP) offers an alternative col-
lection of repeated cross-sections on a
number of countries. The major advan-
tage of these data is that they are based
on responses to a uniform set of ques-
tions attached to country-specific social
surveys. For example, the common ISSP
questions are asked to a subset of re-
spondents to the General Social Survey
in the United States and to respondents
to the British Social Attitudes Survey in
the United Kingdom.'8 The advantage of
cross-national uniformity in the ques-
tionnaire has to be weighted against
three disadvantages. First, the sample
size in each year for each country is con-
siderably smaller than in LIS (roughly
1,500 in ISSP versus 5,000-65,000 or
more in the LIS data sets used here).
This sample size drawback can be par-
tially overcome by pooling years, though
this is problematic when income distri-
butions are changing over time. The sec-
ond limitation of ISSP is that the ques-
tionnaire is designed to be answered in
15 minutes. Because the primary focus
of the survey is on social attitudes only
22 questions are asked about economic
and demographic characteristics. Finally,
most countries report income or earn-
ings in income brackets, with the top
bracket being open-ended. This draw-
back is particularly severe when the
brackets are changed, making compari-
sons over time even more difficult, espe-
cially with small samples.
Full comparability of earnings and in-
come distribution data will never be at-
tainable as long as surveys and institu-
tions differ across countries. While these
limitations must be kept in mind, strong
patterns emerge out of these admittedly
noisy data. As we will show in the follow-
ing sections, surveys with very different
focus and structure give broadly similar
patterns. The issue is not the existence
of noise, which surely exists in all data
sets, but the relative size of the signal to
the noise.
III. Earnings Inequality
A vast literature, reviewed in Levy and
Murnane (1992), has documented the
substantial increases in inequality of
wage rates and annual earnings in the
United States during the 1970s and
1980s. At this point there is a wide con-
sensus in the research community that
an important driving force behind the in-
crease in family income inequality in the
United States was the increased disper-
sion of earnings.19
17
While the LIS project has gone to great
lengths to increase data comparability across na-
tions, not all problems can be overcome. For in-
stance, the LIS data cover a limited number of
years and are thus unable to capture the effects of
business cles on income inequality. Also LIS has
no contro over the questions asked in different
surveys. While all income data used are continu-
ous variables, and while the LIS has up to 38 dif-
ferent categories of cash income for each nation,
some items such as self-employment income may
be measured differently in different nations. For
additional discussion of the technical charac-
teristics of the LIS database, see Atkinson, Rain-
water, and Smeeding (1995a).
18
ISSP started with four countries in 1984 (Aus-
tralia, Germany, the United States, and the United
Kingdom). By 1994 the questions were being
asked in over 20 countries.
19 See Danziger and Gottschalk (1995) and Re-
becca Blank (1994) for links between changes in
the distribution of earnings and income. Danziger
and Gottschalk (1995) attribute the majority of the
change in family income inequality to changes in
the distribution of men's earnings. Karoly and
Burtless (1995) study working age families and
find that change in earnings inequality among men
who work accounts for slightly less than half of the
total change. We return to this topic later in the
paper.
642
Journal of
Economic
Literature,
Vol. XXXV (June 1997)
A. Levels of Inequality
Cross-national studies of earnings in-
equality have focused almost exclusively
on trends, not levels. This largely re-
flects the lack of comparable data across
countries.20 We exploit recently available
data in the LIS database to compare
earnings inequality across a variety of
countries during the late 1980s and early
1990s.21 Table 1 presents summary mea-
sures of the earnings distributions in the
nine countries for which the LIS data-
base provides consistent data on annual
before-tax earnings for males and fe-
males aged 25 to 54. Because it is impos-
sible to separate labor market earnings
from returns to capital in households
with self-employment income we also ex-
clude all persons in such households.22
While we focus on the distribution of
positive earnings we also show the pro-
portion of persons with zero earnings in
Column (3) of Table 1. It should be rec-
ognized that differences in the distri-
bution of positive earnings are very likely
to be affected by these differences in the
proportion of persons with zero earn-
ings. However, without knowing the
earnings that zero earners would have
received if they had worked, it is im-
possible to determine the effect on the
unconditional distribution of potential
earnings. At one extreme one might as-
sume that all zero earners came from the
bottom of the distribution of potential
earnings. It is, however, unlikely (espe-
cially among women) that the full differ-
ence in zero earners reflects additional
persons at the bottom of the distri-
bution. Thus, while Table 1 presents es-
timates of the percentile ratios of the
distribution of positive earnings this
should not be confused with the distri-
bution of potential earnings for all persons.
Our summary measures of inequality
are based on earnings at selected percen-
tile points because these are less sensi-
tive to such inter-country differences as
non-uniform top and bottom coding of
earnings, and underreporting of earnings
at either tail of the distribution. Earn-
ings at selected percentile points are
measured as a proportion of earnings at
the median. For example, the PIO value
of 56.8 for males in Australia signifies
that an Australian male at the tenth per-
centile earned a little more than half as
much as the male at the median. We also
show the 90/10 and 80/20 ratios as sum-
mary mneasures of overall inequality. In-
formation is presented for full-year
full-time workers and all workers with
non-zero earnings.
For full-year workers these countries
can be broken down into three broad
groups. The United States and Canada
stand out as the economies with the
most unequal distributions of earnings
for both males and females, measured
either by the 90/10 or the 80/20 ratio.
For males this largely reflects consider-
ably lower earnings at the bottom of the
distribution.23 For females, low earnings
20A few previous studies have used LIS data to
examine wage and salary differences of heads of
households across nations at a point in time (Gor-
don Green, John Coder, and Paul Ryscavage 1992;
Janet Gornick 1994; Gottschalk and Mary Joyce
1996). Francine Blau and Lawrence Kahn (1996),
who use data from the International Social Survey
Programme, find similar patterns.
21
Earnings is used synonymously with wage and
salary income. Income surveys and, hence, the LIS
database do not usually contain separate measures
of hourly wages. All estimates shown in Table 1
refer to annual earnings except for the United
Kingdom where wages and salaries are measured
during the survey week. The years shown are lim-
ited by data availability in LIS. Therefore differ-
ences in cyclical conditions may affect rankings of
countries with small differences in inequality.
22
The cost of excluding the self-employed is
that the distribution of earnings for the selected
sample will be affected by this sample selection if
the distribution of labor market earnings of the
self-employed is different from the distribution
for all other persons.
23
This is consistent with Blau and Kahn (1996),
who use different data.
Gottschalk and Smeeding: Cross-National Income Inequality 643
TABLE 1
EARNINGS DISTRIBUTIONS IN SELECTED OECD COUNTRIES IN THE MID-1980S AND EARLY 1990s:
PERCENTILE OF MEDIAN AND DECILE RATIoSa
Full-Year,
Full-Time
Workersc
All Workersd
Percentage
with
Country Year Zero Eamingsb PlO P9 P90/P10 P80/P20 PlO P90 P90/P10 P80/P20
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
Males
Australia 1989 20.8 56.8 160.6 2.8 1.9 54.0 161.6 3.0 1.9
Canada 1987 13.2 38.0 174.9 4.6 2.3 36.3 176.0 4.7 2.6
Finland 1987 15.1
-
28.1 169.7 6.0 2.1
Germany 1984 16.3 63.9 162.0 2.5 1.8 58.0 163.9 2.8 1.9
Israel 1992 28.3
- -
47.5 216.5 4.7 2.7
The Netherlands 1987 22.3 71.5 172.8 2.4 1.4 69.3 168.7 2.4 1.7
Sweden 1992 11.1 48.2 166.4 3.5 1.8 43.4 167.0 3.9 1.8
United Kingdom 1986 29.5 61.4 188.1 3.1 2.1 60.7 186.3 3.1 2.1
United States 1991 16.7 33.6 193.1 5.7 3.0 28.1 203.7 7.2 3.5
Females
Australia 1989 35.9 49.2 156.3 3.2 1.9 23.2 183.0 5.7 3.4
Canada 1987 30.6 34.7 179.1 5.2 2.6 27.9 181.8 6.5 3.2
Finland 1987 16.8
-
32.8 152.2 4.6 2.3
Germany 1984 47.9 45.9 156.0 3.4 2.0 23.1 180.6 7.8 3.4
Israel 1992 47.4
- -
35.3 228.3 6.5 3.0
The Netherlands 1987 62.0 72.6 173.5 2.4 1.7 29.9 185.1 6.2 3.1
Sweden 1992 12.3 37.9 153.2 4.0 2.2 30.7 156.6 5.1 2.4
United Kingdom 1986 50.1 64.9 181.0 2.8 2.0 34.6 223.0 6.4 3.5
United States 1991 25.7 40.0 190.0 4.8 2.5 17.7 206.0 11.6 4.0
Source: Authors' tabulations of Luxembourg Income Study database.
a
Persons aged 25 to 54, living in households with zero self-employment income. Wages are net of employer con-
tributions to social insurance (payroll taxes), but gross of employee payroll taxes.
b
Percentage of all persons aged 25 to 54 with zero earnings.
c
Full-Year: 50 full-time weeks or more a year; Full-Time: 35 or more working hours a week. Full-year-full-time
workers cannot be identified in the data for Finland or Israel.
d
All workers with non-zero wage and salary income.
at the bottom are matched by unusually
high earnings at the top of the distri-
bution. These countries are followed by
Sweden, Australia, and the United King-
dom which have 90/10 ratios for males
around 3.0, compared to 4.6 for Canada
and 5.7 for the United States. The coun-
tries with the most equal distribution of
male earnings are Germany and The
Netherlands with 90/10 ratios for full-
time males of around 2.5. The ranking of
countries for women working full-time is
similar to the ranking for males with the ex-
ception of Germany which goes from being
one of the most equal to being more sim-
ilar to Australia and the United Kingdom.
Table 1 shows that the earnings of per-
sons at the tenth percentile are lower
relative to the median in the United
States than in any other country. This
low relative earnings, however, need not
translate into low absolute earnings be-
cause the median is likely to be high in
the United States compared to all of the
644 Journal of Economic Literature, Vol. XXXV (June 1997)
Ratio of Ratio of Real
Low Length of bars represents EIigh High to Low National Median
Earnings the gap between high and Earnings Earnings To Real United
(P1O) low income individuals (P90) (Decile Ratio) States Median
The Netherlands 1987 51 124 2.40 72
Germany 1984 51 128 2.50 79
Australia 1989/90 51 _ 144 2.82 90
United
Kingdom
1986 42 130 3.10 69
Sweden 1992 41 - 140 3.50 84
Canada 1987 35 161 4.60 92
United States 1991 34 I _ _ _ _ 193 5.70 100
0 50 100 150 200
Averageb 44 146 3.52 84
Figure 1. Real Earnings Distribution Comparison for Full-Time Full-Year Males
(all figures in 1991 United States dollars)a
Source: Authors' calculations using the Luxembourg Income Study database.
aNumbers give real earnings (1991 United States dollars) as a percent of the United States median.
bSimple 7 nation average.
countries in Table 1. In order to get a
rough comparison of absolute earnings at
various points in the distribution it is
necessary to translate earnings into a
common currency. While comparing the
purchasing power of different currencies
is fraught with danger, these problems
are considered small enough to warrant
frequent comparisons in average (or me-
dian) incomes across countries. In the
same spirit we compare earnings at sev-
eral points in the earnings distribution
using the Penn World Tables of purchas-
ing power parities which allow us to
translate the values in Table 1 into 1991
United States dollars.24
Figure 1 presents the results. While
the last column shows that the United
States indeed has the highest median
male earnings among the countries
shown, the differences and sometimes
the rankings are quite different at the
PIO and P90. Column 1 shows that the
PIO measured in United States dollars is
higher in all countries than in the United
States. Indeed only Canada has values
nearly as low as the United States. Thus,
persons at the bottom of the earnings
distribution in the United States fare
poorly not only relative to the median in
the United States but also relative to
persons at the PIO in other countries.
For example, a worker at the PIO in the
German distribution earns 51 percent of
the median earnings in the United
States. In contrast, a worker at the PIO
in the United States distribution earns
only 34 percent of the United States me-
dian. At the other end of the distribution
a worker at the P90 in the United States
distribution earns 193 percent of the
United States median. This is by far the
highest value with most other countries
having a P90 at around 130 to 140 per-
cent of the United States median.
B. Trends in Earnings Inequality
The literature on changes in earnings
inequality in developed countries is now
large enough to begin to piece together a
coherent picture of similarities and dif-
ferences in trends. A few countries
closely mirror the United States' experi-
ence while others seem to have avoided
24
See Robert Summers and Alan Heston (1991)
for the basis of the estimates of purchasing power
parity used here. Figure 1 computes the median
high and low income values used in Table 1 as a
fraction of the United States median. For a similar
comparison with similar outcomes, see Richard
Freeman (1994, pp. 2-13).
Gottschalk and Smeeding: Cross-National Income Inequality 645
the increasing inequality of earnings, at
least temporarily. While we are a long
way from fully understanding the causes
for these similarities and differences, a
fairly consistent story is emerging.
Changes in Earnings Inequality in the
United States. Rising earnings and wage
inequality among male workers in the
United States has led to a substantial lit-
erature documenting the trends and at-
tempting to identify the causes. We fol-
low Levy and Murnane (1992) by
updating their summary of changes in
the dispersion in the overall wage distri-
bution. Like them, we further examine
changes in returns to observable skills
and changes in inequality within groups.
The former focuses on increases in wage
differentials between high school and
college graduates and between new en-
trants and older workers. Within group
inequality focuses on increased disper-
sion in the wage distributions within
education and experience groups.
Almost all studies of the United States
use the Current Population Survey
(CPS) to examine the distribution of
weekly or annual wages for males.25 In
order to concentrate on changes in
wages and not changes in hours worked,
most studies select only persons working
full-time and full-year. Because the large
changes in labor force participation of
women make it difficult to separate
changes in the distribution of wages
from changes in the composition of the
female labor force, most studies further
focus on the distribution of earnings of
males. These studies find that wage
growth varied substantially between the
upper, middle, and lower tails of the dis-
tribution. For example, between 1975
and 1992 the P75 ratio for hourly earn-
ings of males in the United States in-
creased by 10 percent and the P90 ratio
increased by 14 percent. In contrast, the
PIO and P25 ratios decreased by 3.2 and
5.1 percent respectively.26 Changes in
the distribution of weekly and annual
earnings were even larger.
Part of the observed change in the dis-
tribution of wages reflects large in-
creases in the returns to education dur-
ing the 1980s. For example, in 1979 the
hourly earnings of recent college gradu-
ates were 23 percent higher than the
earnings of recent high school gradu-
ates.27 By 1989 the college premium in
wage rates for this group had increased
to 43 percent. Because hours worked by
recent high school graduates also fell
relative to the hours worked by college
graduates, the increase in the college
premium in annual earnings was even
larger (from 30 percent to 54 percent).28
The returns to experience also increased
during the 1980s, though not as much as
the returns to education.29 The result of
these trends was a dramatic decline in
the relative position of young high school
graduates and high school dropouts rela-
tive to workers with more experience or
education.
In addition to the increased inequality
between education and experience
groups, studies consistently find large in-
creases in wage dispersion even within
skill groups.30 The wage differential be-
25
There are some exceptions. For example,
Gottschalk and Moffitt (1994) use the PSID.
26
Unpublished data updating Table 2B2 of
Karoly (1993).
27Authors' tabulations of the 1979 and 1989
Census of population for males with less than ten
years of experience.
28These increases in returns to college during
the 1980s are in sharp contrast to the decline in
the returns to education during the 1970s.
29Increases in returns to experience were lim-
ited to less educated workers.
30Thomas Macurdy and Thomas Mroz (1995)
show that the steepening of the cross-sectional ex-
perience profile is a result of downward shifts in
the profiles of more recent cohorts, not the steep-
ening of cohort-specific profiles. The increase in
within group inequality of relative income (i.e.,
ln(P90) -ln(PIO)) reflects constant absolute differ-
ences (i.e., constant p90 -PlO) which translate into
larger relative differences as real earnings decline.
646 Journal of Economic Literature, Vol. XXXV (June 1997)
tween the 90th and 10th percentile in-
creased within the distribution of wages
of young and old workers and within the
distribution wages of high school and
college graduates. Persons in the upper
part of the distribution experienced sig-
nificant growth in real wages while those
in the lower part of the conditional dis-
tribution experienced slight growth or, in
most cases, declines in real wages. The
increase in within group inequality, how-
ever, seems to have started earlier, be-
ginning in the early 1970s.
Explanations for Rise in Earnings
Inequality in the United States.
While there is substantial agreement
about the facts there is still disagree-
ment about the underlying causes. A
variety of changes in the economy, such
as changes in industrial structure, in-
creased foreign trade, increased immi-
gration, skill-based technical changes,
and the decline in institutions that
limit the market (e.g., the fall in the
minimum wage and the decline in
unionization) are consistent with the in-
crease in inequality.3' Disentangling
these explanations is inherently difficult
not only because of data limitations, but
because these explanations are poten-
tially interrelated. For example, if part of
the decline in unionization or the tech-
nological change is the result of in-
creased foreign competition, then one
should attribute these indirect effects to
trade. Likewise, changes in institutional
constraints, such as a decline in union-
ization, may reflect changes in market
forces which limit the options for low
skilled workers.
One set of explanations for the rise in
inequality in the United States focuses
on changes in institutional constraints,
specifically the erosion of the real mini-
mum wage and the decline in union den-
sity.32 During the 1980s the real mini-
mum wage fell by 44 percent. This is
consistent with the decline in wages at
the very bottom of the distribution. But
the decline in the minimum wage can-
not explain the increase at the top of
the distribution or increases in inequal-
ity within high education groups. Spill-
over effects are too small to explain the
large changes elsewhere in the distri-
bution.
However, because much of the change
in inequality in the United States reflects
declines at the bottom of the distri-
bution, the impact of the minimum wage
is not negligible. Several studies estimate
that the decline in the real minimum
wage accounts for roughly 30 percent of
the increase in the dispersion of wage
rates (for example, Nicole Fortin and
Thomas Lemieux, forthcoming). If de-
mand functions are not totally inelastic,
employment will increase as the real
minimum wage declines. This increase in
hours will offset part of the decline in
the wage, leading to a smaller increase in
the dispersion of earnings than wages.
This is consistent with studies that attrib-
ute considerably less than a third of the
decline in the share of earnings received
by the lowest quintile to the decline in
the real minimum wage (for example,
Michael Horrigan and Ronald Mincy
1993).
The decline in unionization is another
measurable institutional change which
could have contributed to the increase in
earnings inequality. The net impact of
unions on the distribution of earnings is
31
For a brief summary of competing explana-
tions see Danziger and Gottschalk (1995, ch. 6).
32
David Gordon (1996) views the change in the
minimum wage and union density as part of a
broader set of institutional changes in which cor-
porations squeezed workers in reaction to in-
creased foreign competition. According to this
broader institutional view, wages are set by corpo-
rations largely independently of market forces. Ac-
cording to Gordon (1996, p. 206), management in-
stitutional change is probably the most important
factor leading to the wage squeeze.
Gottschalk and Smeeding: Cross-National Income Inequality 647
ambiguous. Unions increase the wage
differential between unionized and non-
unionized workers with similar charac-
teristics but lower inequality by reducing
differentials among organized workers
and by raising the wages of persons with
characteristics associated with lower
earnings (e.g., semi-skilled white-collar
occupations). Their impact on the distri-
bution of hours and employment is like-
wise ambiguous. Estimates of the impact
of the decline in unionization suggest
that unions account for roughly 20 per-
cent of the increase in male earnings in-
equality. The decline in unionization,
however, accounts for little of the
changes in the distribution of earnings
among women (Freeman 1994; Fortin
and Lemieux, forthcoming).
Because studies of the impact of re-
ductions in the minimum wage and de-
clines in union density have focused
separately on these institutions, they may
well double count the impact on inequal-
ity. However, even if the estimated im-
pacts of the decline in unions and the
reduction in the minimum wage are not
additive, it is clear that changes in these
two institutional factors had a substantial
combined effect on the rise in earnings
inequality.
Changes in market forces, however,
must also be part of the story. The sharp
increase in both the skill premium and
skill intensity suggests that demand was
shifting faster than supply. The 1980s
were clearly a period of sharp increases
in returns to skill, measured either in
terms of returns to education or experi-
ence. These increases in the relative
price of skilled workers occurred at the
same time that labor markets were ab-
sorbing an increasingly large number of
these workers. The baby bust made older
workers a relatively abundant input and
the continued increase in educational at-
tainment meant that college educated
workers were arriving on the labor mar-
ket in increasing numbers. The fact that
the skill intensity increased at the same
time as the skill premium increased pre-
sents a prima facie case for the impor-
tance of demand shifts in explaining
changes in the earnings distribution in
the United States.
While there is substantial agreement
that shifts in demand are central to the
causal story, this still leaves open many
competing demand side explanations.
The three leading contenders are "de-
industrialization," increased interna-
tional competition, and skill biased tech-
nical change which all predict a shift out
in the demand for skilled labor.
The deindustrialization hypothesis fo-
cuses on shifts in derived demand for
skilled labor resulting from shifts in the
composition of demand for final prod-
ucts.33 While there is no dispute that the
manufacturing sector shrank as the ser-
vice sector grew, especially the high
wage service sector, this change in indus-
trial composition is not likely to be the
major factor causing the increase in de-
mand for skilled workers. Shifts in em-
ployment across sectors can account for
only a fraction of the increase in skill in-
tensity. For example, Kevin Murphy and
Finis Welch (1993, p. 126) estimate that
changes in industrial shares can account
for only 16 percent of the overall change
in demand for college educated work-
ers.34 While the deindustrialization hy-
pothesis properly predicts that both
quantities and prices for skilled labor
would increase, at best it is a part of a
larger story.
Increased international competition
could have also increased the demand
33 The deindustrialization hypothesis and the
foreign trade explanations over ap, to the extent
that some of the industrial shifts reflect changes in
trading volume or patterns.
34 Eli Berman, John Bound, and Zvi Griliches
(1993) also conclude that changes in the industrial
structure were not a major factor causing the in-
creased inequality.
648 Journal of Economic Literature, Vol. XXXV (June 1997)
for skilled labor.35 Two theoretical
frameworks have been used to analyze
the links between changes in interna-
tional trade and changes in the wage dis-
tribution. The factor content of trade ap-
proach, used largely by labor economists,
focuses on changes in relative effective
supplies of less skilled labor (for exam-
ple, see Berman, Bound, and Griliches
1993). Imports embody skilled and un-
skilled labor which, when added to do-
mestic supplies, determine the effective
supplies of these two factors. Because
imports are less skill intensive than do-
mestic production, the opening of trade
increases the relative effective supply of
less skilled workers, which puts down-
ward pressure on their relative wages.
This factor content of trade approach
has been severely criticized by several
trade economists who argue that exoge-
nous output prices, not endogenous fac-
tor quantities, determine relative wages
(for example, see Edward Leamer 1996).
This conclusion is based on the Stolper-
Samuelson theorem that develops links
between factor prices and output prices
which are set on world markets. Two
countries with the same technology, fac-
ing the same output prices, will tend to
have the same wage structure, regardless
of their volume of trade. Therefore,
newly liberalized international trade with
less skill intensive countries changes rela-
tive output prices in the domestic econ-
omy and, hence, the distribution of
wages, no matter what happens to the
volume of trade. According to this trade
theory, it is the decline in the relative
prices of less skill intensive goods, not
the increase in the volume of trade, that
should be the focus of empirical analysis.
The resolution of this ranging debate
is of more than theoretical interest be-
cause estimates of the impact of trade
differ widely across methods. The gen-
eral consensus emerging from factor
content of trade literature is that in-
creased trade accounts for less than 20
percent of the shift in demand.36 There
is a wider range of estimates in the lit-
erature that focuses on the effects of
trade on output prices, with some of the
more vocal critics of the factor content
of trade finding that trade accounts for
40 percent of the decline in wages of less
skilled workers.37
While it is too soon to tell whether a
common ground will be found, Paul
Krugman (1995) offers a possible recon-
ciliation. He argues that the appropriate
counterfactual is what the prices of trad-
ables (and hence wages) would have
been if trade had not expanded. Existing
studies that use observed changes in
prices may over or understate the rele-
vant changes in prices because they in-
clude the impact of numerous factors
other than trade.38 The question is how
to infer the prices that would have oc-
curred in the absence of trade. The an-
swer to this "what if' question about
prices depends crucially on the volume
of trade. If the expansion of trade were
small then there would be little impact
on world prices of tradables. Krugman
develops a CGE model based on com-
monly used supply and demand elastici-
ties to infer the price changes associated
with the increase in trade. As might be
expected, world supply and demand do
not shift very much as long as new trade
35
See David Richardson (1995), Freeman
(1995), and Burtless (1995) for reviews of this lit-
erature.
36
See Freeman (1995) for a review. The excep-
tion is Adrian Wood (1994), who attributes as
much as 50 percent of the decrease in demand for
less skilled workers to international trade.
37 Robert Lawrence and Matthew Slaughter
(1993) find little impact of changes in output
prices in the 1980s while Jeffrey Sachs and
Howard Shatz (1994) and Leamer (1996) find
larger impacts. Leamer's estimate of 40 percent
requires long lags because he uses price changes
in the 1970s to explain wage changes in the 1980s.
38
Factors other than trade may have either rein-
forced or countered the impact of trade.
Gottschalk and Smeeding: Cross-National Income Inequality 649
volumes are small relative to the total.
Given reasonable elasticities, prices have
to adjust only a small amount in order to
absorb the excess demand brought about
by increased trade. Therefore, only a
small part of the observed change in
relative prices is relevant to the ques-
tion: What would prices (and hence
wages) have been if there had been no
change in trade?
The value of this work is that it has the
potential for bridging the analytical gap
between the two approaches by showing
that the volume of trade is indeed rele-
vant to the key "what if' question. Not
surprisingly, the two approaches give
similar empirical results when the hypo-
thetical changes in prices are used rather
than the actual changes in prices.
It should be noted that all explanations
based on increases in international trade
leave unexplained the rising skill inten-
sity in non-traded goods as well as traded
goods sectors. In spite of having to pay
more for skilled workers, employers in
almost all sectors (traded as well as un-
traded goods) chose to hire more skilled
workers
39
Widespread skill biased technological
change would be consistent with in-
creases in both the skill intensity and
skill premium within finely defined
industry occupation cells.40 Firms
would bid up the price of skilled work-
ers as their productivity increased rela-
tive to the productivity of less skilled
workers.
There are two primary objections to
the technological change explanation.
The first is that technological change is
simply a label for our ignorance. Because
changes in technology are difficult to ob-
serve directly, its importance is often in-
ferred by ruling out other factors. As
Steven Davis and Robert Topel (1993)
have colorfully stated, "The argument for
the skill-biased technical change hy-
pothesis is a bit like inferring the exist-
ence of Pluto, because Neptune's orbit
does not otherwise fit the predictions of
theory." Likewise ubiquitous increases in
inequality require some widespread
force, like technology, that cannot be
easily observed. While this critique does
have a ring of truth, there are now a
variety of studies of specific technologi-
cal changes that have increased the de-
mand for the more skilled.41 These
direct sightings of Pluto make the tech-
nological change explanation more com-
pelling.
The second argument against the im-
portance of technological change focuses
on timing (Lawrence Mishel and Jared
Bernstein 1996; David Howell 1995).
Earnings inequality increased most rap-
idly during the 1980s. According to the
critics, this implies that technological
change accelerated during this period.42
But the well-known series on productiv-
ity growth shows a deceleration in out-
put per hour during the 1980s, hardly
strong evidence for an acceleration in
technological change. Furthermore, the
econometric evidence on changes in
capital-skill complementarity and skill
39
Note that other explanations of the rise in in-
equality, such as the decline in real minimum
wages or the decline in unionization, cannot ex-
plain the rise in skill intensity because these
changes would have made less skilled workers
cheaper, leading firms to decrease skill intensity.
40
Trade theory models focus on sector bias,
skill neutral technological change, which also
raises the wages of skilled workers if the techno-
logical change takes place in the skill intensive
sectors.
41 For example, see Thomas Bailey (1988), Ber-
man, Bound, and Griliches (1993), and Peter Cap-
pelli (1993), as well as BLS studies of the impact
on individual industries, such as U.S. Bureau of
Labor Statistics (1994).
42
It should be pointed out that these two criti-
cisms of the tec nological explanation for the
growth in inequality cannot both be right. If tech-
nology cannot be directly observed then it is im-
possible to tell if it accelerated or not. If, on the
other hand, slow growth in average productivity
reflects slow technological change, then Pluto is
observable.
650 Journal of Economic Literature, Vol. XXXV (June 1997)
bias technical change during the 1980s is
mixed.43
The critique based on timing properly
corrects sometimes sloppy use of lan-
guage but it does not deal with the heart
of the argument in favor of the impor-
tance of technological change. It is true
that the literature's stress on demand
side factors sometimes seems to ignore
Marshall's dictum that it takes both
blades of the scissors (demand and sup-
ply) to explain changes in prices. If sup-
ply had grown at a constant rate then de-
mand would have had to accelerate
during the 1980s in order to explain the
increase in the relative wages of skilled
labor. But the supply of educated work-
ers increased at a decreasing rate during
the 1980s, which is consistent with an in-
crease in the college premium in the
face of non-accelerating growth in de-
mand (Katz and Murphy 1992). While it
may be sloppy language to attribute the
rise in the wage premium to demand
side factors, there is nothing inherent in
the argument that requires an accelera-
tion in the shift in demand. Deceleration
in supply will do.
What is required of any explanation
for the increase in inequality is that the
shift in demand be greater than the shift
in supply. Otherwise the explanation will
not be consistent with the rise in skill
intensity in the face of a rise in the skill
premium. Technological change remains
one of the only factors that will result in
a ubiquitous increase in the proportion
of college educated workers employers
are willing to hire in spite of the large
increase in the college premium. Dein-
dustrialization, increases in international
trade, and declines in unionization and
the real minimum wage are all consistent
with the a decline in the relative wages
of less skilled workers but all these theo-
ries predict that firms would choose less
skill intensive production methods, not
more skill intensive methods, as we in
fact observe. Only technological change
is consistent with rising skill intensity in
the face of rising skill prices.
The final explanation for the rise in in-
equality focuses on the distinction be-
tween increases in inequality of perma-
nent earnings and increases in the
volatility of earnings.44 Almost all the ex-
planations reviewed thus far implicitly
assume that the increase in inequality is
a result of increases in the dispersion of
permanent earnings. For example in-
creases in the return to education,
whether caused by skill biased techno-
logical change or increased international
trade, will raise the permanent earnings
of college graduates. Likewise, decreases
in the real minimum wage are assumed
to lower the long-run earnings of less
skilled workers.45
While the focus of most explanations
has been on factors that increase the dis-
persion of permanent earnings, the
cross-sectional evidence that these theo-
ries are attempting to explain cannot
distinguish between changes in perma-
nent and transitory of earnings.46 Longi-
tudinal data is necessary to separate
the relative importance of these two fac-
tors.
Evidence for the 1980s indicates that
increases in the dispersion of permanent
earnings and increases in the variability
of transitory earnings were roughly
43
Mishel and Bernstein (1996) find no evidence
of increased capital-skill complementarity while
Claudia Goldin and Lawrence Katz (1996) find
higher complementarity and increased skill bias
when comparing the 1970s with the 1980s.
44The basis for this distinction is the canonical
error components model.
45
Some institutional explanations, such as de-
creases in unionization, may be consistent with
greater earnings variability.
46
In the simplest model, in which observed log
of earnings is equal to a time invariant person spe-
cific permanent component and transitory compo-
nent, the variance of log earnings is equal to the
variance of the permanent component plus the
variance of the transitory component.
Gottschalk and Smeeding: Cross-National Income Inequality 651
equally important in accounting for the
increase in inequality (both for annual
and weekly earnings).47 Part of the in-
crease in the variability of earnings re-
flects increases in the variance of weeks
worked but weekly earnings also became
less stable.
This suggests that the search for causal
links should focus on factors associated
with greater instability of both weeks
and wages. While this line of research
points in a new direction, we know rela-
tively little about changes in market or
institutional forces- that may have led to
greater year to year (or week to week)
fluctuation in earnings. The decrease in
unionization seems to be part of the
story, but transitory fluctuations in-
creased among unionized as well as
non-unionized workers. Involuntary job
losses from layoffs and firings and volun-
tary quits both increased during the
1980s (Johanne Boisjoly, Greg Duncan,
and Smeeding 1996). Likewise the de-
crease in job duration is a contributing
factor, but instability increased even
among persons who stayed in the same
job.48 The fact that increased instability
accounts for roughly half of the increase
in overall inequality and that we know so
little about its cause opens an obvious
line for future research.
IV. Changes in Earnings Inequality in
Other Industrialized Countries
A. Similarities and Differences
Table 2 provides a summary of
changes in male earnings inequality dur-
ing the 1980s. The table includes the ten
countries for which we have informa-
tion on trends in overall inequality and
trends in returns to education (or occu-
pation) and experience as well as trends
in inequality within education and expe-
rience groups. Because studies of trends
in inequality in other countries summa-
rized in this table vary widely in popula-
tions covered, measures of inequality,
period covered, and a whole host of dif-
ferences that make comparisons across
countries difficult, we focus on studies
that contrast each country with the
United States. This allows us to bench-
mark the change in inequality in each
country to the corresponding change in
the United States. We also include
studies that do not provide specific
comparisons but where authors discuss
their findings in light of changes in the
United States. Our rankings for these
studies, which are marked with an aster-
isk, reflects the authors' qualitative judge-
ment.
The table shows the absolute change
in inequality in each country measured
as a percentage of the absolute change in
inequality in the United States. For ex-
ample, the ++ in Column (3) for Can-
ada signifies that the increase in overall
inequality in Canada was 50 to 80 per-
cent as large as in the United States.49
Because the use of absolute changes is
arbitrary, we also indicate where classifi-
cations would be altered if we compared
relative changes in inequality in the two
countries.50 While these two metrics do
47
See Gottschalk and Moffitt (1994) and Moffitt
and Gottschalk (1995). The latter paper defines
transitory earnings as shocks that die out within
three years.
48
See Gottschalk and Moffitt (1994) for contrib-
uting factors. Henry Farber (1995) finds little
change in job duration for the period covered by
Gottschalk and Moffitt. His more recent unpub-
lished tabulations, however, show a clear decrease
in duration for more recent years.
49 As Appendix Table A indicates, Blackburn
and David Bloom (1994) show the variance of log
earnings increasing by .018 in Canada (from .270
to .288). The change in the United States over the
same period is .036 (from .286 to .320). Thus, in-
equality rose half as much in Canada as in the
United States according to this measure.
50
The classification for Canada is unaltered be-
cause the relative change in Canada of .067 is 77
percent as large as the .119 change in the United
States (.288/ .270 -1 versus .320/ .286 -1). This
continues to fall in the 50 to 80 percent band.
652 Journal of Economic Literature, Vol. XXXV (June 1997)
not exhaust all possible comparisons,
they are the most commonly used meth-
ods.51
The countries shown in Table 2 break
down into four broad groups. The first
consists of countries that experienced at
least as large an increase in inequality as
in the United States. This group includes
only the United Kingdom.52 A second
group which experienced substantial in-
creases in inequality but less than
the United States and the United King-
dom includes Canada, Australia, and Is-
rael.53 France, Japan, The Netherlands,
Sweden, and Finland form a third group
with positive but quite small changes in
earnings inequality over the 1980s
(though inequality started rising in sev-
eral of these countries in the second half
of the decade).54 While even the Nordic
countries experienced some increase in
earnings inequality during the 1980s,
they started from very low levels, result-
ing from a long secular decline in in-
equality.55 Finally, Italy and Germany
form a small group that experienced no
measurable increase in earnings inequal-
ity during the 1980s.56
Thus, what we observe is a diversity of
experiences but with almost all countries
experiencing some increase in earnings
inequality. The hypothesis that inequal-
ity increased only in the United States
can clearly be rejected. However, the hy-
pothesis that all western industrialized
countries experienced as large increases
in inequality as the United States is
equally unsustainable. Clearly the
United States was a leader in the trend
toward greater inequality of labor market
outcomes but most other countries expe-
rienced some changes. Whether one
stresses the differences or the common-
alities is like describing a bottle as half
full or half empty.
When one goes behind changes in
the overall distribution and starts to
examine changes in inequality at differ-
ent points in the distribution or trends
in returns to education or experience,
further similarities and differences
emerge. The United States earnings
distribution became less equal both
because of growth at the top and decline
in absolute and relative earnings at the
bottom. While the absolute decline in
real earnings at the bottom of the distri-
bution is limited to Australia, the United
States, and Canada, the decline in rela-
tive earnings in the lower deciles is
common across a large number of coun-
tries, including Japan, The Netherlands,
Sweden, and the United Kingdom
(OECD 1993, Table 5.2). While less
skilled workers lost ground during the
1980s in most countries, the gains at
the top of the distribution were more
modest than in the United States. Only
the United Kingdom rivaled the United
51
Because any monotonic transformation of an
inequality measure maintains ordinal ranks, there
is no natural metric for comparisons. Each metric
reflects an implicit social welfare function. Our
absolute (relative) classification scheme
implicitly
assumes that equal changes in absolute (relative)
changes in inequality are equally valued.
52As a result of allowing market forces to influ-
ence wages, Russia, Hungary, and the former East
Germany experienced considerably larger percent-
age changes in earnings inequality than the United
States or the United Kingdom. However, these na-
tions are outside the scope of our study.
53 In Canada and Australia the rise in inequality
was largely a result of declines at the bottom of
the distribution. Israel saw very modest declines at
the bottom but large increases at the top.
54 Tom Eriksson and Jantti (1994) show that the
rise in inequality in Finland after 1985 was as
large as the increase in the United States. Pierre
Concialdi (1997, Table 1) shows moderate in-
creases in France between 1984 and 1989.
55 Eriksson and Jantti (1994) show that inequal-
ity increased in Finland between 1985 and 1990
but this followed a sharp decline during the 1970s
and early 1980s. Likewise, what is anomalous
about the 1980s in Sweden (Douglas Hibbs 1990)
and France (Concialdi 1997) is not the rise in in-
equality, which was small, but the ending of a long
period of rapidly falling inequality.
56
Richard Hauser and Irene Becker (1993, Ta-
ble 4), who exclude households with a foreign
head, show a 2.7 percent increase in the Gini coef-
ficient for West Germany between 1983 and 1990.
Gottschalk and Smeeding: Cross-National Income Inequality 653
States in the increase in the P90/P50
ratio.
