Estimating the impact of the global financial crisis on poverty and deprivation.
Saunders, Peter ; Wong, Melissa
Introduction
Despite the controversy over its measurement, poverty is a major
social problem and its alleviation remains an important policy priority.
The current government has acknowledged this, noting that
'indicators of poverty and financial stress continue to be present
despite close to two decades of strong economic growth' (Australian
Government 2009: 7) and poverty rates are included among the indicators
being used to guide and monitor its social inclusion agenda (Australian
Social Inclusion Board 2009; Australian Government 2012).
Australia's poverty rate in the mid-2000s was above the OECD
average (OECD 2008: Figure 5.1; OECD 2010: Figure 4.4B) and concern has
been expressed about the extent of poverty among specific groups,
including Indigenous Australians, sole parents and people with a
disability. Even the OECD--normally reluctant to be directly critical of
specific national programs --has noted that the low level of Newstart
Allowance 'has raised concerns about its adequacy [and]
effectiveness in providing sufficient support for those experiencing a
job loss, or enabling someone to look for a suitable job' (OECD
2010: 127-8). Growing international social unrest over rising levels of
inequality has further focused attention on poverty since there is clear
evidence (at least for OECD countries) of a positive cross-country
relationship between overall income inequality and national poverty
rates (Nolan & Marx 2009: Figure 13.2).
These developments illustrate that, despite their limitations,
poverty line studies still have a role to play in assisting government
agencies to monitor trends andassess income adequacy for those dependent
on the social security system. The recent Pension Review, for example,
presented comparisons between existing pension levels and two
widely-used poverty lines (the Henderson poverty line and one set at 50
per cent of median income), even though it cautioned that 'it
considers neither of these to be a particularly robust measure of
wellbeing' (Harmer 2009: 34, Chart 6). In a similar vein, Wilkins
and colleagues (2011: 34) argue that a poverty line set at 50 per cent
of median income is 'based on a degree of public and researcher
consensus' but acknowledge that it (along with all other poverty
lines) contains 'an element of arbitrariness'.
Support for the use of a relative poverty line benchmarked against
median income also emerged from a recent review by Fair Work Australia
of the use of alternative living standards indicators to assess how well
the needs of the low-paid are being met (Pech 2011). It argued that
among the commonly used alternative income benchmarks, relative poverty
lines 'would seem to offer the most potential for research and
policy analysis' (Pech 2011: 44)--although it also argued that
income should be broadly defined and supplemented by 'information
on financial and other outcomes' (Pech 2011: 76).
This paper applies a poverty line approach to examine the impact of
the global financial crisis (GFC) using survey data that span the years
when it was at its peak. It also examines how deprivation changed over
this period and what happened to a consistent poverty measure that
combines elements of low-income and deprivation. A brief review of the
indicators used and the economic context that accompanied the GFC is
presented in the next section, followed by a description of the survey
data. We then present results showing how poverty, deprivation and
consistent poverty changed between 2006 and 2010 and discusses what the
alternative approaches imply for the extent and nature of Australian
poverty. The final section provides a summary of the main conclusions.
Poverty, deprivation and consistent poverty
Previous Australian studies have measured poverty on the basis of
income alone (for example, Harding et al. 2001; Saunders & Hill
2008; Wilkins 2008; Wilkins et al. 2011: Chapter 7), generally using a
poverty line set at 50 per cent of median income, with sensitivity
analysis conducted to check the robustness of the estimates. An
important finding to emerge from these studies is that estimated poverty
rates can be sensitive to small shifts in the poverty line, particularly
for those whose income is mainly derived from social security payments.
This is because many social benefits are close to one-half of median
income and the fiat-rate, income-tested nature of the Australian social
security system leads to a bunching of incomes around this part of the
distribution particularly for groups like the aged (Tanton et al. 2009).
(1) It is thus important to examine poverty using a variety of poverty
lines, including those set at both 50 per cent and 60 per cent of the
median (see Saunders & Bradbury 2006; Saunders, Hill & Bradbury
2007). It is also relevant to note in this context that most European
poverty studies (and official poverty targets) use a poverty line set at
least at 60 per cent of median income, although the 50 per cent
benchmark is most commonly used by the OECD (2008) and in studies based
on the Luxembourg Income Study (LIS) database (for example, Smeeding
2006).
Despite this income focus in Australian poverty research, there
have been significant advances in data availability and in theoretical
understanding of the concept of poverty. In relation to the former, new
longitudinal data has made it possible to examine the dynamics of
poverty using, for example, data from the Household, Income and Labour
Dynamics in Australia (HILDA) survey (see Headey et al. 2005;
Buddelmeyer & Verick 2008; Wilkins et al. 2011; OECD 2008: Chapter
6). In relation to the latter, an important development --emphasised by
Jenkins & Micklewright (2007)--is the shift away from a
uni-dimensional (income-based) focus to a multi-dimensional framework
building on work on deprivation by Townsend (1979), or on Sen's
ideas of functioning and capabilities (Sen 1985).
More generally, poverty has come to be seen as not lust an issue in
itself, but as one of a number of factors that can result in exclusion.
Low income may increase the risk of poverty but will not always result
in poverty, so that defining and measuring poverty on the basis of
income alone lacks credibility because the existence of poverty has not
been demonstrated. The conceptual limitations of using income alone to
identify poverty have been compounded in the Australian context by the
problems involved in producing accurate measures of income. The
Australian Bureau of Statistics (ABS) has argued that this is a
particular problem for those with lowest (reported) income and now
excludes those in the lowest decile when identifying 'low income
households' (ABS 2010). Although it has progressively introduced a
series of measures designed to improve the quality of its income
statistics across the entire distribution (ABS 2003; 2009a; 2011), this
has made it harder to track changes over time in poverty (and income
distribution)--particularly over longer periods. This in turn has
undermined the usefulness of poverty research for monitoring social
change and assessing the impact of policy. However, there is still a
need for research that can identify which groups are most at risk of
poverty and to help identify where additional support is most needed
(see ACOSS 2008).
