RBA Annual Conference – 1998 Discussion

1. Peter Kenyon

One of the objectives in obtaining the necessary and sufficient conditions for rational choice under majority decision is to motivate purposive research on actual patterns of preferences.

A.K. Sen (1970, p. 171)

I have no real problem with Jeff and Steven's paper. It is an important survey of the dimensions of Australia's unemployment problem. When read in conjunction with Jeff's complementary 1997 survey (Borland 1997), a complete picture of the dimensions of unemployment in this country and the policy options for dealing with it emerges. I particularly like the fact that, at last, the social costs of unemployment, such as its effects on health, crime rates and so on, are being addressed in an economics survey. Isn't it amazing to see how long it has taken for an economist to address these important dimensions of unemployment; dimensions which have, for the most part, not been much addressed in the economics literature?

As the paper is a survey of the dimensions of unemployment, there is little to argue with. It essentially sets the scene for the rest of the discussion that is the business of this conference. So, I will use my time to talk about unemployment policy. I wish to discuss two questions: what determines a policy response and how does this change over time?

But before addressing these questions, I want to contrast two policy issues: unemployment and microeconomic reform. I want to look at the welfare costs and benefits of not getting unemployment down to something like full employment with the welfare benefits that may flow from microeconomic reform.

We have been saying that unemployment is a ‘serious problem’ for a long time. But it doesn't mean that a lot gets done about it. It is serious – more so than just about anything else I can think of as an economic policy imperative.

How big a problem is it? We can quickly approximate the loss of GDP due to total unemployment by a simple expression (we will deal with unemployment in excess of full-employment unemployment shortly):

Loss due to unemployment = (GDP/E) * U

where E = total employment and U = the total number of unemployed.

That is, the total dollar loss is the average product per worker multiplied by the number of unemployed workers (assuming for the moment that the average product of unemployed workers is the same as that of employed workers). Expressed as a proportion of GDP, this becomes simply:

Loss due to unemployment = U/E

Which, after some manipulation, becomes:

Loss due to unemployment = ur/(1–ur)

where ur is the unemployment rate.

However, we need to take into account two factors; first, the unemployment which would occur even at full employment and, second, we have to make allowances for differences in the average productivity of unemployed workers when compared with employed workers. Both of these adjustments require judgments to be made. Conceptually, however, the formula that captures these effects is easy to write. First, let ur′ be the unemployment rate in excess of the full-employment unemployment rate, so ur′ = urur(fe) where ur(fe) is the unemployment rate at full employment.

From Borland and Kennedy's paper we know that the unemployed tend to be less-educated, younger workers. Their previous job (if they had one) was typically as a labourer or tradesperson in either the manufacturing, construction or the hospitality sectors. Recent immigrants are also more than proportionately represented among their number. Consequently, the average productivity of the typical unemployed worker is likely to be less than that of the average employed worker. So, let π (0<π = 1) be the adjustment factor by which the average productivity of the average employed worker needs to be adjusted to obtain the average productivity of the average unemployed worker. Therefore:

Loss due to unemployment in excess of full employment = π * ur′/(1−ur′)

Readers may make their own judgments as to what π is in reality. I shall let π take values of 0.5, 0.75 and 1.0, so as to give a representative range of values.

What is the unemployment rate at full employment? This is virtually impossible to know. Borland and Kennedy discuss various attempts at measuring the so-called ‘natural rate’. The first point to make is that they observe that the confidence intervals around such estimates are typically very large. The second observation from their and other papers at this conference is that it would appear that the ‘natural rate’ has increased from 2 per cent in the 1970s to over 6 per cent by the early 1990s. Why has the ‘natural rate’ drifted upwards? Explanations include hysteresis, whereby the ‘natural rate’ follows the actual rate upwards (and presumably downwards) due to such factors as skill atrophy among the unemployed, state dependence and scarring, structural change brought about by globalisation and/or technological change, demographic changes and so on. The third observation is that none of this can be directly applied to the counterfactual of what would have been the unemployment rate had we had continuous full employment since the early 1970s. We simply do not know what the full-employment unemployment rate was during the years since 1970.

