ArticlePDF Available

Poverty Alleviation versus Social Insurance Systems: A Comparison of Lifetime Redistribution

Authors:
POVERTY ALLEVIATION VERSUS
SOCIAL INSURANCE SYSTEMS:
A COMPARISON OF
LIFETIME REDISTRIBUTION
Jane Falkingham
and
Ann Harding
Discussion Paper No. 12
April 1996
National Centre for Social and Economic Modelling
• Faculty of Management • University of Canberra •
The National Centre for Social and Economic Modelling
was established on 1 January 1993, following a contract
between the University of Canberra and the then federal
Department of Health, Housing, Local Government and
Community Services (now Health and Family Services).
NATSEM aims to enhance social and economic policy
debate and analysis by developing high quality models,
applying them in relevant research and supplying
consultancy services.
NATSEM’s key area of expertise lies in developing and
using microdata and microsimulation models for a range
of purposes, including analysing the distributional impact
of social and economic policy. The NATSEM models are
usually based on individual records of real (but
unidentifiable) Australians. This base produces great
flexibility, as results can be derived for small subgroups
of the population or for all of Australia.
NATSEM ensures that the results of its work are made
widely available by publishing details of its products and
research findings. Its technical and discussion papers are
produced by NATSEM’s research staff or visitors to the
centre, are the product of collaborative efforts with other
organisations and individuals, or arise from
commissioned research (such as conferences). Discussion
papers present preliminary research findings and are only
lightly refereed.
It must be emphasised that NATSEM does not have
views on policy and that all opinions are
the authors’ own.
Director: Ann Harding
National Centre for Social and Economic Modelling
• Faculty of Management • University of Canberra •
POVERTY ALLEVIATION VERSUS
SOCIAL INSURANCE SYSTEMS:
A COMPARISON OF
LIFETIME REDISTRIBUTION
Jane Falkingham
and
Ann Harding
Discussion Paper No. 12
April 1996
iii
ISSN 1320 3398
ISBN 0 85889 508 0
© NATSEM, University of Canberra 1996
National Centre for Social and Economic Modelling,
GPO Box 563
Canberra ACT 2601
Australia
Phone: +61 6 275 4900 Fax: +61 6 275 4875
E-mail: Client services hotline@natsem.canberra.edu.au
General natsem@natsem.canberra.edu.au
Core funding for NATSEM is provided by the federal Department of
Health and Family Services.
iii
Abstract
Among other objectives, modern social security and taxation systems
seek to redistribute income from those with high to those with low
incomes and to reallocate income across the life cycle of individuals.
While there are numerous studies of the extent to which these aims are
achieved during a short period, such as a week or a year, there are
almost no studies of the lifetime redistributional impact of tax and
transfer systems.
In this paper, two broadly comparable dynamic cohort microsimulation
models are used to assess the lifetime redistributional impact of the
British and Australian social security and direct taxation systems. The
analysis suggests that the Australian system, with its emphasis on
poverty alleviation, results in greater lifetime interpersonal
redistribution than the British social insurance-based system.
iv
Authors’ note
Jane Falkingham is a lecturer in population studies in the Department of
Social Policy at the London School of Economics and a Research
Associate of the Welfare State Program at the London School of
Economics. Ann Harding is Professor of Applied Economics and Social
Policy at the University of Canberra and inaugural Director of the
University’s National Centre for Social and Economic Modelling.
Acknowledgments
The authors would like to thank Steve Gribble of Statistics Canada for
detailed helpful comments on an earlier version of this paper. This is a
revised version of a paper presented at the December 1993 International
Association for Research in Income and Wealth Special Conference on
Microsimulation and Public Policy in Canberra, and the authors would
also like to thank participants at that conference for their comments.
General caveat
NATSEM research findings are generally based on estimated
characteristics of the population. Such estimates are usually derived
from the application of microsimulation modelling techniques to
microdata based on sample surveys.
These estimates may be different from the actual characteristics of the
population because of sampling and non-sampling errors in the
microdata and because of the assumptions underlying the modelling
techniques.
The microdata do not contain any information that enables identification
of the individuals or families to which they refer.
v
Contents
Abstract ....................iii
Authors’ note ....................iv
Acknowledgments ....................iv
General caveat ....................iv
1. Introduction ......................1
2. The models ......................3
2.1 Similarities ..........................3
2.2 Differences ..........................5
2.3 Caveats and qualifications ..........................6
3. The systems ......................8
3.1 The British social insurance system ..........................8
3.2 The Australian social assistance model ........................11
4. Lifetime redistribution: social insurance versus
poverty alleviation systems ....................13
5. Intrapersonal versus interpersonal redistribution ....................23
5.1 Financing from direct taxes ........................24
5.2 Financing from all taxes ........................28
6. Summary ....................29
References ....................34
1
1. Introduction
All modern social security systems serve two linked, but distinct, aims.
On the one hand, they are an updated ‘Robin Hood’ mechanism for
transferring resources raised through taxation from those with high
incomes to those with low incomes — termed interpersonal redistribution.
On the other hand, they are also intended to transfer resources from the
time in life when incomes are relatively high to other times when the
incomes of the same individuals are low — intrapersonal redistribution.
To date, most distributional analyses have focused on the first of these
aims, largely consisting of studies of the annual net fiscal incidence of
taxes and cash transfers (Central Statistical Office (UK) 1992; Harding
1984; Saunders 1984). These studies thus examine the redistributional
impact of government during a short period such as a week or a year.
However, a very different distributional picture emerges when the
effects of social security and taxation are examined over lifetimes
(Falkingham and Hills 1995; Harding 1993a; Nelissen 1993).
Social security systems can be broadly classified as:
social insurance-based systems, where individuals secure
entitlements to a range of income support payments because of their
identifiable social insurance contributions while in the labour
market; and
social assistance systems, where entitlement to a benefit is based on
citizenship rather than contribution, but where entitlement is subject
to the imposition of income and assets tests and, frequently for those
of working age, requirements to engage in job search and training
activities.
In reality, few social security systems are purely insurance-based, most
having some elements of social assistance to cover individuals who fail
to meet the contributory requirements.
One might predict that a primarily social insurance-based system would
lead to a higher degree of intrapersonal redistribution of resources over
a lifetime and that a primarily social assistance-based system, with its
emphasis on poverty alleviation, would result in a greater degree of
interpersonal redistribution. However, some degree of convergence in
the redistributional impact of different systems when looked at from a
2 Discussion Paper No. 12
lifetime perspective might also be expected. The ‘insurance’ principle
leads, after the event, to
interpersonal redistribution,
as some will suffer
the contingencies insured, such as sickness and unemployment, while
others will not. Similarly, the length of retirement will vary because of
different life spans. The ‘social’ insurance principle that contributions
should not reflect risk, that ‘none should claim to pay less because he is
healthier or in more regular employment’, also leads to interpersonal
redistribution (Beveridge 1942, p. 13).
Likewise, recipients of income support payments under a poverty
alleviation system may, when viewed across the life cycle, simply be
receiving the taxes they had paid earlier. At any point in time, a large
proportion of those with low incomes are retirees who might have
enjoyed higher incomes in the past, or students who will probably earn
much higher incomes in the future. Social assistance schemes therefore
implicitly reallocate income over the life cycle, introducing an element of
intrapersonal distribution.
To examine whether a poverty alleviation system results in greater
lifetime
interpersonal redistribution
requires data for complete lifetimes,
which are exceedingly rare. Given this lack of data, an alternative is to
examine what Summers (1956) referred to as the ‘latent lifetime income
distribution’ — that is, the lifetime distribution that would result if
current conditions continued indefinitely. Economists and economet-
ricians have used a number of techniques to simulate lifetime income
profiles, but these have not allowed for the real life continual changes in
the circumstances of individuals. The relatively recent technique of
dynamic microsimulation, in contrast, allows the characteristics of
individuals within the model to change constantly so that individuals
may, for example, enter or leave the labour force, get married or
divorced, and bear children (see Harding 1993a). One of the most
striking findings from longitudinal data is how much diversity people
experience over their lifetimes. Thus, dynamic microsimulation models
are potentially a significant improvement on the static models used by
econometricians in the past.
This paper investigates the hypothesis using data from two dynamic
cohort microsimulation models developed in parallel for Australia
(HARDING) and Great Britain (LIFEMOD). The examples of Australia
and Britain are particularly germane to the question at hand. In the
1980s Australia undertook a radical reform of its social security system
Poverty Alleviation versus Social Insurance Systems 3
that left it with perhaps the ‘purest’ social assistance system in the
industrialised world. The British system, in turn, still has its foundations
in the Beveridge report and remains the epitome of a social insurance-
based system.
Chapter 2 discusses the construction of the two dynamic cohort models,
focusing on their similarities and differences. Although developed in
parallel and sharing a similar modular structure, the two models are not
identical, perhaps more akin to ‘kissing cousins’ than sisters. Therefore,
it is important to highlight where differences may be due to methodol-
ogy rather than the operation of the respective social security systems.
Chapter 3 summarises the social security systems in Australia and
Britain, as modelled in HARDING and LIFEMOD. Chapters 4 and 5
contrast how these examples of social insurance and poverty alleviation
systems perform in the lifetime income redistribution stakes. The
chapters seek to provide answers to such questions as: how redistri-
butional overall are the systems; what balance between intrapersonal
and interpersonal redistribution do they produce; how far do the two
systems converge in outcomes when compared over the lifetime; and,
finally, is a poverty alleviation system really just that or does it, over the
lifetime, imitate a social insurance system? Finally, chapter 6
summarises the main findings.
2. The models
The construction of both of the dynamic cohort models employed in this
paper has been usefully described elsewhere (for HARDING see
Harding 1993a; for LIFEMOD see Falkingham and Lessof 1991, 1992).
Nevertheless, it is instructive to briefly describe their development,
pointing out their similarities and differences.
2.1 Similarities
Both models simulate complete life histories for a pseudocohort of 2000
males and 2000 females. These individuals are dynamically aged from
birth to death and experience major life events such as schooling,
marriage, childbirth, children leaving home, employment and
4 Discussion Paper No. 12
retirement. At this stage, the cohort experiences no immigration or
emigration; the only way in which the cohort changes size is from
attrition due to mortality.
