RBA Annual Conference – 1998 Solutions to Unemployment and Avoiding the ‘Diabolical Trade-off’: A Discussion Peter Dawkins

1. Introduction

The papers for this conference canvass a wide range of issues that are relevant to understanding the causes of high unemployment in Australia. They also provide considerable insights into a range of possible policy responses. This paper draws on many of the points made in these papers. It does not, of course, confine itself to these papers alone.

The aim of the paper is to help in the search for the appropriate policy responses to unemployment in Australia. In so doing, it must first address the question of what has caused the persistence of high unemployment in the past quarter of the century, following thirty years of very low unemployment (Figure 1). In the subsequent discussion of solutions, a distinction is highlighted between policies aimed at increasing employment in general, and policies that are more focused on the bottom end of the labour market. It is concluded that if we wish to make a substantial reduction in unemployment on a long-term basis, it is appropriate to adopt policies of both kinds. Further it is argued that they would be complementary.

Figure 1: Unemployment this Century

It is noted that there are uncertainties about the likely magnitude of the effects of the kind of policy package proposed, and also important political issues about who would be winners and losers and whether such a policy package is politically feasible. It is contended, however, that the costs of not embarking upon policies of this kind are likely to be such as to make such a package potentially politically feasible.

Another theme of the paper is whether we can avoid the ‘diabolical trade-off’. The papers about the United States and Europe by Katz (1998) and Jackman (1998) respectively, tend to lead one to the conclusion that the first-best solution to the unemployment problems of many OECD countries such as Australia is, in Jackman's words, one that ‘favours policies of flexibility, deregulation and generally the reduction of government intervention in the labour market’.

The downside of such policies is that they tend to lead to a wider distribution of earnings, although as the paper by Harding and Richardson (1998) for this conference points out, a redistribution even from low-wage earners to unemployed people would make the overall distribution of income more equal. Nonetheless, Harding and Richardson argue that it does not follow that lower incomes for low-wage earners would necessarily be egalitarian. In their words: ‘a tax on low-wage workers is a much less equitable way to finance this than a tax on all wage and salary earners’.

As Katz points out, however, the story might be a little bit more complicated than the existence of a ‘diabolical trade-off’. Further, he argues that in the US, policies should be adopted to improve the position of the disadvantaged in the labour market without losing the benefits of labour market flexibility. In this paper I argue that what we need in Australia is to increase the flexibility of the labour market in order to reduce unemployment, while adopting such policies for the disadvantaged that avoid the creation of a severe working poor problem and the potential detrimental effects on equity of freer labour markets.

2. The Causes of High Unemployment

2.1 Economic growth

The Green Paper on unemployment (Committee on Employment Opportunities 1993), and the resultant Working Nation policy (Keating 1994), strongly emphasised the importance of the rate of economic growth as a determinant of the rate of unemployment, and its main conclusion was that the rate of economic growth needed to be increased to substantially reduce unemployment. Its main prescription as a result (alongside a considerable expansion of labour market programs), was to maintain or increase the rate of microeconomic reform, in order to raise the rate of economic growth.

Since that time, the unemployment debate appears to have reflected a growing view that while raising the rate of economic growth is a good aim, we should not place too much emphasis on this as a solution to the unemployment problem (see for example Borland 1997).

One paper in this conference, however, that strongly asserts the importance of growth is Dungey and Pitchford (1998). They point out that:

Historically, periods of high activity and growth such as in World War 2 and during 1945–75 were accompanied by low unemployment. It would be surprising if growth were not the cause of such favourable employment experience.

They also emphasise the benefit to be had from less drastic recessions. Nonetheless, they point out that higher growth is not a panacea because it brings with it the potential for rising inflation. They suggest that we should aim to achieve as high a rate of growth as is possible without causing rising inflation. They also provide an estimate of the ‘steady inflation rate of growth’ (SIRG) for Australia, since the floating of the Australian dollar, of 4.37 per cent per annum. They suggest that, as a result, it may be possible to get unemployment down to 5 or 6 per cent over the next four years, without causing rising inflation.

How optimistic their scenarios are, depend partly on the assumed rate of wage inflation. If real wages increase at half the rate of economic growth, we get about a one per cent lower unemployment outcome in four years than if they increase at the same rate as output growth. This highlights how both economic growth and real wages are influential in determining the rate of unemployment.

2.2 Real wages

Probably the greatest degree of consensus exists about the importance of real wages in determining the level of employment and unemployment. The two wages explosions of the mid 1970s and early 1980s were very significant episodes in the long-run growth of unemployment in Australia and we have not yet recovered from them (Figures 2 and 3).

Figure 2: Real Average Weekly Earnings – Trend and Actual
Figure 2: Unemployment Rate – Trend and Actual

In their paper for this conference, Debelle and Vickery (1998) estimate an aggregate model of the labour market, and an associated NAIRU, in which a slower growth of real wages of 2 per cent for a year would lead to a permanent reduction in the unemployment rate of around 1 percentage point. They estimate that the NAIRU or natural rate of unemployment is about 7 per cent, and that to get it down requires a structural change in the labour market to produce a significant decline in the equilibrium real wage.

2.3 A collapse in the demand for unskilled workers

So far we have been thinking at a macroeconomic level. Turning to a more microeconomic perspective, a major issue is the extent to which the growth in unemployment has been concentrated in certain parts of the labour market.

