Ask the Scholar
Document scope · 1 page
Scholar
Ask about this object, its catalog metadata, its source description, or the page inventory.
For page-specific OCR and visual context, open one of the page chats.
Scholar Source Context
Document identity
localId
54979491
label
[Review of Income and Wealth Series 39, Number 3, September 1993, Poverty, Inequality, and Family Living Standards Impacts Across Seven Nations: The Effect of Noncash Subsidies for Health Education, and Housing] [loose]
core
doc
dtoType
document
citationUrl
pageCount
1
Source metadata
id
54979491
sourceUrl
contentType
document
title
[Review of Income and Wealth Series 39, Number 3, September 1993, Poverty, Inequality, and Family Living Standards Impacts Across Seven Nations: The Effect of Noncash Subsidies for Health Education, and Housing] [loose]
citationUrl
collections
Records of the Council of Economic Advisers (Clinton Administration)
Subject Files
imageCount
1
hasImages
yes
source
import
hasTranscription
no
Source extras
naId
54979491
levelOfDescription
fileUnit
otherTitles
42-t-5290267-20140224F-003-019-2017
recordType
description
ocrSource
nara-archive
Single page context
seq
1
pageIndex
0
type
document
mediaId
a2f56942901ca194
ocrText
FOIA Number: 2014-0224-F
FOIA
MARKER
This is not a textual record. This is used as an
administrative marker by the William J. Clinton
Presidential Library Staff.
Collection/Record Group:
Clinton Presidential Records
Subgroup/Office of Origin:
Council of Economic Advisers
Series/Staff Member:
Robert Schoeni
Subseries:
OA/ID Number:
21167
FolderID:
Folder Title:
[Review of Income and Wealth Series 39, Number 3, September 1993, Poverty, Inequality, and
Family Living Standards Impacts Across Seven Nations: The Effect of Noncash Subsidies for Health,
Education, and Housing] [loose]
Stack:
Row:
Section:
Shelf:
Position:
S
21
3
3
1
Review of Income and Wealth
Series 39. Number 3. September 1993
POVERTY, INEQUALITY, AND FAMILY LIVING STANDARDS
IMPACTS ACROSS SEVEN NATIONS: THE EFFECT OF NONCASH
SUBSIDIES FOR HEALTH, EDUCATION AND HOUSING
BY TIMOTHY M. SMEEDING
Syracuse University and Luxembourg Income Study
PETER SAUNDERS
University of New South Wales
JOHN CODER
U.S. Bureau of the Census
EXECUTIVE
STEPHEN JENKINS
OF THE PRESIDENT
University College Swansea
JOHAN FRITZELL
DEC 22 1993
University of Stockholm
NEW EXECUTIVE
ALDI J. M. HAGENAARS
OFFICE BUILDING
Erasmus University
RICHARD HAUSER
University of Frankfurt
AND
MICHAEL WOLFSON
Statistics Canada
The main aim of this paper has been to summarize the impact of noncash income-health and health
education benefits, and imputed rent-on living standards, income distribution and poverty in seven
nations at the beginning of the 1980s using the Luxembourg Income Study database. Our results do
not give rise to a pattern of national differences in poverty rates or income inequality which are
markedly different from that which emerges from previous LIS research based on cash income alone.
While these results may be sensitive to the techniques used to measure and value noncash benefits in
this paper, it appears that noncash income reinforces the redistributive impact of conventional (cash)
tax-transfer mechanisms rather than acting to offset them in any major way.
Note: This paper summarizes a six-year project which was conducted in conjunction with, and
under the auspices of, the Luxembourg Income Study. Sponsors have included the United States
National Institute on Aging, the United States National Science Foundation, and the LIS member
countries. The authors would also like to thank an anonymous referee, Dirk Wolfson, and various
individuals who helped us as project participants. These include Grant Cameron. Wolfhard
Dobröschke-Kohn, Flip de Kam, Peter Hedstrom, Brigitte Buhmann, Michael O'Higgins, Uwe
Warner. Julie Tapp, and Inge O'Connor. A shorter and less complete draft of this paper was presented
to the 22nd General Conference of the IARIW in Films, Switzerland, August 1992. However, the
authors retain the right to all errors of commission and omission. Additional information on the LIS
research project may be obtained from Smeeding, care of Luxembourg Income Study at CEPS,
B.P. 65. L-7201 Walferdange, Luxembourg.
229
INTRODUCTION
Economic resources, including both cash and noncash income, determine the
economic well-being of households in all nations. Cash income is the most widely
employed measure of household economic well-being, but it excludes considerable
amounts of resources received in a noncash form. These include health care,
housing, education, child care, transportation, food, and other subsidies from
governments or from other third parties (i.e., employers), production for own
consumption by farmers and by other individuals living mainly in rural areas,
and in-kind transfers received from relatives, friends and others in the form of
food, clothing and/or shelter. Moreover, the distribution of these resources may
vary systematically by country, by regime or by population subgroup, thus affect-
ing measures of relative economic well-being among households.¹ The omission
of noncash income from microdata based measures of economic well-being is not
purely unintentional. In most countries aggregate income in-kind is measured by
systems of national income and/or social accounting. However, the problems
inherent in the measurement, valuation, and imputation of noncash income to
individual households on the basis of microdata files are formidable. While a few
countries (e.g., United States, Netherlands) have partially accomplished this task
with some difficulty and while others have achieved at least some limited microdata
accounting of selected income sources (Germany, Australia, Switzerland, United
Kingdom), some countries (Canada, Sweden) have never before systematically
attempted such a task. Moreover, none of these countries have ever attempted a
joint project aimed at producing measures of noncash income which are inter-
nationally comparable among such nations.
The authors of this paper and several colleagues have been working on
such a project in Western nations for the past several years under the auspices
of the Luxembourg Income Study (LIS). This paper presents a summary of
the results of this project as they relate to income inequality, living standards
for several types of families, and poverty measurement. Additional detail is
available by contacting the authors and also in Smeeding, Saunders and Jenkins,
et al. (1993).
The remainder of this paper discusses the importance of noncash income
in Western nations, and summarizes our conceptual and empirical approach
to measuring the size and impact of noncash income in seven of the countries
participating in the LIS project. For data reasons, the scope of the project is
restricted to noncash incomes associated with education (schooling), health
services, and, for five of the seven countries, comparative estimates of noncash
housing benefits accruing to home owners. The paper encompasses noncash
benefits accruing to individuals as a result of direct (subsidized) public
'Noncash income does not include off the books cash or noncash income (grey economy) and
hence, this topic is not discussed in this paper. While we do not include any of the Reforming Socialist
Economies (RSEs) in this paper, it should be noted that many such countries, including Poland and
Hungary, have elaborate systems of national accounts, consumption studies, and income distribution
estimates which include a wide range of goods and services provided in-kind. Yet even these nations
exclude large amounts of these subsidies. For more on this topic, see Smeeding and Torrey (1991)
and Teglarsky and Struyk (1990).
230
provision, tax concessions and provisions subsidized by employers for health
determine the
and education benefits. After explaining the methodology and data sources,
the most widely
the paper discusses and analyzes the results, focusing specifically on comparisons
considerable
of the distribution of noncash income across countries. A classification accord-
health care,
ing to life-cycle category and family-type allows the results to be analyzed
subsidies from
more thoroughly and highlights the role of noncash income in redistribution
duction for own
both across and within the life course of individuals and families and its
in rural areas,
impact on the living standards of such families. Particular attention is given
in the form of
to comparisons of the distributions of final (cash plus noncash) income in
resources may
each country as well as to the impact of noncash income on the incidence
"oup, thus affect-
and structure of relative poverty.
The omission
well-being is not
Noncash Income and the Luxembourg Income Study Project
I is measured by
Γ, the problems
Comparative research on the distribution of economic well-being has made
cash income to
considerable progress in recent years. That progress has been facilitated by
ble. While a few
advances in both methodological procedure and data availability. Methodolog-
plished this task
ically, recent income distribution research has achieved greater clarity on
'nited microdata
questions relating to the appropriate unit of analysis, the basis for ranking
zerland, United
those units and their weighting in deriving aggregate measures of inequality
e systematically
(Atkinson, 1983; Atkinson, Rainwater and Smeeding, 1993). These develop-
ver attempted a
ments have permitted analysis of the distribution of income among households
which are inter-
to translate more readily into the distribution of economic well-being among
individuals. Much, though not all, of the empirical application of this new
en working on
methodology has been undertaken within a comparative context. That, in turn,
er the auspices
has been made possible by advances in data availability, specifically by the
a summary of
production of microdata sets which generally conform to agreed upon and
ving standards
standardized concepts and definitions.
ional detail is
At the forefront of this research effort has been the Luxembourg Income
is and Jenkins,
Study (LIS), an international, cooperative research endeavor which began in 1983
with the aim of improving comparative measures of economic well-being. There
ncash income
are, in fact, three distinct components of the research undertaken as part of the
rical approach
LIS project. The first involves the reorganization of national microdata sets in
the countries
order that they conform to a common, standard, conceptual and definitional
the project is
framework. The second involves the use of the data thus generated to analyze
oling), health
various aspects of economic well-being and inequality within a comparative frame-
es of noncash
work. The third is to make the standardized data sets easily available to the
isses noncash
international research community, in order that researchers within national boun-
dized) public
daries can utilize them, confident in the knowledge that national differences in
data concepts and definitions have, as far as possible, been eliminated. The LIS
database currently covers over 20 countries with data covering various periods
y economy) and
from 1969 to 1990.
forming Socialist
Analysis of the effect cash transfers and benefits on income distribution for
riding Poland and
come distribution
the countries included here can be found in Smeeding, O'Higgins and Rainwater
en these nations
(1990), and in O'Higgins, Schmaus and Stephenson (1989). In fact, all of the
d Torrey (1991)
research undertaken as part of the LIS project to this point has been based on
measures of cash income. The income concepts around which the LIS database
231
has been constructed-factor income, gross income, disposable income and
equivalent income-are all based on a conception of income expressed in terms
of cash only. Noncash elements which form part of income in its broader meaning
have, with few exceptions, been excluded.² This segregation was inevitable in the
early phases of the LIS project, but its continuation has become increasingly
difficult for at least two reasons. First, because economic well-being is, in fact,
determined by more than just receipts of cash income, there is a need to begin to
Count
expand cash income measures to reflect a broader range of noncash components.
Australi
Second, studies based on cash income may give a distortionary picture of the
Canada
impact of government budgetary policies because within this limited framework
Netherl:
government (cash) transfers and (direct) taxes do not balance-even in the remote
Sweden
U.K.
sense which characterizes the actual overall fiscal situation-but also because
U.S.
governments may seek to achieve their redistributive goals through programs
(West) (
which provide noncash benefits rather than just through tax-transfer mechanisms.