Columns (4) to (6) of Table 2 summa-
rize changes in returns to experience and
education (or occupation), as well as
trends in inequality within skill groups.
All countries, except Finland, Israel, and
Italy, experienced increases in inequality
within skill groups and most countries
experienced an increase in the returns
to experience. The United States stands
out in two important respects. First, it is
the country with the largest increase in
returns to education. Second, it experi-
enced large increases in all three com-
ponents-increases in returns to both
education and experience, as well as
increases in inequality within groups.
Only the United Kingdom also experi-
enced large increases in all three compo-
nents.
The commonalities suggest that simi-
lar factors may have affected these coun-
tries. The differences suggest that these
forces were either not equally strong
in all countries or that they were coun-
tered by country-specific factors. For
example, some countries may have
experienced supply shifts that countered
the demand shifts, leaving relative
wages constant. Countries with declining
proportions of young people in the labor
market should have experienced smaller
increases in the experience premium,
and fewer young people competing for
a dwindling number of jobs should
have limited the decline in their wages.
Similarly, countries with large increases
in college enrollments should have ex-
perienced relatively small increases in
the college premium as the growth in
supply offset some of the increase in de-
mand.
Differences in wage setting institu-
tions may account for some of the differ-
ences in growth in inequality. There is
certainly a prima facie case that coun-
tries with high union coverage or cen-
tralized wage setting were able to limit
the growth in inequality. (Freeman and
Katz 1993; and Fortin and Limieux,
forthcoming, discuss the role of institu-
tions.) Germany, Italy, and the Nordic
countries have fairly centralized wage
setting and a high proportion of their
workforce covered by collective bargain-
ing agreements (Lars Calmfors and John
Driffill 1988; OECD 1994b). At the
other extreme, unionization rates de-
clined in both the United States and the
United Kingdom and wage bargaining
became less centralized in the United
Kingdom (David Blanchflower and Free-
man 1992).
B. Impact of Changes in Relative
Supplies
We start by turning to the cross-
national relationship between changes in
the rates of return to education and ex-
perience and changes in the relative sup-
plies of persons classified by education
and experience. If market forces were
responsible for the diversity of changes
in returns to education, then countries
with faster growth in college educated
workers would have experienced smaller
growth in the education premium.57
Likewise, countries with a baby bust en-
tering the labor market would have expe-
rienced smaller than average increases in
the experience premium as less experi-
enced workers became relatively scarce.
The question we ask is whether these
supply shifts are sufficient to explain the
small increases in relative factor prices
in countries with centralized wage set-
ting. If they are, then this suggests that
institutional constraints may not have
been binding.
Exploring the importance of shifts in
supply requires estimates of changes in
57
This assumes that shifts in demand were
roughly equal across countries, or at least that
shifts in demand were not strongly positively cor-
related with supply shifts.
654 Journal of Economic Literature, Vol. XXXV (June 1997)
TABLE 2
CHANGES IN MALE EARNINGS INEQUALITY OVER THE 1980s IN INDUSTRIALIZED COUNTRIESa
Overall Returns Earnings
Earnings Returns to to Education Inequality
Country and Authors Years Inequality Experience or Occupationb Within Group
(1) (2) (3) (4) (5) (6)
Australia 1981-89 + ++ mixed ++
*
Borland (1992) 1981-85 ++ +++ -+++
Gottschalk and Joyce (1995) 1976-90 +++ na na
Gregory (1993)
Canada 1979-87 ++ ++ -+++
Blackburn and Bloom (1994) 1981-87 ++ ++ + ++
Gottschalk and Joyce (1995)
Finland 1980-90
0 0 0 0
* Ericksson and Jantti (1994)c 1987-91 + -
- 0
Gottschalk and Joyce (1995)
France 1976-87 + + (0) mixed
Katz, Loveman, Blanchflower (1995) 1979-84
++a +++b (-) +
Gottschalk and Joyce (1995)
Germany 1983-88 0 0 0 na
*
Abraham and Houseman (1995)
Israel 1979-86 + +++ ++a 0
Gottschalk and Joyce (1995)
Italy 1978-87 Oc oc
Oa -c
Ericksson and Ichino (1995)
Japan 1974-90 +b mixed + na
Katz, Loveman, Blanchflower (1995)
The Netherlands 1979-89 0 0 +
* Hartog, Oosterbeek, Teulings (1992) 1983-87 +b +++ +
Gottschalk and Joyce (1995)
Sweden 1984-91 ++ + ++ +++
*
Edin and Holmlund (1995)d 1981-87 +b - (+++) +++
Gottschalk and Joyce (1995)
United Kingdom 1979-90 +++ ++ (++) +++
Katz, Loveman, Blanchflower (1995) 1979-86 +++ +++ (+++) +++
Gottschalk and Joyce (1995)
aClassification for studies that compare country to the United States in same time period (for measures, see
Appendix A):
+ + + increase in inequality at least 80 percent as large as in the United States
+ + increase 50 to 80 percent as large as in the United States
+ increase 10 to 50 percent as large as in the United States
0 increase from -10 to +10 percent of change in the United States
-
decrease greater than -10.
Classification for other countries based on authors' qualitative comparison.
bParentheses signify returns to higher paid occupations (e.g., non-manual). Wherever possible, returns to
education are for recent labor market entrants.
c
Small changes over decade reflect decline followed by sharp increase after 1985.
dInequality was constant from 1974-84 in this study.
Gottschalk and Smeeding: Cross-National Income Inequality 655
returns to education and experience in
each country. Two types of evidence are
available. The first comes from LIS
which allows similar earnings functions
to be estimated across a variety of coun-
tries using similar samples and variable
definitions (Gottschalk and Joyce 1996).
The second source of evidence comes
from country-specific studies that pro-
vide less comparability across countries
but greater detail on the specific country
being studied.
The data from LIS provide evidence of
the importance of market forces. There
is a systematic negative relationship be-
tween the size of supply shifts and
changes in education and experience
premium across countries. The relation-
ship is particularly strong for the educa-
tion premia. Sweden, Finland, and Can-
ada experienced relatively small growth
in the relative supply of young workers
during the years coverd by LIS
(Gottschalk and Joyce 1996). This was
accompanied by small increases in the
age premium in these countries. In con-
trast, The Netherlands experienced a
large inflow of young workers and a sub-
stantial decline in their relative earnings,
which is again consistent with a market
explanation for changes in the age pre-
mium. Changes in the education pre-
mium also show a negative relationship.
The Netherlands experienced the largest
yearly growth in the proportion of work-
ers with a college degree and it experi-
enced an actual decline in the college
premia. Likewise, large increases in the
supply of college workers in Israel and
Australia are consistent with the small
increase in their college premia. At the
other extreme, the United States and
Canada experienced relatively small in-
creases in the supply of college workers.
This was accompanied by substantially
larger increases in the college premium
than in the above countries.
The other sources of evidence on the
importance of shifts in supply are coun-
try studies. While it is difficult to make
comparisons across countries because
concepts and measures often differ
across studies, the general pattern is
similar to that found in LIS. These stud-
ies find fairly consistent effects of
changes in the education composition of
the workforce and some weaker support
for the proposition that the age composi-
tion affected the experience premium
(Katz, Gary Loveman, and Blanchflower
1995). Increases in the college premium
in Japan, the United Kingdom, and the
United States are consistent with de-
mand shifting faster than supply. Like-
wise, Sweden, Canada, and Australia
offer support for the importance of
changes in the supply of college edu-
cated workers. In Sweden, the ratio of
workers with a college degree to those
with a gymnasium degree rose steadily
during the 1970s and early 1980s. This
was followed in the late 1980s by a de-
cline in the proportion of workers with a
college degree and a modest increase in
the returns to education, which is consis-
tent with a simple supply/demand expla-
nation (Pers-Ander Edin and Bertil
Holmlund 1995). The smaller increases
in returns to education in Canada than in
the United States are also largely ex-
plained by the substantially larger
growth in the proportion of the work
force with a college degree in Canada
than in the United States.58 The lack of
growth in the educational premium in
Germany, The Netherlands, and Austra-
lia (during the 1970s) can also be ex-
plained by shifts in the relative supplies
of college educated workers.59 Australia
58
See Freeman and Karen Needels (1991). John
DiNardo and Lemieux (1993) conclude, however,
that changes in unionization and the minimum
wage, not relative supplies, were the driving forces
in Canada.
59 Katharine Abraham and Susan Houseman
(1995) show that educational attainment contin-
ued to accelerate during the 1980s, which is con-
656 Journal of Economic Literature, Vol. XXXV (June 1997)
offers a stark example of the impact of
an accelerating supply of college work-
ers. The most rapid growth in educa-
tional attainment occurred prior to 1978,
a period during which the education pre-
mium actually fell (Robert Gregory and
Frank Vella 1992). Returns to education
also dropped substantially in The Neth-
erlands during the 1980s while the pro-
portion of workers with a college degree
increased dramatically as a result of gen-
erous government subsidies for educa-
tion during the 1970s (Hartog, Hessel
Oosterbeek, and Coen Teulings 1993).
The evidence from LIS and country-
specific studies strongly suggests that
market forces played a role in limiting
the increase in inequality. This state-
ment applies also to countries with cen-
tralized wage setting institutions. While
it is possible that some omitted variable
is responsible for the negative correla-
tion between changes in factor prices
and changes in relative factor supplies,
these studies at least provide the empiri-
cal basis for the presumption that market
forces matter, even in countries with
centralized wage setting.
C. Impact of Differences in Institutions
Given the large proportion of workers
covered by collective bargaining agree-
ments and the centralization of wage set-
ting in many OECD countries, one
should look beyond market forces to ex-
plain changes in the structure of wages.
Most western European countries place
a greater emphasis on distributional is-
sues than the United States and many
have centralized wage setting institutions
that can be used to limit the impact of
market forces. The question is how
much of the diversity in trends in in-
equality can be explained by these insti-
tutional factors?
Again, conceptual as well as measure-
ment issues must be confronted. First,
how should one measure the degree to
which wages are set by "institutions" in
different countries? One common mea-
sure is the union density rate (the pro-
portion of the workforce belonging to a
union) but this measure potentially
misses workers who do not belong to a
union but who are covered by union-ne-
gotiated wage agreements. For example,
in France, 85 percent of the workforce
was covered by collective bargaining
agreements in 1980 but only 17.5 per-
cent belonged to unions (OECD 1994c,
Table 5.8). Furthermore, neither union
density nor union coverage necessarily
captures the degree to which wage set-
ting is centralized. For example, Japan
has low union coverage rates and bar-
gaining is at the company level but wage
demands are coordinated through a na-
tionwide Shunto (spring offensive) which
sets guidelines that form the basis for
company level bargaining. Thus, while
institutional wage setting in an environ-
ment of high coverage rates in countries
such as Germany and Norway points to
the importance of unionized wage set-
ting, the picture is less clear for coun-
tries that share some, but not all, of
these attributes.
While it is easy to contrast the decen-
tralized labor markets of the United
States with the more centralized or
unionized labor markets in most of
Europe, ranking European countries is
more problematic. Countries differ in
many dimensions, making it difficult to
aggregate into a single summary mea-
sure. For example, bargaining is fairly
decentralized in France but the bottom
of the wage structure is tightly con-
trolled by a widely applied minimum
wage (the SMIC). If market forces
changed primarily at the bottom of the
distribution, this seemingly minor insti-
tutional factor might be paramount.
sistent with the stability of the education premium
in Germany.
Gottschalk and Smeeding: Cross-National Income Inequality 657
Given the idiosyncratic nature of coun-
try-specific institutions, it comes as no
surprise that indices of centralization
have come to different rankings depend-
ing on the weights they attach to dif-
ferent attributes (Calmfors and Driffill
1988; Alberto Alesina and Roberto
Perotti 1994).
The diversity of institutional arrange-
ments across countries suggests that
using a single measure may be inappro-
priate, but this leaves a great deal of
room for ex-post rationalization. If in-
equality increases, one may be tempted
to infer that the particular institutions in
that country were not effective in limit-
ing the impact of market forces. But this
is not a test of the institutional hypothe-
sis, because it assumes that institutions
matter and infers the effectiveness of the
particular institution from the outcome.
The second measurement issue fo-
cuses on the distinction between levels
and changes. For example, is inequality
expected to increase in countries with
high but declining centralization of wage
negotiations? Sweden entered the 1980s
with bargaining at the national level but
moved somewhat away from this highly
centralized system in the early 1980s as
employers withdrew from this arrange-
ment. It was only by starting with suffi-
ciently centralized labor markets that
countries like Sweden managed to end
the decade with markets that were still
as centralized as countries like Germany
that did not experience similar institu-
tional changes. Likewise, the union cov-
erage declined in many countries during
the 1980s, including the United King-
dom and Australia.
The key conceptual question is
whether the level of the institutional
constraint or the change in level is rele-
vant. Many countries with centralized
wage setting or high rates of union cov-
erage saw these institutions weakened
during the 1980s. This could be used to
rationalize either stable wages (the insti-
tutions remained strong enough to block
the impact of market forces) or rising in-
equality (the institutions weakened).
With relatively few countries and a great
deal of latitude in prediction, it is hard
to test institutional explanations.60
Finally, if changes in institutional
structures are central to explanations of
changes in inequality, then one must at
least consider the possibility that causa-
tion runs partially in the opposite direc-
tion. Surely some of the weakening of in-
stitutional barriers was a response to the
increased pressure brought about by
changes in market forces. If this is the
case, then it is inappropriate to treat
changes in institutional factors as exoge-
nous.
The importance of centralized wage
setting is often based on the observed
negative cross country correlation be-
tween the degree of centralization or
unionization and the trend in inequality.
Almost all countries with institutional
limits on market forces managed to have
either small increases in inequality
(France and the Nordic countries) or no
change in inequality (Germany). At the
other extreme, the United States and the
United Kingdom, two countries with de-
centralized labor markets, experienced
the largest increases in inequality.
The fact that countries with small in-
creases in earnings inequality also had
some form of institutional wage setting
does not necessarily mean that these
constraints were binding. Differences in
market forces could have been responsi-
ble for the small increase in inequality.
We, therefore, examine two additional
sources of information which may shed
light on the importance of these institu-
tions.
First, if these institutions were limit-
60
Fortin and Lemieux (forthcoming) are careful
to stipulate that they focus only on change in insti-
tutional factors.
658 Journal of Economic Literature, Vol. XXXV (June 1997)
ing prices from reaching market clearing
levels then we might expect to see an in-
crease in the relative unemployment rates
of less skilled workers. Unless unions or
public agencies were able to guarantee
employment as well as wages, workers
would find it increasingly difficult to find
jobs as the demand for their services de-
clined but wages were unable to adjust.61
Gottschalk and Joyce (1996) find some
evidence that unemployment rates of the
young did increase more than for the old
in countries with centralized labor mar-
kets and small increases in the age pre-
mium. This is consistent with the hy-
pothesis that these countries were using
the institutional constraints at their dis-
posal to limit the decline in wages of the
young. Relative wages did not fall as
much as they otherwise would have, but
this wage policy was at the cost of in-
creases in the relative unemployment
rates of the young.62
Another piece of evidence that sheds
light on the importance of institutional
constraints in limiting the rise in in-
equality comes from distinguishing be-
tween increases at the top and declines
at the bottom of the distribution. It is
commonly assumed that institutional
constraints were used primarily to pro-
tect those at the bottom. If this is the
case then we should observe relatively
small declines in the P1O/P50. For exam-
ple, the high and rising minimum wage
in France (the SMIC) should have kept
the bottom of the distribution from fall-
ing. The fact that the PIO and P50 grew
roughly equally is evidence that this con-
straint was binding. The fact that the
P90 grew faster than either indicates
that France was also experiencing a shift
in demand for high skilled workers
(OECD 1996). Similar patterns of floors
under the PIO are found for Belgium,
Finland, and Germany but not for Aus-
tralia, New Zealand, Italy, and Sweden.
In the latter countries the increases in
the P90/P10 reflect declines in the
P10/P50 as well as increases in the
P90/P50. This indirect evidence also sug-
gests that institutional constraints were
binding in some but not all countries.
Finally, some countries with coordi-
nated wage setting institutions did not
stem the tide of inequality. For example,
Australia's enactment of the accord be-
tween the government and trade unions
allowed unions to coordinate and cen-
tralize wage setting.63 This agreement
had the potential of limiting increases in
inequality as well as reducing inflation-
ary pressures. However, Australia experi-
enced a large increase in earnings in-
equality (see Table 2). This suggests that
the power to limit wage adjustments may
not have been used to offset the increase
in earnings inequality (Gregory and
Vella 1992, p. 92). However, the accord
was weakened over the 1980s and the
largest increase in inequality occurred at
the end of the decade. This could imply
that institutions were important because
inequality increased when they were
weakened (Borland 1992, p. 16). This
61
John Pencavel (1991) points to the paucity of
sound empirical studies showing the employment
impact of unions. This suggests that institutional
constraints and wages may not lead to higher un-
employment rates for less skilled workers.
62
Gottschalk and Joyce (1996) and Blau and
Kahn (1996) find evidence of both market and in-
stitutional factors limiting the wage decline for the
young in several OECD countries. Edin and
Holmund (1995) find similar increases in official
youth unemployment rates and hidden youth un-
employment (increases in enrollment in govern-
ment training programs). Stephen Nickell and
Brian Bell (1995) track changes in unemployment
rates over a longer period and find changes in the
relative unemployment rates of less educated
workers in Germany, The Netherlands, and Swe-
den which are similar in magnitude to changes in
the United States and the United Kingdom. David
Card, Francis Kramarz, and Lemieux (1995) find
changes in employment rates in France that are
similar to those in the United States.
63
By increasing noncash benefits, the accord
may have had greater impact on cash inequality
plus noncash income. See Saunders (1994).
Gottschalk and Smeeding: Cross-National Income Inequality 659
difference in focus on the level or the
change in institutions vividly illustrates
the difficulty of subjecting institutional
explanations to a rigorous test.
D. Institutional Differences and
Changes in Aggregate
Unemployment Rates
In the previous section we focused on
relative unemployment rates of different
groups to see if protected groups experi-
enced increases in relative unemploy-
ment rates. Increases in aggregate un-
employment in many OECD countries
during the 1980s has also been cited as
evidence that institutional constraints on
wage adjustments were binding.64 This
argument takes several forms. The first
is that these institutions were used to
raise average wages faster than produc-
tivity. According to this argument, the
United States opted for stagnant real
mean wages but high employment while
other OECD countries opted for wage
growth but paid a price in terms of
higher aggregate unemployment rates.
This form of the argument, has no impli-
cations for inequality because it focuses
on the mean, not the dispersion, of the
wage distribution.
The second form of the argument is
that centralized wage setting institutions
were used to counter the shift in relative
demand away from less skilled workers
(Nickell and Bell 1995). According to
this line of reasoning, aggregate unem-
ployment rates grew in countries that
were willing to accept higher unemploy-
ment rates for the least skilled in order
to keep low skilled wages from falling to
market driven levels. This interpretation
for the rise in aggregate unemployment
rates in OECD countries has two empiri-
cal implications. First, if this were the
cause for the rise in aggregate unemploy-
ment rates in OECD countries, then
those countries experiencing the smallest
increases in inequality should have expe-
rienced the largest increases in aggre-
gate unemployment rates.65 Second, the
rise in aggregate unemployment should
reflect larger increases in the unemploy-
ment rates of low skilled workers than in
the unemployment rates of more skilled
workers.
While unemployment rates in many
OECD countries did increase during the
1980s to levels similar in the United
States, the increases in aggregate unem-
ployment rates were not unusually large
in countries with more centralized labor
markets. In fact, the increases in unem-
ployment rates were much less pro-
nounced in Nordic countries than in
other OECD countries during the
1980s.66 Furthermore, the rises in aggre-
gate unemployment rates were largely
driven by increases in unemployment
rates of more skilled workers. As Chart 1
indicates, the relative unemployment
rates of the young actually declined in
almost all countries with centralized la-
bor markets.67 Youth unemployment did
become more of a problem in many
OECD countries but this reflects an in-
crease in the aggregate number of unem-
ployed workers, not an increase in the
proportion of the unemployed who were
young.
64
For alternative explanations of the rise in un-
employment rates in OECD countries, see Charles
Bean (1994) and Guiseppe Bertola and Andrea
Ichino (1995).
65
This assumes that the institutional barriers to
a decline in the wages of the least skilled did not
also operate to limit quantity adjustments. Institu-
tions such as unions (whether centralized or not)
could have bargained both on prices and qjuanti-
ties. In this case, however, institutions would not
be responsible for the higher aggregate unemploy-
ment rates in these countries.
66
OECD (1994c, Chart 1.13). The Nordic coun-
tries did experience large increases in unemploy-
ment rates in the late 1980s and early 1990s but
this would require over five year lags in the effect
of rigid wages on unemployment rates.
67See Gottschalk and Joyce (1996) and Nickell
and Bell (1995) for changes in relative unemploy-
ment rates by education as well as age.
660 Journal of Economic Literature, Vol. XXXV (June 1997)
7-
6.5
-
6- 1979
5.5
-
~ ~ ~~~~~~~~~~~~~LI1990
5
-
;4 4.5
-
4
-
,
3.5
3 3
2.5-
2
1.5-
0.5-
0
0~~~~~~~~~~~~~~
Chart 1. Youth Unemployment Rates Relative to Adults
Source: OECD (1994c Job Studies, Table 1.17, p. 43).
Thus, these aggregate data do not pro-
vide strong support for the hypothesis
that the higher unemployment rates in
OECD countries reflect the conse-
quences of distributional policies. While
careful studies of the relationship be-
tween changes in relative wages and
relative unemployment rates using micro
data may provide stronger support for this
theory, this work remains to be done.
E. Summary
The strength of current research on
changes in earnings inequality has been
to develop a set of stylized facts that any
theory must fit. It is clear that the
United States and the United Kingdom
were not the only countries to experi-
ence an increase in earnings inequality.
However, the changes in these two coun-
tries were unusually large. The challenge
is to understand why some countries
managed to escape the forces of inequal-
ity which affected the United Kingdom
and the United States to a much greater
degree. Supply shifts are clearly a part of
the explanation, including in many coun-
tries with centralized labor markets.
Gottschalk and Smeeding: Cross-National Income Inequality 661
Length of bars represents Ratio of
Lowb the gap between high and Highc High to Lowd Gini
(PlO) low income individuals (P90) (Decile Ratio) Coefficiente
Finland 1991 58 158 2.74 0.227
Sweden 1992 57 _ 159 2.78 0.229
Belgium 1992 58 * 163 2.79 0.230
Norway 1991 56 _ 158 2.80 0.230
Denmark 1992 54 _ 155 2.86 0.239
Austria 1987f 56 _
1
163 2.89 0.227
Luxembourg 1985 59 174 2.95 0.238
Germany 1984 57 171 3.01 0.249
The Netherlands 1991 57 173 3.05 0.268
Italy 1991 56 176 3.14 0.255
Switzerland 1982 54 185 3.43 0.311
France 1984 55 _ _ 193 3.48 0.294
Canada 1991 47 I _ 183 3.90 0.285
Spain 1990 49 I 198 4.02 0.306
Israel 1992 50 _
1
205 4.12 0.305
Ireland 1987 50 _ _ _
1
209 4.23 0.328
Australia 1989/90 45 f _ _ _ 193 4.30 0.308
United Kingdom 1991 44 d _ * 206 4.67 0.335
United States 1991 36 *
I_
_ 208 5.78 0.350
0 50 100 150 200 250
Averageg 53 180 3.52 0.274
Figure 2. Comparisons of Levels of Income Inequality: The Gap between Low and High Incomea
Individuals (numbers given are percent of median in each nation and Gini coefficient)
Source: Authors' calculations using the Luxembourg Income Study database.
aIncome is household disposable income per equivalent adult using an equivalence scale factor of E=0.5.
bRelative income for individuals who are lower than 90 percent of the individuals in the country and higher than
10
percent of the individuals, as a percent of national median.
cRelative income for individuals who are higher than 90 percent of the individuals in the county and lower than
10
percent of the individuals, as a percent of national median.
dRatio of 90th to 10th percentiles, or decile ratio.
eGini coefficients are based on incomes which are bottom coded at one percent of disposable personal income and top
coded at ten times the median income.
fAustria excludes self-employment income in its survey.
gSimple 19 nation average.
While institutional constraints do not
seem to have been binding in all coun-
tries, they are also clearly part of the
story. The question should not be
whether it was market forces (i.e., shifts
in supply that offset the shifts in demand)
or institutional constraints that limited the
increase in inequality. Both are clearly
necessary to explain cross-national dif-
ferences in the growth in inequality.
V. Income Inequality and Redistribution
The preceding section has docu-
mented the substantial changes in labor
markets which led to greater earnings in-
equality in the United States and many
other industrialized countries. In this
section we broaden the focus to the dis-
tribution of post-tax and transfer in-
come. How does the distribution of in-
come in the United States compare with
that in other industrialized countries?
Were changes in the distribution of labor
market earnings matched by correspond-
ingly large changes in the distribution of
family income? Did changes in taxes and
transfers cushion or exacerbate changes
in labor market incomes?
662 Journal of Economic Literature, Vol. XXXV (June 1997)
A. Relative Levels of Income Inequality
Figure 2 shows the distributions of
post-tax and transfer income in 19
OECD countries for the most recent
year available in LIS. In the United
States, a person in a household at the
tenth percentile received 36 percent of
the median income (Column 2), while a
person at the 90th percentile received
208 percent of the median (Column 4).
This results in a decile ratio of 5.78, indi-
cating that a person living in a household
at the 90th percentile enjoys over five
and three-quarters times the income of a
person at the 10th percentile.
The United States has the largest
value of the 90/10 ratio recorded in Fig-
ure 1, with the next largest being the
United Kingdom with a value of 4.67.
The lower part of the distribution of dis-
posable income appears to be substan-
tially different in the United States than
in other countries. The United States
person at PIO has 36 percent of the me-
dian, compared with values averaging 53
percent for the other nations. This dif-
ference owes in part to the relatively low
values of PIO for the United States earn-
ings distribution, shown in Table 1.
However, Canada, which has similar low
PIO values for earnings, has an 11 point
higher value for PIO (47 percent) for ad-
justed household income in Figure 2.
Thus, individual earnings distributions
may be quite different from household
income distributions.
At the top of the distribution, the
United States does not stand out to the
same extent. In Ireland, the income at
the top decile is 209 percent of the me-
dian, just about that in the United
States. The top deciles are noticeably
lower in Austria (which excludes self-em-
ployment income), Belgium, and the
Scandinavian countries.
While percentile ratios have some ob-
vious appeal (e.g., insensitivity to top
and bottom coding, ease of under-
standing), they have the disadvantages of
focusing on only two points in the distri-
bution. An alternative is to use a sum-
mary measure of inequality such as the
Gini which is shown in the final column
of Figure 2.68 While this ranking of na-
tions according to the Gini differs
slightly from that produced by the decile
ratios, there appears to be a clear group-
ing of nations. Scandinavia, Austria, and
the BENELUX countries have the least
inequality followed by central Europe,
then the Commonwealth countries, Is-
rael, and southern Europe, with the
United States, the United Kingdom, and
Ireland at the bottom.
B. Absolute Levels of Income Inequality
Because countries differ substantially
in terms of real GDP per capita, most
authors have made comparisons across
nations using relative income measures
such as those used in Figure 1. Measures
of real or absolute income differences
across nations again require comparisons
of the purchasing power of currencies
across nations. Such comparisons can be
used to test the argument that the higher
the average standard of living in a par-
ticular nation, the better off are its citi-
zens.69
Figure 3 presents the PIO, P50, and
P90 in each country measured as a
proportion of the United States median
using the same Penn World Tables and
methods used in Figure 1.70 Just as
68
Still another method would involve rank or-
derings based on Lorenz dominance. Such an or-
dering produces a very similar ranking of nations.
See Atkinson, Rainwater, and Smee ding (1995a,
Figure 4.4) for such a ranking.
69 See Atkinson (1995); Smeeding and
Gottschalk (1996); and Rainwater and Smeeding
(1995) for additional real disposable income com-
parisons across nations.
70
Excluded from Figure 3 are nations with real
median incomes below 70 percent of the United
States median (Austria, Israel, Italy, Spain, and
the United Kingdom).
Gottschalk and Smeeding: Cross-National Income Inequality 663
Ratio of Real
Low High Ratio of National
Disposable Length of bars represents Disposable High to Median To
Incomeb the gap between high and Incomec Low Incomes Real United
(PlO) low income individuals (P90) (Decile Ratio) States Median
Finland 1991 44 122 2.74 77
Sweden 1992 49 136 2.78 86
Belgium
1992 49 136 2.79 83
Norway
1991 46 128 2.80 81
Denmark 1992 48 137 2.86 89
Luxembourg
1985 48 143 2.95 83
Germany
1984 44 132 3.01 77
The Netherlands 1991 45 136 3.05 83
Italy
1991 42 132 3.14 75
Switzerland 1982 47 163 3.43 88
France 1984 40 138 3.48 72
Canada 1991 45
1
174 3.90 95
Australia 1989/90 38 _ _ 161 4.30 83
United States 1991 36 J __ _ 208 5.78 100
0 50 100 150 200 250
Averaged 44 146 3.36 84
Figure 3. Real Income Distribution Comparison
(numbers given are percent of United States median income in 1991 United States dollars)a
Source: Authors' calculations using the Luxembourg Income Study database.
aUnit of aggregation is the household and units are weighted by the number of persons in the household. Incomes are
adjusted by E = 0.5 where adjusted disposable income (DPI)=actual DPI divided by household size (s) to the power E:
Adjusted DPI = DPI/sE.
bRelative income for individuals who are below 90 percent of the individuals in the country and more affluent than 10
percent of the individuals in the country. Numbers give real income (1991 United States dollars) as a percent of the
United States median.
cRelative income for individuals who are more affluent than 90 percent of the individuals in the country and below 10
percent of the individuals in the country. Numbers give real income (1991 United States dollars) as a percent of the
United States median.
dSimple average, excluding United States.
we found for earnings, the wider distri-
bution of United States incomes means
that "low income" persons living in
households at the P10 level in the
United States had lower living standards
than did similarly situated persons in
each of the 14 other nations compared
here, despite the United States' clear
advantage at the median. For instance,
a person at the 10th percentile of
the Finnish distribution had an income
that was 58 percent of the Finnish
median (Figure 2). In terms of purchas-
ing power parity (Figure 3), the same
person has a real income which is 44
percent of the United States median.
However, while Finland's median in-
come person enjoys a standard of living
that is 77 percent that of the United
States (Figure 3), the person at the 10th
percentile of the Finnish distribution
still has a real income which is higher
than that found in the United States (44
percent versus 36 percent). In fact, Fig-
ure 3 shows nations with real median in-
comes as low as 72 percent of the United
States median, but no nation with a
lower standard of living at the 10th per-
centile. At the other end of the scale,
"high income" Americans enjoyed real
living standards far above those experi-
enced in other nations. At the P90 level,
the real income of Americans was almost
half again as high as the average incomes
of persons at the P90 point in their dis-
tribution.
664 Journal of Economic Literature, Vol. XXXV (June 1997)
.50
-
Household Incomea
.45 -
.40 - . - -- --
Adjusted Disposable Incomeb
.35
.30
1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995
Figure 4. Income Inequality in the United States: 1967-1995
(Gini Coefficient for Family Income and Adjusted Disposable Income)
Sources: U.S. Department of Commerce, Bureau of the Census (1995b, Table B-6).
allousehold income series is weighted by households (all persons sharing the same living facilities) and includes all
sources of money income, including earnings and transfer income.
bAdjusted disposable income adds food stamps and other cash income components and subtracts federal income and
payroll taxes. It is based on households as an income aggregation unit, but weighed by the number of persons in each
household, 1979-1993 only. This concept is the same as the one used in LIS. The figures are shown in Table 3.
C. Trends in Income Inequality
In this section we show how post-tax
and transfer income inequality has
changed over the past 10 to 25 years. We
start with an overview of trends in the
United States and then turn to cross-
national comparisons.
Trends in the United States. Income
inequality in the United States increased
steadily during the 1980s.71 Figure 4
shows the Gini coefficient for income
before taxes but after transfers over the
period 1967 to 1993. By this and almost
all other measures, inequality remained
relatively stable from 1967 to the mid-
1970s and then started increasing.72 Ad-
justing for household size and federal in-
come and payroll taxes, weighting by
persons, and thereby using an income
definition which is similar to that used
in the cross-national comparisons indi-
cates an even greater increase in in-
equality.73
The Gini values which underlie the
adjusted disposable income line in Fig-
ure 3 has been reproduced in Table 3
along with the corresponding percentile
points of the associated income distri-
bution. Relative incomes fell at the
71 See Jeffrey G. Williamson and Peter H. Lin-
dert (1980); Plotnick and Eugene Smolensky
(1992); and Goldin and Robert Margo (1991) for a
longer-term perspective on the United States in-
come distribution.
72
Data on inequality among families, which
goes back to 1947, shows a secular decline in in-
equality through 1970 and an increase after 1979.
Karoly and Burtless (1995) find the United States
increase robust with respect to unit of observation,
adjustments for unit size and unit of income
aggregation (weighting by persons, households or
families).
73 The trend in post-tax and transfer disposable
income can go back only to 1979 due to data re-
strictions. See also Karoly (1995) for a similar
trend in equivalence-adjusted family income from
1974-1993.
Gottschalk and Smeeding: Cross-National Income Inequality 665
TABLE 3
TRENDS IN UNITED STATES INCOME INEQUALITY: 1979-1993
PERCENTILES OF ADJUSTED DISPOSABLE PERSONAL INCOME
Relative
Year PIO/P50 P20/P50 P80/P50 P90/P50 P90/PLO P80/P20 Gini
1979 40.4 57.3 154.6 190.2 4.71 2.70 0.313
1980
39.9 56.7 154.2 189.7 4.75 2.72 0.310
1981
39.2 55.6 155.2 192.7 4.92 2.79 0.318
1982
37.7 54.3 159.0 200.2 5.31 2.93 0.331
1983 36.7 53.1 161.8 203.3 5.55 3.05 0.339
1984 36.5 53.1 162.2 204.4 5.60 3.06 0.340
1985 36.5 53.3 162.9 205.0 5.61 3.05 0.342
1986
35.5 52.9 162.6 204.7 5.77 3.07 0.341
1987 34.8 52.3 161.4 201.2 5.78 3.09 0.342
1988
35.1 52.5 162.4 205.0 5.85 3.10 0.347
1989
35.8 52.6 162.1 205.9 5.75 3.08 0.351
1990
35.9 53.1 163.6 207.6 5.79 3.08 0.352
1991
35.5 52.8 162.9 207.7 5.79 3.09 0.350
1992 34.7 52.0 164.6 209.2 6.03 3.16 0.357
1993a
34.4 51.9 167.9 214.1 6.22 3.23 0.363
1991/1979*100 86 92 109 113 131 119 116
Source: U.S. Department of Commerce, Bureau of Census (1995b).
a
1993 income is topcoded at the 1983-1991 level of $299,000 per household and reflects population weights
from the 1990 census.
bottom (P10) by about as much as they
rose at the top (P90) over this period. As
a result, the decile ratio rose by more
than 30 percent from 1979 to 1993,
while the Gini value rose by 16 percent
over the period.
While household income inequality in
the United States rose over this period,
the effects of the 1981-82 and 1990-91
recessions hastened the trend toward
greater inequality. What is unusual by
historical standards is that inequality
grew during the 1983-89 recovery as
well. (See also Burkhauser et al., 1996,
on this point.)
Trends in OECD Countries. The avail-
able empirical evidence concerning
recent trends in income inequality in
different nations is summarized in
Table 4.74 Countries are listed in order
of changes in disposable income inequal-
ity (as measured by the change in the
Gini coefficient) from largest to smallest
change.
The largest changes in income distri-
bution took place in the United King-
74
Figures
in Table 4 are based on Appendix Ta-
ble B and are not comparable across countries be-
cause they come from a wide variety of studies.
Differences in data, concepts of income, or
method of calculation may affect the measure of
the level of inequality. These differences, which
are explained in the notes to Appendix Table B,
are less likely to affect trends. However, they do
limit comparisons of percentage change in Ginis
because different weighting patterns and different
equivalence scales will produce different absolute
values for the Gini within and across nations. See
also Atkinson, Rainwater, and Smeeding (1995a,
Ch. 4) on this point.
666 Journal of Economic Literature, Vol. XXXV (June 1997)
TABLE 4
CHANGES IN MARKET AND DISPOSABLE INCOME INEQUALITYa
Years Market Income Disposable Income
Country Source Change Inequalityb Inequality
United Kingdom Goodman and Webb (1994) 1981-91
+++ ++++
Atkinson (1993)
United States U.S. Bureau of the Census (1995A) 1980-93 +++ +++
Sweden Gustafsson and Palmer (1993) 1980-93 +++ +++
Statistics Sweden (1995)
Australia Saunders (1994) 1980-81 + +
1989-90
Denmark Aaberge et al. (1995) 1981-90 + +
New Zealand Saunders (1994) 1981-89 + +
Japan Tachabanaki and Yagi (1995) 1981-90 + +
Bauer and Mason (1992)
The Nether- Atkinson, Rainwater, and Smeeding 1981-89 + +
lands (1995a)
Muffels and Nelisen (1996)
Norway Epland (1992) 1982-89 + +
Belgium Cantillon et al. (1994) 1985-92 + +
Canada Beach and Slotsve (1994) 1980-92 + 0
Statistics Canada (1994)
Israel LIS (1995) 1979-92 + 0
Finland Uusitalo (1995) 1981-92 +++ 0
France Concialdi (1996) 1979-89 0 0
Portugal Rodrigues (1993) 1980-90 0 0
Spain LIS (1995) 1980-90 na 0
Ireland Callan and Nolan (1993) 1980-87 + 0
West Germany Burkhauser and Poupore (1997) 1983-90 + 0
Hauser and Becker (1993)
Italy Brandolini and Sestito (1993) 1977-91
Eriksson and Ichino
(1995)
See source notes at end of Appendix Table B.
a
Degree of change is based on Appendix Table B and is coded as follows:
Designation Interpretation Range of Change in Gini
small decline -5 percent or more
0 zero -4 to +4 percent
+ small increase 5 to 10 percent
++ moderate increase 10 to 15 percent
+++ large increase 16 to 29 percent
++++ extremely large increase 30 percent or more
b
Most studies show changes in market income inequality, while still others do not discuss market income
changes at all. The latter are marked "na."