It is important to point out that the microunit that both HARDING and
LIFEMOD simulates is the individual rather than the family or the
household. Therefore it is only the characteristics of the cohort
individuals that are modelled. For example, there is no comprehensive
information on the children of each cohort member, but simply their age,
parity, and whether they are participating in full-time education (that is,
whether they are classified as dependent children). HARDING and
LIFEMOD do not contain detailed information about the household
composition of cohort members, unlike other microsimulation models
such as that developed at Tilburg University (Nelissen 1993). However,
since cohort members are selected to marry each other, the character-
istics of an individual’s spouse are available at each stage during their
union. This is of particular importance in the labour market, income tax
and social security modules, where these characteristics interact and the
characteristics of both partners in the preceding period determine the
characteristics of the present period. Specification of an individual’s
income also depends on the spouse’s income.
The simulation process requires transition probabilities or behavioural
equations as input for every variable generated. The various prob-
abilities of demographic and other events occurring to people are
estimated in the two models from a variety of sources (most notably
published official statistics and secondary analysis of large sample social
surveys such as the Australian income distribution survey and the UK
general household and family expenditure surveys).
Given the uncertainty surrounding future changes in marriage and birth
rates, labour force participation rates and so on, both models assume a
steady state world and, as far as possible, all input data are calculated
for a specific year. Therefore, the HARDING cohort is ‘born’ into and
lives in a world that looks like Australia in 1986. Unfortunately, because
of data availability, the two models were not constructed on the same
base year. The LIFEMOD cohort lives in a world that looks like Britain in
1985. Although the steady state assumption results in a highly stylised
‘population’, it has been found to provide a useful benchmark against
which current government policies and changes to those policies can be
evaluated. Both the Canadian DEMOGEN and German SFB3 dynamic
Poverty Alleviation versus Social Insurance Systems 5
cohort models also assume a steady state world when evaluating the
impact of existing and potential government policies (Wolfson 1988,
p. 233; Hain and Helberger 1986, p. 63).
2.2 Differences
Apart from the transition probabilities to which each cohort member is
exposed, and the institutional structure of taxes and benefits of the
Australian and British social security systems, there are several key pro-
gramming differences that should be highlighted. The two models use
common core code for the initial modules, particularly those concerned
with family formation and dissolution. However, the models diverge in
the approach taken to simulate labour force participation and earnings.
HARDING models the number of hours spent in paid employment, self-
employment or unemployment at a given age, whereas LIFEMOD
models the number of weeks spent in any given state during the year.
HARDING also uses techniques to generate the inequality apparent
within the labour market, such as the existence of concentrated long
term unemployment. LIFEMOD, like the DEMOGEN and SFB3 models,
uses econometric techniques to simulate the decision to enter or leave
the workforce, while the HARDING model follows the DYNASIM
approach of using simple matrices of probabilities of participation
(Orcutt, Caldwell and Wertheimer 1976). However, the HARDING
matrices include most of the dependent variables in the LIFEMOD
equations and both studies found educational attainment and workforce
participation in the previous year to be among the main explanatory
variables.
Because of data limitations, female labour force participation was
modelled independently of spouse’s employment patterns in
HARDING. In LIFEMOD, however, the employment and earnings of
married women are dependent on their husband’s employment status
and income. This might create more ‘double-unemployed’ couples in
Britain than in Australia, given the well known disincentive effect of a
husband’s unemployment on a wife’s labour supply. However, there is
little evidence of such an effect when the lifetime distribution of earnings
is examined.
Another difference is in the treatment of maintenance income. Both
HARDING and LIFEMOD model the receipt of maintenance income by
6 Discussion Paper No. 12
sole mothers following divorce. However, in LIFEMOD payment of
maintenance income by ex-husbands was also modelled. This is taken to
be ‘negative’ income and thus depresses men’s original income. This
may account for the somewhat flatter distribution of original income in
Britain than in Australia. However, this paper compares the change in
the distribution between original and gross income (that is, the role of
cash benefits) and between gross income and net income (that is, the role
of taxes) rather than examining the shape of the original income
distribution. Consequently, differences in modelling methodology are
not expected to greatly influence the overall results.
2.3 Caveats and qualifications
The use of HARDING and LIFEMOD in distributional analysis of the
kind presented in this paper involves a number of difficulties whose
significance should be fully appreciated when interpreting the sub-
sequent results. The problems apply equally to HARDING and
LIFEMOD and a common approach has been taken to their resolution.
Thus, if biases exist, they should apply equally to the two models.
Indirect and local government taxes
Both models currently ignore indirect taxes and are limited to the major
cash transfers and income taxes administered by the central govern-
ments. Neither HARDING or LIFEMOD include taxes levied by other
levels of government
1
, the inclusion of which could (along with expendi-
ture taxes) make a substantial difference to the findings. In the United
Kingdom in 1985, income taxes and national insurance contributions
amounted to 52 per cent of total central government revenue. In
Australia in 1986, income taxes amounted to 51 per cent of all federal
government revenue. During the 1980s the British government pursued
a policy of shifting the tax burden from one of ‘tax as you earn’ to ‘tax as
you spend’, and this has obvious implications for redistribution now.
1
In Britain in 1985 this was local authority rates, which were replaced by the
infamous ‘poll tax’ in 1989-90 and succeeded by the council tax in 1993-94. In
Australia, all three levels of government levy taxes or rates that are not included in
this analysis.
Poverty Alleviation versus Social Insurance Systems 7
Benefits in kind from other government funded services also affect
lifetime income redistribution. Le Grand (1982) identified the education
system as being particularly regressive over the course of life, with the
middle classes benefiting disproportionately from expenditure on
education (particularly higher education), although this conclusion is
disputed by Harding using Australian data (1993a, p. 72). Both models
do allow for the allocation of expenditure on education and, in the case
of LIFEMOD, on the national health service; however, these expendi-
tures are not included in this particular analysis (for a discussion of
education transfers, see Harding 1995 and Barr and Falkingham 1993).
The counterfactual
In assessing the impact of the social security system on the distribution
of income, the distribution after intervention necessarily has to be
compared with the distribution before intervention. This immediately
raises the question of what is the most appropriate ‘before’ benchmark
or counterfactual. This study follows convention and measures the
redistributional effect of the taxation and transfer systems against the
original distribution of pre-tax and pre-transfer income. While it is
clearly invalid to assume that the pattern of factor incomes would
remain the same if there were no government, such an assumption is
implicit in this study, as there are no data available indicating how the
lifetime distribution of factor incomes in Australia or Britain would
change if their governments were to disappear.
Assumed incidence of cash transfers and income taxes
The benefit of cash transfers is assumed to be fully incident on those to
whom the transfers are paid. This assumption is controversial. It can be
argued that the benefit of family allowances is incident equally on a
husband and wife, or indeed on the children, rather than solely on the
mother who is the formal recipient. Yet alternative assumptions are also
not straightforward. Both governments have explicitly made the mother
the recipient of child benefits (family allowances) because of doubts as to
who received the benefits when it was paid to the husband. Similarly it
is not clear that the benefits are fully incident on children (Barro 1974).
Likewise, the burden of taxation is assumed to be fully incident on those
legally liable to pay the taxes. Those with such liabilities are also
8 Discussion Paper No. 12
assumed to pay them in full and no account is taken of possible tax
evasion or of the ‘underground’ economy. However, full compliance is
not mirrored in the case of benefit receipt, and both models incorporate
assumptions of take-up rates. In LIFEMOD take-up of means-tested
benefits is related to the size of entitlement (the higher the entitlement
the greater the likelihood of an individual claiming it). In HARDING,
take-up of the family income supplement is affected by self-employment
status, the number of children and the size of the entitlement.
Real economic growth
A fourth issue is whether and how to allow for real economic growth.
The ‘unit of account’ in both models is based on current earnings levels.
Allowance could be made for real earnings growth as well as career
progression in earnings as people become older, but then there would
need to be a discount for the lower value of later receipts. Implicitly the
approach adopted assumes that the effects of overall real economic
growth and a real discount rate cancel each other out. In LIFEMOD, this
means that, once in payment, benefits that are indexed to prices rather
than earnings (such as public sector pensions or State Earnings Related
Pension Scheme rights in retirement) are assumed to slip back each year
at the rate of real earnings growth.
3. The systems
3.1 The British social insurance system
In LIFEMOD the individuals in the model live their entire lives in the
demographic and economic environment of Britain as it was in 1985.
Therefore, the social security benefits they receive and their direct tax
liabilities are calculated under the rules of the taxation and benefit
systems as they were in April 1985. The simulation of the systems is
described in detail in Hills and Lessof (1993).
Entitlement to the majority of benefits within the British social security
system depends on an individual’s national insurance contribution
record. The contribution in any year depends on earnings levels and the
number of weeks of employment or self-employment. In 1985, employee
Poverty Alleviation versus Social Insurance Systems 9
contributions had a ‘slab structure’, with reduced rates payable in
certain earnings bands. Earnings above the upper earnings limit were
exempt. The social security module of LIFEMOD follows this banded
structure in modelling contribution liabilities.
In addition to the national insurance contribution, contributions to the
State Earnings Related Pension Scheme (SERPS) and, where applicable,
contributions to occupational pensions were also modelled. In 1985 it
was possible for a (legally) married couple to choose between
assessment of their joint income or assessment of individual incomes. In
LIFEMOD the tax liabilities under both options are calculated. It is
assumed that couples choose the option giving the lowest total liability:
tax calculated jointly or tax on the part of a wife’s income that is deemed
to be part of her husband’s income for tax purposes is then apportioned
between the husband and wife.
Old age pensions constitute by far the largest cash transfer program on
an annual basis and are the most significant cash benefit when viewed
over a lifetime. Income support in old age in Britain takes the form of a
two-tier pension. The first tier is provided by a universal flat rate basic
state pension for which an individual must have sufficient qualifying
years — that is, years in which a given amount of national insurance
contribution is paid. People with home care responsibilities may be
credited with contribution years toward the basic state pension even if in
those years they paid insufficient national insurance contributions either
because of low or no earnings. This is known as home responsibility
protection. Receipt of this protection does not affect income in the year it
is received but may affect future income streams.
The second tier is furnished by earnings-related pensions, which can be
provided under SERPS or by an employer under an occupational
pension scheme. In 1985 the number of private personal pensions were
low, both in payment and in accumulation, and so these were not
included in the modelling process. The majority of occupational pension
schemes in Britain are defined benefit schemes with the pension linked
to the level of final salary, whereas personal pensions pay a defined
contribution. The latter are much more difficult to calculate in a steady
state world. In LIFEMOD, pension rights are simulated along with
SERPS and other pension entitlements as the cohort individuals age. The
pension rights accumulated may be transferred to an individual’s
10 Discussion Paper No. 12
spouse in the event of them dying before pensionable age. Widows can
also inherit occupational pensions.
