Alternative sources of welfare (1).

Abstract

Direct transfers from government are not the only forms of institutional support in Australia for people not earning or not fully supported by the market or families. Employers provide paid leave to employees through sick leave, recreational leave, long service leave and, in the public sector, paid parental leave. These are paid at a rate equal to full or half salary. Employers also contribute to compulsory workers compensation, which provides a pool of funds for employees who are unable to work because of accident or illness. Superannuation is another form of provision for retired workers, and in some cases for those with permanent disability. There are two major forms--defined benefit, which provides pensions determined by the employee's final salary; and accumulation, which is funded through the contributions made on an employee's behalf and the earnings on that capital More recently there have been proposals for other forms of support--one promoted by Dr Bruce Chapman is for income contingent loans, the principle underlying the HECS scheme. Another promoted by Professor Peter Dawkins and others in the mid 1990s was an earned income refundable tax credit to support low wage workers with dependent children. The Chifley Research Centre and Mark Latham explored matched savings or nest-egg accounts, for young people and others to facilitate asset building. What are the comparative advantages and disadvantages of these various forms of support? What are the outstanding problems that need addressing at this point in Australia? Which of these models could be usefully applied to such issues? This paper discusses the pros and cons of the existing systems and other proposals in the Australian context, at the present time.

Introduction

At June 2004, 60 per cent of adult Australians derived their income fully or partly from earnings (ABS 2004). About 30 per cent received income support through social security (FACS 2004). Given that about 3 per cent received both, this leaves about 13 per cent dependent on intra-family transfers, savings and investments or other sources of income.

Children and young people under the age of 16 and many others above that age are supported within the family, but the majority are also supported through family payments. Around 1.6 million families who are not receiving income support receive family payments.

The cost of social security is $62 billion in 2004-05 (Treasury 2004), over 40 per cent of Commonwealth expenditure. While many argue that it is low compared with the social insurance schemes in other advanced countries, in the current culture of small government it is a constant pre-occupation. In the words of a Centre for Independent Studies publication

Current welfare policy is clearly inadequate. The cost to government has blown out, yet poverty continues (Humphreys 2001, p 19)

The paradoxical perception that it costs too much and leaves people in poverty leads to calls from the Right for its abolition--on the assumption that poverty would be miraculously cured by an abrupt massive increase in employment. However in other quarters there is a growing sense that it is inadequate--but some of the proposals for alternatives to social security raise significant questions of social equity.

This paper looks existing and proposed alternatives and explores their advantages and disadvantages in comparison with our social security model. First, what is the problem?

Risk and Who Should Pay?

Schmid and Schomann describe the contest between the State, market, family and individual in the bearing of risks. They define a risk as

   an event that deviates substantially below or above from the
   expected or 'normal' (which could also be normative) medium
   outcome. A risk becomes social if it is considered not only as a
   matter of the individual but as a matter of some collectivity,
   especially the family, the enterprise or the state. (2004, p. 2)

The responses to this contest that are offered by various economic and social policy analysts reflect different political standpoints. Neo-liberal economists will tend to argue for a reduced role for the State and even less for the market, shifting the burden of risk towards the family and the individual. Feminists will tend to argue for a reduced role for the family in terms of women's financial dependence on men, and for a greater role for the State and the market in facilitating their employment and financial independence. Those representing the disadvantaged would argue for greater economic redistribution across society and a reduced burden of risk for the individual. It is hard to apply Rawls' (1990) 'veil of ignorance' in determining the most equitable allocation of the burden of risk.

The different responses are based on different moral positions on the desirability of collective risk sharing and redistribution, different views on the acceptable level of inequality, and different perceptions of the degree to which individuals are responsible for their own circumstances. That is, how clearly they understand the consequences of their own actions in increasing or reducing risk, how much control they have over their circumstances, and how they have responded to their opportunities.

Measures to reduce risk through collective means tend to be subject to moral hazard: that is people will be freer to behave in ways that increase their risk because the consequences are not so dire. A significant factor in policy design is concerned with constraining individual choice and minimising perverse incentives. But as Schmid points out (1998), this reduction in risk allows people to take advantage of various positive opportunities and allows for greater social and economic flexibility.

Who are the People Not Covered by the Market?

People who are not fully engaged in the market economy, and deriving an adequate income from it, are in a range of circumstances. Not all of these fit the definition of risk given by Schmid and Schomann (2004) above as abnormal, nor do they meet the criteria of being unforeseeable and outside the individual's control. As the TLM framework shows us some are doing socially and economically desirable things.

