Wrong Way---Go Back

The Effects Of The Social Welfare System On Unemployment

Des Moore

"We can have exactly as many paupers as the country chooses to pay for;"
(Thomas McKay in Methods of Social Reform 1896, London John Murray)

"11% unemployment is a matter of choice."
(Ted Evans in Reducing Unemployment at National Forecasting and Economic Policy Conference
28 October 1993)

Executive Summary

This paper reviews some recent research in Australia and overseas relating to the effects on unemployment of increasing generosity in assistance provided for the unemployed.

There are sound grounds from a theoretical viewpoint to expect that increasing replacement rates (increasing proportions of income from work that are replaced by unemployment benefits) will increase the disincentive to work and lead to higher unemployment. The effects of increasing replacement rates will be exacerbated in circumstances where the labour market is highly regulated, as it has been and remains in Australia.

A Lesson From History

Perhaps foreshadowing the Higgins Harvester award, the "liberalised" poor laws in eighteenth and early nineteenth century England attempted to set a subsistence income related to family size. Experience under these arrangements (which were reformed in the early 1830s) show the adverse effects of excessive generosity in relief, including higher unemployment and demoralisation. "Only" one sixth of the English population then "lived at the expense of public charity", according to Alexis de Tocqueville. In Australia today about one quarter rely on government benefits for their principal income source.

Some International Comparisons

In an important change of direction within the OECD, research in recent years has increasingly highlighted the adverse effects from excessive government intervention in the labour market, rather than market failure. And, as even governments have come to recognise the reality that continued high unemployment may partly reflect excessive generosity in unemployment benefits, many have sought to reduce structural unemployment by cutting benefit entitlements.

However, OECD research shows that to date only four countries have reduced their structural unemployment rate (the rate below which wages growth starts to accelerate) in recent years. Australia is not one of them and its structural rate is "stuck" at 8.5%, compared with the OECD average of 7.1 %.

That research also shows that, measured in relation to unemployment assistance only, replacement rates in OECD countries have risen significantly over the past thirty years. However, an important development has been the increasing acknowledgement of the need to take into account the effects of the whole range of social benefits. Thus, the OECD calculation of replacement rates after taking account of the taxation treatment of benefit entitlements as well as some other social welfare benefits, reveals that they come up to 70-80% of earnings that are two-thirds of average earnings for both the OECD and Australia.

Most importantly, another piece of OECD research concludes that not only has the increase in replacement rates been a significant contributor to some countries' higher unemployment rates, but it is also a significant factor in the difference between some countries' current rates and the OECD average rate. In the case of four OECD countries the "generosity" factor is assessed as accounting for 3-5 percentage points of the differential, though only 1 percentage point in the case of Australia. However, this calculation of replacement rates and their effects takes account only of unemployment assistance.

Even so, the decomposition of factors explaining these OECD estimates of differences between Australia's structural unemployment rate and the OECD average suggests that the relative generosity of unemployment benefits is almost as important as the "job protection, unionisation and co-ordination" factor (Attachment A)

The Australian Evidence

The "jumps" in Australian unemployment in the mid 1970s and again in the early 1980s and 1990s were not caused by the significant liberalisation of social security and welfare benefits since the mid 1970s. However, the fact that our structural unemployment rate has increased so much since that time, and has apparently been maintained at such a high level for an extended period, points to the liberalisation of social security and welfare as a contributing factor.

Australian research has generally focused on replacement rates calculated by reference mainly to unemployment assistance only and, hitherto, has tended to suggest very limited effects on unemployment from increases in such rates. One 1994 study in this category shows that increases in replacement rates relative to minimum wages brought them to relatively high levels by 1993-94 (ranging from 75% to over 90% for couples and from 52% to 65% for single people over 21). (Attachment B shows trends in both minimum wages and unemployment payments). This implies that, if account were also taken of other cash and non-cash benefits, replacement rates would be over 100% in most cases. Given that most unemployed start at the lower end of the wage scale, and that many OECD countries have no minimum wage because they treat the minimum welfare safety net payment as a 'reservation wage', this suggests a significant disincentive effect to work in a section of the labour market particularly vulnerable to unemployment.

The 1996 National Commission of Audit (NCA) report did in fact highlight the increase in other social welfare benefits of both a cash and non-cash nature and included a case study showing that the real value of a range of cash benefits to an unemployed family increased by nearly 24% between 1984 and 1996, when it amounted to almost $22,000 pa, or about two thirds of average earnings. The Report also pointed out that, for households whose principal income was unemployment assistance in 1993-94, the Household Expenditure Survey shows non-cash entitlements as adding an average $150 per week to household income. (The Survey also shows that, after taking account of all taxes and selected non-cash benefits, such households had an average income per person---including children and the aged---of just over $150 per week).

The NCA Report drew attention to the need to take account of the total social security package, expressed "concern" about the reduction in work incentives and stated (cautiously) that this "may drive average rates of unemployment (and government spending thereon) up over the longer term."

