6. Poverty and income inequality

6.1 Adequacy of payments
6.2 Trends in poverty and inequality
6.3 The Henderson and other poverty lines
6.4 The benefit safety net

6.1 Adequacy of payments

Given that the alleviation of poverty is one of the primary objectives of the Australian income support system, it should be regarded as a key measure of the success or otherwise of social security spending.

In assessing the impact of the system on trends in poverty, there are a number of conflicting considerations to bear in mind. On the one hand, levels of unemployment and of non- participation among men have increased substantially. These trends, together with the increase in the extent of sole parenthood,³⁸ are likely to have contributed to an increase in the extent of vulnerability to poverty among people of workforce age. These trends are also likely to be associated with significant changes in the composition of the low-income population. In the past, it was the retired elderly who were most likely to be in poverty, whereas now it appears to be lone parents and the unemployed.

Earnings surveys also show a widening of wages dispersion among individuals over the past 15 years, particularly associated with a fall in the real earnings of the low paid. It is difficult to determine whether this is associated with changes in wage rates or changes in the composition of the workforce. On the other hand, the wages of women have increased relative to male earnings, thus tending to reduce dispersion. Analysis of data from income surveys over the 1980s (Eardley 1997) suggests an overall reduction in the size of the low paid workforce (from 15 to 13 per cent of all workers), because a marginal increase in low pay among men (from 10 to 11 per cent) was swamped by a large fall in the proportion of low paid women (from 23 to 16 per cent). The extension of more generous family assistance to those in low paid work could be expected to reduce poverty among this group.

In addition, the adequacy of basic income support payments has been very substantially increased over time (Table 15). For example, in 1996 the real level of the single adult rate of unemployment payments was 2.35 times its level in 1965, the real level of payment for a sole parent with one child was 1.6 times the 1965 level, the real level for a couple without children was 2.45 times the 1965 level, and the real level for a couple with two children was 2.56 times higher (2.83 times including rent assistance).

These real increases in basic payment rates were largely achieved in the period 1970 to 1975, when the McMahon and the Whitlam Governments increased pensions and benefits. Most basic payments (apart from the single rates of unemployment allowances) were subsequently indexed to inflation. Payments for children were substantially increased from the mid-1980s onwards and rent assistance was increased in the late 1980s and early 1990s. Virtually all payments are now indexed at least to the Consumer Price Index (CPI); pension rates are now also adjusted in line with Male Total Average Weekly Earnings (MTAWE).

These real increases in payment levels could be expected to have reduced poverty, as could the increase in the receipt of additional income from earnings and child support.