In recognition of the limitations (empirical and theoretical) of
relying on income alone to estimate poverty, indicators of financial
stress or hardship have been used in several Australian studies to
better identify who lacks adequate economic resources--either as an
alternative to income or in combination with it (Bray 2001; Headey 2007;
Marks 2007; Hahn & Wilkins 2008; Wilkins et al. 2011, chapter 9).
Other studies have combined low income with other indicators of low
economic capacity as a way of better identifying who is actually
experiencing poverty (Saunders & Hill 2008; Scutella et al. 2009).
Work being undertaken by the ABS on developing a measure of
'consumption possibilities' fits within this latter tradition,
although it does not overcome the challenge of determining who is poor,
or where action is needed (see ABS 2009b; Billing et al. 2010.) In a
recent development on which the current paper draws, Saunders and Naidoo
(2009)--extending earlier work by Saunders, Naidoo and Griffiths
(2008)--combined low-income with deprivation estimates to produce a
measure of what has come to be called consistent poverty (Nolan &
Whelan 1996; Whelan et al. 2006).
The concept of consistent poverty builds on a new approach to
poverty measurement that extends Townsend's work on deprivation
that was first used to identify poverty over 30 years ago (Townsend
1979). Since then, a number of improvements have been made to the
approach and it has received official endorsement by governments who
have set and monitored poverty reduction targets using an approach that
combines low income with deprivation (such as in the UK
Government's commitment to eradicate child poverty and under the
EU's Social Agenda being pursued as part of the Lisbon
Strategy--see Atkinson (2007)). Under the deprivation approach, people
are identified as deprived when they face 'an enforced lack of
socially perceived necessities' (Mack & Lansley 1985: 39). in
practice, this involves identifying a list of items that are widely
regarded as necessary or essential for people to attain an acceptable
standard of living, and then establishing who does not have and cannot
afford these items. It should be noted that deprivation is not the same
as financial stress or hardship referred to earlier, because the
indicators used to identify the latter are not based on community views
about what is necessary or essential.
One of the key findings to emerge from this body of work is that
there is a relatively low degree of overlap between poverty, measured on
the basis of income alone, and deprivation. Thus, the authors of one
recent review of literature in the field have noted that:
'In the poverty context, it has been argued forcefully that
low income may fail in practice to distinguish those experiencing
distinctively high levels of deprivation or exclusion, and studies using
direct measures of deprivation for a range of countries have lent some
support to this assertion' (Nolan & Marx 2009: 319)
There are many potential explanations for this, including the
inaccuracy of the income statistics used to estimate poverty and the
failure of most equivalence scales to make allowance for the costs of
disability or the entrenched disadvantage (and hence greater needs)
faced by some groups. These factors suggest that there is a clear need
to complement income-based poverty estimates with evidence that taps
more directly into the actual living conditions experienced, and the
deprivation approach provides a method for doing just that.
The first national study of deprivation in Australia was conducted
by Saunders, Naidoo and Griffiths (2007), and this paper builds on that
work using data from the original survey and a more recent one,
conducted in mid-2010, which replicated many of the questions included
in the earlier survey. Given their timing, these two surveys allow the
impact of changes experienced between 2006 and 2010 on deprivation and
other indicators of social disadvantage (including conventional income
poverty rates) to be examined using a common set of indicators. It is
important to acknowledge at the outset that although the period examined
spans the GFC, the observed changes cannot be attributed solely to the
GFC, since many other factors also changed over the period (see Saunders
& Wong 2011a; 2012). These include the series of fiscal stimulus
measures introduced by the government in 2008 and 2009 to avoid the
recession that was widely predicted at the time, which proved to be
effective in moderating its effects. These one-off measures, along with
the substantial increase in the age pension introduced following the
Pension Review mentioned earlier, increased the incomes of many
low-income (and other) Australian households, thereby reducing their
exposure to both poverty and deprivation. Because of this, it is not
possible to conclude by comparing outcomes observed in 2006 and 2010
that any difference was attributable solely to the GFC. It is
nonetheless of interest to document how disadvantaged Australians fared
during this period of turmoil and uncertainty.
This is all the more important because when the GFC originally
emerged, grave warnings were voiced about its likely impact on those
least able to protect themselves against its effects. In the immediate
aftermath of the outbreak of the GFC, several government and
non-government agencies responsible for addressing social problems
undertook or commissioned studies of its impact on disadvantaged and
vulnerable Australians. The clear message to emerge from these studies
(reviewed in more detail in Saunders & Wong 2011a) was that the
crisis was likely to have a disproportionate impact on those with lowest
incomes. For example, a report commissioned by a consortium of leading
community sector NGOs and conducted by Access Economics in 2008 argued
that as a result of the GFC:
'Economic growth will inevitably slow, the extent to which is
uncertain ]but the] impact will vary across different segments of
society, with the unemployed and other vulnerable groups particularly
hard hit' (Access Economics 2008: 2; emphasis added)
A series of subsequent studies confirmed this bleak prediction,
including those undertaken by the Department of Families, Housing,
Community Services and Indigenous Affairs (FAHCSIA 2009), The Salvation
Army (2010) and the Wesley Mission (2010). Although the surveys that
underlie these reports are not nationally representative and their
timing means that they would have captured the impact of factors other
than the GFC (including rising living costs, which were of increasing
concern), the overwhelming impression they present is that the GFC
resulted in a considerable negative impact on those least able to absorb
it. For example, the Foreword to the Wesley Mission report noted that:
'Many are still struggling to recover from the job losses or
reduced working hours caused by the GFC. Rising rents and mortgage costs
add to the problem as does the relentless increase in utility
costs.' (Wesley Mission 2010: 5).