An examination of the total available package of cash and non-cash benefits implies that income support is extensive for those who choose not to work and/or who are eligible for unemployment assistance. Apart from unemployment assistance itself, potentially available benefits include rental assistance, family payments, parenting payments (which have replaced the sole parent pension and the parenting allowance), child care assistance, disability support pension, health cards, free medical and public hospital treatment, subsidised pharmaceuticals, and subsidies for school and post school education. Most of these benefits are available in addition to unemployment assistance and some do not count for purposes of the income test for such assistance, which has itself been progressively liberalised to the point where 15% of those working part time are also receiving a partial unemployment allowance.

In fact, the 1993-94 Household Expenditure Survey shows that, for households whose principal source of income is the unemployment allowance, the average amount of cash and (selected) non-cash government benefits was $434 per week per household in that year. Of that amount, unemployment assistance itself constituted only $231 per week, or about 53 percent.

While the range of benefits available to those who become eligible for unemployment assistance itself provides a considerable incentive to secure such assistance, the extent to which people apply for and receive unemployment allowances also depends on the stringency of work and activity tests. That is difficult to assess. Although activity testing and "reviews" of recipients have increased in recent years, there remains more than a suspicion that there is still a significant capacity for the unemployed to earn other income and avoid declaring it, and/or to be selective in taking job offers. Certainly, the actual and projected results of reviews from increased "data matching" suggest that 15-20% of payments are subject to "over-payments" of one degree or another. Most importantly, the penalties applied to those caught (mostly only repayments of, "over-payments") have not been calculated to act as much of a deterrent to future avoidance, though- an additional penalty has been introduced with effect from 1997-98.

The recent Clemenger Advertising survey showed that over 70 per cent of people surveyed nominated dole bludging as their principal concern and that it was the highest ranking such concern. This concern was apparently self-generated rather than prompted from a list. It adds weight to other perceptions that the official figures of those receiving unemployment allowances overstate the extent of unemployment. However, this has to be balanced against those receiving such allowances but who are really partially unemployed in the sense that they would like to work more; and those who have dropped out of the work force because they have become discouraged by the difficulty of obtaining jobs but can also be regarded as unemployed.

The Scope For Action

Action to reduce the generosity of the present unemployment assistance scheme would desirably be combined with substantive deregulation of the labour market. Indeed, action in one area without action in the other would limit the potential for reducing unemployment.

Even so, drawing on action taken in other countries, there is a range of measures that could "encourage" more of the unemployed to work and thus help reduce unemployment. These include tightening work activity tests in regard to jobs judged by unemployed as "unsuitable", reducing the amount of assistance after a period (say, twelve months), requiring a longer wait for assistance for those who quit jobs voluntarily (about 16 % of unemployed), requiring greater use of liquid funds before payment of allowances, eliminating the de facto indexation of unemployment allowances to changes in average weekly earnings (as proposed by the NCA), and imposing penalties (in addition to repayments of "over-payments") on those who are detected as cheating the system.

Another possible measure to encourage the unemployed to accept work would be to provide an income tested tax credit payable only to those in work, perhaps by making income supplements to low income families under the Family Tax Payment scheme conditional on the recipient being in work. By supplementing incomes of those on low wages, such an approach would help counter objections that deregulation of the labour market would "only" lead to the creation of a class of "working poor".


The publication of OECD research showing that differences in "generosity" of unemployment assistance are a significant explanatory factor in countries having higher unemployment levels than others, and the increasing recognition that many other welfare benefits also have disincentive effects on work, highlights the need to undertake a comprehensive review of the effects on unemployment of the "generosity" of the whole range of relevant cash and non-cash benefits in Australia. There seems little doubt that such generosity accounts for a not insignificant proportion of Australia's present unemployment rate. Publication of estimates of the adverse effects of present arrangements would at least reduce pressure on the Government to adopt other interventionist policies.

There is in any event a strong case for adopting a range of measures to "encourage" more of the unemployed to work, and thus help reduce unemployment. Such action would preferably be taken in conjunction with substantive deregulation of the labour market, but should not await that.

More generally, while Australia provides less middle class welfare than Europe, the case for reducing the welfare liberality line and welfare dependency here is also a strong one. If combined with lower taxation and deregulatory reforms of the labour market, additional low income jobs would be created and that would reduce the number of two income households in poverty. Overall, such a result would surely be beneficial both economically and socially.


The H R Nicholls Society believes that Government intervention in the labour market has adverse effects on employment and economic activity generally. The main focus of the Society's attention has been on the regulation by government or government agencies of the terms and conditions on which employers can take on employees, that is, primarily on the extent to which the demand for labour is adversely affected by regulation, resulting in high levels of unemployment.

But there is also another form of government intervention with implications more for the supply side of the labour market and which is increasingly being recognised as a major cause of high unemployment. 1 am referring to Government intervention that involves the provision of unemployment and related welfare benefits.

It is usually argued that the subsidy which unemployment benefits provide for job search improves the functioning of the labour market and thus tends to reduce unemployment. Further, some people will choose to work even if they could obtain a reasonable, free income without doing so. Nonetheless, the availability of such benefits also has the potential to increase unemployment by acting as a disincentive both to searching for and accepting jobs. By also reducing the restraint on wage claims by those who are employed unemployment benefits tend to add to the cost of employment and, hence, to unemployment itself. The additional taxation required to finance unemployment benefits should not be overlooked either: as the OECD has emphasised, "taxes increase the cost of employing workers, particularly low-wage workers".1

The extent of the disincentive to become employed depends on a number of factors. Much obviously depends on the generosity of unemployment benefits relative to wage levels and on the stringency of the conditions which have to be met in order to become (and remain) eligible to receive unemployment assistance, including the work tests and the penalties for not declaring income earned when on the dole.