Nevertheless, an estimate needs to be made. I am going to follow the writers of the 1993 Green Paper Restoring Full Employment (Committee on Employment Opportunities 1993) and assume that the current full-employment unemployment rate is 5 per cent. For simplicity, I will assume that the full-employment unemployment rate has grown continuously since 1972, when it was 2 per cent. Table 1 shows the loss due to unemployment in excess of full employment under these assumptions since 1972. As a further check, I also include Mitchell and Watts' (1997) estimates expressed as a percentage of GDP. Their estimates are derived from calculating the number of potential workers available for work above a constant 2 per cent unemployment rate and calculating a ‘GDP bonus’ if each of these workers were employed. Their estimates are considerably greater than mine.

Table 1: Estimates of the Welfare Loss Due to Unemployment
Per cent
Year Unemployment rate Full-employment unemployment rate Unemployment above full-employment rate GDP loss
(1) (2) (1) – (2) π = 0.5 π = 0.75 π = 1.0 Mitchell and Watts
1972 2.5 2.0 0.5 0.3 0.5 0.6 3.3
1973 1.8 2.1 1.3
1974 2.4 2.1 0.3 0.2 0.3 0.4 2.7
1975 4.6 2.2 2.4 1.2 1.8 2.4 9.7
1976 4.7 2.3 2.4 1.2 1.8 2.4 8.3
1977 5.7 2.4 3.3 1.7 2.6 3.4 10.1
1978 6.2 2.5 3.7 1.9 2.9 3.8 11.2
1979 5.9 2.6 3.3 1.7 2.6 3.4 9.5
1980 5.9 2.7 3.2 1.7 2.5 3.3 8.5
1981 5.6 2.8 2.8 1.5 2.2 2.9 7.1
1982 6.7 2.9 3.8 2.0 3.0 4.0 8.9
1983 9.9 3.0 6.9 3.7 5.6 7.4 13.2
1984 8.5 3.1 5.4 2.9 4.3 5.7 10.9
1985 7.9 3.2 4.7 2.5 3.7 4.9 9.1
1986 8.0 3.3 4.7 2.5 3.7 4.9 8.3
1987 7.8 3.4 4.4 2.3 3.5 4.6 7.8
1988 6.8 3.6 3.2 1.7 2.5 3.3 5.9
1989 5.7 3.7 2.0 1.0 1.5 2.0 4.5
1990 7.0 3.8 3.2 1.7 2.5 3.3 5.3
1991 9.5 4.0 5.5 2.9 4.4 5.8 7.8
1992 10.5 4.1 6.4 3.4 5.1 6.8 9.1
1993 10.7 4.3 6.4 3.4 5.1 6.8 9.2
1994 9.2 4.4 4.8 2.5 3.8 5.0 7.9
1995 8.1 4.6 3.5 1.8 2.7 3.6 6.4
1996 8.5 4.7 3.8 2.0 3.0 4.0 6.3
1997 8.4 4.9 3.5 1.8 2.7 3.6
1998 8.0 5.0 3.0 1.6 2.3 3.1

Sources: ABS and author's calculations; Mitchell and Watts (1997).

The numbers are striking in their magnitude. Even under the most pessimistic (and almost certainly unrealistically low) estimates of the productivity of unemployed workers, the welfare losses due to unemployment are enormous. Australia at a minimum has lost well over 1.5 per cent of GDP per year for the past twenty years because it has not implemented policies which would have resulted in full employment. Much more likely is a welfare loss which averaged over 3 per cent of GDP every year for over 20 years. Australia has essentially given away one whole year's worth of its real GDP over the past two decades or so because it has not been willing to implement policies which would have generated full employment.[1]

Now I would like to contrast these welfare losses with the welfare gains that are alleged to flow from ‘microeconomic reform’. It would not be an exaggeration to suggest the microeconomic reform agenda has come to dominate economic policy in this country.

The recent obvious heavy involvement of the Commonwealth Government in the dispute following the replacement of Maritime Union of Australia members with non-union labour on the Australian waterfront by the Patricks Stevedoring company is but one (of many) examples of the priority which is currently being placed on the microeconomic reform agenda. The examples are myriad – the (partial?) sale of Telstra, the almost daily reports of privatisation of virtually all of the Federal Government's Department of Administrative Services functions (the government car fleet, asset services, construction services, architecture and interior design functions and so on), the corporatisation, and in many cases, the privatisation of state electricity, gas and water authorities, the privatisation of many of the capital cities' public transport services, inquiries into just about every state statutory marketing authority, and so on, indicate that the major focus of government (federal and states) economic policy is the ‘microeconomic reform agenda’. It almost goes without saying that unemployment has virtually disappeared as a major focus of economic policy, let alone the primary focus.