Evidence derived from waves 7 and 9 of the HILDA survey is
consistent with a modest increase in financial stress between 2007
(prior to the onset of the GFC) and 2009 (by which time fears of its
negative effects in Australia had largely dissipated). Over that period,
five of seven indicators of financial stress increased, with the largest
increase being the percentage of respondents who asked for help from a
welfare or community organisation because of a shortage of money. (2)
Although the increases were all modest in size--less than one percentage
point--no indicator had recorded an increase in any two-year period
between 2001 and 2007 (Wilkins et al. 2011: Table 9.1). There is thus
some evidence to support the claims made in the smaller studies referred
to above, although it is clear that the whole issue warrants further
examination, and the rest of this paper addresses that task.
Data sources and methods
The Community Understanding of Poverty and Social Exclusion (CUPSE)
survey was distributed by mail to 6,000 adult Australians randomly
selected from the electoral rolls in April 2006. (3) It generated 2,704
responses, equivalent to a response rate of 46.9 per cent--somewhat
higher than that achieved by other similar social surveys conducted
around that time. Thus, for example, the 2003 Australian Survey of
Social Attitudes (AuSSA) achieved a response rate of 44 per cent--see
Wilson and colleagues (2005: 7). The Poverty and Exclusion in Modern
Australia (PEMA) survey was distributed to a new sample of 6,000 adults
in May 2010 and generated 2,645 responses--equivalent to a response rate
of 46.1 per cent. (4) The 2010 survey included the same questions as
were asked in 2006 (with some minor modifications to address
difficulties encountered in the earlier survey), although some questions
about attitudes to poverty and inequality were removed, and new
questions relating to the impact of the GFC and to aspects of community
participation and location were included.
Detailed comparisons between the composition of both samples and
relevant ABS data (reported for 2006 by Saunders, Naidoo and Griffiths
(2007: Table A.3) and for 2010 in Saunders and Wong (2011b: Figure 1)
and Saunders and Wong (2012: chapter 3)) indicate that, although
generally representative, the following groups are under-represented in
both the CUPSE and PEMA samples: males; those aged under 30; those who
have never been married; those who live alone; Indigenous Australians;
those with lower levels of education; those in private rental
accommodation; and those with incomes over $1,000 a week. (5) Most of
these differences are small, and while some of them are inter-related,
others may reflect the difficulties involved in conducting (and
responding to) a mail survey.
The one area where the difference between sample composition and
the structure of the adult population was most pronounced is in relation
to age structure. As is common for mail surveys, older people (aged 50
and over) were over-represented relative to younger people (particularly
those aged under 30) among the respondents of both surveys. This
age-related bias can affect key aspects of the survey results (such as
when identifying whether an item attracts majority support for being
essential) and in these instances, population-based weights have been
applied to the raw data before drawing any conclusions. Previous
analysis reported in Saunders, Naidoo and Griffiths (2007, 2008)
indicates that aside from identifying which items receive majority
support for being essential, weighting the survey data to reflect the
age structure of the population has relatively little impact on the
results, so (except where indicated) only the unweighted estimates are
presented below.
Both surveys included questions about each of a series of items (61
in the case of CUPSE, 73 in PEMA) that asked respondents to provide a
'Yes' or 'No' response to three questions about each
item: Is it essential (where essential was defined as 'things that
no-one in Australia should have to go without today')? Do you have
it? And, if not (and where the item is purchasable by individuals), is
this because you cannot afford it? The selection of the items was based
in part on focus groups held with low-income Australians who were asked
to identify what they thought was necessary to achieve a decent standard
of living (see Saunders & Sutherland 2006). Other items were drawn
from deprivation studies conducted in New Zealand, Ireland and Britain,
and the list also included modified versions of some of the indicators
used to identify hardship or financial stress in the Australian studies
referred to earlier.
The responses to these three key questions were used to identify
deprivation as follows: First, following international practice (see
Gordon 2006; Pantazis et al. 2006), only those among the 61 items
included in both surveys that attracted majority support for being
essential were considered. Deprivation was then defined to exist when
people did not have and could not afford each of these items. This
approach to defining deprivation embodies community opinion about which
items are essential, with the lack of each item being indicative of an
unacceptable standard of living, with the fact that this was because the
item could not be afforded providing the link to poverty.
A total of 26 items received majority support for being essential
in 2006, although one of these (the television) was subsequently dropped
after conducting reliability and validity tests (see Saunders &
Naidoo 2009). One additional item (a separate bedroom for children aged
10 and over) failed to attract majority support in 2010 (and only just
did so in 2006) and this too was dropped. The remaining 24 items were
regarded as essential by a majority in both years, and these are shown,
along with the degree of support attracted by each item, in Table 1.
These 24 'essentials of life' items form the basis of the
analysis of deprivation that follows.
Results
The two surveys described above have been used to estimate the
social impact of the GFC by examining changes in three indicators:
conventional (income-based) poverty rates, the incidence and severity of
deprivation, and the overlap between these two, or the incidence of
consistent poverty. The sensitivity of the results to variations in how
each indicator is defined has also been examined. Thus, poverty rates
have been estimated using poverty lines set at 50 per cent and 60 per
cent of median (equivalised) disposable income. (6) Three summary
indicators of deprivation have been used: the percentage of households
that are deprived of at least two, or at least four of the 24 essential
items, and a sum-score index equal to the average number of essential
items that households of a given type are deprived of. A similar
sum-score index forms the basis of the index of social exclusion being
developed at The Melbourne Institute (see Scutella et al. 2009), and
there is statistical support for the use of such an index as a summary
measure of multiple hardship or deprivation (see Butterworth &
Crosier 2006; Cappellari & Jenkins 2007).
Following Saunders and Naidoo (2009: Table 9), consistent poverty
is defined as those with incomes below each of the two poverty lines,
who are also deprived of two or more (or four or more) of the 24
essential items. Results are presented in aggregate and for the
following six household types: working-age single adults (aged 18-64);
older single adults (aged 65 and over); working-age couples without
children; older couples; couples with children; and sole parent
families. Children are defined as being aged under 18 years and mixed
households containing three or more adults (including non-dependent
children aged 18 and over) have been excluded from the disaggregated
analysis.