In an article entitled ‘The Collapse in Demand for the Unskilled and Unemployed Across the OECD’, Nickell and Bell (1995) highlighted the fall in the demand for unskilled labour across OECD countries over the previous twenty years and suggested that the demand had been falling faster since the beginning of the 1980s. Evidence was presented for Australia that between the mid 1980s and the early 1990s, the ratio of the unemployment rate of the high skilled to the rate of unemployment of the low skilled (based on educational attainment) had reduced significantly. This was a typical, though not universal, experience for an OECD country.

The paper by Borland and Kennedy (1998) for this conference includes data on unemployment rates by educational attainment, which also suggest that degree and diploma holders have fared much better in the market than either those who leave education after Year 12 or before (reproduced here as Figure 4). The former group had an increase in their unemployment rate from about 4 per cent to about 5 per cent between 1980 and 1994, while the latter had an increase from about 8 per cent to about 14 per cent. Of particular interest in Borland and Kennedy's figure, is that this has been particularly prevalent in the 1990s, although we are still to find out how much of this is cyclical and how much structural.

Figure 4: Rate of Unemployment by Educational Attainment

Nickell and Bell (1995) incidentally went on to compute that the fall in the share of employment of the high skilled in OECD countries between the 1970s and the 1980s, accounts for about 20 per cent of the rise in aggregate unemployment. It seems likely that this order of magnitude also applies to Australia.

In the paper by Katz (1998) for this conference, there is a discussion of this shift in demand towards higher-skilled labour. He mentions that Krugman's diagnosis is that skill-biased technological change has worked against low-skilled workers, but the way that it affects them depends upon the flexibility of the labour market (Krugman 1994). In the flexible US labour market it shows up as greater wage inequality. In Europe it shows up in higher unemployment.

Katz argues that ‘labour market adjustments to changes in the relative demand for skill also depends upon education and training policies, macroeconomic policies and experiences, and wage-setting institutions in a manner possibly more complicated than suggested by a simple diabolical trade-off between inequality and unemployment’ (Freeman and Katz 1994, 1995). According to Katz (1998), however, Krugman's intepretation ‘probably contains substantial grains of truth’.

2.4 Institutional arrangements

2.4.1 Introduction

This then takes us on to institutional arrangements. As Katz (1998) points out, this needs to be conceived of far more broadly than wage-setting institutions. In particular, he argues that unemployment benefit arrangements are very important.

2.4.2 Unemployment benefits

In a simple theoretical framework used to illustrate various hypotheses, Katz (1998) points out that an increase in unemployment benefits increases the reservation wage relative to productivity and shifts up the ‘wage-setting curve’. Increased transfers are also likely to require increased taxes on the employed and thereby increase the tax wedge and produce a downward shift in the ‘demand wage relation’. Both these effects are predicted, in this framework, to lead to higher equilibrium unemployment. Katz refers to empirical evidence which tends to support this effect including research by Layard, Nickell and Jackman (1991). The paper by Jackman (1998) for this conference shows that studies of the European experience suggest that increases in unemployment benefits have had a small upward effect on the unemployment rate. Australian studies vary in their estimates but it seems likely that the significant rise in the replacement ratio in the mid 1970s had such an effect, perhaps of the order of one percentage point. Pissarides (1991) attributes a larger effect to this increase in the replacement ratio.

Katz (1998) points out, however, that there is a puzzle that needs to be resolved. That is, unemployment benefits in the US were less generous than in Europe and Australia in the 1950s, 1960s and 1970s, when unemployment was actually higher in the US than in Europe. One possible explanation he puts forward is that ‘the generosity of the treatment of the unemployed affects the dynamic response to adverse macroeconomic shocks’. After various shocks to OECD economies in the 1970s and 1980s there was an increase in unemployment in most OECD countries that became sustained. Katz (1998) states: ‘The periods of high unemployment were of shorter duration in the United States, and the lower generosity of unemployment benefits and lower insider bargaining power, meant that the unemployed continued to put strong pressure on wage-setting in the United States’. Another possible explanation that he discusses is that higher unemployment benefits may have become more costly in the face of the decline in the relative demand for less-skilled workers.

2.4.3 The wage-setting system and labour market regulation

Discussions of the role of the wage-setting system are controversial and the associated disputes hard to settle empirically. Jackman (1998) points out that there have been two basic approaches to wage-setting and labour market regulation in Europe in the past twenty years. In the UK the thrust has been to deregulate the labour market, reduce union power and emphasise market forces. In most continental European countries a much more interventionist model has been adopted, to a greater or lesser degree. In the corporatist model of Austria and Sweden for example, this includes centralised co-ordination of wage determination by the social partners.

A view that has achieved considerable credibility is that of Calmfors and Driffill (1988). They argued that the best outcomes for real wages, employment and unemployment tend to be either in the most centralised systems, which enable co-ordination to achieve wage restraint, or in the most decentralised systems such as the United States where real wage growth has been very low by international standards. This view receives some support in the paper for this conference by Wooden and Sloan (1998), who point out that the experiences of Australia, New Zealand and the UK are consistent with this framework.

Calmfors and Driffill's (1988) classification had Australia in the middle of their scale of centralised versus decentralised systems, or on ‘the hump’, which they argue, is the worse place to be.

This framework provides an interesting way to think about the Australian experience over the past thirty years. Attempts to decentralise wage-setting in 1974–75 and 1980–81 produced wage inflation and unemployment, arguably because the wage-setting system still had strong unions and a strong principle of comparative wage justice, which resulted in leap-frogging wage claims and associated wage explosions. It was not a true decentralisation of wage-setting.

