Productivity Commission
Research Paper
Rising inequality?
A stocktake of the evidence
August 2018
 Commonwealth of Australia 2018
ISBN 978-1-74037-663-1 (PDF)
ISBN 978-1-74037-662-4 (print)
Except for the Commonwealth Coat of Arms and content supplied by third parties, this copyright work is
licensed under a Creative Commons Attribution 3.0 Australia licence. To view a copy of this licence, visit
http://creativecommons.org/licenses/by/3.0/au. In essence, you are free to copy, communicate and adapt the
work, as long as you attribute the work to the Productivity Commission (but not in any way that suggests the
Commission endorses you or your use) and abide by the other licence terms.
Use of the Commonwealth Coat of Arms
Terms of use for the Coat of Arms are available from the Department of the Prime Minister and Cabinet’s
website: https://www.pmc.gov.au/government/commonwealth-coat-arms
Third party copyright
Wherever a third party holds copyright in this material, the copyright remains with that party. Their permission
may be required to use the material, please contact them directly.
Attribution
This work should be attributed as follows, Source: Productivity Commission 2018, Rising inequality? A
stocktake of the evidence.
If you have adapted, modified or transformed this work in anyway, please use the following, Source: based on
Productivity Commission data, Rising inequality? A stocktake of the evidence.
An appropriate reference for this publication is:
Productivity Commission 2018, Rising inequality? A stocktake of the evidence, Commission Research Paper,
Canberra.
Publications enquiries
Media, Publications and Web, phone: (03) 9653 2244 or email: mpw@pc.gov.au
The Productivity Commission
The Productivity Commission is the Australian Government’s independent research and advisory
body on a range of economic, social and environmental issues affecting the welfare of Australians.
Its role, expressed most simply, is to help governments make better policies, in the long term interest
of the Australian community.
The Commission’s independence is underpinned by an Act of Parliament. Its processes and outputs
are open to public scrutiny and are driven by concern for the wellbeing of the community as a whole.
Further information on the Productivity Commission can be obtained from the Commission’s website
(www.pc.gov.au).
CONTENTS iii
Contents
Acknowledgements v
Abbreviations vi
Executive summary 1
1 About this study 7
1.1 What this study is about 7
1.2 An overview of what we do 8
1.3 The broader economic context 11
2 Framework and approach 19
2.1 The conceptual framework 19
2.2 Operationalising the conceptual framework 25
2.3 Our analytical approach 27
3 Income and consumption inequality 37
3.1 Trends in income and income inequality 38
3.2 The distribution of income in detail 44
3.3 The demographics of the income distribution 58
3.4 Comparing the distributions of income and consumption 62
4 Wealth inequality 69
4.1 Trends in wealth and wealth inequality 70
4.2 Trends in the distribution of components of wealth 77
4.3 The demographics of wealth and income 80
5 Economic mobility 89
5.1 How does mobility relate to inequality? 90
5.2 Intergenerational mobility in Australia 91
5.3 Life course mobility in Australia 94
iv RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE
6 Economic disadvantage 107
6.1 How disadvantage relates to inequality 108
6.2 The prevalence of poverty 110
6.3 The demographics of poverty 117
6.4 How long does poverty last? 121
6.5 Material deprivation 127
6.6 Social exclusion 133
Glossary 141
References 147
ACKNOWLEDGEMENTS v

Acknowledgements
The Commission is grateful to all those who have given their time to share their experiences
and expertise in support of this flagship research paper.
The Commission wishes to particularly thank our external referees — Roger Wilkins
(Melbourne Institute, University of Melbourne) and Peter Whiteford (Crawford School of
Public Policy, Australian National University) — for helpful feedback.
This paper was produced by Patrick Jomini, Joshua Craig, Lisa Leong, Henry Williams,
Claire Prideaux and Thithi Nguyentran. The study was overseen by Commissioner
Jonathan Coppel.
This paper uses unit record data from the Household, Income and Labour Dynamics in
Australia (HILDA) Survey. The HILDA Project was initiated and is funded by the Australian
Government Department of Social Services (DSS) and is managed by the Melbourne
Institute of Applied Economic and Social Research (Melbourne Institute). The findings and
views reported in this paper, however, are those of the Commission and should not be
attributed to either DSS or the Melbourne Institute.

vi RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE
Abbreviations
ABS Australian Bureau of Statistics
ACOSS Australian Council of Social Services
CEDA Committee for Economic Development of Australia
GDP Gross domestic product
GFC Global Financial Crisis
GST Goods and services tax
HES Household Expenditure Survey
HILDA Household, Income and Labour Dynamics in Australia
IMF International Monetary Fund
OECD Organisation for Economic Cooperation and Development
PC Productivity Commission
PPP Purchasing power parity
RBA Reserve Bank of Australia
SEM Social Exclusion Monitor
SIH Survey of Income and Housing
EXECUTIVE SUMMARY 1
Executive summary
Over nearly three decades, inequality has risen slightly in Australia
In all societies some inequality occurs due to differences in ability, opportunity, effort and
luck. Institutional and policy constructs can add to this, or detract from it.
Moreover, excessive inequality and entrenched disadvantage can erode social cohesion and
hinder growth. It can also sap investment in education and skills and slow productivity
growth. Yet there is no precise causative relationship, let alone a consensus on how much
inequality matters. It is a topic that continues to draw diverse and competing views.
This study does not directly enter these debates. Rather, its purpose is to contribute a
foundation to an informed discussion on inequality and its social impacts, bringing together
and taking stock of the latest and most complete evidence measuring the level of and trends
in inequality, poverty and disadvantage in Australia via multiple means.
While comprehensive, this study is not exhaustive; other studies examine geographic, racial
or gender inequality.
Even this modest level of ambition is not without its challenges. No single metric is sufficient
to give a definitive answer to the seemingly straightforward question: have inequality, poverty
and disadvantage in Australia risen, fallen or remained steady in recent years?
Our focus, therefore, eschews the specific and often self-serving use of any one measure of
inequality. Instead we use an array of indicators that examine the distributions of household
incomes, consumption and wealth, their composition and importantly, movement within the
distributions over time, and in response to life events, such as transitions to work, divorce
and retirement. For poverty and disadvantage our approach goes beyond the standard
metrics, giving weight to measures that capture the experience of those households in the
bottom part of the distribution.
The broader context for this study has been an evident reduction in global income inequality
and poverty since the late 1980s, the time-frame we most often apply. At the same time,
however, there has been rising inequality within many developed countries. We review the
Australian experience, which is less dire than some would have it, but not exemplary.

2 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE

Sustained growth has delivered significantly improved living standards for
the average Australian in every income decile
What also distinguishes Australia from most other developed countries has been its
unprecedented 27-year period of uninterrupted economic growth, prompting many to ask
how the economic gains from growth have been shared. While growth is no guarantee
against a widening disparity between rich and poor, we show that it has delivered for the
average Australian household in every income decile significantly improved living
standards. This is in contrast with the United States (which had a similar rate of increase in
income inequality as Australia) where the distribution is much more uneven, with income
growth in the lower deciles about a quarter of that for Australian households.
What matters more than economic growth for understanding trends in inequality are the
sources of income growth (labour, capital and transfers). These fluctuate in ways that
sometimes favour those on high incomes and sometimes favour those on low incomes. For
example, the mining boom was a period that favoured high income earners and capital income,
lifting measures of inequality. In contrast, the post-Global Financial Crisis period has benefited
lower income groups, despite weak overall growth in labour income. Among the various forces
acting on inequality and poverty, the one constant that matters is having a job.
Over recent decades income growth rates by age group have also varied substantially, but
for the most part, the variation reflects overall trends in the strength of income growth. That
is, when the economy is strong, all age groups tend to benefit from higher income growth
and when the economy is weak, all age groups tend to experience lower income growth. But
at different times, some age groups have benefited more or less than others. Most recently,
young people’s incomes have grown relatively slowly. On average, however, each new
generation has earned more income than the last at a given age, and reaches the same level
of income earlier in life.
Examining more closely the demographics of the income distribution provides additional
insights. We know for example that Australians in their prime working years are more likely
to be in the middle and upper income deciles, whereas over-65 year olds are over-represented
in lower income deciles, reflecting retirement and reliance on the Age Pension. We also
know that individuals living in households where no person is in paid work are strongly
concentrated in the lower deciles, especially if there are dependent children in the household.
Similarly, households with dependent children and two or more income earners are
over-represented in middle and upper income deciles, and working households without
dependent children tend to be over-represented at the top of the distribution.
Australia’s progressive tax and highly targeted transfer systems
substantially reduce inequality
Another clear message from the data is that Australia’s progressive tax system and highly
targeted transfer system substantially reduce income inequality. Income tax and government
EXECUTIVE SUMMARY 3
transfers have typically lowered the measure of overall income inequality (the Gini
coefficient) by 30 per cent, an equalising effect that far outweighs the overall increase in the
measure since the late 1980s. This equalising effect has fluctuated over time, but overall
there has been no material change in the past thirty years. Redistributive tax policies can,
however, also have unintended negative consequences on economic efficiency, for example,
inciting a reduction in labour supply.
While income is usually given prominence in debates about inequality, how evenly
consumption is distributed is often a better measure, as consumption contributes most
directly to wellbeing. Moreover, income patterns alone do not capture the importance of
in-kind transfers from government, such as health, education, childcare subsidies and
government housing. These in-kind transfers have an additional equalising effect, because
people with low incomes (and households with children) receive the largest amount of
in-kind transfers. When the more expansive measure of final consumption is used, overall
inequality (the Gini coefficient) is about 30 per cent lower again than that for disposable
household income. In-kind transfers can also bear on future inequality by opening doors to
greater opportunities and lifting incomes later on.
The distinction between income and consumption comes out most strongly in analysis by
age. For example, while 25 to 34 year olds are over-represented in upper deciles for income,
they are over-represented in lower deciles for final consumption. This reflects reduced
reliance on the education and health system in this age group, as well as higher rates of
savings. On the other hand, those aged 65 or older, who are strongly over-represented in
lower deciles for income, are under-represented in lower deciles for final consumption.
The distribution of wealth is relevant too. Household welfare depends not just on resources at
a point in time, but over time as well, and wealth provides a sense of financial security. Wealth
can also provide an important safety net for older Australians, many of whom have relatively
low incomes but high wealth, in terms of managing aged care costs and longevity risk.
Similar to income, growth in wealth has been spread widely across the population. On
average, households in all but the bottom decile experienced real increases in wealth,
predominantly in housing assets and superannuation balances over the past fifteen years.
However, with the growth in wealth strongest in the upper deciles, some measures of wealth
inequality have risen. While wealth distribution in Australia somewhat predictably is more
unequal than income or consumption, Australia’s wealth distribution remains less skewed
than in other countries. Among 28 OECD countries, Australia ranks eighth most equal, as
measured by the Gini coefficient of wealth.
The fact that inequality levels are so different among developed countries hints strongly at
the scope for policies, institutions and political environments to shape inequality.