Table 2 presents the estimated poverty rates in each year using the
two poverty lines. The estimates in columns 1 and 2 compare for 2006 two
methods of imputing an exact income figure from the survey data that
provides incomes in ranges (see endnote 5): the estimates in column 1
assign all incomes to the mid-point of the relevant income range, and
are thus the same as those used by Saunders and Naidoo (2009), while
those in column 2 use a more sophisticated imputation method in which a
random assignment method has been used to generate incomes within each
income range. (7) Not surprisingly, there are large differences in the
estimated poverty rates produced by the two approaches, particularly for
older households. The net impact is to reduce the poverty rate by
between three and seven percentage points, depending upon which poverty
line is used. (8) It is important to note that, despite the limitations
of the survey income and household composition variables, the resulting
poverty rate estimates are similar to those produced using better
quality (ABS or HILDA) data. Thus, the estimated poverty rate of 11.1
per cent based on the lower poverty line in 2006 is identical to that
estimated for 2006-07 using ABS income data by Saunders, Hill and
Bradbury (2007: Table B.4), although the poverty rate at the higher
poverty line shown in Table 2 is below their estimate of 19.4 per cent.
Wilkins and colleagues (2011: Figure 7.1) estimate a poverty rate of
just below 12 per cent in 2005-06 using HILDA data.
Columns 3 and 4 of Table 2 show the impact in 2010 of applying two
alternative measures of household size and structure. The
"comparable" estimates in column 3 replicate the methods
applied to the 2006 data by Saunders and Naidoo (2009), while the
'best' estimates in column 4 are based on the more complete
information on household structure that was collected in 2010. (9) The
impact of these alternatives on estimated poverty rates is small and
restricted to those (mixed) households (not shown) that contain three or
more adults (or non-dependent children).
The estimates for the two years shown in columns 2 and 3 of Table 2
(which are most directly comparable in terms of how they were generated)
imply that there was a slight increase in poverty using the lower
poverty line and a small fall in the poverty rate using the higher
poverty line. However, the picture of change over the period is
different for different households and depends for some of them on which
poverty line is used. (10) The lower poverty line indicates that poverty
fell among single older people and households with children,
particularly sole parents. The latter declines reflect a range of
factors (including employment growth over the period) but their impact
would need further examination that is beyond the scope of this paper.
While these declines still exist when the higher poverty line is used,
the poverty rate among single older people is now estimated to have
increased. This is despite the substantial increase in the single rate
of age pension in September 2009 following the Pension Review (Harmer
2009: see also Saunders & Wong 2011c) and reflects the bunching of
pensioner incomes referred to earlier.
Before proceeding, it is important to acknowledge the limitations
of all of the poverty estimates in Table 2. They are all based on income
data provided in surveys that were not specifically designed to estimate
poverty and many adjustments have had to be made to the raw data before
the estimates could be produced. Two of the most important of these
relate to the methods used to derive a point estimate of gross income
and those used to derive disposable income by imputing tax paid. Table 2
is thus not intended to contain definitive estimates of poverty rates,
but to provide a benchmark against which the estimates of deprivation
and consistent poverty can be compared. It is important that this caveat
is kept in mind when assessing the implications of this aspect of the
results.
The estimated changes in deprivation presented in Table 3 avoid the
complexities involved in imputing a precise value for income from
grouped data (or modeling tax liabilities), and do not require a choice
to be made about where to set the poverty line or what equivalence scale
to use. They do not, however, avoid the need to make judgments about how
to identify which items are essential (the majority support criterion)
and to decide how many lacked essential items constitute being deprived.
Reflecting these judgments, three alternative indicators are shown in
Table 3, two based on the number of essential items that are lacking and
the third equal to the mean deprivation score (or sum-score index)
described earlier. Comparing these alternatives allows the sensitivity
of the estimates to be examined and their robustness to be established.
Whichever indicator is used to measure deprivation, the results
show a clear overall improvement over the period, as well as in the
circumstances of virtually all household types. These declines in
deprivation are more pronounced and widespread than the modest falls in
poverty rates shown in Table 2. It is notable, for example that
deprivation among single older people is much lower in 2010 than in
2006, confirming the significant impact of the 2009 pension increase.
The improved circumstances of single working-age people (and sole
parents) also shows up more clearly in the deprivation figures in Table
3 than in the poverty rates in Table 2, although the change experienced
by couples with children is much smaller (and not consistent across all
indicators) for deprivation than for poverty.
Overall, the deprivation estimates in Table 3 provide clearer
evidence of improvement over the period than the poverty rates presented
in Table 2, and although the picture in both cases is one of only modest
improvement, this is true even among those groups most at risk. This
finding contrasts with the claims described earlier about the negative
impact of the GFC being greatest for those least able to afford a
further set-back, and with the modest increases in many of the HILDA
financial stress indicators described earlier.
Table 4 estimates how consistent poverty changed over the
period--in aggregate and for different household types--when alternative
poverty lines and deprivation indicators are used to identify it. These
poverty estimates are based on the 'comparable' estimates
shown in Table 2, although the 2006 consistent poverty estimates differ
slightly from those presented by Saunders and Naidoo (2009: Tables 4
& 9) because of differences in the methods used to produce them.
Looking first at the overall figures, the consistent poverty estimates
are all considerably below the conventional poverty rates shown in Table
2, and they either declined or were unchanged between 2006 and 2010. Our
preferred measure is based on a poverty line set at 60 per cent of
median income and being deprived of at least two essential items,
because the use of a higher poverty line reflects European practice (see
Whelan et al. 2006), although we accept that the number of deprivation
items is a matter for judgment. On this measure, consistent poverty
declined overall from 8.9 per cent in 2006 to 7.9 per cent in 2010 (11).
In proportionate terms, this represents a decline of around 11 per
cent--somewhat lower than the declines in all three indicators of
deprivation in Table 3, but higher that the proportionate reduction in
the poverty rates in Table 2. Table 4 also indicates that consistent
poverty fell among single older people and had virtually disappeared by
2010, with most other groups (with the exception of older couples and
couples without children) also experiencing declines between 2006 and
2010.