On the other hand, in the mid 1980s, the then Labor Government chose to adopt a wages policy as part of the Accord with the Australian Council of Trade Unions, which represented a move towards the corporatist approach of countries such as Austria and Sweden. In the early stages of the Accord, this was a very centralised wages policy with common wage increases across the workforce. This was loosened in the late 1980s and early 1990s.

Reviews of the success of the Accord suggest that especially in the early stages it was very successful in restraining the growth of real wages and as a result, led to stronger employment growth and lower unemployment than would otherwise have occurred. In the context of the Calmfors and Driffill framework, this was a move away from the top of the ‘hump’ towards the more centralised system, which produced superior macroeconomic outcomes.

This move towards centralisation, however, was not sustained. First it did not prove possible to ‘lock in’ the idea of corporatism. This was partly because employer groups never consented to the process. Secondly, in the latter stages of the Accord, consistent with other policies that had been introduced by the Labor Government – which had moved towards deregulation of markets – the Accord itself started to embody much more decentralisation in wage-setting. The Labor Government started to encourage enterprise bargaining, a process that has been further accelerated by the subsequent Coalition Government. The election of the Coalition Government, of course, represented the official end of the Accord. Thus there was a reversal of the move away from the ‘hump’ towards centralisation in the Calmfors and Driffill model. It seems that now we are moving away from the hump again, this time in the decentralising direction.

It should be added that the ACTU have indicated that they do not favour a return to the corporatist approach of the Accord and there would have to be significant doubt that moving back in the corporatist direction is a feasible option in Australia. Following the Calmfors and Driffill framework, this suggests perhaps that we should keep moving in the decentralising direction, in order to lock in the benefits that can be obtained from being towards that end of the distribution.

3. Solutions

3.1 The role of economic growth

There is some evidence that the long-run rate of economic growth may have risen in recent times because of an increase in total-factor productivity growth (Industry Commission 1997; Melbourne Institute 1997a). Dungey and Pitchford's estimate of a 4.37 per cent stable inflation rate of growth is also suggestive that there has been an increase in the long-run rate of economic growth in recent times, although it would be interesting to see how sensitive this estimate is to the choice of time period.

This does beg the question of why employment growth has not been stronger and why unemployment has not come down more. A related question is why has the current business cycle recovery not brought with it much greater increases in employment and reductions in unemployment?

There are a number of possible answers to these questions, all of which could be contributory explanations. One is that we have not fully exploited the potential non-inflationary growth rate yet. We can perhaps run the economy faster on the demand side.

The second is that that the rate of productivity growth has been high and the rate of economic growth required to reduce unemployment has become commensurately higher. Part of the possible explanation for rising productivity growth is a period of substantial downsizing of Australian companies, which might, in principle, allow for the shed labour to be productively utilised elsewhere, though this could take a significant amount of time to happen.

A third explanation is that in a period in which inflation has reduced considerably, it has taken some time for wage expectations to adjust downward, and thus wage increases in the mid 1990s have been higher than would normally be associated with such a low level of inflation. Thus real wages have been growing faster in the current recovery phase of the business cycle than would normally be the case.

A fourth explanation could be that enterprise bargaining has been part of the cause of productivity increase but most of the benefit has gone into higher real wages which has tended to confine the benefits to the ‘insiders’.

Having said that, unemployment is now around 8 per cent. It was as high as 11 per cent in the last recession and if the economy continues to grow above trend for another three or four years we could see unemployment come down towards 5 per cent, which looks possible in the context of the Dungey and Pitchford model. The chance of economic growth producing substantial employment gains will be further enhanced by the fact that consumers' inflationary expectations as monitored by the Melbourne Institute (1997b, 1998a) have declined substantially over the past eighteen months and wages growth also appears to be moderating (Melbourne Institute 1998b).

We noted above that there is some evidence that the long-run rate of economic growth has become enhanced in recent times. This could be partly due to the microeconomic reform that has occurred in the past fifteen years and partly due to enhanced technical progress. As Dungey and Pitchford (1998) suggest, it seems very likely that faster economic growth was a major cause of low unemployment in the 1950s and 1960s and we should be able to derive some such benefits from faster growth in the 1990s and 2000s. The evidence both of Dungey and Pitchford (1998) that there is scope to run the economy a little faster, and of Debelle and Vickery (1998) that we are still above the natural rate of unemployment, suggest that we can get unemployment down somewhat through faster growth in demand. Further continuing to implement microeconomic reforms aimed at productivity growth could be expected to be helpful, in the long run, to employment prospects. Policies aimed at enhancing technical progress especially where there are associated externalities are also worth focusing on (Rogers 1997).

However, evidence suggests that what happens to real wages is a critical issue in determining whether faster economic growth delivers substantially lower unemployment on a long-term basis. Indeed it may even be possible to substantially reduce unemployment without faster economic growth, if equilibrium real wages are reduced.

3.2 Restraining the general level of real wages

3.2.1 Introduction

Real wage restraint in the Accord produced positive employment benefits and reduced the level of unemployment. In the past few years of enterprise bargaining and very low inflation, real wages have moved ahead again. However, the growth in real wages appears to be declining now that low inflationary expectations are getting ‘locked in’ although they are still rising significantly.

There is widespread agreement amongst economists that holding down real wages would offer great hope for significantly lower unemployment. Debelle and Vickery (1998) have a 1 per cent decline in the equilibrium rate of unemployment for a decline in the growth of real wages for one year of 2 per cent.

This begs the question, how can we get a decline in the growth of real wages by 2 per cent below what it would otherwise be?