Table 15: Trends in the real value of social security payments for different family types, 1965 to 1997, $ per year ($ 1996-97)
Year	1965	1972	1976	1982	1983	1989	1996	1997
Unemployment allowance,no rent assistance
Single	$3,493	$4,006	$7,859	$6,476	$6,487	$7,720	$7,845	$8,088
Couple,no children	$6,033	$6,524	$13,079	$13,248	$13,224	$13,782	$14,793	$14,998
Couple,one child	$6,880	$7,980	$14,689	$14,795	$14,735	$15,681	$17,196	$17,456
Couple,two children	$7,939	$9,680	$16,403	$16,332	$16,371	$18,167	$20,330	$20,653
Couple,three children	$9,209	$11,661	$18,322	$17,972	$18,008	$20,307	$22,733	$23,111
Couple,four children	$10,479	$13,717	$20,293	$19,610	$19,647	$22,889	$26,063	$26,506
Pension,no rent assistance
Single,no children	$5,032	$5,556	$7,859	$7,947	$7,932	$8,269	$8,867	$8,991
Single,one child	$7,572	$8,968	$10,704	$10,411	$10,265	$10,978	$12,084	$12,362
Single,two children	$8,631	$10,668	$12,418	$12,133	$11,979	$13,464	$15,218	$15,559
Single,three children	$9,901	$12,649	$14,337	$14,138	$13,932	$15,604	$17,621	$18,017
Single,four children	$11,171	$14,705	$16,308	$16,144	$15,885	$18,186	$20,951	$21,412
Couple,no children	$9,217	$9,830	$13,079	$13,248	$13,224	$13,782	$14,793	$14,998
Couple,one child	$10,064	$11,293	$14,689	$14,795	$14,735	$15,681	$17,196	$17,456
Couple,two children	$11,122	$12,992	$16,403	$16,332	$16,371	$18,167	$20,330	$20,653
Couple,three children	$12,392	$14,973	$18,322	$17,972	$18,008	$20,307	$22,733	$23,111
Couple,four children	$13,662	$17,029	$20,293	$19,610	$19,647	$22,889	$26,063	$26,506
Unemployment allowance,with rent assistance
Single	$3,493	$4,006	$7,859	$6,476	$6,487	$8,734	$9,711	$10,024
Couple,no children	$6,033	$6,524	$13,079	$13,248	$13,224	$14,796	$14,997	$16,825
Couple,one child	$6,880	$7,980	$14,689	$14,795	$14,735	$16,702	$19,357	$19,715
Couple,two children	$7,939	$9,680	$16,403	$16,332	$16,371	$19,188	$22,497	$22,912
Couple,three children	$9,209	$11,661	$18,322	$17,972	$18,008	$21,328	$25,200	$25,666
Couple,four children	$10,479	$13,717	$20,293	$19,610	$19,647	$23,911	$28,535	$29,061
Pension,with rent assistance
Single,no children	$5,455	$6,206	$8,888	$8,652	$8,891	$9,282	$10,735	$10,927
Single,one child	$7,996	$10,086	$11,733	$11,114	$11,219	$11,999	$14,236	$14,621
Single,two children	$9,054	$11,786	$13,447	$12,836	$12,933	$14,485	$17,376	$17,818
Single,three children	$10,324	$13,767	$15,366	$14,842	$14,887	$16,626	$20,079	$20,572
Single,four children	$11,594	$15,823	$17,337	$16,848	$16,840	$19,208	$23,414	$23,967
Couple,no children	$9,640	$10,480	$14,108	$13,953	$14,184	$14,796	$14,997	$16,825
Couple,one child	$10,487	$11,943	$15,718	$15,501	$15,694	$16,702	$19,357	$19,715
Couple,two children	$11,546	$13,642	$17,432	$17,037	$17,331	$19,188	$22,497	$22,912
Couple,three children	$12,816	$15,623	$19,351	$18,677	$18,968	$21,328	$25,200	$25,666
Couple,four children	$14,086	$17,679	$21,322	$20,315	$20,607	$23,911	$28,535	$29,061

Source: Calculated from Moore and Whiteford, 1986 and personal calculations.

6.2 Trends in poverty and inequality

Despite the central importance of these issues, the problems of measurement are such that there is no consensus on the trend or level of inequality and poverty in Australia. This is shown in Tables 16 and 17. Some studies find rising inequality, while some find inequality to be stable or falling. The most notable point is that the trend in income inequality, whatever it might be, is not particularly strong, with many of the differences in the Gini coefficients unlikely to be statistically significant.

The estimates of poverty rates by different authors are more substantial. It is possible to argue either that poverty increased by around 60 per cent over the 1980s (Saunders 1994) or that it fell by more than 20 per cent (Harding and Mitchell 1992).

These contradictory results arise because researchers focused on different aspects of income and used differing methods of analysis. Technical choices may have a decisive influence on apparent trends, as well as on the picture of the underlying extent of poverty. Having said this, the lack of strong or clear trends is consistent with the view that, while underlying social and economic problems have increased, improvements in social security benefit levels and greater targeting of benefits will have tended to some extent to offset these trends.

6.3 The Henderson and other poverty lines

It should be particularly emphasised that the common perception that poverty in Australia has increased substantially over time reflects the very substantial limitations of the 'Henderson poverty line', which has been used since the late 1960s. When it was originally developed, the Henderson measure was closer to an 'absolute' poverty line, since it was defined as the basic wage plus child endowment for a couple with two children. This arbitrary assumption was justified on the basis that it produced a standard 'so austere as, we believe, to make it unchallengeable. No one can seriously argue that those we define as poor are not so' (Henderson, Harcourt and Harper 1970).