Table 5 provides a summary of the impact of the alternative
indicators used in Tables 2-4 on the composition of households who fall
into the disadvantaged category when each indicator is used. Looking
first at the impact within each year, the main effect of replacing the
conventional (income-based) poverty measure by the consistent (income
and deprivation) poverty measure is to increase the proportions of the
poor who are single working-age adults and either couples with children
or sole parent families (depending on the year), and to reduce the
proportions who are older people (single or couples) and working-age
couples without children. The decline in the latter group--childless
couples--is particularly striking.
The overall impact is thus that the group identified as being poor
contains a greater proportion of families with children when the
consistent poverty measure is used than when a conventional poverty line
approach is adopted. Thus, for example, in 2010 the percentage of
disadvantaged households that contains older people (singles or couples)
declines from 11 per cent to under seven per cent when the conventional
poverty measure is replaced by the consistent poverty measure. In
contrast, the same switch results in the proportion consisting of
households with children (couples and sole parents) increasing from 55
per cent to 62 per cent.
When it comes to comparing the situation between years, the changes
over the period resulted in a substantial decline in the proportion of
those in poverty or deprivation who have children (from 78 per cent to
62 per cent for the consistent poverty measure) and a slight increase in
the proportion that consists of older households (particularly older
couples). Single working-age people also represented a larger proportion
of those identified as poor or deprived in 2010 than they did in 2006.
Clearly, the differences described above (within and between years) are
not just of academic interest, but have profound implications for the
social policy settings that affect the living standards of people in
different circumstances and at different stages of the life cycle.
Conclusions
This article has used survey data to examine changes in different
concepts of poverty in the wake of the GFC, between 2006 and 2010. This
four-year window includes the period of intense community concern and
policy activity that followed the onset of the financial crisis towards
the end of 2008. A range of indicators have been used, with a focus on
investigating the sensitivity of the findings to the methods used to
produce them. A secondary objective has been to demonstrate the
challenges involved in using survey data to estimate changes in poverty
rates using different approaches. The analysis indicates that some of
the findings are highly sensitive to which measures and methods are
used, and this suggest that caution should be applied when relying on
any single measure or method to examine the issues addressed in this
paper.
Most of the indicators examined indicate that things improved
overall between 2006 and 2010, although it is not possible to attribute
this solely to the GFC having a more benign effect than had been
predicted because many other things changed over the period (including
the policy response to the crisis, which explicitly directed additional
resources to many low-income Australians). There are, however, marked
differences in the changing fortunes of different groups, although some
of these are sensitive to the methods used to estimate circumstance and
change. More importantly, the different indicators paint a different
picture of which groups are most affected by poverty at a point in time,
with those based on deprivation alone or in combination with income
revealing a different profile from those based on conventional
income-based poverty lines.
Finally, it is important to acknowledge that although the results
presented here do not appear to confirm the picture produced by some of
the welfare agency studies cited earlier in the paper, this does not
imply that those results are invalid. There is evidence (from the HILDA
survey) which suggests that the GFC exerted a detrimental short-run
impact on many of those who were already doing it tough. Our results
indicate that by 2010 things had improved for many of these groups, but
this does not necessarily mean that they did not experience significant
financial and other forms of social and psychological distress in the
period immediately following the financial crisis.
Acknowledgements
This paper was originally presented at the Australian Conference of
Economists in July 2011. The authors wish to acknowledge the helpful
comments provided by Roger Wilkins, who was the Discussant of the paper
at the conference, and those provided by two anonymous referees. The
usual caveats apply.
References
Access Economics (2008) The Impact of the Global Financial Crisis
on social services in Australia, Canberra, Access Economics.
Atkinson, A. B. (2007) 'EU social policy, the Lisbon agenda
and re-imagining social policy', Henderson Oration, 21 February,
University of Melbourne
http://public-policy.unimelb.edu.au/_data/assets/pdf_file/0015/509010/atkinson. pdf
ABS (Australian Bureau of Statistics) (2003) 'Revised
household income distribution statistics'. In Australian Economic
Indicators, June, Cat. No. 1350.0.
-- (2009a) Household Income and Income Distribution, Australia,
2007-08, Cat. No. 6523.0.
-- (2009b) 'Report on low consumption possibilities research
project'. In Methodological News, Cat. No. 1504.0
-- (2010) Measures of Australia's Progress 2010, Cat. No.
1370.0.
-- (2011) Household Income and Income Distribution, Australia,
2009-10, Cat. No. 6523.0.
ACOSS (Australian Council of Social Service) (2008) Who Is Missing
Out? Hardship Among Low Income Australians, Sydney, Australian Council
of Social Service.
Australian Government (2009) A Stronger, Fairer Australia. National
Statement on Social Inclusion, Canberra, Department of Prime Minister
and Cabinet.
-- (2012) Social Inclusion in Australia. How Australia is Faring,
2nd Edition, Canberra, Department of Prime Minister and Cabinet.
Australian Social Inclusion Board (2009) A Compendium of Social
Inclusion Indicators. How's Australia Faring? Canberra, Department
of Prime Minister and Cabinet.
Billing, J., Kindermann, B., McColl, B. & Rolfe, N. (2010)
'Improving our understanding and measures of economic hardship:
Australia's development of a low consumption possibilities
framework', paper presented to the 31st General Conference of the
International Association for Research on Income and Wealth, St Gallen,
Switzerland, 22-28 August.
Bray, J. R. (2001) Hardship in Australia: an analysis of financial
stress indicators in the 1998-99 Australian Bureau of Statistics
Household Expenditure Survey, Occasional Paper No. 4, Canberra,
Department of Family and Community Services.
Buddelmeyer, H. & Verick, S. (2008) 'Understanding the
drivers of poverty dynamics in Australian households', Economic
Record, 84 (266), 310-21.
Butterworth, P. & Crosier, T. (2006) 'Deriving a measure
of financial hardship from the HILDA survey', Australian Social
Policy 2005, 1-12.
Cappellari, L. & Jenkins, S.P. (2007) 'Summarising
multiple depriviation indicators'. In S.P. Jenkins & J.