4 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE
Economic mobility is high in Australia, with almost everyone moving across
the income distribution over the course of their lives …
The standard inequality measures considered thus far give a snapshot of the distribution at a
point in time. While they show some widening of the gap between households, that does not
mean that the rich and the poor households at the beginning and the end of the period are the
same households.
The distinction is important because a society with a given level of inequality, and where
household incomes are static over time, faces different and more serious policy challenges
than a society with the same level of inequality but where household incomes are mobile.
There are two types of mobility: intergenerational mobility and life course mobility.
Intergenerational mobility refers to the relationship between a person’s economic position
and that of their parents, and life course mobility refers to changes in a person’s economic
position throughout their life. The limited timeframe of Australian longitudinal data limits
our capacity to assess intergenerational mobility. Instead, we present original analysis on the
degree of life course mobility in Australia using the HILDA dataset. In other words, how
much people move across the distribution for income or wealth from year to year.
It turns out that almost everyone moves across the income distribution over the course of
their lives. Over a 16-year period, the average Australian was classified in five different
income deciles; and for less than one per cent of people, the decile to which they belonged
remained unchanged over the whole period. And nine per cent spent time in both the top and
the bottom income decile. A lower, but still significant level of mobility was also apparent
across the wealth distribution. This highlights the fluid nature of income and wealth: over
time, any given decile consists of a different group of people — most of the people in the
top decile today were not there fifteen years ago.
Life events — such as transitioning from education into work, career advancement,
household formation, having children, divorce and retirement — underpin some of the
observed trends in economic mobility. Typically, income rises during the working years,
though this can be interrupted by childbearing and other life events, such as ill health.
Similarly, Australians accumulate wealth in their middle years, and draw on this wealth in
retirement when their earnings drop. These changes in income and wealth allow people to
‘smooth’ their consumption.
… but some Australians experience entrenched economic disadvantage
While life course mobility affects households across the entire distribution, the ends of the
distribution are ‘stickier’ than the middle. Households in the top and bottom two income
deciles at the beginning of the period were the most likely to be in the same decile fifteen
years later. About three per cent of households were stuck in one of the bottom two deciles
throughout the period. Stickiness at the ends of the distribution is indicative of some
entrenched inequality.
EXECUTIVE SUMMARY 5
Accordingly, the last chapter of this study updates earlier Commission research on the nature
and extent of deep and persistent disadvantage in Australia. Disadvantage is a
multidimensional concept that can take the form of low economic resources (poverty),
inability to afford the basic essentials of life (material deprivation) or being unable to
participate economically and socially (social exclusion). Because the elements of
disadvantage encompass a diverse range of indicators, it is difficult to reach a single
conclusion about the overall trend in disadvantage.
Many Australians experience economic disadvantage at some stage in their lives, but for
most, it is temporary. About nine per cent of Australians (2.2 million people) experienced
relative income poverty (income below 50 per cent of the median) in 2015-16, with children
and older people having the highest rates of relative income poverty. This aggregate figure
has fluctuated since 1988-89 but, despite 27 years of uninterrupted growth, has not declined.
Persistent and recurrent poverty affects a small, but significant proportion of the population.
About three per cent of Australians (roughly 700 000 people) have been in income poverty
continuously for at least the last four years. People living in single-parent families,
unemployed people, people with disabilities and Indigenous Australians are particularly
likely to experience income poverty, deprivation and social exclusion. For people in these
circumstances, there is an elevated risk of economic disadvantage becoming entrenched,
limiting their potential to seize economic opportunities or develop the skills with which to
overcome these conditions.
These risks are particularly elevated for children living in jobless households, which is a
group that has stood out among the multiple measures of inequality and disadvantage.
How the future of inequality in Australia evolves will depend on the opportunities that
citizens have to improve their living standards today. For by far the largest number of us,
sustained economic growth and reliable access to employment — complemented by skills
and education improvements as specified in our 2017 report, Shifting the Dial — will offer
these opportunities and the ability to embrace them.
But for a significant albeit much smaller group, the challenges are much more complex.
Growth and complementary improvements in skills and education policies will not be
enough. In some previous research, we found that needs in housing or health policies could
better be fashioned to address more directly than today quite specific needs — what might
be termed ‘hand-made’ policy — as we look out towards a fourth decade of uninterrupted
economic growth.
ABOUT THIS STUDY 7
1 About this study
1.1 What this study is about
Researchers, policy makers and the general public have always been interested in questions
of economic inequality, but the level of interest in Australia and other developed economies
has escalated in recent years. There is a vigorous debate about how inequality affects
people’s wellbeing, its causes and consequences, and whether and what type of policy
responses might be required. It is a topic that continues to draw diverse and competing views
(box 1.1).
Influential economists such as Thomas Piketty and Branko Milanović, organisations
including the OECD and the IMF, and numerous commentators have contributed to this
renewed interest. The OECD has identified inequality as one of the key challenges facing
the world (OECD 2018c). So have some political leaders internationally and at home, the
latter including a 2014 Senate Inquiry into the extent of income inequality in Australia
(SCARC 2014, p. 18).
This study does not directly enter the debates these gathering comments have triggered.
Rather, its purpose is to contribute to an informed discussion in Australia by bringing
together and taking stock of the latest and most complete evidence measuring the prevalence
of, and trends in, inequality, economic mobility and disadvantage across Australian society.
It updates earlier Commission research for developments in the post-global financial crisis
period, and adds to recent analyses on inequality, such as work by Whiteford (2018), Kaplan,
La Cava and Stone (2018) and the Committee for Economic Development of Australia
(CEDA 2018b).
While comprehensive, this study is not exhaustive. Other dimensions of inequality — such
as the distribution of income, consumption and wealth by location or gender — are not
examined in depth. Some analysis of economic conditions by geographic region can be
found in the Commission’s Transitioning Regional Economies study (PC 2017c). Nor does
the study include a detailed analysis of disadvantage as it affects Indigenous Australians. For
detailed discussion of such issues, readers can refer to the Commission’s series of work on
Overcoming Indigenous Disadvantage (SCRGSP 2016).
Even this study’s modest level of ambition is not without its challenges. No single metric is
sufficient to give a definitive answer to the seemingly straightforward question: have
inequality, economic mobility and disadvantage in Australia risen, fallen or remained steady
in recent years? This is the case because these concepts are multidimensional, and they link
to each other — and to broader notions of wellbeing — in complex ways. It is also because
different datasets can show different results. Our focus, therefore, eschews the specific and
often self-serving use of any one measure of inequality.
8 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE
Box 1.1 Contrasting views on inequality
Some commentators argue that inequality is not a major problem in Australia and does not require
policy action.
The best available evidence demonstrates that income inequality is low and declining in Australia.
… Access to basic material goods is close to universal. Absolute poverty has been virtually eradicated,
while relative poverty is low and has been declining in recent years. (Wild and Bushnell 2017, p. 40)
Most measures suggest income inequality [in Australia] has now stabilised or diminished … And there is
no reason to think that stabilisation has occurred at a level of inequality that is dangerously high,
undermining equality of opportunity, worsening poverty or in other ways eroding the social fabric. On the
contrary, far from income groups hardening into hereditary castes — with all the inequity and inefficiency
that would bring — social mobility in Australia remains relatively high. (Ergas 2017)
Economic inequality is not intrinsically bad and equality does not equate to fairness. Policies need to
give more attention to incentive. Australia’s tax, social security and welfare systems are already highly
redistributive towards a more equal income distribution and very effective in saving people from absolute
poverty. (Carling 2017)
Others maintain that inequality is a growing problem that should be addressed.
A high-inequality society is a highly unstable society. When assets in a society are distributed in a way
that most people believe is unfair, it creates a form of systemic risk. … When we look at the rise of
populist politics around the world, it’s clear that inequality helps foster the sense that mainstream politics
isn’t delivering. … A more equal nation will have higher levels of wellbeing, more mobility, and more
stability. (Leigh 2017)
At the heart of Australia’s society and economy is the idea of the ‘fair go’: the notion that, if we work hard
enough, we will be able to get ahead no matter our gender, ethnicity, or our post code. But in recent
years, the fair go has been under threat, particularly as wage and income inequality has widened, leaving
more Australians behind. (Rajadurai 2018, p. 12)
Income inequality is indeed a drag on the ability to convert wealth to well-being. … Such findings also
serve as a vivid reminder for politicians and public servants alike that inequality is an issue that can be
kicked down the road no longer. It needs confronting here and now … (Chin 2017)
Human beings have deep-seated psychological responses to inequality and social hierarchy.
… [I]nequality colours our social perceptions … which affect the way we relate to and treat each other.
… A growing body of research shows that inequality damages the social fabric of the whole society.
(Wilkinson and Pickett 2014)
1.2 An overview of what we do
We use an array of measures and indicators to explore changes in the distribution of income
and consumption (chapter 3) and wealth (chapter 4), drawing on the now extensive set of
information in household surveys — namely the Household Expenditure Survey and the
Household, Income and Labour Dynamics in Australia survey — collected over the past
15 years or more.
This provides for a richer analysis of economic inequality than using annual data or income
alone, because consumption and wealth are distributed across the population in different
ways, and it is the goods and services people consume, not the income they earn, that
contributes to economic wellbeing. Wealth also contributes directly to wellbeing by
providing a sense of financial security and social prestige.
ABOUT THIS STUDY 9
Measures of income, consumption and wealth are also decomposed into their constituent
parts. This gives a better understanding of the forces influencing the distribution, and allows
a top-down perspective on how Australia’s tax and transfer system affects summary
distribution measures, such as the Gini coefficient (described in chapter 2).
Economic opportunity and mobility
The inequality measures presented in chapters 3 and 4 reflect a snapshot of inequality among
households at a point in time. This will change over time, as households’ economic resources
fluctuate — often by significant amounts. Yet, observing a widening gap over time between
‘rich’ households and ‘poor’ households does not mean that the rich and the poor households
are the same ones at the beginning and the end of the period.
The distinction is important, because a society with a given level of inequality and where
household incomes are static does not face the same challenges as a society with the same
level of inequality but where household incomes are mobile. Indeed, there is greater
consensus that policy should be more concerned about significant, persistent inequality of
opportunities and barriers to economic mobility than inequality of outcomes (Argy 2006,
p. 51). Accordingly, information on the degree and form of mobility is relevant, and can
offer richer insights on how to interpret trends in income inequality.
Economic mobility is manifest in two forms.
 Intergenerational mobility refers to the relationship between parents’ socioeconomic
status and their children’s socioeconomic status, and often focuses on the transmission
of advantage or disadvantage across generations. When based on analysis of income
alone, intergenerational mobility is sometimes framed in terms of ‘intergenerational
earnings elasticity’ — a quantitative measure of how a person’s earnings are affected by
their parents’ earnings, where a higher earnings elasticity indicates less income mobility.
 Life course (intragenerational) mobility refers to the movements of individuals across
the economic distribution throughout their lives (d’Addio 2007, p. 12). This form of
mobility incorporates life events such as transitioning from education to work,
cohabitation and family formation, and retirement. It can also reflect the impact of
unexpected changes in circumstances (such as illness, injury or disability) that affect a
person’s capacity to work.
Due to the limited timeframe of Australian longitudinal data, intergenerational analysis of
full life cycles is not possible, although we review existing studies that estimate Australia’s
intergenerational mobility using a variety of methods. We primarily examine people’s
changing life circumstances (life course mobility) — such as transitions from study to work,
household formation or dissolution, and retirement — and use some cross-sectional analysis
to draw out insights into the effect of these events (chapter 5).
10 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE
As Australian longitudinal datasets continue to mature, future possibilities for a fuller
analysis will open up. For example, there could be value in undertaking econometric analysis
of the determinants of income mobility (demographic and otherwise).
Economic disadvantage
The presence of inequality does not necessarily mean that some people always live in
poverty (a state of insufficient economic resources) or experience economic disadvantage
more generally. Rich countries like Australia typically have a low prevalence of absolute
poverty. But in practice, higher economic inequality often indicates that a greater proportion
of people experience inadequate resources and opportunities (Bureau for Development
Policy 2013, p. 27; Douglas et al. 2014, p. 22).
Poverty and other forms of disadvantage can suppress a person’s ability to improve their
economic situation, such as by affecting their ability to find work or to invest in their own
skills through education and training. Those constrained opportunities can in turn limit a
person’s capacity to improve their economic circumstances or ‘climb up the ladder’, and
may feed into a cycle of widening inequality if people with few economic resources are
‘stuck on the bottom rung’. Looking at the prevalence of economic disadvantage is also
relevant in its own right.
For these reasons, metrics of the prevalence of disadvantage, and people’s movements into
and out of disadvantage, add an important dimension to the discussion of inequality
(chapter 6). For example, concerns about inequality may be assuaged if poverty spells are
mostly temporary, if the prevalence of poverty is decreasing, and/or if income mobility
allows people to improve their economic positions over time. It is also possible that
increasing inequality could be accompanied by declining poverty if, for example, all incomes
were rising but the income distribution was only ‘stretching out’ in the top deciles.
More detailed analysis of the severity and duration (persistence) of disadvantage, the
characteristics of people experiencing disadvantage, and the causes, costs and consequences
of disadvantage can also be found in the Productivity Commission’s earlier analysis of Deep
and Persistent Disadvantage in Australia (McLachlan, Gilfillan and Gordon 2013).
Framework and approach
The overarching framework for thinking about inequality, mobility and disadvantage from
a broader wellbeing perspective, and how these concepts are related to each other, is outlined
in chapter 2. Chapter 2 also explains the Commission’s analytical approach.
This study uses the most up-to-date data sources. They include the late 2017 release of new
waves of two major nationally representative household surveys: the ABS Household
Expenditure Survey (HES) and the Melbourne Institute Household, Income and Labour
ABOUT THIS STUDY 11
Dynamics in Australia (HILDA) Survey. The study also uses the ABS Survey of Income and
Housing (SIH), which has been integrated with HES since 2003-04.
HES has been used as the main data source for cross-sectional analysis, because it is
available over the longest period of time and (unlike SIH and HILDA) includes
comprehensive consumption data in addition to data on income and wealth. All
cross-sectional results based on HES have been compared with HILDA and, where
discrepancies between HES and HILDA are apparent, these are indicated in footnotes.
HILDA has been used for longitudinal analysis and for some cross-sectional analysis where
it contains relevant variables that are not in HES. Finally, the SIH provides more frequent
estimates for some summary measures, as the SIH is run more often than HES.
In longitudinal studies such as HILDA, the representativeness of the sample atrophies over
time. One reason for this is attrition, which occurs when people drop out of a survey, and
which affects all longitudinal studies. In HILDA, this attrition is not random (Summerfield
et al. 2017, p. 180), which could bias the results in this study. Specifically, people who are
relatively young (15–24 years), born in a non-English speaking country, unemployed or
working in low-skilled occupations are particularly likely to leave the survey (Summerfield
et al. 2017, p. 180).
The Melbourne Institute makes adjustments for this attrition by reweighting the sample; that
said, these adjustments are not perfect. Another source of declining representativeness of the
HILDA Survey is that immigrants arriving in Australia after 2001 could not be included in
the initial sample. In 2011, the sample was topped up with new participants, including recent
immigrants. However, Wilkins (2014, p. 78) found that including or excluding the top-up
sample had little effect on 2010-11 income statistics.
1.3 The broader economic context
Changes in the distribution of household income, consumption and wealth occur in the
context of changes in the broader economic landscape, internationally and within Australia.
The international context
Internationally, the forces of globalisation, technological advances and structural reforms
have underpinned economic growth since World War II. Initially, rising incomes were
concentrated in developed economies, and a few rapidly developing economies, such as
South Korea. However, the late 1990s marked a dramatic change in the spread of economic
growth, as emerging economies (most notably China and India) overtook developed
economies in their rates of growth (Subramanian and Kessler 2013, p. 2). These economies
continue to catch up with the developed world.
This pattern of economic growth has altered the distribution of global incomes over the past
twenty years. Notably, high rates of income growth experienced by many in the bottom half
12 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE
of the global income distribution (figure 1.1) greatly reduced poverty: between 1990 and
2013, the number of people in extreme poverty (living on less than US$1.90 per day) fell
from 1.85 billion to 767 million (World Bank 2016, p. 5). This growth pattern suggests there
has been some reduction in inequality across countries, albeit accompanied by high rates of
income growth at the top of the distribution (depicted by the tip of the elephant’s trunk
in figure 1.1).
Figure 1.1 The ‘elephant curve’ — high income growth has lifted a large
number of the world’s population out of poverty
Growth in mean equivalised real income for each percentile of the world
population, 1988–2008a
a Vertical axis shows the change in real income (the difference between 1988 real income and 2008 real
income for each percentile, as a percentage of 1988 real income), in constant international dollars.
Source: Corlett (2016, p. 5), based on data from Milanović and Lakner (2013).
In contrast, inequality has been increasing within many countries, particularly in advanced
market economies. The Gini coefficient of equivalised disposable income — a measure of
overall income inequality — rose by an average of 9 per cent, across 13 OECD countries,
between 1989 and 2011.1 Chapter 3 provides further discussion on trends in income
distribution in OECD countries.
1 Or nearest years available. Commission estimate based on OECD (2015), Webber and Mallett (2017),
LIS (2017).
ABOUT THIS STUDY 13
The Australian context
To place trends in inequality in Australia into a broader context, it is instructive to look at
the Australian economic landscape over the past three decades.
What distinguishes Australia from most other developed countries has been its
unprecedented 25-year period of uninterrupted economic growth, which included a sustained
period of disposable income growth between 1992-93 and 2007-08. Over the same period,
the unemployment rate decreased from about 11 per cent to 4 per cent (figure 1.2).
In contrast, the post-mining boom period has included a four-year span of falling per capita
disposable incomes (figure 1.2), with low wage increases and ‘labour productivity … lower
than [during] both the “golden era” of the mid-1990s, and the lengthy prosperous period
from 1950 to 1970’ (PC 2017b, p. 29).
Figure 1.2 Australia has had sustained growth in real incomesa,b
a Chain volume measures of real net national disposable income per capita for year ended June (reference
year 2015-16). b Unemployment rate is trend data, as at June.
Sources: ABS (Australian National Accounts: National Income, Expenditure and Product, Sep 2017,
Cat. no. 5206.0; Labour Force, Australia, Dec 2017, Cat. no. 6202.0).
This period of growth coincided with a number of structural and cyclical developments.
Microeconomic reforms implemented through the 1980s and 1990s have been credited with
underpinning sustained growth in output, productivity and incomes (PC 2005, p. xii). These
included reductions in industry assistance measures and tariffs, privatisation of government
business enterprises, the shift away from centralised wage determination to enterprise
bargaining, and reducing barriers to competition in markets for essential infrastructure
0
2
4
6
8
10
12
-6
-4
-2
0
2
4
6
8
1988 1991 1994 1997 2000 2003 2006 2009 2012 2015
P
er
c
en
t
P
er
c
en
t
Growth in real net national
disposable income per
capita (LHS)
Unemployment rate (RHS)

14 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE

services, such as electricity, gas, road transport and water. For example, productivity
improvements and price changes that occurred during the 1990s in selected essential
infrastructure services resulted in an estimated 2.5 per cent permanent addition to Australia’s
gross domestic product (GDP) (PC 2005, p. 35).
Significant growth in housing prices since the early 2000s (particularly in eastern seaboard
capital cities) has been an important driver of household wealth among homeowners,
particularly among older Australians (PC 2015a, p. 12). Residential property prices in capital
cities more than doubled in nominal terms between 2003 and 2017 (figure 1.3). However,
the growth in house prices has also been accompanied by a significant increase in household
debt (Lowe 2017).
The mining investment boom between 2005 and 2013 contributed significantly to
economic growth, employment and incomes. Downes, Hanslow and Tulip (2014, p. 1)
estimated that by 2013, the mining boom had increased real per capita disposable income by
13 per cent and real wages by 6 per cent, and reduced the unemployment rate by
1.25 percentage points. Mining regions experienced particularly strong growth in average
incomes during the boom, but slower growth in the aftermath (PC 2017c, p. 96).
Figure 1.3 Capital city house prices have grown strongly since the early
2000s, contributing to large increases in measured wealth
(a) Residential property prices across
Australian capital citiesa

(b) Median house prices in
Sydney and Melbourneb
a Residential Property Price Index (weighted average of eight capital cities). September quarter. Reference
period is September 2011 = 100.0. b Median price of established house transfers, September quarter.
Source: ABS (Residential Property Price Indexes: Eight Capital Cities, Sep 2017, Cat. no. 6416.0).
Australia’s experience after the global financial crisis (GFC) suggests a certain resilience
to economic shocks. The GFC was characterised by a small and brief drop in per capita GDP
growth, and a much smaller increase in unemployment than was experienced in the
60
70
80
90
100
110
120
130
140
150
160
2003 2006 2009 2012 2015
R
es
id
en
ti
al
p
ro
p
er
ty
p
ri
ce
in
d
ex
On average,
house prices
in capital
cities have
doubled
200
300
400
500
600
700
800
900
1,000
2003 2006 2009 2012 2015
M
ed
ia
n
h
o
u
se
p
ri
ce
(
$'
00
0)
In Melbourne,
they have more
than doubled
Sydney
Melbourne

ABOUT THIS STUDY 15

early 1990s recession. It also marked the beginning of an extended period of low interest
rates, and prompted a substantial fiscal expansion between 2008 and 2012 (Makin 2016,
pp. 4–5). The post-GFC period has also featured:
 slower wage growth, with a trend towards smaller and less frequent wage increases
(Bishop and Cassidy 2017, p. 17)
 higher labour productivity growth than the mid-2000s (attributed to increased
physical capital), although lower labour productivity growth than the mid-1990s (PC
2017b, p. 33)
 an increased household saving ratio, reflecting increased precautionary savings due to a
reduction in expected future incomes (Price and Finlay 2014, p. 1).
Employment and demographic trends also form an important part of the broader economic
context underpinning trends in inequality, and they bear on trends in equivalised measures
of household resources. Ongoing, structural changes in Australia’s labour market
(figure 1.4) are characterised by:
 a rising labour force participation rate (from 63 per cent in 1989 to about 65 per cent
in 2016), with increased labour market participation among women (figure 1.4, panel a)
as well as a steadily increasing share of part-time employment (Cassidy and
Parsons 2017, p. 19)
 an ageing population (over the longer term), with an increasing proportion of the
population aged 65 years and over (McDonald 2016, p. 65). However, more people aged
65 years and over are continuing to work: between 1989 and 2016, the labour force
participation rate of people in this age group more than doubled to reach nearly
13 per cent (figure 1.4, panel b). This has contributed to a shift in the age composition of
the workforce towards older people (figure 1.4, panel c).
A gradual sectoral shift has also been evident, as the proportions of workers employed in the
manufacturing and agricultural sectors have halved since 1989 (figure 1.4, panel d).
Australia’s strong economic performance and resilience have underpinned steady growth in
average real household income and consumption. Average equivalised disposable household
income rose from $31 000 in 1988-89 to $54 000 in 2015-16 (in 2016-17 dollars) (chapter 3).
Household consumption rose by less over the same period, largely because the household
saving rate rose following the GFC (ABS 2018b).
Changing perceptions about living standards
These broadly based increases in material living standards bear on trends in economic
disadvantage and open up many new economic opportunities. That said, the overall increase
in average incomes and consumption belies the perceptions of many Australians.
Many Australians report they are ‘doing it tough’. ACOSS et al. (2015, p. 17) found that
‘there is a widespread belief that a rising proportion of Australians are struggling financially’

16 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE

and, in a recent survey, 51 per cent of Australians believed that their income had fallen
behind the cost of living in the past two years (Essential Research 2018, p. 9). In another
recent survey, 44 per cent of respondents felt that they had not personally benefited at all
from 26 years of economic growth (CEDA 2018a, p. 14).2

Figure 1.4 The changing structure of Australia’s labour force
(a) Increased female labour force
participation ratea

(b) Increased labour force participation rate
among older Australiansa
(c) Age group shares of total employment
(d) Rising services sector share of total
employmentb
a As at June. b Annual averages (to August) calculated from quarterly data.
Sources: ABS (Labour Force, Australia, Dec 2017, Cat. no. 6202.0; Labour Force, Australia, Detailed —
Electronic Delivery, Jan 2018, Cat. no. 6291.0.55.001; Labour Force, Australia, Detailed, Quarterly,
Nov 2017, Cat. no. 6291.0.55.003).
2 The Commission has not directly investigated perceptions of inequality for this study, but it could form part
of a future stream of research.
45
50
55
60
65
70
75
80
19
89
19
92
19
95
19
98
20
01
20
04
20
07
20
10
20
13
20
16
P
er
c
en
t
Males
All
Females
0
10
20
30
40
50
60
70
19
89
19
92
19
95
19
98
20
01
20
04
20
07
20
10
20
13
20
16
P
er
c
en
t
65+ years old
All age groups
0
2
4
6
8
10
12
14
16
15
–1
9
20
–2
4
25
–2
9
30
–3
4
35
–3
9
40
–4
4
45
–4
9
50
–5
4
55
–5
9
60
–6
4
65
+
S
h
ar
e
(p
er
c
en
t)
2017
1989
0
20
40
60
80
100
19
89
19
92
19
95
19
98
20
01
20
04
20
07
20
10
20
13
20
16
P
er
c
en
t Services
Manufacturing
Agriculture

ABOUT THIS STUDY 17

These views may reflect more on the real declines in per capita incomes in the post-mining
boom years, rather than changes in the distribution of economic resources. Others have
suggested it could be that ‘Australians have lost perspective and … confuse cost of living
with cost of a lifestyle’ (Wade 2017). Regardless of the reason, they indicate a risk of social
fracture that cannot be ignored. This study places these debates on a level footing by
presenting the facts.