Gottschalk and Smeeding: Cross-National Income Inequality 667
dom and in the United States, where
there has been a clear trend toward
greater inequality. Rising earnings in-
equality among men and among two-
earner families, and the growth in the
number of single individuals and single
female headed families were the primary
factors accounting for the increase in in-
equality in the United States since the
mid-1970s. In the United Kingdom, ris-
ing unemployment and higher numbers
of single parents were important in
building a large group at the bottom of
the distribution, while higher earnings
for well-educated men and women, in-
creased capital income, and self-employ-
ment income were all-important in ex-
plaining the growing income share at the
top. (For the United States see Karoly
1995; Danziger and Gottschalk 1995; and
Duncan, Smeeding, and Willard Rodgers
1994; and for the United Kingdom, see
Jenkins 1995a, 1996.)
While the trends in earnings inequality
and in income distribution were similar
in the United States and the United
Kingdom, the degree of change in the
distribution of family income was mark-
edly different. In the United States the
largest increases in inequality were con-
centrated in the early 1980s and contin-
ued into the early 1990s. In the United
Kingdom income inequality fell through
the mid-1970s but the Gini coefficient
rose by more than 30 percent between
1978 and 1991. This is almost double the
increase over a similar period in the
United States, and more than double the
decline in the United Kingdom from
1949 to 1976. (Compare Karoly 1995 to
Atkinson 1996a, Table 1.)
While starting from a much lower
level of inequality, Sweden experienced
a pattern of change in inequality similar
to that in the United Kingdom, down-
ward until 1981, then upward in the
1980s, with the sharpest increases in the
early 1990s. The Swedish Gini increased
by about 20 percent between 1981 to
1993, though the Swedish income distri-
bution remained considerably more
equal than either the United States or
the United Kingdom in spite of these
changes.75 In Australia, Denmark, and
Japan, the upward trend over the 1980s
was slightly less than that experienced in
the United States and Sweden. The same
is true in New Zealand, though all of
these increases came during the late
1980s (Saunders 1994).
Perhaps Atkinson (1996a, p. 43) sums
it up best:
Among the other (non-United Kingdom)
OECD countries, it is certainly wrong to
think in terms of a world-wide trend towards
increased income inequality in the 1980s: the
upward trend was exhibited to differing de-
grees in different countries, and was not to
be found in some countries. At the same
time, those seeking to identify a common pat-
tern for OECD countries other than the
United Kingdom and the United States could
say that continuing progression towards re-
duced inequality was the exception rather
than the rule. Moreover, it may be that these
countries are lagging behind the United
States and the United Kingdom, and that the
1990s will see a rise in income inequality
more generally.
Clearly the remaining evidence in Ta-
ble 4 supports this assertion. Inequality
rose only slightly in three nations (The
Netherlands, Norway, and Belgium), and
eight other countries show no change in
inequality in the 1980s. Only in Italy do
we find a noticeable decrease in inequal-
ity during the 1977-1991 period.
It is also noteworthy that there ap-
75Bjorklund and Freeman (1994) find little in-
crease in inequality among non-aged families with
children over this period. However, they compute
only subgroup inequality trends, excluding the
aged and persons aged 18 and 19. Were we to cal-
culate absolute changes in inequality as measured
by the Ginis, the Swedish increase would be less
in absolute terms than that found in the United
States or in the United Kingdom. Also, the trend
toward greater inequality in Sweden may have
peaked in 1991 and has receded slightly since that
time. See Appendix Table B.
668 Journal of Economic Literature, Vol. XXXV (June 1997)
pears to be no apparent relation between
the trend over the 1980s and the overall
level of inequality at the start of the pe-
riod. Inequality increased both in the
United States, with a high level of in-
equality even before the increase, and in
Sweden, which started from a much
lower level of inequality. Inequality fell
in Italy and rose in the United Kingdom,
though both occupied, intermediate po-
sitions in the mid-1980s (see Atkinson,
Rainwater, and Smeeding 1995a).
Nor is there a consistent country
group story. Among the Scandinavian na-
tions, Sweden experienced a rapid rise in
inequality in the early 1990s, while Fin-
land did not. In Europe we find large
secular increases in inequality in the
United Kingdom, smaller increases in
Denmark, Belgium, and The Nether-
lands, but stasis in Germany, Portugal,
Ireland, and France, with a secular de-
crease in Italy. Canada experienced only
mild increases in inequality of family in-
come while the United States experi-
enced much larger increases, despite
similar changes in earnings inequality
(Card and Freeman 1993).
D. Accounting for the Changes
The changes in the distribution of
family income distribution that we have
documented are a product of a compli-
cated set of forces: changes in labor mar-
kets that affect earnings of individual
family members; changes in returns to
capital; demographic changes, such as
the aging of the population and growth
of single parent households, which affect
both family needs and labor market
decisions; changes in social norms, such
as the women's movement and the
purported decline in the work ethic
among men, which may have affected
demographic and labor market prefer-
ences; and policy changes in tax and
transfer programs which not only af-
fected family income directly but also
may have affected work and investment
decisions.
The inclusion of multiple income
sources received by multiple individuals
thwarts attempts to identify the causal
links that led to variations across time
and across countries in the distribution
of total post-tax and transfer family in-
come. There is ample evidence that fam-
ily members take account of all sources
of income available to the family in de-
ciding not only how much each member
might work, but also how to structure liv-
ing arrangements. Moreover, govern-
ments themselves react differently to
market income changes via changes in
redistribution (tax and transfer) policy,
and via other policies (e.g., government
employment).
Aggregating earnings across all indi-
viduals in a household and adding other
sources of income takes us from the dis-
tribution of individual earnings to the
distribution of family income. Ideally
one would like to know how much of the
change in inequality of total family in-
come is caused by exogenous changes in
each source of income. This would re-
quire a fully articulated model of behav-
ioral responses. For example, if exoge-
nous increases in inequality of male
earnings led wives of low income hus-
bands to work more, then this portion of
the change in overall inequality would be
caused by changes in the distribution of
husbands' earnings, not wives' earnings.
Structural models that include all behav-
ioral links are well beyond the scope of
existing empirical work. Researchers
have, therefore, limited themselves
largely to purely accounting exercises
which decompose changes in overall in-
equality into a set of component parts
that may reflect endogenous as well as
exogenous changes.76
76
The primary drawback of accounting exercises
is that they can easily be misinterpreted because
they do not make a distinction between endoge-
Gottschalk and Smeeding: Cross-National Income Inequality 669
While accounting decompositions can
potentially offer insights into the pat-
terns of changes in inequality, these
methods also raise a set of conceptual
and measurement issues. Most account-
ing decompositions of income by source
are based on identities between inequal-
ity of total income and three attributes
of the joint distribution of the compo-
nent sources: (1) inequality of each mar-
ginal distribution, (2) correlations (or
some other measures of covariance) be-
tween income sources, and (3) the rela-
tive size of each source.77 For example,
the Gini coefficient for total income can
be written as the sum of the products of
the Gini coefficients for each source, the
Gini correlation between the source and
total income, and the share of the total
from each source.78 Alternatively the co-
efficient of variation squared, CV2, of to-
tal income can be written in terms of the
CV2 of each source, the correlations be-
tween all sources, and the share of the
total from each source.
If inequality of a particular source in-
creases then it is easy to attribute the
resulting increase in overall inequality to
that source. However, it is not obvious
how to classify the effects of changes in
the correlations among sources or rela-
tive sizes of each source, because these
factors inherently affect two or more
sources. Thus, while identities allow the
total to be decomposed into parts, it is
often not obvious how to go from this to
a meaningful accounting of the sources
of the change in inequality, even over-
looking the problems caused by behav-
ioral links. Furthermore, these problem-
atic decisions can often lead to very
different conclusions (Maria Cancian
and Deborah Reed, forthcoming). For
example, Karoly and Burtless (1995) and
Karoly (1995) attribute much of the rise
in family inequality to changes in wives'
earnings, while Cancian, Danziger, and
Gottschalk (1993) conclude that most of
the increase in family income inequality
reflects increases in male earnings in-
equality and that changes in the distri-
bution of wives' earnings played a more
modest role.79 This difference in inter-
pretation partially reflects differences in
the ways in which changes in correla-
tions between wives' earnings and other
sources and changes in shares of income
coming from wives' earnings are treated
in these decompositions.
While it is not clear how to apportion
the individual pieces to a specific source,
accounting identities do allow us to iso-
late the pieces.80 Several important styl-
nous and exogenous factors. For example, to say
that changes in mean husbands' earnings account
for X percent of the change in mean family in-
come does not imply that family income would
have dropped by that percentage if husbands'
earnings ad not changed. Other sources could
have responded to the decline in husbands' earn-
ings.
77Other decompositions focus on population
subgroups, for example, how much of the increase
in family income inequality occurs among families
with the same work status of head and how much
comes from differences in means across family
types (see Jenkins 1995a). These decompositions
also make no distinction between endogenous and
exogenous forces.
78
Lerman and Shlomo Yitzhaki (1985) show the
Gini coefficient is a weighted average of Ginis of
individual sources with weights that depend on
the correlation and shares. The overall Gini can be
smaller than the Gini of each source, even if the
correlation is positive. See Anthony Shorrocks
(1982) for an early discussion of the conceptual
issues.
79This study and a similar one for the United
Kingdom by Steven Harkness, Ste phen Machine,
and Jane Waldfogel (1996) refer only to inequality
among married couple families, not the entire
population. Because the impact of wives' earnings
may affect inequality among married couple fami-
lies differently from inequality as a whole, it is
difficult to draw inferences for the entire popula-
tion from these studies.
80
Comparing decompositions across studies of
different countries is further hampered by differ-
ences in measures of inequality, which make it im-
possible to impose a consistent (though arbitrary)
method of decomposing changes in inequality.
Cancian and Robert Schoeni (1992) use consistent
measures across a variety of countries.
670 Journal of Economic Literature, Vol. XXXV (June 1997)
ized facts about the individual pieces
stand out in the literature. Wives' earn-
ings have become an increasingly large
proportion of family earnings but wives'
earnings are only weakly correlated with
husbands' earnings. This weak correla-
tion in annual earnings reflects the nega-
tive correlation between the labor supply
of wives and husbands' earnings, which
partially offsets the high correlation in
wages. While the correlation in earnings
between spouses is low, it has increased
in the United States. However, in spite
of the positive and rising correlation in
spouses' earnings, family earnings are
more equally distributed than husbands'
earnings alone. Thus, if the difference
between the distributions with and with-
out wives' earnings is taken as a measure
of wives' contribution to inequality, then
wives' earnings equalize the level of in-
equality in the United States, while they
are disequalizing with respect to the
trend in inequality.
While the overall tax and transfer sys-
tem in the United States is progressive,
changes in taxes and transfers during the
1980s reduced progressivity.81 However,
changes in taxes and transfers account
for only a small part of the trend in in-
equality during the 1980s and early
1990s. While the real value of unemploy-
ment compensation, welfare benefits and
other cash transfers aimed at the poor in
the United States fell relative to GDP
from 1980 to 1990, this can account for
only a small proportion of the trend in
post-tax and transfer inequality (OECD
1994a, Table lc and Chart 1). Changes
in taxes also account for little of the
trend in inequality in the United States.
This might be expected because lower
marginal tax rates at the top of the distri-
bution were offset by a higher zero
bracket amount and higher personal
income tax exemptions which helped
the working poor after the 1986 tax re-
forms. Furthermore, increases in the
Earned Income Tax Credit during the
1980s and into the 1990s raised the post-
tax earnings at the bottom of the distri-
bution. These changes were, however,
much smaller than the impact of the in-
crease in earnings inequality in the
United States (Edward Gramlich, Rich-
ard Kasten, and Frank Sammartino
1993).
E. Factors Associated with Changes
in Other OECD Countries
Table 4 contrasts the trends in in-
equality in market income and dispos-
able income in the United States with
the experiences of a number of OECD
countries. Market income includes the
earnings of all persons in the household
and all income from interest, dividends,
rents and other market sources. Because
disposable income is equal to market in-
come plus transfers minus taxes, taxes
and transfers have two effects. They lead
to behavioral adjustments in labor supply
that may affect market income inequality
and they add (or subtract) income to
yield the distribution of disposable in-
come.
Because earnings constitute the major-
ity of market income for most house-
holds and because earnings among
family members tend to be positively
correlated, it should come as no surprise
that our ranking of trends in market in-
come in Table 4 closely mirrors the rank-
ing on the basis of individual earnings in
Table 2, though not completely so.
France, The Netherlands, Belgium, Nor-
way, and Portugal had small increases in
the dispersion of market income as well
as individual earnings, at least until
1990. But not all nations followed this
pattern. For instance, overall earnings
inequality in Canada increased less than
81
The difference between the pre- and post-fisc
distributions is, however, small compared to other
modern nations. See Smeeding and Coder (1995).
Gottschalk and Smeeding: Cross-National Income Inequality 671
the distribution of individual earnings
(Charles Beach and George Slotsve
1994).
Increased receipt of capital income
(including deferred capital income from
private pensions) and a growing correla-
tion between high capital income and
high earnings acted to increase market
income inequality in the 1980s in the
United Kingdom, Japan, Australia, and
New Zealand. However, this factor was
not nearly as important as changes in
earned income inequality in any of these
countries.82
Demographic and social change also
played a role in accounting for the rise in
inequality in OECD countries since
1970 though the relative importance of
these changes is still unsettled. Most
find the role of demographic factors to
be smaller than economic factors
(Jenkins 1995a; Jantti and Danziger
1994; Fritzell 1993; Danziger and
Gottschalk 1995). Burtless and Karoly
(1995) and Lerman (1996) attributed a
larger role to demograhic and social fac-
tors than do others.83 The aging of the
population and policies that have en-
couraged early retirement helped reduce
adjusted income inequality in many ad-
vanced countries because the level of ad-
justed income inequality among the aged
is generally less than that found among
the non-aged.84
Demographic change during the 1980s
also led to a sharp increase in the frac-
tion of single parent families. In Ger-
many, the United Kingdom, and The
Netherlands, these changes were par-
ticularly large.85 Because single parent
families have low average income, this
demographic shift served to increase in-
equality.86
In summary, changes in earned in-
come inequality appear to be the prime
force behind changes in market income
during the 1980s in most countries. With
earnings more than 70 percent of market
income, it should not be surprising that
increased individual earning inequality
and other changes in earnings within
the household would be important fac-
tors in accounting for change in in-
come inequality. Other market forces
(such as capital income) and demo-
graphic changes also affected market in-
come inequality, though to a lesser de-
gree.
But market income changes and
demographic factors do not tell the
whole story. More than 25 percent of
all households in major OECD nations
82On Australia and New Zealand, see Saunders
(1994); on the United Kingdom, see Atkinson
(1996a) and Jenkins (1995a); on Japan, see John
Bauer and Andrew Mason (1992); and for the
United States, see Cowell and Jenkins (1993) and
Duncan, Smeeding, and Rogers (1994). Part of the
reason why annual income inequality measures do
not permit a greater role for changes in capital
income is because they report only realized inter-
est, rents, and dividends received, ignoring inter-
est paid and both realized and unrealized capital
gains. Because of nonrealization and deferral of
most asset income, annual income statistics ignore
most changes in net worth and thus true capital
income (or loss).
83
The importance of demographic change in
the United States is larger if we limit our
analysis
to working age families or to families with chil-
dren. But even then, they account for less than
half of the difference between 1971 and 1989 with
most of their effect coming during the 1970s (Ler-
man 1996; Karoly and Burtless 1995). The in-
creased level and correlation of women's earnings
with men's earnings accounts for a large fraction
of the change in family income inequality during
the 1980s (Karoly and Burtless 1995).
84
Exceptions are the United States and Ger-
many. See Smeeding, Rainwater, and Torrey
(1993); and Atkinson and Sutherland (1993).
85
The percentage of single parent families with
at least one child under 15 out of all families with
children under 15 rose from 9.4 percent to 14.6
percent in Belgium, 9.8 to 15.4 percent in Ger-
many, 7.9 to 12.2 percent in The Netherlands, and
13.7 to 19.0 percent in the United Kingdom from
1981-82 to 1990-91. European Commission (1995).
86
Births out-of-wedlock also rose in these coun-
tries. See the European Commission (1995). How-
ever, out-of-wedlock birth does not necessarily in-
dicate low income in countries such as Sweden
where many high income parents live together for
long periods outside of marriage.
672 Journal of Economic Literature, Vol. XXXV (June 1997)
16
-
14
-1979
12 ~~~~~~~~~~~~~~~~~~~~~~1990
12
-
~~~~~~~~~~~~~~~~D1990-91
10
8-
6-
4-
2
0
f
Chart 2. Expenditures on Social Programs Among the Non-aged as Percentage of GDP in 1980, 1985,
and 1990-91
Source: OECD (1994b, Tables lb, Ic)
Note: These include cash benefits for disability and disability services, employment promotion benefits, unemployment
compensation, family allowances, welfare benefits, and other miscellaneous items. Excludes all cash benefits to the aged
and survivors, health benefits, and education benefits.
depend on something other than earn-
ings as the primary source of their
gross incomes. In nations such as the
United Kingdom, The Netherlands, and
Sweden, this figure reaches 30 percent
of income (Atkinson, Rainwater, and
Smeeding 1995b). Countries differed
dramatically both in the amount of so-
cial protection they offered working
families at the beginning of the 1980s
and the changes in expenditures on
these programs.87 Chart 2, which shows
public cash expenditures on social
protection for the non-aged as a percent-
age of GDP in 1980, 1985, and 1990-91,
illustrates the diversity of experi-
ences.88 Countries are ranked according
to spending in 1980. While these expen-
ditures do not cover all forms of trans-
fers to the non-aged population, they
show the same general patterns that
would be found using alternative defini-
tions. Sweden, The Netherlands, Den-
mark, and Finland all spent 10 percent
or more of the GDP on social protection
for the non-aged in 1980, and increased
their expenditures between 1980 and
1990. While Norway spent less than 10
87This includes legislated discretionary changes
and the automatic
response
to changing market in-
come circumstances of households.
88
Social protection is a classification used by
the OECD. It includes disability and disability
services, employment promotion benefits, unem-
loyment
compensation, family allowances, wel-
fare benefits, and other miscellaneous items. So-
cial protection in Chart 2 excludes all cash
benefits to the aged and survivors, health benefits,
and education benefits.
Gottschalk and Smeeding: Cross-National Income Inequality 673
percent of GDP on these programs in
1980, this fraction had risen to 14 per-
cent by 1990. In contrast, Japan spent
only 2.4 percent of GDP on these pro-
grams in 1980 and even less in 1990.
Likewise, expenditures on these pro-
grams fell from only 4.5 percent of GDP
in 1980 in the United States and from
3.4 percent in Italy. Thus, both the level
and trends in expenditures varied widely
across countries. The Nordic and north-
ern European countries, which had the
lowest levels of inequality and then some
of the smallest increases in income in-
equality, were also the countries with the
greatest social protection.
The growth in transfers during the
1980s partially reflects increased take-up
rates as many of these countries experi-
enced greater demands on social protec-
tion programs as a result of widening in-
equality of market income. In fact, in the
more "activist" European and Nordic so-
cial welfare states, social expenditure
trends in the 1980s can be better de-
scribed as adaptions to changing circum-
stances than as alterations in the basic
systems of social protection (Niels Ploug
and Jon Kvist 1994; European Commis-
sion 1994; Karen Gardiner 1993). While
limitations on some types of social insur-
ance benefits (e.g., unemployment, dis-
ability) were introduced, and indexation
formulae were made less generous in
some European and Scandinavian coun-
tries, there were also increases in family
benefits and welfare benefits for the
long-term unemployed and for single
parents in Australia, Denmark, France,
Germany, Norway, and Finland.
Perhaps even more important were
changes in the composition of spending
and its effectiveness in replacing lost
market income due to unemployment
and disability.89 For instance, in nations
such as Canada and Finland, generous
long-term unemployment benefits sig-
nificantly dampened the effects of
higher unemployment on disposable in-
come inequality. As a result of increased
take-up rates and other policy changes,
the decade ended with the vast majority
of countries spending more on social
protection programs than ten years ear-
lier.
While the level of social spending is
negatively correlated with changes in in-
come inequality, there is little relation-
ship between retrenchment and in-
creases in inequality in most countries.
This undoubtedly reflects the fact that
some countries that reduced their expen-
ditures on the non-aged (Belgium, Ger-
many, and Italy) experienced few new
demands in their programs because in-
equality of market income grew only
modestly. Some of the nations with
small- to medium-sized social protection
systems whose transfer systems automat-
ically reacted to the rising tide of market
income inequality with higher outlays
(Australia and Ireland) were unable to
stem that tide. And in two nations (the
United Kingdom and New Zealand), re-
ductions in benefit levels for the non-
aged helped to exacerbate inequality,
even though overall social expenditures
increased in both nations during the
1980s (Atkinson 1993; Jenkins 1996;
Hills 1995).
There were equally large changes in
tax policies during the 1980s. The lower-
ing of top income tax rates was not lim-
ited to the United States. The top in-
come tax rates were cut in 26 of the 28
industrialized countries surveyed in
Messere (1993). These reductions were
not only widespread but large in many of
these countries. Top federal income tax
rates fell from 50 to 28 percent in the
89
See Gottschalk, Gustafsson, and Edward Pal-
mer (1996); OECD (1994a); Gardiner (1993);
Ploug and Kvist (1994); and the European Com-
mission (1994). All deal with this and similar is-
sues.
674 Journal of Economic Literature, Vol. XXXV (June 1997)
45-
* Decile 10 early 1980s
40 Decile 10 late 1980s
1
Decile 2 early 1980s
35
Decile 2 late 1980s
30
-
25-
20
15-
10
5-
0 --
0 45 C ?000)
-$0
C;0
0 ()o
E0
Chart 3. Average Tax Rates for Households in the Second and Tenth Deciles in Selected OECD Countries
Source: Authors' calculations using the Luxembourg Income Study database.
Note: Average tax rates are total income and employee payroll tax as a percentage of gross cash income for households
ranked by household disposable income adjusted for family size (E=0.5).
United States, 70 to 40 percent in the
United Kingdom, 48 to 14 percent in
Norway, and from 78 to 50 percent in
Sweden, though some rose again by
small amounts (e.g., from 28 to 32 per-
cent in the United States in 1993, and 50
to 55 percent in Sweden in 1993). Addi-
tional tax progressivity was introduced
by changes such as the family benefit in
the United Kingdom and the Earned In-
come Tax Credit in the United States,
and taxable income definitions were
broadened in many nations.
Reductions in the top marginal tax
rates did not necessarily lead to declines
in taxes collected on families at the top
of the distribution. Many of these fami-
lies paid higher taxes as a result of in-
creased income subject to tax and an in-
crease in other taxes that make up for
income tax reductions. In fact, overall
tax revenues rose in most OECD coun-
tries, owing mainly to increased payroll
taxes for social retirement, disability and
health care, increased VAT for general
revenue, and increased employment-re-
lated taxes levied on employers to cover
higher unemployment outlays (Messere
1993; OECD 1994).
Chart 3 shows the average federal in-
come and payroll tax rates paid by fami-
lies in the top and second decile groups
of the distribution of disposable income
in the early 1980s and late 1980s to early
Gottschalk and Smeeding: Cross-National Income Inequality 675
1990s. These figures show that in most
countries (The Netherlands, Sweden,
West Germany, and Canada), average tax
rates increased for families at both ends
of the distribution, with the larger in-
creases occurring at the top. In Austra-
lia, France, and the United Kingdom, av-
erage tax rates increased at the top and
decreased at the bottom. It is only in the
United States and Norway that average
federal tax rates declined at the top and
increased at the bottom of the distri-
bution.
Our reading of the limited cross-
national information on changes in tax
and transfer structures is that changes in
taxes paid and transfers received were
largely offsetting to the changes in the
distribution of pre-tax and transfer in-
comes. This would occur automatically
in countries with progressive tax and
transfer systems. How much of these
changes came from explicit policy
changes as compared to changes in eco-
nomic behavior of households is an im-
portant question that remains to be an-
swered.90 The links between changes in
tax and transfer policy and changes in
the distribution of disposable income in
different countries are certainly not well
understood at this stage.
In nations with weak safety nets and
less activist governments, changes in
market incomes were dominant. Here,
the United States, Australia, and Japan
stand out as the three best examples. In
the United Kingdom there appears to
have been such a massive change in mar-
ket income inequality that the British tax
and transfer system was not able to over-
come these forces and may even have
contributed to them. While income and
other tax changes have benefitted the
well-to-do in a small number of coun-
tries, means tested benefits were in-
creased in some nations to cushion re-
ductions in other types of benefits (e.g.,
unemployment), producing some offset
to the disequalizing trend in market in-
come, but also reducing work effort.9'
And there are clearly exceptions. Fin-
land, Canada, and Norway experienced
smaller increases in inequality of dispos-
able income than would be suggested
by their changes in inequality of earn-
ings and market incomes, and Sweden
and New Zealand experienced large in-
creases compared to the small increase
in earnings inequality.
A pressing area for future research is
to isolate the impact of changes in tax
and transfer policies on the distribution
of family income. This will require an
explicit model of the endogenous in-
creases in transfers that accompany de-
clines in earnings at the bottom of the
distribution and an explicit model of the
impact of changes in tax and transfers
on the distribution of pre-tax income.
Only then will it be possible to isolate
the relative importance of exogenous
changes in the distribution of pre-tax
income from both exogenous and endo-
genous changes in taxes and transfers.
V. Summary and Conclusions
Concerns about earnings inequality
and joblessness have moved to the top of
the social agenda in many OECD coun-
tries. The growing internationalization of
the economy and labor market and gov-
ernment reactions to social and eco-
nomic issues such as population aging,
divorce, and increased female labor
force participation rate, have added to
our interest in how successful different
90
See Auerbach and Slemrod (1997) for a sur-
vey of behavioral effects of the United States 1986
Tax Reform.
91See Paul Johnson and Steven Webb (1993)
and Gardiner (1993). It may also be true that ex-
pansion of the safety net for single parents pro-
duced a marked decline in their labor market ac-
tivity. See Moffitt (1992) for a review of United
States evidence on this topic.
676 Journal
of
Economic
Literature,
Vol. XXXV
(June
1997)
economies are in dealing with these is-
sues.
Over the past decade, new data re-
sources have expanded to meet these in-
terests. Much has been learned from
studies of annual cross-sectional house-
hold income microdata. New frontiers
will include increased usage of national
household income panel data sets which
will follow the same individuals over
longer time horizons, and greater usage
of cross-national labor force surveys and
surveys that focus on expenditures and
wealth.
Yet, while great strides have been
made to provide a factual basis for cross-
national studies, much less progress has
been made in providing a tight theoreti-
cal framework to analyze these data. Bet-
ter structural models of income distri-
bution and redistribution that can be
applied across nations are badly needed.
Ideally, an overall framework would si-
multaneously model the generation of all
sources of income (labor income, capital
income, private transfers, public trans-
fers, and all forms of taxation) as well as
the formation of income sharing units.
While most of the components of such a
model were identified as early as the
mid-1960s, our progress toward building
such a model has been slow (James
Meade 1964). Atkinson (1996b) has
made a first step at one component of
such a model and has hinted at other
components. If we are to understand
why we observe the extent and pattern
of inequality levels and trends that are
extant in this review, an overall concep-
tual framework with empirically testable
components is the next big step that
must be taken.
REFERENCES
AABERGE, ROLF ET AL. Income inequality and in-
come mobility in the Scandinavian countries
compared to the United States. Turku, Finland:
Abo Akademi U., 1995.
AARON, HENRY J. Politics and the professors: The
great society in perspective. Washington, DC:
Brookings Institution, 1978.
ABRAHAM, KATHARINE G. AND HOUSEMAN,
SUSAN N. "Earnings Inequality in Germany," in
RICHARD FREEMAN AND LAWRENCE KATZ,
eds. 1995, pp. 371-404.
ALESINA, ALBERTO AND PEROTTI, ROBERTO.
"The Political Economy of Growth: A Critical
Survey of the Recent Literature," World Bank
Econ. Rev., Sept. 1994, 8(3) pp. 351-71.
ATKINSON, ANTHONY B. "On the Measurement of
Inequality," J. Econ. Theory, Sept. 1970, 2, pp.
244-63.
. "What is Happening to the Distribution
of Income in the U.K.?" Proceedings of the
British Acad., 1993, 82, pp. 317-51.
. "Income Distribution in Europe and the
United States." Luxembourg Income Study
Working Paper No. 133, Oct. 1995.
. "Explaining the Distribution of Income,"
in New inequalities. Ed.: JOHN HILLS. Cam-
bridge: Cambridge U. Press, 1996, pp. 19-48.
. "Bringing Income Distribution in from
the Cold" Econ. J., Mar. 1997.
ATKINSON, ANTHONY B. AND MICKLEWRIGHT,
JOHN B. Economic transformation in Eastern
Europe and the distribution of income. Cam-
bridge, Eng: Cambridge U. Press, 1992.
ATKINSON, ANTHONY B.; RAINWATER, LEE AND
SMEEDING, TIMOTHY M. Income distribution
in OECD countries: Evidence from the Luxem-
bourg Income Study (LIS). Paris: OECD,
1995a.
. "Income Distribution in European Coun-
tries," in Welfare and inequality: Essay on Bri-
tain and Europe. By ANTHONY B. ATKINSON.
Cambridge: Cambridge U. Press, 1995b, pp.
41-63.
ATKINSON, ANTHONY B. AND SUTHERLAND,
HOLLY. "Two Nations in Early Retirement?
The Case of Britain," in Age, work, and soci al
security. Eds.: ANTHONY B. ATKINSON AND
MARTIN REIN. London: Macmillan, 1993, pp.
132-60.
AUERBACH, ALAN AND SLEMROD, JOEL. "The
Economic Effects of the Tax Re form Act of
1986," J. Econ. Lit., June 1997, 35(2), pp. 589-
632.
BAILEY, THOMAS. "Education and the Transfor-
mation of Markets and Technology in the Tex-
tile Industry." NCES Paper No. 2, Teachers
College. New York: Columbia U., 1988.
BAUER, JOHN AND MASON, ANDREW. "The Dis-
tribution of Income and Wealth in Japan,"
Rev. Income Wealth, Dec. 1992, 38(4), pp. 403-
428.
BEACH, CHARLES M. AND SLOTSVE, GEORGE A.
"Are We Becoming Two Societies? Income Po-
larization and the Middle Class in Canada."
Queens U. mimeo, Nov., 1994.
BEAN, CHARLES R. "European Employment: A
Survey" in J. Econ. Lit., June 1994, 32(2), pp.
573-619.
BEGUE, JEAN. "Remarques sur une Etude de
Gottschalk and Smeeding: Cross-National Income Inequality 677
l'OCDE Concernant la Repartition des Re-
venus dans Divers Pays," Economie et Statis-
tique, 1976, 84, pp. 97-104.
BELL, CLIVE AND ROSENBERG, CHRISTOPH B.
"Combining Consistency with Simplicity
When Estimating Tax Incidence: Alternative
Assumptions and Findings for Three Coun-
tries." Luxembourg Income Study Working Pa-
per No. 102. Walferdange, Luxembourg, Sept.,
1993.
BERMAN, ELI; BOUND, JOHN AND GRILICHES,
ZVI. "Changes in the Demand for Skilled Labor
within U.S. Manufacturing Industries: Evidence
from the Annual Survey of Manufacturing."
NBER Working Paper No. 4255. Cambridge,
MA: National Bureau of Economic Research,
Jan., 1993.
BERTOLA, GIUSEPPE AND ICHINO, ANDREA.
"Wage Inequality and Unemployment: United
States vs. Europe," NBER Macroeconomics An-
nual, 1995, pp. 13-66.
BJORKLUND, ANDERS AND FREEMAN, RICHARD
B. "Generating Equality and Eliminating Pov-
erty the Swedish Way." NBER Working Paper
No. 4945. Cambridge, MA: National Bureau of
Economic Research, Dec., 1994.
BLACKBURN, McKINLEY L. "International Com-
parisons of Income Poverty and Extreme In-
come Poverty," Amer. Econ. Rev., May 1994,
84(2), pp. 371-74.
BLACKBURN, McKINLEY L. AND BLOOM, DAVID.
"Changes in the Structure of Family Income-
Inequality in the U.S. and Other Industrialized
Nations During the 1980s." Luxembourg In-
come Study Working Paper No. 118, Nov.,
1994.
BLANCHFLOWER, DAVID G. AND FREEMAN,
RICHARD B. "Unionism in the United States
and Other Advanced OECD Countries," Ind.
Relat., Winter 1992, 31(1), pp. 56-79.
BLANK, REBECCA. "Why Were Poverty Rates So
High in the 1980s?" in Poverty and prosperity
in the USA in the late twentieth century. Eds.:
DIMITRI B. PAPADIMITRIOU AND EDWARD N.
WOLFF. NY: Macmillan, 1994,
pp.
21-55.
BLAU, FRANCINE D. AND KAHN, LAWRENCE M.
"International Differences in Male Wage
Inequality: Institutions versus Market Forces,"
J. Polit. Econ., Aug. 1996, 104(4), pp. 791-
836.
BoISJOLY, JOHANNE; DUNCAN, GREG J. AND
SMEEDING, TIMOTHY M. "The Shifting Inci-
dence of Involuntary Job Losses from 1968 to
1992." Mimeo. Northwestern U., July 1996.
BORLAND, JOHN. "Wage Inequality in Australia."
Paper presented at the NBER Conference on
the Labor Market in International Perspective,
Apr. 1992 [revised July 1992].
BRANDOLINI, ANDREAS AND SESTITO, PAOLO.
"La Distribuzione dei Redditi Familiari in
Italia, 1977-1991," Servizio Studi, Banca
d'Italia, 1993.
BUHMANN, BRIGITTE ET AL. "Equivalence Scales,
Well-Being, Inequality, and Poverty: Sensitivity
Estimates across Ten Countries Using the Lux-
embourg Income Study (LIS) Database,"
Rev. Income Wealth, June 1988, 34(2), pp. 115-
42.
BURKHAUSER, RICHARD V. ET AL. . "Where in the
World Is the Middle Class? A Cross-National
Comparison of the Shrinking Middle Class Us-
ing Kernel Density Estimates." Cross-National
Studies in Aging Program Project Paper No.
26, All-University Gerontology Center, The
Maxwell School, Syracuse, NY: Syracuse U.,
Jan. 1996.
BURKHAUSER, RICHARD V. AND POUPORE, JOHN
G. "A Cross-National Comparison of Permanent
Inequality in the United States and Germany,"
Rev. Econ. Statist., Feb. 1997, 79(1), pp. 10-15.
BURTLESS, GARY. "International Trade and the
Rise in Earnings Inequality," J. Econ. Lit., June
1995, 33(2), pp. 800-16.
CALLAN, TIMOTHY AND NOLAN, BRIAN. "Income
Inequality and Poverty in Ireland in the 1970s
and 1980s." ESRI Working Paper No. 43. Dub-
lin: ESRI, 1993.
CALMFORS, LARS AND DRIFFILL, JOHN. "Bargain-
ing Structure, Corporatism and Macroeconomic
Performance," Econ. Policy, Apr. 1988, 6, pp.
14-61.
CANCIAN, MARIA; DANZIGER, SHELDON AND
GOTTSCHALK, PETER. "Working Wives and
Family Income Inequality Among Married Cou-
ples," in SHELDON DANZIGER AND PETER
GOTTSCHALK, eds. 1993, pp. 195-221.
CANCIAN, MARIA AND REED, DEBORAH. "Assess-
ing the Effects of Wives' Earnings on Family
Income Inequality," Rev. Econ. Statist., forth-
coming.
CANCIAN, MARIA AND SCHOENI, ROBERT. "Fe-
male Earnings and the Level and Distribution
of Household Income in Developed Countries."
LIS Working Paper No.84, The Maxwell
School, Syracuse U., Sept. 1992.
CANCEILL, G. AND VILLENEUVE, A. "Les
Inegalites de Revenus: Quasi Statu Quo entre
1979 et 1984 pour les Salaries et les Inactifs,"
Economies et Statistique, 1990, 230, pp. 65-74.
CANTILLON, BEA ET AL. Indicateurs sociaux:
1985-1992. Centre for Social Policy, U. of Ant-
werp, 1994.
CAPPELLI, PETER. "Are Skill Requirements Ris-
ing? Evidence from Production and Clerical
Jobs," Ind. Lab. Relat. Rev., 1993, 46(3), pp.
515-29.
CARD, DAVID AND FREEMAN, RICHARD, eds.
Small differences that matter. Chicago: U. of
Chicago Press, 1993.
CARD, DAVID; KRAMARZ, FRANCIS AND
LEMIEUX, THOMAS. "Changes in the Relative
Structure of Wages and Employment: A Com-
parison of the United States, Canada and
France." Working Paper No. 355. Princeton,
NJ: Industrial Relations Section, Princeton U.,
1995.
CONCIALDI, PIERRE. "Income Distribution in
France: The Mid-1980's Turning Point," in PE-
678 Journal of Economic Literature, Vol. XXXV (June 1997)
TER GOTTSCHALK, BJORN GUSTAFSSON, AND
EDWARD PALMER, eds. 1997
COULTER, FIONA; COWELL, FRANK A. AND
JENKINS, STEPHEN P. "Equivalence Scale
Relativities and the Extent of Inequality and
Poverty," Econ. J., Sept. 1992, 102(1), pp.
1067-82.
COWELL, FRANK AND JENKINS, STEPHEN P. "The
Changing Pattern of Inequality: The U.S. in the
1980s." Economics Department Discussion Pa-
per No. 93-10. Swansea: U. of Wales, 1993.
DANZIGER, SHELDON AND GOTTSCHALK, PETER.
Uneven tides: Rising inequality in America. NY:
Russell Sage Foundation, 1993
. American unequal. Cambridge: Harvard
U. Press, 1995.
DAVIS, STEVEN J. AND TOPEL, ROBERT H. "Inter-
national Trade and American Wages in the
1980s: Giant Sucking Sound or Small Hiccup?
Comment," Brookings Pap. Econ. Act., 1993, 2,
pp. 214-21.
DEINENGER, KLAUS AND SQUIRE, KENNETH.
"Measuring Income Inequality: A New Dat-
abas." World Bank mimeo, Washington, DC.
1995.
DINARDO, JOHN; FORTIN, NICOLE M. AND
LEMIEUX, THOMAS. Labor market institutions
and the distribution of wages, 1973-1992: A
semiparametric approach. Montreal: U. of
Montreal, 1994.
