RBA Annual Conference – 1998 Unemployment and Income Distribution Ann Harding and Sue Richardson[*]

1. Introduction

Unemployment is widely regarded as the foremost economic problem currently facing the Australian economy. Among its many evils, high unemployment – especially long-term unemployment – has become a major source of poverty. The unemployed have now replaced the aged as the predominant group in poverty in Australia (King 1997). It also contributes to the rising levels of anxiety about the increasing risks faced by employed people of involuntary loss of their jobs.[1] Psychologists have concluded that the pain of unemployment arises from two sources: ‘lack of money and lack of structure and purpose in life’ (Feather 1997, p. 39). Detailed interviews with unemployed people have ‘revealed again and again the limiting effects of loss of income and shortage of money on many aspects of family life and the negative effects of relative poverty on psychological well-being … In the absence of financial strain, the negative effects of unemployment were halved’ (Feather 1997, p. 40).[2]

The persistence of high levels of unemployment in Australia and in many European countries has been attributed to ‘structural rigidities that reduce employment opportunities for low-skill workers’ (OECD 1996, p. 22). ‘Structural rigidities’ include minimum wage laws, which are claimed to prevent the wage structure from responding to declines in the relative demand for low-skill workers. Advocates of this view draw attention to the fact that a number of countries which have had relative success in reducing unemployment also operate relatively unregulated labour markets and have experienced substantial rises in inequality in the distribution of earnings, partly because of falls in pay at the bottom. The United States, Canada, the United Kingdom and New Zealand are given as examples.[3] The experience of these countries is contrasted with those of many in the European Union, which have avoided the rising inequality but also have high and persistent unemployment. Australia has experienced rising inequality, but not as much as in the first group of countries and its wage structure is still relatively compressed. It has also experienced high and persistent unemployment, but below that of many of the European countries.

A summary view of the issues may be expressed as follows. Unemployment is, for most, a psychologically distressing state. Much, though not all, of the distress comes from the low income which it entails. The unemployed are much poorer than are low-wage workers. Unemployment is persisting because wages are prevented from falling to levels that would substantially increase demand for low-skill workers. Australia should therefore respond to its high unemployment by reducing minimum wages and other conditions on the employment of low-skill workers.

While the authors do not necessarily share this view of the world, it provides the background against which this paper has been framed. Section 2 of the paper recounts briefly what is known about the changes in income and earnings inequality in Australia over recent decades. The focus then moves to an analysis of the incomes and poverty rates of the unemployed, based on the 1994/95 income survey data produced by the Australian Bureau of Statistics. Section 3 describes key characteristics of this data and the methodology. Section 4 paints a picture of the incomes and other characteristics of the unemployed and contrasts this with the characteristics of wage and salary earners. Section 5 examines the poverty rates of the unemployed and other groups within the Australian population. Section 6 contrasts the incomes of unemployed individuals and families with those of their peers within the labour force and with those of the population more generally. Section 7 compares the incomes of the unemployed with the incomes of low-wage workers. Section 8 concludes and also speculates about whether a wage cut for low-wage workers would appear likely to reduce overall inequality and the extent of low incomes.

2. Earnings and Income Distribution Trends

A substantial number of studies have found an increase in the inequality of the distribution of earnings in Australia during the past two decades or so. For example, Borland and Wilkins (1996) concluded that earnings dispersion had increased for male and female full-time employees between 1975 and 1994, and that real earnings had fallen for male employees with earnings below the 30th percentile during the same period. In more recent work, Borland and Kennedy (1998) have again found an increase in earnings inequality for full-time employees between 1982 and 1994. Interestingly, they conclude that rates of return to education appear to have fallen, and that the growth in inequality has been due to increases in earnings inequality within groups of workers with the same educational attainment and potential years of labour market experience.

Harding (1997) also found a rise in earnings inequality between 1982 and 1994, with declines in the real earnings of both male and female full-time employees with earnings below 50 per cent of the full-time median wage – but very sharp real increases for those with earnings above 1½ times the median wage. Other studies have also pointed to growing earnings inequality (Gregory 1993; King et al. 1992).

How does the rate of growth in earnings inequality in Australia compare with that of other industrialised countries? A range of studies have suggested that the growth in earnings inequality has been much sharper in the US and the UK but has been lower in Sweden, Finland, France and Germany. The Australian experience appears to match with a group of countries (including Canada, Israel and the Netherlands) where increases in earnings dispersion have been significant, but not as substantial as in the UK and the US (Gottschalk and Joyce 1997; Borland and Wilkins 1996).

Has growing inequality in the distribution of earnings among individuals also translated into growing inequality in the distribution of income among families? Although earnings are important – because they are the key component of the average family's income – most analysts of income inequality look at trends in the distribution of disposable income, which is the amount of income families have left to spend after paying income tax. A number of studies of income distribution trends in the 1980s in Australia concluded that the distribution of disposable income among families had become more unequal (Saunders 1993; Harding 1996; EPAC 1995). However, a more recent study spanning a longer time period indicated that there had been no change in the degree of inequality of disposable income between 1982 and 1994, according to aggregate measures of inequality (Harding 1997). This apparent stability in overall inequality, however, disguised real income increases for those families at the top and bottom of the income distribution – and real income losses for the middle 40 per cent of Australians.

A recent international study of 15 countries during the 1980s and the early 1990s, concluded that the increase in disposable-income inequality had been relatively high in Australia in comparison with other countries (Gottschalk and Smeeding 1997, p. 27). This study concluded that the percentage increase in the Gini coefficient for Australia had only been exceeded by that for the UK, Sweden and Denmark. However, data for Australia were only available until 1989 and, as suggested above, it is possible that this trend may not have continued during the 1990s. As in Australia, a number of international studies have also pointed to the ‘disappearing middle class’ (Burkhauser et al. 1996, although see also Wolfson 1997).

3. Data and Methodology

The data used for this study came from the 1994/95 Survey of Income and Housing Costs (SIHC) confidentialised unit-record file, issued by the Australian Bureau of Statistics. The survey contained individual records for almost 14,000 adults aged 15 or more, belonging to 8,675 income units living in private dwellings. It should be appreciated that this is not a large sample size, and that relatively large sampling error is associated with estimates for small population subgroups. All records are weighted by the ABS, so that the results can be grossed up to arrive at estimates for the whole population.