FRAMEWORK AND APPROACH 19

2 Framework and approach

Key points
 Inequality, disadvantage and economic mobility are related concepts that each affect wellbeing
in different ways.
– Inequality affects people’s wellbeing through their values and preferences in relation to the
societal distribution of resources as well as their expectations about acceptable living
standards.
– Disadvantage (including poverty) directly impacts on wellbeing by limiting people’s ability
to achieve the life outcomes they value.
– Economic mobility is an important indicator of the extent of, and access to, opportunities
for people to improve their economic situation.
 Our approach focuses on analysing the distribution of three broad measures of economic
resources: income, consumption and wealth. We consider how the distributions vary over time
and across groups, and examine some of the contributors to their movements.
– The measures of income, consumption and wealth are linked in an accounting sense and
provide better insights than a single measure does. Analysing these three measures
together has become possible with the maturing of Australian household surveys.
– The analysis is based on equivalised household measures of income, consumption and
wealth to account for differences in household composition and ‘economies of scale’ when
sharing living costs.
 Many indicators can be used to analyse and report on inequality, mobility and disadvantage.
It is even possible for two metrics to lead to different findings on the level of, or trends in,
inequality. This is in part because of multiple data collections using different methodologies
and sampling techniques. Accordingly, we present and draw conclusions based on an array
of indicators.


This chapter sets out a conceptual framework for considering inequality from a wellbeing
perspective and describes the analytical approach used to apply the framework. Section 2.1
places inequality in the broader context of wellbeing, and presents a conceptual model of
how economic resources affect people’s wellbeing. Section 2.2 then discusses how the
model is operationalised using income, consumption and wealth as measures of economic
resources. Section 2.3 outlines the study’s analytical approach.
2.1 The conceptual framework
This study frames inequality, mobility and disadvantage — and how they relate to each
other — within a wellbeing framework.

20 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE

There are many definitions of wellbeing, but they all typically encompass ideas such as
‘living well’, ‘functioning well’ and enjoying a good quality of life (MHCNSW 2017, p. 9)
and incorporate both objective and subjective elements (OECD 2011, p. 266; Stiglitz, Sen
and Fitoussi 2010, p. 15).
 Objective aspects of wellbeing are about functioning well in life. These are easier to
measure, using indicators such as employment, health status, education and housing
conditions (OECD 2011, p. 19), but they do not account for differences in the levels of
satisfaction that people derive from these circumstances.
 Subjective aspects of wellbeing are about how people feel and how they evaluate their
lives (MHCNSW 2017, p. 13). They are measured using self-reported indicators, such as
life satisfaction. Subjective wellbeing has the advantage of accounting for preferences
and values, but it might be inaccurate in some situations, such as where people become
acclimatised to extreme adversity (Sen 1992, p. 6).
Wellbeing is also a multidimensional concept (Bureau for Development Policy 2013, p. 16;
Stiglitz, Sen and Fitoussi 2010, p. 14). Sumner and Mallett (2013, p. 675), for example,
present wellbeing along three interrelated dimensions:
 a material dimension, consisting of a person’s living standard and material resources
 a relational dimension, encompassing a person’s relationships and connections within
society that affect what they can do and be, with the resources they have
 a subjective dimension, involving personal values, perceptions and attitudes that affect
how people evaluate particular ways of doing and being.
This broad and multidimensional notion of wellbeing has two important implications. First,
economic resources are an important determinant of wellbeing, in conjunction with many
other factors. Second, wellbeing does not exist in isolation: it is affected by characteristics
of individuals and of society.
A capability model of wellbeing
In considering more specifically how economic inequality might affect wellbeing, this study
uses Sen’s (1992, 1993) capability model. Sen conceptualises wellbeing as a person’s ability
to achieve ways of living that they value. His model depicts how a person’s capability is
related to their economic resources and personal-level and societal-level factors (figure 2.1).
Sen’s approach has been influential in shaping discussion on growth and inequality and on
how economic and social progress is measured (Bureau for Development Policy 2013, p. 25;
McLachlan, Gilfillan and Gordon 2013, p. 7), and departs from traditional approaches that
focus exclusively on the utility derived from consumption (Robeyns 2005, p. 96).
FRAMEWORK AND APPROACH 21
Figure 2.1 Economic resources within a capability model of wellbeing
Source: Adapted from Robeyns (2005, p. 98).
Under the capability model, people convert economic resources into a capability set of
possible outcomes from which people choose a particular combination of realised outcomes.
 Economic resources are the goods and services available as a result of income, in-kind
transfers and non-market production (Robeyns 2005, p. 98).
 The capability set of possible outcomes ranges from the basic, such as achieving adequate
nutrition and physical shelter, to the more complex and intangible, such as having a sense
of control over one’s life, and being socially included (Sen 1993, p. 31). They can include
‘working, resting, being literate, being healthy, being part of a community, being
respected, and so forth’ (Robeyns 2005, p. 95).
 Realised outcomes are a particular combination from the ‘capability set’. They can have
flow-on effects on a person’s future economic resources, and on the resources available
to any dependent children.
The model assumes there is an intrinsic value in being able to make choices (Sen 1992,
p. 41). So, for example, someone who has the means to be adequately nourished, but chooses
to fast, is considered to be better off than someone who is involuntarily under-nourished
(Sen 1992, p. 52).
Economic
resources
(goods and
services that an
individual has at
a point in time)
Capability set
(set of achievable
‘functionings’ or things
that an individual can
‘be’ or ‘do’)
Realised
functionings
(observed
outcomes)
Conversion
factors
Individual
choices
Outcomes affect future resources available to
the individual, and to future generations
Personal conversion
factors
• Personal characteristics
(e.g. physical health)
• Personal values and
preferences
• Family/household
characteristics
Social and environmental conversion
factors
• Social norms and values
• Institutions and policies
• Other people’s behaviour and characteristics
• Environmental conditions (e.g. geographic
location, climate)

22 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE

Personal and social factors influence the extent to which people are able to convert resources
into outcomes.
 Personal conversion factors include a person’s physiological and intellectual
characteristics, skills and knowledge, values and preferences. For example, a person with
a chronic health condition would require a different level of economic resources to
achieve the same outcomes (all else equal) than someone without such a condition, due
to the costs of medical treatment.
 Social conversion factors include social norms and behaviours, social and legal
institutions, public policies, geographic location and climate, environmental amenity,
levels of crime and so on. For example, someone living in a neighbourhood with
relatively high crime could experience lower wellbeing than if they lived in a safer area
(even with the same level of economic resources), due to psychological effects and/or
the costs of personal and home security.
Inequality and wellbeing within a capability model
The capability model depicts a person’s own economic resources as affecting their ability to
achieve valued life outcomes. In addition, it has three attributes that shed light on how
inequality in the distribution of resources could affect wellbeing (without necessarily leading
to a judgment that a particular level of inequality is better or worse than another).
First, the model allows for the possibility that people’s wellbeing is affected by their personal
values and preferences with respect to the distribution of economic resources (via personal
conversion factors). That is, for a given level of resources, people might be able to achieve
a higher or lower level of wellbeing according to whether their perceptions of the societal
distribution of resources aligns with their preferences (Schneider 2016, p. 1727).
 People who prefer economic equality (for example, due to egalitarian values) could be
negatively affected by an increase in societal inequality, even if their own level of
economic resources did not change.
 Conversely, people who prefer relatively higher inequality (for example, due to valuing
the potential for achieving high rewards through effort and risk-taking) could benefit
from increased inequality.
Second, the model includes the possibility that the wellbeing derived from a given level of
economic resources depends on a person’s expectations and aspirations about material living
standards, based on perceptions of others’ living standards. This model attribute makes it
possible for everyone to experience a real increase in economic resources, without a
proportional increase in wellbeing, as people would adjust their expectations accordingly
(Dolan, Peasgood and White 2008, p. 98; Easterlin 1995, p. 41).
Third, the model allows for broader social, cultural, environmental and technological
conditions to affect how a person can use their own economic resources to achieve particular
outcomes. If economic inequality is connected with these societal-level conditions, it could

FRAMEWORK AND APPROACH 23

have flow-on implications for wellbeing. For example, if inequality influences how people
perceive and evaluate themselves and others, it could influence wellbeing by affecting
people’s quality of relationships and social participation (Delhey and Dragolov 2013, p. 154;
Douglas et al. 2014, p. 22; Schneider 2016, p. 1727).
Mobility and wellbeing within a capability model
The capability model also helps in conceptualising the links between economic mobility and
wellbeing.
The model implies that for a given level of economic resources, a person’s access to
opportunities to alter this level of economic resources (in accordance with their own skill,
effort and risk-taking) directly influences their wellbeing. Low observed mobility could
indicate that the outcomes available to someone are constrained by pre-existing factors, such
as their parents’ socioeconomic background (Martinez et al. 2017, pp. 381–2). Such
constraints would represent a direct limitation on people’s ability to influence their own
situation, undermining their wellbeing.
A complicating factor is that a lack of observed mobility could also indicate a relative
stability of people’s resources. Just as a lack of opportunity represents a constraint on
wellbeing, stability and security in one’s level of economic resources could also provide
certainty, contributing positively to people’s wellbeing (this is more likely at the upper end
of the distribution). So, there could be some trade-off between mobility and stability,
particularly as people might differ in their preferences for greater opportunity (and therefore,
risk) relative to certainty.
Disadvantage and wellbeing within a capability model
Finally, the capability model highlights the potential implications of economic disadvantage
for wellbeing. It considers not only a lack of economic resources (poverty) but, more
broadly, a lack of opportunities for social and economic participation, including the inability
to achieve a standard of living that society considers acceptable (box 2.1).
Most obviously, where people have a low absolute level of economic resources, this can
directly undermine the possibilities available to them. This occurs where a person lacks the
resources to achieve basic life outcomes, such as adequate food and shelter.
Where people have a low relative level of economic resources, wellbeing is undermined via
social conversion factors. A person in a developed economy such as Australia might be
relatively well-off by world standards, but if they have a low level of income relative to the
average Australian, this could affect their ability to participate in society and the economy.
For example, if a person cannot afford to join a local sporting club, this could affect their
ability to be socially included, constraining their wellbeing.

24 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE


Box 2.1 What does disadvantage entail besides poverty?
Poverty has traditionally been understood as inadequate economic resources, and this condition
in turn has most commonly been defined as receiving a low income (Scutella, Wilkins and
Kostenko 2013, p. 5). But poverty is only one facet of disadvantage:
Two individuals with the same income can have very different living standards if their income does not
measure adequately all the resources that are available to each of them and/or if their needs differ.
(Fusco, Guio and Marlier 2011, p. 10)
Needs and costs of living vary widely across households and as such, a particular threshold level
of income could be sufficient for most people, but would still leave some with insufficient economic
resources to achieve a particular standard of living, due to their specific circumstances
(McLachlan, Gilfillan and Gordon 2013, p. 36).
Disadvantage is best characterised as a lack of opportunities to participate economically and
socially on a par with one’s peers. While poverty might overlap with other aspects of
disadvantage — such as low capabilities, deprivation or social exclusion — in some cases it will
not represent deep or persistent disadvantage.


Furthermore, where people’s realised outcomes are substantially constrained by their
absolute or relative level of economic resources, this could have flow-on effects for their
future economic resources, and those of their children (box 2.2). Through such channels,
poverty — and other forms of disadvantage — can have longer-term effects on family
wellbeing.

Box 2.2 Disadvantage can affect future wellbeing
Economic disadvantage can have a range of dynamic effects, where a lack of resources in one
period affects a person’s future resources and the resources (and opportunities) available to their
children. For example:
 people with few resources can suffer poor physical and mental health, which undermines
wellbeing directly and can also affect access to education and employment opportunities
 a lack of goods and services that are typically available to the average person (such as
communications technologies, clothing for attending job interviews, or access to transport)
could also constrain a person’s employment opportunities, affecting current and future
resources
 inadequate resources could affect children’s social and educational opportunities — such as
by affecting their ability to engage in school or to participate in extra-curricular activities
 being unable to deal with unexpected events could be harmful to future material resources
and, therefore, wellbeing — for instance, a lack of insurance, savings or access to credit leaves
people vulnerable to a sudden drop in income or an unexpected event (such as a broken
appliance, medical emergency or car accident).
FRAMEWORK AND APPROACH 25
2.2 Operationalising the conceptual framework
From concepts to measures
Sen (1992, p. 52) acknowledged that a person’s capability set is not directly observable, as
it encompasses all possible combinations of outcomes that a person could have chosen, but
did not. Even obtaining data on achieved outcomes is challenging, especially where these
are intangible (such as social participation). It is easier to obtain measures of economic
resources, and how these resources are distributed.
Accordingly, the main focus of this study is to assess the distribution of three measures of
economic resources — income, consumption and wealth — and how those distributions vary
over time and across groups. Income, consumption and wealth are also linked in an
accounting sense (figure 2.2), and their individual components shed light on how people
convert the stock and flow of resources into wellbeing.

Figure 2.2 Income, consumption and wealth affect current and future
wellbeing
Income
Income is the most commonly used measure of inequality, reflecting the relative frequency
with which income data are available.
Resources
available to
support future
wellbeing
Resources used
to support current
wellbeing
Financial
resources
available
to fund
Contributes
to
Net wealth
(start of period)
Household
disposable
income
Private
household
consumption
In-kind transfers
(e.g. public health,
education)
+
+
Net wealth
(end of period)
+
Future
disposable
income
Future
in-kind
transfers
+
Household
saving/dissaving

26 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE
Disposable income (private income net of taxes and cash transfers) strongly influences the
capacity for private consumption (current wellbeing). It can also be saved and invested,
contributing to a person’s wealth and, in turn, supporting their future economic
wellbeing (figure 2.2). Income is also used as a proxy for measuring disadvantage (using
absolute and relative income measures). Moreover, given that income affects people’s future
economic resources and those of their dependent children, it is likely to be a determinant of
intergenerational economic mobility.
Nonetheless, using income alone for analysing people’s economic wellbeing has its
limitations. It does not directly measure the value of goods and services that people actually
consume (as people might consume more or less than their current income). Nor does it
account for the stock of other economic resources — that is, wealth — that might support
wellbeing (through, for example, drawing down savings or borrowing against assets).
Consumption
Consumption is the measure of economic resources that contributes most directly to
wellbeing. This is because it measures the value of goods and services that people consume,
not the income they earn. If income and consumption were the same this would not matter,
but typically there is a gap between people’s income and their consumption in a given period.
The gap reflects people choosing their level of consumption based on their expected lifetime
income, rather than their current income (PC 2015b, p. 15). Students and young families
often consume more than they earn and finance the difference through borrowing.
Middle-aged families often earn more than they consume and accumulate wealth, and
retirees typically spend more than they earn and draw down their wealth.
Household consumption also includes two components that are important for people’s
current economic wellbeing: imputed rent and in-kind government transfers.
 Imputed rent captures the consumption value that people derive from living in their own
home, rather than paying market rent for that same home. It is calculated based on the
estimated value of the property.
 In-kind transfers capture the value of government services (such as public health and
education) that a person uses.
Incorporating these two components has a significant equalising effect and provides a clearer
view of economic inequality (particularly given age-related patterns in home ownership and
in the use of government services) that would not be captured by examining income patterns
alone. Furthermore, consumption can be a more accurate indicator of current economic
resources than income and wealth for particular groups, including unincorporated business
owners and people on very low incomes.

FRAMEWORK AND APPROACH 27

That said, even consumption is not a complete measure of household wellbeing. Due to data
limitations, some aspects of household wellbeing are not captured by our estimates,
including goods produced within the home and leisure time.
Wealth
The distribution of wealth is important because economic wellbeing depends not only on the
goods and services people consume today, but on their consumption possibilities over time.
As noted above, wealth enables people to ‘smooth’ their consumption over their life through
saving, borrowing and drawing on their assets. As such, people with low income but
substantial net wealth can use their wealth to achieve a higher level of economic wellbeing
than their current income would otherwise permit.
The distribution of wealth thus complements income and consumption measures by offering
an indicator of both current and future economic inequality. Wealth is also a key determinant
of intergenerational economic mobility (through bequests or by funding children’s
education) and can contribute to people’s wellbeing in its own right by providing a sense of
financial security and social prestige.
This study’s measure of wealth includes accrued capital gains.
2.3 Our analytical approach
The ‘equivalised’ household is the unit of observation
The analysis in this study is based on equivalised household measures of income,
consumption and wealth.3 Equivalised measures of economic resources adjust for household
size and composition. They account for larger households needing more resources to achieve
the same standard of living as a smaller household, and for some ‘economies of scale’ when
sharing living costs. (For example, two people living alone in two separate households
typically incur higher living costs in total than two people living in one household.)
The process of equivalisation involves adjusting household-level variables for differences
in household composition (ABS 2013b, p. 1). The formula used for equivalisation is the
ABS ‘OECD-modified equivalence scale’. It involves allocating points to each
household member:
 1 point for the first adult
 0.5 points for each additional person aged 15 years or older
 0.3 points for each child aged under 15 years.

3 In earlier Commission analysis, Greenville, Pobke and Rogers (2013) also reported individual distributions;
it is not possible to directly compare these with household distributions.

28 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE

Total household income is divided by the sum of these points to yield the equivalised income
(ABS 2013b, p. 1). Each person in a household is then allocated an equivalised income,
consumption and wealth (figure 2.3).