But unemployment benefits are far from being the only form of social assistance with potential effects on the incentive to work and, as the OECD has recently pointed out, "the tax and benefit system must be considered in its entirety when attempting to assess the differing work incentives facing the working population".2 An important development in the analysis of the reasons for high levels of unemployment has been the increasing taking-into-account of the effects of a whole range of social benefits provided by government. Such analyses are showing that, if persons are eligible for a range of benefits, their income from such benefits can amount to a high proportion of what they might be expected to earn from work.

The interplay between the tax and social welfare systems is also relevant. The disincentive to take up employment may be increased if income from work is subject to relatively high rates of tax and withdrawal of benefits.

There is also a link between the extent of unemployment assistance and the extent of regulation of employment conditions. If unemployment assistance is available on relatively liberal terms and conditions in circumstances where work is likely to be unavailable at the authorised rates of pay to certain types of workers because of excessive regulation of the labour market, that will tend to increase both applicants for and recipients of such assistance. Thus, excessive labour market regulation in circumstances where relatively liberal unemployment assistance is available can be said to have a double whammy effect on unemployment levels---it both prevents people from getting jobs and encourages them to register for assistance.

None of the foregoing is to deny that the unemployment figures also understate unemployment. For example, the difficulty experienced by some of obtaining offers of employment in a regulated market will also cause people to drop out from the work force altogether (the so-called discouraged worker effect). As they will then not be available for work, such people will not be counted as "unemployed" even though, in reality, they are. In addition, some of those working part time are also "unemployed" (in the sense that they would like to work more hours) but will not be counted by the ABS as unemployed even though, under the income test for unemployment assistance, the "free" income provisions will allow a proportion of part timers to receive the allowance. On the other hand, a proportion of those receiving unemployment assistance will actually be working and should not be counted as "unemployed".

The growing awareness of the potential for unemployment and related welfare benefits to add to unemployment numbers is resulting in increasing attention being given to the issue.

A Lesson From History

There has been a long history in England of government intervention to help "the poor"---and an on-going debate about the measures that should be used and their effects. The poor laws dated back to the sixteenth century, when the dissolution of the monasteries led the government to make provision for the indigent who had previously been cared for by the church. Under the Elizabethan poor laws, "outdoor" relief was provided by government as well as "indoor" relief for the aged and infirm and temporary shelter and work for the able-bodied in workhouses or poorhouses.

While the poor law system remained in existence until replaced by the welfare state after World War 11, it underwent a major change in the 1830s when it was perceived as being in crisis after the system was liberalised in the late eighteenth century to provide what sounds very much like a precursor to the Higgins Harvester wage award. Thus, in a decree whose application quickly spread, the JPs of Berkshire, meeting at Speenhamland in 1795 at a time of bad harvests, decided that "every poor and industrious man" whose earnings fell below a given standard would receive a subsidy to bring his income up to subsistence level. The standard was determined by the price of bread and the size of the man's family.

The re-issue by the Institute of Economic Affairs in May 1997 of the 1835 Memoir on Pauperism by Alexis de Tocqueville (best known for his Democracy in America) is a timely reminder of an important lesson from that 1795 liberalisation. The introduction by American historian, Dr Gertrude Himmelfarb, argues that the liberalisation set off a "cycle of evils" and her exposition is certainly not without relevance today.3

Thus, Dr Himmelfarb points out that the increase in taxation to pay for the higher subsidies forced wages down while the provision of more generous relief induced a rise in agricultural unemployment. This led to a reduction in the quality of available labour which, in turn, caused a fall in productivity and an increase in food prices. The more generous relief encouraged earlier marriages and more children and the resultant faster growth in population led to still lower wages. More generally, the poor were said to become demoralised.

By the early 1830s the pressure for reform led to a Royal Commission which produced a 400 page preliminary report (of which a remarkable 15,000 copies were sold at four shillings a piece) and a 200 page final report (of which no less than 20,000 copies were either sold or distributed free). That report led to an attempt to distinguish between "the independent poor" (the labouring poor) and the indigent (the paupers), who alone were to be the recipients of relief. The able-bodied were to be assisted only in the workhouse under conditions less favourable than those of the independent labourer, while those regarded as truly indigent would be cared for as they had been before.

Notwithstanding the reforms enacted in 1834, de Tocqueville argued in his Memoir that the whole concept of public charity was flawed insofar as it applied to the able-bodied because, he believed, men will work only to sustain life or to improve their condition. But if men are given a right to public charity they will not seek to improve their condition -"a right to public charity, unlike other rights, degrades the man who claims it by condemning him to a life of dependency and idleness". Private charity is different because it does not carry with it any right or assurance but is as spontaneous and temporary as public relief in times of public calamities. However, de Tocqueville noted that private charity may be insufficient for the developing industrial age. He promised a further tract dealing with the problem of industrial pauperism but that never appeared.

de Tocqueville's analysis is of particular interest because he observed several court sessions where cases involving paupers were heard and had communications with the principal author of the Royal Commission report. His discovery, with "indescribable astonishment", that "one-sixth of the inhabitants of this flourishing kingdom live at the expense of public charity" doubtless also helped form his views. One wonders what he would say today with about a quarter of the population relying on government cash benefits for the major part of their income!4

Some International Comparisons

I want now to jump ahead in time to outline the results of some recent OECD research into the effects on unemployment levels of increasingly generous unemployment and other welfare benefits over more recent years. During my recent visit overseas I had the opportunity of discussing the results of this research, which is part of a broader exercise undertaken as a follow-up to the 1994 OECD Jobs Study.