Sou
This means that measures of economic well-being based on disposable cash income
Not
are subject to the vagaries of the overall fiscal structure within countries, and that
all levels
"Sul
comparisons of both the level and distribution of well-being between countries
are dependent upon the existing fiscal structures. Particularly as we seek to under-
stand and compare the distribution of income in the Reforming Socialist Econom-
the larg
ies of Eastern Europe and Russia with that of the U.S., we need to broaden our
here.³
comparative measures of the distribution of well-being. This paper should be seen
N
as a first exploratory step in this direction.
have co
of hous
would
THE SIGNIFICANCE OF NONCASH ICOME
ies are
Noncash income may be provided to private households by governments, by
service
private third parties such as employers, or by the household itself as in the case
benefit
of imputed return from durables such as owned housing or automobiles. By far
differe
the largest amounts of noncash benefits are provided by governments. Gov-
Since 1
ernments tax and transfer large amounts of total personal (factor) income-
income
ranging from 20 percent (in the U.S.) to over 40 percent (in Sweden)-proceeding
income
from market-determined factor income to final income. In most countries, cash
F
income transfers constitute less than half of government expenditures. Hence, not
mislea
all of the income taxed away by governments, even counting only direct taxes,
both w
emerges as contributing to the post-tax, post-transfer disposable cash income of
sive th
households. The amounts taxed (or borrowed in the case of deficit financing) but
ing, va
not transferred in cash constitute noncash income components. While not all such
well-b
components may be measured, valued and imputed to households, large parts of
A
public noncash income transfers in the form of health care, education and housing
educa
can be so imputed, at least in principle. Moreover, health care and education are
k
benefits
benefit
to 7.1 I
"Near-cash" income-that is, payments made in flexible currency denominations, such as food
Health
stamps in the U.S., or cash benefits contingent on meeting certain needs, e.g., university scholarships
the U.S.
or housing allowances in Sweden or the U.K., are already included in LIS disposable income on the
about :
grounds that these benefits that are denominated in money terms and are very nearly equivalent to
Thus o
an equal cash transfer in the eyes of the recipient.
mediar
232
income
and
TABLE I
in terms
ESTIMATES OF CASH (PENSIONS AND UNEMPLOYMENT BENEFITS) AND NONCASH
oader meaning
(HEALTH AND EDUCATION) SOCIAL EXPENDITURES AS A PERCENTAGE OF
evitable in the
GDP IN 1960, 1975. AND 1981
increasingly
Noncash-Cash
is, in fact,
Cash
Noncash
Differences"
to begin to
Country
1960
1975
1981
1960
1975
1981
1960
1975
1981
components.
picture of the
Australia
3.5
5.7
6.4
5.2
11.7
10.5
1.7
6.0
4.1
Canada
4.3
6.6
6.9
5.4
12.1
11.8
1.1
5.5
4.9
framework
Netherlands
5.4
11.4
14.0
5.8
13.5
13.8
0.4
2.1
-0.2
in the remote
Sweden
4.6
8.4
12.3
8.0
12.9
15.5
3.4
4.5
3.2
also because
U.K.
4.3
7.0
8.8
7.1
11.8
11.2
3.8
3.8
2.4
U.S.
4.8
8.1
7.9
4.9
10.0
9.7
0.1
1.9
1.8
programs
(West) Germany
9.9
14.1
13.9
5.5
12.0
11.78
-4.4
-2.1
-2.2
mechanisms.
Source: OECD (1985).
cash income
Note: Noncash Social Expenditure includes the cost of direct health and education benefits from
tries, and that
all levels of government. Employer and other third party benefits are not included.
"Subtracts cash from noncash benefits.
countries
to under-
Econom-
the largest government noncash subsidies in each of the seven nations examined
broaden our
here.³
should be seen
Not only is the size of noncash income important, its distribution may also
have considerable effects on the distribution of well-being between different classes
of households. Consider, for example, public health and education benefits. Most
would argue that health benefits provided by governments and insurance compan-
ies are most valued by older citizens who are more likely to make use of medical
vernments,
by
services. Similarly children (and/or families with children) are most likely to enjoy
in
the
case
benefits from education subsidies in a given year. One would thus expect that
obiles. By far
differential gains and losses would be realized across different household types.
Gov-
Since the value of noncash benefits is likely to be disproportionate to net (cash)
income-
income, these income components might also have large distributional effects by
-proceeding
income class, as well as by demographic group.
ountries,
cash
For all of these reasons, the distribution of disposable cash income may yield
Hence,
not
misleading inferences about the relative well-being of various types of households
direct
taxes,
both within and across countries. If we accept the axiom that the more comprehen-
income of
sive the definition of income used the better is the measure of welfare, then measur-
nancing) but
ing, valuing and imputing noncash income will give a more complete picture of
not
all
such
well-being than that afforded by cash income alone.
parts
of
An indication of the aggregate importance of public noncash health and
and
housing
education benefits in the seven countries in our study is provided in Table 1.
ducation
are
Within each of the countries examined here. health and education subsidies are the only noncash
benefits separately accounted for by the OECD (1985). They are the largest two types of noncash
benefit in each of the nations studied here. Education varies from 5.2 percent of GDP in Germany
such
to 7.1 percent in the Netherlands, with the U.K. the median country at 5.8 percent of GDP in 1981.
as
food
Health care subsidies vary from 4.2 percent in the U.S. to 9.9 percent in Sweden. However, correcting
scholarships
income
the U.S. aggregate to reflect tax subsidized employer benefits for health care brings the U.S. total to
on
the
about 8.0 percent of GDP. The next lowest nation to the U.S. was the U.K. at 5.4 percent of GDP.
equivalent
to
Thus correcting for U.S. employer subsidies reduces the range of estimates across countries. The
median nation in terms of health care subsidies was Canada at 5.6 percent of GDP in 1981.
233
Noncash expenditures are shown relative to the major elements of cash transfer
to em
spending (pensions and unemployment benefits), both being expressed as a per-
cance
centage of Gross Domestic Product (GDP). The estimates overstate the ratio of
incon
noncash to cash benefits because the OECD noncash expenditures may include
some cash items (e.g., education allowances paid in cash to tertiary students),
while coverage of cash benefits is restricted to pensions and unemployment bene-
Conc
fits. Together, these shortcomings are not likely to be of sufficient importance to
fundamentally change the picture indicated in Table 1.4 In 1981, in all countries
1
except West Germany and the Netherlands, OECD's noncash expenditure
such
exceeded expenditure on cash transfers.⁵ The difference was almost 5 percent of
only
GDP in Canada, and exceeded 4 percent and 3 percent of GDP in Australia and
from
Sweden, respectively. Between 1960 and 1975, noncash benefits grew faster than
parti
cash benefits. However between 1975 and 1981 the pattern was reversed, as cash
in th
transfer spending rose sharply relative to GDP while noncash spending fell relative
Thes
to GDP everywhere except in the Netherlands and Sweden. In the Netherlands,
provi
noncash expenditure fell slightly below cash transfer spending after the mid-
video
seventies due to the rapid rise in cash transfer spending, particularly disability
gove
benefits. West Germany is the only country where cash transfer spending has
treat
consistently exceeded noncash expenditure, owing largely to the generous West
gove
German public pension system. While the aging of the population may further
and
increase public pension outlays in these nations over the next several decades,
wou
noncash benefits still generally dominated cash benefits during the 1979-81 period
and
U.K
during which we observe and measure household income in this paper.
These data confirm that the total size of public noncash benefits is such
our (
as to present the possibility that their inclusion as part of income might well
our
influence the overall level of economic well-being and its distribution. However,
expl
the ranking of countries according to the levels of cash and noncash spending
is similar, except for Canada whose noncash ranking is well above its cash
and
transfer ranking. This suggests that governments have not used cash transfer
quar
and noncash benefit programs as substitutable methods of achieving their social
inco
objectives. It thus implies that while the inclusion of noncash income will
non
increase measured economic well-being, it may also cause the observed degree
of r
of inequality of final income to be more equal than that of disposable income
priv
(both across and within countries) at least if the equalizing redistributive
tativ
impact of cash and noncash incomes are similar. The figures in Table 1 indicate
igno
that the cross-country variation in aggregate noncash expenditure is less than
one
the variation in spending on cash transfers. The variation in noncash spending
imp
in turn largely reflects cross-country variations in health expenditure, spending
acro
on education being a broadly similar proportion of GDP in all countries
ferr
(O'Higgins, 1988 and footnote 3). These points aside, however, the main message
one
app
a m
"For instance, family benefits in the form of child allowances, maternity leave, and other types
of benefits lumped together by OECD are excluded. They totalled less than 7 percent of social expendit-
we
ures in each of the countries studied here in 1981. As mentioned earlier, OECD does not distinguish
or
noncash subsidies other than health care and education.
the
⁵We will refer to West Germany throughout this paper because the measurements and data were
collected subsequent to 1950 and prior to 1990 when East and West Germany were separate states.
tra
234
transfer
to emerge from Table 1 is that noncash income is of sufficient quantitative signifi-
as
a
per-
cance that it need be taken account of in any comprehensive measurement of
he
ratio
of
income and assessment of economic well-being.
include
students).
CONCEPTUAL, METHODOLOGICAL AND EMPIRICAL ISSUES
bene-
portance
Conceptual Approach
to
countries
In practice, the range and type of noncash income to include in a project
xpenditure
such as this is enormous. It has already been noted that government is not the
percent
of
only source of noncash income to private households. The goods and services
stralia
and
from which noncash income is derived may also be provided by private third
aster
than
parties such as employers or charitable organizations, or by the household itself
as
cash
in the form of home-grown food or implicit rent on owner-occupied housing.
relative
These items may be delivered and subsidized directly or, in the case of government
therlands.
provisions, indirectly via tax expenditures or regulatory policies. Employer pro-
the
mid-
vided benefits such as health-care insurance in the United States may also attract
disability
government support if they, or employee contributions, receive concessionary tax
has
treatment. Ideally, one would like to include the net incidence of all types of
West
government and non-government benefits (and to subtract all taxes and charges)
further
and to assign the benefits in cash equivalent terms to each household. Such a task
decades.
would be daunting to say the least (e.g., see O'Higgins and Ruggles, 1981; Ruggles
period
and O'Higgins, 1981, for consistent attempts at such an incidence study in the
U.K. and U.S.). Practically, such an exercise was beyond the range and scope of
is
such
our collective project. The jointly determined goals and criteria which have guided
well
our project choices for selecting noncash benefits should therefore be made
However,
explicit.
spending
Our primary goal was to improve upon measures of economic well-being
its
cash
and the size distribution of well-being within and between countries by adding
transfer
quantitatively important and practically measurable components of noncash
social
income to the LIS cash income database. Moreover, in selecting components of
will
noncash income for imputation, we sought to measure the flow from those sources
degree
of noncash income which have a deliberate (large) and differential impact on
income
private incomes within or between countries. Conceptually acceptable but quanti-
stributive
tatively insignificant noncash income components were for this reason deliberately
indicate
ignored. The principle of international comparability was our sine qua non. Since
less
than
one of our main objectives was the improvement of the LIS database, it was
spending
important to produce measures of noncash income components which were robust
spending
across countries. Following this principle we sometimes chose to abandon pre-
countries
ferred measurement techniques available for practical implementation in only
message
one or two countries and adopted instead less accurate but wholly comparable
approaches to noncash income measurement across all countries, or at least across
a majority of countries, involved.
other
For instance, we were forced to exclude those goods and services for which
types
expendit-
we either did not have the requisite data needed to impute a value to them (higher
distinguish
or tertiary education subsidies) and/or were not of great overall significance at
data
the time of the income surveys with which we were working (child care and
were
states.
