Poverty rates by electoral divisions, 2006

Research Paper Index

Research Paper no. 27, 2008-09

Poverty rates by electoral divisions, 2006

Tony Kryger
Statistics and Mapping Section


Methodology and definitions
Missing data


Executive Summary

This paper provides 2006 estimates, for each Commonwealth Electoral Division, of the number of persons living in poverty and the proportion which this represents of the total population (i.e. poverty rate). The estimates were derived from synthetic small area data produced by the National Centre for Social and Economic Modelling (NATSEM) at the University of Canberra.



Based on a measure of relative poverty that is described in detail below, 11.7 per cent of all Australians (or more than one in nine Australians) were living in poverty in 2006. There were, however, significant regional variations. On a state and territory basis, rates of poverty ranged from 6.2 per cent in the ACT to 14.7 per cent in Tasmania. On an electorate basis, rates ranged from 5.6 per cent in Mitchell (NSW) to 19.0 per cent in Cowper (NSW).

Note that the poverty estimates presented in this paper are synthetic (or model based) estimates and, therefore, should be regarded as indicative only.

It is also important to note that the poverty estimates in this paper cannot be compared with earlier estimates produced by the Parliamentary Library due to numerous methodological differences underlying production of the small area estimates, including refinements in reweighting procedures and changes in definitions, geographical units and source data.

Methodology and definitions

The National Centre for Social and Economic Modelling (NATSEM) has produced a series of population and poverty estimates at the statistical local area (SLA) level[1]. The estimates were generated using spatial microsimulation techniques, and the initial development and validation of the methodology underlying the estimates was funded by an Australian Research Council Discovery Grant (DP0664429).

NATSEM estimates at the SLA level were derived by taking unit record data from the 2002–03 and 2003–04 Australian Bureau of Statistics Surveys of Income and Housing and applying a set of weights to the data. The weights, representing households in SLAs across Australia, were created by matching a number of benchmarks from the 2006 Census of Population and Housing with data from the Surveys of Income and Housing. The benchmarks included type of household, tenure type, household income, dwelling structure, household mortgage and several other variables. Incomes from Surveys of Income and Housing were converted to 2006 dollars using the change in average weekly earnings. NATSEM estimates of population and poverty at the SLA level were then aggregated to electoral divisions by the Parliamentary Library using an SLA to Commonwealth electoral division concordance.

As there is no universally accepted definition of poverty, NATSEM has adopted a widely used measure of relative poverty in which persons are defined to be in poverty if they are living below some community standard. In line with current Australian practice, the poverty line has been set at 50 per cent of median disposable household income[2]—with income first adjusted for household size and composition, and with all members of a household whose income falls below the poverty line deemed to be in poverty. The scale used to adjust household disposable incomes is the modified OECD equivalence scale which gives a value of 1.0 for the first adult in a household, 0.5 for the second and subsequent adults, and 0.3 for any dependent children. Points are calculated for each household using this scale. Disposable household income is then divided by the number of points, to produce a measure of equivalised income for each household.[3]

The process of adjusting household disposable incomes using an equivalence scale recognises that there are economies of scale to be gained in household consumption. For example, by assigning a value of 1.0 to a lone person household, and a value of 2.1 to a standard family household consisting of a couple with two children, the OECD scale is assuming that a lone person household requires about 48 per cent [= (1 / 2.1) * 100] of the income of a standard family household to achieve the same standard of living.


Population and poverty estimates produced by NATSEM at the SLA level exclude persons living in non-private dwellings and non-classifiable households. Aggregations of these population numbers (to state or electorate levels)—the denominator used for the calculation of poverty rates—therefore, will not be exactly the same as published population data from the 2006 Census of Population and Housing.

Missing data

As the results produced by spatial microsimulation are estimates, NATSEM has sought to validate the results to ensure they are sufficiently reliable. In total, NATSEM was forced to exclude 163 SLAs from its analysis, the major reason being an inability to calculate household weights for these areas—generally these were highly unusual SLAs, such as military bases or industrial areas, or those with low populations. Since many of the areas excluded were sparsely populated remote areas, total exclusions covered only 0.72 per cent of the Australian population. Unfortunately, however, the excluded SLAs tended to be concentrated in just a few electorates, thereby affecting the accuracy of the poverty estimates produced for those electorates.