It is worth mentioning at the outset that, as the realities of high European unemployment have come to be recognised, there has been a clear shift within the OECD towards favouring more market oriented solutions, or at the least of pointing out the problems with interventionist policies whose avowed intention is to be "socially equitable". It is evident that there are still "political" constraints on what can be said publicly: I was told, for example, that although research shows that youth unemployment rates are low where wage rates are low, "nothing has been published about this". But the encouraging thing is that the proclivities of most researchers are to highlight failures of government intervention rather than (as in the past) of failures of the market.

It is also noteworthy that the message is even starting to produce action at the political level. Thus, "the response of most governments to high and persistent unemployment has been to cut benefit entitlements."5

The broader exercise undertaken by the OECD has involved a detailed examination of the factors influencing unemployment in individual countries and of the effects of various measures taken to try to reduce the "structural" unemployment rate. (The structural unemployment rate is the rate below which wages growth starts to accelerate and is thus a determining factor in setting the speed limits to growth). Up to 1996 only four OECD countries are assessed as having reduced their structural unemployment rates (the Netherlands, New Zealand, the United Kingdom and Ireland) since 1990, as against 11 countries where it has increased and 8 where it is judged to have remained fairly static.

In Australia's case the structural rate is estimated as having remained fairly static over the last six years at around the very high level of 8.0-8.5%, compared with the OECD average of 7.1%. While the currently assessed rate of 8.5% should not be taken too literally, it is worth pointing out that it is not an average that has been derived from an analysis stretching over a long period. Its calculation is in fact based on the experience of the period 1990-95 and shows a slight increase compared with previous OECD estimates for both 1990 and 1986. One implication is that the Australian labour market has become slightly less flexible since the mid 1980s and even since 1990.

Turning to the question of the effects of unemployment benefits, the OECD's Jobs Study of 1994 found that the increased generosity of unemployment benefits could have contributed to the rise in unemployment over the previous thirty years but that there were long time lags of up to a decade before the full effects were felt. Since then the research has been taken a good way further.

The OECD's July 1996 Employment Outlook concluded that "in the longer term high replacement rates will tend to undermine work incentives" and that, "without counter-vailing factors", the higher are replacement rates "the higher unemployment will be." (The replacement rate is the proportion of income from work that is replaced by unemployment benefits). It also observed that, "by freeing the unemployed from having to take less than ideal jobs", benefits "increase the duration of unemployment spells" and "by altering the incentives in wage bargaining", they tend to push up wages and hence add to unemployment.6

The Outlook shows that the overall average replacement rate has about doubled to about 30 percent between the early 1960s and 1995. (That average replacement rate is calculated by reference to average earnings and two-thirds of average earnings). No historical figures are given for Australia but it is clear from a chart that our average replacement rate has also almost doubled from around 15% since the early 1970s to 27% in 1995.

The Outlook notes that, while such replacement rates suggest that "benefit systems do not impose large work disincentives... a more detailed examination of incomes in and out of work suggests that such a conclusion would be premature". In fact, a further calculation taking account of the taxation treatment of benefit entitlements (which are sometimes untaxed), benefits to children, social assistance and housing benefits reveals a quite different picture. On that basis, the OECD (unweighted) average proportion of income replaced by benefits for average earners comes up to 60-70% (depending on the type of recipient) and the Australian rate is similar to that average. For those earning two-thirds of the average, the replacement rates range from 70-80% for both the OECD average and Australia.7

Perhaps the most important piece of recent OECD research, however, has reached the conclusion that "high levels of unemployment benefits increase structural unemployment significantly" and "high UB replacement rates have a direct impact on unemployment... and also a severe impact on the speed of labour market adjustments."8 It includes estimates that the increase in the replacement rate over the past two decades has had a significant effect on unemployment rates in several OECD countries, especially in Spain and Ireland. It also estimates that the current level of unemployment benefits alone relative to wages accounts for between 3 and 4 percentage points of the differential between the average OECD "structural" unemployment rate of 7.1 % and the rate in countries such as France (9.7%), Denmark (9.7%), Belgium (9.7%), and the Netherlands (9.1%).

For Australia, the estimates from this research suggest that only 1 percentage point of the differential compared to the OECD average is due to the generosity of unemployment benefits and that only 1 percentage point of the increase of 7.6 percentage points in our structural unemployment rate since 1971 is due to an increase in the relative generosity of unemployment benefits. But, if other forms of assistance were also to be taken into account, the effect would surely be considerably greater given that the replacement rates including such other assistance would be much higher.