It is useful to differentiate between reasons for less than full participation:

* those that are normal, predictable and universal: such as childhood and old age

* sanctioned and valued non-market activities, such as child-rearing, caring and volunteering

* those that are associated with improving employability and productivity such as frictional unemployment, education and training, including life-long learning.

* undesirable misfortunes, such as unemployment, sickness and disability

Childhood and retirement can hardly be described as risks, in that they are seen as normal stages of the life-course. Although some children and some older people are employed, at least part-time, there are cultural beliefs and institutionalised barriers that discourage this and justify corresponding means of financial provision.

Bearing and raising children is the largest category of the valued non-market activities. There are different views on whether and for how long that should involve leaving or reducing paid work. Other things could be included in that category--caring for dependent adults, perhaps volunteering, arts, military service, elite sport. These activities do not directly serve the market but they do serve society. They usually involve some element of choice. The fall in the birth rate has underscored concerns that we have not provided an adequate environment for child bearing, in the light of men's and women's aspirations and needs (Mc Donald 2000).

There are certain things that contribute to the efficiency and productivity of the market--education, training, searching for work (frictional unemployment), changing jobs, re-training. The Danish approach of flexicurity aims to reduce the risk to individuals of losing or leaving jobs while allowing for turnover in the workforce.

Then there are those adverse risks that occur unpredictably and only affect some people--unemployment, sickness and disability. Even here however current economic theory is that we require a certain level of unemployment to prevent inflation (Reserve Bank 2005).

Official unemployment rates are at the lowest point they have been for 30 years (Treasury 2004) and labour shortages are said to be emerging in some areas. This gives rise to

   the fear that the present boom may have taken unemployment below
   the "non-accelerating-inflation" rate of unemployment (the NAI RU),
   meaning it's just a matter of time before labour shortages lead to
   wage inflation, which will oblige the economic managers to hit the
   brakes, slow the economy and push unemployment back up to its
   'natural' rate. (Gittins 2004)

The NAIRU analysis implies that a certain amount of unemployment or excess labour is necessary to the stable functioning of the economy.

   The unfortunate feature of a NAIRU or 'inflation first' policy is
   obvious--its success relies on the existence of an unemployed pool
   of labour. In a very real sense, the well-being of a nation depends
   on the misfortune of a minority yet substantial number of
   'sacrificed' citizens (Lawn 2004, p. 272).

Frictional unemployment among highly employable people is thus included among contingencies beneficial to the market; long term, structural unemployment among people with low employability is not. Long term unemployment and low income reduce people's employability, because of a loss of morale and loss of connection with the labour market.

We have heard the problems: lack of skilled workers, chronic long term unemployment, underemployment, mature age unemployment, youth unable to break into the core workforce, low birth rate, unstable families, social exclusion, growing gap between rich and poor. The high numbers of people needing social security are the symptom not the cause.

But more than that, this reflection shows that we need to accept and accommodate a proportion of the population outside the workforce at any particular time, for the reasons just discussed. Many of these need to leave and return to employment, or from part-time to full-time employment. As Schmid states:

the notion of' full employment' in the traditional sense can no longer be sustained. The slogan of 'work for all' has to be qualified anew. The 'all' in the traditional sense has been restricted, according to the breadwinner concept, to the male heads of households. This ideology can no longer be upheld. In other words, the full employment objective in the sense of work for all under the condition of eight hours a day, five days a week, 46 to 50 weeks a year, 40 to 50 years during an average individual lifespan, is both outdated and unachievable (1998, p. 4).

However, recent Government statements imply that employment must be privileged above 'welfare' (Grattan 2005). Does Australia have fair mechanisms in place to support those who are outside the workforce at particular times for various reasons?

In thinking through alternatives to the Australian social assistance system, it is useful to think in terms of three principles--Direct transfers, Income smoothing and insurance. There is also the question of the contest between Government (wider tax paying community), the market (including employers' obligations), the family and the individual.