Figure 2.3 From household income to equivalised income:
a stylised example
The chosen measures of economic resources
The analysis primarily uses broad measures of income, consumption and wealth (figure 2.4).
However, sometimes more narrowly defined measures are used — for example, to gain a
top-down perspective on how Australia’s tax and benefit system bear on measures of
inequality.
More specifically, we use inflation-adjusted4 measures of:
 household disposable income, which includes labour, capital and other income, as well
as the effect of income tax and government cash transfer payments (such as Newstart,

4 All results are presented in 2016-17 dollars, adjusted using the national Consumer Price Index (CPI). This is
the standard approach employed in inequality research to account for changes in purchasing power over time
and it is the only approach that allows for international comparisons of inequality measures. A potential
Household
disposable income
= $60 000
Equivalised
disposable income
= $60 000
Multi-person householdSingle person household
Equivalisation factor:
1 adult = 1 point
Household disposable
income = $105 000
Equivalised disposable income =
$105 000 divided by 2.1 = $50 000
Equivalisation factor:
1 point for 1st adult + 0.5 points for 2nd adult
+ 0.3 points per child = 2.1 points
Equivalised disposable
income = $60 000
Equivalised disposable
income = $50 000
Equivalised disposable
income = $50 000
Equivalised disposable
income = $50 000
Equivalised disposable
income = $50 000
1 data point 4 data
points

FRAMEWORK AND APPROACH 29

Family Tax Benefit and the Age Pension). It does not account for indirect taxes (such
as GST) or in-kind transfers (the provision of government services), including childcare
subsidies
 household consumption, which includes private expenditure on goods and services
(including consumer durables, such as vehicles and household appliances);5 imputed rent
(an estimate of the housing amenity enjoyed by owner-occupiers);6 and in-kind transfers
(the value of government services used by a household, such as health, education and
childcare). In this study, consumption excludes income tax, the principal component of
mortgage repayments, other capital housing costs, superannuation and life insurance.
Figure 2.4 Chosen measures of income, consumption and wealth
shortcoming of this approach is that the CPI uses a single fixed basket of goods and services for all
households — in reality, each household spends its income on a different mix of goods and services. However,
preliminary calculations by the Productivity Commission indicate that using a custom basket of goods and
services for each household has little impact on overall trends in income growth and income inequality.
5 Ideally, consumption would also incorporate a measure of the flow of benefits that arise from using consumer
durables. One way to achieve this would be to calculate an imputed service flow based on the value of
household durables. This has not been possible due to limited data availability.
6 Imputed rent refers to the consumption value that homeowners obtain from the property in which they live. It
is usually calculated as the value the homeowner would have to pay to a landlord if they rented rather than
owned their residence. In this study, imputed rent for homeowners is calculated as 5 per cent of the estimated
sale price of their residence, following Dollman et al. (2015, p. 7). Some studies include imputed rent in
income as well as consumption. In this study we have excluded imputed rent from our income measure.
Excluding imputed rent from income was necessary to allow for international comparisons and for a measure
better aligned with the common understanding of income.
Income
Labour income
(wages, salaries)
Capital income
(dividends, rental
income)
Transfer
payments
(e.g. Newstart,
Age Pension,
Family Tax
Benefit)
less Income tax
Household
disposable income
Household
expenditure on goods
and services (including
consumer durables
such as appliances and
vehicles)
Imputed rent on
owner-occupied
housing
Private consumption
Public consumption
In-kind transfers (e.g.
health, education,
childcare)
Consumption Wealth
Household assets
less Household liabilities
Business wealth (net value of
business assets)
Financial wealth (bank
accounts, shares, bonds, other
financial investments, less
investment loans)
Property wealth (value of
housing less housing debt)
Superannuation balance
Personal wealth (home
contents less student, credit
card and personal debt)

30 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE
Household net wealth (sometimes called net worth) is measured as the excess of total
household assets (including superannuation) over total household liabilities.
Measuring inequality, economic mobility, and economic disadvantage
Indicators of inequality
Summary and quantile-based indicators are the main measures used to describe the
distribution of income, consumption and wealth.
A summary indicator depicts inequality over the whole distribution. The Gini coefficient is
the most commonly used summary indicator, taking a value between 0 and 1 (figure 2.5). A
zero value indicates perfect equality (all people have the same income) and a value of 1
indicates perfect inequality (one person has all the income).

Figure 2.5 Visualising Gini coefficients and quantile-based indicators
Consider a population of individuals ranked by income:
To calculate the Gini coefficient …
To calculate average income by decile … To calculate average income by quintile …
$0 $20,000 $40,000 $60,000 $80,000 $100,000 $120,000 $140,000 $160,000 $180,000
Highest incomeLowest income
... imagine each
person's height
is equal to their
income ...
... then stack
each person,
one by one, so
the furthest right
column includes
every person ...
... the curve of the
columns is the
Lorenz curve. With
perfect equality
it would be
a 45
line ...
... then the Gini
coefficient =
A
B
Cumulative share of people 
C
um
ul
at
iv
e
sh
ar
e
of
in
co
m
e

... take the ranking
above and arrange
people into ten
equal groups ...
1 2 3 4 5 6 7 8 9 10
... average income
by decile is just the
average height of
each group.
Income decile
In
co
m
e

... take the ranking
above and arrange
people into five equal
groups ...
1 2 3 4 5
... average income by
quintile is the average
height of each group.
Income quintile
In
co
m
e


FRAMEWORK AND APPROACH 31
The Gini coefficient has the advantage of being a single statistic that summarises the extent
of inequality for the whole distribution. Gini coefficients are also widely used in the
literature, providing a basis for comparison with other studies and across countries.
However, the Gini coefficient does not indicate where in the distribution any changes in
inequality have occurred.
Quantile-based indicators refer to groups within the distribution. They are formed by ranking
all observations from smallest to largest and then dividing these into equal-sized groups. For
example, these could be quintiles (five equal groups), deciles (10 equal groups) or
percentiles (100 equal groups).
So that each quantile contains the same number of people, the span of quantile ranges varies
greatly (figure 2.2). Quantiles (specifically deciles) are used significantly in this study, in
chapters 3, 4 and 5. There is also brief reference to the top 1 per cent, which has garnered a
lot of interest and is used by others in highlighting inequality.
Quantiles can be used to calculate shares, such as the share of wealth held by the top decile
of the wealth distribution. They can also be used to calculate quantile ratios, which indicate
the relative gap between two points in the distribution. For example, the P90/P10 ratio
represents the ratio of the upper value of the 90th percentile (9th decile) to the upper value
of the 10th percentile (bottom decile) (ABS 2013b, p. 3).
While quantile ratios can be more intuitive than the Gini coefficient, there are drawbacks.
They do not capture changes in the extremes of the distribution (for example, within the
top or bottom decile). Quantile ratios can also be very volatile if the underlying data are
volatile. For example, if the wealth held by the person at the 10th percentile doubles —
say from $5000 to $10 000 — then the P90/P10 ratio will halve (assuming there is no
‘shuffling’ of the income ranking, and that the wealth held by the person at the 90th
percentile does not change).
In this study, distributional patterns in income, consumption and wealth are primarily
presented using decile charts. These show the absolute levels, absolute change or percentage
changes in the average values of these variables for each decile. This metric has been chosen
to aid reader interpretation and enable more detailed comparison across groups. That said,
where appropriate we also report Gini coefficients, quantile ratios and average and median
values, as it is possible for different metrics to reveal a different take on the level or trends
in inequality (box 2.3).
Indicators of economic mobility
Indicators of economic mobility gauge the extent to which people move across a distribution
of economic resources over time. As noted in chapter 1, they seek to depict either
intergenerational mobility (the extent to which someone’s relative position in the economic
distribution relates to the position of earlier generations of their family), or life course
(intragenerational) mobility over shorter periods of time.

32 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE
Figure 2.6 Decile ranges vary greatly
Equivalised disposable income and equivalised wealth distributions, and decile
ranges, 2015-16a
Income

Dollars
Wealth

Dollars

a 2016-17 dollars. Ranges are rounded to the nearest one thousand. Income bins are $2000 wide, wealth
bins are $20 000 wide. The charts do not include those with very high or very low wealth or income.
Source: Productivity Commission estimates using ABS (Microdata: Household Expenditure, Income and
Housing, 2015-16, Cat. no. 6540.0, released 25/10/17).

0
200,000
400,000
600,000
800,000
1,000,000
1,200,000
1,400,000
$0 $50,000 $100,000 $150,000
Dollars
N
u
m
b
e
r
o
f
p
e
o
p
le
Top decile: [$90,000 plus]
9th: [$73,000 – $90,000]
Bottom decile:
[up to $23,000]
2nd: [$23,000 – $28,000]
3rd: [$28,000 – $34,000]
4th: [$34,000 – $39,000]
5th: [$39,000 – $46,000]
6th: [$46,000 – $52,000]
8th: [$62,000 – $73,000]
7th: [$52,000 – $62,000]
Dollars
N
um
be
r
of
p
eo
pl
e
0
200,000
400,000
600,000
800,000
1,000,000
1,200,000
1,400,000
$0 $500,000 $1,000,000 $1,500,000 $2,000,000
Dollars
N
u
m
b
e
r
o
f
p
e
o
p
le
Top decile: [$1,135,000 plus]
9th: [$746,000 – $1,135,000]
Bottom decile:
[up to $28,000]
2nd: [$28,000 – $70,000]
3rd: [$70,000 – $134,000]
4th: [$134,000 – $210,000]
5th: [$210,000 – $305,000]
6th: [$305,000 – $408,000]
8th: [$544,000 – $746,000]
7th: [$408,000 – $544,000]
Dollars
N
um
be
r
of
p
eo
pl
e

FRAMEWORK AND APPROACH 33


Box 2.3 Different indicators of inequality can lead to different
conclusions
The choice of indicator(s), and the way in which indicators are used and reported, can lead to
different findings about inequality. This is because indicators measure different aspects of the
distribution. For example, in the figure below, the Gini coefficient depicting the whole distribution
suggests an increase in income inequality from 1988-89 to 2015-16, whereas the P50/P10 ratio
suggests little overall change in income inequality for the same period.

a Based on equivalised disposable income, 2016-17 dollars. Index (1988-89 = 100).
Sources: Productivity Commission estimates using: ABS (Microdata: Household Expenditure, Income and
Housing, 2015-16, Cat. no. 6540.0, released 25/10/17); ABS HES Basic confidentialised unit record files
for years 1988-89 through 2009-10 as available at 25/10/17; ABS SIH Basic confidentialised unit record
files for years 1993-94 through 2013-14 as available at 25/10/17.
A commonly used indicator of intergenerational mobility is the relationship between fathers’
and sons’ earnings (at the same age), measured as an intergenerational earnings elasticity.
The higher the value of this elasticity, the more that knowing a parent’s place in the earnings
distribution will tell us about where we can expect the child’s place to be. And the lower the
value, the less ‘stickiness’, so that a parent’s relative earnings are a weak predictor of the
child’s ‘rung’ on the earnings ladder of the next generation (Corak 2013, p. 81).
Measuring economic mobility requires longitudinal data to compare outcomes for the same
individuals over time. There are few longitudinal databases in Australia, and those that do
exist have only been in operation for comparatively short periods of time, making it difficult
to build a comprehensive picture of mobility. For this reason, this study does not present
original estimates of intergenerational mobility.
Instead, the direction and magnitude of shifts across income, consumption and wealth deciles
are used to report on life course mobility, presented using decile shift matrices and ribbon
charts. Decile shift matrices show the proportion of people in each decile who moved to each
of the other deciles (and the proportion who did not change deciles) and ribbon charts
represent the decile shift matrices in a visual manner.
90
95
100
105
110
115
120
In
d
ex
Year
Gini coefficient
P50/P10

34 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE
A commonly used indicator of intergenerational mobility is the relationship between fathers’
and sons’ earnings (at the same age), measured as an intergenerational earnings elasticity.
This approach provides richer and more information on the specific transitions made by
people at all parts of the distribution. Demographic analysis is also used to illustrate the
effect of life events on observed mobility.
Indicators of economic disadvantage
There is an even greater diversity of measures to analyse and report on economic
disadvantage (box 2.4).
As with inequality, the choice of indicator can critically influence findings about the extent
of disadvantage. For example, analysing the prevalence of poverty using an income-only
measure could classify people as poor who have temporarily low incomes but high net wealth
(such as business owners experiencing a loss in one year: ABS 2013a, p. 3), thus overstating
the extent of poverty.
This problem can be partly addressed by including some measure of both income and wealth
in a poverty measure. That said, measures of economic resources could also give a more
limited picture of broader disadvantage, compared with multidimensional indicators that
incorporate people’s level of access to educational, social and employment opportunities.
The use of broader indicators helps to avoid misclassifying people as poor due to transitory
variations in income (or, conversely, excluding people from classification as poor due to
incomes above the poverty threshold, despite those incomes being insufficient to cover some
people’s specific needs).
For this study, a suite of indicators are used to present a detailed picture of the extent of
economic disadvantage in Australia. They include:
 relative poverty — having less than 50 per cent of median income or consumption
 anchored poverty — having current income less than 50 per cent of the median income
in 1988-89 (adjusted for inflation)
 financial poverty — having low levels of income, consumption, and liquid assets
simultaneously
 the poverty ‘gap’ — the average dollar value that would be required to lift a person out
of poverty (by raising their income or consumption above the relevant poverty threshold)
 the persistence of poverty — time spent in poverty, and the extent of poverty recurrence
 material deprivation — being deprived of goods and services that most people agree to
be essential for life, due to a lack of economic resources (relative to household expenses)
 social exclusion — limited access to material resources, as well as limited opportunities
to participate socially, educationally and in employment.
FRAMEWORK AND APPROACH 35
Box 2.4 Indicators of economic disadvantage
Income poverty
Indicators of absolute poverty measure the proportion of people in poverty with reference to a
minimum level of income calculated as necessary for supporting a minimum acceptable standard
of living (Marks 2007, p. 1). In Australia, an absolute poverty line is not calculated, but it is
common to see such indicators used in reference to developing countries — for example, the
international (extreme) poverty line of $US1.90 per person per day (World Bank 2016, p. 3).
Relative poverty thresholds are usually set at a given percentage of median household
disposable income, most commonly 50 per cent (Wilkins 2017, p. 33). This indicator recognises
that in developed economies, disadvantage is increasingly understood to depend in part on the
material living standards of others in society. However, relative poverty indicators can be highly
sensitive to changes in transfer payments and median incomes (ABS 2013a, p. 3). For example,
an increase in relative poverty could indicate that real incomes at the bottom of the distribution
are falling, but could also reflect strong growth in median incomes.
Persistence
To examine the dynamics of people’s experiences of poverty, researchers use indicators such as
the survival rate, which measures the likelihood that an episode of poverty lasts beyond a
particular time period; and the hazard rate, which measures the likelihood that people exit poverty
after a given time period (Azpitarte and Bodsworth, in CEDA 2015, p. 39).
Deprivation
Indicators of material deprivation measure the extent to which people lack access to goods
and services that most people agree to be essential for life, for affordability reasons (Wilkins 2016,
p. 83). For example, Saunders and Wong (2012, p. 38) constructed a deprivation indicator based
on 24 essential items and activities (such as medical treatment if needed, secure locks on doors
and windows, and a substantial meal at least once a day).
Deprivation indicators offer a more direct view of whether people have their basic needs met. It
can also be particularly useful for measuring child poverty, given that resources may not always
be shared equitably within a household (Wilkins 2016, p. 84).
Social exclusion
A broad conceptualisation of disadvantage includes the idea that it is not just a lack of economic
resources or essential goods and services that affect people’s wellbeing, but also a lack of access
to opportunities to ‘fully participate in social and economic life’ (Azpitarte and Bodsworth, in CEDA
2015, p. 38). Social exclusion centres on the latter concept (participation, and engagement with
society more broadly), though it also incorporates income poverty.
In Australia, social exclusion indicators are multidimensional, calculated by summarising
information on people’s level of exclusion across several life domains. For example, the Social
Exclusion Monitor (measured by the Brotherhood of St Laurence and the Melbourne Institute)
uses multiple indicators in each of seven domains (material resources, employment, education
and skills, health and disability, social connection, community and personal safety) to compile a
single summary statistic representing the extent of a person’s social exclusion (Azpitarte and
Bowman 2015, p. 1).
ECONOMIC DISADVANTAGE 107

6 Economic disadvantage
Key points
 Disadvantage is about ‘impoverished lives’, not just ‘depleted wallets’. It encompasses poverty
(low economic resources), material deprivation (an inability to afford the ‘basic essentials of life’),
and social exclusion (an inability to fully participate in the ordinary activities of a community).
 Nine per cent of Australians (2.2 million people) lived below the relative income poverty line (half
of median disposable income) in 2015-16.
– The relative income poverty rate in Australia has fluctuated, but is currently close to its average
level over the past three decades.
– Changes in relative income poverty are often driven by changes in median income. This form
of poverty increased during the mining boom when median income grew strongly.
– Using a different poverty line anchored to the 1988-89 median income in real terms, the rate of
poverty fell from 9 per cent in 1988-89 to 3 per cent (700 000 people) in 2015-16.
 The demographics of poverty reveal that jobless households, particularly those with children,
experience the highest poverty rates. Age-wise, children and older people (65+ years) have
been the most likely to experience both income and consumption poverty.
 The length of time people spend in poverty is as important as the rate of poverty.
– About half of Australians experienced income poverty at some point between 2001 and 2016.
Most of these experiences (79 per cent) lasted less than three years.
– However, a small proportion of people get ‘stuck’ in poverty for extended periods. Six per cent
of poverty spells lasted six years or longer.
 Many people who exit poverty re-enter at a later date. Of those who were in income poverty
in 2001, 30 per cent had returned to (or were still in) income poverty in 2016.
 People in poverty often experience more fluctuations in their incomes than others. Between
2006 and 2016, people below the poverty line experienced more than twice as much income
volatility, year on year, as people above the poverty line.
 Deprivation metrics provide a more accurate reflection of the balance between resources available
and basic needs that have to be met.
– Material deprivation affects a slightly higher proportion of Australians (a little under 12 per cent)
than does income poverty. But the two often do not overlap; many people experience
deprivation without being in poverty, and vice versa.
– Children, lone parents, those with a disability, the unemployed and Indigenous Australians are
most at risk of multiple deprivation.
 Social exclusion metrics are closely related to deprivation, but incorporate a focus on participation
in the economic and social activities of a community. They help us to examine the relationship
between poverty and the characteristics that make it difficult to participate economically.
– The prevalence of marginal social exclusion was relatively steady between 2006 and 2015, but
deep social exclusion showed a small and sustained rise after 2012.
108 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE
6.1 How disadvantage relates to inequality
Disadvantage in Australia is most often conceptualised as synonymous with poverty (low
economic resources, particularly incomes). But this is too restrictive, as disadvantage entails
not only low economic resources and material possessions, but also low capabilities with
which to obtain those economic resources in the future. Or, as Sen (2000, p. 3) observed, it
is about ‘impoverished lives, and not just depleted wallets’.
Poverty should not, therefore, be thought of as equivalent to economic disadvantage. Rather,
poverty is only one of three elements of disadvantage (box 6.1). The other two are
deprivation, being unable to afford to buy items or undertake activities that are widely
regarded in society as things that everyone should have, and social exclusion, being unable
to fully participate in the ordinary activities of a community.
There are several reasons why metrics of disadvantage add an important dimension to the
story of inequality.
 Concerns about inequality may be assuaged if the prevalence of poverty is decreasing, if
poverty spells are mostly temporary, or if opportunities exist for people to improve their
economic positions. It is also possible that increasing inequality could be accompanied
by increasing mobility or decreasing disadvantage (if, for example, all incomes were
rising, but high incomes were growing faster than low incomes, as was the case during
the mining boom).
 Greater inequality might also indicate that more people are at risk, or on the brink, of
poverty. The concept of ‘the precariat’, meaning a large group of people in a precarious
or vulnerable economic position, has garnered significant attention in the past decade
(Standing 2016; Wright 2016).
 Finally, the presence of higher economic inequality might indicate a rationale for
redistribution to alleviate disadvantage and, thereby, to minimise some of the associated
personal and social costs.
Beyond the prevalence of disadvantage, it is also worth considering the costs of
disadvantage, both to the people directly affected and to the community at large. Some of
the costs cannot be measured or avoided, but conceptually it is useful to think about social
costs (costs to the quality of life for the affected individual and others) and economic costs
(such as forgone employment income). These issues are explored further in McLachlan,
Gilfillan and Gordon (2013, p. 149).