transportation services). We also excluded, reluctantly, noncash income in the
235
form of chronic (long-term) health care subsidies-provided in the form of both
on the following four gene
domiciliary and institutional care-for the frail elderly and for younger people
(i) In order to imp
with severe disabilities. This was partly due to lack of reliable comparative data
benefits and costs, with onli
on the cost of these services, but also because the institutionalized population is
Thus, the benefits associa
excluded from most household survey datasets. Finally, we did not control for
income, just as any costs (n
differences in spending for education within nations. While these differences may
from total (gross) benefits
be relatively small in some countries, they can also be very large. For instance,
and thus receive no nonca
the variance in education spending in the U.S. can be 100 percent or more across
(ii) The total (gros
states. To the extent that these expenditure levels also vary with income, e.g.,
amount of money a goven
lower education expenditures for poor minority children in the southern United
has been made to estima
States, we may overstate the value of educational benefits to some groups simply
benefits. This implies that
because we fail to control for variance within larger nations.
stated in some cases, part
Three broad classes of benefits were included in our study: imputed rental
well have chosen to spend
ncome from home ownership (in all but the U.K. and Australia where the LIS
had these been provided a
version of the datasets excluded the data needed to make such imputations),
(iii) The household
health care and education. Tertiary education spending and its associated noncash
to be the only household
income was excluded because the LIS tapes did not permit those studying in
specific (private) externali
tertiary institutions and their subsidies to be identified. Cash scholarship support
estimating them.
of living expenses for tertiary education was identifiable for those who received
(iv) We include be
such support. However, these are a very small minority of such students in most
public noncash benefits f.
of the countries studied.
have been estimated whe
In the area of housing benefits, data limitations and comparability forced us
where they were not, five
to focus on imputed rent to owner occupiers. Housing benefits paid in the form
used.
of cash allowances were already included in the LIS database. Other types of
We now turn to spec
benefits were too elusive to include in most countries. These include purchase
the field of education subsi.
subsidies for low-income home buyers (Netherlands), and the net value of subsid-
(primary) and secondary $
ized rental housing (U.S., U.K.).
outlays have been allocate
Thus, this study estimates noncash income provided by government and
methods involve calculat
employers in most of the health and education areas, and in the area of imputed
student from data on tot:
rental value for owner-occupiers. Noncash income provided through tax expendit-
averages as noncash inco
ures are also included. although these are implicitly incorporated into the LIS
of education. Adjustment
cash income framework because they affect taxable income and are thus allowed
public subsidies for priv:
for when deriving disposable income from gross income. In a limited sense,
example) have been alloc:
therefore, the project can lay claims to incorporate all three elements of Titmuss'
incomes are allocated to a
social divisions of welfare spending (public, occupational and fiscal) at least within
private schools, we assum
the education and health areas. The inclusion of housing benefits reflects some
thus also provide benefits
mixture of public subsidies and of home production.
that government subsidie
in government schools. Ir
Imputation Rules
are assumed to place a ze
Having described the scope of noncash income, the next set of issues relates
ance. Finally, we have dec
to the identification and valuation of noncash benefits necessary for the imputation
of noncash income. Again. given space constraints it is only possible here to
describe our methods in general terms.⁶ Our imputation procedures were based
'The indirect effects of go
excluded. The implicit counterf:
in the absence of any governm:
"Clearly other analysts could make alternative judgements about rules for inclusion or exclusion
"While recipient or cash eq
of benefits. or for imputation and valuation of benefits within or across countries. We are able to
overstate the value of noncash
only outline a subset of the detailed choices we made in this paper. For additional detail on our
be false. For instance, in the c
choices, see Smeeding. Saunders. Jenkins et al. (1993). Chapter 2.
subsidies above their market Va
236
of both
on the following four general principles:
people
(i) In order to impute noncash income, account must be taken of both
data
benefits and costs, with only the resulting net subsidy being imputed to households.
pulation is
Thus, the benefits associated with a partial subsidy are included as noncash
control
for
income, just as any costs (whether third-party charges or taxes) must be subtracted
ences
may
from total (gross) benefits. If there is no subsidy, households pay market prices
instance,
and thus receive no noncash income.⁷
across
(ii) The total (gross) value of noncash benefits is assumed equal to the
e.g.,
amount of money a government (or employer) spends on each item. No attempt
United
has been made to estimate the recipient or cash equivalent value of noncash
simply
benefits. This implies that the recipient's value of noncash income may be over-
stated in some cases, particularly for those families on low incomes who might
rental
well have chosen to spend the monetary value of noncash subsidies in other areas
the
LIS
had these been provided as cash transfers.
outations),
(iii) The household which directly receives each noncash benefit is assumed
noncash
to be the only household to benefit. We thus disregard all general (social) or
tudying
in
specific (private) externalities, largely because of the practical impossibilities of
support
estimating them.
received
(iv) We include both operating and capital outlays when allocating
in
most
public noncash benefits for education and health care. Annual capital outlays
have been estimated where data on interest and depreciation were available;
forced
us
where they were not, five year averages of actual capital expenditures have been
the
form
used.
types
of
We now turn to specific imputation procedures which we have followed. In
purchase
the field of education subsidies, our analysis has been restricted to public elementary
of
subsid-
(primary) and secondary schooling. The benefits of current (operating) and capital
outlays have been allocated to families with children in education. Our estimation
and
methods involve calculating, for each level of education, average outlays per
imputed
student from data on total outlays and student enrollments, and imputing these
expendit-
averages as noncash income to families with children participating in each level
the
LIS
of education. Adjustments for early-leavers ("drop-outs") have been made and
allowed
public subsidies for private education (which are important, in Australia, for
sense,
example) have been allocated on a randomized basis. Since the resulting noncash
Titmuss'
incomes are allocated to all students, whether they attend public (government) or
within
private schools, we assume that subsidies to government schools are of value and
some
thus also provide benefits to those with children in private schools, and likewise
that government subsidies to private schools are of value to those with children
in government schools. In contrast, families whose children "drop-out" of school
are assumed to place a zero value on their foregone opportunity of school attend-
relates
ance. Finally, we have deducted property tax payments from homeowners in order
nputation
here
to
based
'The indirect effects of government subsidies or taxes on market prices, e.g., housing, were also
excluded. The implicit counterfactual is therefore that the market price is the price which would prevail
in the absence of any government intervention via taxes or subsidy.
exclusion
'While recipient or cash equivalent valuation is the ideal, and while market value will on average
are
able
to
overstate the value of noncash benefits to low income households. one can find cases where this may
on
our
be false. For instance, in the case of poor health, a low income family may value health insurance
subsidies above their market value.
237
to arrive at a net subsidy figure. This is because property taxes are the major
worth of h
financing mechanism for local schools in most countries in the study.
with the e.
In the field of health care subsidies, our imputations have been based on a
which con
risk-related insurance premia approach. That is, we view health care as an insur-
education
ance benefit received by all coverees, independently of their actual use of health
other nati
care benefits, and also that the benefits (and hence premia) differ by age and
benefits W.
gender in line with differences in need. According to this line of argument, insur-
Other Mea
ance premia should be actuarially adjusted (age and sex related) to account for
differences in the need-related value of being covered by health insurance. Thus,
There
benefits received are estimated by age and sex-specific outlays spread over all
turning to
coverees in each age-sex cell of the population. The actual cells used to estimate
a basic unc
benefits and the method for allocating nontax (user) charges for health insurance
choice of t}
are derived from national data sources on utilization rates for different elements
units, i.e., 1
in the health care system, differentiated by age and gender, and national data on
of noncash
the incidence of any tax or user charges. In cases where freely available public
more relate
health insurance is all that exists (e.g., in Sweden), gross benefits only are imputed,
or single pe
the taxes to support them already being deducted. In cases where public and
This definit
private third-party charges are levied on households and employers (the Nether-
quarters ar
lands, U.S. and West Germany) an allocation of costs is also specified. In the
to ignore ec
case of direct payments to providers (e.g., out of pocket charges, deductibles,
unmarried
etc.), no imputation of costs or benefits is undertaken. Finally, in cases where
The m
total third-party premia equal expected benefits (i.e., no subsidy is realized), no
Sweden wh
imputation is made. Thus, only subsidized and insured benefits and payments to
marriage-lik
insurers are taken into account here.¹⁰
counted as
For housing, the correct measure of implicit rent is the opportunity cost of
(e.g., elderl:
the housing used, i.e., the counterfactual private market rent minus cost of owning
even where
(including depreciation, property taxes, maintenance, etc.). However, these data
holds or fai
are not available for all countries. Alternatively in competitive markets, the
usual defini
implicit rental value of owned homes can be measured as a fixed interest return
Having
on the net worth of the home. Economic theory holds that ignoring transactions
a number of
costs and differential risk, investment funds (financial capital) will flow between
types. The
sectors to equilibrate the marginal rate of return on all types of investments.
families in 1
Hence, the implicit rate of return on housing equity will equal a safe private
Table A-1 C
market rate of return (or the return on relatively riskless long-term government
relation to tl
bonds) on an equal value of investment. The annual rate of return which is used
categorizatic
in this case is approximated by a 2 percent real return (2 percent on top of the
and policy F
change in overall consumer prices for a country in the year studied). Inflation
1. Far
plus 2 percent was, thus, multiplied by home equity to estimate imputed rent.
(a)
In summary, our imputation methods have involved combining the existing
LIS data set with additional data on noncash expenditure aggregates, on the
(b)
utilization rates of education and health services, and on estimates of the net
(c)
Where property txes are not used in this way (e.g., in Australia), the deduction of property taxes
from noncash education benefits was not undertaken. Further, we have ignored the potential incidence
of property taxes on renters.
10In the U.S., the 12 percent of the population without health insurance were not assigned benefits.
"Thus, wi
While this properly reflects the value of their insurance coverage, it likely also understates the value
attempted wou
of health care benefits which they actually receive due to provision of free or charity care.
group of nation
238
the major
worth of homeowners. In order to achieve this, we have had to combine familiarity
9
with the existing LIS data with detailed knowledge of the national data sources
based on a
which comprise LIS, an understanding of the structure and operation of the
IS an insur-
education and health insurance schemes in each country and expertise in bringing
e of health
other national data sources to bear on how the detailed allocation of noncash
y age and
benefits was to be imputed.
ent, insur-
ccount for
Other Measurement Issues
nce. Thus,
There are a number of additional issues that have to be discussed before
d over all
turning to our results. These are discussed only briefly, so that readers can gain
10 estimate
a basic understanding of our actual procedures. The first such issue relates to our
insurance
choice of the basic unit of analysis. Noncash benefits have to be imputed to income
it elements
units, i.e., to persons, families, or households. Our emphasis is on the distribution
al data on
of noncash benefits between families defined to include either a group of two or
ble public
more related persons living together as a family and sharing their housekeeping,
e imputed,
or single persons who are assumed to independently keep their own housing units.
public and
This definition implies that two unmarried individuals sharing the same living
he Nether-
quarters are treated as independent families. The bias implicit in this treatment is
ied. In the
to ignore economies of scale in housing (and other domestic arrangements) among
eductibles,
unmarried people living together.
ises where
The major exceptions to these general rules are in the Netherlands and
alized), no
Sweden where, again as noted earlier, unmarried persons living together in a
lyments to
marriage-like relationship (i.e., sharing living quarters, facilities and expenses) are
counted as a single family, and in Canada, where related generations of families
ity cost of
(e.g., elderly mother and adult children) are treated as separate economic units,
of owning
even where they live together. In general, while these procedures treat some house-
these data
holds or families differently than others, they come closest to the preferred and
rkets, the
usual definition of families within each country.
rest return
Having defined families for the purposes of analysis, the next step is to specify
ansactions
a number of different family types for measuring the impact of subsidies on family
W between
types. The results presented in the following section of the paper disaggregate
vestments.
families in two different dimensions, according to eight family types (Appendix
ife private
Table A-1 contains the frequency distribution of families by type of family.) In
overnment
relation to this disaggregation, we adopted the following exclusive and exhaustive
ich is used
categorization which was chosen in part because of its relevance for both analytical
top of the
and policy purposes:
Inflation
1. Families with Children (children are 17 or younger),
ed rent.