Direct Transfers

Government

The Australian social security system is paid directly by central government from tax revenue. It provides pensions for the aged, people with disabilities, sole parents, carets and some other small groups: the basic rate for a single pensioner ($235.35 a week) is 50 per cent of the Federal minimum wage for fulltime adult employees of $467.40 per week (AIRC 2004) and 25 per cent of male total average weekly earnings. The rate for a member of a couple is slightly less, $196.50. Allowances (benefits) are provided for unemployed people, students, widows, older partners of people on a pension or benefit and partnered parents caring for children, people with a short term illness preventing them from working and certain other groups. These are paid at a lower weekly rate--$197.30 for a single person, $177.94 for a member of a couple. Austudy, for adults studying full-time, is substantially less again.

Other OECD countries also have a direct cash payment system underlying their social insurance. These are less developed, meaner and usually more discretionary than the Australian system and of course only provide for the small proportion of people in need who are not covered by social insurance.

The Australian system is simple to administer. It is targeted to need so is more redistributive than social insurance. It has the capacity to cover everyone, not only those who have contributed through employment. It is flexible--entitlements can be extended or changed as contingencies arise, without the need to build up entitlements over a long period. Finally it involves much less churning of funds through Government: Australian taxes tend to be low in comparison to the combined tax and social security take of other countries.

By tweaking our system it can be used to provide supplements to people earning very low incomes. Around 10 per cent of people on payments have some earnings, including 30 per cent of sole parents. It is difficult to provide earning supplements through a social insurance system, as it is based generally on income replacement rather than supplementation.

However the Australian means tested system tends to lack legitimacy in comparison with social insurance models. Many tax payers perceive themselves as unlikely ever to benefit and resent the transparent re-distribution involved. In social insurance regimes, nearly everyone has an entitlement and it has very strong popular support. In Australia, recipients tend to be stigmatised not only by the public but also by Government. There is little popular objection to quite harsh and intrusive surveillance and conditions.

Payment rates are therefore kept very low to maintain an incentive to work. This means that those leaving employment to bear or care for children, take up fulltime education or training or retiring must take a serious drop in income. Similarly people who lose their jobs or become disabled can experience a sudden fall in income, if they are not cushioned by other employment benefits.

Social security has been based on a model of the family comprising a single breadwinner with dependent spouse and children and payments are therefore means tested on family income and assets. A person in any of the situations above, whose partner is employed, is not compensated in his or her own right. Although we provide for parental responsibilities through parenting payments, family payments and the baby bonus, these provide limited compensation for women and families for lost income. Women who are not earning because, for example, of child-bearing and raising are expected to rely financially on employed partners. However, as women are increasingly integrated into the labour force, they aspire to their own financial rights. Men who can't work are also forced to depend on their employed partners, if they are lucky enough to have them.

The single breadwinner model also underlies the rates of payment but, as community norms are now based on two income families, there is a fast growing gap in relative incomes between families dependent on social security and those in paid work.

Income Supplements for People in Low Paid Work

Market incomes are diverging--particularly between those in part-time or intermittent work and those in permanent full-time work, but also on an hourly pay basis. Some have argued that wages at the bottom end need to fall further to reduce unemployment (Dawkins 2002). This has sparked an interest in whether the incomes of low wage workers should be supplemented, both to maintain their living standards and to maintain adequate work incentives.

The pension and benefit income tests provide some capacity to combine social security payments with paid employment. A single pensioner can receive some supplementation with an earned income up to around 1.4 times the Federal faward minimum and a pensioner couple the figure is 2.3 times that award rate. This is higher for pensioners with dependent children, who also receive family payments. This capacity is far more restrictive for people on Newstart or other benefits--the limit for combining part payment with earned income is earnings of 60-70 per cent of the Federal award minimum

Some proposals have looked at ways to improve the net income of low paid workers or families--this could be used particularly for supporting people combining part-time work with caring or education, although they are designed for low income people in general. Peter Dawkins and the Five Economists proposal for an earned income refundable tax credit (Dawkins 2002) similar to that existing in the US. Humphreys (2001) has proposed a negative income tax along similar lines. In the latest CEDA Information paper Perry (2005) I proposed instead a version of a guaranteed minimum income, integrated with the income support system. I wrote my paper because I believed that the others had overlooked the fact that substantial support is already provided to low to middle income families with children through family payments.

Another variant of public direct payments advocated by some is basic income--flat payments without condition or means tests.

Direct Payments by Employers

Employers provide direct payments in the form of paid leave. The level of leave payments is usually equivalent to full or half earnings.

Long service and recreation leave applies equally to all ongoing employees.

