Income distribution in Australian states

Although the Occupy movement is under less media spotlight than in late 2011, there have been numerous claims in the media of increasing wealth inequality in the developed world, and most notoriously in the United States. The 2011 Australian census presents an opportunity to analyse income distribution and subsequently determine to what degree claims of inequality are borne out in the data.

Household income distributions across Australian states and territories, based on 2011 census data

At the lower end of the income scale, Tasmania stands out as having a significant spike with its mode at 14.5% of respondent households reporting income of $900 per week, which is equivalent to $47,000 per year. South Australia also has a similar, but less extreme, mode with 12% of households reporting income of $900 and similarly 12% reporting $1,000 per week. The tall, narrow profile of these the income distributions for these two southern states paints a picture of states with relatively high levels of income equality. These higher levels of income equality are quantified through the smallest standard deviations of all states (see chart below).

The big population states: NSW, Victoria and Queensland

While Tasmania and South Australia stand out as the lowest earning states, buried beneath them in the chart is a tangle of lines representing the core of Australia's population: New South Wales, Victoria and Queensland. These three states comprise 77% of the Australian population, and follow very similar income distributions (sharing correlation coefficients > 0.98). Victoria has the lowest mean income and smallest standard deviation, News South Wales has the highest, and Queensland is in between; however these differences are subtle and all three have similar levels.

Of the three high population states, one element of note is the minor spike in New South Wales at the $2,600 per week level ($135,662 per annum). This income level is at the 97th percentile in New South Wales, the 98th percentile in Queensland and the 99th percentile in Victoria.

Overall, however, these three states can be summarised as residing in the middle in terms of both mean income and levels of inequality. Victoria, New South Wales and - to some degree - Queensland export a relatively large quantity of services. While this can temper a boost from resource booms, it also reduces their exposure to negative phases in the commodity cycle. This level of diversification, combined with higher populations,  leads to more moderate income distribution profiles.

Measurements of inequality

Limitations on the available household income data - such as a maximum possible result of $5,000 per week - mean that an accurate Gini coefficient is not possible to derive. Further to this, income as a metric represents something different from wealth. As an alternative indicator, I am using standard deviation as an inequality metric. This measurement looks at the spread along the income distribution levels from the mean; the theory is that the wider the spread, the greater the inequality because there are more poor people and more wealthy people further from the mean. Whilst this is an imperfect indicator, it is useful for quantifying the range and relative frequency of incomes encountered within a given state.

Income inequality ranked using standard deviation as a measure of inequality. Higher standard deviation indicates higher levels of inequality. Mean income for each state is also shown.

When using standard deviation as an income inequality metric, the Northern Territory and the ACT show the largest spread of incomes in the country. Tasmania and South Australia show the lowest levels of inequality, although the potential range of incomes are constrained by relatively low average earnings in these states.

Western Australia

Western Australia follows a broadly similar distribution to New South Wales, Queensland and Victoria, albeit shifted to the right and with a significant shoulder around the $1,800 per week / $94,000 per year level. Its rightward position relative to New South Wales, Queensland and Victoria indicates that its earnings are higher. This view is backed up by a mean income of $1,465, third in the country and only behind the two lowly populated territories.

Western Australia's standard deviation is only $3 more than New South Wales, yet its mean earnings are $131 per week larger. This indicates that the lower earners in Western Australia have a larger income than their equivalents in New South Wales, and therefore possibly providing some evidence the mining boom is benefiting the broader economy. It should be noted though that these metrics do not take into account relative cost-of-living levels within Western Australia and New South Wales.

The two territories of Australian Capital Territory and the Northern Territory

The Northern Territory and the Australian Capital Territory have the highest earners of the states and territories in Australia. Both present an interesting and distinct distribution from the states.

Income distribution in the Northern Territory and Australian Capital Territory. Figures are based on the 2011 Australian Census.

The Australian Capital Territory has by far the highest mean household income at $1,909 per week and has its mode (i.e. peak value) at $1,800. Its distribution is offset significantly to the right of the states, reflecting the higher mean, and it displays a significant shoulder around the $2,100 level, exhibiting bimodal characteristics. This second peak can perhaps be explained by a groundswell of territory or Commonwealth public sector salaries supporting households that earn approximately $110,000 per year.

The results from the Northern Territory also exhibit some interesting traits. At $1,474, its mean is significantly below the ACT, yet its most frequent income level - at $1,900 - is higher. This difference between its mean and mode is caused by a heavy tail at its lower income levels, which is pulling down the mean. The Northern Territory also exhibits a significant amount of households with zero income, which also reduces the mean, while leaving its mode unchanged.

Relative income distribution standard deviation by state. The Northern Territory has both the largest absolute standard deviation, and the largest relative standard deviation. ACT has the lowest relative standard deviation.

The chart above looks at the relative standard deviation across state incomes. Notice that the Australian Capital Territory and the Northern Territory are at opposite ends of the spectrum. This further paints a picture of significant wealth inequality in the Northern Territory. It has the second highest mean income in Australia, but also the largest range of incomes relative to its average.

The 2011 Australian census captured that 27% of residents of the Northern Territory are indigenous Australians. It has been reaping many of the economic gains from the recent mining boom, and despite only having 211,945 residents - i.e. 1% of the population - it contains Australia's 20th highest earning town in Nhulunbuy. Its capital, Darwin, contains some of the wealthiest city centre residents in the country. These are the underlying details that support a picture of high levels of income inequality.

Canberra, in contrast, has managed to sustain both extremely high levels of average income and a relatively narrow band of income distribution, indicating that the territory has its wealth fairly evenly distributed.

Of course, with only around 2.5% of the Australian population residing in these two territories, statistical extremities can perhaps be expected. It is perfectly plausible that perhaps buried within outback Queensland or Western Australia, there are similar levels of income disparity to the Northern Territory. Or perhaps swathes of the Perth area contain similar patterns of high wealth to Canberra.

As an aside, it is worth paying attention to the relative standard deviation in New South Wales. It sits at the top of this particular ranking, alongside the Northern Territory. This metric indicates a high level of inequality relative to other areas in Australia. I have been performing further analysis on income distribution in major Australian cities, however that is a subject for a future post. I'm sure that particular future post will shed further light on this.

The ABS's Statistical Area Level 1 (SA1) data sets have been used as a proxy for determining income levels. These provide a level of data granularity to geographical areas containing approximately 500 people.