(a) Non-aged couples (head under 65, couple may or may not be
he existing
married)
es, on the
(b) Single parents (one adult only plus children),
of the net
(c) Other families with children (including a few units with head 65 and
over);
roperty taxes
tial incidence
ened benefits.
"Thus, while the focus of our research effort has been explicitly comparative, what we have
the value
attempted would almost certainly not have been possible except as a joint venture undertaken by a
group of national researchers committed to such a task.
239
2. Elderly Families (head 65 or older),
and express that
(a) Single elderly persons (one person unit 65 or older),
income is thus <
(b) Elderly couple (head 65 or older);
income and per
3. Non-aged Families Without Children,
adjusted or final
(a) Single persons,
to poverty. Whe
(b) Childless couples (of any marital status),
by family type -
(c) Other childless families (more than two adults or families with
to incomes are
young adults age 18 or over).
As already noted, the basic income concepts we have used are those developed
as part of the LIS project and other research in the area (Smeeding, O'Higgins,
LEVI
and Rainwater, 1990). However, to the familiar (cash income) concepts of factor
income, gross income and disposable income, we now add the concept of dispos-
The Level of No
able income and imputed noncash income in the form of education and health
The overall
care, and another which includes housing as well. These will be called "full income
1" (health and education only) and "full income 2" (health, education and
income are pres
for each countr
housing).
Since family size and structure have a considerable influence on the well-
lower panel are
being of individual family members, account must be taken of differences in family
It is important t
need in order to derive measures of individual well-being for poverty measurement.
costs and charge
This is done by applying a set of equivalence scales-which express relative family
ers. This explair
needs-in order to derive measures of equivalent income, or family income
(gross) governnt
adjusted for family needs. Equivalent income is a preferable measure of individual
ranking of coun
well-being to per capita family income because the latter makes no allowance for
the two compar
economies of scale in family financial arrangements. We regard the extent of such
exceptions are A
economies as an empirical issue which is incorporated into the scales themselves,
1, and the U.K
and not something which is a matter for pre-judgement.
according to Ta
education incon
There remains the question of which set of equivalence scales to use, an issue
on which there currently exists little consensus, but which is known to influence
percent in the 1
cross-country comparisons of inequality and poverty, at least under certain
Sweden, the imp
nent of noncash
circumstances (Buhmann, Rainwater, Schmaus and Smeeding, 1988). We have
selected as our base case a simple set of equivalences that lie about midway
except Canada.
of noncash edu
between the two extreme scales produced by recent research on the topic. These
scales allocate a weight of 1.0 for the first adult in each family, 0.4 for each
subsidy to heal
additional adult in the family and 0.3 for each child. They approximate what
fairly small.
The additi
Buhmann et al. (1988) refer to as the Budget Studies/Program equivalences. These
nounced. While
are based in turn on equivalence scales estimated from budget study data on
expenditure patterns for different family types, as well as on family size differentials
health, the disti
in benefit levels built into social programs. The scales imply, for example, that a
the distribution
single parent with one child and a married couple with two children have needs
largest amount
the final rankir
which are 30 percent and 100 percent greater than the needs of a single adult,
leaves Canada
respectively.
Although these equivalence scales have been used to derive equivalent
Germany and 1
disposable cash income, the question arises of whether the same scales should
be applied to adjust noncash income. Since noncash income does not depend
Living Standar
upon family size or structure (only on characteristics pertaining to indi-
viduals)-which suggests that there are no economies of scale in noncash
The effect
income-we decided to aggregate all noncash income for the family as a whole
types-the effe
240
and express that in per capita terms. Our welfare-based measure of final family
income is thus equal to the sum of equivalent (or adjusted) disposable cash
income and per capita noncash income. This income concept is referred to as
adjusted or final income (1 or 2) in subsequent tables and discussion referring
to poverty. When investigating the effect of noncash income on living standards
by family type or on the overall income distribution equivalence adjustments
milies
with
to incomes are not made.
developed
O'Higgins,
LEVELS OF NONCASH INCOME AND LIVING STANDARDS
of factor
of dispos-
The Level of Noncash Income
and
health
full
income
The overall mean amounts of (unadjusted disposable) cash and noncash
and
income are presented for each country in Table 2. The figures in the top panel
for each country are expressed in national currencies, while the figures in the
the
well-
lower panel are standardized relative to each country's mean disposable income.
in family
It is important to recall that the noncash incomes shown in Table 2 are net of all
asurement.
costs and charges, including only the net subsidy from government and/or employ-
family
ers. This explains the differences between the patterns shown in Table 2 and the
income
(gross) government expenditures shown in Table 1. Despite these differences, the
individual
ranking of countries according to the relative importance of noncash income on
owance
for
the two comparable elements (health and education) is broadly similar. The two
of
such
exceptions are Australia, whose ranking is higher according to Table 2 than Table
themselves,
1, and the U.K. whose rank changes from fifth according to Table I to second
according to Table 2. Given these changes, and the fact that noncash health and
an
issue
education income averages 16.6 percent of disposable income, ranging from 13
influence
percent in the U.S. and West Germany to almost 22 percent in the U.K. and
certain
Sweden, the importance of noncash income is again reinforced. The health compo-
We
have
nent of noncash income is greater than the education component in all countries
midway
except Canada, U.K. and the U.S. In the U.K., this mainly reflects the high level
These
of noncash education income, while in the U.S. it reflects the relatively low net
for
each
subsidy to health care. The differences between these categories in Canada are
what
fairly small.
These
The addition of housing benefits makes many of these differences more pro-
data
on
nounced. While on average, housing income in-kind is less than education and
ifferentials
health, the distribution of housing benefits across countries is very different than
that
a
the distribution of the other types of benefits. Canada and Germany have the
have
needs
largest amounts of noncash income of this sort. particularly Canada. As a result
adult,
the final ranking of noncash income in the five countries with housing as well
leaves Canada at the top of the heap followed by Sweden and Netherlands. West
equivalent
Germany and the U.S. bring up the rear.
should
depend
to
indi-
Living Standards
noncash
The effect of noncash income on the average income levels of household
a
whole
types-the effect on living standards of different household types-is shown in
241
Table 3 (for health and education only) and Table 4 (including housing). Living
standard impact was calculated by comparing overall average group income-
unadjusted disposable income and final income-to the national mean. Net differ-
ences in impact by family type are shown at the bottom of each table.
The bottom panel of Table 3 indicates that, relative to average incomes,
noncash income is greatest for middle-aged families with children and the very
elderly. The biggest relative losers in most countries are younger families without
children and childless couples, so-called "yuppies," and those approaching retire-
ment age. The size of the relative gains for families with children are greater than
those for the elderly in all countries. Among families without children, relative
losses are generally greater for couples than for the other groups. The impacts on
the elderly are generally more modest than one would have thought given their
relatively higher benefits from health care. The differences in health subsidy for
the aged versus other groups (adults, children) are clearly less than the differences
in education which directly benefit only one group: families with younger children.
Before the addition of noncash income, single parents with children, single
adults-aged and nonaged-and aged couples had below average disposable
incomes. Nonaged married couples, with and without children, and larger famil-
ies-generally those included under "other"-had higher incomes. Since we did
not adjust these incomes for family size, we clearly create upward bias in the
measured well-being of "other" categories. Still, the addition of noncash income
in the form of health and education most improves the position of single parents
with children. Single aged persons gain a small amount and aged couples hold
TABLE 2
their own (except in Sweden where the gains are large for the aged). The childless
nonaged lose-both the couples and others who already had above average
incomes, and also nonaged single persons whose cash incomes were below average
to start.
If one were to double the living standards of the singles (or double the
incomes of the couples), they would be much closer to each other, indicating that
overall living standards per capita are more for these people than shown in Table
3. Single parents with children-the least well-off group in cash terms in several
OVERALL MEAN AMOUNTS OF CASH AND NONCASH INCOME IN NATIONAL CURRENCY BY COUNTRY
of these nations-appear to gain most from this exercise. Their full incomes remain
below average (except for Germany and Sweden), but they are higher once income
in-kind is added in, than they were before.
The addition of housing benefits (Table 4) changes this picture only margin-
ally. The aged now gain more-due particularly to the higher fraction which own
homes in Canada and the U.S.-but otherwise the "winners" and "losers" are
still, respectively, the childful and the childless. It, thus, appears that the benefits
of home ownership as we have measured them, are fairly evenly distributed across
the eight groups of family types shown here.
INEQUALITY
The effect of these benefits on the overall size distribution of income is cap-
tured most simply in Table 5 (health and education only) and Table 6 (adding in
education). No adjustments to income are made for family size or type. The
bottom panel of each table again captures the difference due to noncash benefits.
242
ash benefits.
or type. The
6 (adding in
come is cap-
buted across
the benefits
"losers" are
n which own
only margin-
once income
omes remain
ms in several
own in Table
dicating that
r double the
'elow average
ove average
The childless
couples hold
single parents
cash income
1 bias in the
Since we did
larger famil-
e disposable
ildren, single
ble.
ager children.
he differences
1 subsidy for
it given their
e impacts on
dren, relative
greater than
aching retire-
nilies without
and the very
age incomes,
an. Net differ-
income- dne
using). Living
TABLE 2
OVERALL MEAN AMOUNTS OF CASH AND NONCASH INCOME IN NATIONAL CURRENCY BY COUNTRY
Final Income I
Final Income 2
Health and
Total Noncash
Health and
All Noncash
Country
Disposable
Education
Income
Education Only
Benefits
(Year)
Income
Education
Health
(3+4)
Housing
(3+4+6)
(2+5)
(2+7)
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
I. Amounts in National Currency
Australia (1981 82)
14,699
948
1.124
2,072
na
na
16,741
na
Canada (1981)
21,505
1,631
1,537
3,168
2,820
5,988
24,673
27,498
Netherlands (1983)
31,377
2,502
3,037
5,539
1,800
7,339
36,916
38,716
Sweden (1981)
64,283
5,399
8,653
14,052
3,717
17,769
78,335
82,052
243
U.K. (1979)
5,290
638
509
1,147
na
na
6,437
na
U.S. (1979)
14,338
1,091
774
1,865
508
2,373
16,203
16,711
West Germany (1981)
31,302
1,573
2,497
4,070
2,626
6,696
35,372
37,998
II. As Percent of Disposable Income
Australia
100
6.5
7.7
14.1
na
na
114.1
na
Canada
100
7.6
7.1
14.7
13.1
27.8
114.7
127.9
Netherlands
100
8.0
9.7
17.7
5.7
23.4
117.7
123.4
Sweden
100
8.4
13.5
21.9
5.8
27.6
122.0
127.6
U.K.