As well as old age support there is a range of contingent benefits in
Britain that are not subject to a means test — child benefit (which
depends simply on the number of children and which is assumed in
LIFEMOD to be incident on the mother except in the case of widowers);
one parent benefit paid to lone parents; unemployment benefit paid for a
maximum of 52 weeks in any two years to those who are unemployed
and who have a good enough recent employment history; and invalidity
benefit, severe disablement allowance, mobility allowance, attendance
allowance and invalid care allowance, all of which depend on the degree
of disability and age of original receipt. Income from all of these benefits,
and tax liabilities, are modelled prior to calculating any eligibility for
means-tested support.
Means-tested benefits include the supplementary benefit (now income
support) or family income supplement (now family credit). The latter
applies only to those who have children and have worked full-time for a
period in the year. The amount received depends on the number and the
ages of the children and on weekly income in relation to particular
thresholds. Income support is available for individuals or couples with
periods without full-time work during a year. The entitlement depends
on family circumstances, age, previous receipt of income support,
income and capital. Family credit is assumed to be received by women
except in the case of widowed fathers. Where a couple is entitled to
family credit, the amount received is taken to be part of the man’s
income. Additions to family credit for housing costs, along with other
housing-related cash transfers such as housing benefit and rent and rate
rebates, were not modelled in this version of LIFEMOD. Similarly
subsidies available to owner occupiers through mortgage interest tax
relief were also not included.
Means-tested benefits are assessed on a weekly rather than an annual
basis. Where income varies over the year (for instance, because of a
period of unemployment) benefit entitlement would generally be
underestimated if receipt were calculated on the basis of average income
over the whole year. Thus, annual income in LIFEMOD is divided
between two periods — the period in which someone is employed full-
time and the rest. Assessment for means-tested benefits is on a joint basis
for couples (including those cohabiting but not married).
Poverty Alleviation versus Social Insurance Systems 11
Student grants were simulated, but were treated as part of original
income rather than as a social security type of cash benefit. This is
different from the approach adopted in the Australian model.
3.2 The Australian social assistance model
The Australian social security system principally consists of a range of
income-tested payments that are available to those with particular
characteristics (the simulation of the system is described in detail in
Harding 1993a). The social security cash transfers simulated in the
model, with 1986 payment rates and income tests, are the age pension,
invalid pension, wife’s pension, carer’s pension, widow’s pension,
supporting parent’s benefit, unemployment benefit, sickness benefit and
special benefit. All of these transfers were income tested on the income
of the individual (or husband and wife for married or de facto couples),
and bore no relation to previous periods in the labour force or previous
earnings. In addition to the basic rates, all of those receiving the above
payments could receive a number of additional allowances, of which
additional pension and benefit (paid for dependent children) and the
mother’s or guardian’s allowance (paid to those receiving the widow’s
pension or the supporting parent’s benefit) were included in the model.
Rent assistance, which was paid to pensioners and beneficiaries who
were private renters, was not included in the model because of the lack
of housing data from the 1986 income distribution survey.
Family income supplement was payable to low income families with
dependent children who were not receiving any other form of income
support from the federal government — that is, effectively those who
were in the labour force. Only two payments in 1986 were not income
tested and included in the model. One was the family allowance, which
was payable monthly to people with dependent children aged less than
16 years, full-time dependent students aged 16 and 17 years not
receiving education transfers, or similar students aged 18–24 years in
very low income families. The multiple birth payment to the parents of
triplets or quads aged under 6 years was also modelled.
All benefits and supplements for couples were paid by the Department
of Social Security to the husband, while pensions were split equally
between partners but any additional payments for the children of
pensioners were paid to the wife. The family allowance, multiple birth
12 Discussion Paper No. 12
payment and family income supplement were all expressly paid to the
mother in married couples. All of these provisions were fully
incorporated within the simulation.
The simulation of social security transfers was very comprehensive. A
small number of payments were not modelled but, overall, 97 per cent of
the outlays on pensions, 99 per cent of the outlays on benefits and 98 per
cent of the outlays on child transfers were captured in the simulation.
Apart from waiting periods and a few special provisions, eligibility for
social security cash transfers was essentially based on weekly income.
Consequently, eligibility for most payments was based on the income
and circumstances of each family during each week of each year in the
model.
In 1986 all pensions and the supporting parent’s benefit were both
income and asset tested. It was not possible to model the assets test
adequately because of the lack of data about assets in Australia. A
number of techniques were therefore used to produce an appropriate
degree of take-up for the age pension, sickness and special benefit and
the family income supplement. (The age pension was the principal
source of income for most retirees in Australia, with 79 per cent of the
population of age pension age receiving the age pension in 1986. Others
received occupational pensions from their employers, which were
generally linked to final average salary and length of service.)
The Department of Education also provided two major transfers in 1986,
income tested on the income of both parents and students — the
Secondary Allowances Scheme (SAS) and the Tertiary Education
Assistance Scheme (TEAS). Both of these were simulated in the model,
as was the Postgraduate Awards Scheme. These payments were counted
as part of the cash transfer system when assessing the redistributional
impact of government (in contrast to the British methodology, which
counted them as original income).
Moving to the income tax system, the tax unit in Australia is the
individual, although those with particular family circumstances can
claim a number of special deductions or rebates. The 1985-86 Australian
income tax scales were modelled, as were the dependent spouse rebate,
the sole parent rebate, and the pensioner and beneficiary rebates for
social security recipients. The Medicare levy, which was a special tax
supplement set at 1 per cent of taxable income, was also modelled. The
Poverty Alleviation versus Social Insurance Systems 13
Australian and British income tax scales are compared in table 1, along
with some cash transfer payment rates.
Table 1: Australian and British direct tax scales and selected cash
transfers
A$1 = £0.485
T
axable income Marginal tax rate (for single taxpayer)
B
ritain 1984-85
$0–$3645
9%
a
$3645–$4546
9% (above lower earnings limit for national insurance contribution)
$4546$26 804
39% (above single person allowance, until upper earnings limit)
$26 804$39 587
40%
$39 587–$50 309
45%
$50 309$66 598
50%
$66 598$82 887
55%
$82 887 +
60%
A
ustralia 1985-86
$0–$4595
0%
$4595–$12 500
25%
$12 500–$19 500
30%
$19 500–$28 000
46%
$28 000–$35 000
48%
$35 000 +
60%
B
ase pension in Australia at June 1986
Single person
$102.10 per week
Married couple
$170.30 per week
S
upplementary benefit in Britain at June 1985
Single person
$57.84 per week
Married couple
$93.90 per week
a
If earnings are less than A$3645 in a year, no national insurance is paid. Once earnings exceed
A$3645, 9 per cent national insurance contribution is levied on the entire amount.
4. Lifetime redistribution: social insurance versus
poverty alleviation systems
This chapter compares the overall lifetime redistributional impact of the
British social security system as it was in 1985 with the impact of the
Australian system in 1986. Individuals are ranked in decile groups (ten
equal sized groups) by a measure of lifetime living standards. Rather
than using total lifetime income, the ranking uses each individual’s
annualised lifetime equivalent net family income.
Four distinct decisions are
being made here about the ranking measure.
14 Discussion Paper No. 12
First, lifetime income is
annualised
by averaging total lifetime net income
over each year of life from the age of 16 years. This has the advantage of
avoiding locating people in the upper part of the income distribution
just because they have a long life and have the longest time to
accumulate total income. Conversely, if annualised income were not
used, those who died in their 20s and 30s would appear to have a low
lifetime income and thus a low standard of living, whereas their incomes
while they were alive might have been very high. Cohort individuals
who die before the age of 17 years are excluded from the analysis, as the
majority of this group would not have entered the labour force and thus
would have zero annualised income. Annualised income measures can
be viewed as the average amount of income received during each year of
adult life.
Second,
net
or disposable ‘after-tax’ income is used in preference to gross
income as it provides a more accurate measure of the living standards
attained by individuals and families.
Third, income is
equivalised
to allow for different family circumstances of
cohort individuals. If equivalent income were not used, a person with an
annualised income of half a million dollars and no dependants, for
example, would be regarded as having experienced the same lifetime
standard of living as another person with the same annualised income
who supported a spouse and three children.
Fourth, although it is the individual who is the unit of analysis, when a
cohort member is living with other adults the equivalent income of the
family is attributed to them. This assumes that each person in a family
experiences the same standard of living — that is, the income is shared
equally within the family. Research by Edwards (1981) and Pahl (1990)
suggests that this is not always the case. In other work the sensitivity of
varying this assumption of equal sharing has been explored
(Falkingham and Hills 1995; Harding 1993a), but here analysis is limited
to this central case.
There are numerous difficulties with equivalence scales and it is not
clear which scale is the most appropriate — especially for international
comparative analysis (see Coulter, Cowell and Jenkins 1992 for a full
discussion of the problems). The equivalence scale used in both the
Australian and British results is that implicit in the British social security
system in 1985 — the so-called ‘McClements scale’ — but it has been
Poverty Alleviation versus Social Insurance Systems 15
rescaled so that a single adult is given the value of 1.00 (that is, to give
family income per equivalent adult). (Additional adults are given the
value of 0.64, children aged under 2 years 0.15, aged 5–7 years 0.34, aged
8–10 years 0.38, aged 11–12 years 0.41, aged 13–15 years 0.44 and aged
16–17 years 0.59.)
Table 2 presents the composition of lifetime income for the two model
cohorts in decile groups ranked by the measure of lifetime living
standards (that is, equivalent net family income averaged over each year
of life from the age of 16 years). One thing that is immediately clear from
the table is that lifetime incomes, however measured, appear to have
been higher in Australia than in Britain. This may be due to different
lengths of life or other demographic or social factors, or it may be due to
differences in the modelling. For example, the British world of 1985 has
been modelled while Australian incomes a year later in 1986 should be
expected to have been somewhat higher as the Australian consumer
Table 2: Composition of lifetime income by deciles ranked by
annualised lifetime equivalent net family income
A$1 = £0.485
Decile by annualised lifetime equivalent net family income
12345678910All
A$’000 A$’000 A$’000 A$’000 A$’000 A$’000 A$’000 A$’000 A$’000 A$’000 A$’000
Australia
Earnings
279
400
461
533
609
696
784
852
1 007
1 478
710
Original income
297
430
498
591
672
785
903
1 028
1 257
1 946
841
Cash transfers
89
107
94
87
75
64
55
40
29
13
65
Gross income
386
537
592
678
747
849
957
1 068
1 286
1 959
906
Income tax
45
77
97
128
154
193
240
290
395
745
237
Net income
341
460
495
551
593
656
717
778
891
1 214
670
Equiv. net income
359
477
518
577
619
676
724
800
887
1 166
680
Annualised equiv.
net income 5 864 7 540 8 468 9 395 10 265 11 184 12 220 13 530 15 450 20 157
11
Britain
Earnings
222
330
402
437
530
614
680
682
792
1 070
558
Original income
241
357
431
474
538
579
674
752
876
1 221
612
Cash transfers
134
132
134
140
136
146
136
146
123
120
136
Gross income
375
487
563
614
674
726
810
899
1 000
1 342
745
(National insurance
contribution) 3
8
57 74 87 103 115 140 167 200 315 39
(Income tax)
16
25
29
31
37
39
43
47
54
62
130
Total direct taxes
52
82
105
118
140
154
183
214
254
377
169
Net income
323
404
458
497
534
571
627
682
746
965
581
Equiv. net income
328
421
476
525
555
604
643
715
769
964
599
Annualised equiv.
net income 5 670 7 320 8 124 8 783 9 402 10 041 10 845 11 856 13 260 16 620
10
16 Discussion Paper No. 12
price index, for example, increased by 8.4 per cent between June 1985
and June 1986. The difference may also be due to the exchange rate used
to convert the pound sterling to dollars (the average rate during the
Australian financial year 1985-86 — the best that could be achieved in a
comparison of Australia in 1986 with Britain in 1985). The absolute
amounts are very sensitive to the exact rate and so direct comparisons of
absolute amounts between the two countries should be treated with
some caution. Of more interest is the
pattern of income composition within
deciles and the distribution of income.