The decomposition of the factors explaining differences compared with the OECD average unemployment rates puts the role of generosity of unemployment benefits in perspective. Attachment A summarises those factors and is of particular interest in bringing out the fact that active labour market programs have played a role that is tiny---and which for some countries, including Australia, actually increased the unemployment differential. The chart suggests that the relative generosity of unemployment benefits explains almost as much of Australia's unemployment differential as the "job protection, unionisation and co-ordination" factor.

The Australian Evidence

The mid 1970s marked a major turning point in Australia's experience with provision of social security type benefits. The general story is one where unemployment and spending on social welfare both jumped sharply in the mid 1970s, with social welfare subsequently growing at least in line with GDP and, if anything, edging up slightly. Such spending now accounts for 15 percent of GDP9 compared with less than 8 percent in the early 1970s and, although most benefits are subject to means and/or asset testing, the widening of eligibility means that middle and upper income groups probably now receive cash and non-cash benefits totalling around $20 billion a year.10

While the "jumps" in unemployment in the mid 1970s, and again in the early 1980s and the early 1990s, were due to factors not connected with the level of welfare benefits, there has (as noted) also been a large increase in Australia's structural unemployment rate which requires explanation ie why does it now appear that we cannot get unemployment below about 8% without engendering inflationary pressures and why is that rate so much higher than it was in the early 1970s?

While Australia's continued excessive regulation of the labour market undoubtedly constitutes the major part of the explanation, it is also relevant that Australia's social security and welfare arrangements were significantly liberalised in the mid 1970s and that, since then, the real value of most welfare benefits has been increased further and eligibility has been widened.

Possibly the most comprehensive recent Australian study of the trend in unemployment benefits and their possible effects is that undertaken by Barbara Whitlock of the Social Policy Division of the Department of Social Security as a thesis for the Graduate Diploma in Economic History at the ANU. This study was published in January 1994 by the ANU Centre for Economic Policy Research and again by the Department in February 1994.11

The study shows that between 1970-71 and 1974-75 replacement rates for unemployment payments alone more than doubled to nearly 30% of average weekly earnings (AWE) in the case of single persons, and to nearly 50% in the case of couples. Since then the rate has been broadly maintained relative to AWE in the case of couples but has decreased to around 25% in the case of single persons.

However, given that most unemployed are likely to start at the lower end of the wage scale, it is more relevant to examine replacement rates relative to minimum wages. It is relevant that many OECD countries see no need for a minimum wage because the minimum welfare safety net payment provides a 'reservation wage' below which few will accept a job.

The Whitlock study does in fact show that replacement rates related to minimum wages for adults and couples without children not only increased significantly over the twenty years to 1993-94 but ranged up to 90% in that year. For couples with children they ranged between 76 and 86% and for single adults over 21 between 52% and 65%. Part time employment to supplement unemployment benefits was shown in some situations to result in replacement rates beyond 100%, implying that disposable income falls if a full time job is obtained.12

Attachment B showing. trends in both minimum wages and unemployment payments confirms that there has been a downward trend in minimum wages in real terms since the mid 1970s while unemployment payments have maintained their real value. Thus, the disincentive to accept work at the lower end of the wage scale has been steadily increasing.

It is important to recognise that the replacement rates used in the Whitlock study take account of unemployment payments only. If the full range of cash and non-cash social security benefits were taken into account, replacement rates would surely be well over 100% for most people.

However, perhaps partly reflecting the patrons of the author, this study did "not find that increases in the level of unemployment payments encourage more people to become unemployed and take up unemployment payments." It did conclude, though, that changes in payment levels have a reasonably large impact on the length of time people spend drawing unemployment benefits when these people are unemployed for a short time and that, where eligibility tests are stringently administered, unemployment beneficiaries fall.

The effect of taking account of the potential availability of other forms of assistance is shown by the 1996 National Commission of Audit (NCA) report, which included a case study showing that the real value of all cash benefits potentially available to an unemployed family increased by nearly 24% between 1984 and 1996. Benefits included in the case study, which covered parenting allowance, family payment, and rental assistance in addition to the unemployment payment itself, amounted to almost $22,000 pa or $844 per fortnight in 1996.13 Average fortnightly earnings in 1996 were about $1,300

The NCA report also found that the average real value of components of non-cash entitlements for an average of various income units showed an increase of nearly one third between 1981-82 and 1993-94 (from $68 per week to $90 per week). The report further noted, however, that the ABS Household Expenditure Survey for 1993-94 showed that, for households whose principal income was unemployment assistance averaging about $225 per week, non-cash entitlements (which are effectively non-taxable) added an average $150 per week to household income. (These non-cash benefits do not, moreover, purport to be a complete list). That Survey shows that, after taking account of non-cash benefits and direct and indirect taxes, households whose principal source of income was unemployment allowance in 1993-94 had an average total income of $412 per week. With an average of 2.7 persons (including children and aged persons) per household, that indicates an average income per person of just over $150 per week.