The SIHC was conducted continuously throughout 1994/95, and the estimates for current weekly income can thus refer to any month during this financial year. The period-income estimates refer to annual income during the 1993/94 financial year. To maintain comparability with earlier studies of low-wage workers undertaken by the authors, negative income-unit incomes were reset to zero and income units with zero annual incomes were excluded entirely. Income units regarded by the ABS as being out of scope for period-income analysis were also excluded.

The measure of gross weekly cash income is defined as regular and recurring cash receipts before the deduction of tax or any other items. It includes wages and salaries, investment income, and government cash transfers such as unemployment payments. Receipts excluded from income because they were not regular and recurring included income-in-kind, inheritances and capital gains and losses. Income tax was imputed by the ABS.

A decision has to be made about how widely income is assumed to be shared between individuals. The income unit used in this study is the ABS income unit, which is a restricted family grouping which assumes that income is shared between partners in couple families and between parents and dependent children. Dependent children are defined as children aged less than 15 years or full-time students aged 15–24 years and still living at home. An unemployed 17-year-old son still living in the parental home is therefore defined as a separate income unit. The four types of income units are thus single-person units, couples with or without dependent children, and sole-parent income units. Because ‘income unit’ is an unwieldy term, the term ‘family’ is used throughout the rest of this paper instead of ‘ABS income unit’.

When comparing the living standards of the unemployed with those of the rest of the population, it is important to take account of differences in family size and composition. Thus, most people would not assume that a single wage and salary earner with an income of $300 a week experiences the same standard of living as an unemployed couple with six children with a family income of $300. One standard method is to use equivalence scales, which estimate how much more (or less) income families with various characteristics require to achieve the same standard of living as a ‘benchmark’ family. In this study, the OECD equivalence scales were applied to the after-income-tax incomes of income units, in order to better assess the relative living standards of the unemployed. The scales used attribute a value of 1.0 to the reference person in the income unit, 0.7 to the second person, and 0.5 to third and subsequent persons. Thus, an income unit consisting of a couple with two dependent children was assumed to need 2.7 times as much income as a single person in order to achieve the same standard of living. This equivalence scale can be criticised on the grounds that it takes no account of the differential costs facing those in different labour-market or housing-tenure states. It has the advantage, however, of being relatively transparent.

Section 5 examines the poverty rates of the unemployed. Poverty was defined as a situation where the income of the income unit fell below half equivalent median current disposable income. (Disposable income means after the payment of income tax.) Two ‘poverty’ measures were constructed, based on the median incomes of the entire population, and of only those living in income units with a member in the labour force. It should be appreciated that this is an entirely arbitrary definition of poverty (although it is one that has been used in a number of international studies). In the Australian context, this measure should arguably be regarded as a measure of inequality rather than of poverty, and the term ‘poverty’ is used mainly as a shorthand method of saying ‘income below half equivalent median current disposable income’.

In Section 6, the income distribution for unemployed and other persons is analysed. In order to allocate every person in the population to a decile of income, each individual was assumed to enjoy a standard of living measured by the equivalent disposable income of the income unit to which they belonged. When constructing the deciles, children were included: the bottom decile thus consists of the 10 per cent of the population with the lowest equivalent incomes, rather than the 10 per cent of persons aged 15 or more with the lowest incomes (in other words, allocation to deciles was based on ‘person weighting’).

Unemployed persons were basically defined as persons aged 15 or more who were not employed during the reference week and had actively looked for full- or part-time work at any time in the four weeks up to the end of the reference week, or who were waiting to be called back to a job from which they had been stood down without pay for less than four weeks up to the end of the reference week, for reasons other than bad weather or plant breakdown. This definition is the same as that used in the ABS Labour Force Survey series. In this study, long-term unemployed are categorised as those who have been unemployed continuously for 39 weeks or more.

4. Incomes and Other Characteristics of the Unemployed

There has been extensive analysis of the characteristics of the unemployed. For example, as shown in the Appendix, the unemployed tend to be younger, migrants and less well educated (see also ABS 1993). The Appendix enables us to see the family relationships of unemployed people and to compare these with wage earners. About one third of unemployed men are husbands with dependent children and this is reasonably similar to the proportion of this group among wage earners. Wives with dependent children, in contrast, are somewhat under-represented among the unemployed. It may well be that women with dependent children who cannot find a job simply withdraw from the workforce and are numbered among the hidden unemployed. Children living at home are prominent among the unemployed – both those who are full-time students and the larger number who are not.[4]

Table 1 provides related and additional information on the unemployed which is derived from the Survey of Income and Housing unit-record data for 1994–95. The advantage of this data set is that it enables us to examine not just the characteristics, but also the income – both personal and family – of unemployed workers, and to compare their situation with others, such as workers.

Table 1 provides information on all unemployed workers, on those who have been unemployed for 39 weeks or more (described as long-term unemployed), and on all wage and salary earners. The data are disaggregated by gender. We caution readers, however, that the sample size becomes unreliably small when we disaggregate the long-term unemployed in particular. The absolute number of men and women within the ABS sample in each of the employment categories is shown in the second row. The estimated number they represent for the whole population is shown in the first row. For example, we estimate that there were 107,000 long-term unemployed women in 1994/95, and this estimate is based on 108 observations from the unit-record data set.

Table 1: Income and Other Characteristics of Unemployed and Wage and Salary Earners 1994/95
Unemployed Long-term unemployed All wage and salary workers
Female Male All Female Male All Female Male All
(a) Denotes less than 20,000 weighted observations, or a sampling error greater than approximately 25 per cent.
Number weighted ('000) 298 437 735 107 214 321 2,879 3,795 6,674
Number of unweighted observations in sample survey 300 430 730 108 195 303 2,923 3,700 6,623
Principal source of current family income (per cent):
No income 4(a) 5(a) 4 1(a) 2(a) 2(a) 0 0 0
Wage and salary 31 14 21 15(a) 10 12 93 97 95
Own business 2 2(a) 2 1(a) 1(a) 1(a) 4 1 2
Government cash benefits 58 73 67 78 84 82 3 1 2
Other 5(a) 6 5 5(a) 3(a) 3(a) 1(a) 1 1
Principal source of period (annual) family income (per cent):
No income 0(a) 2(a) 1(a) 0(a) 1(a) 1(a) 0(a) 0(a) 0(a)
Wage and salary 49 39 43 23 19 20 90 93 92
Own business 4(a) 2(a) 3 1(a) 0 0 4 2 3
Government cash benefits 45 53 50 73 77 75 4 4 4
Other 2(a) 4(a) 3 3(a) 3(a) 3(a) 1 1 1
Average personal income from govt cash benefits $91 $121 $109 $123 $149 $141 $11 $8 $10
Average gross personal income $96 $139 $122 $129 $161 $150 $461 $672 $581
Average family income from govt cash benefits $129 $175 $156 $163 $207 $192 $22 $21 $21
Average gross family income $472 $304 $372 $384 $298 $326 $941 $903 $920
Average equivalent disposable family income $314 $256 $279 $298 $264 $276 $684 $674 $678
Proportion of all persons receiving unemployment allowances UA (per cent) 40 67 56 63 80 74
Proportion of married persons receiving UA (per cent) 22 71 52 40 80 69
Average UA received by all unemployed $53 $99 $81 $87 $120 $110
Average UA for only those receiving UA $135 $148 $144 $139 $151 $148