In contrast to traditional cross-sectional or annual distributional studies,
the distribution of lifetime income appears to be more equitable. Most
notably, even the lowest income deciles have significant amounts of
income from earnings, constituting over half of total lifetime gross
income. Higher lifetime
original incomes
are the product of higher
earnings and greater investment income and increased access to
occupational superannuation (pensions). Although not shown in the
table, the distribution of private pension income is highly skewed
towards those in the top three deciles of lifetime income.
Table 3 shows the share of various income and tax measures received by
each decile. It simply re-expresses the dollar values shown in table 2 as
decile shares, to aid interpretation. Table 3 suggests that, despite the
higher dollar values of earnings and original factor income in Australia,
the share of earnings received by each decile was very similar in the two
countries. Nonetheless, at the top end of the earnings distribution there
was some divergence, with the top 10 per cent of the population in
Australia receiving 20.8 per cent of total gross lifetime earnings but the
top decile in Britain receiving only 19.1 per cent.
Other components of original income (that is, private occupational
pensions and investment income) appear to have been more unequally
distributed in Australia than in Britain. The top 10 per cent of
Australians received 23.2 per cent of total gross original income accord-
ing to the HARDING model, while the top 10 per cent of the British
population received 19.9 per cent of British gross original income
according to LIFEMOD (table 3). Some of these discrepancies may be
due to different methods of modelling earnings, investment income and
superannuation in the two models. However, it should be noted that this
finding is not out of step with comparisons of the inequality of the
annual
(rather than lifetime) distribution of income in the two countries.
Poverty Alleviation versus Social Insurance Systems 17
For example, Mitchell (1991, p. 123) found that the distribution of
original income in Australia was more unequal than the distribution of
original income in the United Kingdom.
Cash transfers were much more important in Britain than in Australia
for every decile. As table 2 shows, the 10 per cent of the Australian
population with the lowest lifetime incomes received on average
A$89 000 in cash transfers during their lives. This compares with the
A$134 000 received by the poorest 10 per cent of the British population.
In Australia, cash transfers generally declined in absolute value as
income increased. In contrast, in Britain the absolute value of lifetime
total cash benefits was relatively flat across the whole income
distribution (with a slight rise in the middle, followed by a slight dip).
Thus, while the top decile of Australians received on average only
A$13 000 of cash transfers during their lifetimes, the top decile of the
British population averaged A$120 000 of benefits. And the bottom
Table 3: Shares of lifetime income by deciles ranked by annualised
lifetime equivalent net family income
Decile by annualised lifetime equivalent net family income
12345678910All
%%%%%%%%%%%
Australia
Earnings
3.9
5.6
6.5
7.5
8.6
9.8
11.1
12.0
14.2
20.8
100
Original income
3.5
5.1
5.9
7.0
8.0
9.3
10.7
12.2
14.9
23.2
100
Cash transfers
13.6
16.4
14.4
13.4
11.4
9.8
8.4
6.1
4.4
2.0
100
Gross income
4.3
5.9
6.5
7.5
8.2
9.4
10.6
11.8
14.2
21.7
100
Income tax
1.9
3.2
4.1
5.4
6.5
8.2
10.2
12.2
16.7
31.5
100
Net income
5.0
6.8
7.4
8.2
8.8
9.7
10.7
11.6
13.2
18.2
100
Equiv. net income
5.2
7.0
7.6
8.5
9.1
9.9
10.7
11.8
13.0
17.1
100
Annualised equiv.
net income 5.1 6.6 7.4 8.2 9.0 9.8 10.7 11.8 13.5 17.7
100
Britain
Earnings
4.0
5.9
7.2
7.8
8.9
9.5
11.0
12.2
14.2
19.1
100
Original income
3.9
5.7
7.0
7.7
8.8
9.4
10.9
12.2
14.2
19.9
100
Cash transfers
10.0
9.7
9.9
10.4
10.0
10.8
10.1
10.8
9.2
8.9
100
Gross income
5.0
6.5
7.5
8.2
9.0
9.7
10.8
12.0
13.3
17.9
100
National insurance
contribution 4.1 6.2 7.8 8.3 9.4 10.0 11.3 12.6 14.0 16.1
100
Income tax
2.3
4.5
5.8
6.7
8.0
8.9
10.8
12.9
15.4
24.2
100
Total direct taxes
3.0
4.9
6.2
7.0
8.3
9.2
10.9
12.9
15.1
22.4
100
Net income
5.6
7.0
7.9
8.5
9.2
9.8
10.8
11.8
12.8
16.6
100
Equiv. net income
5.5
7.0
8.0
8.7
9.2
10.1
10.7
11.9
12.8
16.1
100
Annualised equiv.
net income 5.7 7.1 7.9 8.4 9.2 9.7 10.7 11.5 13.0 16.8
100
18 Discussion Paper No. 12
decile in Australia received almost 600 per cent more in cash transfers
than the top decile — in sharp contrast to the bottom decile in Britain,
which received only 12 per cent more than the top decile.
Nonetheless, cash transfers in Britain were still progressive, amounting
to a higher proportion of the income of the lifetime poor than of the
lifetime rich. Figure 1 shows the average amounts of cash transfers
received by deciles, ranked by annualised lifetime equivalent net family
income, expressed as a percentage of their lifetime average gross income.
In Britain the lifetime cash transfers received by the bottom decile
amounted to just over 35 per cent of the group’s lifetime gross income,
while the comparable figure in Australia was about 23 per cent. For the
top income decile, cash transfers in Britain amounted to 9 per cent of
lifetime gross income while, in Australia, the comparable figure was less
than 1 per cent.
What about direct taxes? For Australia, federal income tax was included
in the simulation and for Britain the comparable taxes were income taxes
plus national insurance contributions. Direct taxes amounted to a higher
proportion of lifetime gross income in Australia than in Britain — 26 per
cent on average in Australia compared with 23 per cent in Britain (table
2). For the poorest 10 per cent of the Australian cohort, total lifetime
direct taxes amounted to A$45 000 on average, compared with A$52 000
Figure 1: Total lifetime cash transfers received as percentage of total
lifetime gross income by deciles ranked by annualised
lifetime equivalent net family income
0
5
10
15
20
25
30
35
40
12345678910
Deciles by annualised lifetime equivalent net family income
%
Australia
Britain
Poverty Alleviation versus Social Insurance Systems 19
for the British cohort. For the top decile in Australia, direct taxes
averaged just under A$750 000 while for the top decile in Britain total
lifetime direct taxes totalled only A$377 000.
The poorest 10 per cent of the Australian cohort contributed 1.9 per cent
of total lifetime direct taxes, compared with 3 per cent for the bottom
British decile (table 3). The top decile shouldered a larger proportion of
the lifetime direct tax burden in Australia than in Britain — 31.5 per cent
compared with 22.4 per cent. (This is in part a product of the interaction
of higher incomes of the top decile in Australia, but is also due to a much
more progressive income tax scale.)
Figure 2 shows direct tax burdens as a percentage of gross lifetime
income for deciles ranked by annualised lifetime equivalent net family
income. For the bottom decile, lifetime direct taxes amounted to 12 per
cent of gross lifetime income in Australia and 14 per cent in Britain. The
two tax incidence lines crossed at the fifth decile, with direct taxes
amounting to 38 per cent of the gross income of the top decile in
Australia and only 28 per cent for that decile in Britain. On the face of it,
figures 1 and 2 suggest that both the cash transfer and direct tax systems
were more progressive over the lifetime in Australia than in Britain.
Does this necessarily mean that the net effect of the tax and transfer
systems was more progressive in Australia?
Figure 2: Total lifetime direct taxes paid as percentage of total
lifetime gross income by deciles ranked by annualised
lifetime equivalent net family income
0
5
10
15
20
25
30
35
40
12345678910
Deciles by annualised lifetime equivalent net family income
%
Australia
Britain
20 Discussion Paper No. 12
The direct comparison of the redistributional impact of the taxation and
benefit systems is clouded by the fact that the two countries start from
different underlying distributions. What is of central concern to the
hypothesis, however, is not that lifetime incomes appear to be more
unequally distributed in Australia than in Britain; rather it is the impact
that the taxation and benefit systems have on that distribution. The effect
of the social security system is shown by the difference between the
distribution of original income and the distribution of gross income,
while the impact of the direct taxation system is reflected in the
difference between the distribution of gross income and the distribution
of net income (disposable income). The redistributional effect of the
other main instrument of redistribution, the family, can be illustrated by
comparing net income with equivalent net income.
Figure 3 shows the difference made by the taxation and transfer systems
to the shares of lifetime income accruing to each decile within the two
countries. The figure shows that the cash transfer system increased the
bottom decile’s share of lifetime income more in Britain than in
Australia. That is, while the bottom British income decile received 3.9
per cent of lifetime original income, this share was increased by cash
transfers to 5.0 per cent of gross income — the 1.1 per cent increase
illustrated in figure 3. In contrast, the share of income accruing to the
bottom income decile in Australia was increased by only 0.6 per cent as
a result of the cash transfer system. This suggests that, although the cash
transfer system appeared less progressive in Britain, it nonetheless had a
larger redistributional impact over the lifetimes of those with the lowest
lifetime incomes than the Australian system did — no doubt due to the
higher absolute level of transfers. However, as figure 3 shows, this effect
was apparent for only the bottom decile; for deciles two to five, the
impact of the cash transfer system on income shares was very similar in
both countries.