The NCA report reached the rather cautious conclusion that this indicates "cause for concern that social security benefits are at least reducing the incentive for job search by those unemployed" and noted that, along with the business cycle, this "may drive average rates of unemployment (and government spending thereon) up over the longer term". It acknowledged that empirical studies of any link between benefits and unemployment rates have produced "mixed results", with the main conclusion being that there appears to be a correlation between higher benefit levels and higher unemployment but that it tends to be lagged. However, the report also pointed out that such studies are subject to a number of difficulties, including that the impact of any given replacement rate varies according to the total social security package and individual "client".14

The NCA report included in the inside back cover a large folded sheet which summaries the main cash benefits that are available and the main eligibility conditions which apply. (Attachment C sets out the provisions applying to unemployment benefits or, as it is now known, Newstart Allowance). It should be noted that most cash and non-cash benefits are available in addition to unemployment assistance and some do not count for purposes of the income test for such assistance. Also, assistance such as the family payment and sole parent pension are available regardless of whether work is being sought. Most benefits listed are subject to income/assets tests but are available even if the applicant is not eligible for unemployment payments. Some (though not unemployment assistance itself) are not taxable.

The ABS Household Expenditure Survey for 1993-94 shows that, for households whose principal source of income is the unemployment allowance, the average amount of cash and (selected) non-cash government benefits was $434 per week per household in that year. Of that amount, unemployment assistance itself constituted only $231 per week or about 53 percent.).15 Once the full extent of these cash and non-cash benefits, and their eligibility conditions, is appreciated it becomes difficult not to conclude that, when considered as a whole, the social security system is not responsible for a significant proportion of Australia's unemployment,

The main benefits relevant to those with potential to work are:

  • Rental assistance---available to low income and unemployed who are renting,. State public housing tenants are eligible for rental rebates only,
  • Family payments---non-taxable means tested payments to families with children. The minimum payment is available to all families with taxable incomes up to $64,000 and an additional amount is payable to low income families.
  • Family tax payment---a new additional non-taxable payment to families with children through tax concessions or cash payments to low income families.
  • Parenting payment---a single income support payment for those with child caring responsibilities (replacing the sole parent pension and the parenting allowance). Subject to income and assets tests.
  • Child care assistance---fee relief and cash rebates are available to low and middle income families using child care services. The operating costs for some services are also subsidised. A potentially major encouragement to join the work force.
  • Disability support pension---requires at least a 20% level of medical impairment and continuing, inability to work or undertake training within the following two years. The very large increase in the number of such pensioners in recent years indicates that those in a position to do so are using it in lieu of unemployment assistance because it is more generous,
  • Medical and public hospital treatment---available to all without charge. Low income earners are exempt from the Medicare levy. No means/asset test. Subsidised pharmaceuticals.
  • Education---AUSTUDY provides income support for students 16 years and over. Subject to income and assets tests. Higher education grants subsidise the cost of tertiary education.

The income test has also been progressively liberalised in recognition of the potential adverse effects of high replacement rates on work incentives. For single people, the "free" area was about 20% higher in 1993 than it was in 1969 and for couples the free area more than doubled in real terms over this time. In addition, a shift in 1995 to independent entitlements for husband and wife has made it easier for a member of the household to have some earnings without losing all benefit entitlements. About 15% of recipients of unemployment payments now admit to working part-time. As a person earning is not counted as "unemployed" for purposes of the ABS unemployment figures, that reduces the official unemployment rate but not the numbers receiving allowances.

The higher the replacement rate the greater the significance assumed by the requirements that applicants have to meet before being eligible for unemployment payments. This assumes particular importance in Australia given the unlimited duration of assistance.16 Although there has been an increase in job search activity requirements and reviews in recent years, and an increase in "downward variations" of unemployment assistance, anecdotal and other evidence suggests that there is still a considerable capacity for unemployed to earn other income that is not declared and to be selective in taking job offers. Some regions which have consistently high unemployment rates are in locations with favourable climatic conditions and Queensland as a whole has had an unemployment rate above the Australian average in every year bar one in the last twenty years.17 It now has the highest rate except for Tasmania.

Moreover, the penalty for exploiting the system is generally small: a Department of Social Security review published in February 1997 identified $54 million of "over-payments" of unemployment benefit in the six months ended December 1996 though in the great majority of cases the only penalty imposed was the repayment of the amount over-paid. (However, from 1997-98 Newstart recipients failing to declare other income will be liable to repay and will have their payments reduced by 18% for 26 weeks). An indication of the possible extent of cheating is provided by a new data matching measure announced in the 1997-98 Budget to check identity and income information provided by people receiving assistance: it is estimated that, out of the additional 50,000 who will be checked, some 11,000 or 22% will have their payments either cancelled or reduced.18

But, while there seems little doubt that Australia's "black economy" is a considerable source of income that is not taxed and not declared for social security eligibility purposes, it may not be as large as Italy's appears to be. Reflecting the considerable interest in the American media in the relative performance of labour markets, in June the New York Times published a story from Sicily about the reasons for the 30 per cent unemployment rate in that region. The story included an interview in which the official in charge of the unemployment office in Paterno stated "If those figures were true, it would be frightening, so you have to look at them with a pair of pincers. There is massive illegal work, and there always will be." The story went on to report a recent survey showing that one out of three jobs in Southern Italy is "black" and it noted that, for employers, ''irregular' jobs are desirable because they mean not having to pay either taxes or benefits. Yet, despite the loss of revenue, out of 100,000 reports of irregular jobs in Italy in 1996, only 5,000 were even examined by the Government. No mention was made of any penalties being imposed.