ECONOMIC DISADVANTAGE 109
Box 6.1 What does disadvantage entail besides poverty?
Poverty is only one element of disadvantage, as a household’s needs and costs of living vary
widely. A particular threshold level of income could be sufficient for some, but would still leave
others with insufficient economic resources (McLachlan, Gilfillan and Gordon 2013, p. 36),
because:
 they might have special expenditure needs (for example, as a result of sickness or disability),
 they might have high work-related expenses (such as childcare or transport costs), or
 they might face high costs of living (as often occurs in remote areas).
And, equally, for some people a low level of income may not represent significant disadvantage:
Low income … does not automatically imply that disadvantage is present. It is possible, for example, that
some people who have low incomes are simply moving between jobs and are not disadvantaged in any
meaningful way. We then need to look at indicators other than income to decide who is actually
disadvantaged. (Derby, in CEDA 2015, p. 13)
Conceptually, the three main elements that contribute to disadvantage — poverty, material
deprivation, and social exclusion (see figure below) — are related, and often overlap, but it is also
possible for someone to experience only one element at a time (McLachlan, Gilfillan and
Gordon 2013, p. 53).
The prevalence and persistence of disadvantage are the focus of this chapter. We provide
recent estimates on each of the three elements of disadvantage: poverty, material deprivation,
and social exclusion.50 Sections 6.2, 6.3 and 6.4 examine estimates and aspects of poverty

50 This study does not analyse ‘welfare dependence’ (long-term reliance on government transfers as a major
source of income: see, for example, Cobb-Clark et al. 2017). This subject is rapidly becoming one of major
policy interest in analyses of disadvantage — in May 2018, the House of Representatives launched an
Inquiry into Intergenerational Welfare Dependence (SCIWD 2018). In late 2017, the ABS and the
Department of Social Services released the first version of the Australian Priority Investment Approach to
Welfare Research Dataset (ABS, Cat. no. 4490.0.55.001), a longitudinal sample of income support
Social exclusion
Unable to participate
in the normal
economic and social
activities of the
community
Material deprivation
Cannot afford the essentials
for an acceptable standard
of living
Poverty
Low economic resources,
relative to the average
110 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE

rates for income, consumption, and ‘financial’ poverty, and their persistence over time.
Section 6.5 then examines updated estimates of the prevalence of material deprivation in
Australia (and considers how deprivation overlaps with poverty) and section 6.6 does the
same for measures of social exclusion.
6.2 The prevalence of poverty
Measuring poverty
The most common method of measuring poverty is using income, as it is straightforward to
measure and income data are widely collected. The adequacy of a household’s income is
compared to some externally-defined threshold of ‘need’ (box 6.2). However, income is only
one type of economic resource. People may also have access to assets or borrowings, so
poverty metrics can offer more accurate insights by incorporating wealth or financial
resilience information, or by adjusting for major expenses (such as housing costs) to better
assess purchasing power.
This study includes five different measures of poverty, estimated using data from the ABS
Household Expenditure Survey (HES), over the period from 1988-89 to 2015-16. They are:
 relative income poverty
 anchored income poverty (based on the 1988-89 relative poverty threshold)
 relative private consumption poverty
 relative final consumption poverty (though, as noted in chapter 3, our estimates of final
consumption do not take into account indirect taxes) and
 ‘financial’ poverty — a metric combining relative income poverty, relative consumption
poverty, and a minimum level of liquid assets.
Relative income poverty rates are also estimated annually, using data from the Melbourne
Institute Household Income and Labour Dynamics in Australia (HILDA) survey, for the
period 2000-01 to 2015-16. The threshold for relative income and consumption poverty is
set at 50 per cent of the median equivalised household disposable income, or 50 per cent of
the median equivalised household consumption. And a person is defined as being in financial
poverty if they simultaneously fulfil three conditions:
 they have less than 50 per cent of the median income
 they have less than 50 per cent of the median private consumption
 their total liquid assets (cash, bank deposits, and equity, plus superannuation if at least
one person in the household is over the age of 65) are less than three months’ worth of
the equivalised income poverty line for their household size.

recipients currently spanning 14 years. This dataset’s availability offers rich new possibilities for exploring
this facet of disadvantage.

ECONOMIC DISADVANTAGE 111
Box 6.2 Income poverty: relative, absolute or anchored?
Identifying who is living in poverty requires a threshold that separates the disadvantaged from the
rest of the population (McLachlan, Gilfillan and Gordon 2013, p. 32). Indicators of income poverty
use either relative, absolute, or anchored thresholds.
A relative poverty threshold is simple and self-adjusting, but is arbitrary
The relative income poverty approach considers that people are living in poverty if their income is
below a certain percentage of median household income. For example, the main thresholds used
by the OECD are 50 and 60 per cent of median equivalised household income. A relative income
poverty threshold therefore reflects movements in the median standard of living.
This measure is simple to calculate, and self-adjusts to movements in incomes and to differences
in cultural and temporal preferences (unlike an absolute threshold). It also takes into account the
possibility that a low relative level of economic resources can bear on people’s wellbeing
(chapter 2).
However, the selection of a threshold is arbitrary and can produce counterintuitive results.
Saunders, Wong and Bradbury (2016, p. 97) note that a fall in the United Kingdom’s median
income resulted in a lowering of the relative poverty line and a consequent decline in the proportion
of people living in relative poverty — even though deprivation increased as living standards
declined (Fahmy 2014). Conversely, an increase in all incomes could produce a rise in the relative
poverty rate, as was observed in Australia during the mining boom.
An absolute threshold reflects ‘essential needs’, but is complex
Under an absolute income poverty approach, people are considered to be poor if their income is
not sufficient to cover the costs of a basket of ‘necessary’ goods and services, which is updated
as community norms evolve.
This approach focuses on the minimum resources required to live at an acceptable standard,
rather than pegging those needs to the rest of the population. However, as incomes grow, so do
community expectations as to what constitutes an acceptable standard of living
(Phillips et al. 2013, p. 8). Furthermore, absolute poverty thresholds can be complex, as they
depend on the changing prices of hundreds of goods and services and the selection of the basket
is affected by cultural trends and norms (Boarini and d’Ercole 2006, p. 12).
Absolute poverty thresholds are rarely used in the Australian context. However, the concept of a
minimum acceptable level of economic resources has been used in setting minimum wages
(stretching back to the Harvester decision of 1907), and Saunders and Bedford (2017, p. 1) note
that the ‘budget standards’ produced by the University of New South Wales’ Social Policy
Research Centre are used by governments in setting minimum wages and assessing the
adequacy of income support payments.
An anchored threshold lies somewhere in the middle, conceptually
An anchored poverty threshold is one that is not based on a fixed basket of goods and services,
but whose real value (and therefore purchasing power) is held constant over time. Often the
starting value is the relative poverty threshold from a particular year, which is then adjusted only
for inflation, rather than keeping pace with median incomes.
112 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE

How has the prevalence of poverty changed?
Figure 6.1 shows all four measures of relative poverty in Australia. The major features of
each poverty measure over the past 30 years are as follows.
 Income poverty was just over nine per cent in 2015-16, representing about 2.2 million
people. Of the four poverty measures, it fluctuated the most over the period, but
nevertheless remained within a relatively narrow band around an average of roughly
10 per cent.
– There was a prolonged increase between 1993-94 and the late 2000s, peaking at close
to 12 per cent in 2009-10, probably reflecting the period’s strong growth in median
incomes as a result of the mining boom, as well as the impacts of the Global Financial
Crisis (GFC). The post-GFC period has seen income poverty fall by about
one-quarter, and in doing so return to the same level it was in 1988-89.

Figure 6.1 Trends in relative poverty measures
Relative poverty rates, 1988-89 to 2015-16a
a Income and consumption are equivalised as described in chapter 2. For all indicators, estimates of
disposable household income are based on total current weekly household income (including transfers),
less income tax.
Source: Productivity Commission estimates using: ABS (Microdata: Household Expenditure, Income and
Housing, 2015-16, Cat. no. 6540.0, released 25/10/17) and ABS HES Basic confidentialised unit record files
for years 1988-89 through 2009-10 as available at 25/10/17.
 Private consumption poverty is estimated to be slightly higher than income poverty, at
just under 10 per cent of the population, representing about 2.4 million people in 2015-16.
Private consumption poverty, like relative income poverty, was at its lowest in
1993-94 — these results were partly due to low median income and private consumption
0
2
4
6
8
10
12
1988-89 1993-94 1998-99 2003-04 2009-10 2015-16
Year
P
e
r
c
e
n
t
in
p
o
v
e
rt
y Income
Private
consumption
Final
consumption
Financial

ECONOMIC DISADVANTAGE 113

in the wake of the early 1990s recession. Private consumption poverty has risen in every
survey interval since 1993-94.
 Final consumption poverty is much lower, at about 2.7 per cent of the population
(650 000 people) in 2015-16. This measure has showed less movement than income or
private consumption poverty, fluctuating by less than 1 per cent, with a longer-term
average of about 2.5 per cent.
 Financial poverty is lower again, at about 2.4 per cent (580 000 people) in 2015-16.
Prior to 2009-10, financial poverty had increased slightly; interestingly, the financial
poverty rate has tracked the final consumption poverty rate closely over the last three
HES surveys.51
How do the different relative poverty metrics compare?
The gap between income poverty and private consumption poverty fell between 2009-10 and
2015-16, a period during which incomes in the bottom decile grew at a faster rate than any
other decile.
The difference between private and final consumption poverty was significant over the entire
time period and has increased steadily over the past 25 years (in 2016, the gap was about
seven percentage points, representing just under 1.7 million people). This suggests that
in-kind transfers are relatively effective at alleviating consumption poverty by helping to lift
many poor households — but not all — above the consumption poverty line.52
The difference between the low rate of financial poverty and the roughly one-tenth of people
living in income poverty may indicate that many of those with low incomes (such as age
pensioners and other retirees) have access to borrowings or assets with which to fund their
consumption.
As Wilkins (2017, p. 34) notes, one reason why income poverty estimates fluctuate more
than consumption poverty estimates is because many income support recipients in Australia
have incomes close to 50 per cent of the median income, so that relatively small changes to
government transfers can bring about sizable movements in the poverty rate (see figure 6.4,
below, for more detail).

51 Some or all of the above estimates are likely to slightly understate the true prevalence of poverty, because
surveys often under-sample the most disadvantaged Australians. Section 6.4 provides more detail.
52 Note that households in final consumption poverty are not necessarily also in private consumption poverty.
A household may have private consumption above the poverty line, but simultaneously consume no value
(or very little value) via in-kind transfers. Regardless of the degree of overlap between the two cohorts,
though, it stands to reason that for every household experiencing private consumption poverty but not final
consumption poverty, something — such as in-kind transfers — is lifting that household above the final
consumption poverty line.
114 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE
Anchored poverty has fallen significantly as real incomes have grown
The anchored income poverty rate is calculated based on a threshold set at 50 per cent of the
median equivalised household disposable income in 1988-89, with this threshold inflated
according to the Consumer Price Index (figure 6.2).
Anchored income poverty fell substantially, from 9.1 per cent in 1988-89 to 2.9 per cent in
2015-16. This most recent estimate represents about 700 000 people who have an equivalised
disposable household income of less than $13 900 (the 1988-89 relative poverty line).
The sustained fall in anchored poverty sits in contrast to the relative poverty rates presented
in figure 6.1. One of the major reasons why the relative income poverty rate has not fallen
significantly below its long-term average of roughly 9 per cent is because median incomes
have grown substantially in real terms over the period (chapter 3, figure 3.4).
Figure 6.2 Anchored poverty is less than one-third of its initial level
Anchored income poverty rate, 1988-89 to 2015-16a
a Incomes are equivalised as described in chapter 2. Estimates of disposable household income are based
on total current weekly household income (including transfers), less income tax.
Source: Productivity Commission estimates using: ABS (Microdata: Household Expenditure, Income and
Housing, 2015-16, Cat. no. 6540.0, released 25/10/17) and ABS HES Basic confidentialised unit record files
for years 1988-89 through 2009-10 as available at 25/10/17.
Poverty gap measures have been broadly steady
The poverty gap indicates the difference between a poverty threshold and someone’s income
or consumption. This indicator is often used to examine the ‘depth’ of poverty, with a larger
gap representing deeper poverty. We have estimated the average (population-wide) poverty
0
2
4
6
8
10
1988-89 1993-94 1998-99 2003-04 2009-10 2015-16
Year
P
e
r
c
e
n
t
in
p
o
v
e
rt
y
Anchor point: 1988-89
relative poverty

ECONOMIC DISADVANTAGE 115

gaps from 1988-89 (or 1993-94) to 2015-16, in 2016-17 dollars and as a percentage of the
relative poverty lines for income, private consumption and final consumption (figure 6.3).
While the relative income poverty gap has trended upwards over the past decade in real
dollar values, the income and consumption poverty gaps as a percentage of the poverty line
have remained relatively stable. Of note is the spike in the income poverty gap at the time of
the early 1990s recession, after which the poverty gap took close to a decade to return to its
pre-recession level. (The difference between this spike and the decrease in the prevalence of
income poverty at the same time highlights the manner in which relative poverty rates can
sometimes reflect movements in the median wage rather than in actual standards of living.)
The absence of a similar spike in the income poverty gap at the time of the GFC suggests it
had a far smaller impact on the depth of poverty than the 1990s recession.
Figure 6.3 Income poverty gaps have seen the greatest fluctuations
Average relative poverty gaps, 1988-89 to 2015-16
a Dollars are constant 2016-17 dollars.
Source: Productivity Commission estimates using: ABS (Microdata: Household Expenditure, Income and
Housing, 2015-16, Cat. no. 6540.0, released 25/10/17) and ABS HES Basic confidentialised unit record files
for years 1988-89 through 2009-10 as available at 25/10/17.
The persistent difference between the (percentage) poverty gaps for private and final
consumption — about one-third the size of the final consumption poverty gap — suggests
that a reasonable proportion of people in private consumption poverty utilise in-kind
transfers such as Medicare, public schools or public housing, thereby reducing the depth of
their consumption poverty (though, as noted above, there is not necessarily any overlap
between the two cohorts).
0
2,500
5,000
7,500
10,000
1988-89 2003-04 2015-16
Year
D
o
ll
a
rs
0
10
20
30
40
50
60
70
1988-89 2003-04 2015-16
Year
P
e
r
c
e
n
t
Income
Private
consumption
Final
consumption
Income
Private
consumption
Final
consumption
- 15-16
Year Year
D
ol
la
rs
a
116 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE
Long survey intervals can hide sizable fluctuations in the poverty rate
One shortcoming of the poverty rate analysis presented so far is the length of time (four to
six years) between surveys. There are significant year-on-year variations observable in our
estimate of annual poverty rates according to HILDA data (figure 6.4), which are not
apparent in the HES poverty rates shown above.
Figure 6.4 HILDA and SIH show more detail, compared to HES, of
year-on-year movements in poverty rates
Annual relative income poverty rates, 2000-01 to 2015-16a,b
a HILDA poverty rate is based on equivalised disposable income. b ACOSS poverty rates are calculated on
an after-housing costs basis and using SIH data from 2003-04 to 2013-14 inclusive. Where ACOSS has
calculated multiple poverty rates for one year to reflect changing definitions of income in the SIH, we have
shown the estimate that uses the most recent definition of income. Identical result for 2003-04 is a
coincidence.
Sources: ACOSS (2016, p. 17); Productivity Commission estimates based on Melbourne Institute
(Household, Income and Labour Dynamics in Australia (HILDA) Survey, Release 16).
There are many other measures of poverty that utilise a range of different methodologies and
datasets. For example, the Australian Council of Social Services (ACOSS) uses data from
the ABS Survey of Income and Housing (SIH) to estimate ‘after-housing’ poverty rates
(relative income poverty after accounting for housing costs), also presented in figure 6.4.
The most recent update of the ACOSS poverty measure estimated that 13.3 per cent of
people, or just under 3 million, were living in after-housing poverty in 2013-14 (ACOSS
2016, p. 11).
While the levels of the two poverty rate estimates have often been quite different (given that
housing costs do not necessarily move in tandem with median incomes, and taking into
account that the estimates use two different data sources), there are some consistent trends.
Both estimates show an increase in relative income poverty in, or shortly after, 2006-07,
0
2
4
6
8
10
12
14
16
2000-01 2002-03 2004-05 2006-07 2008-09 2010-11 2012-13 2014-15
P
er
c
en
t
in
p
o
ve
rt
y
Year
ACOSS (SIH)
HILDA

ECONOMIC DISADVANTAGE 117

a year that saw significant restrictions introduced around eligibility for Parenting Payments
(Brady and Cook 2015, pp. 2, 19; Summerfield et al. 2010, p. 74).53 Recalling that many
income support recipients have incomes quite close to the poverty line, these eligibility
changes may have been a major driver of this spike in the poverty rates shown. This was
followed by a significant decrease in both relative poverty rates, in part reflecting slowing
growth in median incomes in the period after the GFC (chapter 3, figure 3.6).
6.3 The demographics of poverty
Not all demographic groups are equally likely to be in poverty. Rates of poverty are much
higher among some household types and some age groups than others. This section discusses
these differences as well as trends in rates of poverty over time for particular demographic
groups.54
Who was poor in 2015-16?
Poverty rates vary significantly between household types (figure 6.5). Not surprisingly, both
income and private consumption poverty are highest for working-age households in which
no person has paid work (with the major source of income for those households likely to be
public transfers, such as Austudy, Newstart, Parenting Payments or the Disability Support
Pension).
Jobless families (households with at least one child under the age of 15 and no paid work)
experience the highest poverty rates, followed by jobless child-free households.55 Private
consumption poverty is lower than income poverty for both types of jobless households,
suggesting that some jobless households may have access to credit (or financial assistance
from family or friends) with which to lift their consumption.

