It will not surprise anyone that Indigenous people are twice as likely as other Australians to be in the lowest income quintile, and almost four times less likely to be in the highest quintile (see Table 8.1). Indigenous people are more likely than other Australians to rely on income from government pensions and allowances and be unable to raise $2000 within a week for something important. This is consistent with Indigenous employment disadvantage and the fact that the mean income of Indigenous households is around 60 per cent of that of other households (for discussion of the former, see Chapman & Gray in this volume).

a. The population for this table is persons aged 18 and over.

b. All the differences between Indigenous and non-Indigenous statistics in this table are significant at the 5% level.

Source: Table 4 in ABS (2004c)

a. The population for this table is Indigenous persons aged 15 and over.

b. The differences between remote and non-remote areas are statistically significant at the 5% level.

Source: Table 1 in ABS (2004c)

The OECD equivalence scale does not take into account differences in the cost of living across areas, so it is particularly important to look at the income distributions and other characteristics in various areas. Table 8.2 documents the income differences in remote and non-remote areas and illustrates that the overall distributions are not very different: there is a similar proportion of residents in all quintiles, except the second, in which remote areas have 12 percentage points more. The bottom line, in the bottom line, is that the average equivalised income in remote areas is lower, albeit only slightly lower. However, the proportion of people who cannot raise $2000 in a week for something important is much higher than in non-remote areas (73.0% and 47.3% respectively). This may indicate that the cost of living is placing more financial stress in remote areas or that Indigenous people in remote areas do not have access to social networks from which they might borrow the money (i.e. a form of social capital). The latter possibility can probably be discounted, since Indigenous people often are said to have a substantial amount of bonding social capital (Hunter 2004b). While Indigenous people in remote areas tend to have large social networks, the ‘problem of collective action’ may mean that it is difficult to coordinate the large number of competing claims on group resources (Olson 1965).

Before moving onto poverty measures, I want to describe financial stress by labour force status and remoteness. Figure 8.1 presents the incidence of being able to raise $2000 within a week in the bars. As in the rest of the monograph, the whiskers denote the 95 per cent confidence intervals, and as a rough rule if whiskers do not overlap, there is a significant difference between the estimates. As noted above, financial stress is higher in remote Australia. It is interesting to note that even non-CDEP employees in ‘very remote’ Australia have close to 60 per cent of workers suffering financial stress. This may reflect the methodology, with CAs being predominantly in such areas, but the size of the difference is remarkable and in my opinion unlikely to be explained by non-sampling error.

Also note that the level of financial stress is high for both unemployed and NILF categories in all areas. CDEP appears to have less financial stress in more urban environments but the reliability of these estimates is particularly low. That is, overall financial stress is high for all groups that are likely to be reliant on some form of government payment, and the best way out of financial stress is to have a job outside the CDEP scheme.

a. The difference between remote and non-remote statistics is significant at the 5% level.

Source: Table 18 in ABS (2004c)

It is possible to explore the differences between remote and non-remote areas in more detail by looking at selected income characteristics (table 8.3). Remote areas are more likely to report their main source of income as the CDEP scheme, but are less likely than other areas to rely for most of their income wages from other jobs or government pensions and allowances. The reason for this is simply that CDEP income is identified as a separate source of income. If one argues that CDEP is in some sense a government payment, albeit with a work requirement, then the reliance on government payments in remote areas is substantially higher than in other areas. Remote areas are substantially less likely than other areas to have had government pensions and allowances as their main source of income in the last two years, again presumably because of the flow of people through CDEP schemes.

How do Indigenous people access their money? People in remote areas are significantly less likely to have a bank account, access money over the counter or engage in internet banking. While the incidence of internet banking is much lower than for the rest of Australians in both areas, the particularly low level of internet banking in remote areas is due to the lack of access to personal computers and the internet in such areas (see Radoll in this volume).

Table 8.4 explores several factors that are likely to be correlated with equivalised income. In the first CAEPR Working Paper, ‘Three Nations Not One’, I described how Indigenous respondents to the 1994 NATSISS only had a weak correlation between such factors and equivalent income. That study was hindered by the grouped nature of much of the income data for the 1994 survey, which meant that my calculations were only rough approximations.

Source: Table 9 in ABS (2004c)

The 2002 data, which permits a more robust estimate of equivalised income than was possible in 1994 (at least using the CURF), shows that there is a significant difference between long-term health in the first quintile and the other quintiles. However, the incidence of such conditions is still very high in the top income group, with over one quarter having a long-term health condition. This level of illness is unprecedented in other high income groups in Australia. Similarly, almost one third of the fourth and fifth quintiles were current daily smokers. All of the variables in this table exhibit a significant difference between the bottom quintile and the fourth and fifth quintiles except risky alcohol consumption, which is slightly higher in the top income group. Indigenous people in the bottom quintile are significantly less likely than the top income group to have an educational qualification, be employed outside the CDEP scheme, or participate in the labour force. However, they are more likely to be in financial stress, and to have been arrested, incarcerated or a victim of physical or threatened violence.