3.2.2 Wages policy – a wage-tax trade-off

One solution, as adopted under the Accord would be a wages policy. Under the Accord the submissions of the Commonwealth and the ACTU to the Industrial Relations Commission proposed modest wage increases that actually entailed for some time a reduction in real wages. It is implausible that the ACTU will argue along these lines with a Coalition Government in power and somewhat unlikely even with a Labor Government.

However, it would be open to the Commonwealth to mount this argument. The main problem they face under the current wage-setting system is that many fewer employees now depend essentially upon award wages handed down by industrial tribunals. Whereas it was about 90 per cent of employees at the time of the Accord, it is now thought to be between 30 and 40 per cent.[1] It is likely, however, that many wages determined by enterprise bargains may also be influenced by what happens to award wages.

If there were no adjustments to the ‘wages safety net’, that is, wages in awards were frozen for, say, four years, this could be expected to lead to a reduction in their real value of the order of 10 to 12 per cent depending upon the rate of price inflation. If about 25 per cent of employees' wages were held constant as a result and say a further 40 per cent increased slower than they otherwise would have, then this could have the effect of holding back the average growth in real wages by the order of 3 to 4 per cent, bearing in mind that those employees who depend on award wages are generally low in the distribution of earnings.

This policy would need to be complemented by various others discussed later but as a guide, if we apply Debelle and Vickery's estimates of the effect of lower real wage growth this could be expected to reduce the equilibrium level of unemployment by about 1.5 to 2 percentage points, bringing it down to about 5 to 5.5 per cent.

One of the main side effects of such a policy would be to cause a widening in the distribution of earnings as higher-paid employees on enterprise bargains would be less affected than lower-paid employees.

It seems unlikely therefore that the Industrial Relations Commission would agree to a freezing of award wages, unless low-paid workers could be compensated in some other way. Here, a possible solution is for the Commonwealth to compensate low-paid workers through the tax-transfer system, by awarding them tax credits. One advantage of such a policy is that low-paid workers who are in families with children could be awarded higher tax credits than those without children, thus incorporating the Australian tradition embodied in the famous Harvester Judgement that we should be concerned about the needs of low-paid workers, which are affected by their family circumstances. Whereas the traditional view of the average worker as a married man with three children, might have applied at the time of the Harvester Judgement, the workforce is far more heterogeneous in the 1990s.

A second advantage could be that a tax credit may turn out to be more valuable to low-paid workers than a wage increase. This would depend upon the way it was implemented. The problem at present, is that for many low-wage earners, especially those with children, an increase in their wage has little effect on their net income. First they pay tax on the extra income. Second, they lose some of their social security benefits because of the high taper rates that apply to the associated means tests.

3.2.3 The wage-setting and industrial relations system

In Section 2.4.3 it was noted that Calmfors and Driffill's analysis of the effect of the wage-setting system suggests that there are two broad strategies to achieve lower real wages and therefore greater employment and lower unemployment. One is for centralised co-ordination of wages so that powerful unions agree to wage moderation so that everyone can benefit (thus internalising an externality). The other is to have highly decentralised wage-setting, so that any individual wage bargain that seeks to achieve higher wage increases than the norm does not get passed on through the principle of comparative wage justice with the danger of pricing the associated labour out of the market.

We noted that while Australia experimented with moving to a highly centralised and co-ordinated wage-setting system of the corporatist variety under the Accord, this does not appear to be an option now. In the latter stages of the last Labor Government, we started moving towards a more decentralised wage-setting system that the current Coalition Government is seeking to lock in through further industrial relations reform. This reform further emphasises decentralised wage-setting and reduces the power of the unions to have a general effect on wage growth, through such things as disallowing secondary boycotts etc.

If Australia is to capture the benefits that can be attained from fully decentralised wage-setting, it would appear to be the best approach to continue the kind of industrial relations reforms that have set Australia in this direction.

3.2.4 The unemployment benefit system

We noted in Section 2.4.2 that a more generous unemployment benefit system is likely to raise the natural level of unemployment and that particularly the rise in unemployment benefits in the mid 1970s probably had this effect. It was also noted that the duration of unemployment benefits is an issue. Countries like the US where unemployment benefits cut out after a while, probably have an advantage in terms of keeping the rate of unemployment, especially long-term unemployment, down. The problem, however, that we face is that cutting unemployment benefits or reducing their duration would have significant negative effects, at least in the short term, on many people towards the bottom of the income distribution.

It should be added that this issue needs to be thought of in the context of the social security system in general and the interface of that system with the tax system. The combination of means-tested social security benefits such as unemployment benefits and family payments, along with the effect of income taxation, mean that for many individuals, and especially families, the monetary benefits from taking a job or moving from part-time to full-time work are often very low, and especially when you add in the cost of work-related expenses (such as transport, child care, work clothes etc.) may even be negative (for a lengthier discussion of this, see Dawkins et al. 1997).

3.2.5 Negative income tax

As a radical solution to this problem, Dawkins and Freebairn (1997) recommended part or all of social security benefits be replaced by a tax-credit system, which is a version of what is sometimes called ‘negative income taxation’. A full ‘basic income flat-tax system’ is the most radical version of this and would reduce effective marginal tax rates considerably for low-income families and increase the incentive to work. The main problem with this radical proposal, is that if it were introduced in one go, it would probably need a flat tax rate in excess of 50 per cent – if it was to be revenue neutral in its effect – which would become the effective marginal tax rate for everyone (Dawkins et al. 1997). This would reduce the effective marginal tax rate for many low and middle income families but increase it for those on high incomes. In time, it is likely that the behavioural benefits from such a system would increase the tax base and allow a reduction in the tax rates. It seems likely that such a radical reform, with an apparently high tax rate and significant numbers of winners and losers may not be politically feasible in the short term. It appears more likely that we could move in this direction gradually, in such a way that when it is finally achieved, it could be achieved at a significantly lower tax rate (Garnaut 1998).