DINARDO, JOHN AND LEMIEUX, THOMAS. "Di-
verging Male Wage Inequality in the United
States and Canada, 1981-1988: Do Unions Ex-
plain the Difference?" Mimeo, U. of California
at Irvine, 1993.
DUNCAN, GREG: SMEEDING, TIMOTHY AND
RODGERS, WILLARD. "Whither the Middle
Class?" in Poverty and prosperity in the USA in
the late twentieth century. Eds.: DIMITRI B.
PAPADIMITRIOU AND EDWARD WOLFF. New
York: Macmillan, 1994, pp. 202-71.
EDIN, PER-ANDERS AND HOLMLUND, BERTIL.
"The Swedish Wage Structure: The Rise and
Fall of Solidarity Wage Policy?" in RICHARD
FREEMAN AND LAWRENCE KATZ, eds. 1995,
pp. 307-44.
EPLAND, JON. "Inntektsfordelingen i 80-CErene,"
Ekonomiske Analyser, 1992, 2, pp. 17-26.
ERIKSSON, CHRISTOPHER L. AND ICHINO, AN-
DREA C. "Wage Differentials in Italy: Market
Forces, Institutions, and Inflation," in RICHARD
FREEMAN AND LAWRENCE KATZ, eds. 1995,
pp. 265-306.
ERIKSSON, TOM AND JANTTI, MARCUS. "The Dis-
tribution of Earnings in Finland, 1971-1990."
Mimeo, Turku, Finland: Abo Akademi U., 1994.
EUROPEAN COMMISSION, DIRECTORATE-GEN-
ERAL FOR EMPLOYMENT, INDUSTRIAL RELA-
TIONS AND SOCIAL AFFAIRS. "Social Protection
in Europe, 1993." CE-79-93-978-EN-C. Brus-
sels, Belgium: Office for Official Publications of
the European Communities (EC), 1994.
. "The Demographic Situation in the Euro-
pean Union: 1994 Report." CE-84-94-202-EN-
C. Brussels, Belgium: Office for Official Publi-
cations of the European Communities (EC),
1995.
FARBER, HENRY S. "Are Lifetime Jobs Disappear-
ing? Job Duration in the United States, 1973-
1993." Working Paper No. 341. Princeton, NJ:
Industrial Relations Section, Princeton U.,
1995.
FORSTER, MICHAEL. "Poverty in OECD Coun-
tries," OECD Social Policy Studies, Oct. 1993.
10.
FORTIN, NICOLE M. AND LEMIEUX, THOMAS.
"Institutional Changes and Rising Wage In-
equality: Is There a Linkage?" J. Econ. Perspec-
tives, forthcoming.
FREEMAN, RICHARD B., ed. Working under dif-
ferent rules. New York: Russell Sage Founda-
tion, 1994.
- . "Are Your Wages Set in Beijing?" J. Econ.
Perspectives, Summer 1995, 9(3), pp. 15-32.
FREEMAN, RICHARD B. AND KATZ, LAWRENCE
F., eds. Differences and changes in wage struc-
tures. Chicago: U. of Chicago Press, 1995.
. "Rising Wage Inequality: The United
States vs. Other Advanced Countries," in RICH-
ARD B. FREEMAN, ed. 1994, pp. 29-62.
FREEMAN, RICHARD B. AND NEEDELS, KAREN.
"Skill Differentials in Canada in an Era of Ris-
ing Labor Market Inequality." Paper presented
at NBER Conference on United States and Ca-
nadian Labor Markets, Jan. 1991.
FRITZELL, JOHAN. "Income Inequality in Five Na-
tions," Acta Sociologia, 1993, 36(1).
GARDINER, KAREN. "A Survey of Income Inequal-
ity Over the Last Twenty Years: How Does the
U.K. Compare?" Welfare State Programme Dis-
cussion Paper WSP/100. London: London
School of Economics, 1993.
GOLDIN, CLAUDIA AND KATZ, LAWRENCE F.
"Technology, Skill, and the Wage Structure: In-
sights From the Past," Amer. Econ. Rev., May
1996, 86(2), pp. 252-57.
GOLDIN, CLAUDIA AND MARGO, ROBERT A. "The
Great Compression: The Wage Structure in the
United States at Mid-Century." NBER Working
Paper No. 3817. Cambridge, MA: National Bu-
reau of Economic Research, 1991.
GOODMAN, ALISSA AND WEBB, STEVEN. For
richer, for poorer. IFS Commentary No. 42.
London: Institute for Fiscal Studies, 1994.
GORDON, DAVID M. Fat and mean: The corpo-
rate squeeze of working Americans and the
myth of managerial "downsizing." New York:
Martin Kessler Books, 1996.
GORNICK, JANET. "Economic Gender Gaps in the
Industrialized Countries: Evidence from LIS."
Paper prepared for the United Nations Human
Development Report Office, Baruch College,
CUNY, Nov. 20,1994.
GOTTSCHALK, PETER; GUSTAFSSON, BJORN A.
AND PALMER, EDWARD, eds. The distribution
of economic welfare in the 1980s. Cambridge U.
Press, 1997.
- . "What's Behind the Increase in Inequal-
Gottschalk and Smeeding: Cross-National Income Inequality 679
ity?" (1997a) in PETER GOTTSCHALK, BJORN A.
GUSTAFSSON, AND EDWARD PALMER, eds.
1997, pp. ??.
GOTTSCHALK, PETER AND JOYCE, MARY. "Is
Earnings Inequality Also Rising in Other Indus-
trialized Countries?" Mimeo. Boston College,
1995.
. "Changes in Earnings Inequality in
OECD Countries: The Role of Market and In-
stitutional Factors." Mimeo. Boston College,
1996.
GOTTSCHALK, PETER AND MOFFITT, ROBERT.
"The Growth of Earnings Instability in the U.S.
Labor Market," Brookings Pap. Econ. Act.,
1994, 2, pp. 217-72.
GRAMLICH, EDWARD M.; KASTEN, RICHARD AND
SAMMARTINO, FRANK. "Growing Inequality in
the 1980s: The Role of Federal Taxes and Cash
Transfers," in. SHELDON DANZIGER AND PE-
TER GOTTSCHALK, eds.,1993, pp. 225-49.
GREEN, GORDON; CODER, JOHN F. AND RYSCAV-
AGE, PAUL. "International Comparisons of
Earnings Inequality for Men in the 1980s," Rev.
Income Wealth, Mar. 1992, 38(1), pp. 1-15.
GREGORY, ROBERT AND VELLA, FRANK. "Aspects
of Real Wage and Employment Changes in the
Australian Male Labour Market." Mimeo. Aus-
tralian National U., July 1992.
GROSHEN, ERICA L. "Rising Inequality in a Salary
Survey: Another Piece of the Puzzle." Fed. Res.
Bank of Cleveland Working Paper 9121, Dec.
1991.
GUSTAFSSON, BJORN AND PALMER, EDWARD.
"Changes in Swedish Inequality: A Study of
Equivalent Income, 1975-1991." Mimeo. U. of
Gothenburg, 1993.
HARKNESS, STEVEN: MACHIN, STEPHEN AND
WALDFOGEL, JANE. "Women's Pay and Family
Incomes in Britain 1979-1991," in New inequal-
ities: The changing distribution of income
and wealth in the UK. Ed.: JOHN HILLS.
Cambridge: Cambridge U. Press, 1996, pp.
158-80.
HARTOG, JOOP; OOSTERBEEK, HESSEL AND
TEULINGS, COEN. "Age, Wages and Education
in The Netherlands," in Labour markets in an
ageing Europe. Eds.: PAUL JOHNSON AND
KLAUS F. ZIMMERMAN. London: Cambridge U.
Press, 1993, pp. 182-211.
HAUSER, RICHARD AND BECKER, IRENE. "The
Development of the Income Distribution in the
Federal Republic of Germany During the Sev-
enties and Eighties." Cross-National Studies in
Aging Program Project Paper No. 26, All-U.
Gerontology Center, The Maxwell School,
Syracuse U., Dec. 1993.
HIBBS, DOUGLAS A., JR. "Wage Dispersion and
Trade Union Action in Sweden," in Generating
equality in the welfare state: The Swedish expe-
rience. Ed.: INGA PERSSON. Oslo: Norwegian
U. Press, 1990, pp. 181-200.
HILLS, JOHN. "Inquiry into Income and Wealth
Volume II: A Summary of the Evidence." York:
Joseph Rowntree Foundation, 1995.
HORRIGAN, MICHAEL W. AND MINCY, RONALD
B. "The Minimum Wage and Earnings and In-
come Inequality," in SHELDON DANZIGER AND
PETER GOTTSCHALK, eds., 1993, pp. 251-75.
HOWELL, DAVID. "The Collapse of Low-Skill
Male Earnings in the 1980s: Skills Mismatch or
Shifting Wage Norms?" New School for Social
Research, mimeo, 1995.
JANTTI, MARCUS AND DANZIGER, SHELDON.
"Child Poverty in Sweden and the United
States: The ETfect of Social Transfers and Pa-
rental Labor Force Participation," Ind. Lab. Re-
lat. Rev., Oct. 1994, 48(1), pp. 48-64.
JENKINS, STEPHEN P. "The Within-Household
Distribution and Why It Matters: An Econo-
mists's Perspective," in Familien zwischen gere-
chtigskeitsidealen und benachteiligungen. Ed.:
C. BADELT. Vienna: Bohlau-Verlag, 1994, ch. 4,
pp. 74-98.
. "Accounting for Inequality Trends: De-
composition Analyses for the U.K., 1971-86,"
Economica, Feb. 1995a, 62(245), pp. 29-63.
"
Did the Middle Class Shrink during the
1980s? U.K. Evidence from Kernel Density Es-
timates," Econ. Letters, Oct. 1995b, 49(4), pp.
407-13.
. "Recent Trends in the UK Income Distri-
bution: What Happened and Why?" Oxford
Rev. Econ. Pol., 1996,12(1), pp. 29-46.
JENKINS, STEPHEN P. AND COWELL, FRANK.
"Parametric Equivalence Scales and Scale Rela-
tivities," Econ. J., July 1994, 104(425), pp. 891-
900.
JOHNSON, PAUL AND WEBB, STEVEN. "Explaining
the Growth in UK Income Inequality: 1979-
1988" Econ. J., Mar. 1993, 103(415), pp. 429-
35.
KAROLY, LYNN A. "The Trend in Inequality
among Families, Individuals, and Workers in
the United States: A Twenty-five Year Perspec-
tive," in SHELDON DANZIGER AND PETER
GOTTSCHALK, eds. 1993, pp. 19-97.
. "Anatomy of the United States Income
Distribution: Two Decades of Change." Mimeo.
RAND Corporation, Santa Monica, CA, Sept.
1995.
KAROLY, LYNN A. AND BURTLESS, GARY. "Demo-
graphic Change, Rising Earnings Inequality and
the Distribution of Peronal Well-Being, 1959-
1989," Demography, Aug. 1995, 32(3), pp. 379-
405.
KATZ, LAWRENCE; LOVEMAN, GARY AND
BLANCHFLOWER, DAVID. "A Comparison of
Changes in the Structure of Wages in Four
OECD Countries," in RICHARD FREEMAN AND
LAWRENCE KATZ, eds. 1995, pp. ??
KATZ, LAWRENCE AND MURPHY, KEVIN.
"Changes in Relative Wages, 1963-1987: Sup-
ply and Demand Factors," Quart. J. Econ., Feb.
1992, 107(1), pp. 35-78.
KRUEGER, ALAN. "How Computers Have
Changed the Wage Structure? Evidence from
Microdata, 1984-1989," Quart. J. Econ., Feb.
1993, 108(1), pp. 33-60.
680 Journal of Economic Literature, Vol. XXXV (June 1997)
KRUGMAN, PAUL. "Technology, Trade, and Factor
Prices." NBER Working Paper No. 5355. Cam-
bridge, MA: National Bureau of Economic Re-
search, Nov., 1995.
LAMPMAN, ROBERT J. Ends and means of reduc-
ing income poverty. Chicago: Markham Pub.
Co., 1971.
LAWRENCE, ROBERT Z. AND SLAUGHTER, MI-
CHAEL J. "International Trade and American
Wages in the 1980s: Giant Sucking Sound or
Small Hiccup?" Brookings Pap. Econ. Act., Mi-
croeconomics, 1993, 2, pp. 161-210.
LEAMER, EDWARD E. "Wage Inequality from
International Competition and Technologi-
cal Change: Theory and Country Experi-
ence," Amer. Econ. Rev., May 1996, pp. 309-
14.
LERMAN, ROBERT. "The Impact of the Changing
U.S. Family Structure on Child Poverty and In-
come Inequality, "Economica, Supplement
1996, 63(250), pp. S119-39.
LERMAN, ROBERT AND YITZHAKI, SHLOMO. "In-
come Inequality Effects by Income Source," Rev.
Econ. Statist., Feb. 1985, 67(1), pp. 151-56.
LEVY, FRANK AND MURNANE, RICHARD J. "U.S.
Earnings Level and Earnings Inequality: A Re-
view of Recent Trends and Proposed Explana-
tions," J. Econ. Lit., Sept. 1992, 30(3), pp.
1333-81.
LIS. Authors' calculations from the Luxembourg
Income Study database, 1995.
MACURDY, THOMAS AND MROZ, THOMAS. "Mea-
suring Macroeconomic Shifts in Wages from
Cohort Specifications." Presented at the NBER
Labor Markets conference in Aug. 1989.
Mimeo, Department of Economics, Stanford
U., revised, Sept. 1995.
MEADE, JAMES E. Efficiency, equality, and the
ownership of property. London: Allen and Un-
win, 1964.
MESSERE, KENNETH. Tax policy in OECD coun-
tries: Choices and conflicts. Amsterdam, The
Netherlands: BV Publications, 1993.
MISHEL, LAWRENCE AND BERNSTEIN, JARED.
"Technology and the Wage Structure: Has
Technology's Impact Acce erated Since the
1970's?" Economic Policy Institute. July, 1996.
MOFFITT, ROBERT. "Incentive Effects of the U.S.
Welfare System: A Review," J. Econ. Lit., Mar.
1992, 30(1), pp. 10-61.
MOFFITT, ROBERT A. AND GOTTSCHALK, PETER.
"Trends in the Autocovariance Structure of
Earnings in the U.S.: 1969-1987." Mimeo,
Johns Hopkins U., July 1995.
MUFFELS, RUUD AND NELISSEN, JAN. "The
Distribution of Economic Well-Being in The
Netherlands, its Evolution in the 1980s and
the role of Demographic Changes," in PETER
GOTTSCHALK, BJORN A. GUSTAFSSON, AND
EDWARD PALMER, eds. 1977.
MURPHY, KEVIN M. AND WELCH, FINIS. "Indus-
trial Change and the Rising Importance of
Skill," in SHELDON DANZIGER AND PETER
GOTTSCHALK, eds. 1993, pp. 101-32.
NICKELL, STEPHEN AND BELL, BRIAN. "The Col-
lapse in Demand for the Unskilled and Unem-
ployment across the OECD," Oxford Rev. Econ.
Pol., Spring 1995, 11(1), Oxford U. Press, pp.
40-62.
ORGANISATION FOR ECONOMIC CO-OPERATION
AND DEVELOPMENT (OECD). "Earnings
Inequality: Changes in the 1980s," in Employ-
ment outlook. Paris: OECD, 1993, ch. 5, pp.
157-78.
. "New Orientations for Social Policy." So-
cial Policy Studies Paper No. 12. Paris: OECD,
1994a.
. The jobs study: Facts, analysis, strategies.
Paris: OECD, 1994F.
. "The OECD
jobs
Study: Evidence and
Explanations," in Labor market trends and
underlying forces of change. Paris: OECD,
1994c.
. "Earnings Inequality, Low-Paid Employ-
ment and Earnings Mobility" in Employment
outlook. Paris: OECD, 1996, ch. 3, pp. 59-108.
PEN, JAN. Income distribution. Harmondworth
Penguin, 1971.
PENCAVEL, JOHN. Labor markets under trade
unionism: Employment, wages, and hours.
Cambridge: Blackwell, 1991.
PLOTNICK, ROBERT AND SMOLENSKY, EUGENE.
"Inequality and Poverty in the United States:
1900 to 1990." Working Paper #193, Graduate
School of Public Policy. Berkeley, CA: U. Of
California-Berkeley, July 8, 1992.
PLOUG, NIELS AND KVIST, JON, eds. Recent trends
in cash benefits in Europe. Copenhagen: Danish
National Institute of Social Research, 1994.
RADNER, DANIEL B. "Adjusted Estimates of the
Size Distribution of Family Money Income," J.
Bus. Econ. Statist., Apr. 1983, 1, pp. 135-46.
RAINWATER, LEE AND SMEEDING, TIMOTHY M.
"Doing Poorly: The Real Income of American
Children in a Comparative Perspective."
Luxembourg Income Study Working Paper
No. 127, Center for Policy Research, The Max-
well School. Syracuse, NY: Syracuse U., Aug.
1995.
RICHARDSON, J. DAVID. "Income Inequality and
Trade: How to Think, What to Conclude," J.
Econ. Perspectives, Summer 1995, 9(3), pp. 33-
55.
RODRIGUES, CARLOS. "The Measurement and
Decomposition of Inequality in Portugal,
1980/81-1989/90." Microsimulation Unit Dis-
cussion Paper No. 9302. Cambridge: U. of
Cambridge, 1993.
SACHS, JEFFREY D. AND SHATZ, HOWARD J.
"Trade and Jobs in U.S. Manufacturing," Brook-
ings Pap. Econ. Act., 1994, 1, pp. 1-84.
SAUNDERS, PETER. "Rising on the Tasman Tide:
Income Inequality in Australia and New Zea-
land in the 1980s." SPRC Discussion Paper No.
49. U. of New South Wales, 1994.
SAWYER, MICHAEL. "Income Distribution in
OECD Countries." Paris: OECD, 1976.
SHORROCKS, ANTHONY F. "Inequality Decompo-
Gottschalk and Smeeding: Cross-National Income Inequality 681
sition by Factor Components," Econometrica,
Jan. 1982,50(1), pp. 193-212.
SMEEDING, TIMOTHY M. AND CODER, JOHN. "In-
come Inequality in Rich Countries During the
1980s," J. Income Distribution, Spring 1995,
5(1), pp. 13-29.
SMEEDING, TIMOTHY M. AND GOTTSCHALK, PE-
TER. "The International Evidence on Income
Distribution in Modern Economies: Where
Do We Stand?" in Contemporary economic de-
velopment reviewed. Eds.: MICHAEL BRUNO
AND YUSET MUNDLAB. London: Macmillan,
1996.
SMEEDING, TIMOTHY M.; O'HIGGINS, MICHAEL
AND RAINWATER, LEE, eds. Poverty, inequal-
ity, and income distribution in comparative per-
spective: The Luxembourg Income Study. Wash-
ington, DC: Urban Institute; Brighton:
Harvester Wheatsheaf, 1990.
SMEEDING, TIMOTHY M.; RAINWATER, LEE AND
TORREY, BARBARA. "Going to Extremes: The
U.S. Elderly in an International Context." Lux-
embourg Income Study Working Paper No. 89,
Nov. 1993.
SMEEDING, TIMOTHY M. ET AL. "Poverty, In-
equality, and Family Living Standards Impacts
Across Seven Nations: The Effect of Noncash
Subsidies for Health, Education and Housing,"
Rev. Income Wealth, Sept. 1993, 39(3), pp.
229-56.
STATISTICS CANADA. "Income After Tax Distri-
bution By Size." Catalogue No. 13-210, Table
VIII. Ottawa: Statistics Canada, 1994.
STATISTICA SWEDEN. As provided by KJELL JANS-
SONS, Sept., 1995.
SUMMERS, ROBERT AND HESTON, ALAN. "The
Penn World Table (Mark 5): An Expanded Set
of International Comparisons, 1950-1989,"
Quart. J. Econ., 1991, 105(2), pp. 327-68.
TACHABANAKI, TOSHIAKI AND YAGI, TADASHII.
"Distribution of Economic Well-Being in Japan:
Towards a More Unequal Society," in PETER
GOTTSCHALK, BJORN A. GUSTAFSSON, AND
EDWARD PALMER, eds. 1995, pp. ??.
U.S. BUREAU OF LABOR STATISTICS. Technology
and Its Effect in the Steel Industry. 1994. Wash-
ington, DC: U.S. GPO.
U.S. DEPARTMENT OF COMMERCE, BUREAU OF
THE CENSUS. Alternative methods for valuing
selected in-kind transfers and measuring their
impact on poverty. Technical Report No. 50.
Washington, DC: U.S. GPO, 1982.
. "Income, Poverty and Valuation of Non-
cash Benefits: 1993," Current Population Re-
ports, Feb. 1995a, Series P-60, No. 188. Wash-
ington, DC: U.S. GPO.
. Unpublished tabulations by John Coder,
1995b.
- . "Money Income in the United States:
1995," Current Population Reports, Sept. 1996,
Series P-60, No. 193. Washington, DC: U.S.
GPO.
UUSITALO, HANNU. "Changes in Income Distri-
bution During a Deep Recession." Mimeo.
Finnish Trade Council, Aug. 1995.
WAGNER, GERT G.; BURKHAUSER, RICHARD V.
AND BEHRINGER, FRIEDERIKE. "The English
Language Public Use File of the German Socio-
economic Panel," J. Human Res., Spring 1993,
28(2), pp. 429-33.
WHITEFORD, PETER AND KENNEDY, STEVEN.
"Incomes and Living Standards of Older Peo-
ple: A Comparative Analysis." Final Report,
Volumes I and II, DSS 1211. Social Policy Re-
search Unit. United Kingdom: U. of York, Aug.
1994.
WILLIAMSON, JEFFREY G. AND LINDERT, PETER
H. American inequality: A macroeconomic his-
tory. New York: Academic Press, 1980.
WOLFE, BARBARA AND MOFFITT, ROBERT. "A
New Index to Value In-Kind Benefits," Rev. In-
come Wealth, Dec. 1991, 37(4), pp. 387-408.
WOOD, ADRIAN. North-south trade, employment,
and inequality. Oxford: Clarendon Press, 1994.
682 Journal of Economic Literature, Vol. XXXV (June 1997)
APPENDIX TABLE A
ABSOLUTE AND RELATIVE CHANGES IN INEQUALITY
Absolute Changes
Value in Value in U.S.
Inequality Initial Terminal Change Change (5)(6)
Source Measure Year Year Per Year Per Year
(1) (2) (3) (4) (5) (6) (7)
Blackburn and Bloom (1994)
Canada (1979-87)
Overall Table 7.8 In var .270 .288 .002 .004 .529
Between Ed Table 7.9 College coef .475 .465 -.001 .010 -.122
Between Ex Table 7.9 Age coefs eval at 24 .224 .027 .000 .000 3.080
Within (univ ed) Table 7.9 Sd of resid .485 .501 .002 .002 -1.778
United States (1979-87)
Overall Table 7.8 In var .286 .320 .004 004 1.000
Between Ed Table 7.9 College coef .570 .652 .010 .010 1.000
Between Ex Table 7.9 Age coefs eval at 24 .028 .029 .000 .000 1.000
Within (univ ed) Table 7.9 Sd of resid .484 .475 -.001 -.001 1.000
Ericksson and Ichino (1995)
Italy (1978-87)
Overall Table III.4A St dev In (w) .402 .355 -.055 .012 -.439
Between Ed Table III.3A College coef .220 .260 .004 .009 .500
Between Ex Table III.3A Exp coefs eval at 2 .037 .026 -.001 .000 -10.036
Within (univ ed) Table III.4A Sd of resid .350 .308 -.005 .010 -.447
United States (1978-87)
Overall Table III.4A St dev In(w) .531 .638 .012 .012 1.000
Between Ed Table III.3A College coef .350 .430 .009 .009 1.000
Between Ex Table III.3A Exp coefs eval at 2 .042 .043 .000 .000 1.000
Within (univ ed) Table III.4A Sd of resid .462 .556 .010 .010 1.000
Full-Time Workers
Gottschalk and Joyce (1995)
Australia (1981-85)
Overall Table 1 Coef Variation .334 .357 .006 .008 .732
Between Ed Table 2 College coef .390 .390 -.040 .024 -1.647
Between Ex Table 3 Age coefs eval at 24 .025 .034 .002 .001 1.508
Within Table 4 Sd of resid .500 .620 .030 .014 2.100
Finland (1987-91)
Overall Table 1 Coef Variation .460 .474 .003 .010 .336
Between Ed Table 2 College coef .610 .560 -.013 .023 -.547
Between Ex Table 2 Age coefs eval at 24 .070 .032 .009 .001 -6.619
Within Table 4 Sd of resid 1.130 1.130 .000 .021 .000
France (1979-84)
Overall Table 1 Coef Variation .396 .434 .008 .010 .729
Between Occ Table 2 Manager coef .390 .350 -.008 .009 -.933
Between Ex Table 2 Age coefs eval at 24 .034 .041 .002 .003 .493
Within Table 4 Sd of resid .450 .480 .006 .020 .300
Israel (1979-86)
Overall Table 1 Coef Variation .470 .512 .006 .010 .575
Between Occ Table 2 College coef .100 .210 .016 .009 1.833
Between Ex Table 2 Age coefs eval at 24 .035 .068 .005 .003 1.554
Within Table 4 Sd of resid .530 .540 .001 .020 .071
Gottschalk and Smeeding: Cross-National Income Inequality 683
APPENDIX TABLE A (cont.)
ABSOLUTE AND RELATIVE CHANGES IN INEQUALITY
Relative Changes
Percentage U.S. Percentage
Change Change (8)(9)
Source Per Year Per Year
(1) (8) (9) (10)
Blackburn and Bloom (1994)
Canada (1979-87)
Overall Table 7.8 .008 .014 .575
Between Ed Table 7.9 -.003 .017 -.158
Between Ex Table 7.9 .015 .004 3.446
Within (univ ed) Table 7.9 .004 -.002 -1.732
United States (1979-87)
Overall Table 7.8 .014 .014 1.000
Between Ed Table 7.9 .017 .017 1.000
Between Ex Table 7.9 .004 .004 1.000
Within (univ ed) Table 7.9 -.002 -.002 1.000
Ericksson and Ichino (1995)
Italy (1978-87)
Overall Table III.4A -.014 .020 -.677
Between Ed Table III.3A .019 .023 .812
Between Ex Table III.3A -.040 .003 -13.914
Within (univ ed) Table III.4A -.014 .021 -.690
United States (1978-87)
Overall Table III.4A .020 .020 1.000
Between Ed Table III.3A .023 .023 1.000
Between Ex Table III.3A .003 .003 1.000
Within (univ ed) Table III.4A .021 .021 1.000
Full-Time Workers
Gottschalk and Joyce (1995)
Australia (1981-85)
Overall Table 1 .017 .017 .987
Between Ed Table 2 -.132 .079 -1.670
Between Ex Table 3 .076 .045 1.686
Within Table 4 .054 .022 2.404
Finland (1987-91)
Overall Table 1 .007 .021 .352
Between Ed Table 2 -.021 .078 -.274
Between Ex Table 2 -.192 .034 -5.695
Within Table 4 .000 .030 .000
France (1979-84)
Overall Table 1 .018 .021 .860
Between Occ Table 2 -.022 .034 -.628
Between Ex Table 2 .041 .078 .526
Within Table 4 .013 .028 .457
Israel (1979-86)
Overall Table 1 .012 .021 .574
Between Occ Table 2 .106 .034 3.077
Between Ex Table 2 .096 .078 1.239
Within Table 4 .003 .028 .094
684 Journal of Economic Literature, Vol. XXXV (June 1997)
APPENDIX TABLE A (cont.)
Absolute Changes
Value in Value in U.S.
Inequality Initial Terminal Change Change (5)(6)
Source Measure Year Year Per Year Per Year
(1) (2) (3) (4) (5) (6) (7)
United States (ed) (1979-86)
Overall Table 1 Coef Variation .454 .527 .010 .010 1.000
Between Ed Table 2 College coef .220 .380 .023 .023 1.000
Between Ex Table 2 Age coefs eval at 24 .037 .047 .001 .001 1.000
Within Table 4 Sd of resid .640 .790 .021 .021 1.000
United States (occ) (1979-86)
Between Occ Table 2 Manager coef .220 .280 .009 .009 1.000
Between Ex Table 2 Age coefs eval at 24 .030 .051 .003 .003 1.000
Within Table 4 Sd of resid .640 .780 .020 .020 1.000
Katz, Loveman, and Blanchflower (1995
France (1979-87)
Overall Figure 1 Ln (90/10) 1.200 1.220 .003 .020 .125
Between Occ Figure 6 Ln(Manual/non) .550 .545 -.001 .024 -.026
Between Ex Figure 6 Ln(41-50/21-25) .030 .540 .010 mixed
Within Figure 6 Ln(90/10) mixed .015
Gottschalk and Smeeding: Cross-National Income Inequality 685
APPENDIX TABLE A (cont.)
ABSOLUTE AND RELATIVE CHANGES IN INEQUALITY
Relative Changes
Percentage U.S. Percentage
Change Change (8)(9)
Source Per Year Per Year
(1) (8) (9) (10)
United States (ed) (1979-86)
Overall Table 1 .021 .021 1.000
Between Ed Table 2 .078 .078 1.000
Between Ex Table 2 .034 .034 1.000
Within Table 4 .030 .030 1.000
United States (occ) (1979-86)
Between Occ Table 2 .034 .034 1.000
Between Ex Table 2 .078 .078 1.000
Within Table 4 .028 .028 1.000
Katz, Loveman, and Blanchflower (1995
France (1979-87)
Overall Figure 1 .002 .014 .144
Between Occ Figure 6 -.001 .057 -.020
Between Ex Figure 6 .020 mixed
Within Figure 6 mixed .012
686 Journal of Economic Literature, Vol. XXXV (June 1997)
APPENDIX TABLE B
TREND IN INCOME INEQUALITY IN OECD AND OTHER NATIONS: 1970-1993
Year AU CA IR SP JA NZ IS UK US
1970 98
1971 106 101
1972 105 102
1973 105 102 98
1974 103 95
1975 104 92
1976 107 92
1977 103 90
1978 105 91
1979 101 100 96 101
1980 101 100 100 98 100
1981 100 100 100 100 100 103
1982 101 100 107
1983 103 102 109
1984 103 109 103 110
1985 107 102 96 108 110
1986 103 102 111 110
1987 102 87 108 117 110
1988 101 124 112
1989 111 101 110 125 113
1990 101 114
1991 102 96 113 101 130 113
1992 102 130 115
1993 117
Notes and Sources to Appendix Table B and Table 4
Australia (AU) Saunders (1994), Table 7; income per equivalent adult.
Canada (CA) Beach and Slotsve (1994); Statistics Canada (1994, Table VIII); family income after tax,
weighted by households, unadjusted for family size.
Ireland (IR) Callan and Nolan (1993, Table 4); household disposable income with no adjustment for
household size and with household weights.
Spain (SP) Luxembourg Income Study Database (Fall 1995); disposable income per equivalent adult,
person weights, E =.5 equivalence scale.
Japan (JA) Tachabanaki and Yagi (1995); Bauer and Mason (1992). Unadjusted disposable family
income; dataset excludes single person families living alone.
New Zealand (NZ) Saunders (1994); income per equivalent adult.
Israel (IS) Luxembourg Income Study Database (Fall 1995); disposable income per equivalent
adult, person weights, E =.5 equivalence scale.
United Kingdom (UK) Atkinson (1993); Goodman and Webb (1994, p. A2; equivalence scale, British Households
Below Average Income scale; disposable household income with person weights.
United States (US) U.S. Department of Commerce, Bureau of the Census (1995a, 1995b), special tabulations
using disposable personal income as defined in the text, weighted by persons and
adjusted using an equivalence scale with E =.5. Some series shown in Table 3.
Belgium (BE) Cantillon et al. (1994. Table 30); equivalence (scale 1.0 for the first adult, 9.7 for the
second adult, and 0.5 per child); disposable income with person weights.
Denmark (DK) Aaberge et al. (1995 Table A.4); unadjusted household disposable personal income
weighted by person.
Finland (FI) Uusitalo (1995; Table 2); equivalence scale 1.0 for the first adult, 0.7 for the second adult
and 0.5 per child; disposable income with person weights.
Gottschalk and Smeeding: Cross-National Income Inequality 687
APPENDIX TABLE B (cont.)
Year BE DK FI FR GEA/G EB IT NL NO PO SWA/S WB
1970 109
1971
1972 131
1973
1974
1975 105 112
1976 105 109
1977 114 107
1978 109 105
1979 100 110 103
1980 102 100 102
1981 100 100 100 100 100
1982 99 94 100 102
1983 98 100/100 95 98 102
1984 100 102/100 /98 98 107
1985 100 100 97 104/96 99 107
1986 100 100 /98 99 97 112
1987 102 97 101/97 104 104 107
1988 104 105 99 102 107
1989 106 100 102 /97 97 105 104 98 110
1990 111 99 104/97 123/114
1991 98 95 129/121
1992 105 97 /117
1993 102 /119
France (FR) Canceill and Villeneuve (1990, p. 71); Concialdi (1996, Table 13); household income with
no adjustment for household size and with household weights.
West Germany (GE) A: Hauser and Becker (1993, Table 7)
B: Burkhauser and Poupore (forthcoming, Table 3)
Italy (IT) Brandolini and Sestito (1993, Table 2a); equivalent disposable income with household
weights.
The Netherlands (NL) Data provided by Central Bureau of Statistics, see Atkinson, Rainwater, and Smeeding
(1995a, Chapter 5); household disposable income (deducting from net income interest
paid, health care and life insurance premiums, wealth tax payments, and alimony paid)
with no adjustment for household size and with household weights.
Norway (NO) Epland (1992, Table 4); equivalence scale 1.0 for first adult, 0.7 for second household
member and 0.5 for subsequent members; disposable income with person weights.
Portugal (PO) Rodriques (1993, Table 3); equivalence scale with family size elasticity E + .5; adjusted
household disposable income with person weights.
Sweden (SW) A: Gustafsson and Palmer (1993, Annex); equivalence scale; Swedish social assistance
scale; disposable income (including an allowance for imputed rent on owner-occupied
homes) with person weights. Bjorklund and Freeman (1994).
B: Statistics Sweden (1995)

Journal of Economic Literature Vol. XXXV (June 1997), pp. 633-687

Cross-National Comparisons of Earnings and Income Inequalit
PETER GOTTSCHALK Boston College and TIMOTHY M. SMEEDING Syracuse University
We are grateful for the assistance and encouragement given by our colleagues Anthony B. Atkinson and Lee Rainwater as well as the large number of country experts who provided useful comments on earlier drafts and participants at seminars at Princeton, Rand, Stanford, and U.C. Berkeley. Supportfor this project was provided for Smeeding by the Russell Sage Foundation and by NSF #SBR-9511521. Helpful comments were received from several referees, Anthony Atkinson, Anders Bj5rklund, Gary Burtless, Sheldon Danziger, John Fitzgerald, Johan Fritzell, Joop Hartog, MarkusJdntti, StephenJenkins, Robert Lerman,John Myles, Robert Plotnick, James P. Smith, and Peter Saunders. Esther Gray, Katherin Ross, and Ann Wicks provided excellent assistance with the manuscript. The authors assume responsibility for all errors of commission.

I. Introduction
IN THE distribution of earnings and the distribution of household income was largely a parochial backwater of economic research in the United States until the early 1980s. This lack of interest reflected the view that both the functional and personal distributions of income in the United States showed little change between the end of the 1940s and the mid-1970s.1 This led Robert Lampman (1971, p. 47) to remark that the stability of the income distributions was "remarkable in view of the great changes which have occurred in economic structure and in income and wealth levels." He further noted that predictions of increasing concentration JNTEREST
1 In contrast, British researchers such as Atkinson (1970) and Dutch researchers such as Jan Pen (1971) and their predecessors made key contributions to both inequality theory and measurement during the 1970s.

of wealth "have been proved completely wrong by the American experience." Taking a somewhat more laid-back perspective, Henry Aaron (1978, p. 17) noted that tracking changes in the distribution of income in the United States "waslike watching the grass grow." The lack of interest in distributional issues in the United States in general, and cross-national comparisons in particular, changed for several reasons in the early 1980s. First, the view that the shape of the income distribution was one of the great constants of economics came into question by a series of studies, reviewed in Frank Levy and Richard Murnane (1992), that showed rising inequality of labor market income in the United States and a smaller set of studies that showed that these changes in the earnings distribution were being translated into greater inequality in the distribution of total family income.

633

634

Journal of Economic Literature, Vol. XXXV (June 1997) tion was at least partially responsible for the increase in inequality. Likewise, cross-country comparisons of changes in industrial structure or unionization would at least provide some stylized facts that might inform the debate on the causes of the increase in inequality. The third factor contributing to the increased interest in distribution issues stems not from the positive interest in understanding the causes of change but the normative issues coming out of the debate in the public policy arena over the "fairness issue." The distributive effects of changes in government policies, which had always been a key policy issue in European, Nordic, and British Commonwealth countries, have become an increasingly important policy issue in the United States. In this article, we further develop Levy and Murnane's (1992) review of changes in earnings inequality in the United States in two directions. First, we expand our review to other major industrialized countries, largely OECD nations. Second, we broaden the focus from earnings to household income. As we will show, the increases in the dispersion of both individual earnings and total household income in the United States were larger than in almost all other countries. However, the United States was not the only country to experience an increase in inequality during the 1980s and early 1990s. While most countries experienced at least modest increases in earnings and market income (income before direct taxes and public income transfers) inequality, these were largely offset by changes in other sources of income, producing a more modest increase in the inequality of disposable incomes in most nations. We review not only what we know about what has happened to earnings but also why experiences differed across countries. While causal explanations are

Second, it became considerably easier to perform cross-national comparisons of income distributions. Before 1980 such comparisons were of a rough and ready nature and did not withstand close scrutiny.2 Yet international comparisons of income distribution can provide important benchmarks of how one nation differs from or is similar to other nations. In so doing, they can provide useful information, just as do cross-national comparisons of rates of economic growth, savings, inflation, and unemployment. Fortunately, cross-sectional micro data became publicly available for a variety of rich OECD countries.3 This opened several avenues for research, primarily by allowing greater richness in cross-national comparisons. While cross-national comparisons of average income had been widely used to measure differences in standard of living across countries, these comparisons had focused on the typical or average family. Data about the relative standard of living of persons elsewhere in the distribution could now provide a more complete picture of cross-national differences. These new data also contributed to the literature on trends in inequality. Researchers were not only able to address the factual question of whether inequality grew in other countries but also to get further hints as to possible causes. For example, if countries with binding trade barriers experienced smaller increases in inequality then this would be consistent with the view that increased foreign competi2 See, for example, Michael Sawyer (1976) and the strong negative response to Sawyer by Jean Begue (1976). 3 Database projects such as the Luxembourg Income Study LIS), described in Smeeding, Michael O'Higgins, and Rainwater (1990), and related efforts to make longitudinal household panel data comparable, for example, the United StatesGerman comparative pane project described in Gert Wagner, Richard Burkhauser,and Friederike Behringer (1993), have facilitated cross-national comparisons of inequality.