One of the most striking – but not unexpected – results in Table 1 is that the unemployed have much lower personal and family incomes than their wage and salary earner peers. For example, the average personal income of the unemployed of $122 a week is about one-fifth of the average $581 received by wage and salary earners. Interestingly, the personal current incomes of the long-term unemployed are higher than the personal incomes of all unemployed ($150 versus $122 a week). In part this reflects the waiting periods associated with the receipt of unemployment allowances from the government. While three-quarters of the long-term unemployed receive unemployment allowances, only just over half of all the unemployed do so (Table 1).

There is less disparity between the gross incomes of the families in which the unemployed and wage and salary earners live, with wage and salary families enjoying incomes about 2.5 times greater than those of families with an unemployed member. Gross family income, however, tells us little about relative living standards, largely because it takes no account of the number of people each family has to feed and house. ‘Equivalent disposable family income’ moves us closer towards a comparable measure of living standards, and indicates that the average equivalent income of wage and salary families of $678 is about 2.4 times higher than the average income of unemployed families.

Among unemployed individuals, there are marked differences in both personal and family incomes by gender. The personal incomes of unemployed men are about 45 per cent higher than those of unemployed women, presumably because a greater proportion of unemployed women live in families where their husband earns an income – thereby rendering them ineligible for unemployment assistance, which is means-tested on a family basis. On the other hand, the family incomes of unemployed women are about 50 per cent higher than those of unemployed men, again reflecting the increased probability of a second income earner in the family for women (Figure 1).

Figure 1: Average Personal and Gross Family Incomes of Unemployed and Long-term Unemployed Persons

These differences in personal and family income by gender are both less marked for the long-term unemployed, as a result of long-term unemployed women being less likely to live in a family where the husband earns a reasonable income. As Table 1 indicates, being married makes little difference to the probability of receiving unemployment assistance for men: for example, 80 per cent of both married and all long-term unemployed men receive some unemployment assistance from the government. In contrast, being married has a substantial negative impact on the likelihood of unemployed women receiving unemployment allowance – but this impact is much less for long-term unemployed women.

Table 1 shows that the sources of income for the family in which unemployed people live are quite different from the sources of income for employees. Not surprisingly, wages and salaries are the overwhelming source of current family income for employees. Income from self-employment, from government cash benefits and from all other sources (e.g. interest and dividends) together are the main source of current weekly income for just 5 per cent of the families which have a wage earner: the other 95 per cent rely primarily on wages. In contrast, only about 21 per cent of the unemployed and 12 per cent of the long-term unemployed live in families which rely principally on wage and salary income. Instead, they mostly rely on government cash welfare benefits – the more so if they are long-term unemployed (Figure 2). Four out of five of the long-term unemployed lived in families which had government benefits as their principal source of income in the survey week. This contrasts starkly with the overwhelming reliance of employees on wages and salaries.

Figure 2: Principal Current Family Income Source of Unemployed Persons

There is no inevitability about these figures. A person who is classified as employed may be working only a few hours a week and live in a family in which they and other members draw social welfare payments or are self-employed. The earned wage may then contribute only a small fraction of the family income.[5] And a person who is unemployed may live in a family in which other members have full-time jobs, and hence the family's main source of income is from the wages earned. What the figures tell us is that the preceding relations are not common. People who are employed, even part-time, overwhelmingly live in families which support themselves from their own earnings. People who are unemployed do not. Unemployed and employed people are not much to be found in the same family. This separation of employed and unemployed families is, however, more true for men than it is for women. One-third of unemployed women live in families where wages are the principal source of current income, whereas for men this situation is only half as common.

In interpreting this, recall that a young person, over age 15 and living at home but not a full-time student, is classified as a separate family. Thus they are not counted as sharing in their parents' income. Non-student children living at home comprise 21 per cent of all the unemployed and 15 per cent of adults (aged 21 or more) who are unemployed. They are thus a large enough group to affect the overall image one gets of the unemployed family and may be living quite adequately on their parents' incomes. However, even when the analysis is confined to unemployed people aged 21 or more, the picture of principal source of family income for unemployed families shows little difference – with a slightly higher proportion nominating government cash benefits as the principal source of family income (71 per cent), and a slightly lower proportion nominating wages and salaries (17 per cent).

The distinction between employed and unemployed families becomes less sharp if we examine annual rather than weekly income. Those who were employed at the time of the survey still rely overwhelmingly on their earnings for their annual income in the preceding financial year. But 43 per cent of people who were unemployed at the time of the survey lived in families which had wages and salaries as their principal source of income over the preceding year. Not surprisingly, this figure is halved for the long-term unemployed. It is not clear how to interpret this more encouraging figure. On the one hand, it may mean that people who are now unemployed were, for much of the year, earning a wage. On the other hand, it may mean that during the year, other members of the family had a wage and now do not.

Men are more likely than women to have their own employment status reflected in the principal source of income for the family. Thus 73 per cent of men who are unemployed live in income units which rely on government cash benefits, whereas this figure for unemployed women was 58 per cent. This is despite the fact that unemployed men are more likely to be married than are unemployed women: a woman has a greater likelihood of being a single parent and thus the only income earner for her income unit (Appendix).

Not only are government cash benefits the principal source of income for the majority of families in which there is an unemployed person, but these benefits provide almost all of the personal income of unemployed people (Table 1 and Figure 3). Indeed, for the long-term unemployed they provide 94 per cent of their personal income. And this is despite the fact that only about half of unemployed people and three-quarters of the long-term unemployed actually receive an unemployment allowance.[6] In striking contrast, government cash benefits provide on average only 2 per cent of the current personal income of people who have jobs. The welfare system is very tightly targeted and its beneficiaries are mostly those – individuals and families – who do not earn an income for themselves.