Micklewright (eds), Inequality and Poverty Re-examined, Oxford, Oxford
University Press.
FAHCSIA (Department of Families, Housing, Community Services and
Indigenous Affairs) (2009) Preliminary Report: Survey on the Impact on
Families of the Economic Downturn, Canberra, FAHCSIA.
Gordon, D. (2006) 'The concept and measurement of
poverty'. In C. Pantazis, D. Gordon & R. Levitas (eds), Poverty
and Social Exclusion in Britain. The Millennium Survey, Bristol, Policy
Press.
Hahn, M. & Wilkins, R. (2008) 'A muitidimensional approach
to investigation of living standards of the low-paid: income, wealth,
financial stress and consumption', Research Report No. 5/09,
Melbourne, Australian Fair Pay Commission.
Harding, A., Lloyd, R. & Greenwell, H. (2001) Financial
Disadvantage in Australia 1990 to 2000. The Persistence of Poverty in a
Decade of Growth, Sydney, The Smith Family.
Harmer, J. (2009) Pension Review Report, Canberra, Department of
Families, Housing, Community Services and Indigenous Affairs.
Headey, B. (2007) 'Financial poverty in Australia. Developing
a valid measure based on wealth, income and consumption for use by
Commonwealth and State governments', paper presented to the HILDA
Survey Research Conference, Melbourne, 19-20 July.
Headey, B., Marks, G. & Wooden, M. (2005) 'The dynamics of
income poverty in Australia: evidence from the first three waves of the
HILDA survey', Australian Journal of Social Issues, 40 (4), 541-52.
Jenkins, S. P. & Micklewright, J. (2007) 'New directions
in the analysis of inequality and poverty', In S.R Jenkins & J.
Micklewright (eds), Inequality and Poverty Re-examined, Oxford, Oxford
University Press.
Mack, J. & Lansley, S. (1985) Poor Britain, London, George
Allen and Unwin.
Marks, G.N. (2007) Income poverty, subjective poverty and financial
stress, Social Policy Research Report No. 29, Canberra, Department of
Families, Housing, Community Services and Indigenous Affairs.
Melbourne Institute (2011) Poverty lines, Australia. December
quarter 2010, Melbourne Institute of Applied Economic and Social
Research, University of Melbourne.
Nolan, B. & Marx, I. (2009) 'Economic inequality, poverty,
and social exclusion'. In W. Salverda, B. Nolan & T.M. Smeeding
(eds), The Oxford Handbook of Economic Inequality, Oxford, Oxford
University Press.
Nolan, B. & Whelan, C.T. (1996) Resources, Deprivation and
Poverty, Oxford, Clarendon Press.
OECD (2008) Growing Unequal? Income Distribution and Poverty in
OECD Countries, Paris, OECD.
OECD (2010) OECD Economic Surveys, Australia, Volume 2010/21,
Supplement 3, Paris, OECD.
Pantazis, C., Gordon, D. & Townsend, P. (2006) 'The
necessities of life'. In C. Pantazis, D. Gordon & R. Levitas
(eds), Poverty and Social Exclusion in Britain. The Millennium Survey,
Bristol, Policy Press.
Pech, J. (2011) Relative living standards and needs of low-paid
employees: definition and measurement, Research Report 2/2011,
Melbourne, Fair Work Australia.
Saunders, P. (2008) 'Social exclusion: challenges for research
and implications for policy', Economic and Labour Relations Review,
19 (1), 73 91.
-- (2011) Down and Out--Poverty and Exclusion in Australia,
Bristol, Policy Press.
Saunders, P. & Bradbury, B. (2006) 'Monitoring trends in
poverty and income distribution: data, methodology and
measurement', The Economic Record, 82 (258), 341 64.
Saunders, P. & Hill, P. (2008) 'A consistent poverty
approach to assessing the sensitivity of income poverty measures and
trends', Australian Economic Review, 41 (4), 371 88.
Saunders, P., Hill P. & Bradbury, B. (2007) Poverty in
Australia: sensitivity analysis and recent trends, SPRC Report 4/08,
Sydney, Social Policy Research Centre, University of New South Wales.
Saunders, P. & Naidoo, Y. (2009) 'Poverty, deprivation and
consistent poverty', The Economic Record, 85 (271), 417 32.
Saunders, P., Naidoo, Y. & Griffiths, M. (2007) Towards new
indicators of disadvantage: deprivation and social exclusion in
Australia, Sydney, Social Policy Research Centre, University of New
South Wales.
--(2008) 'Towards new indicators of disadvantage: deprivation
and social exclusion in Australia', Australian Journal of Social
Issues, 43 (2), 175 94.
Saunders P. & Sutherland, K. (2006) Experiencing poverty: the
voices of low income Australians. Towards New Indicators of Disadvantage
Project stage I: focus group outcomes, Sydney, Social Policy Research
Centre, University of New South Wales.
Saunders, P. & Wong, M. (2011a) 'The social impact of the
global financial crisis in Australia', Australian Journal of Social
Issues, 46 (3), 291 309.
--(2011b) 'Measurement and change in deprivation and exclusion
in Australia: a report on research in progress', paper presented to
the Second Townsend Memorial Conference, University of Bristol, 22 23
January.
--(2011c) "Pension adequacy and the pension review',
Economic and Labour Relations Review, 22 (3), 7 26.
--(2012) Promoting Inclusion and Combating Deprivation: Recent
Changes in Social Disadvantage in Australia, Sydney, Social Policy
Research Centre, University of New South Wales.
Scutella, R., Wilkins, R. & Kostenko, W. (2009) Estimates of
poverty and social exclusion in Australia: a multidimensional approach,
Working Paper No. 26/09, Melbourne, Melbourne Institute (if Applied
Economic and Social Research.
Sen, A. K. (1985) Commodities and Capabilities, Amsterdanl,
North-Holland.
Smeeding, T.M. (2006) 'Poor people in rich nations: the United
States in comparative perspective', Journal of Economic
Perspectives, 20 (1), 69-90.