53 Some may attribute this to measurement error. While errors are possible, it is unlikely that the same error
would occur across two different datasets (SIH and HILDA) with different time intervals.
54 Results presented in this section should be treated with caution. General trends in income poverty rates for
demographic groups are mostly consistent across data sources, as are the rankings of demographic groups
by income poverty rate. But specific poverty rates for individual demographic groups sometimes differ
considerably depending on whether the source dataset is HES or HILDA. The largest discrepancy is
apparent in the poverty rate for people aged over 65 — in 2015-16, HES data indicate a poverty rate
of 9 per cent, whereas HILDA data indicate a poverty rate of 22 per cent. These differences likely reflect a
number of factors including sampling error, slight differences in median income translating to differences
in poverty thresholds, and differences in the capture of income for low-income people (particularly retirees).
55 Note that joblessness is a broader categorisation than unemployment, as it also includes people who are not in
the labour force (for example, because of disability, illness or caring responsibilities) but are of working age.
118 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE
Figure 6.5 Poverty rates by household type, 2015-16a

a The general ranking of poverty rates by household type is consistent with HILDA estimates, but specific
poverty rates differ.
Source: Productivity Commission estimates using ABS (Microdata: Household Expenditure, Income and
Housing, 2015-16, Cat. no. 6540.0, released 25/10/17).
Retirees also experience higher poverty levels than working households, but less than
unemployed households, perhaps reflecting a combination of superannuation income and
the higher rates of the Age Pension compared to the Newstart Allowance (DHS 2018a,
pp. 13, 28). In contrast to unemployed working-age households, income poverty for
retirees was lower than private consumption poverty, suggesting that some retirees are still
saving, as discussed in chapter 4.
In general, income poverty levels reflect the household type composition of income deciles
(chapter 3, figure 3.17), where jobless households and retirees are disproportionately
represented in the bottom two income deciles.
Household poverty trends fluctuate significantly
While the rankings of poverty rates by household type have been relatively consistent,
individual household types show far more variation over time (figure 6.6).
0
10
20
30
40
50
60
Family, at least
one in paid work
Family, no
paid work
Working age,
in paid work
Working age,
no paid work
Retiree
P
er
c
en
t i
n
po
ve
rt
y
Household type
Income poverty Consumption poverty
Private consumption poverty
Income poverty
Household type
P
er
c
en
t i
n
po
ve
rt
y


ECONOMIC DISADVANTAGE 119
Figure 6.6 Poverty rates by household type
1993-94 to 2015-16a
a General trends are mostly consistent with HILDA estimates, but specific poverty rates differ.
Source: Productivity Commission estimates using: ABS (Microdata: Household Expenditure, Income and
Housing, 2015-16, Cat. no. 6540.0, released 25/10/17) and ABS HES Basic confidentialised unit record files
for years 1988-89 through 2009-10 as available at 25/10/17.
The greatest variation is apparent in the poverty rates of jobless families. After a sharp initial
decline in the early 1990s, income poverty grew rapidly among jobless families, almost
tripling between 1993-94 and 2009-10. In 2015-16, the private consumption poverty rate for
jobless families reached a new high, while the income poverty rate for jobless families was
just below the peak reached in 2009-10. Poverty also increased a great deal among
working-age people without paid work — particularly income poverty. This may reflect
income support payments (which are indexed to the Consumer Price Index, and therefore do
not increase in real terms) growing more slowly than median incomes.
The sharp drop in the rate of income poverty among retirees after 2009-10 is not reflected in
their private consumption poverty rate. This suggests that small increases in the Age Pension
during this timeframe may have pushed large numbers of retirees from just below the income
poverty threshold to just above it, without substantially increasing their purchasing power.
Children and older people are the most likely to be in poverty
Income and consumption poverty exhibit a handful of patterns with regard to age groups
(figures 6.7 and 6.8), which generally reflect the household types discussed above.
Income Private consumption
'88-89 '93-94 '98-99 '03-04 '09-10 '15-16 '88-89 '93-94 '98-99 '03-04 '09-10 '15-16
0
10
20
30
40
50
60
Year
P
e
r
c
e
n
t
in
p
o
v
e
rt
y
I rivate consu ption
' ' - ' - ' - ' -04 '09-10 '15-16
P
e
r
c
e
n
t
in
p
o
v
e
rt
y
i i Income
'8 ' ' ' - '88-89 '93-94 '98-99 '03-04 '09-10 '15-16
0
10
20
30
40
50
60
Year
P
e
r
c
e
n
t
in
p
o
v
e
rt
y
19 8-89 93-94 98-99 03-04 09-10 20 5-16
Private consumption
19 93-94 98- 9 03-04 09-10 20 5-16
Year
P
e
r
c
e
n
t
in
p
o
v
e
rt
y
Family, no
paid work
Working
age, no
paid work
Retiree
Family, at least one
in paid work
Working
age, in
paid
work
Income
60
50
40
30
20
Private consumption Income
' ' ' - ' - ' -99 '03-04 '09-10 '15-16
P
e
r
c
e
n
t
in
p
o
v
e
rt
y
60
50
40
30
20
10
0
120 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE
Figure 6.7 Income poverty rates by age groupa
1988-89 to 2015-16
a General trends across age groups are mostly consistent with HILDA estimates, but group rankings and
specific poverty rates differ.
Source: Productivity Commission estimates using: ABS (Microdata: Household Expenditure, Income and
Housing, 2015-16, Cat. no. 6540.0, released 25/10/17) and ABS HES Basic confidentialised unit record files
for years 1988-89 through 2009-10 as available at 25/10/17.
Figure 6.8 Consumption poverty rates by age groupa
1993-94 to 2015-16
a These estimates have not been compared to HILDA estimates because equivalent variables are
unavailable in HILDA.
Source: Productivity Commission estimates using: ABS (Microdata: Household Expenditure, Income and
Housing, 2015-16, Cat. no. 6540.0, released 25/10/17) and ABS HES Basic confidentialised unit record files
for years 1988-89 through 2009-10 as available at 25/10/17.

0
5
10
15
20
25
1/
01
/1
98
7
1/
01
/1
98
8
1/
01
/1
98
9
1/
01
/1
99
0
1/
01
/1
99
1
1/
01
/1
99
2
1/
01
/1
99
3
1/
01
/1
99
4
1/
01
/1
99
5
1/
01
/1
99
6
1/
01
/1
99
7
1/
01
/1
99
8
1/
01
/1
99
9
1/
01
/2
00
0
1/
01
/2
00
1
1/
01
/2
00
2
1/
01
/2
00
3
1/
01
/2
00
4
1/
01
/2
00
5
1/
01
/2
00
6
1/
01
/2
00
7
1/
01
/2
00
8
1/
01
/2
00
9
1/
01
/2
01
0
1/
01
/2
01
1
1/
01
/2
01
2
1/
01
/2
01
3
1/
01
/2
01
4
1/
01
/2
01
5
1/
01
/2
01
6
1/
01
/2
01
7
1/
01
/2
01
8
1988-89 93-94 98-99 03-04 09-10 2015-16
0
5
10
15
20
25
1/
01
/1
98
7
1/
01
/1
98
8
1/
01
/1
98
9
1/
01
/1
99
0
1/
01
/1
99
1
1/
01
/1
99
2
1/
01
/1
99
3
1/
01
/1
99
4
1/
01
/1
99
5
1/
01
/1
99
6
1/
01
/1
99
7
1/
01
/1
99
8
1/
01
/1
99
9
1/
01
/2
00
0
1/
01
/2
00
1
1/
01
/2
00
2
1/
01
/2
00
3
1/
01
/2
00
4
1/
01
/2
00
5
1/
01
/2
00
6
1/
01
/2
00
7
1/
01
/2
00
8
1/
01
/2
00
9
1/
01
/2
01
0
1/
01
/2
01
1
1/
01
/2
01
2
1/
01
/2
01
3
1/
01
/2
01
4
1/
01
/2
01
5
1/
01
/2
01
6
1/
01
/2
01
7
1/
01
/2
01
8
1988-89 93-94 98-99 03-04 09-10 2015- 6
P
er
c
en
t i
n
po
ve
rt
y
Year
25–34
Under 15
years
15–2435–44
65+
years
55–64
45–54
0
5
10
15
20
25
1993-94 98-99 03-04 09-10 2015-16
0
5
10
15
20
25
1993-94 98-99 03-04 09-10 2015-16
P
er
c
en
t i
n
po
ve
rt
y
Year
25–34
Under 15
years
15–24
35–44
65+
years
55–64
45–54

ECONOMIC DISADVANTAGE 121

For both measures, children under 15 years and people over 65 years have generally had the
highest poverty rates, but for over 65s these rates have declined substantially in recent
years.56 Another movement that stands out is consumption poverty for people aged 15–24,
which has more than doubled since 2003-04.
Estimates from HES indicate that children under the age of 15 exhibit the highest levels of
both income poverty (11.5 per cent, or about 530 000 children) and consumption poverty
(12.9 per cent, or a little under 600 000) in 2015-16. Child poverty is of particular concern
because of the damage poverty may do to a child’s development, their future productive
capacity, and their life prospects more generally (Hampshire, in CEDA 2015, p. 51;
Conti and Heckman 2012, pp. 363–364; Wilkins 2017, p. 35). For example, Najman et al.
(2018, p. 10) recently concluded that childhood poverty acts as a statistically significant
predictor of subsequent adversities, and that adverse events themselves predict subsequent
poverty (underscoring the concept of a ‘cycle of disadvantage’).
6.4 How long does poverty last?
Why poverty duration matters
Concerns about entrenched disadvantage relate not only to concepts of equity and fairness,
but also to the potential for social and political problems to arise from a lack of economic
opportunities (Kelly 2000; Pare and Felson 2014; Weatherburn and Lind 1998).
Disadvantage imposes economic and social costs — both directly and indirectly — on the
people and families who experience it and on the broader community, including forgone
employment income and economic activity, a lower quality of life, and a lower level of social
cohesion (McLachlan, Gilfillan and Gordon 2013, p. 157). Persistent disadvantage heightens
these costs to all parties.57
From a policy perspective, therefore, the length of time that people spend in disadvantage is
as important as the prevalence of disadvantage. One reason for this is that policies to alleviate
persistent disadvantage can differ from those aimed at temporary disadvantage
(Whiteford 2013, p. 58). For example, short-term loans may help those with temporarily low
incomes, but are unlikely to alleviate (and may exacerbate) persistent poverty. Moreover,
long-term disadvantage is of particular concern because it can have deleterious effects on
human capital (including education, skills, work history and professional networks),
reducing a person’s ability to move out of disadvantage later.

56 This holds true in both HES and HILDA data.
57 The literature on disadvantage does not often quantify specific costs, due to conceptual difficulties
distinguishing the causes and consequences of disadvantage, making it hard to construct a realistic
counterfactual. Moreover, there are significant non-economic costs (relating to the quality, or enjoyment,
of peoples’ lives, rather than their material standards of living) arising out of disadvantage, which are
difficult to measure with any accuracy. Rather, the literature has focused on proxy measures of the social
and economic impacts of disadvantage, particularly those which may affect a person’s chances of exiting
disadvantage. See, for example, McLachlan, Gilfillan and Gordon (2013, pp. 147–184).

122 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE
Longer periods of income poverty, at least, appear more difficult to exit
Due to the wider coverage and availability of income data (compared to consumption or
wealth data), the majority of existing poverty duration analyses focus on income poverty.58
Across a range of countries and time periods, persistent or heavily recurrent income poverty
appears to significantly lower the likelihood that someone moves out of poverty in the future,
compared to a single short episode of income poverty (McLachlan, Gilfillan and
Gordon 2013, pp. 55, 66). (This phenomenon is broadly referred to as ‘negative duration
dependence’.)
In Australia, for example, Azpitarte (2012) estimated (based on HILDA data) that, for people
who had experienced income poverty for only one or two years out of the previous nine
years, the probability of exiting poverty in the following year (62 per cent) was about three
times higher than that of someone who had experienced poverty for six or more years in the
same interval (23 per cent). Older analyses, such as Buddelmeyer and Verick (2007), have
drawn similar conclusions.
Azpitarte and Bodsworth (in CEDA 2015, p. 41) also found that the HILDA survey cohort
experienced significant poverty recurrence. From 2000-01 to 2011-12, roughly 27 per cent
of those who had exited from income poverty became poor again between one and two years
after their initial exit.
However, a substantial proportion of those who exited poverty also remained out for a
relatively long time — more than 35 per cent of those who exited did not become
income-poor again within the first 11 years of the HILDA survey. Moreover, the probability
of returning to poverty declined as the time spent out of poverty increased, in something of
a complementary phenomenon to negative duration dependence. From 2000-01 to 2011-12,
less than 10 per cent of respondents who had been out of poverty for more than four years
returned to living below the poverty line.
In this section we estimate income poverty durations, rates of exit, and rates of recurrence
across the life of the HILDA survey thus far (2000-01 to 2015-16).
Most, but not all, poverty spells are short
Income poverty is a common experience, with about half of Australians spending at least
one year in income poverty between 2001 and 2016. Short-term spells were most common,
with about 61 per cent of people exiting income poverty after one year, and another 18 per
cent exiting after two years (figure 6.9). For respondents who entered and exited poverty
during the period (including those who experienced multiple instances of poverty), the
average poverty spell duration was 1.8 years.

58 As such, the patterns observed cannot necessarily be extrapolated to all facets of poverty without further
longitudinal analysis taking place.
ECONOMIC DISADVANTAGE 123
Figure 6.9 Most, but not all, poverty spells are short
Proportion remaining in relative income poverty at each spell duration,
2000-01 to 2015-16a
a Proportions represent the number of unbroken poverty spells recorded in at least one — and up to 16 —
consecutive HILDA surveys, relative to the total number of poverty spells recorded (n = 11 726 953). Because
the HILDA survey takes place annually, the minimum recorded poverty spell is one year (hence, 100 per cent
of poverty spells are shown as lasting at least one year).
Source: Productivity Commission estimates using Melbourne Institute (Household, Income and Labour
Dynamics in Australia (HILDA) Survey, Release 16).
However, there was significant poverty recurrence. Among people who spent at least one
year in poverty at any point, the average total amount of time spent in poverty over the period
was 4.2 years. This substantially exceeds both the most common poverty spell duration (of
just one year) and the average poverty spell duration, suggesting that many people
experience multiple short spells of poverty.
Other key estimates of poverty recurrence support this observation. It was found that:
 exiting poverty comes with a substantial risk of re-entry. Of all respondents who
moved out of income poverty during the period, 56 per cent later fell back into poverty.
This represented almost a quarter of all respondents who had spent at least one year in
poverty during the period.
Poverty spells lasting at least x years
0
10
20
30
40
50
60
70
80
90
100
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
P
er
c
en
t
o
f
al
l p
o
ve
rt
y
sp
el
ls
Poverty spells lasting at least x years
79% of poverty
spells last less
than 3 years …
… but 6% last
for 6 years or
more … … and 1.5%
exceed 10
years

124 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE

– Of those falling back into poverty, about half (47 per cent) did so after remaining out
of poverty for only one year.
– A further 18 per cent spent two years out of poverty before re-entering, and the
remaining 35 per cent spent at least three years (and potentially several more) out of
poverty before their incomes fell below the poverty line again.
– The latter result suggests that, in terms of future earning potential, the detrimental
effects of poverty may be long-lasting in many cases. However, life cycle effects
should not be discounted in this analysis. Some of the HILDA survey cohort reached
retirement age during the period, in which case a move into income poverty is not
entirely unexpected if someone’s major source of income is the Age Pension (which
is set at about 27 per cent of Male Total Average Weekly Earnings). Furthermore, as
McLachlan, Gilfillan and Gordon (2013, p. 65) observe, a person’s level of income
is unlikely to change significantly over time once they have left the workforce (absent
major changes to the Age Pension); consequently, income poverty in the
post-retirement phase could be expected to persist for longer periods than income
poverty during working age.
 recurrent poverty spells are relatively close together. For respondents who exited and
re-entered income poverty, the average amount of time between poverty spells
was 2.6 years.
– Taking into account Wilkins’s (2017, p. 34) observation regarding the impact of
small welfare changes on poverty rates, as well as the generally fluctuating nature of
relative poverty thresholds, this may suggest that many people spend several
consecutive years close to the poverty threshold — ‘oscillating on the margins of
poverty’ (Saunders and Bradbury 2006, p. 361) — before increasing their income in
the longer term.
How many people were in long-term poverty?
Long-term — or frequently recurrent — income poverty affects a small but significant
proportion of the population.
 Fifteen per cent of people spent any five or more years (consecutive or non-consecutive)
in income poverty over the period, while a little under 9 per cent experienced at least one
period of 5 or more consecutive years in poverty.
 Of the respondents who were in income poverty in 2001 (that is, those who started the
HILDA survey already in poverty), 30 per cent were still in income poverty, or had
returned to income poverty, in 2016. This is not dissimilar to the proportion of people
who were in the lowest income decile in both 2001 and 2016 (chapter 5).
In terms of the dynamics of poverty persistence, survival analysis indicates that the
likelihood of experiencing long-term or recurrent poverty (defined as being in poverty at
both ends of a five-year period), given that a household was already in poverty, fluctuated
slightly over the period — between 44 and 47 per cent — but does not show a clear upward

ECONOMIC DISADVANTAGE 125

or downward trend (figure 6.10). There is, therefore, no clear indication that poverty
recurrence — or the ease, or difficulty, of permanently exiting poverty — has either
improved or worsened over the past 15 years.