100
12.1
9.6
21.7
na
na
121.7
na
U.S.
100
7.6
5.4
13.0
3.5
16.6
113.0
116.6
West Germany
100
5.0
8.0
13.0
8.4
21.4
113.0
121.4
Simple Average"
100
7.9
8.7
16.6
7.3b
23.4b
116.6
123.4b
"Simple average is sum divided by the number of countries with each type of income.
"Averaged over five countries.
TABLE 3
EFFECT OF NONCASH INCOME (EDUCATION AND HEALTH ONLY) ON LIVING STANDARDS:
EFFECT OF
NET BENEFITS AS A PROPORTION OF OVERALL AVERAGE INCOME BY
FAMILY TYPE AND COUNTRY
Country
West
Family Type
Australia
Canada
Netherlands
Sweden
U.K.
U.S.
Germany
Family Type
I. Disposable Income
Families with
Families will
Children
Children
a. Nonaged couples
119
119
115
159
122
126
125
a. Nonaged
b. Nonaged single
b. Nonaged
parents
50
58
69
99
71
55
93
parents
C. Others"
163
151
127
128
176
146
149
C. Others"
Elderlyᵇ
Elderly
a. Single person
37
42
56
56
31
41
50
a. Single per
b. Couple
66
80
82
100
58
87
87
b. Couple
Nonaged without
Nonaged with
Children
Children
a. Single
60
59
58
68
58
61
65
a. Single
b. Couple
117
119
118
140
120
125
119
b. Couple
c. Other
154
135
120
na
135
135
139
c. Other
II. Final Income 1: Health and Education Only
Families with
Families with
Children
Children
a. Nonaged couples
125
126
121
167
137
133
131
a. Nonaged
b. Nonaged single
b. Nonaged
parents
61
71
79
114
91
70
100
parents
C. Others"
170
158
147
153
176
158
159
c. Others"
Elderlyᵇ
Elderly
a. Single person
39
47
56
69
33
43
48
a. Single per
b. Couple
68
84
84
111
57
86
86
b. Couple
Nonaged without
Nonaged will
Children
Children
a. Single
55
53
51
59
52
55
60
a. Single
b. Couple
108
108
104
124
104
114
112
b. Couple
C. Otherᶜ
147
126
115
na
119
126
133
c. Otherᶜ
III. Difference (II-I)
Families with
Families with
Children
Children
a. Nonaged couples
6
7
6
8
15
7
6
a. Nonaged
b. Nonaged single
b. Nonaged
parents
11
13
10
15
20
15
7
parents
C. Others
7
7
20
24
0
12
10
c. Others"
Elderlyᵇ
Elderlyᵇ
a Single person
2
5
0
13
2
2
-2
a Single pers
b. Couple
2
4
2
11
-1
-1
-1
b. Couple
Nonaged without
Nonaged with
Children
Children
a. Single
-5
-6
-7
-9
-6
-6
-5
a. Single
b. Couple
-9
-11
-14
-16
-16
-11
-7
b. Couple
C. Otherᶜ
-7
-9
-5
na
-16
-9
-6
C. Otherᶜ
"Other families with children include those with at least one parent over age 65 or children living
"Other f.
with more than three adults.
with more th
"Elderly are families with head or spouses over age 65.
Elderly
'Other families without children include those with three or more adults.
°Other f.
244
TABLE 4
TANDARDS:
EFFECT OF NONCASH INCOME (EDUCATION. HEALTH AND HOUSING) ON LIVING STANDARDS:
NET BENEFITS AS A PROPORTION OF OVERALL AVERAGE INCOME
BY FAMILY TYPE AND COUNTRY
Country
West
West
Germany
Family Type
Australia
Canada
Netherlands
Sweden
U.K.
U.S.
Germany
1. Disposable Income
Families with
Children
125
a. Nonaged couples
119
119
115
159
122
126
125
b. Nonaged single
93
parents
50
58
69
99
71
55
93
149
C. Others"
163
151
127
128
176
146
149
Elderlyᵇ
50
a. Single person
37
42
56
56
31
41
50
87
b. Couple
66
80
82
100
58
87
87
Nonaged without
Children
65
a. Single
60
59
58
68
58
61
65
119
b. Couple
117
119
118
140
120
125
119
139
c. Otherᶜ
154
135
120
na
135
135
139
II. Final Income 1: Health and Education Only
Families with
Children
131
a. Nonaged couples
na
125
122
166
na
131
133
b. Nonaged single
100
parents
na
72
77
116
na
71
96
159
C. Others"
na
157
149
152
na
156
160
Elderly
48
a. Single person
na
49
55
70
na
47
49
86
b. Couple
na
89
83
110
na
89
86
Nonaged without
Children
60
a. Single
na
51
50
61
na
55
59
112
b. Couple
na
109
104
122
na
114
112
133
c. Otherᶜ
na
128
115
na
na
126
135
III. Difference (II-1)
Families with
Children
6
a. Nonaged couples
na
6
7
7
na
5
8
b. Nonaged single
7
parents
na
14
8
17
na
16
3
10
C. Others"
na
6
22
24
na
10
11
Elderlyᵇ
-2
a Single person
na
7
-1
14
na
6
-1
-I
b. Couple
na
9
I
10
na
2
-1
Nonaged without
Children
-5
a. Single
na
-8
-8
-7
na
-6
-6
-7
b. Couple
na
-10
-14
-18
na
-11
-7
-6
C. Otherᶜ
na
-7
-5
na
na
-9
-4
Iren living
"Other families with children include those with at least one parent over age 65 or children living
with more than three adults.
"Elderly are families with head or spouses over age 65.
°Other families without children include those with three or more adults.
245
For the most part, noncash benefits from education and health are equalizing,
increasing the income share at the bottom and decreasing it at the top (Table 5).
EFFECTS OF
Effects are largest by far in Germany, followed by the U.K. and Canada. Effects
are least in the U.S. and even slightly disequalizing in Sweden at the top of the
distribution. The rank order of nations in terms of the income shares of the lowest
Quintile Share
of Income
quintile are unaffected by the addition of health and education benefits with the
exception of Germany which jumps to the highest with Sweden second. In all
nations, the bottom quintile does better with noncash benefits included. The U.S.
Lowest
Second
still has the lowest share for the bottom quintile, but it is now much closer to
Middle
Australia (second lowest) than before. Effects on the top quintile are generally
Fourth
small except in Germany. Here rank order changes slightly, with Germany again
Highest
Total
becoming the most equal (lowest upper quintile share), and Sweden moving to
second most equal. The rest of the rankings remain intact.
The addition of housing benefits (Table 6) has a substantial impact in Ger-
Lowest
Second
many, greatly reducing the size of the gains in distributional equality made by
Middle
health and education. Noncash benefits are still equalizing in West Germany, but
Fourth
not nearly so much as they were when only health and education were counted.
Highest
Total
No doubt this is in part due to the relatively small fraction of German households
which are home owners (35 percent). In contrast, the addition of housing benefits
Lowest
is decidedly more equalizing in the Netherlands, Sweden and Canada. In these
Second
nations, the housing effects reinforce those of health and education. The Nether-
Middle
lands now has the most equal final income distribution with the highest bottom
Fourth
Highest
quintile share and the lowest top quintile share. Sweden is second and Germany
in the middle. The U.S. remained most unequal with Canada second. The large
"Disposable
Final incom
amounts of noncash housing benefit in Canada therefore, seemed to have only a
modest import on their overall inequality rankings.
Clearly, the relationship between the relative size of benefits and their distri-
poverty which
butional impact is complex. The U.S. has the smallest total expenditure, yet has
the distributic
a larger impact than in Sweden. Sweden, on the other hand, has the largest
widely held vi
noncash sector but the overall least distributional effect. Canada and the Nether-
cannot be est:
lands tend to have the largest equalizing impacts from all three types of noncash
which needs a
benefits combined. West German housing benefits counteract the strong equalizing
We, thus
impact of health and education, leaving only a modest net impact on distribution.
defensible. He
the median le'
differences in
POVERTY
relative pover
are irrelevant
Measuring Poverty: Methods
relative pover
Of the many dimensions of comparative economic well-being that we are in
dependent on
a position to investigate comparatively on the basis of noncash income, perhaps
income) within
their poverty impact is most important. The effect of noncash income on living
as the percenta
standards and on the distribution of final income have been presented. Neither
below half of :
of these made adjustments for family size or need. Poverty measurement must,
half as compat
however, address these issues.
median incom
element which
The first step in this exercise is to select a poverty line. We decided against
the use of an absolute poverty line, partly on the grounds that the concept itself
"We will, he
conveys an unwarranted objectivity, but also because it would result in levels of
percent and 125 1
246
equalizing,
TABLE 5
(Table
5).
EFFECTS OF NONCASH INCOME ON THE OVERALL INCOME DISTRIBUTION BY COUNTRY
Effects
of
Country
top
the
the lowest
Quintile Share
West
of Income
Australia
Canada
Netherlands
Sweden
U.K.
U.S.
Germany
with
the
In
all
1. Disposable Income
The
U.S.
Lowest
5.4
5.4
6.9
8.3
5.9
4.7
7.0
Second
11.7
12.0
13.2
13.2
11.4
11.3
closer
13.1
to
Middle
18.0
18.2
18.0
17.6
18.2
17.7
17.7
generally
Fourth
24.9
25.0
23.7
24.3
25.0
25.5
24.1
again
Highest
40.0
39.4
38.2
36.7
39.5
4.7
38.1
Total
100.0
100.0
100.0
100.0
100.0
100.0
100.0
moving
to
II. Final Income 1:ᵇ Health, Education Only
in Ger-
Lowest
5.7
6.1
7.6
8.6
6.2
5.3
10.2
Second
11.8
12.4
13.2
13.0
11.6
11.6
15.7
made by
Middle
17.9
18.4
18.3
17.2
18.6
17.7
18.8
rmany, but
Fourth
24.9
25.0
23.8
24.4
25.4
25,4
22.7
counted.
Highest
39.7
38.1
37.2
36.8
38.2
40.0
32.6
Total
100.0
100.0
100.0
100.0
100.0
100.0
100.0
households
benefits
III. Difference (II-I)
Lowest
0.3
0.7
0.7
0.3
0.3
0.6
3.2
In these
Second
0.1
0.4
0.0
-0.2
0.2
0.3
2.6
he
Nether-
Middle
-0.1
0.2
0.3
-0.4
0.4
0.0
1.1
bottom
Fourth
0.0
0.0
0.1
0.1
0.4
-0.1
-1.4
Highest
-0.3
-1.3
-1.0
0.1
-1.3
-0.7
-5.5
Germany
The large
"Disposable income includes all forms of cash income net of income and payroll taxes.