Tanton, R., Vidyattama, Y., McNamara, J., Vu, Q.N. & Harding,
A. (2009) 'Old, single and poor: using microsimulation and
microdata to analyse poverty and the impact of policy change among older
Australians', Economic Papers, 28 (2), 102-20.
The Salvation Army (2010) Perceptions of poverty: an insight into
the nature and impact of poverty in Australia, Sydney, The Salvation
Army http:// www.salvos.org.au/about-us/media-centre/documents/PerceptionsofPoverty.pdf
Townsend, P. (1979) Poverty in the United Kingdom, Harmondsworth,
Penguin Books.
Whelan, C. T., Nolan, B. & Maitre, B. (2006) Measuring
consistent poverty in Ireland with EU SILC data, Working Paper No. 165,
Dublin, The Economic and Social Research Institute.
Wesley Mission (2010) Making ends meet: financial stress is not
just about money, Sydney, Wesley Mission
http:Hwww.wesleymission.org.au/Research/
Financial_Stress/Report/Online%20report/The%20Wesley%20Report%20
Oct%202010.pdf
Wilson, S., Meagher, G., Gibson, R., Denemark, D. & Western, M.
(2005) Australian Social Attitudes: The First Report, Sydney, UNSW
Press.
Wilkins, R. (2008) 'The changing socio-demographic composition
of poverty in Australia: 1982 to 2004', Australian Journal of
Social Issues, 42 (4), 481-501.
Wilkins, R., Warren, D., Hahn, M. & Houng, B. (2011) Families,
incomes and jobs, volume 6: a statistical report on waves 1 to 8 of the
Household, Income and Labour Dynamics in Australia Survey, Melbourne:
Melbourne Institute of Applied Economic and Social Research.
Endnotes
(1.) Although much has been made of the sensitivity of poverty
rates to where the poverty line is set, less attention has been paid to
the fact that they are equally sensitive to reporting errors in the data
used to estimate median income and hence to set the poverty line: see
Saunders anti Bradbury (2006).
(2.) The seven indicators, all of which are the result of a
shortage of money, are: could not pay electricity, gas or telephone
bills on time; could not pay the mortgage or rent on time; pawned or
sold something; went without meals; was unable to heat home; asked for
financial help front something, or friends; and asked for help from a
welfare or community organisation. The incidence of all but the first
two indicators increased between 2007 and 2009.
(3.) Because voting is compulsory in Australia, the electoral rolls
provide a reasonably good sampling flame for the adult (aged 18 and
over) population. There is an under-representation of younger people
(aged 25 and below) on the electoral rolls, although this cannot explain
the under-representation of younger people in the survey data, as
Saunders and colleagues (2007: 24) have shown.
(4.) Returned surveys indicating an incorrect address were excluded
before estimating response rates.
(5.) Income refers to gross household weekly income and was
collected in $100 ranges up to $1000, in $500 ranges between $1,000 and
$2,000 and an open-ended top category for incomes over $2,000. There was
also a category for zero or negative income.
(6.) Disposable income was derived from the survey data by applying
the tax scales in each year to an estimate of gross income (derived
using the randomised assignment method described later). Where an exact
income figure (or a precise source like the age pension that implied an
exact figure) was provided it was used. The estimates were then adjusted
using the modified OECD equivalence scale to derive equivalised income.
Zero and negative reported incomes were excluded from the analysis--this
resulted in the removal of less than one per cent of cases.
(7.) For those respondents whose main source of income was an age
pension, the randomised method has been applied so that no
respondent's income can fall below the maximum rate of pension
(single or married, as appropriate).
(8.) It should be noted that the alternative income imputation
methods and the different treatment of household structure both cause
the value of median income and hence the poverty lines used in Table 2
to change.
(9.) New questions were introduced in the 2010 survey to address
the limitations of the household structure questions asked in 2006. They
involved asking respondents to specify the numbers of adults (between
the ages of 18 and 64, and 65 and over) and children (aged under 18) in
their household. This was not asked explicitly in 2006 and assumptions
had to be made about how many adults (and children) were living in group
households.
(10.) The survey estimates of median (equivalised) disposable
income on which the poverty lines in Table 2 are based are around 32 per
cent below those derived for relevant years from the ABS Survey of
Income and Housing (ABS 2011: Table 1) because of the under-reporting of
higher incomes in the CUPSE and PEMA surveys. This obviously lowers the
poverty line, but the impact on the poverty rate also depends upon the
extent of mis-reporting of income among households close to the poverty
line and the accuracy of the method used to derive a point estimate of
income from the grouped survey data. The increase in survey-based median
income between 2006 and 2010 is consistent with the rise in seasonally
adjusted household disposable income per head published by the Melbourne
Institute (2011).
(11.) The consistent poverty estimates in Table 4, which are based
on 60 per cent of median income and at least 4 deprivations, are very
close in both years to the estimates of the incidence of multiple
disadvantage produced recently by the Australian Government (2012: 23).
Table 1: Support for items identified as essential in 2006 and 2010
(weighted percentages)
Item 2006 2010
Warm clothes and bedding, if it's cold 99.8 99.9
Medical treatment if needed 99.9 99.9
Able to buy medicines prescribed by a doctor 99.3 99.5
A substantial meal at least once a day 99.6 99.4
Dental treatment if needed 98.5 98.4
A decent and secure home 97.3 97.1
Children can participate in school activities & 94.7 95.8
outings
A yearly dental check-up for children 94.3 94.9
A hobby or leisure activity for children 92.5 92.7
Up to date schoolbooks and new school clothes 88.5 92.8
A roof and gutters that do not leak 91.5 91.3
Secure locks on doors and windows 91.6 92.4
Regular social contact with other people 92.5 91.6
Furniture in reasonable condition 89.3 89.0
Heating in at least one room of the house 87.4 87.0
Up to $500 in savings for an emergency 81.1 81.4
A separate bed for each child 84.0 81.3
A washing machine 79.4 77.7
Home contents insurance 75.1 72.4
Presents for family or friends at least once a year 71.6 71.4
Computer skills 68.7 72.6
Comprehensive motor vehicle insurance 60.2 59.9
A telephone 81.1 59.7
A week's holiday away from home each year 52.9 53.9
Source: Saunders and Wong (2011c: Table 1).