Figure 6.10 Nearly half of those in poverty in one year are also in poverty
five years later
Per cent of people who remained in or returned to relative income poverty at the
end of a five-year period, given they were in income poverty at the beginning,
2000-01 to 2015-16

Source: Productivity Commission estimates using Melbourne Institute (Household, Income and Labour
Dynamics in Australia (HILDA) Survey, Release 16).
(Note that these estimates of long-term poverty, in particular, should be treated with caution,
due to the likelihood of the HILDA survey under-sampling the most disadvantaged
Australians: box 6.3.)
How does income volatility compare?
Many people in poverty experience greater fluctuations in their incomes than those living
above the poverty line. This can contribute to frequent poverty recurrence and make it
difficult to plan their finances effectively for the future. Between 2006 and 2016, people
below the poverty line experienced more than twice as much income volatility, year-on-year,
as people above the poverty line (figure 6.11).
0
10
20
30
40
50
2000-01 2005-06 2010-11
P
er
c
en
t
Initial year of poverty

126 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE
Box 6.3 A caveat about using HILDA to examine poverty persistence
While longitudinal surveys are essential for analysing poverty duration, they are also limited in
their coverage of the most disadvantaged people in the community, for a number of reasons.
 People most likely to be experiencing disadvantage can also be difficult to contact, and so are
more likely to be excluded from surveys. For example, the HILDA survey excludes homeless
people and people living in very remote areas (McLachlan, Gilfillan and Gordon 2013, p. 192).
 The most disadvantaged people are less likely, or able, to respond if contacted in regards to
surveys (McLachlan, Gilfillan and Gordon 2013, p. 192).
 Disadvantaged people are more likely to drop out from survey samples over time (Watson and
Wooden 2004, p. 300). Efforts to replace these participants, in order to ensure the survey
remains representative, may run into either, or both, of the two problems described above.
There are three main options for dealing with these difficulties.
 One is to undertake special-purpose surveys targeted at those with a higher risk of becoming
disadvantaged, such as Journeys Home (which surveyed homeless people and those at risk
of becoming homeless) and the Longitudinal Survey of Indigenous Children.
 Another option is to over-sample disadvantaged groups in the main longitudinal surveys, such
as HILDA and the Longitudinal Study of Australian Children, which may allow a more granular
examination of factors contributing to disadvantage as well as helping to compensate for
attrition rates (McLachlan, Gilfillan and Gordon 2013, p. 193).
 Administrative records are also a potential source of large and granular datasets, but come
with a range of privacy concerns and a corresponding web of privacy legislation (PC 2017,
p. 138).
Recently, Hérault (2017), with Azpitarte and Johnson, analysed the effect of this under-sampling
(of disadvantaged people) in measuring mobility out of poverty. They found that HILDA poverty
exit rates were substantially higher than those measured in Journeys Home (by at least
28 per cent: pp. 10–12), and that, when decomposed, this difference could not be fully explained
by observable characteristics (pp. 22–26). This research suggests that mobility out of poverty
may have been overestimated for the most disadvantaged Australians.
However, for the volatility of private consumption, the measures were much more similar.
This seemingly reflects the general tendency for people to smooth their consumption over
longer periods of time.59
Who ‘gets stuck’ in poverty?
This study does not undertake detailed analysis of the characteristics of Australians most
likely to experience persistent poverty (or, indeed, other facets of disadvantage). However,
previous work has concluded that some of the groups displaying the highest rates of

59 Note, though, that the bottom income decile — which, given a long-term average poverty rate of about 10 per
cent, includes the vast majority of people in poverty — typically contains a large number of self-employed
people reporting negative labour incomes (chapter 3), particularly in the lowest 3–4 per cent of the
distribution (Saunders and Bradbury 2006, p. 346). Even when negative incomes are set to zero, this drives
down the average income for the bottom decile. The average income therefore tends to be small relative to the
average change in income, and may magnify the volatility of incomes below the poverty line.

ECONOMIC DISADVANTAGE 127

persistent disadvantage include lone parents, Indigenous Australians, people with low
educational attainment, and people with disabilities or other long-term health conditions. For
more detail, see chapter 4 and appendix A of McLachlan, Gilfillan and Gordon (2013).

Figure 6.11 Economic insecurity is higher among those below the
poverty linea,b
Volatility of income and private consumption, 2005-06 to 2015-16
a Volatility is measured as the standard deviation of the 2-year arc percentage change in income (see
Hardy 2017 for more detail). Only survey respondents of working age (25–59 years) were included in the
calculation of volatility. Negative incomes were set to zero. b Consumption does not include expenditure on
consumer durables, such as vehicles or home appliances.
Source: Productivity Commission estimates using Melbourne Institute (Household, Income and Labour
Dynamics in Australia (HILDA) Survey, Release 16).
6.5 Material deprivation
Material deprivation exists ‘when people do not have and cannot afford to buy items or
undertake activities that are widely regarded in society as things that everyone should have’
(Saunders and Wilkins, in Wilkins (ed.) 2016, p. 83). Deprivation metrics aim to provide a
more accurate reflection (compared to poverty alone) of the balance between the resources
available to a household and the basic needs that have to be met.
0
20
40
60
80
Income Consumption
A
nn
ua
l v
ol
at
ili
ty
(
pe
r
ce
nt
)
Below poverty line Above poverty line
The incomes of those in poverty
are more than twice as volatile
as everybody else’s incomes …
… but consumption
volatility is not that
different.

128 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE

Our analysis uses data from deprivation surveys conducted by Saunders (with Naidoo, Wong
and Bradbury) and the Social Policy Research Centre in 2006 and 2010. These surveys
(carried out by post) use a two-stage approach.
 The first stage involves a set of questions around the goods, services and activities
respondents consider essential, and identifies 25 items regarded as essential by a majority
of the population.
 The second stage examines whether respondents possess those items (or have access to
the services and activities) and, if not, whether this is because respondents cannot afford
them, or simply chose not to have them (Saunders and Wilkins, in Wilkins (ed.) 2016,
p. 84).
Saunders’s survey methodology has also been incorporated into a new material deprivation
module of the HILDA survey. From 2014, every fourth survey will include a suite of material
deprivation questions. This means that in the future, it will be possible to conduct
longitudinal analysis of deprivation. At this point, however, only cross-sectional analysis is
possible.
Due to methodological variations between Saunders’s surveys and HILDA, the aggregate
deprivation scores from the two earlier surveys are not comparable to the 2014 results. In
particular, the use of postal surveys for Saunders’s 2006 and 2010 studies resulted in much
lower response rates than the HILDA face-to-face interviews (Saunders and Wong 2012,
p. 23; Watson 2006, p. 15). There is, therefore, a risk that postal survey respondents may
have self-selected based upon specific characteristics. For example, people may have been
more likely to answer if they felt that they personally experienced material deprivation,
which could bias the earlier results upwards.
The prevalence of material deprivation
Saunders and Wilkins’ (in Wilkins (ed.) 2016) analysis of the first HILDA material
deprivation module examined 22 goods, services and activities that were regarded as ‘the
essentials of life’ by a majority of respondents. All but one of these items were also
considered essential by a majority of respondents to the 2006 and 2010 postal surveys
(a motor vehicle being the exception). These items, their deprivation rates (the percentage
of respondents who did not have and could not afford them), and overall deprivation scores
(the average number of essential items of which people are deprived) are listed in table 6.1.
Of the seven essential items with the highest deprivation rates, it is notable that five (at least
$500 in savings for an emergency, home contents insurance, comprehensive motor vehicle
insurance, dental treatment when needed, and a yearly dental check-up for each child) relate
in some way to risk management. For example:
 a lack of insurance means that in the event of damage or theft, people are forced to pay
the entire replacement costs of their possessions

ECONOMIC DISADVANTAGE 129

 a lack of regular dental treatment can lead to much more serious, and costly, dental issues
later on
 a lack of savings may result in people turning to payday lenders, or other high-interest
credit, in the event of an emergency.
That is, deprivation of these items — due to insufficient economic resources — can itself
render a person’s economic situation even worse.

Table 6.1 Rates of item-specific material deprivation
2006 2010 2014
Survey method Postal Postal In-person
interview
(HILDA)
Mean deprivation score (population-wide) 1.43 1.30 0.47
Essential item % who did not have it and could not afford it
At least $500 in savings for an emergency 19.6 17.8 12.2
Home contents insurance 11.1 9.5 8.3
New school clothes for school-age children every yearc 4.0 3.4 6.8
Dental treatment when needed 14.5 13.1 5.2
Comprehensive motor vehicle insurancea 9.8 9.1 4.6
A hobby or a regular leisure activity for childrenb 5.7 5.2 3.7
A yearly dental check-up for each childb 9.8 8.0 3.3
Getting together with friends or relatives for a drink or meal
at least once a month
4.7 4.9 2.5
A roof and gutters that do not leak 4.8 5.1 2.3
Children being able to participate in school trips and school
events that cost moneyc
3.7 3.0 2.1
A motor vehicle .. .. 1.9
Medical treatment when needed 2.1 1.7 1.1
A separate bed for each childb 1.7 2.1 0.8
A home with doors and windows that are secure 5.0 4.4 0.7
When it is cold, able to keep at least one room of the
house adequately warm
2.1 2.5 0.6
Medicines when prescribed by a doctor 4.5 3.5 0.5
Furniture in reasonable condition 2.8 2.2 0.4
A washing machine 1.1 1.0 0.3
A decent and secure home 7.1 6.7 0.3
A substantial meal at least once a day 1.2 0.9 *0.1
Warm clothes and bedding, if it is cold 0.3 0.4 *0.1
A telephone (landline or mobile) 1.9 3.8 *0.1

a Households that have a motor vehicle. b Households with children aged under 15. c Households with
children aged under 15 and attending school. .. Not available. * Estimate not statistically reliable.
Sources: Saunders and Wilkins (2016, p. 85); Saunders and Wong (2012, p. 46).
130 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE
Who suffers from multiple material deprivation?
While it could be argued that the inability to afford any one of the identified ‘essentials of
life’ indicates material deprivation, it is common for deprivation studies to set a higher
threshold (or multiple thresholds for different levels of severity) in order to estimate the
incidence of deprivation. Saunders and Wilkins (2016, p. 86) use two-item deprivation as
their minimum threshold, to allow for response errors and other factors that might otherwise
cause deprivation to be exaggerated.
In addition to examining the overall proportions of people experiencing multiple deprivation,
Saunders and Wilkins (2016, p. 87) present the deprivation rates of various specific groups
(figure 6.12).
Some of their major observations include the following.
 Age-wise, the highest rates of multiple deprivation are found in children under the age
of 15 years; the lowest are found in respondents 65 and over.
– This low level of material deprivation in older Australians is in direct contrast to their
higher relative poverty rates (compared to the population overall), but potentially
reflects their tendency to have a higher level of wealth compared to the general
population (chapter 4, figure 4.12).
– Note, however, that there are more deprivation items that apply to households with
children (table 6.1). This may contribute to the higher deprivation rates of children
(Saunders and Wilkins 2016, p. 86).
 Lone parents experience about three times the deprivation of partnered parents.
 There is a clear ordering of deprivation rates according to both labour force status and
income quintile. The multiple deprivation rate for jobless households was almost four
times that of households in which at least one person is employed full-time. Households
in the bottom income quintile averaged a multiple deprivation rate of 26.3 per cent —
3.5 times that of the middle quintile and 17.5 times that of the top quintile.
 People whose main sources of income are public transfers experience about three times
as much deprivation as wage-earners.
 Indigenous Australians have very high rates of multiple deprivation, while immigrants
from English-speaking backgrounds have low rates.
 Finally, there appears to be a strong relationship between work-restricting disabilities
and deprivation: a quarter of people with a severely work-restricting disability
experience multiple deprivation, compared to 10 per cent of people with a
non-work-restricting disability and 9 per cent of people without a disability.
ECONOMIC DISADVANTAGE 131
Figure 6.12 Children, lone parents, Indigenous Australians and those
with a disability are most at risk of multiple deprivation
Per cent of people deprived of at least two essential items, 2014
Source: Saunders and Wilkins (2016, p. 87).
Relatedly, Redmond and Skattebol (2017, pp. 9–14) have found that youth material
deprivation in Australia (specifically regarding food and clothing) was also more
concentrated among young people with disabilities and Indigenous young people, as well as
young people who cared for a family member with illness or disability. The authors also
0 10 20 30 40
Disability with severe work restriction
Disability with no work restriction
No disability
Non-English speaking foreign country
English-speaking foreign country
Other Australian-born
Indigenous
Bottom quintile
2nd quintile
Middle quintile
4th quintile
Top quintile
Public transfers
Wages
Employed full-time
Employed part-time
Not in the labour force
Unemployed
Elderly couple
Lone parent
Couple with dependent children
65+ years
Under 15 years
Per cent deprived of 2 or more essentials
A
ge
H
ou
se
ho
ld

ty
pe
E
m
pl
oy
m
en
t
st
at
us
In
co
m
e
qu
in
til
e
B
ac
kg
ro
un
d
M
ai
n
in
co
m
e
D
is
ab
ili
ty

132 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE

concluded that youth deprivation was associated with exclusion from participation in normal
activities and low levels of engagement in education, which suggests a dynamic relationship
between material deprivation and social exclusion.
Does income poverty drive deprivation?
As noted in box 6.1, a given level of income alone does not guarantee that a person can —
or cannot — afford to meet their basic needs. Some people with low incomes have savings
to draw upon; others have assets (such as a fully paid-off dwelling) that reduce particular
expenditure needs and therefore free up income to be spent on other essentials. And
conversely, some people have incomes above the poverty line, but nevertheless cannot afford
all of the essentials due to high costs of living or special expenditure needs. Productivity
Commission analysis of the 2014 HILDA deprivation module indicates that, for a majority
of people experiencing income poverty or multiple material deprivation, there is little
overlap (figure 6.13).
Figure 6.13 The overlap between material deprivation and income
poverty is small (2014)a,b

a Percentages shown are proportions of the entire population. b ‘Income poverty’ refers to relative income
poverty (less than 50 per cent of the median equivalised disposable household income).
Source: Productivity Commission estimates using Melbourne Institute (Household, Income and Labour
Dynamics in Australia (HILDA) Survey, Release 16).
Income PovertyMaterial Deprivation
2.3 million (9.8%)Deprived of ≥2 essentials:
2.7 million (11.6%)
440 000
(1.9%)
Deprived of
≥3 essentials:
1.5 million
(6.6%)
680 000
(2.9%)
ECONOMIC DISADVANTAGE 133
Roughly 680 000 people (2.9 per cent of the population) are simultaneously in poverty and
deprived of two or more essential items; this represents about 29 per cent of all people in
income poverty, and 25 per cent of all people experiencing multiple deprivation. And about
440 000 people, or 1.9 per cent of the population, experience both poverty and deprivation
of 3 or more essential items (comprising 19 per cent of those in poverty, and 28 per cent of
those in 3-item deprivation).
Are all deprivations equally bad?
One major limitation of this approach to measuring deprivation is its failure to take into
account the seriousness of different forms of deprivation (because all the essential goods,
services and activities hold equal weights as far as deprivation scores are concerned). For
example, it is questionable whether children going without new school clothes has as serious
an effect on quality of life as not being able to afford prescription medication when it is
needed (McLachlan, Gilfillan and Gordon 2013, p. 38). An alternative to this approach is to
place greater weight on deprivation of those items that were considered essential by larger
proportions of the population (see, for example, Saunders 2011, pp. 133–7).
6.6 Social exclusion
The concept of social exclusion originally arose as a metric of deprivation (Scutella, Wilkins
and Horn 2009, p. 7), but extends to a wider range of life domains, with a focus on
participation in the economic and social activities of a community. In this way, it recognises
the multi-dimensional nature of disadvantage — acknowledging that a lack of economic
resources, inadequate access to services, and low levels of human capital all make it difficult
for people to participate in society (The Smith Family 2003), and that these effects may be
intergenerational. It also has some clear parallels with Sen’s capability model (chapter 2):
Poverty is just one part of this picture. Language and cultural barriers, locational disadvantage or
discrimination because of a disability can also play a part. Social exclusion is often the outcome
of people or communities suffering from a range of problems such as unemployment, low
incomes, poor housing, crime, poor health, disability and family breakdown. In combination,
these problems can result in cycles of disadvantage, spanning generations and geographical
regions. (QCOSS 2009, p. 1)
There is no generally accepted definition of social exclusion, but the concept is widely used
among OECD countries. McLachlan, Gilfillan and Gordon (2013, p. 50) note that it
frequently includes aspects of:
 a lack of agency — exclusion lies beyond the narrow responsibility of the individual, and
generally the individual would like to participate rather than being excluded
 dynamic effects — the individual is excluded not only because of their current situation,
but also because they have little prospect for the future (that is, their resources and
capabilities are likely to remain low for the foreseeable future)
134 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE
 relativity — social exclusion is relative to the norms and expectations of society at a
given point in time (Atkinson and Hills 1998, p. 10).
In Australia, there are two long-running measures of social exclusion: NATSEM’s Child
Social Exclusion Index (Phillips et al. 2013, p. 29) and the Brotherhood of St Laurence and
Melbourne Institute’s Social Exclusion Monitor (SEM) (Horn, Scutella and Wilkins 2011).
Due to its community-wide coverage, this analysis focuses on the SEM.
Developed in 2009, the SEM is based on seven life domains — material resources,
employment, education and skills, health and disability, social, community, and personal
safety — with 29 indicators spread across these domains (box 6.4). It uses a cumulative
scoring system, where the greater the number of indicators or higher the score, the greater is
the depth of an individual’s social exclusion (Horn, Scutella and Wilkins 2011, p. 2). The
SEM is based on HILDA survey data; as such, it lends itself to longitudinal analysis, and is
able to be updated annually.