Figure 3: Government Cash Benefits as a Proportion of Personal and Family Income

The figure for the average contribution of cash welfare payments to the incomes of employed families conceals the fact that for some such families these payments are indeed quite important. In their submission to the 1998 ‘Safety Net Review’ before the Australian Industrial Relations Commission, the Joint Governments show that for people who live in employed families, cash welfare payments are equal to 20 per cent of the private income of people in the bottom quintile of the income distribution. This drops to 5 per cent for the second quintile and to virtually zero for the remaining families (Joint Governments 1998, p. 185, Table 11.1). Again, the cash benefits received are tightly targeted to those at the bottom. But this does suggest that the difference between unemployed and low income employed families in their sources of income may not be as great as they appear to be when we look only at the average situation of all employed people.

In addition, it should be emphasised that this analysis has only taken account of the cash benefits received by unemployed families from the government. Families also receive significant non-cash benefits from government, principally via their usage of publicly funded health, education and housing services. The above report, for example, also showed that the bottom quintile of wage and salary earner households received an estimated $252 a week in health, education and housing non-cash benefits – about 2.7 times as much as the $93 received by the top quintile of households (Joint Governments 1998, p.187). While these figures did not deal with unemployed households, the results are suggestive (see also Johnson et al. 1995; Harding 1995a).

Whereas government cash benefits provide about the same proportion of own income for unemployed men and women (around 89 per cent for all, and 94 per cent for long-term unemployed), there is a substantial gender difference when we look at the contribution of these benefits to the income of the whole family. For unemployed women, government cash benefits contribute less than 30 per cent of the income of the family in which they live. For unemployed men the figure is 58 per cent. Although the percentages are higher, the gender pattern is similar for the long-term unemployed. This again implies that women are more likely than men to have another person in the income unit who is earning an income.

The data also enable us to observe the socioeconomic status of the geographic area in which workers and the unemployed live. Gregory and Hunter (1995) have shown that there has been a large increase in the geographic concentration of advantage and disadvantage, so we would expect unemployed workers, especially the long-term unemployed, to be heavily concentrated in the bottom deciles of the ranking of areas by socioeconomic status. Figure 4 shows the distribution of all wage earners, of all unemployed workers, and of long-term unemployed workers by SEIFA decile – a geographic measure of socioeconomic status constructed by the ABS.[7] It shows that there is a concentration of unemployed workers in the lower SES deciles, and that this is more severe for the long-term unemployed than for all unemployed. If you are long-term unemployed, the chance that you live in an area which ranks in the lowest two deciles is more than twice that of an employed person. The chance that you live in one of the top two decile areas is less than half that of an employed person. But still, 16 per cent of the unemployed (10 per cent of the long-term unemployed) live in the top 20 per cent of areas. This compares with 26 per cent of employees.

Figure 4: SEIFA Decile of Unemployed and Wage and Salary Earners

We may conclude that the unemployed are not a random sample of the workforce, but neither are they remarkably different from it. While young people are noticeably over-represented among the unemployed, unemployment – and especially long-term unemployment – can strike at any age. One-third of the unemployed and 44 per cent of the long-term unemployed are aged 35–54, and the age distribution of unemployment is remarkably similar for men and women.

The unemployed are more different in their education and ethnicity. They are much more likely than the workforce at large to have no post-school qualifications and to be immigrants. They are also less likely to be married (even if we confine the comparison to those over the age of 21). Whereas almost two-thirds of all employees are married, less than one-half of the unemployed are.

But the most distinctive thing about the unemployed is their source of income. Even though only 70 per cent of men and less than half of women who are unemployed receive unemployment allowances, government cash benefits provide most of their personal income and often the majority of the total income of the families in which they reside. The dominance of social welfare payments is even greater for the long-term unemployed. The contrast with those who have jobs – including part-time jobs – is striking. For both individuals who have jobs and their families, government cash payments are a tiny part of their overall income.

5. Unemployment and Poverty

From the evidence of the previous section we can be confident that, on average, unemployed people have low incomes. Because of the means-tested character of the Australian social welfare system, it is inevitable that people who rely upon it mostly have relatively low incomes (even if it is possible for their asset position to be moderately comfortable). Table 2 displays a summary set of information, which examines the extent to which unemployed people and their families are in poverty. As noted earlier, we have selected a totally arbitrary measure of poverty – half the equivalent median current disposable income of families – and enquire whether unemployed people are disproportionately found to have incomes below this level. This question has a number of dimensions.

Table 2: Poverty Rates for Various Groups using Two Poverty Lines
Per cent; 1994/95
Entire-population poverty line Labour-force poverty line
Persons aged 15+:
Unemployed 28 45
Long-term unemployed 20 42
Wage and salary earners 2 4
All 8 12
Dependent children in:
Unemployed families 27 50
Long-term unemployed families 27 50
Wage and salary families 3 6
All families 12 19

Income is, in all cases, expressed as the current equivalent disposable income of income units, adjusted by the OECD equivalence scale. We compute two income distributions: one contains the whole population; the other contains only people who live in income units which have at least one person in the labour force, termed the labour-force distribution, for convenience. Unemployed people, as well as the self-employed and employees, are included as part of the labour force. The rationale for looking separately at the distribution of income among the labour force is that we would expect that people who seek to, and indeed do, support themselves with paid work would generally have a higher income than people who rely predominantly on social welfare payments. A high proportion of income units which are out of the labour force do indeed rely on social welfare payments for their income – sole parents, the aged and invalids for example. We wish on the one hand to compare the situation of the unemployed with their peers – those who are in the labour force. We wish also to assess their fortunes in relation to the whole population, including the other groups who rely on social welfare payments. The two distributions have different median incomes: the median for those in the labour force being higher than the median for all people in the population ($494 and $426 a week respectively). An unemployed family is defined as one which contains an unemployed person aged 15 or over.