Table 2: Poverty rates in 2006 and 2010--sensitivity analysis
(unweighted percentages)
Household type Original-- Revised--
income set at incomes
mid-points (1) randomised (2)
2006
Poverty line = 50% of median income
Single adult (working age, WA)) 9.2 9.7
Single adult (older person, OP) 7.8 2.1
Couple, no children (WA) 3.4 3.4
Couple, no children (OP) 14.1 3.6
Couple plus children 15.4 11.7
Sole parent households 28.9 28.0
All households 14.4 11.1
Poverty line = 60% of median income
Single adult (WA) 28.1 18.5
Single adult (OP) 50.0 2.1
Couple, no children (WA) 8.6 6.6
Couple, no children (OP) 24.1 4.9
Couple plus children 21.7 19.3
Sole parent households 33.3 38.5
All households 23.8 16.9
Household type "Comparable" "Best"
estimates estimates
(3) (4)
2010
Poverty line = 50% of median income
Single adult (working age, WA)) 10.1 10.1
Single adult (older person, OP) 1.4 1.4
Couple, no children (WA) 5.3 5.3
Couple, no children (OP) 4.3 4.3
Couple plus children 6.9 6.9
Sole parent households 21.2 21.9
All households 11.4 10.9
Poverty line = 60% of median income
Single adult (WA) 16.7 16.7
Single adult (OP) 4.1 4.1
Couple, no children (WA) 8.5 8.5
Couple, no children (OP) 6.2 6.2
Couple plus children 12.0 12.2
Sole parent households 35.2 35.2
All households 16.7 16.7
Note: The 'All households' category includes those containing
more than two adults, i.e. group households, parents living with
non-dependent children and related adults living together.
Table 3: The incidence of deprivation and mean deprivation scores in
2006 and 2010 (unweighted percentages)
Household type At least 2 At least 4 Mean
deprivations deprivations deprivation
(DEP (DEP score
[greater [greater
than or than or
equal to] 2) equal to] 4)
2006
Single adult (WA) 39.3 23.4 2.12
Single adult (OP) 25.3 12.7 1.29
Couple, no children (WA) 17.5 7.5 0.82
Couple, no children (OP) 11.6 5.5 0.52
Couple plus children 24.6 12.4 1.22
Sole parent households 54.9 37.7 3.38
All households 25.7 13.7 1.32
Household type At least 2 At least 4 Mean
deprivations deprivations deprivation
(DEP (DEP score
[greater [greater
than or than or
equal to] 2) equal to] 4)
2010
Single adult (WA) 30.9 15.9 1.66
Single adult (OP) 16.6 9.2 0.84
Couple, no children (WA) 17.4 7.5 0.82
Couple, no children (OP) 9.4 3.8 0.46
Couple plus children 23.6 12.8 1.18
Sole parent households 40.5 27.0 2.60
All households 22.1 11.6 1.15
Table 4: The impact of alternative definitions of consistent poverty
(unweighted percentages)
Household type Poverty line = 50% median income AND:
DEP [greater DEP [greater
than or than or
equal to] 2 equal to] 4
2006
Single adult (WA) 7.7 5.1
Single adult (OP) 1.4 0.7
Couple, no children (WA) 1.0 0.6
Couple, no children (OP) 1.0 1.0
Couple plus children 5.9 4.1
Sole parent households 13.9 8.9
All households 5.9 3.8
Household type Poverty line = 50% median income AND:
DEP [greater DEP [greater
than or than or
equal to] 2 equal to] 4
2010
Single adult (WA) 6.6 4.6
Single adult (OP) 0.0 0.0
Couple, no children (WA) 2.0 1.5
Couple, no children (OP) 1.6 0.9
Couple plus children 4.2 3.1
Sole parent households 10.4 7.6
All households 5.6 3.8
Household type Poverty line = 50% median income AND:
DEP [greater DEP [greater
than or than or
equal to] 2 equal to] 4
2006
Single adult (WA) 13.3 8.7
Single adult (OP) 1.4 0.7
Couple, no children (WA) 1.6 0.8
Couple, no children (OP) 1.0 1.0
Couple plus children 9.8 7.0
Sole parent households 22.8 16.5
All households 8.9 6.0
Household type Poverty line = 50% median income AND:
DEP [greater DEP [greater
than or than or
equal to] 2 equal to] 4
2010
Single adult (WA) 10.7 7.1
Single adult (OP) 0.7 0.0
Couple, no children (WA) 2.7 2.0
Couple, no children (OP) 2.2 0.9
Couple plus children 7.1 5.1
Sole parent households 16.0 10.4
All households 7.9 5.2
Table 5: The composition of the poor using alternative definitions
(percentages)
Household type Poverty Deprivation Consistent
(60% of (DEP [greater poverty
median) than or (60% AND DEP
equal to] 2) [greater than
or equal to]
2)
2006
Single adult (WA) 10.1 13.1 14.5
Single adult (OP) 0.9 6.3 1.1
Couple, no children (WA) 9.3 15.1 4.5
Couple, no children (OP) 4.2 6.0 1.7
Couple plus children 58.0 44.7 58.1
Sole parent households 17.5 14.8 20.1
All households 100 100.0 100.0
Household type Poverty Deprivation Consistent
(60% of (DEP [greater poverty
median) than or (60% AND DEP
equal to] 2) [greater than
or equal to]
2)
2010
Single adult (WA) 14.0 13.6 18.1
Single adult (OP) 2.5 5.9 0.9
Couple, no children (WA) 19.9 21.5 13.0
Couple, no children (OP) 8.5 7.0 6.0
Couple plus children 39.4 42.1 47.4
Sole parent households 15.7 9.9 14.7
All households 100.0 100.0 100.0