Gottschalk and Smeeding: Cross-National Income Inequality never easy to pin down, the issues are fairly well defined. Labor economics provides a rich theoretical framework that has been applied with some success, at least in a partial equilibrium setting, to explain changes in the structure of wages. The expansion from individual earnings to household disposable income, however, raises a whole host of analytical as well as measurement issues.4 Economic and demographic decisions within the household are endogenous and so complex that empirical research is far from being able to sort out the linkages. The problem of endogeneity would arise even in the context of a single country. However, the problem of endogeneity is further aggravated by the expansion to the international context. Social and political institutions that may affect how other household members and government taxes and transfers respond to changes in market conditions differ considerably across countries. As a result, the responses of household disposable income to changes in the structure of wages, interest rates, or other prices will vary across countries. Given the lack of any unifying theoretical structure to analyze household income, we will largely limit ourselves to presenting the basic facts that any theory would have to explain. However, there is a strong need for a better theoretical structure in which to understand these outcomes. We begin our review by laying out a set of stylized facts both for the United States and for other nations. We present summaries for both the level and trend in earnings and income inequality. In Section II we briefly turn to conceptual and comparability issues. This is fol4 Disposable household income includes all sources of income received by all household members, including income transfers from governments and other parties, net of income and payroll taxes.

635

lowed in Sections III and IV by an overview of what we know about changes in earnings inequality in a variety of countries and the causes for these changes. Section V turns to family income inequality to answer the same questions. While we cover a wide range of topics, not everything under the rubric of changes in inequality is addressed. We are concerned with highly developed countries, almost exclusively the OECD nations, and do not address inequality in developing nations or in the transition countries of Eastern Europe and the former U.S.S.R.5 Issues related to wealth inequality, consumption and expenditure inequality, the tradeoff between equality and efficiency, social choice theory, and the theoretical and empirical literature on inequality measurement are largely excluded. Other pertinent issues, such as the burgeoning literature on growth and inequality, the dynamics of income, and intergenerational mobility are also not covered. Finally, due to constraints of space and time, the literature on crossnational comparisons of low incomes or poverty is also excluded.6 A. Stylized Facts The growing interest in national and cross-national differences in earnings and income inequality has produced a wide range of recent comparative studies of the level and trend in inequality along with dozens of studies and reports on individual countries. Our summary of the stylized facts emerging from these studies is as follows:
5 See Klaus Deinenger and Kenneth Squire (1995) on income inequality in developing and developed countries and Atkinson and John Micklewright (1992) on inequality in Eastern and Central Europe. See Smeeding and Gottschalk (1996) for comparisons which include the OECD nations, selected Eastern European nations, and Taiwan. 6 Interested readers should consult Michael Forster (1993), McKinley Blackburn (1994), Atkinson, Rainwater, and Smeeding (1995a), Rainwater and Smeeding (1995).

636

Journal of Economic Literature, Vol. XXXV (June 1997) nations, with the greatest inequality in the United States and the least inequality in Nordic and Northern European countries. (2) Post-tax and transfer disposable money income is more equally distributed than market income in all OECD nations, and there is a noticeable correlation between public cash income transfer expenditures and disposable income inequality. (3) Even after adjusting for real income differences across countries (using purchasing power parity), low income United States citizens have real living standards below those found in most other rich OECD countries. Trends (1) Increases in household income inequality were more muted than were changes in earnings inequality in most nations. Still, increased earnings inequality among men was probably the most important factor in explaining rising income inequality. (2) Income inequality increased in most, but not all, OECD nations during the 1980s and early 1990s. Trends in inequality were not closely associated with levels of inequality. Some nations with low levels of inequality experienced some of the largest increases. (3) Reductions in social welfare spending for the non-aged and regressive changes in the structure of income taxes for some countries during the 1980s account for only a small part of the trend in post-tax and transfer inequality in most nations. (4) Married women's labor force participation rates, hours, and wages increased substantially in almost all countries during the 1980s. The

1. Earnings. Levels (1) At any given time there are wide differences across modern countries in the level of earnings inequality for both men and women. (2) Nations with centralized wage bargaining (e.g., Sweden, Germany) have greater equality than nations with less centralized bargaining (e.g., the United States and Canada). Trends (1) Almost all industrial economies experienced some increase in wage inequality among prime aged males during the 1980s (Germany and Italy are the exceptions). (2) But large differences in trends also exist across countries, with earnings inequality increasing most in the United States and the United Kingdom and least in Nordic countries. (3) The increasing demand for more skilled workers, coupled with differences across countries in the growth in supply of skilled workers, explains a large part of differences in trends in returns to education and experience. (4) Institutional constraints on wages also seem to matter. The rise in the relative unemployment rates of the least skilled in some, but not all, countries with centralized wage setting institutions suggests that constraints were at least partially responsible for limiting the rise in inequality. B. Disposable Income Levels (1) There is substantial diversity in the inequality of household disposable income across major OECD

particularly for those surveys from the Luxembourg Income Study. the existing evidence shows that rankings of countries with respect to income inequality are robust with respect to changes in the accounting period (Rolf Aaberge et al.g. The main source of United States income data is the March Supplement to the Current Population Survey (CPS). and too short for families that can smooth consumption over multiple years. There is even greater diversity in the decision of what to include under total household income. the relevant measure of income might be a few pay periods for families who do not have sufficient assets to smooth consumption and cannot borrow against future income. measure income on a yearly basis. 1995.. which differs nation by nation in quality of data reported and in its economic importance. France). Finland. Surveys may also differ in the income sources they include as earnings. Thus. Israel). may also be included in earnings.g. Almost all nations include vacation pay ("13th month earnings") and salary bonuses in their measures of earned income. For families that are not credit constrained this might require measures of lifetime post-tax and transfer income adjusted for family size.7 637 A. would be adjusted for economies of scale 7 For discussion of the problems of comparability across countries and for additional information on survey differences. At the other extreme. Sweden. in effect an income supplement to a labor force survey. in consumption using an appropriate equivalence scale. Comparability and Data Quality In this section we address the measurement problems raised when making comparisons of earnings and income distributions across countries. This is certainly too long an accounting period for families that are severely credit constrained. Cross-national comparisons of income inequality have focused primarily on the distribution of disposable money income after direct taxes (income and employee payroll) and . Income Definitions Ideally income would be measured on a post-tax and transfer basis consistent with the Haig-Simons income concept of real consumption plus (or minus) change in net worth. II.8 Income would include both cash and noncash components. and would cover the period over which families can smooth consumption by lending or borrowing.. especially Chs. Rainwater. Other nations use specific income surveys or have extensive surveys of expenditures with detailed income components sections (e. thus tending to increase income inequality. almost all of the existing data sets. While the problem raises important conceptual issues. Self-employment income. In a few nations (Sweden. Norway) survey respondents give the statistical office permission to go directly to government records to measure incomes and report only demographic information to the survey takers.Gottschalk and Smeeding: Cross-National Income Inequality positive correlation between husbands' and wives' earnings also increased moderately. For example. and Smeeding (1995a. Richard Burkhauserand John Poupore 1997). 2 and 3 and Appendices 1 through 5 8 This broad definition of income is an attempt to get closer to the distribution of lifetime utility. including the CPS in the United States. Other countries similar to the United States have annual or periodic surveys of consumer finances or income (Canada and Australia). the type and purpose of surveys used for international comparisons vary widely by country. United Kingdom. Unfortunately. The Netherlands. unemployment insurance and/or sick pay are included as a transfer in most countries but are included in earnings in a few (e. see Atkinson.

Germany. New Zealand. Total household income is divided by the number of 9 Direct taxes are most often estimated from tax imputation models rather than official tax records. varies between 0 and 1.g. An important and non-obvious lesson from these papers is that the relationship between inequality measures and elasticities is non-monotonic. XXXV (June 1997) equivalent adults in order to arrive at a measure of household "equivalent" income. Frank Cowell. Further comparabilityissues are raised by definitions of income sharing units and the unit of analysis. and the United States in the Luxembourg Income Study are obtained using a tax imputation model at the household level to estimate direct taxes. the larger is E. the range of possible values is rather evenly covered. See Jenkins and Cowell (1994). typically excluding much of capital gains.9 While this definition of post-tax and transfer disposable income is broad. by age) as well as for family size. Equivalence scales are designed to accomplish this adjustment by taking into account those household characteristics deemed to affect economies of scale and economies of scope as reflected by differences in household size and composition. Sweden. it falls considerably short of the Haig-Simons comprehensive definition. and Smeeding 1995a. In general. . Belgium. assuming that income is equally shared by all members of the household. Survey-based research on income inequality sometimes focuses attention on the household as the unit of income sharing and as the unit of analysis.. These adjustments for family size can have a large effect on the level of measured inequality within and across nations. and Norway use official records of taxes paid. other times the unit of analysis is the individuals within the household. Jenkins (1994). using different equivalence scales preserves the general rank order of countries. Sweden). Adjustmentfor Household Size and Composition: Equivalence Scales Most studies of income inequality adjust income to take account of differences in material needs for families of different sizes. Inequality rankings at a point in time are fairly robust to choice of equivalence scales (Atkinson.11 However. equal sharing within the household assumption. Vol. or even more narrowly related persons according to income tax regulations (e.g. For example the after-tax data for Australia.638 Journal of Economic Literature.. shows that the estimates of overall household income inequality derived from three different methods of estimating within household inequality are very different from those derived using the conventional. however. (1988). Between these extremes. and income in kind. Finland.g. And definitions of income sharing units themselves may differ across nations. (1988) first proposed a succinct parametric approximation to equivalence scales which summarized the wide range of scales in use: Adjusted Income = Disposable Income/SizeE The equivalence elasticity. Rainwater. imputed rents. Due to lack of a long time series of com11 See Fiona Coulter. the smaller are the economies of scale assumed by the equivalence scale. for example: all related persons living together or a famify (e. indirect taxes. a ew use a more narrow definition. Italy. home production. 10While most nations aggregate income across all members of a househol. Brigitte Buhmann et al. The literature has moved beyond the one parameter equivalence scale used here to two parameter scales which include adjustments for types of individuals (e.1). after transfer payments. albeit at different levels of inequality. The various studies reviewed in this survey make use of equivalence scales ranging from E = 0 (no adjustment) to E = 1 (per capita income which ignores economies of scale). Most studies of cross-national distribution make no adjustments for differences in incomes within households. Canada). E. and Stephen Jenkins (1992) and Buhmann et al. or of the benefits from public spending other than cash transfers. no account is taken of leisure. and Luxembourg surveys report only after-tax income.10 B.. Figure 4.

S..g. VAT) taxes across OECD nations. 14 See Clive Bell and Christoph Rosenberg (1993). more uniform than is the distribution of cash benefits alone. and goods and services (expenditure. (1993) and Peter Whiteford and Steven Kennedy (1994). the literature has made little progress in arriving at a true Hicksian equivalent variation measure of their cash equivalent value to households (U. Barbara Wolfe and Robert Moffitt 1991). These estimates indicate that including noncash benefits affects the level but not the trend in inequality since 1979. income.. see U. the literature cannot determine if choice of equivalence scale affects trends in measured inequality across countries.'4 D. However. See Smeeding et al. health insurance and more substantial social retirement pensions in most other nations).12 Including noncash benefits in estimates of the level and trend of income inequality also requires the valuation of these benefits. Bureau of the Census 1982. Department of Commerce. Kenneth Messere (1993) presents aggregate data on the tax mix across countries. the manner in which taxes are collected may affect the results of crossnational comparative analyses. and occupational pensions in many nations). (1993) find that imputation of health and education benefits and some housing benefits had an equalizing impact in all countries. health insurance for workers in the United States.Gottschalk and Smeeding: Cross-National Income Inequality parative data. 639 In-kind benefits also tend to be a small share of total social transfers relative to cash benefits in nations with small shares of GDP spent on cash benefits. the limited evidence indicates that the exclusion of noncash benefits does not have a large impact on the income inequality rankings of countries. In fact. Data Quality Comparison with National Accounts One common criticism of earnings and income distribution data derived from 13 Smeeding et al. however. Hence. property. A similar type of difference may occur if countries rely on employers to provide some types of benefits (e. .S. we know of only one study that has included the full burden of direct and indirect taxes in cross-national studies of income distribution. 12The mix of cash plus noncash benefits across OECD nations is. C. Because high cash benefit nations tend to also be high in-kind benefit nations. Department of Commerce. one might suspect differences across countries depending on a nation's preferences for cash versus noncash transfers. While several national studies of noncash benefits have assessed their impact on the income distribution as measured by the cost of benefits to the supplier. and because of the differential mix of personal.13 Most studies of income distribution employ either a measure of all sources of money income prior to the deduction of all taxes ("gross income") or a measure that subtracts "direct taxes"-income and employee payroll taxes-to arrive at disposable income. while governments provide others (e.g. Because of differential reliance on employer and employee social security contributions across nations. business. In general. or if demographic composition of nations are very different. Noncash Benefits and Taxes The disposable money income measures used in most studies include only public cash and near cash benefits (food stamps and other similar benefits denominated in cash). but did not affect the inequality ranking of countries. Bureau of the Census (1995a). earnings. evidence for differences in trends within the United States indicates that choice of equivalence scale may affect the level of measured inequality but not its trend (Lynn Karoly and Burtless 1995). studies do not count personal property or wealth taxes as direct taxes. For one estimate of the effect of including in-kind benefits in income distribution in the United States.

2 and Appendix 6). and survey data collection practices) are constant across time. XXXV (June 1997) E. What is relevant is not only the amount of underreporting but its distribution. Italy. On the other hand. Rainwater. it is possible to test sensitivity to alternative choices of units. Vol. definitions. if data quality changes over time. One way of determining the size of the underreporting is to compare the total income of different types reported in the household surveys with external information drawn from national accounts and country data registers. and the United States). Evidence in the United States derived from a direct matching of individual responses to administrative and tax records (Daniel Radner 1983). but no evidence on direct matching is available for other nations. these differences will cancel. and adjustments for differences in family size. but part of the difference can be explained by the fact that the totals are not fully comparable. In two nations (Australia and West Germany) there is an aggregate shortfall of some 20 percent. Still the available information indicates that total income estimates based on the surveys used for income distribution studies are about 90 percent of the comparable national income totals in six of the eight countries for which comparison data are available in the Luxembourg Income Study (Canada. the earnings and income data presented in this paper come from the same source. Level versus Trend Point in time comparisons of the level of inequality across countries impose much stronger data requirements than comparisons of trends in inequality. Wage and salary income is. household surveys is that they are incomplete in coverage of income. income definition. . if income components that are less (or more) well reported increase in significance over time. or if factors such as top coding have different impacts over time. The Netherlands. Moreover. and other measurement issues such 'as top and bottom coding of income. country-specific idiosyncrasies would affect levels of inequality but not trends in inequality. LIS collects none of its own data.15 While underreporting may be small for the most important income sources. 15 See Atkinson. then trends as well as levels will be affected.640 Journal of Economic Literature. Rather it takes data collected mainly by national statistical agencies and applies consistent measures and concepts across countries to produce greater uniformity in cross-national comparisons. But while the aim of the LIS project is to increase the degree of crossnational comparability. and Smeeding (1995a. adjustments for income sharing. quality of income reporting. this would affect both measures of central tendency and dispersion. complete cross16 For instance. Other data. Table 3. cover earnings well have but have very limited information on incomes. this may be of little comfort for distributional measures. As long as differences across countries (in income measures. indicates that the problem of property income underreporting is primarily found among upper income households with heads aged 65 and over. the United Kingdom.'6 Access to micro data in LIS also makes it possible to impose consistency on additional elements such as the unit of observation. such as that found in the International Social Survey datafiles. If underreporting were small but non-random. however. The Luxembourg Income Study (LIS) data sets were assembled specifically to overcome many of the problems addressed in these sections. As a result. generally well reported in all countries. importance of income components. Finland. Not all countries have been able to compare survey data with national accounts or other external data. which have been adjusted to make them comparable to the microdata sources.

The issue is not the existence of noise. Because the primary focus of the survey is on social attitudes only 22 questions are asked about economic and demographic characteristics.'8 The advantage of cross-national uniformity in the questionnaire has to be weighted against three disadvantages. see Atkinson.. the sample size in each year for each country is considerably smaller than in LIS (roughly 1. By 1994 the questions were being asked in over 20 countries. tionnaire is designed to be answered in 15 minutes. and the United Kingdom). a country survey's choice of a singular unit of income aggregation). . strong patterns emerge out of these admittedly noisy data. While these limitations must be kept in mind. surveys with very different focus and structure give broadly similar patterns. For instance. We return to this topic later in the paper. Rainwater. which surely exists in all data sets. While all income data used are continuous variables.'7 641 The International Social Survey Programme (ISSP) offers an alternative collection of repeated cross-sections on a number of countries. At this point there is a wide consensus in the research community that an important driving force behind the increase in family income inequality in the United States was the increased dispersion of earnings.000 or more in the LIS data sets used here). First. This drawback is particularly severe when the brackets are changed. the common ISSP questions are asked to a subset of respondents to the General Social Survey in the United States and to respondents to the British Social Attitudes Survey in the United Kingdom. and Smeeding (1995a). Also LIS has no contro over the questions asked in different surveys. Germany. the United States. Full comparability of earnings and income distribution data will never be attainable as long as surveys and institutions differ across countries. not all problems can be overcome. This sample size drawback can be partially overcome by pooling years. The second limitation of ISSP is that the ques17While the LIS project has gone to great lengths to increase data comparability across nations. though this is problematic when income distributions are changing over time. Karoly and Burtless (1995) study working age families and find that change in earnings inequality among men who work accounts for slightly less than half of the total change. For additional discussion of the technical characteristics of the LIS database. III. Earnings Inequality A vast literature. The major advantage of these data is that they are based on responses to a uniform set of questions attached to country-specific social surveys. making comparisons over time even more difficult. As we will show in the following sections.Gottschalk and Smeeding: Cross-National Income Inequality national uniformity will never be possible because the country surveys that form the starting point for LIS vary in focus and scope. reviewed in Levy and Murnane (1992).g. the LIS data cover a limited number of years and are thus unable to capture the effects of business cles on income inequality.19 19See Danziger and Gottschalk (1995) and Rebecca Blank (1994) for links between changes in the distribution of earnings and income. especially with small samples. For example. and while the LIS has up to 38 different categories of cash income for each nation. Danziger and Gottschalk (1995) attribute the majorityof the change in family income inequality to changes in the distribution of men's earnings. with the top bracket being open-ended. 18ISSP started with four countries in 1984 (Australia. Finally. some items such as self-employment income may be measured differently in different nations. most countries report income or earnings in income brackets.500 in ISSP versus 5. has documented the substantial increases in inequality of wage rates and annual earnings in the United States during the 1970s and 1980s.000-65. but the relative size of the signal to the noise. and because certain aspects of surveys cannot be adjusted ex post facto (e.

unlikely (especially among women) that the full difference in zero earners reflects additional persons at the bottom of the distribution. and underreporting of earnings at either tail of the distribution. without knowing the 20A few previous studies have used LIS data to examine wage and salary differences of heads of households across nations at a point in time (Gordon Green. For example. The United States and Canada stand out as the economies with the most unequal distributions of earnings for both males and females. who use data from the International Social Survey Programme. it is impossible to determine the effect on the unconditional distribution of potential earnings. measured either by the 90/10 or the 80/20 ratio. At one extreme one might assume that all zero earners came from the bottom of the distribution of potential earnings. while Table 1 presents estimates of the percentile ratios of the distribution of positive earnings this should not be confused with the distribution of potentialearnings for all persons. Vol. It should be recognized that differences in the distribution of positive earnings are very likely to be affected by these differences in the proportion of persons with zero earnings. 21 Earnings is used synonymouslywith wage and salaryincome. A.8 for males in Australia signifies that an Australian male at the tenth percentile earned a little more than half as much as the male at the median. not levels.20We exploit recently available data in the LIS database to compare earnings inequality across a variety of countries during the late 1980s and early 1990s. Information is presented for full-year full-time workers and all workers with non-zero earnings. For full-year workers these countries can be broken down into three broad groups. This largely reflects the lack of comparable data across countries.23For females. All estimates shown in Table 1 refer to annual earnings except for the United Kingdom where wages and salaries are measured during the survey week. Income surveys and. XXXV (June 1997) earnings that zero earners would have received if they had worked. However.22 While we focus on the distribution of positive earnings we also show the proportion of persons with zero earnings in Column (3) of Table 1. the LIS database do not usually contain separate measures of hourly wages. who use different data. Therefore differences in cyclical conditions may affect rankings of countries with small differences in inequality. Thus. John Coder. the PIO value of 56. however. Janet Gornick 1994.642 Journal of Economic Literature.21Table 1 presents summary measures of the earnings distributions in the nine countries for which the LIS database provides consistent data on annual before-tax earnings for males and females aged 25 to 54. Gottschalk and Mary Joyce 1996). Levels of Inequality Cross-national studies of earnings inequality have focused almost exclusively on trends. The years shown are limited by data availability in LIS. Our summary measures of inequality are based on earnings at selected percentile points because these are less sensitive to such inter-country differences as non-uniform top and bottom coding of earnings. hence. We also show the 90/10 and 80/20 ratios as sumof mary mneasures overall inequality. . and Paul Ryscavage 1992. Earnings at selected percentile points are measured as a proportion of earnings at the median. find similar patterns. 22 The cost of excluding the self-employed is that the distribution of earnings for the selected sample will be affected by this sample selection if the distribution of labor market earnings of the self-employed is different from the distribution for all other persons. It is. low earnings 23 This is consistent with Blau and Kahn (1996). Francine Blau and Lawrence Kahn (1996). Because it is impossible to separate labor market earnings from returns to capital in households with self-employment income we also exclude all persons in such households. For males this largely reflects considerably lower earnings at the bottom of the distribution.

1 1.0.3 43.4 188.0 12.0 2. livingin households income.9 2.9 3.9 30.4 3.7 167.6 - 54.6 37.5 3. The countries with the most equal distribution of male earnings are Germany and The Netherlands with 90/10 ratios for fulltime males of around 2.0 4.2 3.Gottschalk and Smeeding: Cross-National Income Inequality TABLE1 1990s: OECD COUNTRIES THE MID-1980S AND EARLY IN IN EARNINGS DISTRIBUTIONS SELECTED OF PERCENTILE MEDIANAND DECILERATIoSa 643 Country (1) Males Australia Canada Finland Germany Israel The Netherlands Sweden United Kingdom United States Females Australia Canada Finland Germany Israel The Netherlands Sweden United Kingdom United States Full-TimeWorkersc Percentagewith Full-Year.0 - 72.6 1. This low relative earnings.7 3.4 2.1 156.8 13.6 172.8 - 63.9 156.5 2.7 6. need not translate into low absolute earnings because the median is likely to be high in the United States compared to all of the .1 3. c Full-Year: full-timeweeks or more a year.0 47. Year ZeroEamingsb PlO P9 P90/P10 P80/P20 PlO (7) (8) (4) (5) (6) (2) (3) 1989 1987 1987 1984 1992 1987 1992 1986 1991 1989 1987 1987 1984 1992 1987 1992 1986 1991 20. These countries are followed by Sweden.1 58. however.1 2.9 40.0 186.0 2.Wagesare net of employerconwith zero self-employment to tributions socialinsurance(payroll taxes).5 4.0 1.7 179.3 34.Full-Time: or moreworkinghoursa week.1 5.2 156.6 17.1 193.8 152.7 163.8 47.6 228.2 181.0 - 2.7 1.4 3. with non-zerowage and salary at the bottom are matched by unusually high earnings at the top of the distribution. and the United Kingdom which have 90/10 ratios for males around 3.8 1.0 2.4 1.8 166.9 - AllWorkersd P90 (9) 161.9 2.2 5.1 23.0 174.4 4.7 2.2 5.2 180.3 1.0 169. tabulations Luxembourg a Personsaged 25 to 54.1 3.1 16.5 16.7 2.5 6.9 162. Full-year-full-time 35 50 workerscannotbe identifiedin the datafor Finlandor Israel.3 185.9 47.3 29.0 181. Australia.3 203.1 7.8 6.9 2.4 60.5 153.9 2.9 32.8 2.6 16.3 50.5 48.2 2.0 190.0 206.5 3.4 11.1 45.but grossof employeepayrolltaxes. The ranking of countries for women working full-time is similarto the rankingfor males with the exception of Germanywhich goes from being one of the most equal to being more similar to Australiaand the United Kingdom. d Allworkers income.1 25.7 for the United States.9 216.7 49.2 15.9 30.1 6.2 5.7 183.4 33.7 35.7 6. Table 1 shows that the earnings of persons at the tenth percentile are lower relative to the median in the United States than in any other country.0 P90/P10 P80/P20 (10) (11) 3.6 for Canada and 5.4 3.0 3.8 160.0 2.2 27.8 2.7 34.8 4.9 64.0 36.6 7.2 61.6 2.6 2.0 173.5 Source: Authors' of Income Studydatabase.4 62.8 23.1 35.5 69.0 1.4 3.5.5 4.6 176.7 1.0 - 71.3 22.7 28. b Percentageof all personsaged 25 to 54 with zero earnings.2 2.6 38.1 29.3 28.3 11.6 223.8 4.3 3.7 56.8 4.1 1.3 28.4 3.5 168. compared to 4.

persons at the bottom of the earnings distribution in the United States fare poorly not only relative to the median in the United States but also relative to persons at the PIO in other countries. a worker at the PIO in the German distribution earns 51 percent of the median earnings in the United States. B. While the last column shows that the United States indeed has the highest median male earnings among the countries shown. In contrast. these problems are considered small enough to warrant frequent comparisons in average (or median) incomes across countries. At the other end of the distribution a worker at the P90 in the United States distribution earns 193 percent of the United States median. This is by far the highest value with most other countries having a P90 at around 130 to 140 percent of the United States median.82 3.644 Journal of Economic Literature.70 Ratio of Real National Median To Real United States Median 72 79 90 69 84 92 100 The Netherlands 1987 Germany 1984 Australia1989/90 United Kingdom 1986 Sweden 1992 Canada 1987 United States 1991 Averageb 51 51 51 42 41 35 34 44 _ I 0 _ _ _ _ 50 100 150 Figure 1. For a similar comparison with similar outcomes.60 5.50 4. For example. Vol. see Richard Freeman (1994. Column 1 shows that the PIO measured in United States dollars is 24 See Robert Summers and Alan Heston (1991) for the basis of the estimates of purchasing power parity used here. Thus. 2-13). aNumbersgive real earnings (1991 United States dollars)as a percent of the United States median. Real Earnings Distribution Comparison for Full-Time Full-Year Males (all figures in 1991 United States dollars)a Source: Authors'calculationsusing the LuxembourgIncome Study database. While comparing the purchasing power of different currencies is fraught with danger. Trends in Earnings Inequality The literature on changes in earnings inequality in developed countries is now large enough to begin to piece together a coherent picture of similarities and differences in trends. bSimple 7 nation average. In order to get a rough comparison of absolute earnings at various points in the distribution it is necessary to translate earnings into a common currency.40 2. countries in Table 1. A few countries closely mirror the United States' experience while others seem to have avoided .24 Figure 1 presents the results. XXXV (June 1997) Low Earnings (P1O) Length of bars represents the gap between high and low income individuals EIigh Earnings (P90) 124 128 144 130 140 161 193 200 146 3. Figure 1 computes the median high and low income values used in Table 1 as a fraction of the United States median.52 84 Ratio of High to Low Earnings (Decile Ratio) 2.50 2. pp. higher in all countries than in the United States.10 3. In the same spirit we compare earnings at several points in the earnings distribution using the Penn World Tables of purchasing power parities which allow us to translate the values in Table 1 into 1991 United States dollars. a worker at the PIO in the United States distribution earns only 34 percent of the United States median. the differences and sometimes the rankings are quite different at the PIO and P90. Indeed only Canada has values nearly as low as the United States.

1 percent respectively.30The wage differential be26Unpublished data updating Table 2B2 of Karoly(1993). 27Authors' tabulations of the 1979 and 1989 Census of population for males with less than ten years of experience. For example..e. 29Increases in returns to experience were limited to less educated workers. Almost all studies of the United States use the Current Population Survey (CPS) to examine the distribution of weekly or annual wages for males. ln(P90) -ln(PIO)) reflects constant absolute differences (i.2 and 5. the increase in the college premium in annual earnings was even larger (from 30 percent to 54 percent). and lower tails of the distribution. Because hours worked by recent high school graduates also fell relative to the hours worked by college graduates. Rising earnings and wage inequality among male workers in the United States has led to a substantial literature documenting the trends and attempting to identify the causes. In contrast. 30Thomas Macurdy and Thomas Mroz (1995) show that the steepening of the cross-sectional experience profile is a result of downward shifts in the profiles of more recent cohorts. The former focuses on increases in wage differentials between high school and college graduates and between new entrants and older workers. at least temporarily. These studies find that wage growth varied substantially between the upper.Gottschalk and Smeeding: Cross-National Income Inequality the increasing inequality of earnings. most studies select only persons working full-time and full-year. 645 increased by 14 percent.28 The returns to experience also increased during the 1980s. Like them. constant p90 -PlO) which translate into larger relative differences as real earnings decline.26 Changes in the distribution of weekly and annual earnings were even larger. For example. 28These increases in returns to college during the 1980s are in sharp contrast to the decline in the returns to education during the 1970s. We follow Levy and Murnane (1992) by updating their summary of changes in the dispersion in the overall wage distribution. Part of the observed change in the distribution of wages reflects large increases in the returns to education during the 1980s.27 By 1989 the college premium in wage rates for this group had increased to 43 percent.e. in 1979 the hourly earnings of recent college graduates were 23 percent higher than the earnings of recent high school graduates. most studies further focus on the distribution of earnings of males. For example. between 1975 and 1992 the P75 ratio for hourly earnings of males in the United States increased by 10 percent and the P90 ratio 25 There are some exceptions.29 The result of these trends was a dramatic decline in the relative position of young high school graduates and high school dropouts relative to workers with more experience or education. Changes in Earnings Inequality in the United States. we further examine changes in returns to observable skills and changes in inequality within groups. middle. Gottschalk and Moffitt (1994) use the PSID.25 In order to concentrate on changes in wages and not changes in hours worked. not the steepening of cohort-specific profiles. a fairly consistent story is emerging. though not as much as the returns to education. The increase in within group inequality of relative income (i. In addition to the increased inequality between education and experience groups. the PIO and P25 ratios decreased by 3. .. While we are a long way from fully understanding the causes for these similarities and differences. studies consistently find large increases in wage dispersion even within skill groups. Because the large changes in labor force participation of women make it difficult to separate changes in the distribution of wages from changes in the composition of the female labor force. Within group inequality focuses on increased dispersion in the wage distributions within education and experience groups.

then one should attribute these indirect effects to trade. because much of the change in inequality in the United States reflects declines at the bottom of the distribution. Several studies estimate that the decline in the real minimum wage accounts for roughly 30 percent of the increase in the dispersion of wage rates (for example.. One set of explanations for the rise in inequality in the United States focuses on changes in institutional constraints. skill-based technical changes. If demand functions are not totally inelastic. Nicole Fortin and Thomas Lemieux. 6).646 Journal of Economic Literature. For example. Likewise. forthcoming). changes in institutional constraints. beginning in the early 1970s. management institutional change is probably the most important factor leading to the wage squeeze. such as a decline in unionization. however. A variety of changes in the economy. the impact of the minimum wage is not negligible. Michael Horrigan and Ronald Mincy 1993). in most cases. but because these explanations are potentially interrelated. if part of the decline in unionization or the technological change is the result of increased foreign competition. wages are set by corporations largely independently of market forces. declines in real wages. Persons in the upper part of the distribution experienced significant growth in real wages while those in the lower part of the conditional distribution experienced slight growth or. According to this broader institutional view. . But the decline in the minimum wage cannot explain the increase at the top of the distribution or increases in inequality within high education groups. leading to a smaller increase in the dispersion of earnings than wages. Explanations for Rise in Earnings Inequality in the United States. tween the 90th and 10th percentile increased within the distribution of wages of young and old workers and within the distribution wages of high school and college graduates. and the decline in institutions that limit the market (e. seems to have started earlier.32 During the 1980s the real minimum wage fell by 44 percent. increased immigration. specifically the erosion of the real minimum wage and the decline in union den31 For a brief summary of competing explanations see Danziger and Gottschalk (1995. Spillover effects are too small to explain the large changes elsewhere in the distribution. p. may reflect changes in market forces which limit the options for low skilled workers. This increase in hours will offset part of the decline in the wage. such as changes in industrial structure. The decline in unionization is another measurable institutional change which could have contributed to the increase in earnings inequality. However. XXXV (June 1997) sity. Vol.g. This is consistent with studies that attribute considerably less than a third of the decline in the share of earnings received by the lowest quintile to the decline in the real minimum wage (for example. employment will increase as the real minimum wage declines. While there is substantial agreement about the facts there is still disagreement about the underlying causes. ch. The net impact of unions on the distribution of earnings is 32 David Gordon (1996) views the change in the minimum wage and union density as part of a broader set of institutional changes in which corporations squeezed workers in reaction to increased foreign competition.3' Disentangling these explanations is inherently difficult not only because of data limitations. The increase in within group inequality. This is consistent with the decline in wages at the very bottom of the distribution. the fall in the minimum wage and the decline in unionization) are consistent with the increase in inequality. increased foreign trade. According to Gordon (1996. 206).

Increased international competition could have also increased the demand 33 The deindustrialization hypothesis and the foreign trade explanations over ap. The deindustrialization hypothesis focuses on shifts in derived demand for skilled labor resulting from shifts in the composition of demand for final products. p. Shifts in employment across sectors can account for only a fraction of the increase in skill intensity.Gottschalk and Smeeding: Cross-National Income Inequality ambiguous.g. Kevin Murphy and Finis Welch (1993. even if the estimated impacts of the decline in unions and the reduction in the minimum wage are not additive. For example. The baby bust made older workers a relatively abundant input and the continued increase in educational attainment meant that college educated workers were arriving on the labor mar- 647 ket in increasing numbers. John Bound. semi-skilled white-collar occupations). Their impact on the distribution of hours and employment is likewise ambiguous. accounts for little of the changes in the distribution of earnings among women (Freeman 1994. However. it is clear that changes in these two institutional factors had a substantial combined effect on the rise in earnings inequality. however. this change in industrial composition is not likely to be the major factor causing the increase in demand for skilled workers. 126) estimate that changes in industrial shares can account for only 16 percent of the overall change in demand for college educated workers. and skill biased technical change which all predict a shift out in the demand for skilled labor..33 While there is no dispute that the manufacturing sector shrank as the service sector grew. they may well double count the impact on inequality. While there is substantial agreement that shifts in demand are central to the causal story. Unions increase the wage differential between unionized and nonunionized workers with similar characteristics but lower inequality by reducing differentials among organized workers and by raising the wages of persons with characteristics associated with lower earnings (e. Estimates of the impact of the decline in unionization suggest that unions account for roughly 20 percent of the increase in male earnings inequality. however. at best it is a part of a larger story. The three leading contenders are "deindustrialization.34 While the deindustrialization hypothesis properly predicts that both quantities and prices for skilled labor would increase. The 1980s were clearly a period of sharp increases in returns to skill. The sharp increase in both the skill premium and skill intensity suggests that demand was shifting faster than supply." increased international competition. especially the high wage service sector. this still leaves open many competing demand side explanations. 34 Eli Berman. Fortin and Lemieux. Changes in market forces. measured either in terms of returns to education or experience. must also be part of the story. The decline in unionization. . forthcoming). These increases in the relative price of skilled workers occurred at the same time that labor markets were absorbing an increasingly large number of these workers. to the extent that some of the industrial shifts reflect changes in trading volume or patterns. Because studies of the impact of reductions in the minimum wage and declines in union density have focused separately on these institutions. and Zvi Griliches (1993) also conclude that changes in the industrial structure were not a major factor causing the increased inequality. The fact that the skill intensity increased at the same time as the skill premium increased presents a prima facie case for the importance of demand shifts in explaining changes in the earnings distribution in the United States.

with some of the more vocal critics of the factor content of trade finding that trade accounts for 40 percent of the decline in wages of less skilled workers. the distribution of wages. He argues that the appropriate counterfactual is what the prices of tradables (and hence wages) would have been if trade had not expanded.36There is a wider range of estimates in the literature that focuses on the effects of trade on output prices. 37 Robert Lawrence and Matthew Slaughter (1993) find little impact of changes in output prices in the 1980s while Jeffrey Sachs and Howard Shatz (1994) and Leamer (1996) find larger impacts. focuses on changes in relative effective supplies of less skilled labor (for example. Existing studies that use observed changes in prices may over or understate the relevant changes in prices because they include the impact of numerous factors other than trade. and Griliches 1993). The factor content of trade approach. As might be expected. The resolution of this ranging debate is of more than theoretical interest because estimates of the impact of trade 35 See David Richardson (1995). not endogenous factor quantities. who attributes as much as 50 percent of the decrease in demand for less skilled workers to international trade. for skilled labor.648 Journal of Economic Literature. not the increase in the volume of trade. Vol. Two countries with the same technology. 38 Factors other than trade may have either reinforced or countered the impact of trade. facing the same output prices. and Burtless (1995) for reviews of this literature. XXXV (June 1997) differ widely across methods. which puts downward pressure on their relative wages. determine relative wages (for example. see Edward Leamer 1996). Freeman (1995). The general consensus emerging from factor content of trade literature is that increased trade accounts for less than 20 percent of the shift in demand. This factor content of trade approach has been severely criticized by several trade economists who argue that exogenous output prices. hence. Imports embody skilled and unskilled labor which. Paul Krugman (1995) offers a possible reconciliation. Bound. . used largely by labor economists. This conclusion is based on the StolperSamuelson theorem that develops links between factor prices and output prices which are set on world markets. If the expansion of trade were small then there would be little impact on world prices of tradables. The exception is Adrian Wood (1994).37 While it is too soon to tell whether a common ground will be found. Therefore. that should be the focus of empirical analysis. no matter what happens to the volume of trade. will tend to have the same wage structure. the opening of trade increases the relative effective supply of less skilled workers. According to this trade theory.35 Two theoretical frameworks have been used to analyze the links between changes in international trade and changes in the wage distribution. The answer to this "what if' question about prices depends crucially on the volume of trade. regardless of their volume of trade. Because imports are less skill intensive than domestic production. determine the effective supplies of these two factors. it is the decline in the relative prices of less skill intensive goods. when added to domestic supplies.38 The question is how to infer the prices that would have occurred in the absence of trade. world supply and demand do not shift very much as long as new trade 36 See Freeman (1995) for a review. Krugman develops a CGE model based on commonly used supply and demand elasticities to infer the price changes associated with the increase in trade. see Berman. newly liberalized international trade with less skill intensive countries changes relative output prices in the domestic economy and. Leamer's estimate of 40 percent requires long lags because he uses price changes in the 1970s to explain wage changes in the 1980s.