Interestingly, despite the far larger cash transfers paid to upper income
families in Britain than in Australia, for the top 20 per cent of the cohort
the British cash transfer system nonetheless produced a larger decline in
income share than the Australian transfer system did. The top decile’s
share of income in Britain, for example, declined by 2 per cent (from 19.9
per cent to 17.9 per cent) as the income measure was shifted from
original to gross income. In contrast, the top decile’s share of income in
Australia was reduced by only 1.5 per cent by cash transfers.
Poverty Alleviation versus Social Insurance Systems 21
Moving to the taxation system, the Australian taxation system appears
to have changed net income shares to a greater extent than the British
system does, except for the lowest income decile. Consequently, the net
effect of the Australian taxation and transfer systems was to increase the
income share accruing to deciles two to five by more than the taxation
and transfer systems in Britain did. Similarly, the more progressive
Australian direct tax structure resulted in the income share of the top 30
per cent of the cohort being reduced by more as a result of the taxation
and transfer systems than was the case with the British systems. It is
notable that in both countries the crossover point between net gain and
net loss occurs at the seventh decile.
A useful summary measure of inequality is provided by Gini
coefficients. Table 4 shows these for various lifetime income measures
for the HARDING and LIFEMOD cohorts. The Gini coefficient has a
value of one when one person receives all of the income in a society and
Figure 3: Changes in lifetime income shares accruing to each decile,
ranked by annualised lifetime equivalent net family
income, as a result of the transfer and direct taxation
systems
-5
-4
-3
-2
-1
0
1
2
12345678910
Australia G-O
Britain G-O
Australia N-O
Britain N-O
Deciles by annualised lifetime equivalent net family income
%
Note
: Australia G-O measures the impact of the cash transfer system, and is the percentage share of
gross income accruing to a particular decile minus the percentage share of original income accruing to
the same decile. A positive result thus means that the cash transfer system increases the income share
accruing to a decile. Australia N-O measures the combined impact of the taxation and transfer
systems, and shows the percentage share of net income accruing to a particular decile minus the
percentage share of original income accruing to that decile.
22 Discussion Paper No. 12
a value of zero if income is equally distributed among all persons. The
higher the Gini coefficient, the more unequal the distribution of income
being measured.
The cash transfer system in Australia reduced the Gini coefficient (from
original income to gross income) by 0.038. In Britain, although the
absolute amounts of cash benefits in table 2 were much more evenly
spread across the deciles, the Gini coefficient was reduced by 0.055.
Thus, the cash transfer system had a greater equalising effect in Britain
than in Australia.
The direct taxation system further reduced the Gini coefficients. The
additional reduction in Britain was relatively small — only 0.027, even
though the absolute amounts of direct taxes paid exceeded cash transfers
received. In Australia, however, the reduction was larger — 0.059.
Overall, the net effect of the tax and transfer systems in Britain, reflected
in the difference between the disposable and original income
distributions, was a reduction in the Gini coefficient of 0.082. In
Australia the net effect was greater — 0.097. Nonetheless, the lifetime
distribution of disposable income — as measured by the Gini coefficient
— remained more equal in Britain than in Australia.
All of the preceding analysis simply shows the income personally
received by cohort individuals within the two models. In many cases
this would not provide an accurate indicator of the standard of living
achieved by individuals. This is because those with a low original
income of their own might have lived in a family where they shared, for
example, the earnings of their spouse. To examine the redistributional
impact of the family, the difference between the distribution of the
disposable incomes of individuals is compared with the distribution of
Table 4: Gini coefficients for lifetime income measures
Income measure Australia Britain
Original income 0.370 0.327
Gross income 0.332 0.272
Net (disposable) income 0.273 0.245
Equivalent net family income 0.220 0.204
Differences
Gross income minus original income -0.038 -0.055
Net income minus gross income -0.059 -0.027
Net income minus original income -0.097 -0.082
Equivalent net income minus net income -0.053 -0.041
Poverty Alleviation versus Social Insurance Systems 23
the disposable incomes of the families in which those individuals lived.
Table 4 suggests that the role played by families in redistributing
lifetime income was as great as that of the taxation system in Australia.
Thus, while the Australian taxation system reduced the Gini coefficient
by 0.059 the impact of the family on disposable income reduced the Gini
coefficient by 0.053. In Britain the family played less of a role but was
still very influential, reducing the Gini coefficient by 0.041. One reason
for this is that the LIFEMOD cohort experienced a higher incidence of
marital dissolution than did the HARDING cohort and thus spent a
greater proportion of their lifetimes in single adult households where
there was no opportunity to benefit from income sharing.
5. Intrapersonal versus interpersonal redistribution
From the models it is possible to identify receipts of cash benefits that
were ‘paid for’ at another stage in an individual’s life (intrapersonal
redistribution) and those that represent net transfers from others
(interpersonal redistribution). To separate the net transfers within the
model populations, it was necessary to make an assumption about the
taxes allocated to pay for them. This can be done in two ways:
assuming only
direct taxes were used to finance benefits
— thus, it is
assumed that in Britain cash benefits were financed by all of the
national insurance contributions and 58 per cent of the income tax
paid by each group and that in Australia 27.6 per cent of all income
taxes paid by the cohort financed all cash transfers received by the
cohort; and
assuming that a proportion of
both direct and indirect taxes was used.
Indirect taxes are not modelled in the original versions of HARDING
and LIFEMOD (although see Harding 1993b) but in recent years the
combined effect of direct and indirect taxes in Britain has come close
to being proportional to gross income
2
. The combination of direct
and indirect taxes is therefore proxied by assuming financing from
2
For instance, for the United Kingdom the Central Statistical Office (1992, appendix
4, table 2) shows the Gini coefficients of equivalised gross income and equivalised
post-tax income (that is, after both direct and indirect taxes) as equal — 0.32 in
1985.
24 Discussion Paper No. 12
the percentage of each individual’s gross income required to pay for
benefits in aggregate. In HARDING 7.2 per cent of the lifetime gross
income received by all cohort members would finance all cash
transfers received by the cohort. In Britain the corresponding figure
was 16.3 per cent.
Two possible financing scenarios are thus tested. By adopting such an
approach, the counterfactual assumptions used are that, in the absence
of cash transfers from the welfare state, individuals would have to pay
correspondingly less tax :
in proportion to their direct tax liabilities; or
in proportion to their gross income.
Thus it is hoped that systemic differences in taxation structures between
the two countries are controlled for.
5.1 Financing from direct taxes
The first line of data in table 5 shows for deciles ranked by the measure
of living standards — annualised lifetime net family income — total
Table 5: Lifetime net taxes and benefits funded by ‘allocated’ direct
taxes by deciles ranked by annualised lifetime equivalent net
family income
A$1 = £0.485
Decile by annualised lifetime equivalent net family income
12345678910All
A$’000 A$’000 A$’000 A$’000 A$’000 A$’000 A$’000 A$’000 A$’000 A$’000 A$’000
Australia
Cash benefits
89
107
94
87
75
64
55
40
29
13
65
Allocated direct tax
12
21
27
29
43
54
66
80
109
206
65
Lifetime net benefit
78
88
71
58
42
27
15
6
2
0
40
Lifetime net tax
1
2
4
6
10
17
26
46
82
193
40
Self-financed benefits
11
19
23
23
33
37
40
34
27
13
25
Average lifetime
gain
+77
+86
+67
+52
+32
+10
-11
-40
-80
-193
Britain
Cash benefits
134
132
134
140
136
146
136
146
123
120
136
Allocated direct tax
41
68
84
97
113
126
148
173
204
297
136
Lifetime net benefit
97
74
64
64
52
54
39
35
21
12
52
Lifetime net tax
4
10
14
21
29
34
51
62
102
189
52
Self-financed benefits
37
58
70
76
84
92
97
111
102
108
84
Average lifetime
gain
+93
+64
+50
+43
+23
+20
-12
-27
-81
-177
Poverty Alleviation versus Social Insurance Systems 25
lifetime cash benefit receipts. The following line shows what each group
would contribute in direct taxes to finance the cash benefits of the entire
cohort. The calculations in the remainder of the table show net receipts
and payments using these ‘allocated’ tax payments.
During each year of life individuals might pay tax, receive cash benefits
or (for perfectly good reasons) both. In the last case, some or all of the
benefits they received in a year would be paid for by their own direct tax
in the same year — that is, there would be intrapersonal redistribution
within the same year
(sometimes known as ‘churning’). For example, some
people would receive:
taxable benefits (such as the state pension);
non-taxable benefits but pay tax on other income (for instance,
women with earnings receiving child benefit); or
benefits because they are out of work for part of a year, but pay tax
on earnings in the rest of it.
People might also pay for their own benefits in other years of their lives.
Over their whole lives, individuals are either net lifetime taxpayers or
net lifetime benefit recipients (or, conceivably, neither if they break
even).
3
3
Algebraically the various measures can be expressed as follows. Let
Bi
be the gross
cash benefits received by individual i in year t, and Ti be the gross (allocated) tax
paid by them in that year. Then the aggregate gross cash benefits, G, for the model
population over its entire lifetime is given by:
G =
i Bi
=
i Ti
.
Self-financed benefits, SFB, are given by:
SFB
i
= Ti (if Bi > Ti)
= Bi (if Bi < Ti).
An individual’s net lifetime gain,
Ni
, is:
Ni
=
Bi – Ti.
Total
interpersonal
redistribution,
IR
, is the sum of positive net lifetime gains and is
given by:
IR
=
i
(
Ni Ni
> 0)
.
Total
intrapersonal
redistribution,
IA
, is the sum of self-financed benefits:
IA
=
i SFBi
(and by construction,
G
=
IR
+
IA
).
26 Discussion Paper No. 12
Table 5 illustrates the scale of such ‘self-financing’ of benefits (intra-
personal redistribution), together with the net lifetime taxes or benefits
of individuals in excess of self-financed benefits (interpersonal
redistribution). For example, the bottom income decile in Australia
received a total of A$89 000 in cash transfers during their lifetimes.
About 28 per cent of the total income tax that they paid during their
lifetimes financed all cash transfers, so that their allocated direct tax
burden was about 28 per cent of the A$45 000 of lifetime direct tax
shown in table 2 (that is, about A$12 000). In other words, out of all the
cash transfers paid to the entire cohort during their lifetimes, the bottom
decile contributed on average A$12 000 of the direct taxes that financed
those cash transfers.