Microeconomic reform can mean many things. However, following the Industry Commission's brief to analyse the effects of ‘Hilmer and related reforms’ on economic growth and the fiscal position of the federal and state governments (which resulted in the Industry Commission's 1995 Report), microeconomic reform has been interpreted very broadly as those reforms which focus on competition policy and which are derived from the Hilmer Report (Hilmer, Rayner and Taperell 1993). These ‘Hilmer and related reforms’ comprise reform of Telstra, Australia Post, the Federal Airports Corporation, the Civil Aviation Authority, rail, road and port authorities, state electricity, gas and water authorities, the competitive tendering for the provision of public services, deregulation of the building industry and the move to self-regulation of many other industries.

The estimates of the welfare gains that would allegedly flow from the implementation of this microeconomic reform agenda range from about 5.5 per cent of GDP (Industry Commission 1995) to, at most, about one per cent of GDP (Quiggin 1997). Quiggin subjects the assumptions and calculations of the IC estimates to close scrutiny. His much lower estimate of the gains from the reform agenda results from his finding that most of the IC estimates of productivity gains that would come from reform are over-optimistic due to the use of inappropriate benchmarks and inappropriate assumptions about productivity growth, and therefore represent upper bounds rather than most likely outcomes. Further, he observes that many of the workers displaced from employment by the reform agenda will not be entirely absorbed elsewhere in the economy in equivalent jobs, but at least some of them will be permanently displaced from employment and others will almost certainly move into jobs which require fewer skills, both of which imply a semi-permanent shift in the effective supply of labour (Quiggin 1997). We should not lose sight of the fact that much of the reform agenda is about reducing employment in the public sector.

In summary, it is apparent that the gains in terms of additional GDP from microeconomic reform are at most comparable with the losses in GDP that have resulted from 25 years of less than full employment and are probably considerably less in magnitude, yet microeconomic reform dominates the policy agenda.

So the question is, why is it that microeconomic reform has come to dominate the policy attention of governments and unemployment has slipped very much into the background? Indeed, not only is unemployment not at the forefront of policy concerns, expenditure on active labour market programs to shift the unemployed, particularly the long-term unemployed, into employment has been savagely slashed.[2]

This question gives rise to several further questions. Whose preferences does this shift in policy focus represent? Have these preferences changed over time? If so, why? How are community preferences about the role of government ascertained by government and its agencies and how are these translated into policy priorities?

We know very little about whose preferences affect unemployment policy (or any policy, for that matter). This is a blind spot for the economics profession as far as I can see. We have some notion that policy preferences are revealed through the political process. We also understand that political parties spend a lot of time and effort (and money) surveying public opinion in various ways – marginal seats, various demographic groups and so on. But, surely it is naive to believe that there is a one-to-one correspondence between public preferences about such matters as the quantity and type of public goods and services required, the direction of macroeconomic policy concerning unemployment (which is a form of public good) and the public policy which results.

Even a passing acquaintance with public-choice theory (even if as only revealed through re-runs of Yes Minister/Prime Minister) tell us that the general public's preferences are only very inadequately revealed through the ballot box.

A few years ago Michael Pusey addressed this issue (Pusey 1991). His answer? Economists with a predisposition for economic liberalism or economic rationalism, as it is known in this country.[3] Following a tradition in political sociology that dates at least from Weber and which received its most influential articulation in this country in the work of Sol Encel (Encel 1970), Pusey argues that the top echelons of the Commonwealth public service not only work closely with ministers in the implementation of policy, but are the brokers of interests and the articulators of national ideals and goals and, as such, contribute greatly to the formulation of policy. This process works in parallel with, but also transcends, the formal democratic process in a way which is captured so brilliantly in the BBC television model of Sir Humphrey and Jim Hacker.