The Henderson poverty line was again used by the Commission of Inquiry into Poverty in the early 1970s, and by researchers and welfare organisations subsequently. This required some means of updating the Henderson line over time. Initially, the line was adjusted according to movements in average weekly earnings, but since the early 1980s it has been updated in line with household disposable income per capita (HDIPC) from the National Accounts.

Since the Poverty Inquiry Survey in 1973, HDIPC has grown in real terms by around 30 per cent, implying that the real level of living of someone just at the poverty line has risen by the same percentage over this period. This means that the statement that the 'poor have got poorer' when based on incomes compared to the Henderson line is not correct. In fact, the absolute living standards of those at the poverty line have risen. The real increases in social security rates noted above should also be remembered in this context.

Table 16: Results of studies of inequality in Australia
Study	Income Concept	Period	Data source	Main Results
Gregory 1993	Individual gross earnings,not equivalised	1976 to 1990	Weekly Earnings of Employees (WEED)	Growth in low paid and high paid jobs-the 'disappearing middle'
Borland and Wilkins 1996	Individual gross earnings,not equivalised	1975 to 1994	WEED; Income Distribution Survey (IDS)	Real weekly earnings of males fell at 10^th percentile and rose at 90th percentile.
Bradbury and Doyle 1992	Cash disposable income,equivalised	1983/84 to 1989/90	Microsimulation, IDS	Gini increased from.367 to.370
Saunders 1993	Cash disposable income,equivalised	1981/82 to 1989/90	IDS	Gini increased from.27 to.29
Harding 1994	Gross income, equivalised	1981/82 to 1989/90	IDS	No change in Gini
Whiteford 1994	Cash disposable income,equivalised	1982/83 to 1989/90	Microsimulation, IDS	Gini fell from.328 to.319
OECD 1995	Cash disposable income,equivalised	1981/82 to 1985/86	IDS	Gini increased from.287 to.295;P90/P10 fell from 4.05 to 4.01
Raskall and Urquhart 1994	Social wage income (health,schooling), equivalised	1982/83 to 1989/90	Microsimulation, IDS	Gini increased from.272 to.276
Harding 1995	Social wage income (health,education, housing,childcare), equivalised, after housing	1994	Microsimulation, IDS	Gini for cash disposable income of.308, for final income of.289
Johnson, Manning and Hellwig 1995	A.Cash disposable income, equivalised B. Social wage income (health, education, housing, childcare, concessions), equivalised	1981/82 to 1993/94	Microsimulation, HES	A. Gini fell from.308 to.296 B. Gini fell from.255 to.226
ABS 1996	Final income (social wage plus indirect taxes), not equivalised	1984 to 1993/94	Household Expenditure Survey (HES)	Q5/Q1 increased from 4.5 to 4.7
Gregory and Hunter 1995	Gross household income of areas, not equivalised	1976 to 1991	Census	Gini increased from.14 to.18; incomes fell for low income areas between 1976 and 1981 and rose for rich between 1981 and 1991

Note: The Gini coefficient ranges between 0 and 1 with a higher Gini implying greater inequality. The P90/P10 ratio is the income of the unit at the 90th percentile relative to that at the 10th percentile (from the bottom), with a higher ratio implying greater inequality. The Q5/Q1 ratio is the ratio of the income share o the richest 20 per cent to that of the poorest 20 per cent, with a higher ratio implying greater inequality.