As an alternative short-term measure, in the same spirit, Dawkins and Freebairn (1997) recommend the idea of earned-income tax credits, borrowed from the United States.[2] This is a policy which the paper by Katz (1998) for this conference also endorses as having been a good move in the US. Earned-income tax credits are a kind of partial negative income tax system aimed solely at individuals or families with low incomes because they are on low wages. This idea was recommended earlier in this paper as a complement to freezing award wages, so that those on low wages, especially families with children, were not detrimentally affected, and many could become better off. This would contribute to avoiding the diabolical trade-off. While we may need to accept a wider distribution of earnings and reforms to the unemployment benefit system that increases the incentive to work so that it improves the wage-setting curve in the framework outlined by Katz (1998), we may be able to do it in a way that does not have overall adverse effects on equity.

3.2.6 Is there a case for a jobs levy?

My predecessor as Director of the Melbourne Institute, Richard Blandy, proposed in 1993, the idea of a jobs levy, which was a variant on a proposal that had been floated by the ACTU at the time that Working Nation was under development.

Blandy's idea (1998) was that all employees should pay a levy on their wages, the revenue from which would be used to enable employers to take on new employees. Indeed the interesting twist in this proposal was that firms could reclaim the levy paid by their employees, if they were creating net new jobs in their firms.

It is an interesting idea in that it seeks to achieve consent for the tax by making the benefits of paying it more visible to the taxpayers who actually observe the new jobs being created in their enterprise. The way that it would be making new jobs possible, of course, is effectively by reducing real wages to firms which increase their employment.

One argument against such a proposal is that many of the new jobs that it would subsidise would have been created anyway. This is a conventional argument against wage subsidies. Another argument against it is that while it effectively reduces real wages in ‘upsizing firms’ it does not have the effect of reducing real wages in firms that are ‘downsizing’. This might be defended as an attempt to encourage upsizing rather than downsizing, but it is also arguable that it is downsizing firms that have the greatest need to be able to reduce real wages as a way of avoiding employment reductions. A third argument is that it would create pressure for larger wage increases, which would have a counteracting effect, and that the proposal could even be counterproductive as a result.

Blandy, however, puts up a spirited defence of the proposal. I choose to mention it because it is an imaginative idea which seeks to address a central issue of how a reduction in the growth in real wages can be achieved. This is not an easy task.

If the proposal was varied so that employers received wage subsidies if they engaged long-term unemployed persons, whether they were upsizing or downsizing, it could be argued that the problem of deadweight loss would be counteracted by the equity benefit of increasing the employment of long-term unemployed persons. This proposal would, in effect, represent an extension of existing labour programs paid for by a tax on wages.

At present, I am not convinced about the proposal. The major thrust of the approach that I am proposing is one which seeks to make the labour market work more efficiently, reduce wage inflation etc. Bearing in mind the Calmfors and Driffill analysis and the argument for a greater reliance on market forces, as proposed in the paper by Jackman (1998) in this conference, this proposal does not seem to fit naturally into such an agenda. It could distort the operation of labour markets and might add to the sclerosis problem that Jackman discusses. I do believe, however, that it is worthy of discussion.

3.3 Focusing on the low skilled

We noted in Section 2.3 that there appears to have been a collapse in the demand for low-skilled workers. This suggests that while we should consider policies aimed at reducing the general level of unemployment, there is also a special case for policies targeted at the low skilled who have the highest rate of unemployment and one that has been growing faster than for other employees.

3.3.1 Education and training – long-term policies

An important long-term issue here is whether it may be possible over time to substantially reduce the proportion of the labour force that can be labelled as unskilled. The first place to look here, would be at improving the whole system of education and training in Australia. I do not propose, in this paper, to provide a blueprint for the future of the education and training system in Australia, but it is clearly worthy of attention.

3.3.2 Labour market programs

Martin (1998) has presented an excellent paper on labour market programs to this conference, reviewing OECD experience in this area.

Given the high incidence of unemployment among unskilled workers, it seems inevitable that active labour market polices will remain very important for some time to come. One of the criticisms of Australia's approach to active labour market policies is that they tend to be introduced or expanded as a short-term political expedient and chopped and changed around without much ongoing systematic analysis of strategy.

The case to be more systematic and improve our evaluation of these policies is very strong, and Martin's proposals look like a very good basis for such a systematic approach. He has seven proposals of strong relevance to the Australian case which I repeat here:

  1. integrate the referral to active programs as closely as possible with benefit and placement work;
  2. ‘profile’ new benefit claimants to identify those at risk of becoming long-term unemployed; provide the latter immediately with counselling and job-search assistance;
  3. make passive income support as ‘active’ as possible by using instruments like re-employment bonuses, in-work benefits etc.;
  4. enforce availability to work and job-search tests;
  5. make continued receipt of income support conditional on accepting to participate in active programs after a certain minimum duration of unemployment spell;
  6. ensure that participants in training and public sector employment programs continue to be available for work and actively seek jobs; and
  7. explore ways of making the public employment service more effective by giving greater play to the role of market signals.