Final income I adds the market value of health and education benefits to disposable income.
ave only a
heir distri-
poverty which differed according to the national standard of living as well as to
re, yet has
the distribution of income within each nation. It would, thus, conflict with the
he largest
widely held view among scholars working in this arena that a poverty standard
e Nether-
cannot be established independently of the economic and social context within
of noncash
which needs arise and are defined (Smeeding, Rainwater and O'Higgins, 1990).
equalizing
We, thus, regard the choice of a relative poverty measure as much more
tribution.
defensible. Here we pick a poverty line which is equal to the same fraction of
the median level of living (median disposable cash income after adjustment for
differences in need using the equivalence scale) in each nation. The use of this
relative poverty line implies that differences in living standards across countries
are irrelevant to the measurement of national poverty rates. The choice of a
relative poverty measure does. however, make the level of poverty in a country
we are in
dependent on the distribution of resources (adjusted cash or cash plus noncash
perhaps
income) within each nation. We have chosen to measure the incidence of poverty
on living
as the percentage of all families with adjusted incomes (cash or cash plus noncash)
1. Neither
below half of median adjusted income, even though there is nothing special about
must,
half as compared to 40 percent or 60 percent or some other percentage of adjusted
median income. Our basic poverty standard, thus, contains an explicitly subjective
di against
element which we accept as inevitable in any exercise such as this. 12
ept itself
"We will, however, test the sensitivity of our results using alternative poverty lines set at 75
levels of
percent and 125 percent of the half median disposable income poverty standard.
247
TABLE 6
income) poverty standard to
EFFECTS OF NONCASH INCOME ON THE OVERALL INCOME DISTRIBUTION BY COUNTRY
transfer (cash) income and po
Regardless of the choice (
Country
valued is very important. As
Quintile Share
West
market value or cost to the
of Income
Australia
Canada
Netherlands
Sweden
U.K.
U.S.
Germany
$2,500) of education benefits
1. Disposable Income"
$2,500) transfer in cash. Since
Lowest
5.4
5.4
6.9
8.3
5.9
4.7
7.0
choose to spend noncash trar
Second
11.7
12.0
13.2
13.2
11.4
11.3
13.1
Middle
18.0
18.2
18.0
17.6
18.2
17.7
17.7
of in education outlays or the (
Fourth
24.9
25.0
23.7
24.3
25.0
25.5
24.1
in income at their government
Highest
40.0
39.4
38.2
36.7
39.5
40.7
38.1
the true level of well-being of
Total
100.0
100.0
100.0
100.0
100.0
100.0
100.0
choice between £800 of educ:
II. Final Income 2:ᵇ Health and Education, and Housing
(e.g., £600) might prefer the (
Lowest
na
6.2
9.3
8.9
na
5.9
7.2
transfer overstates the true in
Second
na
12.4
14.6
13.3
na
11.9
12.8
Middle
na
18.5
18.3
17.4
na
17.9
17.7
this reason, differences betwe
Fourth
na
25.4
23.1
24.1
na
25.1
24.8
receipt of noncash income she
Highest
na
37.6
34.7
36.3
na
39.2
37.6
income. To the extent that far
Total
na
100.0
100.0
100.0
na
100.0
100.0
cost, their real incomes and h
III. Difference (II-I)
estimated here.
Lowest
na
0.8
2.4
0.6
na
1.2
0.2
Second
na
0.4
1.4
0.1
na
0.6
-0.3
In computing the effect o
Middle
na
0.3
0.3
-0.2
na
0.2
0.0
based on health and educatic
Fourth
na
0.4
-0.6
-0.2
na
-0.4
0.7
for homeowners is thus not c
Highest
na
-1.8
-3.5
-0.4
na
-1.5
-0.5
"Disposable income includes all forms of cash income net of income and payroll taxes.
Final income adds the market value of health, education and housing benefits to disposable
Measuring Poverty: Results
income.
Since we have chosen a
of adding in noncash income
These choices still leave unresolved the issue of whether the same poverty
what we find in Table 7. Hov
standard should be used to measure poverty on the basis of cash income alone
benefits varies across the seve
and according to the sum of cash and noncash income. There are good arguments
in poverty, noncash income I
both ways here. It can be argued to be most appropriate to define the poverty
ada, Australia, and the U.S.
line on the same basis as that used to define income itself. Thus, a cash poverty
Germany, the Netherlands, a
line should be used in conjunction with cash income, but when noncash income
was already much lower in
is included in the income measure, the poverty line should be re-defined accord-
initially highest in the first
ingly so as to comprise cash and noncash elements. 13 The main disadvantage of
marked when expressed in p:
this approach is that it is difficult to unravel the impact of noncash income on
The only significant cha
poverty when the poverty line itself is also changing.
in the U.K., where noncash
Since our interest is primarily in estimating the impact of noncash income
impact on poverty. The ranki
on poverty, we have rejected this approach in favor of one where the poverty line
noncash income is included
is fixed independently of the definition of income. We thus use a poverty line
both income measures, and
based on median adjusted disposable cash income throughout our analysis. This
basis of final income than or
allows us to see what difference the inclusion of noncash income makes to the
When poverty is measu
incidence of relative poverty when a common poverty standard is used. This
four European countries an
approach is no different in principle from that used in studies which estimate the
colonial nations becomes me
effects of government taxes and transfers on poverty by using a common (cash
"See, for example the poverty
"An obvious contender would be to set the poverty line equal to one-half of median adjusted
based) poverty standard was also us
final income rather than median adjusted disposable income.
in their calculations of the impact (
248
income) poverty standard to compare poverty estimates based on pre-tax, pre-
UNTRY
transfer (cash) income and post-tax, post-transfer (cash) income. 14
Regardless of the choice of poverty line, the issue of how noncash income is
valued is very important. As explained earlier, we value noncash income at its
West
Germany
market value or cost to the government. We therefore assume that £800 (or
$2,500) of education benefits for a family with one child is equal to an £800 (or
$2,500) transfer in cash. Since low income families with few cash resources might
7.0
choose to spend noncash transfers differently if they were given in cash instead
13.1
17.7
of in education outlays or the cost of medical insurance, the decision to count them
24.1
in income at their government cost or market value may lead to an overestimate of
38.1
the true level of well-being of such families. Stated differently, a family offered a
00.0
choice between £800 of education expenses or a lesser amount of cash income
(e.g., £600) might prefer the cash. If so, the £800 imputed value of the noncash
7.2
12.8
transfer overstates the true increase in the economic welfare of the family. For
17.7
this reason, differences between the estimates of poverty before and after the
24.8
receipt of noncash income should be treated as the maximum impact of noncash
37.6
00.0
income. To the extent that families would value these benefits at less than market
cost, their real incomes and hence, their poverty rates will change by less than is
estimated here.
0.2
-0.3
In computing the effect of noncash income on poverty, we only present results
0.0
based on health and education or final income 1. The addition of imputed rent
0.7
-0.5
for homeowners is thus not captured in the following analyses.
lisposable
Measuring Poverty: Results
Since we have chosen a common cash income based poverty line, the effect
of adding in noncash income can only be to reduce poverty. And, in fact, this is
poverty
what we find in Table 7. However, the results of adding in health and education
alone
benefits varies across the seven nations studied. In terms of the absolute reduction
:uments
in poverty, noncash income has the biggest impact in the U.K. followed by Can-
poverty
ada, Australia, and the U.S. The impact in the remaining three countries-West
poverty
Germany, the Netherlands, and Sweden-is much smaller, partly because poverty
income
was already much lower in these three countries. Since the poverty rates were
accord-
initially highest in the first group of four countries, these differences are less
tage
of
marked when expressed in proportionate terms, but they nonetheless, remain.
on
The only significant change in the ranking of national poverty rates occurs
in the U.K., where noncash income has the largest (absolute and proportional)
income
impact on poverty. The ranking of all other countries stays much the same whether
rty
line
noncash income is included or not. The U.S. has the lowest ranking according to
line
both income measures, and looks worse relative to the other countries on the
This
basis of final income than on the basis of disposable income.
to
the
When poverty is measured using final income, the distinction between the
This
four European countries and the three remaining larger and younger, former
the
colonial nations becomes more marked. Within these European nations, there is
(cash
"See. for example the poverty studies by Smeeding, Torrey and Rein (1988). A common (cash-
adjusted
based) poverty standard was also used by Paglin (1980) and by the U.S. Bureau of the Census (1982),
in their calculations of the impact of noncash benefits on poverty.
249
TABLE 7
benefits. The effects
FAMILY POVERTY RATES IN SEVEN NATIONS BASED ON ADJUSTED DISPOSABLE INCOME AND
The biggest impact (
FINAL INCOME I
poverty among singl
Adjusted Disposable Cash
Adjusted Final
U.K. There is also a
Income
Income 1
the effect of noncash
National
National
Here we also find de
Country (Year)
Amount"
Rank
Amount
Rank
Difference
and the U.K. The im
Australia (1981-82)
15.1
2.5
7.4
2
7.7
and the single elderly
Canada (1981)
15.1
2.5
7.2
3
7.9
terms and, in the latt
Netherlands (1983)
6.6
6
4.7
5
1.9
Sweden (1981)
5.6
7
4.3
6.5
1.3
U.K. (1979)
13.5
4
4.3
6.5
9.2
U.S. (1979)
18.5
I
12.1
I
6.4
West Germany
7.5
5
5.4
4
2.1
FAMILY POVERTY RAT
Note: Poverty rates are calculated as the percentage of families with adjusted incomes less than
half of national median adjusted disposable cash income.
"Adjusted disposable cash income is after-tax cash income adjusted for differences in family size
Fam
using the budget studies program equivalence scale.
Adjusted final income is adjusted disposable cash income plus the estimated market value of in-
kind benefits in the form of education and health care.
Nonag
The difference between the poverty rate based on cash income only and the poverty rate based
Country
Coupi
on cash plus noncash income.
Australia
8.8
Canada
8.9
little variation in overall poverty, the incidence of poverty being between 4 percent
Netherlands
1.6
and 5 percent. In the former colonies, in contrast, poverty ranges from 7 percent
Sweden
3.2
to 12 percent-far higher overall than in Europe and with a much more diverse
U.K.
3.6
U.S.
8.8
pattern. It is interesting to note that the lowest poverty rate in the non-European
West Germany
1.3
countries after the inclusion of noncash income (7.2 percent in Canada) is about
11
the same as the highest poverty rate in continental Europe before noncash income
Australia
2.6
is included (7.5 percent in West Germany). This is a dramatic indication of the
Canada
1.5
Netherlands
0.4
extent of the differences between relative poverty rates in the two groupings of
Sweden
0.8
countries included in this study.
U.K.
0.1
Differences in the level of poverty and the impact of noncash income are
U.S.
3.4
West Germany
0.4
shown by family type in Table 8. We begin by noting the wide variation in cash
income based poverty rates across countries. In no country do we find nonelderly
Australia
6.2
couples with or without children. to have double digit poverty rates. In contrast,
Canada
7.4
the poverty rate for nonaged single people exceeds 10 percent in all countries,
Netherlands
1.2
Sweden
2.4
while single elderly people have the highest poverty rates almost everywhere except
U.K.
3.5
in the Netherlands and Sweden. In all countries except the Netherlands and
U.S.
5.4
Sweden, the risk of poverty is much higher in families (with or without children)
West Germany
0.9
with only a single adult member than in families with two (or more) adults present.