Sick leave applies only to those who need it, and it is quite probable that employers try to avoid taking on staff who they assume are likely to take more than average sick leave.

Some employers, particularly the Commonwealth and State public services, provide paid parental leave. If paid parental leave was introduced into universal conditions of employment, it might lead to more discrimination in employment against women of child bearing age, as it would add to the employment costs. With a move towards individual workplace agreements, other employees could be expected to trade off parental leave entitlements, leaving young women disadvantaged.

Some employers also provide scope for employees to take some paid time off for study purposes, and, where the study is likely to improve the worker's productivity, may cover the cost of tuition as well.

A paper at this conference by Gahan and Stricker proposed converting long service leave entitlements into a leave bank--extending entitlements to casual workers and allowing those who change jobs to transfer their entitlements. This could cover people who wished to take leave for further education or maternity leave. This would be more restrictive than the present system, for those who are eligible--it can now be taken when the employee chooses. While the idea of a leave bank warrants further thought, using it for parental leave could disadvantage childbearing women who would then not have the same scope as men be able to use it for continuing education.

In other papers here, John Buchanan raised ideas for extending pooled entitlement funds--such as group training. Michael Keating mentioned 'learning entitlements' which could be jointly funded by individuals and government. There is a potential role for employers.

The idea of funding by employers, whether for universal entitlements for all workers or on an insurance basis, has merit to cover contingencies that increase flexibility and productivity of the labour market. Naturally levies on employers or an extension of employee leave entitlements add to the total cost of employment, adding a 'levy wedge' to the wages paid. However, arguments can be made that where employers and businesses are co-beneficiaries of activities such as training in particular, that they should be expected to bear a proportion of the cost.

Income Smoothing

Income smoothing is redistribution across the individual life course. This can be either prospective or retrospective. The advantage is that it addresses some of the problems raised yesterday by Schmid--myopia or people's underestimation of future risks. The prospective schemes also have the advantage that the cash put in in advance is greatly augmented by the earnings of that capital over time. The disadvantage of individual income smoothing is that it does not involve redistribution across people to address the great disparity between different people's lifetime incomes.

The income smoothing rationale works equitably for people whose life-time earnings are average or above average, for a universal contingency such as retirement or when it provides future financial advantages, as in the case of higher education. It is inequitable when individual re-distribution over the life-course replaces social redistribution from rich to poor. Individualisation of funding exacerbates lifetime inequality.

The classic prospective example is superannuation, where workers in effect forgo part of their wages for retirement income, assisted by tax concessions. This provides an alternative, or supplement, to age pension. One way of looking at it is that those earning higher incomes during their working life should be made to provide for their own retirement rather than expecting other citizens to pay for it.

Home ownership is an example of income smoothing, where typically a person or family spends years of working life paying off a home which then provides secure housing in retirement with low costs compared with renting. In addition to this income smoothing, there is a further associated level: Both the Commonwealth Government and some financial institutions offer loans or annuities to retired people against the capital invested in their homes, repayable from their estates.

Bruce Chapman (2004) promotes retrospective income smoothing through income contingent loans, the principle underlying the HECS scheme, where a loan is repayable only when a person's income reaches a certain level. He has suggested that these could be used for parental leave, training for elite athletes and various other contingencies.

The justification for the HECS scheme is that as university degrees are likely to lead to higher earnings, students should contribute to the cost, when and if they realise that advantage. Originally the threshold for paying HECS debts was set at average earnings. It then fell to well below that level.

At one stage Chapman proposed income contingent loans for maternity leave (ABC 2002) which would mean that an individual woman should bear the costs of lost earnings, if her income subsequently returns to a higher than average level. In earlier work (Beggs and Chapman 1988), Chapman himself calculated the high lifetime costs for women from reduced labour force participation due to having children. This showed that child bearing women were bearing costs not shared by men or childless women. Although a woman's earnings may, at a later point in time, reach a level at which the debt is repayable, this is not the same as saying her lifetime earnings will be higher than average. I regard this use of income contingent loans as failing the test of universality or income advantages.

Chapman at this conference advocates HECS for Technical College (TAFE) students.

Another problem with the concept is that expenditure is not smooth. The years of working life have become shortened by later entry and earlier retirement. For some people, the need to pay off HECS debts and student loans and save for retirement come at a critical time for purchasing a house and rearing children. At such times their expenditure in relation to their income might be higher than at any other time in their life-course.