Some of these ideas are already embodied to a greater or lesser extent in Australian policies. For example, Australia has recently embarked on a major new initiative with respect to (vii) and we will watch with interest how the reform of job-placement services works. Also we do have some in-work benefits. The idea of tax credits proposed in Section 3.2.2 above would extend this further. The idea of re-employment bonuses as a way of improving the incentive structure of the unemployment benefit system is well worth examining. What is in some doubt is that the policies that have been adopted in Australia that are in keeping with Martin's proposals, have been conceived of as an integrated package of employment/welfare policies, with a central focus of moving people from welfare to work. This is what is needed.

3.3.3 Award wages

The proposal put forward in Section 3.2.2 to freeze award wages was made in the context of reducing the growth in average real wages.

We noted, however, that this would have a greater impact on the real wages of low-skilled and relatively low-paid workers. It could therefore be expected to create more employment opportunities for such workers than for higher-skilled workers.

The paper by Freebairn (1998) for this conference discusses the evidence on the elasticity of demand for workers of different categories. His main point is that there is a paucity of evidence about this and the need for more research in this area.

However, a few pertinent points can be made both from a theoretical perspective and from the limited empirical evidence that does exist.

First, in an extreme case where low-skilled and high-skilled workers are perfect complements, a reduction in the real wage of low-skilled workers would have an equal effect on the demand for high-skilled workers as on the demand for the low skilled. However, as we move to higher values for the elasticity of substitution of workers of different skill categories, the more such a change in the real wage tends to favour low-skilled workers.

Second, evidence presented by Hamermesh (1993) suggests that low-skilled workers have a higher own wage elasticity of demand than high-skilled workers. The determinants of the own wage elasticity are the price elasticity of demand for low-skilled intensive products, the elasticity of substitution of low-skilled for high-skilled employees, the elasticity of substitution of low-skilled employees for capital, and the share of the low-skilled workforce in labour costs. It seems likely that a major reason why low-skilled workers tend to have a higher elasticity of demand than high-skilled labour is that they are more easily substituted for by capital than high-skilled workers who tend to be complementary with capital.

Although the own wage elasticity of demand for low-skilled workers is generally found to be higher than for high-skilled workers, in a number of studies in various countries, the value of the elasticity in the limited number of studies reported by Hamermesh is quite variable. Some estimates suggest quite high elasticities for less-skilled labour and for younger workers (e.g. Grant 1979; King 1980; Lewis 1985) whereas some are substantially lower, but still normally higher than for more-skilled workers (e.g. Bresson, Kramarz and Sevestre 1992; Okunade 1991). The lower ones are either much the same or lower than the typical values estimated for the aggregate elasticity of labour demand. The two possible reconciliations of such an apparent contradiction could be as follows. First, the studies are often of particular industries or firms that may have lower labour demand elasticities than firms or industries in general. Secondly, it could be that the elasticity of substitution between low-skilled and high-skilled employees may be low enough that a reduction in the wage of one group, not only increases the demand for its labour but also the demand for the other group as well. That is, there are negative cross-price elasticities.

Thus the freeze on award wages could increase the employment of all types of workers but would be expected to have a disproportionately positive effect on the employment of low-skilled workers. Further, given the higher estimated elasticities of demand for low-skilled workers, it seems likely that for any given reduction in the average real wage, it will have a greater effect on employment the more it relates to low-skilled workers.

One argument that is sometimes raised against the idea of restraining the wages of low-skilled workers, is that recent studies, especially by Card and Krueger (1994, 1995) in the US, have found that increases in minimum wages do not have negative employment effects. Freebairn (1998) discusses this evidence and points to various reviews on the subject. In common with a number of other critics he expresses doubt that the Card and Krueger evidence does provide a good natural experiment of the effect of minimum wages, particularly emphasising a point made by Hamermesh (1995) that much of the adjustment to the change may have been made before the change in minimum wages, as it had been anticipated a long time in advance.

There is an emerging consensus anyway, that the Card and Krueger evidence is consistent with there being a small negative effect of raising minimum wages on the employment of workers on these wages (see for example Seltzer 1997).

Further, whether you doubt the Card and Krueger evidence or not, its relevance to assessing the effect of freezing all award wages in Australia is not clear. First, the minimum wage is significantly higher in Australia than in the US. Secondly, we are talking here not only about restraining the minimum wage but the wages of all people reliant on awards, amounting to over 30 per cent of employees. In this context, the evidence that labour demand curves generally slope downwards and that the elasticity of demand for low-skilled employees tends to be higher than for high-skilled employees, would seem to be more pertinent.

3.3.4 Social security reforms

The proposal outlined earlier, about negative income taxation or earned-income tax credits, while relevant to the improvement of the operation in the labour market in general, would be particularly beneficial at the bottom end of the labour market.

3.4 The case for a reform package drawing on a range of policies

One of the points that comes through very strongly from Katz (1998), from Jackman (1998), and from Martin (1998) is that monocausal solutions are inappropriate. It is not good enough just to change the unemployment benefit system. It is not good enough just to make the wages system more flexible. It is not good enough just to enhance labour market programs to increase the work-readiness of the unemployed. It is necessary to undertake a combination of all these and other things.

4. Should We Aim to Reduce Unemployment Substantially?

4.1 The case for aiming to reduce unemployment substantially

Australia has had an unemployment rate averaging over 7.5 per cent for about the past twenty-five years. I argue that we should seek to adopt a reform package aimed at reducing the level of unemployment to around 5 per cent or below. It could be argued that if this were possible, why have we not done it before? Are the costs of achieving it likely to be too great to justify the policy package?

Against this, however, there is a strong argument that there is a danger of underestimating the benefits of reducing unemployment. Firstly as shown in one of the papers for this conference (Borland and Kennedy 1998) and previously by Dawkins (1996), unemployed people generally feel worse off in terms of their life satisfaction than is likely to be explained by their low incomes. In turn we can expect them to feel better off if they became employed.