Earnings inequality increased most rapidly during the 1980s. and Griliches (1993). David Howell 1995). In spite of having to pay more for skilled workers. . cannot explain the rise in skill intensity because these changes would have made less skilled workers cheaper. as well as BLS studies of the impact on individual industries. 39 Note that other explanations of the rise in inequality. Bound. 42 It should be pointed out that these two criticisms of the tec nological explanation for the growth in inequality cannot both be right. on the other hand. slow growth in average productivity reflects slow technological change. like technology. its importance is often inferred by ruling out other factors. such as the decline in real minimum wages or the decline in unionization. If technology cannot be directly observed then it is impossible to tell if it accelerated or not.S. It should be noted that all explanations based on increases in international trade leave unexplained the rising skill intensity in non-traded goods as well as traded goods sectors. 649 The first is that technological change is simply a label for our ignorance. the econometric evidence on changes in capital-skill complementarity and skill 41 For example. the two approaches give similar empirical results when the hypothetical changes in prices are used rather than the actual changes in prices. and Peter Cappelli (1993). because Neptune's orbit does not otherwise fit the predictions of theory.Gottschalk and Smeeding: Cross-National Income Inequality volumes are small relative to the total. then Pluto is observable. Not surprisingly. According to the critics. Bureau of Labor Statistics (1994). Given reasonable elasticities. "The argument for the skill-biased technical change hypothesis is a bit like inferring the existence of Pluto. If. prices have to adjust only a small amount in order to absorb the excess demand brought about by increased trade. only a small part of the observed change in relative prices is relevant to the question: What would prices (and hence wages) have been if there had been no change in trade? The value of this work is that it has the potential for bridging the analytical gap between the two approaches by showing that the volume of trade is indeed relevant to the key "what if' question. see Thomas Bailey (1988).40 Firms would bid up the price of skilled workers as their productivity increased relative to the productivity of less skilled workers. which also raises the wages of skilled workers if the technological change takes place in the skill intensive sectors. Furthermore. Berman.42 But the well-known series on productivity growth shows a deceleration in output per hour during the 1980s. employers in almost all sectors (traded as well as untraded goods) chose to hire more skilled workers39 Widespread skill biased technological change would be consistent with increases in both the skill intensity and skill premium within finely defined industry occupation cells. skill neutral technological change. hardly strong evidence for an acceleration in technological change. such as U. that cannot be easily observed. There are two primary objections to the technological change explanation.41 These direct sightings of Pluto make the technological change explanation more compelling. Therefore. this implies that technological change accelerated during this period." Likewise ubiquitous increases in inequality require some widespread force. As Steven Davis and Robert Topel (1993) have colorfully stated. leading firms to decrease skill intensity. Because changes in technology are difficult to observe directly. While this critique does have a ring of truth. 40 Trade theory models focus on sector bias. there are now a variety of studies of specific technological changes that have increased the demand for the more skilled. The second argument against the importance of technological change focuses on timing (Lawrence Mishel and Jared Bernstein 1996.

If supply had grown at a constant rate then demand would have had to accelerate during the 1980s in order to explain the increase in the relative wages of skilled labor. . Evidence for the 1980s indicates that increases in the dispersion of permanent earnings and increases in the variability of transitory earnings were roughly 44The basis for this distinction is the canonical error components model. XXXV (June 1997) with the a decline in the relative wages of less skilled workers but all these theories predict that firms would choose less skill intensive production methods. 46 In the simplest model. 45 Some institutional explanations. decreases in the real minimum wage are assumed to lower the long-run earnings of less skilled workers. But the supply of educated workers increased at a decreasing rate during the 1980s. there is nothing inherent in the argument that requires an acceleration in the shift in demand. the variance of log earnings is equal to the variance of the permanent component plus the variance of the transitorycomponent. in which observed log of earnings is equal to a time invariantperson specific permanent component and transitory component. the cross-sectional evidence that these theories are attempting to explain cannot distinguish between changes in permanent and transitory of earnings. increases in international trade. Deceleration in supply will do. Otherwise the explanation will not be consistent with the rise in skill intensity in the face of a rise in the skill premium. It is true that the literature's stress on demand side factors sometimes seems to ignore Marshall's dictum that it takes both blades of the scissors (demand and supply) to explain changes in prices. Vol.650 Journal of Economic Literature. For example increases in the return to education. and declines in unionization and the real minimum wage are all consistent 43 Mishel and Bernstein (1996) find no evidence of increased capital-skill complementarity while Claudia Goldin and Lawrence Katz (1996) find higher complementarity and increased skill bias when comparing the 1970s with the 1980s.46Longitudinal data is necessary to separate the relative importance of these two factors. may be consistent with greater earnings variability. Technological change remains one of the only factors that will result in a ubiquitous increase in the proportion of college educated workers employers are willing to hire in spite of the large increase in the college premium. While it may be sloppy language to attribute the rise in the wage premium to demand side factors.44Almost all the explanations reviewed thus far implicitly assume that the increase in inequality is a result of increases in the dispersion of permanent earnings.43 The critique based on timing properly corrects sometimes sloppy use of language but it does not deal with the heart of the argument in favor of the importance of technological change. such as decreases in unionization. Only technological change is consistent with rising skill intensity in the face of rising skill prices. not more skill intensive methods. whether caused by skill biased technological change or increased international trade. Likewise. The final explanation for the rise in inequality focuses on the distinction between increases in inequality of permanent earnings and increases in the volatility of earnings. which is consistent with an increase in the college premium in the face of non-accelerating growth in demand (Katz and Murphy 1992). will raise the permanent earnings of college graduates. bias technical change during the 1980s is mixed.45 While the focus of most explanations has been on factors that increase the dispersion of permanent earnings. What is required of any explanation for the increase in inequality is that the shift in demand be greater than the shift in supply. Deindustrialization. as we in fact observe.

This continues to fall in the 50 to 80 percent band. His more recent unpublished tabulations.270 to . 651 ing the 1980s. Changes in Earnings Inequality in Other Industrialized Countries A. This allows us to benchmark the change in inequality in each country to the corresponding change in the United States. but instability increased even among persons who stayed in the same job.48 The fact that increased instability accounts for roughly half of the increase in overall inequality and that we know so little about its cause opens an obvious line for future research. Thus.288/ . IV. 50 The classification for Canada is unaltered because the relative change in Canada of . but transitory fluctuations increased among unionized as well as non-unionized workers.50 While these two metrics do 49As Appendix Table A indicates. The latter paper defines transitory earnings as shocks that die out within three years. . Greg Duncan. however. Likewise the decrease in job duration is a contributing factor.320/ .320). 48 See Gottschalk and Moffitt (1994) for contributing factors.270 -1 versus .288).036 (from .Gottschalk and Smeeding: Cross-National Income Inequality equally important in accounting for the increase in inequality (both for annual and weekly earnings).49 Because the use of absolute changes is arbitrary. While this line of research points in a new direction. This suggests that the search for causal links should focus on factors associated with greater instability of both weeks and wages. we know relatively little about changes in market or institutional forces.119 change in the United States (. We also include studies that do not provide specific comparisons but where authors discuss their findings in light of changes in the United States. Our rankings for these studies. The table shows the absolute change in inequality in each country measured as a percentage of the absolute change in inequality in the United States.286 -1).286 to . Henry Farber (1995) finds little change in job duration for the period covered by Gottschalk and Moffitt. we focus on studies that contrast each country with the United States. and a whole host of differences that make comparisons across countries difficult. Blackburn and David Bloom (1994) show the variance of log earnings increasing by . The decrease in unionization seems to be part of the story.067 is 77 percent as large as the .018 in Canada (from . Similarities and Differences Table 2 provides a summary of changes in male earnings inequality dur47 See Gottschalk and Moffitt (1994) and Moffitt and Gottschalk (1995). reflects the authors'qualitativejudgement. inequality rose half as much in Canada as in the United States according to this measure. measures of inequality.that may have led to greater year to year (or week to week) fluctuation in earnings. and Smeeding 1996). show a clear decrease in duration for more recent years.we also indicate where classifications would be altered if we compared relative changes in inequality in the two countries. The table includes the ten countries for which we have information on trends in overall inequality and trends in returns to education (or occupation) and experience as well as trends in inequality within education and experience groups. Because studies of trends in inequality in other countries summarized in this table vary widely in populations covered. For example. The change in the United States over the same period is . the ++ in Column (3) for Canada signifies that the increase in overall inequality in Canada was 50 to 80 percent as large as in the United States. period covered.47 Part of the increase in the variability of earnings reflects increases in the variance of weeks worked but weekly earnings also became less stable. Involuntary job losses from layoffs and firings and voluntary quits both increased during the 1980s (Johanne Boisjoly. which are marked with an asterisk.

Israel saw very modest declines at the bottom but large increases at the top.51 The countries shown in Table 2 break down into four broad groups. these nations are outside the scope of our study. what is anomalous about the 1980s in Sweden (Douglas Hibbs 1990) and France (Concialdi 1997) is not the rise in inequality. The Netherlands. who exclude households with a foreign head. show a 2. Table 1) shows moderate increases in France between 1984 and 1989. they are the most commonly used methods.53 France. Japan. Russia. but the ending of a long period of rapidly falling inequality. Sweden. what we observe is a diversity of experiences but with almost all countries experiencing some increase in earnings inequality. When one goes behind changes in the overall distribution and starts to examine changes in inequality at different points in the distribution or trends in returns to education or experience. they started from very low levels. and Finland form a third group with positive but quite small changes in earnings inequality over the 1980s (though inequality started rising in several of these countries in the second half of the decade). and Israel. which was small. Whether one stresses the differences or the commonalities is like describing a bottle as half full or half empty.2). This group includes only the United Kingdom. Our absolute (relative) classification scheme implicitly assumes that equal changes in absolute (relative) changes in inequality are equally valued. there is no natural metric for comparisons. Vol. further similarities and differences emerge. and the former East Germany experienced considerably larger percentage changes in earnings inequality than the United States or the United Kingdom.54 While even the Nordic countries experienced some increase in earnings inequality during the 1980s. Clearly the United States was a leader in the trend toward greater inequality of labor market outcomes but most other countries experienced some changes. Table 4). However. and the United Kingdom (OECD 1993. Hungary. 55 Eriksson and Jantti (1994) show that inequality increased in Finland between 1985 and 1990 but this followed a sharp decline during the 1970s and early 1980s. Italy and Germany 51 Because any monotonic transformationof an inequality measure maintains ordinal ranks. Only the United Kingdom rivaled the United 56 Richard Hauser and Irene Becker (1993. While less skilled workers lost ground during the 1980s in most countries. the United States. Likewise.56 Thus. 52As a result of allowing market forces to influence wages. However. Table 5. 54 Tom Eriksson and Jantti (1994) show that the rise in inequality in Finland after 1985 was as large as the increase in the United States.7 percent increase in the Gini coefficient for West Germanybetween 1983 and 1990. resulting from a long secular decline in inequality. The first consists of countries that experienced at least as large an increase in inequality as in the United States. including Japan. the decline in relative earnings in the lower deciles is common across a large number of countries. Sweden. the hypothesis that all western industrialized countries experienced as large increases in inequality as the United States is equally unsustainable. XXXV (June 1997) form a small group that experienced no measurable increase in earnings inequality during the 1980s.52 A second group which experienced substantial increases in inequality but less than the United States and the United Kingdom includes Canada.55 Finally. The United States earnings distribution became less equal both because of growth at the top and decline in absolute and relative earnings at the bottom. and Canada. Each metric reflects an implicit social welfare function. While the absolute decline in real earnings at the bottom of the distribution is limited to Australia. The Netherlands. not exhaust all possible comparisons. . 53 In Canada and Australia the rise in inequality was largely a result of declines at the bottom of the distribution. Australia.652 Journal of Economic Literature. Pierre Concialdi (1997. The hypothesis that inequality increased only in the United States can clearly be rejected. the gains at the top of the distribution were more modest than in the United States.

Israel. Columns (4) to (6) of Table 2 summarize changes in returns to experience and education (or occupation). The differences suggest that these forces were either not equally strong in all countries or that they were countered by country-specific factors. experienced increases in inequality within skill groups and most countries experienced an increase in the returns to experience. There is certainly a prima facie case that countries with high union coverage or cen- 653 tralized wage setting were able to limit the growth in inequality. Countries with declining proportions of young people in the labor market should have experienced smaller increases in the experience premium. and Fortin and Limieux. OECD 1994b). Differences in wage setting institutions may account for some of the differences in growth in inequality. At the other extreme.Gottschalk and Smeeding: Cross-National Income Inequality States in the increase in the P90/P50 ratio. Exploring the importance of shifts in supply requires estimates of changes in 57 This assumes that shifts in demand were roughly equal across countries. as well as trends in inequality within skill groups. then this suggests that institutional constraints may not have been binding. The United States stands out in two important respects. . The question we ask is whether these supply shifts are sufficient to explain the small increases in relative factor prices in countries with centralized wage setting. (Freeman and Katz 1993. For example. Only the United Kingdom also experienced large increases in all three components. leaving relative wages constant. unionization rates declined in both the United States and the United Kingdom and wage bargaining became less centralized in the United Kingdom (David Blanchflower and Freeman 1992). countries with large increases in college enrollments should have experienced relatively small increases in the college premium as the growth in supply offset some of the increase in demand. as well as increases in inequality within groups. some countries may have experienced supply shifts that countered the demand shifts. it is the country with the largest increase in returns to education. If market forces were responsible for the diversity of changes in returns to education. All countries. except Finland. or at least that shifts in demand were not strongly positively correlated with supply shifts.57 Likewise. discuss the role of institutions. then countries with faster growth in college educated workers would have experienced smaller growth in the education premium. and the Nordic countries have fairly centralized wage setting and a high proportion of their workforce covered by collective bargaining agreements (Lars Calmfors and John Driffill 1988. Second. Similarly. First. and Italy. it experienced large increases in all three comin returns to both ponents-increases education and experience. If they are. and fewer young people competing for a dwindling number of jobs should have limited the decline in their wages. B. countries with a baby bust entering the labor market would have experienced smaller than average increases in the experience premium as less experienced workers became relatively scarce. Italy. Impact of Changes in Relative Supplies We start by turning to the crossnational relationship between changes in the rates of return to education and experience and changes in the relative supplies of persons classified by education and experience.) Germany. forthcoming. The commonalities suggest that similar factors may have affected these countries.

. bParentheses signify returns to higher paid occupations (e. Loveman. Teulings(1992) and Gottschalk Joyce(1995) Sweden 1979-89 1983-87 1984-91 0 +b ++ +b + - Edin and Holmlund(1995)d and Gottschalk Joyce(1995) * 1981-87 1979-90 1979-86 ++ (+++) (++) (+++) +++ +++ +++ +++ United Kingdom Katz.Loveman. dInequality was constant from 1974-84 in this study. Blanchflower (1995) Gottschalk and Joyce (1995) Germany * Abraham and Houseman (1995) Israel + 0 - ++ 0 0 mixed + na (0) (-) 0 + Oc +b 0 ++a 0 -c and Gottschalk Joyce(1995) Italy and Ericksson Ichino(1995) Japan 1978-87 1974-90 oc mixed 0 +++ Oa + na + + Katz.Blanchflower (1995) The Netherlands * Hartog. Vol.654 Journal of Economic Literature.g. returns to education are for recent labor market entrants.Blanchflower (1995) and Gottschalk Joyce(1995) +++ +++ ++ +++ aClassification for studies that compare country to the United States in same time period (for measures. XXXV (June 1997) TABLE2 CHANGES MALEEARNINGS IN INEQUALITY OVERTHE 1980s IN INDUSTRIALIZED COUNTRIESa Overall Earnings Inequality (3) + ++ +++ ++ ++ 0 + + ++a Country and Authors (1) Australia * Years (2) 1981-89 1981-85 1976-90 1979-87 1981-87 1980-90 1987-91 1976-87 1979-84 1983-88 1979-86 Returns to Experience (4) ++ +++ na ++ ++ 0 + +++b 0 +++ Returns to Education or Occupationb (5) mixed -+++ Earnings Inequality Within Group (6) ++ Borland(1992) Gottschalk and Joyce (1995) na -+++ Gregory(1993) Canada and Blackburn Bloom (1994) and Gottschalk Joyce(1995) Finland * Ericksson and Jantti (1994)c Gottschalk and Joyce (1995) France Katz.decrease greater than -10. see Appendix A): + + + increase in inequality at least 80 percent as large as in the United States + + increase 50 to 80 percent as large as in the United States + increase 10 to 50 percent as large as in the United States 0 increase from -10 to +10 percent of change in the United States .Loveman. Wherever possible. Classification for other countries based on authors' qualitative comparison. c Small changes over decade reflect decline followed by sharp increase after 1985. non-manual).. Oosterbeek.

which is again consistent with a market explanation for changes in the age premium. Changes in the education premium also show a negative relationship. The first comes from LIS which allows similar earnings functions to be estimated across a variety of countries using similar samples and variable definitions (Gottschalk and Joyce 1996).58 The lack of growth in the educational premium in Germany. and Blanchflower 1995). and Australia offer support for the importance of changes in the supply of college educated workers. At the other extreme. The data from LIS provide evidence of the importance of market forces. The Netherlands. which is consistent with a simple supply/demand explanation (Pers-Ander Edin and Bertil Holmlund 1995). Two types of evidence are available. The smaller increases in returns to education in Canada than in the United States are also largely explained by the substantially larger growth in the proportion of the work force with a college degree in Canada than in the United States. and Canada experienced relatively small growth in the relative supply of young workers during the years coverd by LIS (Gottschalk and Joyce 1996). This was followed in the late 1980s by a decline in the proportion of workers with a college degree and a modest increase in the returns to education.Gottschalk and Smeeding: Cross-National Income Inequality returns to education and experience in each country. not relative supplies. Likewise.59 Australia 58 See Freeman and Karen Needels (1991). This was accompanied by small increases in the age premium in these countries. The Netherlands experienced the largest yearly growth in the proportion of workers with a college degree and it experienced an actual decline in the college premia. In Sweden. however. Likewise. were the driving forces in Canada. the United States and Canada experienced relatively small increases in the supply of college workers. There is a systematic negative relationship between the size of supply shifts and changes in education and experience premium across countries. Sweden. the general pattern is similar to that found in LIS. and the United States are consistent with demand shifting faster than supply. The relationship is particularly strong for the education premia. Finland. The other sources of evidence on the 655 importance of shifts in supply are country studies. that changes in unionization and the minimum wage. This was accompanied by substantially larger increases in the college premium than in the above countries. In contrast. Sweden. The Netherlands experienced a large inflow of young workers and a substantial decline in their relative earnings. the United Kingdom. and Australia (during the 1970s) can also be explained by shifts in the relative supplies of college educated workers. large increases in the supply of college workers in Israel and Australia are consistent with the small increase in their college premia. These studies find fairly consistent effects of changes in the education composition of the workforce and some weaker support for the proposition that the age composition affected the experience premium (Katz. Canada. Gary Loveman. the ratio of workers with a college degree to those with a gymnasium degree rose steadily during the 1970s and early 1980s. While it is difficult to make comparisons across countries because concepts and measures often differ across studies. The second source of evidence comes from country-specific studies that provide less comparability across countries but greater detail on the specific country being studied. John DiNardo and Lemieux (1993) conclude. which is con- . 59 Katharine Abraham and Susan Houseman (1995) show that educational attainment continued to accelerate during the 1980s. Increases in the college premium in Japan.

This statement applies also to countries with centralized wage setting institutions. while institutional wage setting in an environment of high coverage rates in countries such as Germany and Norway points to the importance of unionized wage setting. 85 percent of the workforce was covered by collective bargaining agreements in 1980 but only 17. ranking European countries is more problematic. Most western European countries place a greater emphasis on distributional issues than the United States and many have centralized wage setting institutions that can be used to limit the impact of market forces. Hessel Oosterbeek. While it is easy to contrast the decentralized labor markets of the United States with the more centralized or unionized labor markets in most of Europe.656 Journal of Economic Literature. For example.5 percent belonged to unions (OECD 1994c. this seemingly minor institutional factor might be paramount. XXXV (June 1997) Again. and Coen Teulings 1993). The evidence from LIS and countryspecific studies strongly suggests that market forces played a role in limiting the increase in inequality. For example. Vol. in France. these studies at least provide the empirical basis for the presumption that market forces matter. . bargaining is fairly decentralized in France but the bottom of the wage structure is tightly controlled by a widely applied minimum wage (the SMIC). of these attributes. Impact of Differences in Institutions Given the large proportion of workers covered by collective bargaining agreements and the centralization of wage setting in many OECD countries. but not all. Countries differ in many dimensions. Japan has low union coverage rates and bargaining is at the company level but wage demands are coordinated through a nationwide Shunto (spring offensive) which sets guidelines that form the basis for company level bargaining. Returns to education also dropped substantially in The Netherlands during the 1980s while the proportion of workers with a college degree increased dramaticallyas a result of generous government subsidies for education during the 1970s (Hartog. how should one measure the degree to which wages are set by "institutions" in different countries? One common measure is the union density rate (the proportion of the workforce belonging to a union) but this measure potentially misses workers who do not belong to a union but who are covered by union-negotiated wage agreements.8). neither union density nor union coverage necessarily captures the degree to which wage setting is centralized. If market forces changed primarily at the bottom of the distribution. making it difficult to aggregate into a single summary measure. The question is how much of the diversity in trends in inequality can be explained by these institutional factors? sistent with the stability of the education premium in Germany. the picture is less clear for countries that share some. even in countries with centralized wage setting. offers a stark example of the impact of an accelerating supply of college workers. For example. While it is possible that some omitted variable is responsible for the negative correlation between changes in factor prices and changes in relative factor supplies. Table 5. The most rapid growth in educational attainment occurred prior to 1978. Thus. Furthermore. First. one should look beyond market forces to explain changes in the structure of wages. C. conceptual as well as measurement issues must be confronted. a period during which the education premium actually fell (Robert Gregory and Frank Vella 1992).

one may be tempted to infer that the particular institutions in that country were not effective in limiting the impact of market forces. But this is not a test of the institutional hypothesis. The diversity of institutional arrangements across countries suggests that using a single measure may be inappropriate. Almost all countries with institutional limits on market forces managed to have either small increases in inequality (France and the Nordic countries) or no change in inequality (Germany). Alberto Alesina and Roberto Perotti 1994). it is hard to test institutional explanations. therefore. Many countries with centralized wage setting or high rates of union coverage saw these institutions weakened during the 1980s. but this leaves a great deal of room for ex-post rationalization. If inequality increases. The importance of centralized wage setting is often based on the observed negative cross country correlation between the degree of centralization or unionization and the trend in inequality. Likewise. The second measurement issue focuses on the distinction between levels and changes. the union coverage declined in many countries during the 1980s. For example. examine two additional sources of information which may shed light on the importance of these institutions.Gottschalk and Smeeding: Cross-National Income Inequality Given the idiosyncratic nature of country-specific institutions. Surely some of the weakening of institutional barriers was a response to the increased pressure brought about by changes in market forces. At the other extreme. The key conceptual question is whether the level of the institutional constraint or the change in level is relevant. This could be used to 657 rationalize either stable wages (the institutions remained strong enough to block the impact of market forces) or rising inequality (the institutions weakened). it comes as no surprise that indices of centralization have come to different rankings depending on the weights they attach to different attributes (Calmfors and Driffill 1988. First. then it is inappropriate to treat changes in institutional factors as exogenous. With relatively few countries and a great deal of latitude in prediction. If this is the case. the United States and the United Kingdom. Differences in market forces could have been responsible for the small increase in inequality. is inequality expected to increase in countries with high but declining centralization of wage negotiations? Sweden entered the 1980s with bargaining at the national level but moved somewhat away from this highly centralized system in the early 1980s as employers withdrew from this arrangement. if changes in institutional structures are central to explanations of changes in inequality. The fact that countries with small increases in earnings inequality also had some form of institutional wage setting does not necessarily mean that these constraints were binding. then one must at least consider the possibility that causation runs partially in the opposite direction. It was only by starting with sufficiently centralized labor markets that countries like Sweden managed to end the decade with markets that were still as centralized as countries like Germany that did not experience similar institutional changes. two countries with decentralized labor markets. including the United Kingdom and Australia.60 Finally. We. . if these institutions were limit60 Fortin and Lemieux (forthcoming) are careful to stipulate that they focus only on change in institutional factors. experienced the largest increases in inequality. because it assumes that institutions matter and infers the effectiveness of the particularinstitution from the outcome.

The Netherlands. XXXV (June 1997) constraints were used primarily to protect those at the bottom. Relative wages did not fall as much as they otherwise would have. New Zealand. p. 92). For example.63 This agreement had the potential of limiting increases in inequality as well as reducing inflationary pressures. 62 Gottschalk and Joyce (1996) and Blau and Kahn (1996) find evidence of both market and institutional factors limiting the wage decline for the young in several OECD countries. and Sweden which are similar in magnitude to changes in the United States and the United Kingdom. Finally. some countries with coordinated wage setting institutions did not stem the tide of inequality. workers would find it increasingly difficult to find jobs as the demand for their services declined but wages were unable to adjust. the accord was weakened over the 1980s and the largest increase in inequality occurred at the end of the decade. This indirect evidence also suggests that institutional constraints were binding in some but not all countries. In the latter countries the increases in the P90/P10 reflect declines in the P10/P50 as well as increases in the P90/P50. Vol. It is commonly assumed that institutional 61 John Pencavel (1991) points to the paucity of sound empirical studies showing the employment impact of unions. Stephen Nickell and Brian Bell (1995) track changes in unemployment rates over a longer period and find changes in the relative unemployment rates of less educated workers in Germany. Italy. and Lemieux (1995) find changes in employment rates in France that are similar to those in the United States.62 Another piece of evidence that sheds light on the importance of institutional constraints in limiting the rise in inequality comes from distinguishing between increases at the top and declines at the bottom of the distribution. Australia experienced a large increase in earnings inequality (see Table 2). However. The fact that the PIO and P50 grew roughly equally is evidence that this constraint was binding. This is consistent with the hypothesis that these countries were using the institutional constraints at their disposal to limit the decline in wages of the young. If this is the case then we should observe relatively small declines in the P1O/P50. Australia's enactment of the accord between the government and trade unions allowed unions to coordinate and centralize wage setting. This suggests that the power to limit wage adjustments may not have been used to offset the increase in earnings inequality (Gregory and Vella 1992. The fact that the P90 grew faster than either indicates that France was also experiencing a shift in demand for high skilled workers (OECD 1996). ing prices from reaching market clearing levels then we might expect to see an increase in the relative unemployment rates of less skilled workers. but this wage policy was at the cost of increases in the relative unemployment rates of the young. Francis Kramarz. This suggests that institutional constraints and wages may not lead to higher unemployment rates for less skilled workers. and Germany but not for Australia. . For example. This could imply that institutions were important because inequality increased when they were weakened (Borland 1992. See Saunders (1994). p. However. David Card. Similar patterns of floors under the PIO are found for Belgium. This 63 By increasing noncash benefits. the accord may have had greater impact on cash inequality plus noncash income. Unless unions or public agencies were able to guarantee employment as well as wages. and Sweden.658 Journal of Economic Literature. Edin and Holmund (1995) find similar increases in official youth unemployment rates and hidden youth unemployment (increases in enrollment in government training programs). 16). the high and rising minimum wage in France (the SMIC) should have kept the bottom of the distribution from falling.61 Gottschalk and Joyce (1996) find some evidence that unemployment rates of the young did increase more than for the old in countries with centralized labor markets and small increases in the age premium. Finland.

In fact. the increases in aggregate unemployment rates were not unusually large in countries with more centralized labor markets. In this case. however. if this were the 64 For alternative explanations of the rise in unemployment rates in OECD countries. According to this line of reasoning. the relative unemployment rates of the young actually declined in almost all countries with centralized labor markets. of the wage distribution.65 Second. not the dispersion. As Chart 1 indicates. Institutions such as unions (whether centralized or not) could have bargained both on prices and qjuantities. aggregate unemployment rates grew in countries that were willing to accept higher unemployment rates for the least skilled in order to keep low skilled wages from falling to market driven levels. see Charles Bean (1994) and Guiseppe Bertola and Andrea Ichino (1995). 659 cause for the rise in aggregate unemployment rates in OECD countries. 65 This assumes that the institutional barriers to a decline in the wages of the least skilled did not also operate to limit quantity adjustments.64 This argument takes several forms. This interpretation for the rise in aggregate unemployment rates in OECD countries has two empirical implications. 67See Gottschalk and Joyce (1996) and Nickell and Bell (1995) for changes in relative unemployment rates by education as well as age. .Gottschalk and Smeeding: Cross-National Income Inequality difference in focus on the level or the change in institutions vividly illustrates the difficulty of subjecting institutional explanations to a rigorous test.67 Youth unemployment did become more of a problem in many OECD countries but this reflects an increase in the aggregate number of unemployed workers. Increases in aggregate unemployment in many OECD countries during the 1980s has also been cited as evidence that institutional constraints on wage adjustments were binding. The second form of the argument is that centralized wage setting institutions were used to counter the shift in relative demand away from less skilled workers (Nickell and Bell 1995).66 Furthermore. This form of the argument.13). According to this argument. the rise in aggregate unemployment should reflect larger increases in the unemployment rates of low skilled workers than in the unemployment rates of more skilled workers. institutions would not be responsible for the higher aggregate unemployment rates in these countries. D. Institutional Differences and Changes in Aggregate UnemploymentRates In the previous section we focused on relative unemployment rates of different groups to see if protected groups experienced increases in relative unemployment rates. While unemployment rates in many OECD countries did increase during the 1980s to levels similar in the United States. Chart 1. 66 OECD (1994c. the United States opted for stagnant real mean wages but high employment while other OECD countries opted for wage growth but paid a price in terms of higher aggregate unemployment rates. not an increase in the proportion of the unemployed who were young. First. The Nordic countries did experience large increases in unemployment rates in the late 1980s and early 1990s but this would require over five year lags in the effect of rigid wages on unemployment rates. has no implications for inequality because it focuses on the mean. then those countries experiencing the smallest increases in inequality should have experienced the largest increases in aggregate unemployment rates. The first is that these institutions were used to raise average wages faster than productivity. the increases in unemployment rates were much less pronounced in Nordic countries than in other OECD countries during the 1980s. the rises in aggregate unemployment rates were largely driven by increases in unemployment rates of more skilled workers.

Source:OECD(1994cJobStudies. Summary The strength of current research on changes in earnings inequality has been to develop a set of stylized facts that any theory must fit.5 3 2. . It is clear that the United States and the United Kingdom were not the only countries to experience an increase in earnings inequality. including in many countries with centralized labor markets. 43). Vol. these aggregate data do not provide strong support for the hypothesis that the higher unemployment rates in OECD countries reflect the consequences of distributional policies. The challenge is to understand why some countries managed to escape the forces of inequality which affected the United Kingdom and the United States to a much greater degree.p.50 0~~~~~~~~~~~~~~ Chart 1. While careful studies of the relationship between changes in relative wages and relative unemployment rates using micro data may provide stronger support for this theory. Youth Unemployment Rates Relative to Adults Table1. E. 0.660 76. However.4 4.5 5 .5 - ~ ~ ~~~~~~~~~~~~~LI1990 3. Supply shifts are clearly a part of the explanation.54 2 1.5 - Journal of Economic Literature.5- . Thus. the changes in these two countries were unusually large. this work remains to be done. XXXV (June 1997) 6- 1979 5.17.

268 0.238 0.23 4.305 0.230 0.05 3. they are also clearly part of the story.86 2.239 0.328 0.255 0. eGini coefficients are based on incomes which are bottom coded at one percent of disposable personal income and top coded at ten times the median income.67 5.01 3.294 0.335 0. In this section we broaden the focus to the distribution of post-tax and transfer income. While institutional constraints do not seem to have been binding in all countries.52 661 Ratio of High to Lowd Gini (Decile Ratio) Coefficiente 2. gSimple 19 nation average.350 _ * _ _ _ 1 _ I _ I _ f d UnitedStates1991 * 50 _ _ _ _ I_ 1 1 _ _ _ _ _ * 100 150 200 0.14 3.79 2.308 0.e. or decile ratio.306 0. The question should not be whether it was market forces (i.02 4.12 4.74 2. How does the distribution of income in the United States compare with that in other industrialized countries? Were changes in the distribution of labor market earnings matched by correspondingly large changes in the distribution of family income? Did changes in taxes and transfers cushion or exacerbate changes in labor market incomes? .5.as a percent of national median. cRelativeincome for individualswho are higher than 90 percent of the individualsin the county and lower than 10 percent of the individuals..249 0.230 0.229 0.30 4.311 0. bRelativeincome for individualswho are lower than 90 percent of the individualsin the country and higher than 10 percent of the individuals.285 0.78 2.Gottschalk and Smeeding: Cross-National Income Inequality Lowb (PlO) Finland 1991 Sweden 1992 Belgium 1992 Norway 1991 Denmark 1992 Austria1987f Luxembourg1985 Germany 1984 The Netherlands 1991 Italy 1991 Switzerland1982 France 1984 Canada 1991 Spain 1990 Israel 1992 Ireland 1987 Australia1989/90 United Kingdom 1991 58 57 58 56 54 56 59 57 57 56 54 55 47 49 50 50 45 44 36 0 Averageg 53 Length of bars represents the gap between high and low income individuals Highc (P90) 158 159 163 158 155 163 174 171 173 176 185 193 183 198 205 209 193 206 208 250 180 3. fAustriaexcludes self-employment income in its survey. Both are clearly necessary to explain cross-national differences in the growth in inequality. Income Inequality and Redistribution The preceding section has documented the substantial changes in labor markets which led to greater earnings inequality in the United States and many other industrialized countries.43 3.89 2.90 4.48 3. shifts in supply that offset the shifts in demand) or institutionalconstraintsthat limited the increase in inequality.95 3.80 2. dRatioof 90th to 10th percentiles.227 0.274 Figure 2.as a percent of national median. aIncome is household disposable income per equivalent adult using an equivalence scale factor of E=0.78 0. Comparisonsof Levels of Income Inequality:The Gap between Low and High Incomea Individuals(numbers given are percent of median in each nation and Gini coefficient) Source:Authors' calculations IncomeStudy usingthe Luxembourg database.227 0. V.

then the Commonwealth countries. B. XXXV (June 1997) and bottom coding.67. the better off are its citizens. 70Excluded from Figure 3 are nations with real median incomes below 70 percent of the United States median (Austria. 69See Atkinson (1995). and Smeeding (1995a. The United States person at PIO has 36 percent of the median. the United Kingdom. This results in a decile ratio of 5. with the United States. Such an ordering produces a very similar ranking of nations. At the top of the distribution. insensitivity to top . Israel. Belgium. The top deciles are noticeably lower in Austria (which excludes self-employment income). there appears to be a clear grouping of nations. individual earnings distributions may be quite different from household income distributions. Thus. A.662 Journal of Economic Literature. they have the disadvantages of focusing on only two points in the distribution. Spain. In the United States. and the United Kingdom). Such comparisons can be used to test the argument that the higher the average standard of living in a particular nation. While percentile ratios have some obvious appeal (e. which has similar low PIO values for earnings. a person in a household at the tenth percentile received 36 percent of the median income (Column 2).g.4) for such a ranking. just about that in the United States.69 Figure 3 presents the PIO. In Ireland. See Atkinson. Smeeding and Gottschalk (1996). has an 11 point higher value for PIO (47 percent) for adjusted household income in Figure 2. with the next largest being the United Kingdom with a value of 4. An alternative is to use a summary measure of inequality such as the Gini which is shown in the final column of Figure 2. Austria. ease of understanding). compared with values averaging 53 percent for the other nations. Israel.70 Just as 68 Still another method would involve rank orderings based on Lorenz dominance. most authors have made comparisons across nations using relative income measures such as those used in Figure 1. the income at the top decile is 209 percent of the median. the United States does not stand out to the same extent. and Rainwater and Smeeding (1995) for additional real disposable income comparisons across nations. P50. Measures of real or absolute income differences across nations again require comparisons of the purchasing power of currencies across nations. Absolute Levels of Income Inequality Because countries differ substantially in terms of real GDP per capita. Vol. Rainwater. Canada.68 While this ranking of nations according to the Gini differs slightly from that produced by the decile ratios. and southern Europe. and P90 in each country measured as a proportion of the United States median using the same Penn World Tables and methods used in Figure 1. and the BENELUX countries have the least inequality followed by central Europe. The United States has the largest value of the 90/10 ratio recorded in Figure 1. However. The lower part of the distribution of disposable income appears to be substantially different in the United States than in other countries.. Relative Levels of Income Inequality Figure 2 shows the distributions of post-tax and transfer income in 19 OECD countries for the most recent year available in LIS.78. Italy. indicating that a person living in a household at the 90th percentile enjoys over five and three-quarters times the income of a person at the 10th percentile. Figure 4. while a person at the 90th percentile received 208 percent of the median (Column 4). and Ireland at the bottom. Scandinavia. This difference owes in part to the relatively low values of PIO for the United States earnings distribution. and the Scandinaviancountries. shown in Table 1.