Box 6.4 Components of the Social Exclusion Monitor (SEM)
The SEM involves seven life domains, which are measured with varying numbers of indicators
(ranging from two to five per domain, for a total of 29 indicators). Twenty-one of these indicators
are measured in all waves of the HILDA survey, while others are available less frequently (as
rarely as every fourth year). The domains are as follows.
1. Material resources: low income (less than 60 per cent of median household income)*; low net
worth (less than 60 per cent of median household net worth); low consumption (less than
60 per cent of median household consumption); and financial hardship (three or more
indicators of financial stress).
2. Employment: in a jobless household*; long-term unemployed*; unemployed*;
underemployed*; and marginally attached to the workforce*.
3. Education and skills: low formal education*; low literacy; low numeracy; poor English; and little
work experience*.
4. Health and disability: poor general health*; poor physical health*; poor mental health*;
long-term health condition or disability*; and household has a child with a disability*.
5. Social connection: little social support*; and infrequent social activity*.
6. Community: low neighbourhood quality; disconnection from community*; low satisfaction with
the neighbourhood*; low membership of clubs and associations*; and low volunteer activity*.
7. Personal safety: victim of violence and/or victim of property crime; and feeling of being unsafe*.
Indicators are generally (though not always) scored as fractions, so that each domain can garner
a maximum score of 1.0 (with a maximum overall score of 7.0). Threshold-wise, a score of 1.0 or
more signifies some level of exclusion. If respondents receive a cumulative score between
1.0 and 2.0 they are regarded as marginally excluded, while a score of 2.0 or more signifies the
presence of deep exclusion.
* Measured in all waves of the HILDA survey.
Sources: McLachlan et al. (2013, p. 75); Scutella, Wilkins and Kostenko (2013, pp. 282–3).
ECONOMIC DISADVANTAGE 135
The prevalence of social exclusion
The Brotherhood of St Laurence and Melbourne Institute updated the SEM in late 2017,
using data from the 2015 HILDA survey. Figure 6.14 shows the rates at which different
demographic groups experience social exclusion (in both marginal and deep forms).

Figure 6.14 Social exclusion rates across specific groups
Per cent of people deeply and marginally socially excluded, 2015
Source: Brotherhood of St Laurence and Melbourne Institute (2017).
Some of the groups most likely to experience social exclusion include: people aged 65 years
and over; Indigenous Australians; people with long-term health conditions and/or
disabilities; people who have attained less than a Year 12 education; people living in public
0 10 20 30 40 50 60 70
Public housing tenant
Private renter
Mortgagee
Homeowner (outright)
Lone parent
Couple with children
Couple
Single person
Long-term ill health/disability
Indigenous Australian
Non-English speaking
Other English speaking
Australia
65+ years
50–64
24–49
15–24
Under 15 years
Female
Male
Per cent socially excluded
Deep exclusion
Marginal exclusion

136 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE

housing; and lone parents. Several of these groups, such as lone parents and Indigenous
Australians, are also likely to suffer from higher levels of material deprivation (and, as
discussed earlier, older people are also more likely to experience income and consumption
poverty, but not material deprivation).
Overall, more women than men experience both deep and marginal social exclusion. This is
likely linked to the high prevalence of social exclusion among lone parents — in 2015,
women made up 84 per cent of lone parents with children under the age of 15 (ABS 2017b).
Relatedly, a socially excluded parent generally translates (in terms of measurement) to
socially excluded children — the rate of social exclusion among children under 15 years is
also relatively high compared to working-age Australians.
Caution is needed when interpreting these results, as some of the characteristics discussed are
also components of social exclusion, or strongly linked to components. This creates a risk that
a particular characteristic could be interpreted as being related to higher levels of social
exclusion, when it may actually be driving people’s classification as excluded. For example,
low income is generally a criterion of eligibility to obtain public housing; as such, public
housing tenants are very likely to score highly in the material resources domain of exclusion.
Some of the characteristics shown may also be correlated with each other, and a failure to
consider these correlations could lead to misinterpretations of the results. Outright
homeowners, for instance, have higher rates of both deep and marginal social exclusion than
do mortgagees. This seems counter-intuitive, at least with respect to the material resources
domain, but outright homeowners are more likely to be older, and older people display far
higher rates of social exclusion. It is likely to be other characteristics of many older people
(such as low incomes and long-term health conditions) that drive this higher level of social
exclusion, as opposed to their home ownership. Box 6.5 explores some other caveats
regarding the interpretation of the SEM.
Overall, social exclusion rates have risen slightly
The Brotherhood of St Laurence and Melbourne Institute (2017) also estimate
community-wide rates of marginal and deep social exclusion for the period 2006–2015
(figure 6.15). For 2015 (the most recent year analysed) they estimated a deep exclusion rate
of 5.3 per cent and a marginal exclusion rate of 17 per cent.
The prevalence of deep social exclusion remained relatively steady over the decade, with a
small sustained rise after 2012. The rate of marginal social exclusion fluctuated more,
particularly around 2008,60 but did not show a clear upward or downward trend.61

60 Given that the relative poverty rate also fell noticeably at this time (figure 6.4), and the ‘low income’ indicator
is simply a variant of the relative poverty threshold applied in section 6.2, this suggests that the low income
indicator may be a major driver of social exclusion prevalence under the SEM formula.
61 However, as a multidimensional measure, a relatively flat trend in social exclusion overall can nevertheless
involve significant movements in the individual component domains. For example, Martinez and
Perales (2017, p. 491) found that the period between 2009 and 2013 saw increases in aggregate exclusion
ECONOMIC DISADVANTAGE 137
Box 6.5 Care must be taken when interpreting the results of the SEM
While the breadth of the SEM’s many domains helps to capture the complexity of social exclusion
as a concept, it also poses some analytical difficulties. It is important, therefore, to keep in mind
that — as for any multidimensional indicator — the selection of the SEM components requires some
subjective judgment calls, with potential consequences for analysis of the results.
Not all components directly reflect exclusion
Some of the components of the SEM — such as certain disabilities, or poor English — do not
necessarily preclude participation in the normal activities of a community, but rather act more as
proxies for other factors (such as discrimination) that are likely to engender exclusion.
Not all components have the same impact
As with material deprivation, all SEM components are weighted equally. But it is questionable
whether — for example — a lack of volunteer activity has as much of a harmful impact on someone’s
social and economic participation as a low level of education or skills.
Not all components are independent of each other
Some components are likely to be closely related — and in some cases, they may cluster and
become mutually reinforcing. For example, if someone has low literacy and/or low numeracy (and
scores on these components of the education and skills domain accordingly), it is logically more
likely that they also have attained an education of Year 11 or less. This is not because people with
low formal education necessarily struggle with literacy and numeracy — rather, it is the opposite, in
that low literacy and/or numeracy can make it more difficult to attain formal educational
qualifications. Similarly, a disability or long-term health condition is likely to be closely related to
several other components within the health and disability domain.
This could mean that, when a person scores on one component, they are likely to score on others
in the same domain — but these multiple scores effectively reflect the same type of hardship. If the
aim of the SEM is to indicate when people are experiencing multiple forms of hardship at once, this
could run counter to that aim.
There is no ‘offsetting’ of components
The SEM does not include a mechanism for positive characteristics (representing inclusion) to offset
the components of exclusion. Consequently, if a person displays a handful of the more common
SEM components — such as less than 60 per cent of median net worth, low membership of clubs
and associations, or low volunteer activity — they could be classified as socially excluded even if
they are educated, employed, and have a strong social support network.
Despite these caveats, the SEM is a valuable data source
Importantly, none of this goes to say that the SEM is an invalid metric, or that social exclusion is not
likely to be a fairly widespread phenomenon. Nor does it nullify estimates of the persistence of social
exclusion. It is simply a characteristic of multidimensional indicators that their complexity is both a
strength and a shortcoming, and that care must be taken when drawing conclusions from the results.
Source: McLachlan, Gilfillan and Gordon (2013, pp. 50, 80–81, 189).
scores for the health, material resources, employment and social support domains, while aggregate scores
decreased for the education and community participation domains. Relatedly, Azpitarte and Bowman (2015,
pp. 8–9) have previously found significant variation in the extent to which the individual component domains
contribute to the depth and prevalence of social exclusion experienced by different age groups.
138 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE
Figure 6.15 Marginal exclusion has fluctuated more than deep exclusion
Total rate of social exclusion, 2006–2015 (per cent)
Source: Brotherhood of St Laurence and Melbourne Institute (2017).
Social exclusion is mostly short-lived
The 2017 SEM update presents estimates of the survival function for marginal and deep
social exclusion (figure 6.16). This figure indicates that, while almost half of all Australians
experienced some level of social exclusion between 2006 and 2015 (and well over 10 per
cent experienced deep exclusion), social exclusion — particularly deep exclusion — is
usually transitory.
Almost half of all the Australians who experienced deep social exclusion only did so for a
single year over the decade shown, and less than one per cent of Australians experienced
deep social exclusion for more than six years out of ten. (However, as noted in box 6.3, many
people who are likely to be socially excluded will show a higher attrition rate than the total
HILDA sample, so it is improbable that all socially excluded respondents will remain in the
sample in the long term.)
Marginal social exclusion was more persistent than deep social exclusion. Slightly less than
one-third of the people who experienced marginal social exclusion did so for only a single
year, and about 4 per cent of Australians were marginally excluded for more than six years
out of the decade shown (figure 6.16).
0
5
10
15
20
25
2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
P
er
c
en
t
Year
Deep exclusion Marginal exclusion

ECONOMIC DISADVANTAGE 139


Figure 6.16 Social exclusion is mostly transitory, particularly deep exclusion
Years spent in social exclusion, 2006–2015
Source: Brotherhood of St Laurence and Melbourne Institute (2017).
However, as noted in box 6.5, the subjective judgments involved in the selection of SEM
components can affect measurement. This is also the case for the duration of exclusion, as
certain components — for example, having a long-term health condition or disability, or
having a child with a disability — are (by definition) likely to be long-lasting or permanent
conditions. If these components could not realistically be ‘removed’ from someone’s life,
they would increase the likelihood of that person permanently scoring above the threshold
for social exclusion, even if other aspects of their life (such as employment or social
connections) improved significantly. The selection of components could, therefore, be a
driver of some of the persistent exclusion observed in in figure 6.16.
This makes it difficult to compare the prevalence or persistence of social exclusion directly
to those of poverty — especially given that the SEM itself includes a material resources
domain, and therefore overlaps with poverty analyses to an extent. (It also makes it difficult
to link changes in the prevalence of social exclusion to particular events.) However, given
the breadth of indicators encompassed by the SEM — and the subjectivity of some
indicators, such as ‘low satisfaction with one’s neighbourhood’ or ‘feeling of being
unsafe’ — it is certainly possible for poverty (income, final consumption, and financial) to
be decreasing, as seen in figure 6.1, while social exclusion increases.

0
2
4
6
8
10
1 2 3 4 5 6 7 8 9 10
P
er
c
en
t
o
f
ad
u
lt
p
o
p
u
la
ti
o
n
Years spent in social exclusion, 2006-2015
Marginal exclusion Deep exclusion
About 2 per cent of adults spent five or
more years, out of ten, in deep exclusion
GLOSSARY 141

Glossary
Assets See wealth.
Capital income See income.
Consumption Private consumption: Household expenditure (money spent)
on goods and services, including consumer durables (such as
vehicles and household appliances) and imputed rent. In this
study, consumption excludes income tax, mortgage
repayments, other housing costs (such as repairs, council rates
and renovations), superannuation and life insurance.
 Imputed rent: An estimate of the housing amenity enjoyed
by owner-occupiers, valued at market rental rates. In this
study, imputed rent is calculated as 5 per cent of the
estimated sale price of the residence.
Final consumption: Private consumption plus in-kind
transfers from governments.
 In-kind transfers: The value of government services used
by a household (including public education, healthcare,
childcare, government housing and other welfare services).
In this study, in-kind transfers are the same as social
transfers in kind.
Decile See quantile.
Deprivation See disadvantage.
Disadvantage In this study, disadvantage refers to economic disadvantage.
Disadvantage is a multifaceted concept, encompassing three
elements — poverty, material deprivation and social exclusion.
Poverty: Having low economic resources.
 Income poverty: Having low income, compared to some
threshold of ‘need’.
– Absolute income poverty: Having income that is
insufficient to cover the costs of a basket of ‘necessary’

142 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE
goods and services, which is updated as community
norms evolve.
– Anchored income poverty: Having income that is
below a particular threshold, the real value of which is
held constant over time.
– Relative income poverty: Having income below a
certain percentage of median household income. This
study uses a relative income poverty threshold of
50 per cent of median income.
 Consumption poverty: Having a low level of
consumption, compared to some threshold of ‘need’.
 Financial poverty: Simultaneously having less than
50 per cent of the median income, less than 50 per cent of
the median private consumption, and total liquid assets less
than three months’ worth of the equivalised income poverty
line for a given household size.
Deprivation: Being unable to afford to buy items or undertake
activities that are widely regarded in society as things that
everyone should have. In this study, deprivation is the same as
material deprivation.
Social exclusion: Being unable to fully participate in the
ordinary economic and social activities of a community.
Disposable income See income.
Economic disadvantage See disadvantage.
Economic inequality See inequality.
Economic mobility See mobility.
Equivalisation Adjusting household-level variables for differences in the
number and age of people in each household. This enables
comparison of the economic resources available to different
households, by accounting for the fact that larger households
need more income to achieve the same standard of living as a
smaller household, and households generally have some
‘economies of scale’ due to sharing living costs.
Expenditure See consumption.
Final consumption See consumption.

GLOSSARY 143
Financial wealth See wealth.
Gini coefficient A summary indicator of the overall distribution of income,
wealth or consumption. It takes a value between 0 and 1. A
value of 0 indicates perfect equality (all people have the same
income) and a value of 1 indicates perfect inequality (one
person has all the income). The smaller the value, the more
equal the distribution.
Gross income See income.
Household assets See wealth.
Household income See income.
Household liabilities See wealth.
Imputed rent See consumption.
In-kind transfers See consumption.
Income In this study, income refers to equivalised household income
(total cash receipts that are received by a household or by
individual members of the household, during a given period),
adjusted for differences in the number and age of people in each
household (see equivalisation).
Gross income: Total income from labour, capital and
government transfer payments, before tax.
 Labour income: Wages, salaries and other
employment-related income.
 Capital income: Income received in respect of assets, such
as business income, rental income, dividends, interest and
royalties.
 Transfer payments: Income received from direct
government cash payments under the social security
system, such as Newstart, Family Tax Benefit and the Age
Pension.
Disposable income: Gross income minus income tax paid.
Inequality In this study, inequality refers to economic inequality, defined
as differences in people’s access to economic resources to
support their wellbeing.

144 RISING INEQUALITY? A STOCKTAKE OF THE EVIDENCE
Intergenerational
mobility
See mobility.
Intragenerational
mobility
See mobility.
Labour income See income.
Liabilities See wealth.
Life course mobility See mobility.
Liquid assets See wealth.
Material deprivation See disadvantage.
Mobility In this study, mobility refers to economic mobility, defined as
the extent to which people move across the income, wealth and
consumption distributions. There are two types of mobility:
intergenerational mobility and life course mobility.
 Intergenerational mobility (between generations) refers
to the relationship between a person’s economic position
and that of their parents.
 Life course mobility (sometimes referred to as
intragenerational mobility) refers to changes in an
individual’s economic position throughout their lives.
Net wealth See wealth.
Net worth See wealth.
Personal wealth See wealth.
Private consumption See consumption.
Quantile Units of analysis formed by ranking all observations (for
example, incomes of people or households) in a distribution
from smallest to largest and then dividing these into a certain
number of equal-sized groups.
Decile: One of 10 groups formed by ranking all observations
from smallest to largest and then dividing these into
10 equal-sized groups.

GLOSSARY 145

Percentile: One of 100 groups formed by ranking all
observations from smallest to largest and then dividing these
into 100 equal-sized groups.
Quintile: One of five groups formed by ranking all
observations from smallest to largest and then dividing these
into five equal-sized groups.
Quintile See quantile.
Social exclusion See disadvantage.
Social transfers in kind See consumption.
Transfer payments See income.
Wealth Total household assets minus total household liabilities.
 Household assets: The value of entities owned by people
in a household, from which economic benefits can be
derived over time.
 Household liabilities: The value of loans outstanding,
payable by people in a household.
Financial wealth: The value of accounts held with financial
institutions (including offset accounts), shares, debentures and
bonds, loans to non-household members, other financial
investments, silent partnerships, public unit and private trusts,
children’s assets, and assets not elsewhere classified, less
investment loans outstanding (excluding business and property
loans).
Liquid assets: Cash, bank deposits, and equity, plus
superannuation if at least one person in the household is aged
over 65 years.
Net wealth: In this study, wealth is the same as net wealth.
Net worth: In this study, wealth is the same as net worth.
Personal wealth: The value of home contents less student and
credit card debt, and loans for other purposes (such as holidays
or consumer goods).
Wellbeing A person’s ability to achieve ways of living that they value.