Second, the high levels of unemployment that we have been experiencing tend to lead to sub-optimal economic policies, particular when unemployment is rising. Thus for example, the Coalition Government slowed up on the tariff-reform agenda when unemployment was rising in 1997. Similarly, high and rising unemployment tends to foster anti-immigration attitudes, despite the lack of evidence that immigration has had any role in causing unemployment. If we could get unemployment down to 5 per cent or below then such problems might be eliminated.

4.2 Winners and losers and the problem of politics

Nonetheless the package of policies proposed here can be expected to have winners and losers. The main winners would be those employees who would otherwise have been unemployed. There would however, be losers amongst those people who would have been employed anyway but whose net income turns out to be lower. Some people would have lower wages than they would have otherwise, although changes to tax and social security would make some of them better off. Other wages may not have been affected much but their tax rates may be higher.

Major policy packages which have significant winners and losers tend to make politicians nervous. They tend not to like policies with losers. The more the policy package can be designed as a positive-sum game, the more likely the politicians will adopt it. It is very unlikely, however, that there would not be some losers and this reinforces the need to raise the profile of the importance of reducing unemployment for the national interest.

4.3 Uncertainties involved

A further problem is that the effects of embarking upon a policy package are uncertain. I have argued that adopting the policies proposed above might be able to reduce unemployment to 5 per cent or below. Whether this turns out to 6 per cent or 4 per cent or some other number, however, remains uncertain. This is partly because we do not have precise estimates of the elasticities involved etc. For the same reason, estimates of the budgetary cost of such policies are also uncertain. Judgments then have to be made as to whether the uncertain benefits are likely to outweigh the uncertain costs.

My judgment is that a policy package of the kind proposed here is worth embarking on, for the reasons outlined. More attempts to quantify the costs and benefits, and the winners and losers could be made, but given the uncertainties involved, judgments have to be made eventually. One thing that should help such a judgment to be made is that if a policy package aimed at reducing unemployment substantially is not adopted, that is tantamount to saying that we are broadly speaking content with the status quo.

5. Conclusions

Unemployment has averaged over 7.5 per cent over the past twenty-five years. This is generally regarded as having been too high and a strong policy debate about how to get it down has resulted. The papers for this conference, between them, represent a considerable advance to the debate in Australia, about the causes of persistent high unemployment, and clues about where to look if we want to make a substantial reduction in the unemployment rate.

In reviewing the evidence and many of the ideas discussed in these papers, I conclude that a systematic policy package aimed at getting unemployment down to 5 per cent or below, on a sustained basis, should be adopted. This will require a mix of policies aimed at continuous improvement of the education and training system, higher economic growth, less severe recessions, lower equilibrium real wages, improved and more systematic labour market programs, reform of the tax and social security system to reduce effective marginal tax rates to increase work incentives and to supplement low wages, and continuing reform of the wage-setting and industrial relations system to achieve more decentralised wage determination. In so doing, I argue that it is important to get around the ‘diabolical trade-off’ that is sometimes thought to exist between lower unemployment associated with freer labour markets, and greater inequality. In keeping with the argument put forward by Katz (1998) for the US, we should aim for a freer labour market, and reforms to the tax and social security system and to education and training and labour market programs that are in tune with this flexibility but have positive outcomes for equity.


There have been various estimates of this and some of these have been discussed in submissions to the last two Living Wage cases, because it is this proportion of employees who depend upon a wages safety-net adjustment for a wage increase. The Melbourne Institute (1998b) has a regular survey of employees who have been in the same job for the past twelve months, which has produced estimates of between 30 and 40 per cent of such employees whose rate of pay ‘is currently determined by the safety net, that is your wages are solely determined by an award’. This is not to be confused with being an award-based employee as employees on registered enterprise agreements are ‘award based’ but their pay is determined by the agreement. Of course, it is still likely that many agreements would be influenced by what happens to awards. [1]

The Melbourne Institute and NATSEM are currently costing alternative approaches to introducing earned-income tax credits. The findings will be published in Dawkins and Beare (1998). As an example, we have costed an earned-income tax credit equivalent to 5 per cent of wages (plus 1 per cent for each child up to a maximum of 3 children) – up to wage income of $20,000, held at $1,000 (plus $200 per child) until income of $25,000, and tapered down to zero by $40,000. The cost is computed to be $2.6 billion, or $1.5 billion if restricted to families with incomes less than $40,000 (a more gentle means test would be preferred). [2]


Blandy, R. (1998), ‘A Scheme for Full Employment’, mimeo, Austral*asia Economics, Adelaide.

Borland, J. (1997), ‘Unemployment in Australia – Prospects and Policies: An Overview’, Australian Economic Review, 30(1), pp. 391–404.

Borland, J. and S. Kennedy (1998), ‘Dimensions, Structure and History of Australian Unemployment’, paper presented at this conference.

Bresson, G., F. Kramarz and P. Sevestre (1992), ‘Heterogeneous Labor and the Dynamics of Aggregate Labor Demand: Some Estimations Using Panel Data’, Empirical Economics, 17(1), pp. 153–168.

Calmfors, L. and J. Driffill (1988), ‘Bargaining Structure, Corporatism and Macroeconomic Performance’, Economic Policy, 6, pp. 14–61.

Card, D. and A. Krueger (1994), ‘Minimum Wages and Employment: A Case Study of the Fast-food Industry in New Jersey and Pennsylvania’, American Economic Review, 84(4), pp. 772–793.