This is by way of sounding a cautionary note about overstating the extent of the unemployment problem on the basis of the high published figures for unemployment in Europe, and, to some extent, Australia too. Some were startled by the Clemenger Advertising survey published in early August suggesting that over 70 percent of people surveyed nominated dole bludging as their principal concern. This concern was apparently self generated, not promoted from a list.19 In reality, we all know that, behind the published unemployment figures, there is a 'black market' that avoids the regulations and attempts to enforce rules of eligibility for unemployment assistance. One indication of the extent of the Australian 'black' market is provided by the fact that, in 1995-96, cancellations of unemployment assistance amounted to 6% of reviews by Social Security, 'overpayments' to a further 8%, and rate reductions to a further 2% ie something like 15% of the reviews, amounting to over 220,000, were detected as receiving income not declared or of trying to avoid work requirements.

Of course, as noted in the introductory section, this overstatement of unemployment has to be balanced against other influences which lead to an understatement.

The Scope for Action

Action to reduce benefits or tighten eligibility requirements would need to take account of the present relative inflexibility of the labour market and the consequent discouragement for employers to offer jobs, including high minimum wages which make it unprofitable to employ low productivity workers. Any such actions would desirably be combined with substantive deregulation of the labour market, which is the most important measure required to reduce unemployment. Indeed, the interplay between the labour market and social welfare means that action in one area is likely to be less effective without action in a similar direction in the other area.

It is beyond the scope of this paper to examine in detail specific measures to reduce the disincentive to work. Drawing on measures used in other countries, action might include the following:

  • Tighten the work test so that 'unsuitable' jobs have to be accepted after a certain period.
  • Reduce the amount of unemployment assistance by say, 20% after a period of, say, one year.
  • Require a significantly longer wait before receipt of assistance for those who quit their jobs voluntarily (in July 1997 about 121,000 were job leavers out of 377,000 of unemployed persons who had worked full time for two weeks or more in the last two years).
  • Make those who have substantial liquid funds use up more of those funds than presently required.
  • Where practical, make the payment of social security benefits to those of working age conditional on the recipient undertaking job search or, in the case of disabled persons, on undertaking more action than presently required to improve employability.
  • Make all social security benefits subject to income tax.
  • Eliminate the de facto indexation of unemployment payments to changes in male average weekly earnings (as recommended by the NCA).
  • Put the onus on unemployed recipients to show that they are not exploiting the system by, imposing penalties for 'over-payments' of unemployment assistance in addition to repaying over-payments..

Another possible measure to encourage people to work would be to offer a tax credit that would be income tested. The Americans claim that their scheme provides an important encouragement to low income earners to work and some 19 million ( which is about 15% of total employment) benefit from it. The annual cost to the budget is $27 billion, which is equivalent to about 1.7% of total spending. Phasing out begins at around $US 12,000 pa or about $A16,000.

There are two ways of looking at such schemes, which also operate in several other countries apart from the US.

The first is that such a tax credit provides-a social security type income supplement to those who are unable to earn in the market place what is regarded socially as an "adequate" income. Its availability in the US certainly puts something of a different complexion on suggestions that the US system simply leads to the establishment of a group of "working poor" who are left to fend for themselves.

The second interpretation is that this tax credit is a form of wage subsidy which simply allows employers to earn extra profits by paying lower wages than would otherwise be the case. But, while it may be the case that employers are able to pay lower wages, competition in product markets will tend to ensure that "excess" profits are not earned. Moreover, it might equally be argued that some other social security income supplements allow employers to pay lower wages than otherwise. The advantage of this tax credit approach is that it is conditional on being employed. The Australian family tax payment scheme is similar to the US tax credit scheme except that is not employment conditional. Perhaps it should be made so.


The publication of OECD research showing that differences in "generosity" of unemployment assistance are a significant explanatory factor in countries having higher unemployment levels than others, and the increasing recognition that many other welfare benefits also have disincentive effects on work, highlights the need to undertake a comprehensive review of the effects on unemployment of the "generosity" of the whole range of relevant cash and non-cash benefits in Australia. There seems little doubt that such generosity accounts for a not insignificant proportion of Australia's existing unemployment rate. Publication of estimates of the adverse effects of present arrangements would at least reduce pressure on the Government to adopt other interventionist policies.

There is a strong case for adopting a range of measures to "encourage" more of the unemployed to work, and thus help reduce unemployment. Such action would preferably be taken in conjunction with substantive deregulation of the labour market, but should not await that.

More generally, while Australia provides less middle class welfare than Europe, the case for reducing the welfare liberality line and welfare dependency here is also a strong one. If combined with lower taxation and deregulatory reforms of the labour market, additional low income jobs would be created and that would reduce the number of two income households in poverty. Overall, such a result would surely be beneficial both economically and socially.

Attachment A

Attachment B

Attachment B

Attachment C

30/6/96 TO 30/6/96

Basic Conditions of Eligibility

  • Must be unemployed, capable of undertaking, available for and actively seeking work.
  • Over 18 years but under age pension age and registered as unemployed for 12 months or less., OR
  • May do training and voluntary work with CES approval.
  • Willing to enter into a Job Search Activity Agreement if required.
  • JSA recipients incapacitated for work remain on JSA, subject to medical certificates, if the incapacity for work is less than 13 weeks.