Table 2 shows the percentage of various population groups that are in poverty. It shows that only 8 per cent of adults aged 15 or more are in poverty when the entire-population poverty line is used. And that 12 per cent are in poverty when the higher poverty line for those in the labour force is used. It is also clear that unemployed persons live in families with higher poverty rates. Just under 28 per cent of all unemployed persons live in families that are in poverty using the population poverty line, rising to 45 per cent using the labour-force poverty line. Given that there is only $34 a week difference between these two poverty lines, this dramatic increase in recorded poverty underlines the extreme sensitivity of such head-count measures of poverty to small changes in the poverty line. This is caused by the bunching of large numbers of social security recipients around these particular poverty lines. This bunching effect also appears to underlie the changes in the poverty rate for the long-term unemployed, which increases from 20 to 42 per cent as the poverty line increases from the general population poverty line to the labour-force poverty line. The poverty rate among families that declare at least $1 of wage and salary income is only 2 per cent using the population poverty line. This doubles to 4 per cent using the labour-force poverty line.

Children are particularly at risk. Three out of every ten children living in a family with an unemployed member are in poverty, compared with only one out of every ten children in Australia. When the higher labour-force poverty line is used, fully one-half of children living in unemployed families are in poverty. Recall that one-third of unemployed men are fathers with dependent children. Poverty rates are, however, very low among children living in families with some wage and salary income. Using the population poverty line, poverty rates for such children are only 3 per cent. In other words, dependent children living in families with at least one unemployed member have nine times as much risk of being in poverty as those living in families with some wage and salary income.

6. The Unemployed and Income Distribution

In this section we explore the question of where the unemployed, grouped in a variety of ways, lie in the overall distribution of income. Income is again defined as the equivalent disposable income of the family and this income is attributed to each person within the income unit (including children under age 16). Again two different overall income distributions are used. In one, the whole population is ranked from lowest equivalent income to highest. In the other, only people who live in income units which contain a person who is in the labour force are included.

Figure 5 shows where unemployed people are to be found in the overall distribution of income for the entire population. Deciles of the income distribution are measured on the horizontal axis, ranked from lowest to highest. On the vertical axis is measured the proportion of the relevant group which has an income located in each decile. If everyone had the same income, 10 per cent of each group would be found in each decile. A line marking this equal distribution is included in the figure for reference. As Figure 5 shows, the unemployed are heavily concentrated in the bottom third of the overall income distribution. Just over two-thirds of male unemployed and over half of female unemployed are in the bottom two deciles (see also Table 3). As suggested earlier, unemployed men are clustered at the bottom of the distribution to a greater extent than unemployed women.

Figure 5: Distribution of Unemployed People by Decile of Equivalent Current Disposable Income
Table 3: Proportion of Unemployed People in the Top and Bottom Quintiles of the Income Distribution
For labour force For all people
All unemployed 71 2 8 63 2 12
Males 78 2 6 70 2 9
Females 63 2 10 54 2 19
Adults (21 yrs+) 74 1 6 64 1 10
Youth (<21 yrs) 65 6 15 62 6 21
Long-term 78 1 7 68 1 7
With children 64 4 12 58 4 20
Men 75 3 7 71 3 11
Women 50 5 17 41 5 29

Does the picture vary by age or for those with dependent children? The circumstances of the long-term unemployed are shown in Figure 6. In drawing conclusions from this information it is important to recall that, in the sample with which we deal, there were only 303 people in this category. Further division into deciles reduces the number of respondents in the higher deciles to a handful. Our interest, however, is principally in whether the long-term unemployed look different from other unemployed people in the extent to which they are concentrated at the bottom of the income distribution. The precise proportion in each decile above the fifth is not of much concern. Relative to all unemployed and to adult unemployed, the long-term unemployed are more densely clustered in the second decile, and a little less densely clustered in the first decile. But the differences are small and if the information were displayed as quintiles rather than deciles, would not be noticeable. The overall conclusion must be that the income situation of the long-term unemployed is little different from that of unemployed people in general.

Figure 6: Distribution of Unemployed People by Decile of Equivalent Current Disposable Income

This may be interpreted as an outcome of the way that ‘unemployment insurance’ operates in the Australian social welfare system. The level of benefit to which a long-term unemployed person is entitled is the same as that due to any unemployed person in the same family and private income circumstances. Thus, the big difference in income is associated with movement between employment and unemployment, rather than with the duration of unemployment. Indeed, the short-term unemployed may for a time have lower incomes than equally situated long-term unemployed, as they wait to become eligible for benefits and as they use up initial cash reserves which bar them from eligibility. Recall that only 56 per cent of all unemployed receive an unemployment benefit whereas 74 per cent of long-term unemployed do.

The broad equality of circumstances of the long-term unemployed and the short-term unemployed can be attributed to the fact that Australia does not have an unemployment insurance scheme which provides benefits for some defined period which are related to previous earnings. Rather, benefits are determined, whatever the duration of unemployment, by an assessment of current income, assets and dependents. As the Appendix shows, the family circumstances of the long-term unemployed are not strikingly different from those of all unemployed workers.

Figure 6 also looks at the unemployed disaggregated into yet further groups. The new groups are youth – aged 15–20 – and those with dependent children. Most of the difference between the series is found in the first three deciles, although for both youth and those with children, 20 per cent are found in the top half of the income distribution whereas only half this number of all the adult unemployed are.[8]

Almost half of all unemployed youth are in the bottom decile of the distribution of equivalent disposable income unit income, for all people. This suggests intense disadvantage. But it may not be all it seems. Recall that a person living with his or her parents who is over age 15 and not a full-time student is classified as an independent income unit. The provision of board and even cash gifts by parents does not count as part of the income of the young person. Twenty per cent of both all unemployed and the long-term unemployed are living at home with their parents. We do not know the amount of support they receive, but it almost certainly includes housing and most probably includes rather more than that.

The unemployed with dependent children are, like all the other groups, heavily concentrated at the bottom of the income distribution, but not quite so severely as the long-term, all, and all adult unemployed. Although it is not evident in the figure, this is because unemployed women with dependent children are much more evenly spread across the income distribution than are any other group that we looked at. About two-fifths are still in the bottom quintile, but almost one-third are in the top half of the income distribution (Table 3). The comparable figures for men with dependent children are 71 per cent and 11 per cent.

All of these graphs have examined the income distribution of the unemployed relative to that of the entire population, with one-tenth of the population being found in each of the deciles considered above. The percentage of variously defined unemployed groups to be found in the bottom and top quintile, and the top half of the distributions for all people and for people who live in families with a member in the labour force are to be found in Table 3. Not surprisingly, if deciles are constructed for only those families with a member in the labour force, the unemployed appear to be even more heavily concentrated at the bottom of the income distribution.