Figure 3 shows nations with real median incomes as low as 72 percent of the United States median. For instance. At the P90 level.Numbers give real income (1991 United States dollars) as a percent of the United States median. In terms of purchasing power parity (Figure 3).74 2. excluding United States. cRelativeincome for individualswho are more affluent than 90 percent of the individualsin the country and below 10 percent of the individualsin the country.78 2. At the other end of the scale. but no nation with a lower standard of living at the 10th percentile. "high income" Americans enjoyed real living standards far above those experienced in other nations. the person at the 10th percentile of the Finnish distribution still has a real income which is higher than that found in the United States (44 percent versus 36 percent).14 3. However. despite the United States' clear advantage at the median.90 4. the same person has a real income which is 44 percent of the United States median. Real Income Distribution Comparison (numbers given are percent of United States median income in 1991 United States dollars)a Source: Authors' calculationsusing the LuxembourgIncomeStudy database. .05 3. while Finland's median income person enjoys a standard of living that is 77 percent that of the United States (Figure 3).Numbers give real income (1991 United States dollars) as a percent of the United States median. we found for earnings. a person at the 10th percentile of the Finnish distribution had an income that was 58 percent of the Finnish median (Figure 2).80 2.30 5. Incomes are adjusted by E = 0. aUnit of aggregationis the household and units are weighted by the number of persons in the household.78 77 86 83 81 89 83 77 83 75 88 72 95 83 100 _ 50 _ 100 1 __ _ J 150 200 Figure 3.36 84 663 Length of bars represents the gap between high and low income individuals Ratio of Real Ratio of National High to Median To Low Incomes Real United (Decile Ratio) States Median 2.48 3.86 2. bRelativeincome for individualswho are below 90 percent of the individualsin the country and more affluent than 10 percent of the individualsin the country. In fact. the real income of Americans was almost half again as high as the average incomes of persons at the P90 point in their distribution.Gottschalk and Smeeding: Cross-National Income Inequality Low Disposable Incomeb (PlO) Finland 1991 Sweden 1992 Belgium 1992 Norway 1991 Denmark 1992 Luxembourg 1985 Germany 1984 The Netherlands 1991 Italy 1991 Switzerland1982 France 1984 Canada 1991 Australia1989/90 United States 1991 Averaged 44 49 49 46 48 48 44 45 42 47 40 45 38 36 0 44 High Disposable Incomec (P90) 122 136 136 128 137 143 132 136 132 163 138 174 161 208 250 146 3.79 2. dSimple average. the wider distribution of United States incomes means that "low income" persons living in households at the P10 level in the United States had lower living standards than did similarly situated persons in each of the 14 other nations compared here.95 3.01 3.5 where adjusted disposable income (DPI)=actual DPI divided by household size (s) to the power E: Adjusted DPI = DPI/sE.43 3.

See also Karoly (1995) for a similar trend in equivalence-adjusted family income from 1974-1993.S. 1979-1993 only. and thereby using an income definition which is similar to that used in the cross-national comparisons indicates an even greater increase in inequality.50 - Journal of Economic Literature.72 Ad71 See Jeffrey G. Department of Commerce. Relative incomes fell at the equality through 1970 and an increase after 1979. shows a secular decline in in- justing for household size and federal income and payroll taxes. households or families).664 .-. Income inequality in the United States increased steadily during the 1980s. which goes back to 1947. . This concept is the same as the one used in LIS.30 1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 Figure 4.. The figures are shown in Table 3. 73The trend in post-tax and transfer disposable income can go back only to 1979 due to data restrictions.71 Figure 4 shows the Gini coefficient for income before taxes but after transfers over the period 1967 to 1993. Williamson and Peter H. Trends in Income Inequality In this section we show how post-tax and transfer income inequality has changed over the past 10 to 25 years. C.73 The Gini values which underlie the adjusted disposable income line in Figure 3 has been reproduced in Table 3 along with the corresponding percentile points of the associated income distribution. By this and almost all other measures. Trends in the United States. Bureau of the Census (1995b. Table B-6). including earnings and transferincome. Plotnick and Eugene Smolensky (1992). but weighed by the number of persons in each household. XXXV (June 1997) Household Incomea .45 - . weighting by persons. and Goldin and Robert Margo (1991) for a longer-term perspective on the United States income distribution. adjustments for unit size and unit of income aggregation (weighting by persons.-- Adjusted Disposable Incomeb . bAdjusteddisposable income adds food stamps and other cash income components and subtractsfederal income and payroll taxes. It is based on households as an income aggregationunit.40 . inequality remained relatively stable from 1967 to the mid1970s and then started increasing. Income Inequality in the United States: 1967-1995 (Gini Coefficient for Family Income and Adjusted Disposable Income) Sources: U. 72 Data on inequality among families. Karoly and Burtless (1995) find the United States increase robust with respect to unit of observation. allousehold income series is weighted by households (all persons sharingthe same living facilities) and includes all sources of money income. Lindert (1980). Vol. .35 . We start with an overview of trends in the United States and then turn to crossnational comparisons.

2 155.0 204.331 0.78 5.342 0.3 52.8 52.85 5.6 167. the effects of the 1981-82 and 1990-91 recessions hastened the trend toward greater inequality. a 1993 income is topcoded at the 1983-1991 level of $299.2 189.2 205.1 163. they do limit comparisons of percentage change in Ginis because different weighting patterns and different equivalence scales will produce different absolute values for the Gini within and across nations. and reflects population weights bottom (P10) by about as much as they rose at the top (P90) over this period.2 37.S. are less likely to affect trends. What is unusual by historical standards is that inequality grew during the 1983-89 recovery as well.7 55.79 6. or method of calculation may affect the measure of the level of inequality.9 164.7 209.22 131 2.75 5.9 35.7 36.4 205.318 0.05 3.5 52.7 34.0 205.7 201.Gottschalk and Smeeding: Cross-National Income Inequality TABLE 3 TRENDS IN UNITED STATES INCOME INEQUALITY: 1979-1993 PERCENTILES OF ADJUSTED DISPOSABLE PERSONAL INCOME Relative Year PIO/P50 P20/P50 P80/P50 P90/P50 P90/PLO P80/P20 665 Gini 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993a 1991/1979*100 40.7 192.3 52.341 0.351 0.93 3.6 162.3 56.0 51.6 161.05 3. Bureau of Census (1995b).342 0. These differences.4 86 57. Department of Commerce.1 53.77 5.31 5.357 0.313 0.06 3.92 5.6 154.2 203.70 2.08 3.4 39.9 162.9 39.3 204.339 0. and Smeeding (1995a.07 3.1 113 4.000 per household from the 1990 census.5 34. The largest changes in income distribution took place in the United King74 Figures in Table 4 are based on Appendix Table B and are not comparable across countries because they come from a wide variety of studies.72 2.3 53.1 35.6 54.23 119 0.7 200.10 3.4 162.09 3.363 116 Source: U.6 53.310 0.6 207.61 5. (See also Burkhauser et al. which are explained in the notes to Appendix Table B. The available empirical evidence concerning recent trends in income inequality in different nations is summarized in Table 4.9 92 154.60 5..8 35.79 5.08 3.5 36.350 0.4 162.1 53.9 52. 1996.) Trends in OECD Countries.7 36.9 109 190.8 162. concepts of income.347 0.9 207. As a result. while the Gini value rose by 16 percent over the period. on this point.79 2.340 0.5 34.74 Countries are listed in order of changes in disposable income inequality (as measured by the change in the Gini coefficient) from largest to smallest change.2 159. However. 4) on this point.71 4.2 162. While household income inequality in the United States rose over this period.8 35. Differences in data.75 4. See also Atkinson.03 6. Rainwater. Ch.2 214.55 5.1 52. the decile ratio rose by more than 30 percent from 1979 to 1993.0 161.16 3.5 35.352 0.09 3. .

See sourcenotes at end of Appendix a Degree of changeis basedon Appendix TableB andis coded as follows: Rangeof Changein Gini Designation Interpretation smalldecline -5 percentor more 0 zero -4 to +4 percent + 5 to 10 percent smallincrease ++ 10 to 15 percent moderateincrease 16 to 29 percent +++ largeincrease 30 percentor more ++++ largeincrease extremely bMoststudiesshowchangesin marketincome inequality. (1995) Saunders(1994) Tachabanaki Yagi(1995) and Bauerand Mason(1992) and Atkinson. XXXV (June 1997) TABLE4 CHANGES IN MARKET AND DISPOSABLE INCOME INEQUALITYa Country Source Years Change 1981-91 1980-93 1980-93 1980-81 1989-90 1981-90 1981-89 1981-90 1981-89 MarketIncome Inequalityb +++ +++ +++ + + + + + DisposableIncome Inequality ++++ +++ +++ + + + + + United Kingdom GoodmanandWebb (1994) Atkinson(1993) United States U. Bureauof the Census (1995A) and Sweden Gustafsson Palmer(1993) Statistics Sweden (1995) Australia Saunders(1994) Denmark New Zealand Japan The Netherlands Norway Belgium Canada Israel Finland France Portugal Spain Ireland West Germany Italy Aabergeet al. Smeeding (1995a) Muffelsand Nelisen (1996) Epland(1992) Cantillonet al. while stillothersdo not discussmarketincome changesat all. Rainwater. (1994) Beachand Slotsve(1994) Statistics Canada(1994) LIS (1995) Uusitalo(1995) Concialdi(1996) Rodrigues(1993) LIS (1995) Callanand Nolan (1993) and Burkhauser Poupore(1997) Hauserand Becker(1993) and Brandolini Sestito(1993) Erikssonand Ichino (1995) 1982-89 1985-92 1980-92 1979-92 1981-92 1979-89 1980-90 1980-90 1980-87 1983-90 1977-91 + + + + +++ 0 0 na + + + + 0 0 0 0 0 0 0 0 TableB.The latterare marked"na.666 Journal of Economic Literature." . Vol.S.

and more than double the decline in the United Kingdom from 1949 to 1976. and that the 1990s will see a rise in income inequality more generally. and Belgium). 43) sums it up best: Among the other (non-United Kingdom) OECD countries. 1996.) While starting from a much lower level of inequality. This is almost double the increase over a similar period in the United States. and was not to be found in some countries. the degree of change in the distribution of family income was markedly different. (For the United States see Karoly 1995. Also. increased capital income. Inequality rose only slightly in three nations (The Netherlands. Clearly the remaining evidence in Table 4 supports this assertion. and eight other countries show no change in inequality in the 1980s. In the United Kingdom. the Swedish increase would be less in absolute terms than that found in the United States or in the United Kingdom. Moreover. see Jenkins 1995a. At the same time. See Appendix Table B. Smeeding. The same is true in New Zealand. while higher earnings for well-educated men and women.75 In Australia. Table 1. where there has been a clear trend toward greater inequality. It is also noteworthy that there ap75Bjorklund and Freeman (1994) find little increase in inequality among non-aged families with children over this period. However. . though all of these increases came during the late 1980s (Saunders 1994). Danziger and Gottschalk 1995. and Duncan. and Willard Rodgers 1994. the upward trend over the 1980s was slightly less than that experienced in the United States and Sweden.) While the trends in earnings inequality and in income distribution were similar in the United States and the United Kingdom. they compute only subgroup inequality trends. Were we to calculate absolute changes in inequality as measured by the Ginis. with the sharpest increases in the early 1990s. Only in Italy do we find a noticeable decrease in inequality during the 1977-1991 period. Rising earnings inequality among men and among twoearner families. the trend toward greater inequality in Sweden may have peaked in 1991 and has receded slightly since that time.Gottschalk and Smeeding: Cross-National Income Inequality dom and in the United States. In the United States the largest increases in inequality were concentrated in the early 1980s and continued into the early 1990s. it is certainly wrong to think in terms of a world-wide trend towards increased income inequality in the 1980s: the upward trend was exhibited to differing degrees in different countries. downward until 1981. The Swedish Gini increased 667 by about 20 percent between 1981 to 1993. (Compare Karoly 1995 to Atkinson 1996a. Sweden experienced a pattern of change in inequality similar to that in the United Kingdom. In the United Kingdom income inequality fell through the mid-1970s but the Gini coefficient rose by more than 30 percent between 1978 and 1991. those seeking to identify a common pattern for OECD countries other than the United Kingdom and the United States could say that continuing progression towards reduced inequality was the exception rather than the rule. Norway. Perhaps Atkinson (1996a. excluding the aged and persons aged 18 and 19. Denmark. p. it may be that these countries are lagging behind the United States and the United Kingdom. though the Swedish income distribution remained considerably more equal than either the United States or the United Kingdom in spite of these changes. then upward in the 1980s. and Japan. and the growth in the number of single individuals and single female headed families were the primary factors accounting for the increase in inequality in the United States since the mid-1970s. and self-employment income were all-important in explaining the growing income share at the top. rising unemployment and higher numbers of single parents were important in building a large group at the bottom of the distribution. and for the United Kingdom.

and Smeeding 1995a). Accountingfor the Changes The changes in the distribution of family income distribution that we have documented are a product of a complicated set of forces: changes in labor markets that affect earnings of individual family members. XXXV (June 1997) may have affected work and investment decisions.76 76 The primary drawbackof accounting exercises is that they can easily be misinterpreted because they do not make a distinction between endoge- . and The Netherlands. changes in social norms. Rainwater. with a high level of inequality even before the increase. which started from a much lower level of inequality. For example. Belgium. Moreover. and in Sweden. such as the women's movement and the purported decline in the work ethic among men. despite similar changes in earnings inequality (Card and Freeman 1993). changes in returns to capital. Portugal. government employment). and France. Canada experienced only mild increases in inequality of family income while the United States experienced much larger increases. which affect both family needs and labor market decisions. not wives' earnings. In Europe we find large secular increases in inequality in the United Kingdom. Ireland.g. which may have affected demographic and labor market preferences. governments themselves react differently to market income changes via changes in redistribution (tax and transfer) policy. and policy changes in tax and transfer programs which not only affected family income directly but also The inclusion of multiple income sources received by multiple individuals thwarts attempts to identify the causal links that led to variations across time and across countries in the distribution of total post-tax and transfer family income. such as the aging of the population and growth of single parent households. and via other policies (e. Aggregating earnings across all individuals in a household and adding other sources of income takes us from the distribution of individual earnings to the distribution of family income. smaller increases in Denmark. if exogenous increases in inequality of male earnings led wives of low income husbands to work more.. limited themselves largely to purely accounting exercises which decompose changes in overall inequality into a set of component parts that may reflect endogenous as well as exogenous changes. Researchers have. while Finland did not. therefore. with a secular decrease in Italy. pears to be no apparent relation between the trend over the 1980s and the overall level of inequality at the start of the period. but stasis in Germany.668 Journal of Economic Literature. Among the Scandinaviannations. but also how to structure living arrangements. Inequality fell in Italy and rose in the United Kingdom. Structural models that include all behavioral links are well beyond the scope of existing empirical work. Inequality increased both in the United States. demographic changes. D. There is ample evidence that family members take account of all sources of income available to the family in deciding not only how much each member might work. Ideally one would like to know how much of the change in inequality of total family income is caused by exogenous changes in each source of income. Vol. though both occupied. then this portion of the change in overall inequality would be caused by changes in the distribution of husbands' earnings. This would require a fully articulated model of behavioral responses. Sweden experienced a rapid rise in inequality in the early 1990s. intermediate positions in the mid-1980s (see Atkinson. Nor is there a consistent country group story.

However. ings. and Jane Waldfogel (1996) refer only to inequality among married couple families.77 For example. while Cancian.79 This difference in interpretation partially reflects differences in the ways in which changes in correlations between wives' earnings and other sources and changes in shares of income coming from wives' earnings are treated in these decompositions. and the share of the total from each source. how much of the increase in family income inequality occurs among families with the same work status of head and how much comes from differences in means across family types (see Jenkins 1995a). These decompositions also make no distinction between endogenous and exogenous forces. The overall Gini can be smaller than the Gini of each source. and Gottschalk (1993) conclude that most of the increase in family income inequality reflects increases in male earnings inequality and that changes in the distribution of wives' earnings played a more modest role. Stephen Machine. because these factors inherently affect two or more sources. the Gini coefficient for total income can be written as the sum of the products of the Gini coefficients for each source. CV2. these methods also raise a set of conceptual and measurement issues. the Gini correlation between the source and total income. Because the impact of wives' earnings may affect inequality among married couple families differently from inequality as a whole. 80 Comparing decompositions across studies of different countries is further hampered by differences in measures of inequality. Danziger. forthcoming). of total income can be written in terms of the CV2 of each source. and (3) the relative size of each source. not the entire population. for example. accounting identities do allow us to isolate the pieces. to say that changes in mean husbands' earnings account for X percent of the change in mean family income does not imply that family income would have dropped by that percentage if husbands' earnings ad not changed. which make it impossible to impose a consistent (though arbitrary) method of decomposing changes in inequality. Other sources could have responded to the decline in husbands' earndecompositions focus on population subgroups.Gottschalk and Smeeding: Cross-National Income Inequality While accounting decompositions can potentially offer insights into the patterns of changes in inequality. even overlooking the problems caused by behavioral links. Most accounting decompositions of income by source are based on identities between inequality of total income and three attributes of the joint distribution of the component sources: (1) inequality of each marginal distribution. For example. 77Other 669 that source. While it is not clear how to apportion the individual pieces to a specific source. and the share of the total from each source. it is often not obvious how to go from this to a meaningful accounting of the sources of the change in inequality. For example. Thus. Karoly and Burtless (1995) and Karoly (1995) attribute much of the rise in family inequality to changes in wives' earnings. even if the correlation is positive. .80 Several important styl79This study and a similar one for the United Kingdom by Steven Harkness. it is not obvious how to classify the effects of changes in the correlations among sources or relative sizes of each source. these problematic decisions can often lead to very different conclusions (Maria Cancian and Deborah Reed. the correlations between all sources.78 Alternatively the coefficient of variation squared. If inequality of a particular source increases then it is easy to attribute the resulting increase in overall inequality to nous and exogenous factors. Cancian and Robert Schoeni (1992) use consistent measures across a variety of countries. Furthermore. while identities allow the total to be decomposed into parts. 78 Lerman and Shlomo Yitzhaki (1985) show the Gini coefficient is a weighted average of Ginis of individual sources with weights that depend on the correlation and shares. it is difficult to draw inferences for the entire population from these studies. (2) correlations (or some other measures of covariance) between income sources. See Anthony Shorrocks (1982) for an early discussion of the conceptual issues.

welfare benefits and other cash transfers aimed at the poor in the United States fell relative to GDP from 1980 to 1990. in spite of the positive and rising correlation in spouses' earnings. it has increased in the United States. however. They lead to behavioral adjustments in labor supply that may affect market income inequality and they add (or subtract) income to yield the distribution of disposable income. much smaller than the impact of the increase in earnings inequality in the United States (Edward Gramlich.670 Journal of Economic Literature. Vol. it should come as no surprise that our ranking of trends in market income in Table 4 closely mirrorsthe ranking on the basis of individual earnings in Table 2. Market income includes the earnings of all persons in the household and all income from interest. These changes were. XXXV (June 1997) bracket amount and higher personal income tax exemptions which helped the working poor after the 1986 tax reforms. This might be expected because lower marginal tax rates at the top of the distribution were offset by a higher zero 81 The difference between the pre. then wives' earnings equalize the level of inequality in the United States. dividends. France. small compared to other modern nations.81 However. See Smeeding and Coder (1995). Thus.and post-fisc distributions is. Because earnings constitute the majority of market income for most households and because earnings among family members tend to be positively correlated. While the real value of unemployment compensation. Norway. this can account for only a small proportion of the trend in post-tax and transfer inequality (OECD 1994a. Changes in taxes also account for little of the trend in inequality in the United States. overall earnings inequality in Canada increased less than ized facts about the individual pieces stand out in the literature. increases in the Earned Income Tax Credit during the 1980s and into the 1990s raised the posttax earnings at the bottom of the distribution. Table lc and Chart 1). if the difference between the distributions with and without wives' earnings is taken as a measure of wives' contribution to inequality. For instance. and Portugal had small increases in the dispersion of market income as well as individual earnings. however. . While the correlation in earnings between spouses is low. Wives' earnings have become an increasingly large proportion of family earnings but wives' earnings are only weakly correlated with husbands' earnings. though not completely so. at least until 1990. Factors Associated with Changes in Other OECD Countries Table 4 contrasts the trends in inequality in market income and disposable income in the United States with the experiences of a number of OECD countries. rents and other market sources. changes in taxes and transfers account for only a small part of the trend in inequality during the 1980s and early 1990s. Furthermore. This weak correlation in annual earnings reflects the negative correlation between the labor supply of wives and husbands' earnings. changes in taxes and transfers during the 1980s reduced progressivity. Belgium. The Netherlands. E. family earnings are more equally distributed than husbands' earnings alone. While the overall tax and transfer system in the United States is progressive. However. taxes and transfers have two effects. But not all nations followed this pattern. Because disposable income is equal to market income plus transfers minus taxes. which partially offsets the high correlation in wages. while they are disequalizing with respect to the trend in inequality. Richard Kasten. and Frank Sammartino 1993).

Fritzell 1993. See Smeeding. Danziger and Gottschalk 1995). 83 The importance of demographic change in the United States is larger if we limit our analysis to working age families or to families with children. 86 Births out-of-wedlock also rose in these countries. Part of the reason why annual income inequality measures do not permit a greater role for changes in capital income is because they report only realized interest. and The Netherlands.83 The aging of the 82On Australia and New Zealand. and Atkinson and Sutherland (1993).2 percent in The Netherlands. and Torrey (1993). see Cowell and Jenkins (1993) and Duncan. European Commission (1995). these changes were particularly large.9 to 12. see John Bauer and Andrew Mason (1992). they account for less than half of the difference between 1971 and 1989 with most of their effect coming during the 1970s (Lerman 1996.8 to 15. With earnings more than 70 percent of market income. Rainwater.86 In summary. the United Kingdom. on the United Kingdom. Jantti and Danziger 1994.0 percent in the United Kingdom from 1981-82 to 1990-91. and for the United States. out-of-wedlock birth does not necessarily indicate low income in countries such as Sweden where many high income parents live together for long periods outside of marriage. Japan. Other market forces (such as capital income) and demographic changes also affected market income inequality. But market income changes and demographic factors do not tell the whole story. rents. However. this demographic shift served to increase inequality. on Japan. The increased level and correlation of women's earnings with men's earnings accounts for a large fraction of the change in family income inequality during the 1980s (Karolyand Burtless 1995). Increased receipt of capital income (including deferred capital income from private pensions) and a growing correlation between high capital income and high earnings acted to increase market income inequality in the 1980s in the United Kingdom. Because of nonrealization and deferral of most asset income. However.84 Demographic change during the 1980s also led to a sharp increase in the fraction of single parent families.4 percent to 14.6 percent in Belgium. .7 to 19. see Atkinson (1996a) and Jenkins (1995a).4 percent in Germany.85 Because single parent families have low average income. 7. and New Zealand. it should not be surprising that increased individual earning inequality and other changes in earnings within the household would be important factors in accounting for change in income inequality. Smeeding. annual income statistics ignore most changes in net worth and thus true capital income (or loss). and Rogers (1994). 9. ignoring interest paid and both realized and unrealized capital gains. population and policies that have encouraged early retirement helped reduce adjusted income inequality in many advanced countries because the level of adjusted income inequality among the aged is generally less than that found among the non-aged. Australia. changes in earned income inequality appear to be the prime force behind changes in market income during the 1980s in most countries. Most find the role of demographic factors to be smaller than economic factors (Jenkins 1995a. Karoly and Burtless 1995). and dividends received. Burtless and Karoly (1995) and Lerman (1996) attributed a larger role to demograhic and social factors than do others.Gottschalk and Smeeding: Cross-National Income Inequality the distribution of individual earnings (Charles Beach and George Slotsve 1994). and 13. But even then. see Saunders (1994). this factor was not nearly as important as changes in earned income inequality in any of these countries.82 671 Demographic and social change also played a role in accounting for the rise in inequality in OECD countries since 1970 though the relative importance of these changes is still unsettled. More than 25 percent of all households in major OECD nations 84 Exceptions are the United States and Germany. though to a lesser degree. In Germany. See the European Commission (1995). 85 The percentage of single parent families with at least one child under 15 out of all families with children under 15 rose from 9.

The Netherlands. family allowances. and increased their expenditures between 1980 and 1990. and education benefits. 1985. It includes disability and disability services. Denmark. and 1990-91 Source: OECD (1994b. While Norway spent less than 10 88 Social protection is a classification used by the OECD. Vol. and Finland all spent 10 percent or more of the GDP on social protection for the non-aged in 1980. In nations such as the United Kingdom. 1985. illustrates the diversity of experi87This includes legislated discretionary changes and the automatic response to changing market income circumstances of households. The Netherlands. and education benefits. ences.health benefits. this figure reaches 30 percent of income (Atkinson. Excludes all cash benefits to the aged and survivors. and other miscellaneous items. and 1990-91.87Chart 2. Countries differed dramatically both in the amount of social protection they offered working families at the beginning of the 1980s and the changes in expenditures on these programs. welfare benefits. unemployment compensation.672 16 14 -1979 12 10 8642 Journal of Economic Literature. depend on something other than earnings as the primary source of their gross incomes. family allowances.health benefits. employment promotion benefits. Expenditures on Social Programs Among the Non-aged as Percentage of GDP in 1980. XXXV (June 1997) 12 ~~~~~~~~~~~~~~~~D1990-91 ~~~~~~~~~~~~~~~~~~~ 0 f Chart 2. employment promotion benefits. . Ic) Note: These include cash benefits for disabilityand disabilityservices.welfare benefits. While these expenditures do not cover all forms of transfers to the non-aged population. and Smeeding 1995b). Sweden. which shows public cash expenditures on social protection for the non-aged as a percentage of GDP in 1980. Rainwater. they show the same general patterns that would be found using alternative definitions. and other miscellaneous items.88 Countries are ranked according to spending in 1980. and Sweden. Tables lb. Social protection in Chart 2 excludes all cash benefits to the aged and survivors. unemloyment compensation.

There were equally large changes in tax policies during the 1980s. While limitations on some types of social insurance benefits (e. Perhaps even more important were changes in the composition of spending and its effectiveness in replacing lost market income due to unemployment and disability. Norway. in nations 89 See Gottschalk. OECD (1994a). All deal with this and similar issues. In contrast. In fact. Ploug and Kvist (1994). were also the countries with the greatest social protection. the decade ended with the vast majority of countries spending more on social protection programs than ten years earlier. Gustafsson.4 percent of GDP on these programs in 1980 and even less in 1990.to medium-sized social protection systems whose transfer systems automatically reacted to the rising tide of market income inequality with higher outlays (Australia and Ireland) were unable to stem that tide. in the more "activist" European and Nordic social welfare states.g. Germany. The Nordic and northern European countries. both the level and trends in expenditures varied widely across countries. there were also increases in family benefits and welfare benefits for the long-term unemployed and for single parents in Australia. Japan spent only 2. Jenkins 1996. and Italy) experienced few new demands in their programs because inequality of market income grew only modestly. Karen Gardiner 1993). unemployment. Likewise..Gottschalk and Smeeding: Cross-National Income Inequality percent of GDP on these programs in 1980. social expenditure trends in the 1980s can be better described as adaptions to changing circumstances than as alterations in the basic systems of social protection (Niels Ploug and Jon Kvist 1994. and indexation formulae were made less generous in some European and Scandinavian countries. France. there is little relationand inship between retrenchment creases in inequality in most countries. generous long-term unemployment benefits significantly dampened the effects of higher unemployment on disposable income inequality. This undoubtedly reflects the fact that some countries that reduced their expenditures on the non-aged (Belgium.4 percent in Italy. expenditures on these programs fell from only 4. and the European Com- 673 such as Canada and Finland. Germany. reductions in benefit levels for the nonaged helped to exacerbate inequality. The growth in transfers during the 1980s partially reflects increased take-up rates as many of these countries experienced greater demands on social protection programs as a result of widening inequality of market income. These reductions were not only widespread but large in many of these countries. While the level of social spending is negatively correlated with changes in income inequality. Top federal income tax rates fell from 50 to 28 percent in the mission (1994). Thus.5 percent of GDP in 1980 in the United States and from 3. disability) were introduced. which had the lowest levels of inequality and then some of the smallest increases in income inequality. Hills 1995). . Some of the nations with small. and Finland. this fraction had risen to 14 percent by 1990. The top income tax rates were cut in 26 of the 28 industrialized countries surveyed in Messere (1993). And in two nations (the United Kingdom and New Zealand). The lowering of top income tax rates was not limited to the United States. European Commission 1994. and Edward Palmer (1996). Gardiner (1993).89 For instance. As a result of increased take-up rates and other policy changes. Denmark. even though overall social expenditures increased in both nations during the 1980s (Atkinson 1993.

and increased employment-related taxes levied on employers to cover higher unemployment outlays (Messere 1993.674 45- Journal of Economic Literature. though some rose again by small amounts (e.5). and taxable income definitions were broadened in many nations. Vol.0 0 ?000) 45 C 0 E0 ()o Chart 3. United States. OECD 1994).. increased VAT for general revenue. overall tax revenues rose in most OECD countries. 48 to 14 percent in Norway. and from 78 to 50 percent in Sweden. owing mainly to increased payroll taxes for social retirement. Note: Averagetax rates are total income and employee payrolltax as a percentage of gross cash income for households rankedby household disposable income adjusted for family size (E=0.g. 70 to 40 percent in the United Kingdom. Additional tax progressivity was introduced by changes such as the family benefit in the United Kingdom and the Earned Income Tax Credit in the United States. In fact. from 28 to 32 percent in the United States in 1993. disability and health care. Average Tax Rates for Households in the Second and Tenth Deciles in Selected OECD Countries Source: Authors'calculationsusing the LuxembourgIncome Study database. Many of these fami- lies paid higher taxes as a result of increased income subject to tax and an increase in other taxes that make up for income tax reductions. XXXV (June 1997) * 40 35 30 2520 1510 50 --$0 Decile 10 early 1980s Decile 10 late 1980s 1 Decile2 early1980s Decile 2 late 1980s C. and 50 to 55 percent in Sweden in 1993). Reductions in the top marginal tax rates did not necessarily lead to declines in taxes collected on families at the top of the distribution. Chart 3 shows the average federal income and payroll tax rates paid by families in the top and second decile groups of the distribution of disposable income in the early 1980s and late 1980s to early .

These figures show that in most countries (The Netherlands. The growing internationalization of the economy and labor market and government reactions to social and economic issues such as population aging. Finland. average tax rates increased for families at both ends of the distribution. Only then will it be possible to isolate the relative importance of exogenous changes in the distribution of pre-tax income from both exogenous and endogenous changes in taxes and transfers. 675 tries. Sweden. with the larger increases occurring at the top. France. . unemployment).g. and the United Kingdom. In the United Kingdom there appears to have been such a massive change in market income inequality that the British tax and transfer system was not able to overcome these forces and may even have contributed to them. While income and other tax changes have benefitted the well-to-do in a small number of coun90See Auerbach and Slemrod (1997) for a survey of behavioral effects of the United States 1986 Tax Reform. Australia. the United States. and Japan stand out as the three best examples. divorce. means tested benefits were increased in some nations to cushion reductions in other types of benefits (e. average tax rates increased at the top and decreased at the bottom. producing some offset to the disequalizing trend in market income. and Norway experienced smaller increases in inequality of disposable income than would be suggested by their changes in inequality of earnings and market incomes.. and Canada).90 The links between changes in tax and transfer policy and changes in the distribution of disposable income in different countries are certainly not well understood at this stage. This would occur automatically in countries with progressive tax and transfer systems. In Australia. and increased female labor force participation rate. This will require an explicit model of the endogenous increases in transfers that accompany declines in earnings at the bottom of the distribution and an explicit model of the impact of changes in tax and transfers on the distribution of pre-tax income. It is only in the United States and Norway that average federal tax rates declined at the top and increased at the bottom of the distribution. Canada. In nations with weak safety nets and less activist governments. It may also be true that expansion of the safety net for single parents produced a marked decline in their labor market activity.9' And there are clearly exceptions. Our reading of the limited crossnational information on changes in tax and transfer structures is that changes in taxes paid and transfers received were largely offsetting to the changes in the distribution of pre-tax and transfer incomes. Summary and Conclusions Concerns about earnings inequality and joblessness have moved to the top of the social agenda in many OECD countries. How much of these changes came from explicit policy changes as compared to changes in economic behavior of households is an important question that remains to be answered. West Germany. have added to our interest in how successful different 91See Paul Johnson and Steven Webb (1993) and Gardiner (1993). Here. See Moffitt (1992) for a review of United States evidence on this topic.Gottschalk and Smeeding: Cross-National Income Inequality 1990s. A pressing area for future research is to isolate the impact of changes in tax and transfer policies on the distribution of family income. changes in market incomes were dominant. but also reducing work effort. V. and Sweden and New Zealand experienced large increases compared to the small increase in earnings inequality.

New frontiers will include increased usage of national household income panel data sets which will follow the same individuals over longer time horizons. Eng: Cambridge ATKINSON. pp. By ANTHONY B. Yet. ATKINSON. 1978. "What is Happening to the Distribution of Income in the U. 2. AND HOUSEMAN. 132-60. ANTHONY B. 1994. Cambridge: Cambridge 41-63. 351-71. "Income Distribution in European Coun- tries. Washington. private transfers. an overall conceptual framework with empirically testable components is the next big step that must be taken.: The Case of Britain. Much has been learned from studies of annual cross-sectional household income microdata. ALESINA. 19-48. 1994." . Ideally. bridge.K.. bridge: Cambridge in from . 2. SUSAN N." Luxembourg Study Working Paper No. pp. BEACH. HENRY J. 1995. and soci al ANTHONY B. Sept. 244-63. pp.676 Journal of Economic Literature. Press." J. . DC: Brookings Institution.. 317-51. New York: Columbia U. "The Economic Effects of the Tax Re form Act of 1986. 1970. Eds. "Earnings Inequality in Germany. CHARLES M. 38(4). Income Wealth. pp. Lit. 35(2). June 1997. June 1994. and greater usage of cross-national labor force surveys and surveys that focus on expenditures and wealth." NCES Paper No. Turku. ANTHONY B.. THOMAS. "Education in the Texmation of Markets and Technology tile Industry. HOLLY. while great strides have been made to provide a factual basis for crossnational studies. Econ. "Remarques sur une Etude de AARON. 1993. "On the Measurement of Inequality. ." in Welfare and inequality: Essay on Britain and Europe. LEE AND SMEEDING. While most of the components of such a model were identified as early as the mid-1960s. "The Political Economy of Growth: A Critical economies are in dealing with these issues. 1992. ATKINSON.. Politics and the professors: The . 1988. TIMOTHY M. ANTHONY B. our progress toward building such a model has been slow (James Meade 1964). Better structural models of income distribution and redistribution that can be applied across nations are badly needed. Ed. CamU. 32(2). ATKINSON AND MARTIN REIN.?" Proceedings of the British Acad. Finland: Abo Akademi U. Survey" in J. JOHN AND MASON. pp. 1993." larization Queens U. KATHARINE ABRAHAM. Paris: OECD.. of Income. 133. G. If we are to understand why we observe the extent and pattern of inequality levels and trends that are extant in this review. "Income Distribution United Income States.. ROLF ET AL. mimeo. AND MICKLEWRIGHT. Atkinson (1996b) has made a first step at one component of such a model and has hinted at other components. GEORGE A. ROBERTO. 371-404. Economic transformation in Eastern Europe and the distribution of income. Theory. 589632. "Bringing Income Distribution the Cold" Econ. "European Employment: Income inequality and income mobility in the Scandinavian countries compared to the United States. ATKINSON. CamU. Oct. Dec." in Age. Sept. ATKINSON. AUERBACH. Press.: JOHN HILLS. 82. ANDREW. Rev. ANTHONY B. London: Macmillan. BAUER. A BEAN. U.. Econ. pp. pp. 1996. eds. "Two Nations in Early Retirement? security. Nov. Mar. Survey of the Recent Literature. JOEL. capital income.. 1995b. XXXV (June 1997) great society in perspective.. Press. J." in RICHARD FREEMAN AND LAWRENCE KATZ. REFERENCES AABERGE. Vol. and all forms of taxation) as well as the formation of income sharing units. an overall framework would simultaneously model the generation of all sources of income (labor income. Lit. pp. ALAN AND SLEMROD. new data resources have expanded to meet these interests. AND SUTHERLAND. 573-619. CHARLES R. and the TransforBAILEY. JOHN B. Econ. 1997. 1992. much less progress has been made in providing a tight theoretical frameworkto analyze these data. 1995. Income distribution in OECD countries: Evidence from the Luxembourg Income Study (LIS). JEAN. ALBERTO AND PEROTTI. 8(3) pp. 1995." World Bank Econ. RAINWATER. AND SLOTSVE. public transfers. in Europe and the . work. "Are We Becoming Two Societies? Income Poand the Middle Class in Canada. "Explaining the Distribution in New inequalities. 403428." J. BEGUE. "The Disof Income tribution and Wealth in Japan." Rev. 1995a. Over the past decade. Teachers College. pp.