Of this A$12 000 allocated tax burden, A$11 000 was paid back in the
form of cash transfers. The bottom Australian decile thus financed
A$11 000 of the A$89 000 of cash transfers that were paid to them. Most
individuals in the bottom Australian decile were net lifetime benefici-
aries after taking account of cash transfers received and allocated taxes
paid — the average lifetime net benefit for the entire decile being
A$78 000. However, a few individuals within this decile were net
lifetime taxpayers. In other words, a few individuals in the bottom decile
paid out more in allocated taxes during their lifetimes than they received
in cash transfers. For the decile as a whole, the lifetime allocated tax that
they paid out and that financed the benefits of others averaged about
A$1000. Consequently, the average lifetime gain for this group was
A$77 000. Therefore, the ‘self-financed benefits’ rows in table 5 show the
scale of intrapersonal redistribution and the ‘average lifetime gain’ rows
show the scale of interpersonal redistribution.
The bottom decile in Britain received much more in cash transfers than
the bottom Australian decile did, but they also paid much more
allocated direct tax. The poorest decile in Britain paid for about 28 per
cent of the cash transfers that they received (in contrast to only 12 per
cent for the bottom Australian decile). However, the bottom decile in
Britain still received an average lifetime gain of A$93 000 — much
higher than the average lifetime gain of A$77 000 received by the bottom
Australian decile. For deciles two to five, however, the Australian tax
and transfer systems delivered more interpersonal redistribution. The
average lifetime gain for those in the bottom half of the lifetime income
distribution averaged A$63 000 in Australia but A$55 000 in Britain.
Poverty Alleviation versus Social Insurance Systems 27
The proportion of gross benefits received by each decile that represented
intrapersonal rather than interpersonal redistribution grew rapidly as
incomes increased. In Britain, by the third decile the majority of the
group’s gross receipts were self-financed; the same was true by the sixth
decile in Australia. For the top group, A$108 000 out of the A$120 000
received on average in Britain was self-financed and the net taxpayers
within the group paid out A$189 000 (averaged over the whole group).
The pattern for the entire cohort is shown in the final column of table 5.
On average, HARDING model individuals received A$65 000 in cash
transfers over their lifetimes — A$25 000 or 38 per cent representing
benefits that individuals effectively paid for themselves. In Britain,
individuals averaged A$136 000 in lifetime cash transfers — A$84 000 or
62 per cent financed from their own taxes.
This pattern is summarised in figure 4. The top line for each country
shows average gross benefits of each decile group, and the bottom line
the average gross (allocated) taxes paid. Deducting the totals of annual
and lifetime clawback leaves the two unshaded areas in the top and
bottom of the diagram — the net lifetime benefits of lifetime gainers and
the net lifetime taxes of lifetime losers. It is these unshaded areas that
represent the interpersonal redistribution. Judged by Robin Hood
criteria, this redistribution is progressive (although somewhat
Figure 4: Lifetime taxes and benefits by deciles ranked by annualised
lifetime equivalent net family income
A$1 = £0.485
Australia using HARDING Britain using LIFEMOD
-300
-250
-200
-150
-100
-50
0
50
100
150
12345678910
Gross benefit Net lifetime benefits
Gross allocated taxes Net lifetime taxes
Net tax
Net benefit
-300
-250
-200
-150
-100
-50
0
50
100
150
12345678910
Net tax
Net benefit
-
-
Deciles by annualised lifetime equivalent net family income
A$’000
A$’000
28 Discussion Paper No. 12
approximate, with some net lifetime taxpayers at the bottom and a
rather larger number of net lifetime gainers at the top), and is much
more so in Australia than in Britain.
The shaded areas (of equal size for each group at the top and the bottom
of the diagram) show the intrapersonal redistribution. In the Australian
system, on average 38 per cent (A$25 000) of lifetime benefits were paid
for by the same individual at another stage in their life while 62 per cent
were paid for by others. By coincidence, the proportions are reversed in
Britain, with 62 per cent of benefits constituting intrapersonal
redistribution and 38 per cent interpersonal redistribution. So far, the
null hypothesis seems unequivocally unrejected.
5.2 Financing from all taxes
The above results are based on the assumption that cash benefits are
financed by progressive direct taxes. Table 6 shows the results of making
the alternative assumption that financing comes from a share of all tax
revenues. The assumption is that tax is proportional rather than
progressive, with allocated tax amounting to a constant proportion of
gross income in both countries.
Table 6:
Lifetime net taxes and benefits funded by ‘allocated’ direct
tax by deciles ranked by annualised lifetime equivalent net
family income
A$1 = £0.485
Decile by annualised lifetime equivalent net family income
12345678910All
A$’000 A$’000 A$’000 A$’000 A$’000 A$’000 A$’000 A$’000 A$’000 A$’000 A$’000
Australia
Cash benefits
89
107
94
87
75
64
55
40
29
13
65
Allocated direct tax
28
38
42
49
54
61
70
77
93
141
65
Lifetime net benefit
64
73
58
46
33
21
11
5
2
0
31
Lifetime net tax
3
4
6
8
12
18
26
42
66
128
31
Self-financed benefits
25
34
36
41
42
43
44
35
27
13
34
Average lifetime
gain
+61
+69
+52
+38
+21
+3
-15
-37
-64
-141
Britain
Cash benefits
134
132
134
140
136
146
136
146
123
120
136
Allocated direct tax
77
88
103
110
122
130
145
162
178
242
136
Lifetime net benefit
72
55
47
49
41
44
32
33
19
10
40
Lifetime net tax
5
11
16
19
27
28
41
49
74
132
40
Poverty Alleviation versus Social Insurance Systems 29
Self-financed benefits
72
77
87
91
95
102
104
113
104
110
96
Average lifetime
gain
+67
+44
+31
+30
+14
+16
-9
-16
-55
-122
30 Discussion Paper No. 12
Four differences from earlier results stand out.
The proportion of intrapersonal redistribution rises from 38 per cent
to 52 per cent in Australia and from 62 per cent to 71 per cent in
Britain.
The average net gains to the bottom deciles are reduced and,
symmetrically, the net loss of the top groups are decreased. Thus,
there is less interpersonal redistribution.
The interpersonal redistribution involved, while of a smaller scale
than when direct taxes are assumed to be the source of finance, is
nonetheless progressive.
Even controlling for the finance mechanism, the British social
security system results in more intrapersonal redistribution than the
Australian system does and less interpersonal redistribution.
6. Summary
All social security and income tax systems are intended to generate both
interpersonal redistribution of income (from those with high to those
with low incomes) and intrapersonal redistribution (from one period of
an individual’s life cycle to another). Until now, most analyses of the
redistributional impact of the taxation and transfer systems have been
limited to the interpersonal redistribution generated at a single point in
time. The lack of lifetime data has limited the capacity to answer
questions about the intrapersonal and interpersonal redistributional
impact of the modern welfare state over a longer period. This paper
represents a first attempt to undertake international comparative work
about the lifetime redistributional effects of social systems that
emphasise social insurance goals and those that emphasise poverty
alleviation goals.
The work was based on two dynamic cohort microsimulation models
that put 4000 simulated British individuals and 4000 simulated
Poverty Alleviation versus Social Insurance Systems 31
Australian individuals through their life cycles — from birth to death.
Although some aspects of life have been modelled differently, the
models are in many respects unusually comparable, having being
developed with a high degree of cooperation.
The Australian lifetime income distribution appeared more unequal
than the British distribution, and average gross lifetime incomes in
Australia, at A$906 000, were substantially higher than in Britain
(A$745 000). This seemed to be due to a multitude of factors, including
an earlier base year for Britain (1985 as opposed to 1986 for Australia),
higher unemployment and lower education levels in Britain, and
sociodemographic differences, including higher mortality, divorce and
sole-parenthood rates in Britain. It is important to keep these differences
in the original income distribution in mind when assessing the following
results. For example, because the top 10 per cent of the Australian cohort
(the group with the highest incomes) had original incomes that were 40
per cent higher than the top 10 per cent of the British cohort, it would be
expected that their income tax contributions would have been
substantially higher (because of the progressive taxation system). These
underlying differences complicate the comparison, and only broad
conclusions can be drawn about how the two welfare states appear to be
redistributing lifetime income.
Some broad conclusions did, however, emerge. Total lifetime cash
transfers seemed to be much greater in Britain than in Australia,
amounting to A$136 000 on average in Britain but only A$65 000 in
Australia. The absolute value of lifetime cash transfers received declined
considerably as Australian incomes increased. In contrast, the absolute
value remained relatively stable across the whole income distribution in
Britain.
Cash transfers amounted to a higher proportion of gross income in
Britain than in Australia for every decile. For example, cash transfers
amounted to about 35 per cent of the gross income of the bottom decile
in Britain and 23 per cent in Australia. For the top decile, the comparable
figures were 9 per cent in Britain and less than 1 per cent in Australia.
According to Gini coefficients, the British cash transfer system reduced
the inequality of the original income distribution by more than the
Australian system did.
32 Discussion Paper No. 12
Average lifetime income taxes in Australia amounted to A$237 000,
compared with an average A$169 000 in Britain. Interestingly, in both
countries income taxes (plus national insurance contributions in Britain)
amounted to half of total tax revenue. The Australian direct taxation
system looked more progressive, with average tax rates being lower for
the Australian bottom decile than for the corresponding British decile
but higher for the top decile. However, to some extent this apparent
difference was due to discrepancies in the original income distribution
against which progressivity was assessed. The direct taxation system
reduced the Australian Gini coefficient by 0.059, more than double the
0.027 reduction achieved by the British direct taxation system.
It was possible to look at the extent to which receipts of cash benefits by
an individual were paid for by that person at another stage in their life
(lifetime intrapersonal redistribution) and the extent of net transfers
from other people (interpersonal redistribution). Assuming that all cash
benefits were financed from direct taxes, on average out of the A$65 000
each Australian received in cash transfers over their lifetimes, 38 per
cent were effectively paid for by the individuals themselves, whereas 68
per cent were paid for by others. These proportions were reversed in
Britain.
The proportion of cash transfers that represented intrapersonal
redistribution increased as incomes increased. In Britain, by the third
decile, the majority of the group’s gross receipts were self-financed over
the lifetime; the same was true by the sixth decile in Australia.
Alternatively, if it was assumed that cash transfers were financed by a
proportional rather than a progressive tax, the proportion of transfers
that were self-financed increased from 38 per cent to 52 per cent in
Australia and from 62 per cent to 71 per cent in Britain.