Table 1 lists those electorates which were significantly affected by exclusions. For this paper, ‘significantly affected’ means the excluded SLAs made up more than five per cent of the overall population of an electorate. There were six electorates in this category—Kalgoorlie, Kennedy, Leichhardt, Lingiari, Solomon and Sydney. A further 20 electorates were affected to a lesser extent, with exclusions comprising less than two per cent of the overall population in 16 of these electorates and between two and five per cent of the overall population in the remaining four electorates.

Table 1. Electorates 'significantly affected' by exclusions, 2006


Population of excluded SLAs

After exclusions(a)


Number of 


Percentage of
total electorate

Number  in

rate (%)

Kalgoorlie (WA)


18 484


8 615


Kennedy (Qld)


9 352


15 255


Leichhardt (Qld)


15 006


12 604


Lingiari (NT)


43 051


4 218


Solomon (NT)


5 075


6 132


Sydney (NSW)


22 273


12 315


(a) After SLAs with missing data were excluded from the analysis. In other words, the number of persons in poverty and the poverty rate were calculated on the basis of only those SLAs for which data were available.

Source: Derived from NATSEM synthetic estimates of population and poverty data which were produced with NATSEM's SpatialMSM/08C model and applied to data from the 2002–03 and 2003–04 ABS Surveys of Income and Housing.

As a result of exclusions, the number of persons identified as being in poverty in each electorate in Table 1 is potentially severely understated. In addition, since data are missing for a number of SLAs, the poverty rate shown for each electorate should not be regarded as representative of the electorate as a whole.


States and territories

Based on the above definition, 11.7 per cent of all Australians were in poverty in 2006  (Table 2). The state with the highest poverty rate was Tasmania at 14.7 per cent, followed by South Australia and New South Wales with 13.1 and 12.4 per cent of their respective populations in poverty. The ACT had the lowest poverty rate of 6.2 per cent. The poverty rate for the Northern Territory was not available owing to the significant number of SLAs in the Northern Territory for which there were no valid data.

Table 2. Poverty rates(a) by state, 2006


Number  in poverty

Poverty rate  (%)

New South Wales

738 441



530 734



378 920


South Australia

181 423


Western Australia

177 773



63 957


Northern Territory

     -----insufficient data-----


18 194



2 099 792


(a) Percentage of total population in each state who were living in poverty.

(b) Included in Australia are those SLAs in the Northern Territory for which data were available.

Source: Derived from NATSEM synthetic estimates of population and poverty data which were produced with NATSEM's SpatialMSM/08C model and applied to data from the 2002–03 and 2003–04 ABS Surveys of Income and Housing.

Electoral divisions

The electorate with the lowest poverty rate was Mitchell in NSW at 5.6 per cent, closely followed by Canberra at 5.7 per cent (Table 3). Not surprisingly, of the 10 electorates with the lowest poverty rates, four were to be found in a cluster to the north of Sydney Harbour. These were Bradfield, Warringah, Mackellar and North Sydney, all with poverty rates below seven per cent. The highest poverty rate was in Cowper in NSW with 19.0 per cent of its population (or almost one in every five persons) in poverty. Interestingly, while NSW electorates accounted for six of the 10 electorates with the lowest poverty rates, they also accounted for seven of the 10 electorates with the highest poverty rates.

The average poverty rate rose from 10.2 per cent for outer metropolitan electorates, to 10.8 per cent for inner metropolitan electorates, to 12.7 per cent for provincial electorates, and to 13.8 per cent for rural electorates.




[1].       The SLA is a general purpose spatial unit. In aggregate, SLAs cover the whole of Australia.

[2].       Disposable household income is gross household income less the value of income tax and Medicare levy to be paid on the gross income. Median disposable household income is arrived at by first ranking all persons by the income of the household in which they reside. Median household income is that level of income which has an equal number of persons receiving above as below it.

[3].       For a more detailed discussion of equivalised income see Australian Bureau of Statistics (ABS), Household Income and Income Distribution, 2005–06 (Cat. no. 6523.0), ABS, Canberra, 2007.


For copyright reasons some linked items are only available to members of Parliament.