Card, D. and A. Krueger (1995), Myth and Measurement: the New Economics of the Minimum Wage, Princeton University Press, Princeton, New Jersey.

Committee on Employment Opportunities (1993), Restoring Full Employment, AGPS, Canberra.

Dawkins, P. (1996), ‘The Distribution of Work in Australia’, Economic Record, 72(218), pp. 272–286.

Dawkins, P. and G. Beare (1998), ‘Costing Earned Income Tax Credits for Australia’, Melbourne Institute of Applied Economic and Social Research, University of Melbourne Working Paper, forthcoming.

Dawkins, P., G. Beare, A. Harding, D. Johnson and R. Scutella (1997), ‘Towards a Negative Income Tax for Australia’, report by the Melbourne Institute and NATSEM to the Full Employment Project.

Dawkins, P. and J. Freebairn (1997), ‘Towards Full Employment’, Australian Economic Review, 30(4), pp. 405–417.

Dawkins, P. and D. Harding (1997), ‘Developments in the Australian Labour Market: What Chance 5% Unemployment Now?’ Mercer-Melbourne Institute Quarterly Bulletin of Economic Trends, 1.97, pp. 36–50.

Debelle, G. and J. Vickery (1998), ‘The Macroeconomics of Australian Unemployment’, paper presented at this conference.

Dungey, M. and J. Pitchford (1998), ‘Prospects for Output and Employment Growth with Steady Inflation’, paper presented at this conference.

Freebairn, J. (1998), ‘Microeconomics of the Australian Labour Market’, paper presented at this conference.

Freeman, R. and L. Katz (1994), ‘Rising Inequality: The United States vs. Other Advanced Nations’, in R. Freeman (ed.), Working Under Different Rules, Russell Sage, New York, pp. 29–62.

Freeman, R. and L. Katz (1995), Differences and Changes in Wage Structures, University of Chicago Press and NBER, Chicago.

Garnaut, R. (1998), ‘Time to End Unemployment’, Mercer-Melbourne Institute Quarterly Bulletin of Economic Trends, 1.98, pp. 55–62.

Goodridge, S., D. Harding and P. Lloyd (1995), ‘The Long-term Growth in Unemployment’, Melbourne Institute of Applied Economic and Social Research, University of Melbourne, Working Paper No. 2/95.

Grant, J. (1979), Labour Substitution in U.S. Manufacturing, PhD thesis, Michigan State University.

Hamermesh, D. (1993), Labor Demand, Princeton University Press, Princeton, New Jersey.

Hamermesh, D. (1995), ‘Comment’, Industrial and Labor Relations Review, 48(4), pp. 835–838.

Harding, A. and S. Richardson (1998), ‘Unemployment and Income Distribution’, paper presented at this conference.

Industry Commission (1997), Assessing Australia's Productivity Performance, AGPS, Canberra.

Jackman, R. (1998), ‘European Unemployment: Why is it So High and What Should be Done About it?’, paper presented at this conference.

Katz, L. (1998), ‘Reflections on US Labour Market Performance’, paper presented at this conference.

Keating, P. (1994), Working Nation: Policies and Programs, AGPS, Canberra.

King, W. (1980), A Multiple Output Translog Cost Function Estimation of Academic Labor Services, PhD dissertation, Michigan State University.

Krugman, P. (1994), ‘Past and Prospective Causes of High Unemployment’, in Reducing Unemployment: Current Issues and Policy Options, Federal Reserve Bank of Kansas City, Kansas City, Missouri, pp. 49–80.

Layard, R., S. Nickell and R. Jackman (1991), Unemployment: Macroeconomic Performance and the Labour Market, Oxford University Press, Oxford.

Lewis, P. (1985), ‘Substitution Between Young and Adult Workers in Australia’, Australian Economic Papers, 24(44), pp. 115–126.

Martin, J. (1998), ‘What Works Among Active Labour Market Policies: Evidence from OECD Countries’ Experience', paper presented at this conference.

Melbourne Institute (1997a), Mercer-Melbourne Institute Quarterly Bulletin of Economic Trends, 3.97, pp. 1–5.

Melbourne Institute (1997b, 1998a), Survey of Consumer Inflationary Expectations, Melbourne Institute of Applied Economic and Social Research, University of Melbourne.

Melbourne Institute (1998b), Quarterly Wages Report, Melbourne Institute of Applied Economic and Social Research, University of Melbourne, February, May.

Nickell, S. and B. Bell (1995), ‘The Collapse in Demand for the Unskilled and Unemployment Across the OECD’, Oxford Review of Economic Policy, 11(1), pp. 40–62.

Okunade, A. (1991), ‘Translog Cost Estimates of Labor–Labor Substitution in U.S. Hospital Pharmacies’, Journal of Research in Pharmeceutical Economics, 3, pp. 91–113.

Pissarides, C. (1991), ‘Real Wages and Unemployment in Australia’, Economica, 58(229), pp. 35–55.

Rogers, M. (1997), ‘Knowledge, Technology and Productivity in Australia: Lessons From Overseas’, Mercer-Melbourne Institute Quarterly Bulletin of Economic Trends, 3.97, pp. 38–57.

Seltzer, A. (1997), ‘An Evaluation of the International Evidence of the Employment Effects of Minimum Wage Legislation’, Australian Economic Review, 30(2), pp. 208–214.

Wooden, M. and J. Sloan (1998), ‘Industrial Relations Reform and Labour Market Outcomes: A Comparison of Australia, New Zealand and the United Kingdom’, paper presented at this conference.