of Antwerp. Aug. Lab.S. JOHN. LARS AND DRIFFILL. PAOLO. AND POUPORE. and Other Industrialized Nations During the 1980s. JOHN AND GRILICHES.. Jan. 1993. 1990. McKINLEY L. ANDREA. 26. Income Wealth.. G. 1993. May 1994. 1977-1991. CANCIAN." Luxembourg Income Study Working Paper No. pp. "The Shifting Inci- dence of Involuntary Job Losses from 1968 to 1992. 1995. REBECCA. Rev. 14-61. ROBERT. Indicateurs sociaux: 1985-1992. KRAMARZ. 1997. TIMOTHY AND NOLAN.S. 515-29. pp. 1996." Econ. "Generating Equality and Eliminating Poverty the Swedish Way.NY: Macmillan. Europe. ManufacturingIndustries: Evidence from the Annual Survey of Manufacturing. "Income Inequality and Poverty in Ireland in the 1970s and 1980s. ET AL. BRIAN. RICHARD "Unionism in the United States B. BLANCHFLOWER. CANCIAN. PETER. All-University Gerontology Center. FRANCINE "International Differences in Male Wage Inequality: Institutions versus Market Forces. LAWRENCE M. CARD. 31(1). eds. 355." in SHELDON DANZIGER AND PETER GOTTSCHALK. Princeton U. BJORKLUND. Syracuse U. BUHMANN. CANTILLON. JOHN G. BELL. 1993. Princeton. 1993. CANCIAN. "Assessing the Effects of Wives' Earnings on Family Income Inequality. ELI. RICHARD World Is the Middle Class? A Cross-National Comparison of the Shrinking Middle Class Using Kernel Density Estimates. "La Distribuzione dei Redditi Familiari in Italia. forthcoming. 800-16. CALLAN. "Female Earnings and the Level and Distribution of Household Income in Developed Countries.. 791836. BOUND. 1994. RICHARD. eds. BLANK." J. 1992 [revised July 1992]. Sept. GARY. Relations NJ: Industrial Section. "Bargaining Structure. D. Rev. BURKHAUSER." Ind. pp. June 1995. Banca d'Italia. 1994. MARIA AND SCHOENI. MA: National Bureau of Economic Research. 104(4). McKINLEY L. July 1996. "Where in the BURKHAUSER. ANDREAS AND SESTITO. 10-15. pp. AND FREEMAN." NBER Working Paper No. School. Inegalites 1979 de Revenus: Quasi Statu Quo entre pour les Salaries et les Inactifs. in Distribution CONCIALDI." Rev.. Statist. MARIA AND REED. 56-79. A." LIS The Maxwell Working Paper No. "Wage Inequality in Australia. BRIGITTE ET AL. U." et 1984 BLAU. June 1988. Jan. Syracuse. Nov. DAVID." Cross-National Studies in Aging Program Project Paper No." Rev. 46(3). 1993. 1994. BoISJOLY. Well-Being." ESRI Working Paper No."Changes in the Demand for Skilled Labor within U... RICHARD B.. 118. CALMFORS. AND Economies et Statistique." Luxembourg Income Study Working Paper No. 1993. of Chicago Press. RICHARD V. 34(2). pp. FRANCIS AND CARD. NY: Syracuse U. AND KAHN. Econ.Gottschalk and Smeeding: Cross-National Income Inequality l'OCDE Concernant la Repartition des Revenus dans Divers Pays. DANZIGER. 102. "Income France: The Mid-1980's Turning Point. "Changes in the Relative Structure of Wages and Employment: A Comof the United Canada and parison States. "Equivalence Scales. "Les CANCEILL." J. 230. AND BLOOM. Policy. Relat. "International Com- Estimates across Ten Countries Using the Luxembourg Income Study (LIS) Database. 84(2). SMEEDING." Working Paper No. PETER. "A Cross-National Comparison of Permanent Inequality in the United States and Germany. Polit. Econ. Eds. JOHN.. Statist. 4945. Econ. BRANDOLINI. and Poverty: Sensitivity . THOMAS. pp. pp. "Working Wives and Family Income Inequality Among Married Couples. 11542. pp. DAVID.." Servizio Studi. "Changes in the Structure of Family IncomeInequality in the U. Cambridge. 1995. LEMIEUX. and Other Advanced OECD Countries.84. pp. BERTOLA. 677 "Combining Consistency with Simplicity When Estimating Tax Incidence: Alternative Assumptions and Findings for Three Countries. AND VILLENEUVE. TIMOTHY M.. High in the 1980s?" in Poverty and prosperity in the USA in the late twentieth century. 371-74. JOHANNE. Chicago: U. 1976." Ind. France. 97-104. SHELDON AND parisons of Income Poverty and Extreme Income Poverty. Relat. pp. Feb. Dec." Rev. MARIA. DAVID AND FREEMAN. Corporatism and Macroeconomic Performance. BLACKBURN. 195-221. Apr. Centre for Social Policy. BERMAN. Luxembourg. DEBORAH. Apr..: WOLFF. 21-55. CLIVE AND ROSENBERG." NBER MacroeconomicsAnnual. V. PIERRE. . Walferdange. 6. 43. 1996. GOTTSCHALK. pp. 1994. 1993. 79(1). 84. Econ." Economie et Statistique. Winter 1992. ZVI." Small differences that matter. 1992.. DUNCAN.. DAVID G. The Maxwell School. BLACKBURN. BEA ET AL. 1988. pp. Econ. CHRISTOPH B. "International Trade and the Rise in Earnings Inequality.. Cambridge. Inequality.." in PE- Paper presented at the NBER Conference on the Labor Market in International Perspective. GIUSEPPE AND ICHINO." Mimeo. "Why Were Poverty Rates So DIMITRI B. "Wage Inequality and Unemployment: United States vs. Dublin: ESRI." Amer. BORLAND. pp. GREG J. ANDERS AND FREEMAN. Northwestern U. RisCAPPELLI. 33(2). Sept." NBER Working Paper No. Lit. "Are Skill Requirements from Production ing? Evidence and Clerical Jobs. MA: National Bureau of Economic Research. 4255. PAPADIMITRIOU AND EDWARD N. 13-66. BURTLESS. 65-74.

" Amer.K. DEINENGER. Perspectives. 102(1). pp. AND KATZ. of Chicago Press.. ERAL FOR EMPLOYMENT." World Bank mimeo. FORSTER. Press. Eds. Differences and changes in wage structures. INDUSTRIAL RELATIONS AND SOCIAL AFFAIRS. DAVIS. FREEMAN. DAVID M. KAREN. 1994. Jan. FRANK A. Uneven tides: Rising inequality in America. 1991. MA: National Bureau of Economic Research. 1993. 1992. ROBERT A. ALISSA AND WEBB. DANZIGER. RICHARD B. . 9(3). 1995. 1067-82." in RICHARD Forces. Princeton U. JON. COWELL. 1995. AND EDWARD PALMER. eds. LAWRENCE F. GOLDIN. pp. "AreYour Wages Set in Beijing?"J. Summer 1995. Fat and mean: The corporate squeeze of working Americans and the myth of managerial "downsizing. "The Swedish The Rise and Wage Structure: Fall of Solidarity Wage Policy?" in RICHARD FREEMAN AND LAWRENCE KATZ. "The Disin Finland. NICOLE LEMIEUX. 202-71. ROBERT H. GORDON. Nov. 252-57. TIMOTHY AND DUNCAN. A New Dat"Measuring Income Inequality: DC.MICHAEL. ed.S. 307-44.JOHAN. GOODMAN. Skill.. eds. GREG: SMEEDING. RICHARD B." Acta Sociologia. "The Demographic Situation in the European Union: 1994 Report. Perspectives. GUSTAFSSON. 86(2). Working under different rules." CE-84-94-202-EN- "Technology. FARBER. "Rising Wage Inequality: The United States vs. "Income Inequality in Five Na- tions. Cambridge. "What's Behind the Increase in Inequal- BJORN A. SHELDON AND GOTTSCHALK." Mimeo. Cambridge: Harvard Sage Foundation. "Economic Gender Gaps in the Industrialized Countries: Evidence from LIS. London: Institute for Fiscal Studies. PETER.. tribution of Earnings 1971-1990. 1993. LAWRENCE F. XXXV (June 1997) C. NICOLE M. Swansea: U. ERIKSSON. Chicago: U. "Equivalence Scale Relativities and the Extent of Inequality and Poverty. 1997. EDIN." EPLAND. FORTIN. Institutions. JANET. NY: Russell . KENNETH. AND NEEDELS. DINARDO. FIONA. Vol. 341. pp. PAPADIMITRIOU AND EDWARD WOLFF. "Interin the Trade and American national Wages 1980s: Giant Sucking Sound or Small Hiccup? Comment. Act."Poverty in OECD Countries. WILLARD." Econ. 2. Washington. 15-32. NJ: Industrial Relations Section. pp. Princeton. "The Great Compression: The Wage Structure in the United States at Mid-Century. Cambridge U. pp. of Wales.678 Journal of Economic Literature. 265-306. American unequal. eds. 1995. 17-26. Baruch College. FREEMAN. 29-62. CLAUDIA AND KATZ. "Inntektsfordelingen ity Over the Last Twenty Years: How Does the U. ed. 1993. . IFS Commentary No. "Diin the United verging Male Wage Inequality Do Unions ExStates and Canada. HENRY "AreLifetime Jobs DisappearS.: DIMITRI B. Compare?"Welfare State Programme Discussion Paper WSP/100." Working Paper No. ANin Italy: Market DREA C. TOM AND JANTTI."NBER Working Paper No. Econ. Sept. COWELL. TER GOTTSCHALK. 1981-1988: plain the Difference?" Mimeo. 93-10. Belgium: Office for Official Publications of the European Communities (EC). New York: Russell Sage Foundation. STEVEN J. London: London School of Economics. "The Changing Pattern of Inequality: The U. 36(1). AND LEMIEUX." New York: Martin Kessler Books. 1995. Finland: Abo Akademi U. 1997 COULTER. pp." Paper prepared for the United Nations Human Development Report Office. Labor market institutions "Institutional Changes and Rising Wage Inequality: Is There a Linkage?"J. and the Wage Structure: Insights From the Past. eds. PER-ANDERS AND HOLMLUND. Press. THOMAS. Rev. 42. U. AND ICHINO. GOTTSCHALK. 1994. 1995. of California at Irvine. pp.. "Wage Differentials and Inflation. 1992. STEVEN. for poorer. EkonomiskeAnalyser. pp. - AND PALMER. 19731993. ERIKSSON. 1993. forthcoming. 1994. JOHN AND LEMIEUX. 1993. 1994. BERTIL. Econ. AND JENKINS. CHRISTOPHER L. 214-21. Other Advanced Countries. 1995.." in RICHARD B. KLAUS AND SQUIRE. 1993 U. New York: Macmillan. and the distribution of wages. 1994. of Montreal.1994. FREEMAN. the Middle "Skill Differentials in Canada in an Era of Rising Labor Market Inequality. . JOHN. 10. CUNY. Montreal: U. GORNICK. "Whither RODGERS. FRANK AND JENKINS. BJORN GUSTAFSSON. 20. i 80-CErene. 1973-1992: A semiparametric approach. ." OECD Social Policy Studies. May 1996. RICHARD B." Economics Department Discussion Pa- per No. Econ. The distribution .. Belgium: Office for Official Publications the European Communities (EC). FRITZELL. PETER. STEPHEN P. FORTIN. J. AND TOPEL.. "Social Protection in Europe. Oct. Brusof sels. For richer. KAREN. in the 1980s. 3817." CE-79-93-978-EN-C. EUROPEAN DIRECTORATE-GENCOMMISSION. "A Survey of Income Inequal- Class?"in Poverty and prosperity in the USA in the late twentieth century. of economic welfare in the 1980s. STEPHEN P. Brussels. eds. Turku. THOMAS. MARCUS. 1993." Paper presented at NBER Conference on United States and Canadian Labor Markets. AND DINARDO. EDWARD. CLAUDIA AND MARGO. 1994. 1996. Econ. pp. GARDINER. 2. abas. 1993." Brookings Pap. GOLDIN. ing? Job Duration in the United States. THOMAS. FREEMAN AND LAWRENCE KATZ. 1991. FREEMAN. M. 1995. 1994.

: INGA PERSSON.: C. PAUL.12(1). Eds. pp. JOHN F. JOHN. 407-13. HILLS.." Demography. Mar. "Rising Inequality in a Salary Survey: Another Piece of the Puzzle. Press. pp." Fed. 1-15. SHELDON DANZIGER AND PETER GOTTSCHALK. 1995.Gottschalk and Smeeding: Cross-National Income Inequality ity?" (1997a) in PETER GOTTSCHALK. 679 HORRIGAN. "Age. 1971-86. eds. .FRANK. ??. Gerontology Center. 1975-1991. 1995. of Gothenburg." Economica. Ed. 29-46.. 1993. pp.. 181-200. ALAN. LOVEMAN. 1993. AND MINCY. "Demo- Development of the Income Distribution in the Federal Republic of Germany During the Seventies and Eighties.. ch. IRENE." Econ. Feb. MICHAEL W. pp. CODER." Econ." Quart. 217-72. ROBERT. LYNN A.London: Cambridge U. Oct. 1996. HARKNESS." in Generating graphic Change." in SHELDON DANZIGER AND eds. 379405." York: Joseph Rowntree Foundation. Pol. J. pp. pp. "Growing Inequality in the 1980s: The Role of Federal Taxes and Cash Transfers. pp. STEVEN. pp. 1963-1987: Supply and Demand Factors. 1993." Mimeo. KEVIN. Econ. 1993. GOTTSCHALK. EDWARD M. July 1992.. "Explaining "Changes in Swedish Inequality: A Study of Equivalent Income. JOHNSON. MARCUS AND DANZIGER. 4. FRANK. 29-63. Letters. 1990. GREGORY. AND COWELL. . HESSEL TEULINGS. Dec. BJORN AND PALMER. 1995a. GARY. KAROLY. STEVEN: MACHIN. pp. 1992. RAND Corporation. Econ. GOTTSCHALK." in SHELDON DANZIGER AND PETER eds. 251-75. Rev. pp. 1995. equality in the welfare state: The Swedish experience. 1995. pp. HOWELL. 48(1). AND EDWARD PALMER. 103(415). 19-97. 49(4). PETER AND MOFFITT. Santa Monica. pp. 19591989. pp. FRANK. J. PAUL AND WEBB. ERICA L. Dec. "A Comparison of Changes in the Structure of Wages in Four OECD Countries. 62(245)."Ind. Oct. pp. STEPHEN AND the Growth in UK Income Inequality: 19791988" Econ. 891900. Evidence from Kernel Density Estimates. Act. 1984-1989.: PAUL JOHNSON AND KLAUS F. "International Comparisons of Earnings Inequality for Men in the 1980s.. HAUSER. and Workers in the United States: A Twenty-five Year Perspec. Bank of Cleveland Working Paper 9121. Lab. pp. Feb. LAWRENCE. Econ. Income Wealth. JANE. 48-64. 38(1). 1994. LAWRENCE AND MURPHY. "The KAROLY. 2. Rising Earnings Inequality and the Distribution of Peronal Well-Being. 1993. " Did the Middle Class Shrink during the of Real Wage and Employment Changes in the Australian Male Labour Market. J. 32(3). J.. 225-49. DOUGLAS A.K.. 1995. 26." in. "Aspects Distribution and Why It Matters: An Economists's Perspective. RICHARD AND BECKER. "Changes in Earnings Inequality in OECD Countries: The Role of Market and Institutional Factors. HIBBS. "Wage Dispersion and Trade Union Action in Sweden. 33-60. Res. KASTEN. Labor Market. Wages and Education in The Netherlands. RONALD B. SHELDON. "ParametricEquivalence Scales and Scale Relativities. GUSTAFSSON. ZIMMERMAN." Mimeo. eds. ?? KATZ. JENKINS. pp... AND among Families. JANTTI.1993. mimeo. 1995. GROSHEN. The Maxwell School." Brookings Pap. 1997. PETER GOTTSCHALK. Boston College. pp. Econ. "Inquiry into Income and Wealth Volume II: A Summary of the Evidence. Feb. Mar.. PETER AND JOYCE. . GORDON. tive. JR. EDWARD. 1994." in Familien zwischen gerechtigskeitsidealen und benachteiligungen. HARTOG. Relat. "Women's Pay and Family Incomes in Britain 1979-1991. Australian National U. Boston College." in Labour markets in an ageing Europe. JENKINS. 1994. 1980s? U. pp. Ed. "Accounting for Inequality Trends: Decomposition Analyses for the U. pp. BADELT. KRUEGER. "Is Earnings Inequality Also Rising in Other Industrialized Countries?" Mimeo.. MARY. GOTTSCHALK.LYNN A." Cross-National Studies in Aging Program Project Paper No. ROBERT AND VELLA. "The Growth of Earnings Instability in the U. eds. 1996.S. 35-78. "The Collapse of Low-Skill Male Earnings in the 1980s: Skills Mismatch or Shifting Wage Norms?" New School for Social Research. "How Computers Have Changed the Wage Structure? Evidence from Microdata. 1991. OOSTERBEEK. U. Cambridge: 158-80. 42935. JOOP. DAVID. WALDFOGEL." in New inequalCambridge U. 74-98. 1993. GRAMLICH." Rev.: JOHN HILLS. AND BURTLESS. Aug. 108(1). STEPHEN P. GREEN. "The Trend in Inequality ities: The changing distribution of income and wealth in the UK. 1993. "The Within-Household SAMMARTINO. CA. "The Minimum Wage and Earnings and In- come Inequality. "Recent Trends in the UK Income Distribution: What Happened and Why?" Oxford Rev." in RICHARD FREEMAN AND LAWRENCEKATZ. Syracuse U. Sept. Press. July 1994. AND RYSCAV- AGE. All-U. KATZ. 1992. Vienna: Bohlau-Verlag.. "Anatomy of the United States Income Distribution: Two Decades of Change. Press.K." Quart. . GUSTAFSSON. 1996. "Changes in Relative Wages. 1995b. pp. 182-211. DAVID. Ed. Oslo: Norwegian U." Mimeo. COEN. RICHARD AND "Child Poverty in Sweden and the United States: The ETfect of Social Transfers and Parental Labor Force Participation. GARY AND BLANCHFLOWER. 104(425). 107(1).. Individuals. STEPHEN P. BJORN A." Mimeo.

1976. MEADE. Econ." Amer. "Income OECD Countries. 30(3). KENNETH. LIS. THOMAS. NIELS AND KVIST. Econ. MESSERE.S. Sept. SHLOMO. 1333-81. JEFFREY D. pp. 1994." Brookings Pap. MA: National Bureau of Economic Research. Distribution in SAWYER." NBER Working Paper No. MACURDY. BRIAN. ch. RADNER. Tax policy in OECD countries: Choices and conflicts." in PETER GOTTSCHALK.. 3355. PETER. 1-84. 9(3). MUFFELS." Luxembourg Income Study Working Paper No. Co. "Inequality and Poverty in the United States: 1900 to 1990. 5355. Supplement 1996. 67(1). equality. 1995. pp."Rev. and the ownership of property. Cambridge: U. Labor markets under trade unionism: Employment.. RAINWATER. and hours. JOHN.." Oxford Rev. RICHARDSON. ROBERT AND YITZHAKI. wages. Act. The Netherlands: BV Publications. Econ. 1983. "Technology and the Wage Structure: Has Technology's Impact Acce erated Since the 1970's?"Economic Policy Institute. 101-32. 3." Paris: OECD. Apr." in Employment outlook. Johns Hopkins U. MISHEL." Mimeo. HOWARD Distribution of Economic Well-Being in The Netherlands. July 1995. KEVIN M. Sept. 30914. Efficiency. 10-61." Social Policy Studies Paper No. "Wage Inequality International Competition and Technological Change: Theory and Country Experience. ROBERT A. "Earnings Inequality. "Adjusted Estimates of the Size Distribution of Family Money Income. suring Macroeconomic Shifts in Wages from Cohort Specifications. ROBERT J. 157-78. Amsterdam. JARED. analysis. Berkeley. J. pp. ROBERT.. GOTTSCHALK. Chicago: Markham Pub. Trade. LERMAN. RUUD AND NELISSEN. Summer 1995.. The Maxwell School. Econ. SAUNDERS. 135-46. strategies. Press. 1995. "IndusAND PETER trial Change and the Rising Importance of eds. Skill. 1996. EDWARD PALMER. STEPHEN AND BELL. 1992. Recent trends in cash benefits in Europe. "The OECD jobs Study: Evidence and U. "Inequality Decompo- . "The Measurement and "Trends in the Autocovariance Structure of Earnings in the U. SACHS. PLOTNICK. Trade: How to Think. pp.. Act. What to Conclude. Microeconomics. Of California-Berkeley. MURPHY.S. 2. NY: Syracuse U. 30(1). . ch. Paris: OECD. KRUGMAN. ROBERT. LEE AND SMEEDING." in SHELDON DANZIGER FINIS. "Economica.. Bus.680 Journal of Economic Literature. 1989. 1993. Income Inequality in Australia and New Zealand in the 1980s. 127. Paris: OECD. "The Impact of the Changing Inequality: Changes in the 1980s. Econ. 59-108. 1996. BJORN A.: 1969-1987. EDWARD E. Earnings Level and Earnings Inequality: A Review of Recent Trends and Proposed Explanations. 1993. U. Low-Paid Employment and Earnings Mobility" in Employment outlook.. PAUL. ROBERT AND SMOLENSKY..S. MOFFITT. 161-210. 1995. 1993.. Perspectives." Working Paper #193. May 1996. AND GOTTSCHALK. 1985. CA: U. 5. "Incentive Effects of the U. TIMOTHY M. "Rising on the Tasman Tide: J.. . "New Orientations for Social Policy. Explanations. Copenhagen: Danish National Institute of Social Research. of Cambridge. 12. pp. pp. 1992. PENCAVEL. Harmondworth PEN. "Technology. 1. Econ. AND "Tradeand Jobs in U. 40-62. Center for Policy Research. S119-39. DANIEL B. Econ. Cambridge: Blackwell. 63(250). 9302. 1995. AND WELCH. Vol. RICHARD J. its Evolution in the 1980s and the role of Demographic Changes. XXXV (June 1997) NICKELL.. Cambridge. Family Structure on Child Poverty and Income Inequality. 151-56. Statist. PETER. Pol. AND SHATZ. LAMPMAN. Paris: OECD. revised. "Earnings CHAELJ. Syracuse. MI- LEAMER. 1994c. AND SLAUGHTER. pp.July 8." J. pp. RODRIGUES. Lit. 1991. JAN. . "Income Inequality and Welfare System: A Review. Mar." J.ANTHONY F. "In- come Inequality Effects by Income Source. 1994. July. JAN. MOFFITT. Income distribution. from ROBERT Z. FRANK AND MURNANE. Paris: OECD. Econ. pp. Manufacturing. JAMES E. Graduate School of Public Policy. pp. "Mea- LEVY. Aug. "U. Feb. 1993. EUGENE. LAWRENCE. London: Allen and Unwin. 1994. Penguin."J. CARLOS. Thejobs study: Facts. and Factor Prices. Oxford U. 1977. LAWRENCE AND BERNSTEIN. "The Collapse in Demand for the Unskilled and Unemployment across the OECD. 1980/81-1989/90.. 1992. 1964. . Department of Economics. 49. 1. 1994a. Lit. ORGANISATION FOR ECONOMIC CO-OPERATION AND DEVELOPMENT (OECD)." in Labor market trends and underlying forces of change. 1971. "The Decomposition of Inequality in Portugal. Econ. JON. Mimeo. SHORROCKS. 1993. "International Trade and American Wages in the 1980s: Giant Sucking Sound or Small Hiccup?" Brookings Pap. Ends and means of reducing income poverty. Authors' calculations from the Luxembourg Income Study database. THOMAS AND MROZ. 11(1). Statist. 1994F." Presented at the NBER Labor Markets conference in Aug. Paris: OECD." J. Spring 1995. MICHAEL." Microsimulation Unit Discussion Paper No. Stanford U. Rev. PLOUG. LERMAN. Nov.S. "Doing Poorly: The Real Income of American Children in a Comparative Perspective. GUSTAFSSON. of New South Wales. pp.S.eds. eds." SPRC Discussion Paper No. DAVID. 1971. pp. pp.

. GPO. "The English Language Public Use File of the German Socio- M.S. 1995b. 1993. pp. "The PETER AND KENNEDY. STEVEN. BUREAU OF LABOR STATISTICS. "Distribution of Economic Well-Being in Japan: Towards a More Unequal Society. 1994. Sept. Harvester Wheatsheaf. TADASHII. Washington. 1982. As providedby KJELL JANSSONS. "Income. . LEE. 1980. Poverty and Valuation Do We Stand?"in Contemporary economic development reviewed. eds. 1995a. Income Wealth. 429-33.S.S. 1995. and Family Living Standards Impacts Across Seven Nations: The Effect of Noncash Subsidies for Health. 13-29. STATISTICS CANADA. AND EDWARD PALMER. TIMOTHY M. ALAN.S. PETER H. Elderly in an International Context. economic Panel. Unpublished by John Coder. GPO. 5(1). MICHAEL AND RAINWATER.. United Kingdom: U. eds. SMEEDING. ROBERT AND HESTON. J. SMEEDING. pp. Sept. Dec. Social Policy Research Unit. AND GOTTSCHALK. SMEEDING. 188. 1995. Technology come Inequality in Rich Countries During the 1980s."Going to Extremes: The U. Penn World Table (Mark 5): An Expanded Set of International Comparisons. 327-68. Income Wealth. Technical cash Benefits: Report No. DSS 1211. 1996. TIMOTHY ports.. SUMMERS. pp. During Finnish Trade Council. O'HIGGINS. DEPARTMENT OF COMMERCE. BARBARA. Alternative methods for valuing TER. TIMOTHY M. STATISTICA SWEDEN. Feb." in PETER and inequality. DC: U. Rev.Gottschalk and Smeeding: Cross-National Income Inequality sition by Factor Components." Mimeo. "Poverty. No. SMEEDING.. Spring 1995. HANNU. 1950-1989. RICHARD V.S. 89. 1994. Poverty. ??. GPO. "The International Evidence on Income Distribution in Modern Economies: Where selected in-kind transfers and measuring their impact on poverty.50(1). Econ. in Income DistriUUSITALO. 1982. BJORN A. AND LINDERT. SMEEDING. Sept. American inequality: A macroeconomic history. DC: Institute. Education and Housing.. Income Distribution. 681 Jan. pp. WashUrban ington. tabulations . U. "Money Income in the United States: Brighton: LEE AND 1995. BURKHAUSER. New York: Academic Press. AND CODER. AND BEHRINGER. GUSTAFSSON. Ottawa: Statistics Canada. Analysis. Eds. Aug. North-south trade. "Changes bution a Deep Recession. pp. J. 50. 13-210. 1993." Nov. inequal- ity. of NonWashington. BARBARA AND MOFFITT. 193-212. TORREY. Washington. employment. U. "Incomes and Living Standards of Older PeoFinal Report. DC: U. GPO. WOOD. ET AL. ROBERT. TACHABANAKI." Current Population Reports.S. RAINWATER.: MICHAEL BRUNO AND YUSET MUNDLAB. 39(3). "Income After Tax Distribution By Size." Quart. JOHN. TIMOTHY M. 1993." J. Washington. 229-56. 1990. DC: U. London: Macmillan. 1994. "A New Index to Value In-Kind Benefits. FRIEDERIKE. 105(2). Human Res. 193. "In- GOTTSCHALK. TIMOTHY M. Spring 1993. 1991. WHITEFORD. BUREAU OF THE CENSUS. Table VIII. Series P-60.. pp." Current Population Re- 1996. of York. GERT G. PE- and Its Effect in the Steel Industry. Series P-60. 1995. DC: U." ple: A Comparative Volumes I and II." Catalogue No. WILLIAMSON." 28(2). Aug. pp. No.. ADRIAN. ." Rev." Econometrica." Luxembourg Income Study Working Paper No. and income distribution in comparative perspective: The LuxembourgIncome Study. 387-408. Oxford: Clarendon Press. 37(4). TOSHIAKI AND YAGI. WAGNER. Inequality. JEFFREY G. WOLFE.S. 1994. 1991.

041 .006 .100 .000 -.026 .485 .009 .034 .080 -1.071 .300 .012 .638 .002 004 .002 -.035 .619 .402 .029 .000 .357 .002 .030 .068 .003 .010 .010 .4A United States (1978-87) Overall TableIII.9 Table7.8 Overall Table 7.390 .001 .042 .009 .034 .682 Journal of Economic Literature.652 .010 .288 . Vol.610 .336 -.005 .000 .006 -.008 -.470 .010 .462 .009 .021 .036 -.508 2.210 .447 1.001 .3A Within(unived) TableIII.460 .006 .010 .100 .512 .000 1.350 .3A Within(unived) TableIII.025 .530 .130 .009 .010 -.4A Overall Between Ed TableIII.308 .023 .9 Ericksson and Ichino (1995) Italy (1978-87) TableIII.350 .500 .475 .002 .008 .070 1.009 .010 .000 .350 .000 1.004 -.000 -.009 .000 1.480 .012 .355 .729 -.020 .010 .560 .012 .554 .450 .000 1.8 Overall Between Ed Table7.055 .000 .122 3.000 .000 .040 .260 .493 .004 .001 -.3A Between Ex TableIII.001 .000 1.778 1.556 -.430 .270 .004 .4A Gottschalk and Joyce (1995) Australia (1981-85) Table 1 Overall Table2 Between Ed Table3 Between Ex Within Table4 Finland (1987-91) Overall Between Ed Between Ex Within France (1979-84) Overall Between Occ Between Ex Within Israel (1979-86) Overall Between Occ Between Ex Within Table 1 Table2 Table2 Table4 Table 1 Table2 Table2 Table4 Table 1 Table2 Table2 Table4 In var Collegecoef Age coefs evalat 24 Sd of resid In var College coef Age coefs eval at 24 Sd of resid .9 Within(unived) Table 7. Terminal Change Change Year Per Year Per Year (4) (6) (5) (5)(6) (7) Source (1) Blackburn and Bloom (1994) Canada (1979-87) Table 7.010 .390 .130 .390 .020 .933 .028 .000 Full-Time Workers Coef Variation Collegecoef Age coefs eval at 24 Sd of resid Coef Variation Collegecoef Age coefs eval at 24 Sd of resid Coef Variation coef Manager Age coefs eval at 24 Sd of resid Coef Variation Collegecoef Age coefs evalat 24 Sd of resid .434 .224 .540 .043 .002 .001 .3A Between Ex TableIII.032 1.005 .570 .475 .003 -.003 .S.014 .531 .320 .833 1.732 -1.008 .001 .027 .220 .013 .001 .439 .016 .9 Between Ex Within(unived) Table 7.396 .9 Between Ed Between Ex Table 7.000 .501 . XXXV (June 1997) APPENDIXTABLEA ABSOLUTE AND RELATIVE CHANGES IN INEQUALITY AbsoluteChanges Inequality Measure (2) Valuein Initial Year (3) Valuein U.647 1.000 St dev In (w) College coef Exp coefs evalat 2 Sd of resid St dev In(w) College coef Exp coefs evalat 2 Sd of resid .000 1.024 .500 -10.575 1.465 .547 -6.474 .334 .286 .620 .9 United States (1979-87) Table7.037 .484 .4A Between Ed TableIII.529 -.

014 .3A Between Ex Table III.3A Within(unived) Table III.158 3.) ABSOLUTE AND RELATIVE CHANGES IN INEQUALITY 683 RelativeChanges Percentage Change Per Year (8) U.9 Between Ex Table7.860 -.192 .404 .4A Gottschalk and Joyce (1995) Australia (1981-85) Overall Table 1 Table2 Between Ed Between Ex Table3 Within Table4 Finland (1987-91) Overall Between Ed Between Ex Within France (1979-84) Overall BetweenOcc Between Ex Within Israel (1979-86) Overall Between Occ Between Ex Within Table 1 Table2 Table2 Table4 Table 1 Table2 Table2 Table4 Table 1 Table2 Table2 Table4 .017 .021 Full-Time Workers .002 .007 -.000 1.014 .028 .686 2.014 .054 .022 .004 -.S.4A Between Ed TableIII.013 .004 -.022 .034 .021 .008 -.003 .021 .028 .017 .003 .004 -.000 1.003 .017 .079 .000 1.076 .023 .078 .4A TableIII.575 -.021 .9 Table7.018 -.9 Within(unived) United States (1979-87) Table7.3A Between Ed Between Ex TableIII.040 -.004 .020 .352 -.446 -1.000 -.041 .021 -.9 Ericksson and Ichino (1995) Italy (1978-87) Overall TableIII. Percentage Change Per Year (9) (8)(9) (10) Source (1) Blackburn and Bloom (1994) Canada (1979-87) Overall Table7.020 .077 1.457 .695 .014 .4A United States (1978-87) Overall Table III.574 3.021 -.000 1.034 .000 .690 1.8 Between Ed Table7.021 .9 Within(unived) Table 7.3A Within(unived) TableIII.034 .628 .017 .045 .000 .914 -.000 .096 .002 .9 Between Ex Table7.023 .023 .017 -.526 .000 1.030 .078 .106 .8 Overall Between Ed Table7.014 .003 .012 .239 .000 1.002 .132 .987 -1.015 .094 .274 -5.812 -13.020 .677 .732 1.670 1.078 .Gottschalk and Smeeding: Cross-National Income Inequality APPENDIXTABLEA (cont.003 .019 -.

009 .020 .021 .030 1.000 1. Change Per Year (6) (5)(6) (7) Source (1) United States (ed) (1979-86) Table 1 Overall Coef Variation Between Ed Table2 College coef Between Ex Table2 Age coefs evalat 24 Sd of resid Within Table4 United States (occ) (1979-86) Between Occ Table2 coef Manager Between Ex Table2 Age coefs evalat 24 Within Sd of resid Table4 Katz.684 Journal of Economic Literature.001 .003 .540 .640 .020 .037 .023 .023 .545 . Vol.220 .220 .280 .780 .000 1.380 .000 1.003 -.200 .125 -.051 .020 1.454 .000 1.640 .S. XXXV (June 1997) APPENDIXTABLEA (cont.003 . Loveman.550 .001 . and Blanchflower (1995 France (1979-87) Ln (90/10) Overall Figure 1 Between Occ Figure6 Ln(Manual/non) Between Ex Figure6 Ln(41-50/21-25) Within Figure6 Ln(90/10) .220 .030 .021 .010 .015 .527 .024 mixed .010 mixed .010 .790 .000 1.047 .026 .001 .000 1.) AbsoluteChanges Inequality Measure (2) Valuein Initial Year (3) Valuein Terminal Year (4) Change Per Year (5) U.000 1.009 .

Percentage Change Per Year (9) .078 .000 1.034 .000 1.012 .057 mixed . Loveman.030 .000 Source (1) United States (ed) (1979-86) Overall Table 1 Between Ed Table2 Between Ex Table2 Table4 Within United States (occ) (1979-86) Between Occ Table2 Between Ex Table2 Within Table4 Katz.014 .002 -.S.078 .000 1.000 1.020 mixed .020 U.000 1.) ABSOLUTE AND RELATIVE CHANGES IN INEQUALITY 685 RelativeChanges Percentage Change Per Year (8) .028 (8)(9) (10) 1.078 .001 .034 . and Blanchflower France (1979-87) Overall Figure 1 Between Occ Figure6 Between Ex Figure6 Within Figure6 .028 (1995 .021 .034 .034 .Gottschalk and Smeeding: Cross-National Income Inequality APPENDIXTABLEA (cont.144 -.021 .000 1.030 .078 .

0 for the first adult.disposable incomeper equivalentadult.1995b). Unadjusteddisposablefamily Japan(JA) income.5. TableVIII). Vol.GoodmanandWebb (1994. and 0.Bureauof the Census (1995a.incomeper equivalentadult. Finland(FI) Uusitalo(1995.disposable incomewith personweights.specialtabulations using disposablepersonal income as defined in the text. and Tachabanaki Yagi (1995). Beach and Slotsve (1994).and 0. Table 4). weighted by persons and adjustedusingan equivalencescalewith E =.Table2).datasetexcludessinglepersonfamilieslivingalone. household disposableincome with no adjustmentfor householdsize andwith householdweights.0.disposable householdincomewith personweights. New Zealand(NZ) Saunders(1994).5 equivalencescale. United States(US) U. for weightedby households.E =.0 for the first adult. 9.unadjusted familysize. incomeper equivalentadult. Some seriesshownin Table3.personweights.Table7.4). Spain(SP) Luxembourg personweights. Israel(IS) LuxembourgIncome Study Database (Fall 1995). Cantillonet al. Ireland(IR) Callanand Nolan (1993. (1995 Table A.7 for the Belgium(BE) second adult.5 equivalencescale. Departmentof Commerce. StatisticsCanada(1994. equivalence (scale 1.7 for the second adult incomewith personweights.5 per child.686 Journal of Economic Literature.disposable Australia (AU) Canada(CA) . unadjusted household disposable personal income weightedby person.5 per child).p. Bauer and Mason (1992). Table 30). equivalencescale. (1994. A2. Denmark(DK) Aaberge et al.BritishHouseholds BelowAverageIncome scale. equivalencescale 1. Income StudyDatabase(Fall 1995). disposable income per equivalent adult.E =.S. XXXV (June 1997) APPENDIXTABLEB OECD AND OTHER SP JA TREND IN INCOME INEQUALITY IN NATIONS: 1970-1993 IS UK 98 101 102 98 95 92 92 90 91 96 98 100 100 102 103 102 108 111 117 124 125 130 130 US Year 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 AU CA 106 105 105 103 104 107 103 105 101 IR NZ 102 100 100 100 100 109 96 87 108 110 96 113 101 100 101 100 103 107 109 110 110 110 110 112 113 114 113 115 117 100 101 100 101 103 103 102 103 102 101 101 101 102 102 107 111 Notesand Sourcesto AppendixTableB and Table4 Saunders(1994). United Kingdom(UK) Atkinson(1993). familyincome after tax.

and Smeeding The Netherlands(NL) Data providedby Central Bureau of Statistics.and alimonypaid) with no adjustment householdsize andwith householdweights. equivalencescale 1. health care and life insurancepremiums. equivalentdisposableincome with household Italy(IT) weights. for Norway(NO) Epland (1992.householdincomewith no adjustment householdsize andwith householdweights. 0.Table7) (GE) B: Burkhauser Poupore(forthcoming.5 for subsequentmembers.Bjorklund Freeman(1994).Chapter5). (1995a.equivalencescale. disposableincome (includingan allowancefor imputed rent on owner-occupied and homes)with personweights. B: Statistics Sweden (1995) France(FR) . memberand 0.wealth taxpayments.5. adjusted incomewith personweights. Table 2a). 71). Swedish social assistance Sweden (SW) scale. Table 3).7 for second household incomewith personweights. Table 4). householddisposable A: Gustafssonand Palmer (1993.) Year 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 100 104 100 99 98 100 100 100 102 105 106 111 105 BE DK FI FR 109 131 GEA/GEB IT NL NO PO SWA/S WB 105 105 100 100 102/100 97 100 97 99 100 99 98 97 102 100/100 /98 104/96 /98 101/97 102 /97 104/97 95 114 109 110 102 100 94 95 98 99 104 97 100 100 100 98 99 97 104 102 105 104 98 112 109 107 105 103 102 100 102 102 107 107 112 107 107 110 123/114 129/121 /117 /119 Canceilland Villeneuve(1990. Table3) and Brandoliniand Sestito (1993.Gottschalk and Smeeding: Cross-National Income Inequality 687 APPENDIXTABLEB (cont. householddisposableincome (deductingfrom net income interest paid. Table13). equivalencescale with familysize elasticityE + .0 for first adult.see Atkinson. Annex). Concialdi(1996. p.disposable Portugal (PO) Rodriques(1993. for West Germany A: Hauserand Becker(1993.Rainwater.

You're Reading a Free Preview

Descarga
scribd
/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->