Some people have suggested allowing people to use their superannuation for situations other than retirement, such as unemployment, maternity leave, training or home purchase. This may help with periods of high expenditure and low income, but as your entitlement is individual, any draw down before retirement age depletes the amount you will have later on.

The Chifley Research Centre and Mark Latham proposed matched savings or nest-egg accounts,, supported by tax concessions, for children, young people and others to facilitate asset building (See for example Allen Consulting Group 2003; Institute for Public Policy Research 2003). These could be used later for purchasing education or housing. To be in such schemes a family must have enough income to be able to save. The Victorian pilot scheme, Saver Plus, has shown very positive results in assisting families to save relatively small amounts. Saver Plus involves matched contributions from philanthropic organisations (the Brotherhood of St Laurence and the ANZ Bank), and financial management training for participants. The paper by Roslyn Russell at this conference discusses this project.

These proposals are another form of income smoothing. But if the purpose is to enable young people to purchase a house or pay for education, does it replace other existing forms of financial support, such as First Home Owners' grants or public support for education? Would the tax concessions outweigh such support?

Insurance

The second common system is insurance. This involves pooling risks of adversity. The primary difference between the terms 'insurance' and 'income smoothing' as used in this paper is the sharing of risk between individuals in insurance models. While insurance has a number of benefits, its major deficiency is that those who are not in the pool, for example if they have not had enough qualifying employment, are not covered. Also it may be seen as inequitable in that higher income people often pay higher levies and get higher benefits. Its equity aspects depend on the way the contribution and benefit regimes are structured.

Medicare is an example of insurance: the levy is paid by all individuals so that the healthy carry the unhealthy.

The major income support example in Australia is workers compensation where employers make mandatory contributions on a payroll basis which is used only by some employees. Note that sick leave in contrast is not insurance based: it is paid directly by employers to individual workers.

Throughout the twentieth century there were various plans to introduce a national insurance based system, the last by Hancock in 1976 (National Superannuation Committee of Inquiry).

Most other OECD countries have social insurance as the primary social security system. It is usually paid by employers and covers individual employees in the contingencies of unemployment, sickness and disability, parental leave and retirement. These entitlements are not means tested and payments are usually set at some percentage of former earnings. In reality the schemes tend also to be subsidised by central governments and often have a minimum and maximum benefit level modifying the earnings related level.

Unlike individual superannuation, in an insurance system a worker can claim in the case of illness etc without reducing their retirement entitlement.

Another very important feature is that because entitlement is based on employment, it is individual, not based on family need. This means for example that it can cover parental leave, without regard to the other parent's earnings. In many other countries parental leave is paid through national social insurance.

In Australia, paid maternity leave would be best funded on the workers' compensation model: a payroll based levy which could be used by parents taking time off work for child-bearing and care. This would spread the cost of parenthood across the whole workforce on a more equitable basis than the income smoothing or direct paid leave suggestions discussed above. Parental leave entitlements could perhaps be extended to people caring for frail adults.

In unpublished work, I have proposed an insurance based scheme for new migrants (other than refugees) to cover them while they are excluded from social security benefits. All would pay a one-off levy as a condition of entry and the pooled funds would be used to protect those who became unemployed. Those who could not pay before entry would be able to borrow the levy. The funding would be managed by non-government financial institutions, while eligibility for claims could be assessed by Centrelink. This model would be similar in principle to health insurance arrangements for foreign students studying in Australia. I regard this as second best to allowing them normal access, but that is not politically acceptable at the present time.

Other Proposals for Change

Bill Mitchell and others at the Centre for Full Employment and Equity propose a different approach altogether--a job guarantee in which the Commonwealth employs all those who are not employed by the market, at a minimum hourly wage rate (Mitchell & Watts 1997). Bob Gregory and Sue Richardson at this conference mentioned the idea of public employment of low skilled workers.

Conclusion

Australia, like other countries, has components of all these means of support but the predominant one is direct payments through social security. In other Western countries the predominant means is social insurance.

The Australian social security system has both advantages and disadvantages in relation to socia insurance systems.

It is not often feasible to import a measure from an overseas system and graft it onto ours. Thus proposals for new forms should be designed and assessed against our existing measures For example, advocates of adopting the US Earned Income tax credit to support working poor families failed to acknowledge that we already have one built into our system, the Family Tax Benefit. If we were looking for ways to assist the working poor, then we should build on our existing wage, tax and social security institutions to ensure system compatibility.