Table 17: Results of studies of poverty in Australia
Study	Poverty Measure	Period	Data source	Main Results
Saunders and Matheson (1991)	Henderson, HDIPC	1981/82 to 1989/90	Income Distribution Survey (IDS)	Poverty rate rose from 9.2 to 12.8%
Bradbury and Doyle (1992)	A. Henderson, CPI B. Henderson, average survey income	1983/84 to 1989/90	Microsimulation, IDS	A. Poverty rate fell from 11.3 to 9.4% B.Poverty rate rose from 11.3 to 11.4%
Saunders (1990)	A. Henderson, CPI B. Henderson, HDIPC	1982/83 to 1989/90	Microsimulation, IDS	A. Poverty fell from 8.9 to 6.5% B. Poverty rose from 8.9 to 11.6%
Saunders (1994)	Henderson, HDIPC	1981/82 to 1989/90	IDS	Poverty rose from 10.7 to 16.7%
King (1998)	Henderson, HDIPC	1972/73 to March 1996	Income survey and microsimulation	1. Very poor rose from 12.5 to 16.7% 2. Rather poor rose from 20.6 to 30.4 % 3. 'Extremely poor'fell from 3.9 to 3.3%
Harding and Mitchell (1992)	50% of median income	1981/82 to 1989/90	IDS	Poverty fell from 11.0 to 9.5%
Mitchell and Harding (1993)	60% of median income, poverty gap	1981/82 to 1989/90	IDS	Poverty gaps stable or falling slightly
Saunders and Matheson (1993)	50% of median income	1981/82 to 1989/90	IDS	Poverty rose from 9.3 to 9.4%
Harding (1995)	50% of median income, before and after the 'social wage'	1994	Microsimulation, IDS	Poverty substantially reduced by 'social wage' (from 12 to 4% for couples with children)
King and Landt (1996)	A. Henderson, all costs B. Henderson, after housing costs	1995	Microsimulation, IDS	A. Poverty at 11.8% B. Poverty at 9.2%
OECD (1996)	50% of median income	1981/82 to 1989/90	LIS, IDS	Poverty rose from 14.4 to 16.1%
OECD (1998)	50% of median income	1975 to 1994	HES	Poverty fell from 11.9 to 9.5 %

The Henderson poverty line has the additional complication that household disposable income as measured in income surveys does not coincide with HDIPC as measured in the National Accounts. The Income Distribution Survey does not include imputed income from owner- occupied housing or the undistributed earnings of superannuation funds. These are included in the National Accounts, and they have risen more rapidly than other income components. As a result, the Henderson poverty line has actually risen faster than either mean or median survey incomes. It could be argued that this is a problem with the Income Surveys, rather than the poverty line. Nevertheless, the finding that there has been an 'ever-rising tide' (Saunders 1991) of poverty over the 1980s inevitably reflects the fact that the Henderson poverty line has increased in relative and not only absolute terms.

Studies that use a poverty line adjusted only in line with prices (Bradbury and Doyle 1992), or that use a poverty line set at 50 per cent of median income, show inconsistent trends in poverty rates over the past 15 years. For example, Harding and Mitchell (1992) found that the proportion of the population below 50 per cent of the median fell slightly from 11.0 to 9.5 per cent of the population between 1981-82 and 1989-90. Saunders and Matheson (1993) used the same methodology and the same data for the same period, but estimated that relative poverty rose imperceptibly from 9.3 to 9.4 per cent. The OECD (1996), using a similar methodology for the same period, estimated that relative poverty rose from 14.4 to 16.1 per cent. A further OECD study (1998) using the Household Expenditure Surveys found that the proportion of households below 50 per cent of the median fell slightly from 11.9 to 9.5 per cent.

A further complication in assessing and interpreting estimates of this sort is that, for most categories of income support recipients, payments are above the equivalent of 50 per cent of median income, (although income support for youth is below this level). Thus, most of the people below 50 per cent of equivalised median income are not in receipt of income support payments (ABS 1998). Some of this group are self-employed, but others are possibly either between jobs or in waiting periods for benefits.

In summary, the choice of a definition of poverty reflects a set of value judgments. Some of these judgments unavoidably have arbitrary elements. There appear to be major problems, however, with the Henderson poverty line. This means that the Henderson line does not provide consistent measures of trends over time in the number of people with relative low incomes. As noted above, while vulnerability to poverty among people of workforce age has undoubtedly increased, the level of income support has also increased substantially in real terms.

6.4 The benefit safety net

International comparisons of the generosity of benefits are fraught with difficulties. The most common form of comparison involves the use of replacement rates. Under this methodology, the levels of net benefits of defined types of individuals are compared with the disposable incomes of wage earners in similar family types. These sorts of comparisons tend to show

Australia as having very low levels of benefit generosity, precisely because Australia has flat-rate benefits without earnings related features. Thus, by European standards, the Australian pension replacement rate of 25 per cent of average earnings appears to be extraordinarily low.