Thus, controlling for the greater redistributional effect of the more
progressive Australian taxation system does not alter the finding that a
primarily social assistance-based system (such as Australia’s), with its
emphasis on poverty alleviation, results in a greater degree of inter-
personal redistribution of income. Conversely, a primarily social
insurance-based system (such as Britain’s), with its emphasis on the link
between contributions and benefits, results in a greater degree of
intrapersonal redistribution.
Poverty Alleviation versus Social Insurance Systems 33
Finally, it must be acknowledged that this study examines the
redistributional impact of the British and Australian systems by
comparing the post-tax and transfer income distribution with the
original income distribution prevailing in
each
country. An alternative
method would be to simulate the rules of the taxation and transfer
systems of both countries against the
same
original income distribution
(that is, run the rules for both taxation and transfer systems against
either the Australian or the British original income distributions). Such
an exercise would provide a very interesting complement to the study
presented here. However, it would not necessarily provide more
definitive answers about the lifetime redistributional impact of each
country’s systems, as this would depend on the extent to which the
prevailing taxation and transfer rules had already affected behaviour —
and thus the ‘original’ income distribution — within each country.
34 Discussion Paper No. 12
References
Barr, N. and Falkingham, J. 1993,
Paying for learning
, Welfare State Programme,
Discussion Paper WSP/94, London School of Economics.
Barro, R. 1974, ‘Are government bonds net wealth?’,
Journal of Political
Economy
, vol. 82, pp. 1095–117.
Beveridge, W. 1942,
Social Insurance and Allied Services
, Cmd. 6404, HMSO,
London.
Central Statistical Office (UK) 1992, ‘The effects of taxes and benefits on
household income 1989’,
Economic Trends
, no. 459, pp. 127–65.
Coulter, F., Cowell, F.A. and Jenkins, S.P. 1992, ‘Differences in needs and
assessment of income distributions’,
Bulletin of Economic Research
, vol. 44,
no. 2, pp. 77–124.
Edwards, M. 1981,
Financial Arrangements within Families
, National Women’s
Advisory Council, Canberra.
Falkingham, J. and Lessof, C. 1991,
LIFEMOD — The Formative Years
, Welfare
State Programme Research Note WSP/RN/24, London School of
Economics.
—— and ——1992, ‘Playing god: the construction of LIFEMOD, a dynamic
cohort microsimulation model’, in Hancock, R. and Sutherland, H. (eds),
Microsimulation Models for Public Policy Analysis
:
New Frontiers
, STICERD
Occasional Paper no. 17, London School of Economics.
—— and Hills, J. (eds) 1995,
The Dynamic of Welfare
, Harvester-Wheatsheaf,
London.
Hain, W. and Helberger, C. 1986, ‘Longitudinal microsimulation of lifetime
income’ in Orcutt, G., Merz, J. and Quinke, H. (eds),
Microanalytic
Simulation Models to Support Social and Financial Policy
, North Holland,
New York.
Harding, A. 1984,
Who Benefits: The Welfare State and Redistribution
, Social
Welfare Research Centre Reports and Proceedings no. 45, University of
New South Wales, Sydney.
—— 1993a,
Lifetime Income Distribution and Redistribution: Applications of a
Dynamic Cohort Microsimulation Model
, North Holland, Amsterdam.
Poverty Alleviation versus Social Insurance Systems 35
—— 1993b, ‘The lifetime incidence of indirect taxes: wholesale sales tax and
excises’, in Head, J. (ed.),
Fightback: An Economic Analysis
, Australian Tax
Research Foundation, Sydney.
—— 1995, ‘Financing higher education: an assessment of income-contingent
loan options and repayment patterns over the life cycle’,
Education
Economics
, vol. 3, no. 2, pp. 173–203.
Hills, J. and Lessof, C. 1993,
Modelling Direct Tax and Social Security over the
Lifetime
, Welfare State Programme Research Note WSP/RN/25, London
School of Economics.
Le Grand, J. 1982,
The Strategy of Equality: Redistribution and the Social Services,
Allen & Unwin, London.
Mitchell, D. 1991,
Income Transfers in Ten Welfare States,
Avebury, Aldershot.
Nelissen, J.H.M. 1993,
The Redistributive Impact of Social Security Schemes on
Lifetime Labour Income
, Report no. 22, Tilburg University.
Orcutt, G., Caldwell, S. and Wertheimer, R. 1976,
Policy Exploration Through
Microanalytic Simulation
, Urban Institute Press, Washington, DC.
Pahl, J. 1990, ‘Household spending, personal spending and the control of
money in marriage’,
Sociology
, vol. 24, part 1, pp. 119–38.
Saunders, P. 1984,
Evidence on Income Redistribution by Governments
, OECD
Economics and Statistics Department Working Paper no. 11, Paris.
Summers, R. 1956,
An Econometric Investigation of the Size Distribution of Lifetime
Average Income
, Department of Economics Technical Report no. 31,
Stanford University.
Winter, D. 1994, Modelling lifetime earnings, London School of Economics,
mimeograph.
Wolfson, M. 1988, ‘Homemaker pensions and lifetime redistribution’,
Review of
Income and Wealth
, vol. 34, no. 3, pp. 221–50.
NATSEM publications
Copies of NATSEM publications and information about NATSEM may
be obtained from:
Ms Kirrilie Nordsvan
Publications Officer
National Centre for Social and Economic Modelling
University of Canberra
GPO Box 563
Canberra City ACT 2601
Australia
Ph: + 61 6 275 4900 Fax: + 61 6 275 4875
E-mail: natsem@natsem.canberra.edu.au
Periodic publications
NATSEM News keeps the general community up to date with the
developments and activities at NATSEM, including product and
publication releases, staffing and major events such as conferences. This
newsletter is produced twice a year.
The Income Distribution Report (IDR), which is also produced twice a
year, provides information and comment on the average incomes of
Australian families, covering the incidence of taxation for different
family types, the income support provided by the government and how
different family groups are faring. The IDR, which is available on
subscription, presents this information in a simple, easy-to-follow
format.
NATSEM’s Annual Report gives the reader an historical perspective of
the centre and its achievements for the year.
STINMOD/95
STINMOD is an easy-to-use microsimulation model for better
understanding the impact of social and economic change on Australian
families. It is a cutting edge tool for exploring the distributional impact
of possible government policy changes. Microsimulation models are
built from the bottom up, using samples of microdata based on real
Australians. From such a foundation, projections can then be made for
the wider community.
The third release of STINMOD is now available. This release
incorporates changes to social security payments announced in the
1995-96 budget, more flexible output specifications, online benchmark
results and context-sensitive help screens, and enables users to save
policy options so that they can be reused at a later date.
STINMOD Technical Paper series
N
o. Authors Title
1
Lambert, S., Percival, R.,
Schofield, D. and Paul, S.
An Introduction to STINMOD: A Static
Microsimulation Model, October 1994
2
Percival, R. Building STINMOD’s Base Population,
November 1994
3
Schofield, D. and Paul, S. Modelling Social Security and Veterans’
Payments, December 1994
4
Lambert, S. Modelling Income Tax and the Medicare Levy,
December 1994
5
Percival, R. Modelling AUSTUDY, December 1994
6
Landt, J. Modelling Housing Costs and Benefits, December
1994
7
Schofield, D. Designing a User Interface for a Microsimulation
Model, March 1995
8
Percival, R. and
Schofield, D.
Modelling Australian Public Health Expenditure,
May 1995
9
Paul, S. Modelling Government Education Outlays,
September 1995
1
0 Schofield, D., Polette, J.
and Hardin, A.
Modelling Child Care Services and Subsidies,
January 1996
DYNAMOD Technical Paper series
N
o. Authors Title
1
Antcliff, S. An Introduction to DYNAMOD: A Dynamic
Microsimulation Model, September 1993
Discussion Paper series
N
o. Authors Title
1
Harding, A. Lifetime Repayment Patterns for HECS and
AUSTUDY Loans, July 1993
2
Mitchell, D. and
Harding, A.
Changes in Poverty among Families during the
1980s: Poverty Gap Versus Poverty Head-count
Approaches, October 1993
3
Landt, J., Harding, A.,
Percival, R. and
Sadkowsky, K.
Reweighting a Base Population for a
Microsimulation Model, January 1994
4
Harding, A. Income Inequality in Australia from 1982 to 1993:
An Assessment of the Impact of Family,
Demographic and Labour Force Change,
November 1994
5
Landt, J., Percival, R.,
Schofield, D. and
Wilson, D.
Income Inequality in Australia: The Impact of
Non-Cash Subsidies for Health and Housing,
March 1995
6
Polette, J. Distribution of Effective Marginal Tax Rates
Across the Australian Labour Force, August 1995
7
Harding, A. The Impact of Health, Education and Housing
Outlays on Income Distribution in Australia in
the 1990s, August 1995
8
Beer, G. Impact of Changes in the Personal Income Tax
and Family Payment Systems on Australian
Families: 1964 to 1994, September 1995
9
Paul, S. and Percival, R. Distribution of Non-Cash Education Subsidies in
Australia in 1994, September 1995
1
0 Schofield, D., Polette, J.
and Hardin, A.
Australia’s Child Care Subsidies: A Distributional
Analysis, January 1996
1
1 Schofield, D. The Impact of Employment and Hours of Work
on Health Status and Health Service Use,
March 1996
... Our way of raising the question is closer still to research comparing welfare systems by the relative importance of lifecycle financing versus redistribution by income. Falkingham and Harding (1996) The statistical literature on importance of regressors in multivariate regressions differentiates between causal or theoretical importance and dispersion importance of a predictor variable (see e.g. Grömping, 2015). ...
... For a more definitive statement on welfare states in general we will extend our research to further countries in the future. Falkingham and Harding (1996) and Ståhlberg (2007) have found that although reallocation among age groups is always crucial, some welfare systems are more redistributive across income groups than others. Yet, even our single-country case leads to some important consequences for the analysis of Hungarian social policy. ...
Preprint
Full-text available
There seems to be a general consent in the expert community that Hungarian social policy provides poorly targeted benefits and services that are prone to Matthew-effects. Our results confirm this observation but we also find that the data offer an alternative interpretation of what the Hungarian welfare state is actually doing. Instead of supporting the poor it reallocates resources from the working age population to children and elderly people. It functions as an intermediary between overlapping generations that seek to finance their lifecycle by exploiting the opportunity offered by the very overlap, the fact that contemporaries are of different age. In a cross-sectional framework we analyse reallocations by age and income simultaneously and assess the relative importance of these two variables in explaining the access and contribution to public benefits. Our data from 2010 (based on EU-SILC and the Household Budget Survey) covers public transfers (cash and in-kind) and both direct and indirect taxes. We compare the importance of age and income in explaining transfers and taxes in a regression-analysis framework by studying causal importance (comparing coefficients) and dispersion importance of the variables (using Shapley-value decomposition). We find that income is irrelevant in explaining access to benefits and services but age is important. On the contribution side, income proves as important as age. This qualifies our description of the Hungarian welfare system: it serves as a channel through which affluent people in their working age support people in inactive age of all income groups.