Note: Poverty rates
Even in the Netherlands and Sweden, single adult families have among the highest
half of national median
"The elderly are fam
poverty rates. The highest poverty rates of all (well over 40 percent) are found
ᵇOther families with
among single parent families in Australia, Canada, and the U.S., and poverty
with more than two adu
°Other families with
rates for single parent families are well above the national average-two to three
"Adjusted disposable
times the national poverty rate-in all countries except the Netherlands, Sweden
using the budget studies
and West Germany.
"Adjusted final inco
benefits in the form of e
The inclusion of noncash income causes families with children to experience
The difference betw
large reductions in poverty in all nations, due mainly to the impact of education
on cash plus noncash in
250
benefits. The effects of health care benefits are mainly beneficial to the elderly.
INCOME AND
The biggest impact of noncash income in Australia, Canada, and the U.K. is on
poverty among single elderly persons living alone, and on elderly couples in the
U.K. There is also a large impact on single elders in the U.S., but it is less than
the effect of noncash income on single nonaged parents-or so-called lone parents.
Here we also find double digit reductions in their poverty in Australia, Canada
Difference
and the U.K. The impact of noncash income on these two groups (single parents
7.7
and the single elderly) causes their poverty rates to decline substantially in absolute
7.9
terms and, in the latter case, relative to the national poverty rate also. This leaves
1.9
1.3
9.2
6.4
TABLE 8
2.1
FAMILY POVERTY RATES IN SEVEN NATIONS BASED ON ADJUSTED DISPOSABLE INCOME AND
mes less than
FINAL INCOME I BY FAMILY TYPE
in family size
Nonaged without
Families With Children
Elderly"
Children
et value of in-
Nonaged
Nonaged
Single
Single
Single
rty rate based
Country
Couple
Parent
Otherd
Persons Couple
Persons
Couple
Otherᶜ
Total
I. Adjusted Disposable Cash Income
Australia
8.8
54.4
7.0
46.1
7.7
22.1
5.1
3.2
15.1
Canada
8.9
43.9
8.4
41.8
8.9
22.3
5.9
6.7
15.1
n 4 percent
Netherlands
1.6
5.4
19.4
4.9
1.4
15.7
0.9
13.1
6.6
1
7 percent
Sweden
3.2
5.4
0.6
1.1
0.3
12.1
2.4
-
5.6
're diverse
U.K.
3.6
26.3
1.1
50.3
23.5
18.8
2.5
2.8
13.5
U.S.
8.8
48.9
16.7
45.2
17.0
22.4
5.7
9.7
18.5
-European
West Germany
1.3
9.8
4.2
18.1
8.8
11.4
2.2
6.3
7.5
1) is about
II.
Adjusted
Final
Income
1:'
Health
and
Education
sh income
Australia
2.6
21.0
2.2
8.2
4.9
18.7
4.1
1.6
7.4
ion of the
Canada
1.5
18.2
1.0
9.4
1.3
20.5
4.3
2.4
7.2
upings of
Netherlands
0.4
0.0
9.6
4.9
1.0
15.7
0.8
8.7
4.7
Sweden
0.8
2.8
0.0
0.0
0.3
11.1
1.4
-
4.3
U.K.
0.1
0.4
0.2
18.6
1.1
13.2
1.1
1.0
4.3
come are
U.S.
3.4
21.1
3.9
33.9
8.9
21.1
5.3
7.6
12.1
West Germany
0.4
3.3
0.7
14.6
4.4
10.0
1.7
4.1
5.4
on in cash
III.
Difference'
onelderly
Australia
6.2
33.4
4.8
37.9
2.8
3.4
1.0
1.6
7.7
contrast,
Canada
7.4
25.7
7.4
32.4
7.6
1.8
1.6
4.3
7.9
countries,
Netherlands
1.2
5.4
9.8
0.0
0.4
0.0
0.1
4.4
1.9
Sweden
2.4
2.6
0.6
ere except
1.1
0.0
1.0
1.0
-
1.3
U.K.
3.5
25.9
0.9
31.7
22.4
5.6
1.4
1.8
9.2
ands and
U.S.
5.4
27.8
12.8
11.3
8.1
1.3
0.4
2.1
6.4
children)
West Germany
0.9
6.5
3.5
3.5
4.4
1.4
0.5
2.2
2.1
S present.
Note: Poverty rates are calculated as the percentage of families with adjusted incomes less than
the highest
half of national median adjusted cash disposable income.
are found
"The elderly are families with the head or spouse aged over age 65.
ᵇOther families with children include those with at least one parent over age 65 or children living
I poverty
with more than two adults.
to three
"Other families without children include those with three or more adults.
"Adjusted disposable cash income is after-tax cash income adjusted for differences in family size
Sweden
using the budget studies program equivalence scale.
"Adjusted final income 1 is adjusted disposable cash income plus the estimated value of in-kind
benefits in the form of education and health care.
perience
'The difference between the poverty rate based on cash income only and the poverty rate based
ducation
on cash plus noncash income.
251
single nonelderly adults as the group who miss out most from the benefits of
noncash income, having high poverty rates which are least impacted by education
SENSITIVITY OF FAMILY POVERTY ;
HEALTH AND EDUCATION
benefits and health benefits.
Sensitivity Analyses
Families with Chi
Nonaged
When relatively large numbers of families have incomes close to the poverty
Nonaged
Single
line, small changes in the level of the poverty line can have a large impact on
Country
Couple
Parent
estimates of the proportion of the population who are poor. One way to address
1. Poverty Lit
this issue is to use alternative indexes of the depth of poverty. The "poverty gap"
Australia
1.5
9.5
index, for example, is less sensitive than the poverty rate to small changes in the
Canada
0.9
9.7
poverty line. An alternative approach to the sensitivity issue is to examine it
Netherlands
0.3
0.0
Sweden
0.3
1.5
directly by retaining the same poverty rate measure but to consider changes in
U.K.
0.1
0.0
the level of the poverty line itself. We chose this latter method.
U.S.
1.7
13.1
By recalculating poverty estimates for alternative poverty lines set above and
West Germany
0.2
2.0
below the benchmark poverty line of 50 percent of median adjusted disposable
II. Benchmark Poverty Line
cash income, the extent of income clustering in the region of the poverty line can
Australia
2.6
21.0
Canada
1.5
18.2
be ascertained and its significance for our conclusions assessed. We have, thus,
Netherlands
0.4
0.0
recalculated some of the earlier estimates using poverty lines set 25 percent below
Sweden
0.8
2.8
and 25 percent above our benchmark poverty line. 15 As before, these poverty lines
U.K.
0.1
0.4
U.S.
3.4
21.1
are used to estimate poverty before and after the inclusion of noncash income
West Germany
0.4
3.3
(final income 1) in the income measure.
III. Poverty Li:
Table 9 presents the results by family type. At the lower poverty line, the
Australia
5.6
38.4
pattern of poverty indicated by our benchmark poverty line remains virtually
Canada
3.5
31.3
unchanged. In most countries, poverty is highest among nonelderly single people,
Netherlands
0.8
3.5
Sweden
1.5
5.8
single parent families, and single elderly people, in that order. At the higher
U.K.
0.4
1.8
poverty line, poverty amongst the elderly rises sharply, particularly among single
U.S.
6.5
31.3
elderly people, except in the Netherlands and Sweden where overall poverty among
West Germany
0.6
7.0
the elderly remains below the national average. The poverty rates of most non-
Note: Poverty rates are calculated
elderly family types remain unchanged relative to the national poverty rate as the
half of national median adjusted cash d.
poverty line is varied.
"The elderly are families with the he
ᵇOther families with children include
As the poverty line varies, the ranking of countries by the overall final income
with more than two adults.
poverty rate also undergoes several noticeable changes. At our benchmark poverty
"Other families without children inc
line, the total poverty rate is lowest in Sweden and the U.K., followed by the
Netherlands, West Germany, Canada, Australia, and the U.S., in that order. At
the lower poverty line, the U.K. clearly has the lowest poverty rate, while the
SUMMA
rankings of Australia and West Germany improve and those of the Netherlands
The main aim of this pape
and Sweden worsen. At the higher poverty line, Sweden and the Netherlands have
income-health and education
the lowest poverty by a considerable margin, while the U.K.'s ranking drops
income distribution and povert
markedly. These changes in poverty rankings mean that comparisons across our
criticism and may, on average,
seven countries are sensitive to where the poverty line is set, a point noted for
on low (cash) incomes, the resu
cash income based estimates of poverty by Mitchell (1991). The U.S. is the only
The impact of noncash in
country whose ranking is unchanged for all three poverty lines. It has the highest
Education accrues to families ,
poverty in all cases.
fits-though received by all-a
inclusion of noncash income the
"These alternative poverty lines thus corresponds to 37.5 percent and 62.5 percent of median
hence average living standards
adjusted disposable cash income, respectively.
the elderly. In contrast, nonelde
252
of
TABLE 9
SENSITIVITY OF FAMILY POVERTY RATES BASED ON ADJUSTED DISPOSABLE INCOME PLUS
HEALTH AND EDUCATION BENEFITS (FINAL INCOME 1) BY FAMILY TYPE
Nonaged without
Families with Children
Elderly*
Children
overty
Nonaged
Nonaged
Single
Single
Single
on
Country
Couple
Parent
Otherᶜ
Persons Couple
Persons Couple
Otherᶜ
Total
ddress
gap"
I. Poverty Line = 0.75 X Benchmark Poverty Line
in
the
Australia
1.5
9.5
1.4
1.3
1.9
10.3
2.2
0.5
3.7
Canada
0.9
9.7
0.5
1.5
0.4
13.9
2.0
0.8
4.0
it
Netherlands
0.3
0.0
6.3
4.4
0.5
14.0
0.5
6.9
3.8
in
Sweden
0.3
1.5
0.0
0.0
0.2
7.5
0.6
-
2.8
U.K.
0.1
0.0
0.0
0.7
0.0
4.2
0.1
0.4
0.7
U.S.
1.7
13.1
1.7
12.7
3.0
14.1
2.8
4.6
6.6
and
West Germany
0.2
2.0
0.4
4.3
1.9
4.9
1.3
1.0
2.2
sable
II.
Benchmark
Poverty
Line
H
0.50
X
Median
Adjusted
Disposable
Cash
Income
can
Australia
2.6
21.0
2.2
8.2
4.9
18.7
4.1
2.2
7.4
thus,
Canada
1.5
18.2
1.0
9.4
1.3
20.5
4.3
2.4
7.2
below
Netherlands
0.4
0.0
9.6
4.9
1.0
15.7
0.8
8.7
4.7
Sweden
0.8
2.8
0.0
0.0
0.3
11.1
1.4
--
4.3
lines
U.K.
0.1
0.4
0.2
18.6
1.1
13.2
1.1
1.0
4.3
come
U.S.
3.4
21.1
3.9
33.9
8.9
21.1
5.3
7.6
12.1
West Germany
0.4
3.3
0.7
14.6
4.4
10.0
1.7
4.1
5.4
the
III. Poverty Line = 1.25 X Benchmark Poverty Line
Australia
5.6
38.4
3.2
58.2
10.1
28.8
7.4
3.2
16.0
Canada
3.5
31.3
2.9
32.2
5.6
27.9
6.4
5.9
12.7
ople,
Netherlands
0.8
3.5
13.4
5.6
1.4
20.4
1.3
11.6
6.4
igher
Sweden
1.5
5.8
0.0
0.4
0.3
15.4
2.8
-
6.3
U.K.