Whiteford (1995) argues, however, that differing interpretations of the incidence of employer social security contributions can have a substantial impact on the measured generosity of cash benefits. When the purchasing power of basic benefits is compared rather than their replacement rates, the Australian system proves to be significantly more generous than the OECD average.³⁹

Comparisons of income distribution are similarly affected. The United Nations Provisional Guidelines for Income Statistics (1977) advocate the inclusion of employers' contributions to social security schemes in the definition of wage and salary income. This is not commonly done in income surveys such as those held in the Luxembourg Income Study. Because employer contributions are not included in the standard analytical framework, the measured gap between those receiving social security benefits and those with earnings is artificially narrowed in countries such as France and Sweden (as well as other countries with high levels of employer social security contributions).

The most comprehensive data on the level of social assistance are provided by a recent study by Eardley, Bradshaw, Ditch, Gough and Whiteford (1996), which covers all OECD countries40 at 1992. The level of assistance was compared by the use of both benefit replacement rates and benefit levels adjusted by purchasing power parities. The study compares benefit levels for single persons aged 17, 35 or 68 years, couples without children and either 35 or 68 years of age, couples with one child aged either 3 or 7 years, couples with two children aged 7 and 14 years, and lone parents with one child aged either 3 or 7 years. Benefit levels were compared both before and after housing costs (of public rented accommodation) and taking into account income taxes and social security contributions, health care costs and education expenses, and all relevant cash benefits.

Because of differences in the role of employer social security contributions, it has been argued that replacement rates-which compare benefit levels to wage rates-do not provide consistent measures of benefit generosity across countries (Whiteford 1995). The preferred methodology in this study therefore was based on comparisons using absolute benefit levels adjusted by purchasing power parities.⁴¹ When benefit levels were averaged over all the household types noted,⁴² Australian benefit levels were ranked eighth before housing costs and sixth after housing costs. Australian benefit levels were well above the OECD average, being 29 per cent higher than the mean before housing costs and 39 per cent above the mean after housing costs. Overall, Australian benefit levels were very similar to those in Sweden and the Netherlands, with most of the countries with higher benefit levels-Switzerland, Norway, Luxembourg and Canada-having substantially higher levels of national income.

Rankings varied by family types, however, with Australian benefit levels for couples with and without children being between the fourth and sixth highest in the OECD. For sole parents, the Australian rank slipped to 10th and 11th (depending on the age of the child). This poorer ranking is consistent with the finding of higher relative poverty in Australia among this group. After Sweden, Australian benefits had the lowest implicit equivalence scale for sole parents of any OECD country. An Australian sole parent with one child received a disposable income that was 83 per cent of that of a couple without children. In 14 OECD countries, a sole parent with one child receives benefits that are higher than those for couples, and in four other countries the equivalence exceeds 90 per cent. Thus, this study suggests that, rather than being very low, the Australian safety net is set at a relatively high level, although less so for sole parents.

The study also allows comparisons of social assistance and social insurance benefits. When Australian benefits were compared to unemployment insurance and retirement pensions in other countries,⁴³ Australian benefits were 12 per cent below the OECD average before housing costs, but 6 per cent above the average after housing costs. These results reflect the substantial differences between social assistance and social insurance benefits in other countries. For example, in Austria, Belgium, France and Germany, social insurance recipients received around twice as much in benefits as social assistance recipients of the same household composition. Apart from Australia and New Zealand, only a few countries applied fairly uniform benefit levels across both systems. These included Denmark and Switzerland, and to a lesser extent Ireland and Norway.

Thus, in many other countries, there may be marked discrepancies between the standards of living of low-income groups, depending on the type of benefit received. This may also imply the reproduction of inequalities in the primary distribution of income into retirement, unemployment and invalidity. Under the Australian and New Zealand approach, for example, one would expect the compression of previous status inequalities, while in the continental European countries it could be expected that status inequalities would be maintained by the income support system.

Return to top