In addition, Pusey argues that these influential public servants, particularly those concentrated in the key economic bureaucracies such as the Treasury, Prime Minister and Cabinet and Finance, are drawn from similar socioeconomic backgrounds (no prizes for guessing where!), are predominantly trained in economics and as a result, are deeply steeped in the ideology of liberal economic thought/economic rationalism. It is this latter attribute that results from the largely neoclassical economics education that they have received, which differentiates the current top echelons of the Commonwealth bureaucracy from their predecessors, such as H.C. Coombs who espoused a far more catholic set of economic principles, including very healthy doses of Keynesianism.[4]

Like Marx's caricature of Nassau Senior, this analysis was seen to be altogether too simplistic by the economics profession. (See, for example, Dick Blandy's trenchant review in the Australian Quarterly (Blandy 1992)). Nevertheless, Pusey's work did prick a nerve in the Australian economics profession.

In all honesty, I don't think we know precisely (or even imprecisely) how policy preferences are formed and behave over time, especially in the realms where preferences do matter – the demand for public goods, the degree of macroeconomic intervention required to better satisfy the need for economic certainty of the average person, the need for economic regulation to protect the average person from the exercise of monopoly economic power. But one thing is for sure, economists have not been very interested in finding out!.[5] I make several observations:

  • We don't ask people about policy preferences (and even if we did, we probably wouldn't believe the answers, would we?).
  • We don't pay much attention to political scientists/sociologists when they worry about policy preferences and whose preferences are paramount.
  • We rarely study policy formation except in a rudimentary way (usually using public-choice theory or game theory).

Do preferences for what we want governments to do change over time? This could be crucial. (We can get pathological policy responses where people become immune to monstrous acts by governments, as for example in Nazi Germany in the 1930s.) Have we become immune to unemployment?[6] Once Australian governments would either lose office or come very close to it for very small increases in unemployment (for example, Menzies in 1961, McMahon in 1972, Whitlam in 1975 and Fraser in 1983). Not any more.

It is my belief that we need to understand much more about people's attitudes towards the role of government in economic activity, how these attitudes do (and don't) influence policy and whether they change over time. In order to do this, as a first step, we need to know what are people's preferences about the role of government. To use what is now a hopelessly old-fashioned concept, just what is the desired mix between public and private activity in the ‘Mixed Economy?’.

However to answer these questions, I believe that we have to stray a long way from the economist's usual tool kit, so I end with an epistemological plea. Economists, I believe, have to broaden their tool kit. I am rapidly coming to the conviction that to understand the nature of community's preferences about the role of government, we need to explore qualitative research methods, using focus groups, semi-structured interviews and the like. It is only by using these methods that complex attitudes, opinions and the strength of opinion can be ascertained. A person's attitude towards the role of the state in his or her life, his or her attitudes towards fairness and equity, his or her attitudes towards private versus public expenditure, taxation and the like, cannot simply be inferred from the type of data sets which economists typically work with, using the methods usually directed at these data sets. Additionally, I believe that it is necessary to go beyond the use of sample surveys. These simply are not rich enough to explore the depth of beliefs, the subtleties lying behind them and so on, not to mention the practical difficulties of (usually) high non-response rates, non-response bias and the like.[7]

The use of qualitative research methods is a radical jump in economic methodology, but not a new one, nor one which I am alone in suggesting. Indeed, a far more impressive economist than myself has cogently argued for the use of this methodology in respect of understanding more about unemployment and labour markets (see Bewley 1996).

So, what are the issues which I believe ought to be addressed if we are to understand whose preferences are dominating economic policy formation? It is my hypothesis that the general public is much more sympathetic to a traditional Keynesian response to unemployment, that the shift in policy to ‘microeconomic reform’ does not reflect the attitudes of the general public towards the desired role of government, that the preference for the ‘Mixed Economy’ is not some artefact of the past.

I would like, therefore, to explore community attitudes towards the following issues:

  • the provision of traditional government infrastructures such as roads, ports, airports, energy and water utilities and the like;
  • the role of government in the provision of essential services, for example, in police, law and order and the like;
  • the role of government in the provision of health care and the mix between public and private provision;
  • the role of government in the provision of educational services at primary, secondary and tertiary levels and the mix between public and private schools and universities;
  • the role of government in the provision of the arts and cultural services;
  • the role of government in planning, zoning and regulating;
  • issues of the privatisation of public goods and services (electricity, gas, water, public transport etc.);
  • the mix between the public and private sector in banking and other business enterprises where government has often played a role; and
  • the taxation system and the tax mix and possible alternatives (such as a GST).