Does the fact that 15.6 per cent of Indigenous people are in the top two quintiles of Australian household income (see Table 8.1)—and that these people appear to be somewhat better off than other Indigenous people—mean there is now an Indigenous ‘middle class’? While there may be some signs of an Indigenous middle class emerging, the rates of social ills and socioeconomic disadvantage are too high for the ‘rich’ groups in Table 8.4. Obviously, addressing income levels is not the only relevant policy. In spite of the conclusion reached, it is clear that one of the main correlates of high income status is securing a non-CDEP scheme job, and addressing Indigenous labour market disadvantage is likely to be the most effective policy emphasis.

NATSISS allows us to accurately benchmark the top-coding assumptions routinely used in analysis of grouped data in the respective censuses. In the absence of other data, researchers and policy-makers tend to assume that the average income in the top category is some multiple of the bottom end of the income range, usually a factor of around 1.5. The mean income for the 2002 NATSISS respondents in the top category used in the 2001 Census can be calculated to verify the  accuracy  of  this  assumption  for  Indigenous Australians. ^[6] The appropriate multiple of the bottom income range is 1.19 for remote areas and 1.32 for non-remote areas. Therefore, the usual assumptions were far too high for Indigenous people, especially in remote areas, which means previous estimates of average income will tend to understate Indigenous income disadvantage.

The basic measure of poverty is the headcount measure which can be used to illustrate the disadvantage of Indigenous Australians (that even the researchers at the CIS acknowledge). Table 8.5 presents the headcount measures calculated using a poverty line based on one half of the median equivalised income within the GSS. This was considered to be a reasonably robust poverty line by both sides of the poverty war debate. In one-family households, Indigenous people are twice as likely to be under the poverty line. In households with more than one family living in the dwelling, Indigenous people are over 14 times more likely to be classified as poor. Clearly, household structure is an important ‘driver’ of Indigenous poverty. The equivalence scales used to calculate equivalent income is defined on household structure, so the poverty rates are in a sense ‘driven’ by household structure.

Source: Customised cross-tabulations from the 2002 NATSISS and GSS RADL

Having a large number of dependants is another recipe for being in poverty for both Indigenous and non-Indigenous people (i.e. having five or more dependants). Having said this, poverty is also particularly pronounced in Indigenous households with only one dependant.

The variation in poverty by geography is particularly interesting. While poverty is lowest for Indigenous and other Australians in major cities, it is particularly high for Indigenous people living in inner regional areas. Therefore, remote areas are actually less poor than in regional and provincial areas. Despite the lack of an estimate for remote areas from the GSS, poverty for all Australians tends to increase as accessibility to services becomes more difficult.

The tension between the poverty estimates in Table 8.5 and the financial stress data in Figure 8.1 and Table 8.2 is worth noting. Financial stress is higher in remote areas, especially very remote areas. It is particularly noteworthy that non-CDEP scheme employment in very remote areas is also associated with high levels of financial stress. Given that this group is likely to be among the better educated (and most acculturated), this result is unlikely to be caused solely by non-sampling error. The tension between this direct question on financial stress and the standard measures of poverty means that we should question the latter, which are indirectly measured using a series of assumptions.

The theory of poverty is related to utility functions which are used to derive weights for particular households based on cost structures implicit in expenditure surveys (see Hunter, Kennedy & Biddle 2004). Underlying cost structures (and indeed preferences) are not necessarily the same for all sub-population groups, and therefore the imposition of the OECD equivalence scales on NATISS data can—and should—be contested. That is, the following question must be asked: how valid are the OECD equivalence scales when measuring Indigenous poverty?

In technical terms, equivalence scales estimate the ‘economies of scale’ of living in houses of various sizes and composition. The feasible range of equivalent incomes is bound by raw income and per capita income, with the latter representing an equivalence scale where the cost of each extra person is the same of those currently living in the house. At the other extreme, the equivalence scale underlying raw income assumes that each extra person living in a household cost nothing (i.e. perfect economies of scale). The OECD scale is somewhere between these two extreme assumptions.

Table 8.6 documents the headcount measure of poverty estimated by the proportion of people in households that are classified as poor using the various equivalence scales. Where there is only one person in the household, the actual equivalent income should not change when moving from raw income to per capita income, but the rest of the distribution does change which, in turn, results in a change to the poverty line. Consequently, per capita income pushes such households up the income distribution and they appear less poor. For households with two or more people in them, both the poverty line and the household’s equivalent income changes.

Source: Author’s calculations based on the 2002 NATSISS and GSS RADL

It is particularly interesting to note that Indigenous households with between two and four people in them have a tight range of poverty estimates, so we can be reasonably confident that these poverty estimates are reliable irrespective of the equivalence scale used. However, the story is different for large households with over five people in them, as poverty estimates vary wildly for both Indigenous and other Australians. This is not a criticism of the ABS because this has been observed earlier by many authors (Hunter, Kennedy & Biddle 2004; Whiteford 1985). The main point is that measured poverty in Table 8.6 is particularly sensitive to the choice of equivalence scales for large households, which Indigenous people frequently live in.