An alternative way of looking at these issues is to examine the unemployed as a proportion of those in each decile. Figure 7 shows that the unemployed make up about one-fifth of all of those persons aged 15 or more who are in the bottom decile of equivalent family income for the entire population. Wage and salary earners with very low family incomes make up another 10 per cent of all those in the bottom decile. Persons who are not in the labour force are the most important group at the very bottom of the income distribution, contributing half of the bottom decile. And those who fall into the ‘Other’ category (principally the self-employed) make up the final 20 per cent.

Figure 7: Proportion of Persons Aged 15 or More with Specified Labour Force Status, by Decile of Family Equivalent Current Disposable Income

As one would expect, as family income increases, a rapidly rising proportion of people aged 15 and over are wage and salary earners. About 85 per cent of persons in the top decile of family income are wage and salary earners, while less than 1 per cent are unemployed people married to a high income spouse (or living with high income parents). Figure 7 again underlines that the unemployed people are concentrated at the lower tail of the income spectrum.

The clear conclusion from this section is thus that unemployed people are overwhelmingly found at the bottom of the income distribution. It does not matter which group you look at or which income measure is used. Among the modest differences detected, the principal one is that this relative income disadvantage is particularly pronounced for adult men, with and without dependent children.

7. Low-wage Workers and the Unemployed

There is considerable interest in the question of whether low-wage workers are relatively badly off in terms of their personal and family incomes and how they compare on this dimension with unemployed workers. The interest arises in part from the idea that it may be possible to reduce the wages of the low paid and in so doing, generate extra employment for people who are currently unemployed – or reduce the flow of new additions to the unemployment pool. This possible trade-off has been confronted in a very direct way by the Australian Industrial Relations Commission, in its so-called ‘living wage’ cases. These were conducted in 1997 and 1998 and were directed to setting award rates of pay for people who had not participated in enterprise or other bargains. In its submission to the 1998 ‘living wage’ case, the Joint Governments (1998) said:

‘…Large real increases in award rates are likely to have an adverse effect on employment. This obviously does not serve the needs of the low paid. No one should ever lose sight of the fact that the majority of people in poverty are not employed. Anyone who is serious about reducing inequality and poverty should primarily devote their attention to measures which will reduce unemployment’ (p. 178).

Figure 8 traces the income distribution for the unemployed, low- and minimum-wage workers, and all wage and salary earners. Low-wage workers are defined as adults earning less than $10 an hour and juniors earning less than $6 an hour, while minimum-wage earners are defined as adults earning less than $8.50 an hour. The income distribution is for the whole population. The general picture is similar if the distribution for those in the labour force is used. It is clear that unemployed people are more heavily concentrated in the bottom deciles, and are more rarely found in the upper deciles than are minimum-wage earners. Unemployed workers are substantially over-represented in the first three deciles, as was shown previously. In contrast, minimum-wage earners are over-represented in the middle deciles, four to six. This occurs for two reasons. The first is that the minimum wage used in this paper – in effect the then prevailing Federal minimum – is enough to put a full-time worker who has no or few dependents in these middle deciles. The second is that minimum-wage workers often live in families in which there is more than one income earner. Multiple-income earners are much more common among low-wage families than among unemployed ones.

Figure 8: Distribution of Unemployed People, Low, Minimum and All Wage Earners, by Decile of Equivalent Current Disposable Income

The majority of low-wage earners are found further up the income distribution than minimum-wage earners, with about half being concentrated in deciles six to eight. Once again, wage earners generally are better off than low-wage earners, with about three-quarters of all wage earners receiving a family income which places them in the top four deciles of the income distribution.

We conclude from this section that the evidence strongly supports the view that the unemployed are indeed financially very disadvantaged. They comprise a high proportion of people in the lowest deciles of the income distribution and are heavily concentrated in these deciles. These conclusions apply with less force to women than to men, but apply to them nonetheless. Every way that we have looked at it, unemployed workers are much lower in the income distribution than are most low-wage workers – although there are some low-wage workers who are in the lowest deciles and these should not be forgotten.

8. Conclusion

As in other English-speaking countries, the distribution of earned income in Australia has become more unequal in the past decade. Wages and salaries and income from interest, dividends and rent are, separately and together, now less evenly spread across the population than they were (Harding 1997). One reason is a rise in the dispersion of pay, with jobs which pay a lot and jobs which pay a little, growing relative to jobs with middle-level rates of pay. A second reason is the growth in unemployment, including long-term unemployment. It is likely that a third reason is the continuing trend towards people living in smaller and smaller families, including single parent families (Harding 1995b). In theory, this last development should not affect inequality in our income measure because the equivalence scale is intended to capture the economies of shared living arrangements. But it is a crude and simple scale and undoubtedly does injustice to the many ways that sharing accommodation, recreation, transport and so on reduces the per person cost of living.

It is also worth noting that the degree of inequality is exaggerated in our data because young people living at home, who are not full-time students, are treated as if they depend fully on their own income, which clearly they do not. For young men in particular, there are many such people among the low-wage and unemployed groups. The simple fact that these young adults live at home with their parents suggests strongly that for many of them, it is cheaper to do so than to live separately; yet treating them as independent income units implies that they live separately.

The detailed examination of the unemployed presented above enables us to draw a picture of this disadvantaged group. The young are over-represented, but all age groups contribute significantly to the unemployment pool. Migrants, the relatively uneducated and single people are prominent. Many are young people still living – probably reluctantly – at home, a majority of whom are not full-time students. Less than half of all young unemployed receive unemployment allowances (42 per cent), although this rises to 69 per cent of the long-term young unemployed. For the unemployed generally, the comparable figures are 56 and 74 per cent. Yet government cash benefits provide 90 per cent of the personal income of the unemployed and, for adults, a significant proportion of family income. This is in striking contrast to the families which have a wage earner: for them, government cash benefits on average provide just 2 per cent of their income.

The major contribution of this paper is to examine, and display, the income situation of the unemployed. It has been known for some time that the personal incomes of the unemployed are quite low. This is no surprise, since they have little income other than that provided by a modest and tightly targeted social welfare system. What is new is the evidence that not only are the personal incomes of the unemployed low, but so also are the disposable incomes of the families in which they live. Overwhelmingly, unemployed people do not live in families in which there is significant wage or other private income for other family members. They and their families rely heavily on the social welfare system for their income. Since the social welfare system in Australia is of the safety-net kind, providing just enough to enable recipients to make ends meet, it is inevitable that unemployed people and their families have low incomes.