... This common methodological approach carries significant limitations, however. Direct cross-country comparison of results is difficult (rare examples are [43] and [44], both comparing only two countries). Measuring intertemporal effects requires a time-series of retrospective data, which is only available for a few countries. ...
... National samples are reweighed so that each country is represented with the same number of people. The national values for benefits and taxes are then rescaled following the National Transfer Accounts method (national values of benefits and taxes by the national per capita labor income of the age-group [30][31][32][33][34][35][36][37][38][39][40][41][42][43][44][45][46][47][48][49]. The pooling procedure matches the corresponding national age-status groups: the oldest of one country are paired with the oldest of another and people of the lowest status in one country with people of the lowest status in another. ...
Article
Full-text available
Social scientists identify two core functions of modern welfare states as redistribution across (a) socio-economic status groups (Robin Hood) and (b) ‘the lifecycle’ (the piggy bank). But what is the relative importance of these functions? The answer has been elusive, as the piggy bank is metaphorical. The intra-personal time-travel of resources it implies is based on non- quid-pro-quo transfers. In practice, ‘lifecycle redistribution’ must operate through inter-age-group resource reallocation in cross-section. Since at any time different birth cohorts live together, ‘resource-productive’ working-aged people are taxed to finance consumption of ‘resource-dependent’ younger and older people. In a novel decomposition analysis, we study the joint distribution of socio-economic status, age, and respectively (a) all cash and in-kind transfers (‘benefits’), (b) financing contributions (‘taxes’), and (c) resulting ‘net benefits,’ on a sample of over 400,000 Europeans from 22 EU countries. European welfare states, often maligned as ineffective Robin Hood vehicles riddled with Matthew effects, are better characterized as inter-age redistribution machines performing a more important second task rather well: lifecycle consumption smoothing. Social policies serve multiple goals in Europe, but empirically they are neither primarily nor solely responsible for poverty relief and inequality reduction.
... Typically two different approaches are used to construct lifetime profiles from panels that cover a relatively short period of time. One is to estimate processes using a sample that pools together data on different cohorts (see, for example Falkingham and Harding 1996). Another is to adopt a splicing approach that joins individuals observed at different ages within the same short period together to form a complete lifecycle (Bovenberg et al. 2008). ...
Article
Full-text available
This paper examines the lifetime distributional impact of changes to the tax and transfer system. We find that—in contrast to standard snapshot analyses—increases to work-contingent benefits are just as effective at redistributing resources to the lifetime poor as increases to out-of-work benefits. This has important implications for the equity-efficiency trade-off typically thought to apply to work-contingent transfers. We also show that increases to higher rates of income tax are an effective way of targeting the lifetime rich because higher earners tend to exhibit greater persistence in their incomes. Our results illustrate the importance that moving beyond an exclusively snapshot perspective can have when analysing tax and transfer reforms.
... There has been a concerted effort to target welfare payments more closely to those in need. A recent study suggests that this has produced a system which is more efficient at redistributing income to the poor than the British social insurancebased system (Falkingham and Harding 1996). In conjunction with the tax system, however, social security arrangements can create disincentives to accept employment -especially for low-skilled recipients eligible for family entitlements. ...
... This point is of crucial importance in international comparisons of redistribution, since the mix of self-insurance and actual redistribu tion differs between countries. Falkingham and Harding (1996), for example, com pare the redistributive impact of the UK and Australian social security systems. In die targeted Australian system, 38 per cent of lifetime benefits received by indi viduals are paid for at another stage in their life (that is, they are equivalent to selfinsurance); die remaining 62 per cent consists of redistribudon between income groups. ...
... Potencialmente, los modelos multinivel pueden utilizarse para diferenciar las variaciones interindividuales e intraindividuales de variables como los ingresos; esto puede resultar una distinción útil para distinguir los aspectos individuales y contextuales en las estrategias frente a la pobreza (esto requeriría fuentes de datos más desarrolladas y continuas en el tiempo). Al mismo tiempo, los condicionantes de la pobreza pueden ser distintos a los condicionantes de las variaciones y transiciones de la pobreza (Falkingham, 1999;Falkingham et al., 2000;Falkingham & Harding, 1996). ...
Book
Full-text available
El tiempo en las ciencias sociales: un problema que requiere nuevas ideas. También, visiones interdisciplinarias e integración de problemas e ideas que vienen de otras ciencias. Lo que va y no vuelve, causas antes que efectos, obsolescencia de los indicadores, procesos irreversibles, y otros problemas del análisis del tiempo en ciencias sociales. De esto trata mi tesis, editada como libro digital por Editorial Teseo / Facultad de Ciencias Sociales de la UBA, y dirigida por Juan Piovani. Disponible on line en https://www.teseopress.com/desafiosdelanalisisdeltiempo?fbclid=IwAR3wx93e9mW5V8y0XKziTGODfhZCl53SOljBvmGinZFwDHijIxch3pgx9vU
... Several papers have looked at the reallocative effect of taxes and transfers on a lifetime basis (Falkingham and Harding, 1996, Nelissen, 1998, Björklund and Palme, 2002, Pettersson and Pettersson, 2007, Ter Rele et al., 2007, Bovenberg et al., 2008, Bartels, 2012, Levell et al., 2017. This literature systematically finds that the reallocation of lifetime earnings through the tax-and-transfer system partially offsets disparities in lifetime earnings. ...
Conference Paper
Full-text available
In this article, based on a longitudinal study of living conditions, the transitions in incomepoverty in the period 2000 - 2005 in the Tres de Febrero district (Buenos Aires, Argentina)are addressed, and the importance of the longitudinal analysis of the trajectories of poverty.At the same time, different problems of the analysis of the passage of time in socialphenomena are addressed, and some of the methodological, epistemological, and integrationchallenges of the social sciences with other scientific disciplines that can enrich this task.
Article
Full-text available
En este artículo se analizan el número, la velocidad y el sentido de las transiciones en la pobreza por ingresos en el período 2000 – 2005 en el Partido de Tres de Febrero (Buenos Aires), a partir de un estudio longitudinal de tres ondas (en los años 2000, 2002, y 2005). Este estudio refleja la crisis social del período, y al mismo tiempo, vuelve evidente la importancia de la perspectiva longitudinal en el análisis de la pobreza. El análisis de los datos longitudinales se vincula con distintos problemas del análisis del paso del tiempo en los fenómenos sociales, y a los desafíos metodológicos, epistemológicos, y de integración de las ciencias sociales con otras disciplinas científicas que pueden enriquecer esta tarea.
Article
Patterns of spending within a household reflect not only the sources and amount of the household's income, but also who controls money within the household. This idea is explored using data from a study of 102 married couples, who were interviewed both together and separately. Three areas of spending are examined: food and daily living expenses, leisure expenditure, and domestic rates or poll tax. The results suggest that though in absolute terms men contribute more to the domestic economy than do women, in relative terms women contribute a higher proportion of their income. Compared with men, women hold less back, both absolutely and relatively, for their personal expenditure.
Article
There has been growing worldwide interest in income contingent loans (ICLs) for tertiary students in recent years. This paper lists the design issues which have to be tackled when introducing an ICL and describes the solutions adopted by the Australian and New Zealand governments. Although there is keen interest in the likely repayment patterns for ICLs after graduation, it is very difficult to model such repayments using standard economic cohort techniques. This study uses dynamic micro-simulation to simulate the likely future repayment profiles for the two Australian ICL schemes and assesses the proportion of total debt repaid.
Article
Thesis (Ph. D.)--Dept. of Economics, Stanford University. Bibliography: l. 125-127.
Article
The assumption that government bonds are perceived as net wealth by the private sector is crucial in demonstrating real effects of shifts in the stock of public debt. In particular, the standard effects of “expansionary” fiscal policy on aggregate demand hinge on this assumption. Government bonds will be perceived as net wealth only if their value exceeds the capitalized value of the implied stream of future tax liabilities. This paper considers the effects on bond values and tax capitalization of finite lives, imperfect private capital markets, a government monopoly in the production of bond “liquidity services,” and uncertainty about future tax obligations. It is shown within the context of an overlapping-generations model that finite lives will not be relevant to the capitalization of future tax liabilities so long as current generations are connected to future generations by a chain of operative intergenerational transfers (either in the direction from old to young or in the direction from young to old). Applications of this result to social security and to other types of imposed intergenerational transfer schemes are also noted. In the presence of imperfect private capital markets, government debt issue will increase net wealth if the government is more efficient, at the margin, than the private market in carrying out the loan process. Similarly, if the government has monopoly power in the production of bond “liquidity services,” then public debt issue will raise net wealth. Finally, the existence of uncertainty with respect to individual future tax liabilities implies that public debt issue may increase the overall risk contained in household balance sheets and thereby effectively re-duce household wealth.
Article
The literatures on the assessment of income distributions and on differences in needs are not well integrated. The theoretical literature providing results about evaluations and comparisons of income distributions does not, in the main, consider the implications of non-income differences between income-receiving units, while the equivalence scale literature providing results about the relationship between living standards, incomes, and household composition does not consider distributional implications. The aim of this survey is to go some way towards integrating these separate literatures; to consider how to evaluate and compare income distributions while at the same time taking explicit account of differences in household composition. Copyright 1992 by Blackwell Publishing Ltd and the Board of Trustees of the Bulletin of Economic Research
Article
There has been considerable public debate in Canada on the merits of proposals to extend coverage under the public earnings-related pension system (the Canada and Quebec Pension Plans or C/QPP) to homemakers. This paper presents an analysis of one such proposal put forward by a Parliamentary Committee in 1983. The analysis considers both the likely costs and the redistributional impact of this homemaker pension proposal, based on a Monte Carlo lifecycle microsimulation model. The main results are first that the proposal tends to be mildly redistributive from higher to lower lifetime income groups. Secondly, the proposal is of not as much benefit to women as might be expected-it is almost equal in value to men and women. This later conclusion is the result of the fact that the homemaker pension provision was part of a package tt at also included splitting of pension credits accrued during marriage.
LIFEMOD — The Formative Years, Welfare State Programme Research Note WSP
  • J Falkingham
  • C Lessof
Falkingham, J. and Lessof, C. 1991, LIFEMOD — The Formative Years, Welfare State Programme Research Note WSP/RN/24, London School of Economics.