Residential Qualifications

  • Must be an Australian resident and be present in Australia during period of payment.
  • Available to newly arrived migrants after 26 weeks.

Basic Rates and Methods of Payment $ per fortnight

  • Couple, both over 21 or over 18 with children (each) 285.80
  • Couple, no children, one partner over 21:
    • for partner aged 21 & over 285.80
    • for partner aged 18-20 256.70
    • for partner under 18 232.10

  • Single 18-20 at home 169.10
  • Single 18-20 away from home, no children 256.70
  • Single 21 and over, no children 316.70
  • Single over 18 with children 342.60
  • Single aged 60 and over after 6 months 342.60
  • Advances of allowance of up to $1000 may be available
  • Allowees over 60 receive Pharmaceutical Allowance after 6 months
  • Pharmaceutical Allowance is also paid during periods of incapacity

Rent Assistance

  • Ranges between a maximum of $70 and $97 pf if persons first pay a minimum amount themselves ranging between $71 pf and $137 pf
  • Not available to single persons under 25, without dependants, who live with parents or guardians.

Income Test

  • Income up to $60 pf is "free", between $60 pf and $140 pf the allowance is reduced by 50 cents in the dollar. For income over $140 pf, the allowance reduces by 70 cents in the dollar.
  • Earnings Credit Scheme reduces the effect earnings have on payments. Maximum is $100 pf & $500 in total.

Assets Test

  • Assets up to $167, 500 (house owner) or $251,500 (non-home owner) are "free". Part allowance is payable with assets up to $360,000 (house) or $444,000 (non-house).
  • Payment may be deferred when liquid assets exceed $5000 (single) or $ 10,000 (couple).

Taxation Treatment

  • Allowance is taxable but rent assistance is not. Neither are family payments.


1 Employment Outlook July 1996 OECD p 25.

2 Ibid p 28.

3 This section of the paper draws on that introduction. Further details are available in Dr Himmelfarb's The Idea of Poverty, England in the Early Industrial Age Jan 1985 Vintage Books. It is interesting that, although critics of the poor law reforms, social historians G D H Cole (well known for his left wing views) and Raymond Postgate acknowledged in The Common People 1745-1945 that "the advocates of the new Poor Law were broadly right, as far as their case referred to areas in which wages had been regularly subsidised out of the rates. In these areas, after a brief interval of acute misery, wages did rise; and the labourers in the countryside were on the whole soon rather better off than before" (p 278).

4 The 1993-94 Household Expenditure Survey shows that over 4.1 million people relied on government cash benefits as the principal source of their income, or about 24% of the total population. Of those, over 900,000 persons indicated unemployment and sickness allowance as the principal source of their income.

5 Employment Outlook July 1996 p 28 and Box 1 on p 30.

6 OECD Employment Outlook July 1996, Chapter 2 (pp 28 and 36 for above quotes).

7 Ibid pp 28-33.

8 Scarpetta, S., Assessing the Role of Labour Market Policies and Institutional Settings on Unemployment: A Cross country Study, OECD Economic Studies No. 26, 1996/I. The calculations of the effects of increased generosity of UBs imply an average elasticity of unemployment to the UB index of close to 0.5.

9 Broadly defined to include education, health, social security and housing. Not all such spending is 'welfare' in the sense of assisting low income groups. But the income supplements provided, and the tax required to pay for them, have implications for the incentive to work regardless of the income groups to whom the benefits go.

10 This estimate is based on the ABS document "The Effects of Government Benefits and Taxes on Household Income" (Catalogue No. 6537.0) May 1996 and estimates derived from that document by the Chief Economist, ANZ Banking Group, Mr Saul Eslake in his paper "The 1990s Economic Cycle And The Challenges Facing The New Government" (10 July 1996) to the effect that the top 40% of households received from all governments some $16.7 billion in 1993-94 in cash and non-cash benefits from education (other than school education), health, housing and other social security benefits. Projecting that estimate forward at the same rate of growth as Commonwealth budgetary outlays on "social welfare" (as defined) implies that such expenditure would be about $20.5 billion in 1997-98.

11 Whitlock, Barbara (1994) "Does the Social Security Income Support System Remove the Incentive to Work?" ANU CEPR Discussion Paper No. 303.

12 Ibid pp 17-22.

13 National Commission of Audit (NCA) June 1996 p 165.

14 The references in this paragraph are from pp 166-167 of the NCA report.

15 Household Expenditure Survey Australia 1993-94, The Effects of Government Benefits and Taxes on Household Income ABS Cat, No 6537.0 27 May 1996. Table 8.

16 However, although most other countries impose a time limit on such assistance, once the limit is reached some lesser form of assistance continues. See, OECD outlook Ibid p 34.

17 Contrary to popular belief, Queensland has a lower than average proportion of its population in the 65 and over age group and a slightly higher than average proportion in the working age group of 15-64. This means a faster than average growth in the latter age group which is in turn reflected in a higher than average participation rate that would include both genuine and non-genuine job seekers.

18 Budget Measures 1997-98 Budget Paper No. 2 pp 129-130.

19 The Silent Majority III A Clemenger Report August 1997.