It also follows that they have lower incomes than do families which, by virtue of their wage income, are excluded from eligibility for social welfare assistance. In order to maintain some financial reward to employment as opposed to unemployment, it is necessary to set the level of benefits received by most unemployed people below the sums that could be earned at a low wage by a person working more than a few hours per week. And this has been done, even if imperfectly. It is striking that among workers, a tiny proportion of their income, on average, comes from government cash benefits. This does not deny that for some employed families, government assistance for children is significant. But on average it is not. People who rely upon wages and salaries have higher incomes than people who rely upon social welfare payments. This is the intention of the design of the social welfare system, and it works.

Unemployed workers and their families have two to three times the average person's chance of having a very low income. They are also clustered heavily at the bottom of the distribution of equivalent disposable family income: about two-thirds are in the bottom two deciles of the overall distribution. They comprise a large proportion of people in the workforce who have very low incomes. They are substantially worse off in these respects than are people who earn low wages.

It is undoubtedly true, then, that unemployed people mostly have lower equivalent family incomes than do wage earners, even low-wage earners. It follows that a redistribution from low-wage earners to unemployed people would make the overall distribution of income more equal. But it does not necessarily follow that a cut in the wages of the low paid is thereby egalitarian.

First, if one simply wants to increase the incomes of the unemployed and their families, a ‘tax’ on low-wage workers is a much less equitable way to finance this than is a tax on all wage and salary earners.

Second, a cut in the wages of the low paid is helpful to the unemployed only if it causes a substantial increase in the number of low-wage jobs and of all jobs. The evidence on the likely response of employment to a cut in low wages is scarce in Australia. But at the very least, the overseas evidence must cause some pessimism that the employment response would be large. If it is small, then all low-wage workers would be worse off as a result of a cut in their wages, while some previously unemployed workers would be better off and most would not be affected.

Third, an increase in the number of low-wage jobs does not mean a commensurate fall in the numbers of people who are unemployed. Many such jobs would go to people who previously were out of the labour force.

Fourth, a cut in the level of low wages would exacerbate the already severe problem of high effective marginal tax rates for people on the borders between the social welfare system and paid employment (Beer and Harding 1997). The differences between paid work and unemployment assistance would diminish. It is likely that this would produce pressure to lower the level of social welfare benefits, both in order to maintain an incentive to work and in order to maintain the socially preferred position that people who have paid work generally have higher incomes than people who are reliant on social welfare. Since the level of social welfare payments for all forms of social assistance – to single parents, full-time students and the aged for example – are set at a common level, a cut in the level of benefits for unemployed people could bring pressure to cut the levels of all related benefits. (An alternative solution might be to increase the earnings of those in low-paid employment, via earnings credits or wages top-ups, rather than decreasing the payments of those on benefits.)

Overall, therefore, the impact on income inequality of cuts in real wages is unclear. While some proportion of the unemployed might gain jobs, existing low-wage earners could expect their earnings to decline. It is also not clear what the consequent effects might be upon the social security system.[9]


Selected Characteristics of the Unemployed and of Wage and Salary Earners
Per cent
Unemployed Long-term unemployed All wage and salary workers
Female Male All Female Male All Female Male All
(a) Denotes less than 20,000 weighted observations, or a sampling error greater than approximately 25 per cent.
Number weighted ('000) 298 437 735 107 214 321 2,879 3,795 6,674
Number of unweighted observations in sample survey 300 430 730 108 195 303 2,923 3,700 6,623
15–20 28 20 23 17(a) 14 15 11 9 10
21–24 11 13 12 10(a) 12 11 12 10 11
25–34 22 25 24 25 21 23 25 26 26
35–54 34 33 33 43 41 41 47 46 46
55–64 5(a) 9 7 6(a) 12 10 5 8 7
Total 100 100 100 100 100 100 100 100 100
Education qualifications:
Degree 7(a) 6 6 9(a) 5(a) 6 15 15 15
Undergrad diploma 3(a) 2(a) 2(a) 0(a) 1(a) 1(a) 6 2 4
Associate diploma 3(a) 4(a) 4 4(a) 3(a) 3(a) 5 8 6
Skilled vocational 10 17 14 7(a) 13 11 11 25 19
Basic vocational 6(a) 3(a) 5 5(a) 3(a) 3(a) 7 2 4
No qualifications 63 64 64 74 72 73 53 46 49
At school 8 4(a) 6 1(a) 3(a) 2(a) 3 2 2
Total 100 100 100 100 100 100 100 100 100
Marital status:
Married 39 44 42 39 52 48 61 62 62
Never married 45 47 46 31 38 36 31 32 31
Separated, widowed etc. 16 9 12 30 10 16 8 6 7
Total 100 100 100 100 100 100 100 100 100
Year of arrival (for migrants only):
Before 1976 27 46 39 24 47 39 53 57 56
1976–80 9 12 11 12 13 11 11 10 10
1981–85 32 13 21 23 18 20 12 10 11
1986–90 14 19 17 20 12 15 17 17 17
1991–94 18 10 12 21 10 15 7 6 6
Total 100 100 100 100 100 100 100 100 100
Family relationship:
Husband and dependent children 30 18 32 22 37 21
Wife and dependent children 20 10 25 8 34 15
Other husband 14 8 19 12 24 14
Other wife 15 6 14(a) 5(a) 27 12
Lone parent and dependent children 9 2(a) 5 7(a) 3(a) 5(a) 4 1 2
Full-time student child aged 15–24 13 8 10 2(a) 3(a) 2(a) 5 3 4
Other child aged 15+ 18 22 21 19 19 19 12 14 13
Unrelated group 6(a) 8 7 6(a) 7(a) 7 7 9 8
Live alone 8 14 11 11(a) 15 14 8 9 9
Other 7 3(a) 5 15(a) 3(a) 7 3 2 2
Total 100 100 100 100 100 100 100 100 100
Length of time unemployed:
1 to < 39 weeks 64 51 56  
39 to < 52 weeks 5 7 6  
52 to < 104 weeks 12 14 13  
104 weeks+ 19 28 24  


The work on poverty rates was supported by ARC Grant No. A79803294. [*]

See Richardson (1998) for evidence of both the rising risk and the rising feelings of insecurity in the employed workforce. [1]

Recent work by Theodossiou (1998) concludes that the psychological health of unemployed people is significantly worse than that of low-paid workers, with middle-aged people and women less psychologically damaged by unemployment than younger and older age groups and men. [2]