0.4
1.8
0.6
60.1
24.3
22.4
3.4
5.2
14.0
U.S.
6.5
31.3
7.2
51.2
17.7
29.2
7.8
10.8
18.4
West Germany
0.6
7.0
1.0
29.8
10.4
18.8
2.5
7.3
10.6
Note: Poverty rates are calculated as the percentage of families with adjusted incomes less than
the
half of national median adjusted cash disposable income.
"The elderly are families with the head or spouse aged over age 65.
ᵇOther families with children include those with at least one parent over age 65 or children living
tome
with more than two adults.
'Other families without children include those with three or more adults.
the
At
the
SUMMARY AND CONCLUSIONS
have
The main aim of this paper has been to summarize the impact of noncash
income-health and education benefits, and imputed rent-on living standards,
our
income distribution and poverty. Although our valuation methods are open to
for
criticism and may, on average, overstate the value of noncash benefits for those
on low (cash) incomes, the results are nonetheless interesting and informative.
The impact of noncash income is best viewed within a life cycle context.
Education accrues to families with school-age children, while health care bene-
fits-though received by all-are disproportionately high for the elderly. The
inclusion of noncash income thus has the largest impact on the final incomes, and
hence average living standards and poverty rates, of families with children and
the elderly. In contrast, nonelderly single people, particularly young single people,
253
and nonaged families without children find their relative income positions are
U.S. estimates do I
worsened by the inclusion of noncash income. Since single elderly persons and
are so included, the
single parents on average have low living standards, these benefits have a large
decades up to the
impact on their well-being.
increased poverty
Housing benefits, in contrast, have benefits which are difficult to predict. We
fuel the generation:
show only their impact on living standards and inequality, and here for only five
Our results ind
of the seven nations. Patterns of home ownership are likely to benefit the elderly
noncash income ten
and families with children less than they benefit other groups, though other factors,
cycle and although
(e.g., the percent of households who are owners) may be at work as well.
suggest that interge
In all cases, nonaged single persons do least well. They are less likely to be
indicated. What is :
homeowners; they do not have school children; and, their health benefits (and
case among the seve
needs) are only average-somewhere between the lower benefits (and needs) of
debates in, the U.S.
children and the higher ones of the aged.
had to resolve quest:
The distributional results were striking mainly because of two factors. The
and transfer, cash a
strong equalizing impact of noncash benefits in all countries, and also then, the
evened out life cycle i
lack of a large differential net impact on country ranking (with only the exception
that equity tensions h
of Germany where health and education inequality reductions were matched by
which this kind of res
the disequalizing effect of housing benefits).
While different ju
APPENDIX TABLE A-1
yield different findings
of national differences
FREQUENCY DISTRIBUTION OF FAMILIES BY TYPE
different from those V
Country
income alone (e.g., see
West
inclusion of noncash in
Australia
Canada
Germany
Netherlands
Sweden
U.K.
U.S.
the four European COL
more marked. Aside fro
Families with Children
Nonaged couples
28
27
23
28
20
29
23
reinforces the redistribi
Nonaged single
nisms rather than actin
parents
4
5
2
3
4
3
6
Others
7
9
7
9
0
7
9
Elderly
Single person
8
8
16
10
16
12
9
Couple
8
7
10
9
10
10
8
Nonaged without
Children
Atkinson, A. B., The Econo
Single
21
20
18
12
34
12
21
Atkinson, A. B., Rainwater
Couple
14
14
14
17
15
15
14
Evidence from the Luxe
Other
10
11
10
12
na
12
10
Buhmann, B., Rainwater,
Total
100
100
100
100
100
100
100
Inequality and Poverty
Income Study (LIS) D
"Other families with children include those with at least one parent over age 65 or children living
Mitchell, D. Income Transf
with more than two adults (children are age 17 or younger).
OECD. Social Expenditure
"Elderly are families with head or spouses over age 65.
O'Higgins, M., The Allocati
"Other families without children include those with three or more adults, including adults 18
in Palmer, J. L., Sme
years old or over.
Urban Institute, Wash
O'Higgins, M. and Rugg
Previous research using the LIS database has shown that, on a cash income
Households in the U.I
basis, poverty in the early eighties was higher among families with children than
O'Higgins, M., Schmaus,
Microdata Analysis fo
among the elderly (Smeeding, Torrey and Rein, 1988), and that these patterns
June 1989.
changed little during the 1990s (Smeeding, 1992). Furthermore, when noncash
Paglin, M., Poverty and 7
benefits for food, housing and health care are counted, poverty among the elderly
Preston, S., Children and
in the U.S. dropped enormously, further emphasizing the point that the elderly
Vol. 21, pp. 435-457,
Ruggles, P. and O'Higgin
had been doing relatively well (U.S. Bureau of the Census, 1982). However, the
the U.S., Review of /
254
positions
are
U.S. estimates do not take account of education benefits to the young. When they
persons
and
are so included, the differences between these groups drops dramatically. The two
have
a
large
decades up to the 1980s had seen a decline in poverty among the elderly and
increased poverty among families with children. These developments helped to
predict.
We
for
fuel the generational equity debate in the U.S. and elsewhere (Preston, 1984).
only
five
Our results indicate that once both health benefits and education are counted,
the
elderly
noncash income tends to even out fluctuations in the risk of poverty over the life
factors,
well.
cycle and although single nonaged people miss out relatively speaking, our results
likely
suggest that intergenerational inequities may be less than previous research has
to
be
indicated. What is also clear, however, is that the U.S. is something of a polar
enefits
(and
case among the seven countries studied here. This suggests that findings for, and
needs)
of
debates in, the U.S. do not necessarily apply to other nations. Each country has
had to resolve questions of intergenerational equity using a combination of tax
actors.
The
and transfer, cash and noncash subsidy programs. Those policies have clearly
then,
the
evened out life cycle income fluctuations in all countries, but this is not to suggest
exception
that equity tensions have all been resolved. That is essentially a national question
natched
by
which this kind of research cannot answer.
While different judgements on measuring and valuing noncash benefits might
yield different findings, the basic results in this paper do not give rise to a pattern
of national differences in poverty rates or income inequality which are markedly
different from those which emerge from previous LIS research based on cash
income alone (e.g., see O'Higgins, Schmaus and Stephenson, 1989). However, the
inclusion of noncash income thus makes the distinction in poverty profiles between
U.S.
the four European countries and the three colonial nations in this study much
more marked. Aside from this important finding, it appears that noncash income
23
reinforces the redistributive impact of conventional (cash) tax-transfer mecha-
6
nisms rather than acting to offset them in any major way.
9
9
8
REFERENCES
Atkinson, A. B., The Economics of Inequality (Second Edition). Clarendon Press, Oxford, 1983.
21
Atkinson, A. B., Rainwater L., and Smeeding. T., Income Distribution in OECD Countries: The
14
Evidence from the Luxembourg Income Study, Mimco, Walferdange. Luxembourg, April 1993.
10
Buhmann, B., Rainwater, L., Schmaus, G., and Smeeding, T., Equivalence Scales, Well-Being,
100
Inequality and Poverty: Sensitivity Estimates across Ten Countries Using the Luxembourg
living
Income Study (LIS) Database, Review of Income and Wealth, 34(2). pp. 115-142, June 1988.
Mitchell, D. Income Transfer in Ten Welfare States, Aveburg Press, Hampshire. U.K., 1991.
OECD, Social Expenditure 1960-1990. Problems of Growth and Control, OECD. Paris, 1985.
adults
18
O'Higgins, M., The Allocation of Public Resources to Children and the Elderly in OECD Countries,
in Palmer, J. L., Smeeding, T., and Torrey, B. B. (eds.), The Vulnerable. pp. 201-228. The
Urban Institute, Washington, DC, 1988.
income
O'Higgins, M. and Ruggles, P., The Distribution of Public Expenditures and Taxes among
Households in the U.K., 27(3), pp. 298-326, Review of Income and Wealth. September 1981.
than
O'Higgins, M., Schmaus, G., and Stephenson, G., Income Distribution and Redistribution: A
patterns
Microdata Analysis for Seven Countries, Review of Income and Wealth, 35(2), pp. 107-132,
noncash
June 1989.
Paglin, M., Poverty and Transfers In-Kind, Hoover Institution Press, Stanford. 1980.
elderly
Preston, S., Children and the Elderly: Divergent Paths for America's Dependents, Demography,
elderly
Vol. 21, pp. 435-457. 1984.
the
Ruggles, P. and O'Higgins, M., The Distribution of Public Expenditure Among Households in
the U.S., Review of Income and Wealth, 27(2), pp. 137-164, June 1981.
255
Smeeding, T. M., O'Higgins, M., and Rainwater, L. (eds.), Poverty, Inequality and Income
Review of Income and Wealth
Distribution in Comparative Perspective, The Luxembourg Income Study, Harvester Wheatsheaf
Series 39, Number 3, September 19
London, 1990.
Smeeding, T. M., and Torrey, B., Goldmines and Minefields. A Summary of the LIS Conference: The
Changing Structure of Income and Social Policy in Eastern Europe: A Comparative Focus
Luxembourg Income Study, Working Paper # 68. Walferdange, Luxembourg, August 1991.
Smeeding, T. M., Saunders, P., Jenkins, S. et al., Noncash Income, Inequality and Living Standards
THE EARNINGS-RELA
in
Seven Nations, Mimeo, Luxembourg Income Study, Walferdange, Luxembourg, 1993.
LIFETIME F
Smeeding, T. M., Torrey, B., and Rein, M., Patterns of Income and Poverty: The Economic Status
of Children and the Elderly in Eight Countries, in Palmer, J. L., Smeeding, T., and Torrey, B.
(eds.), The Vulnerable, pp. 89-119, The Urban- Institute, Washington, DC, 1988.
Smeeding, T. M., "Why the U.S. Anti-Poverty System Doesn't Work Very Well," Challenge, pp:30-
35, January-February, 1992.
Teglarsky, J. and Struyk, R., Toward a Market-Oriented Housing Sector in Eastern Europe: Develop
ments in Bulgaria, Czechoslovakia, Hungary, Poland, Romania, and Yugoslavia, Urban Institute
Press, Washington DC, 1990.
U.S. Bureau of the Census, Alternative Estimates of the Impact of Noncash Transfers on Poverty,
University of Kent
Technical Paper # 50. U.S. Government Printing Office, Washington, DC, 1982.
E
The redistributive impact of the U.K.
lated using individual lifetime earnin
section data. The scheme is investiga
Bolate the intra-generationally redistr
The results suggest that different
mortality. which outweigh the redistr
the pension scheme are then simula
a greal deal of care is needed in form
How much will it cost to F
mos from pension schemes am
redis tributive impacts of widel
Current and expected pension
holdings in industrialised cour
public debates concerning opt
require detailed microsimulatic
Those involved in pension
ch models and the enforce
nerther clear distributiona
Note: The authors would like to
Centre for Fiscal Policy at the
Department of Employment gave acce
&
1992 Royal Economic Society, at the
the Econometrics of Social Protec
ents as are specifically Rebecca
of this journal, none of whom
On the United States, see inter a
1990. On the United Kingdom,
Wolfson, 1987; on France
256