Another example is Paid Parental Leave. This works well in social insurance schemes because it provides individual entitlement to earnings replacement for women, even if their husbands are employed, and can be tied in with statutory rights to job security. Although we have parenting payment in our social security system it does not provide these features and cannot really without raising issues of inequity with other groups. Bruce Chapman has suggested an income smoothing HECS type financial advance so that women pay for child-bearing later in life. Others have suggested requiring employers to pay directly, along the lines of Sick Leave. These need to be assessed against both the long and the short-term impacts; an exercise that can be assisted by invoking TLM principles and frameworks outlined elsewhere in this collection.

References

ABC (2002), New Ideas in the Maternity Leave, discussion: Interview with Dr Bruce Chapman, Life Matters, 7/10/2002 ABC Radio National.

ABS (2004), Labour Force Survey, June 2004, Australian Bureau of Statistics, Canberra.

AIRC (2004), Safety Net Review Decision, May 2004, Australian Industrial Relations Commission, Melbourne.

Allen Consulting Group (2003), Asset Based Policies: Matched Savings Accounts: Exploring Options, Report to the Chifley Research Centre, Canberra.

Beggs, J. and Chapman, B. (1988), The Foregone Earnings from Childrearing in Australia, Centre for Economic Policy Research, Australian National University, Canberra.

Chapman, B. (2004), Income Related Loans for Public Policy, unpublished, The Australia Institute, Canberra.

Dawkins, P. (2002), The 'Five Economists Plan': the Original Idea and Further Developments, Discussion Paper No 450, CEPR, ANU, Canberra.

FACS (2004), Annual Report 2003-2004, vol 2, Commonwealth Department of Family and Community Services, Canberra.

Gahan, P. and Stricker, P. (2005), The Challenge of Reconciling Working Time Demands and Life Transitions, Transitions and Risk: New Directions in Social Policy Conference, Centre for Public Policy, University of Melbourne, Melbourne.

Gittins, R. (2004), When low jobless rate turns to labour shortage, Sydney Morning Herald, 13 November, Sydney.

Grattan, M. (2005), '"Work first" to be welfare aim', The Age, Melbourne, February 28.

Humphreys, J. (2001), 'Reforming Wages and Welfare Policy: Six Advantages of a Negative Income Tax', Policy, vol. 63 no. 3, p. 242-263.

Institute for Public Policy Research, (2003), Nest-Egg Accounts: Key Questions and Policy Options, Report to The Chifley Research Centre.

Lawn, P. (2004), 'Full employment and ecological sustainability: comparing the NAIRU, Basic Income and Job Guarantee approaches', paper presented to the Full Employment Conference, Centre for Full Employment and Equity, University of Newcastle.

McDonald, P. (2000), Gender equity, social institutions and the future of fertility, Journal of Population Research, 17(1), 1-16.

Mitchell, W. and Watts, M. (1997), 'The path to full employment', Australian Economic Review 30: 436-444.

National Superannuation Committee of Inquiry (1976), Final report, Australian Parliamentary paper, Canberra: AGPS.

Perry, J. (2005), Henderson Revisited: The case for in-work benefits, CEDA Information Paper 82.

Rawls, J. (1990), A Theory of Justice, Oxford: Oxford University Press [Tenth impression, first published 1972].

Reserve Bank (2005), Statement by the Governor, Mr Ian Macfarlane: Monetary Policy, Media Release No: 2005-04, 2 March, Reserve Bank of Australia, Canberra.

Schmid, G. (1998), Transitional Labour Markets: A New European Employment Strategy, Discussion Paper FS I 98-206, Wissenschaftszentrum Berlin fur Sozialforschung.

Schmid, G. and Schomann, K. (2004), Managing Social Risks Through Transitional Labour Markets: Towards a European Social Model, TLM.NET Working Papers No. 2004-01, Amsterdam: SISWO/Institute for the Social Sciences.

Treasury (2004), Federal Budget Paper 4, 2004-05, Commonwealth Treasury.

Endnotes

(1) An earlier version of this paper was presented at the Transitions and Risk: New Directions in Social Policy Conference, Centre for Public Policy, University of Melbourne, 23--25 February, 2005. The author wishes to thank the participants in the workshop and two anonymous referees for their constructive comments to earlier versions of the paper.

Julia Perry, Social Justice and Social Change Research Centre, University of Western Sydney