Gregory (1996) has shown that low-wage jobs grew proportionately as fast in Australia as in the US, despite the fact that in Australia, the minimum levels of wages were higher both relatively and absolutely than in the US. Hancock (1998) quotes data that show that there is no correlation between the degree of dispersion in the wage structure and the employment to population ratio among OECD countries. [3]

Those who met the definition of unemployment outlined in Section 3 were counted by the ABS as unemployed, even if they were engaged in full-time study. [4]

For example, in a married couple family a person earning minimum wages would have to work almost 20 hours a week to contribute an amount which is more than half their combined unemployment benefit – i.e. to have wages as their principal source of income. [5]

Unemployed families with children usually receive family payments, in addition to any unemployment allowances. [6]

SEIFA deciles are computed by the ABS and included in the data set as a variable which may be associated with each individual. They are based on indicators of income, education, rented dwellings, lack of fluency in English, and high unemployment for residents of geographic areas. Decile 1 represents the 10 per cent of geographic areas which have the lowest average socioeconomic status of residents. The data are reported only for NSW, QLD, Victoria and WA, as sample numbers are too small in the other regions to preserve confidentiality. About 15 per cent of the sample is thus excluded. [7]

In the case of youth, this is partly attributable to unemployed 15–24 year olds being counted as part of their parents' income unit if they are also full-time students. Thus, some such parents obviously have high incomes. [8]

These issues were more comprehensively examined in work by us and other authors at the conference ‘Fairly Efficient? Equity and Productivity in the Australian Labour Market’ held at the ANU, 16–17 June 1998. [9]


Australian Bureau of Statistics (ABS) (1993), ‘Australia's Long-term Unemployed: A Statistical Profile’, ABS cat. no. 6255.0.

Beer, G. and A. Harding (1997), ‘Effective Tax Rates’, Income Distribution Report, National Centre for Social and Economic Modelling, University of Canberra, No. 7.

Borland, J. and S. Kennedy (1998), ‘Earnings Inequality in Australia in the 1980s and the 1990s’, mimeo, Australian National University, February.

Borland, J. and R. Wilkins (1996), ‘Earnings Inequality in Australia’, Economic Record, 72(216), pp. 7–23.

Burkhauser, R., A. Crews, M. Daly and S. Jenkins (1996), ‘Where in the World is the Middle Class? A Cross-national Comparison of the Vanishing Middle Class Using Kernel Density Estimates’, University of Essex, ESRC Research Centre on Micro-social Change Working Paper No. 96-8.

Economic Planning Advisory Commission (EPAC) (1995), Income Distribution in Australia: Recent Trends and Research, AGPS, Canberra.

Feather, N. (1997), ‘Economic Deprivation and the Psychological Impact of Unemployment’, Australian Psychologist, 32(1), pp. 37–45.

Gottschalk, P. and M. Joyce (1997), ‘Cross-national Differences in the Rise of Earnings Inequality – Market and Non-market Factors’, Luxembourg Income Study Working Paper No. 160.

Gottschalk, P. and T. Smeeding (1997), ‘Empirical Evidence on Income Inequality in Industrialised Countries’, Luxembourg Income Study Working Paper No. 154.

Gregory, R. (1993), ‘Aspects of Australian and US Living Standards: The Disappointing Decades 1970–1990’, Economic Record, 69(204), pp. 61–76.

Gregory, R. (1996), ‘Wage Deregulation, Low Paid Workers and Full Employment’, in P. Sheehan, B. Grewal and M. Kumnick (eds), Dialogues on Australia's Future, Centre for Strategic Economic Studies, Victoria University, Melbourne, pp. 81–102.

Gregory, R. and B. Hunter (1995), ‘The Macro-economy and the Growth of Ghettos and Urban Poverty in Australia’, Australian National University Centre for Economic Policy Research Discussion Paper No. 325.

Hancock, K. (1998), ‘The Needs of the Low Paid’, in Wealth, Work and Well-being, Academy of the Social Sciences in Australia, Canberra, Occasional Paper No. 1, pp. 1–26.

Harding, A. (1995a), ‘The Social Wage and Income Distribution: the Impact of Health, Education and Housing Outlays in 1994’, Australian Economic Review, 28(3), pp. 71–86.

Harding, A. (1995b), ‘The Impact of Family, Demographic and Labour Force Change Upon Income Inequality in Australia, 1982–93’, Australian Journal of Social Research, 1(1), pp. 47–70.

Harding, A. (1996), ‘Recent Trends in Income Inequality in Australia’, in P. Sheehan, B. Grewal and M. Kumnick (eds), Dialogues on Australia's Future, Centre for Strategic Economic Studies, Victoria University, Melbourne, pp. 283–306.

Harding, A. (1997), ‘The Suffering Middle: Trends in Income Inequality in Australia 1982 to 1993/94’, Australian Economic Review, 30(4), pp. 341–358.

Johnson, D., I. Manning and O. Hellwig (1995), Trends in the Distribution of Cash and Non-cash Benefits – An Overview, Report to the Department of the Prime Minister and Cabinet, AGPS, Canberra.

Joint Governments (1998), ‘Safety Net Review – Wages’, Department of Workplace Relations and Small Business, Canberra.

King, A. (1997), ‘The Changing Face of Australian Poverty: A Comparison of 1996 Estimates and the 1972–73 Finding from the Commission of Inquiry’, National Centre for Social and Economic Modelling Discussion Paper No. 23.

King, J., S. Rimmer and R. Rimmer (1992), ‘The Law of the Shrinking Middle: Inequality of Earnings in Australia 1975–1989’, Scottish Journal of Political Economy, 39(4), pp. 391–412.

OECD (1996), Employment Outlook, OECD, Paris.

Richardson, S. (1998), ‘Who Gets Minimum Wages?’, Australian National University Centre for Economic Policy Research Discussion Paper No. 386.

Saunders, P. (1993), ‘Economic Adjustment and Distributional Change: Income Inequality in Australia in the Eighties’, Social Policy Research Centre Discussion Paper No. 47.

Theodossiou, I. (1998), ‘The Effects of Low-pay and Unemployment on Psychological Well-being: A Logistic Regression Approach’, Journal of Health Economics, 17(1), pp. 85–104.

Wolfson, M. (1997), ‘Divergent Inequalities – Theory and Empirical Results’, Review of Income and Wealth, 43(4), pp. 401–422.