I would like to know about the match between different opinions and socioeconomic circumstances. I would like to know whose preferences affect policy-makers (both elected and non-elected). More ambitiously, perhaps, I would like to know whether preferences in these matters have changed over time, and if so, how.

This is an ambitious research program, but one which I think is essential. To repeat the obvious point that I commenced with: current policy priorities are directed at achieving welfare gains that are, in all probability, only a fraction of what could be achieved if the priorities of policy-makers were, once again, primarily directed at reducing unemployment. As a famous aphorism states, it takes a lot of Harberger triangles to fit an Okun gap.[8] Surely it is incumbent upon the economics profession to find out a little more why policy priorities have become so oblivious of relative costs and benefits. Whose interests are being served by this obliviousness?


Unlike most commentators, I believe that we do have a solution to unemployment (Kenyon 1997). It is not in any way an original solution. Keynes wrote extensively about it in the 1930s and Keynesian policies seemed to work for three decades in the post World War 2 era. One might speculate on the welfare loss due to the macroeconomic intellectual vacuum in economics that has resulted from the demise of Keynesian economics as an operational stabilisation policy package. The reason why the problem is difficult is that the solution is a political decision about fairness and equity as much as it is an economic decision about economic efficiency. [1]

In the 1996–97 Budget, expenditure on labour market programs was reduced by $575 million and by further cuts of $956 million in 1997–98, with still further cuts of $130 million and $175 million scheduled for 1998–99 and 1999–00 (see Budget Statements 1996–97, Budget Paper No. 1, Table 4). [2]

For a discussion of the revival of economic liberalism across the OECD, see Henderson (1995). [3]

For a very detailed analysis of the transition in the ‘Treasury Line’ from eclectic Keynesianism to a more neoclassical bent, see Whitwell (1986). Incidentally, it was Coombs who, I believe, engineered the purchase of the site for the building in which this conference is being held. Presumably this reflected, in part, his concept of public goods. [4]

Of course, in the sphere of private goods that are allocated by and large by competitive markets, there is a tradition in economics which essentially ignores preference formation. And for good reason – preferences are revealed by behavioural responses to changing relative prices and incomes. Of course, for the greater part of publicly produced goods and services, there is no adequate preference revelation mechanism. I submit that this is a fundamental weakness in economists' assessment of the demand for government-produced goods and services and in many areas of the setting of priorities for public economic policy. [5]

Adrian Pagan commented to me during the conference that one possibility is that we have become exhausted by the issue of unemployment. It is simply too hard to deal with, and so policy priorities have moved to areas where policy can be more effective. But whose preferences have dictated this shift? [6]

See the Appendix to Mackay (1997) for a clear argument in favour of qualitative research methods for social science research of the type suggested here. [7]

This wonderful piece of wisdom has been attributed to James Tobin. I thank Bill Mitchell for reminding me of it. But alas, I have been unable to track down the source! [8]


Bewley, T. (1996), ‘A Depressed Labor Market as Explained by Participants’, American Economic Review, 85(2), pp. 250–254.

Blandy, R. (1992), ‘Multiple Schizophrenia: Economic Rationalism and its Critics: A Review Article’, Australian Quarterly, 64(1), pp. 101–106.

Borland, J. (1997), ‘Unemployment in Australia – Prospects and Policies: An Overview’, Australian Economic Review, 30(4), pp. 391–404.

Committee on Employment Opportunities (1993), Restoring Full Employment: A Discussion Paper, AGPS, Canberra.

Encel, S. (1970), Equality and Authority, Cheshire, Melbourne.

Henderson, D. (1995), ‘The Revival of Economic Liberalism: Australia in an International Perspective’, Australian Economic Review, 28(1), pp. 59–85.

Hilmer, F., M. Rayner and G. Taperell (1993), National Competition Policy: Report by the Independent Committee of Inquiry, AGPS, Canberra.

Industry Commission (1995), The Growth and Revenue Implications of Hilmer and Related Reforms, AGPS, Canberra.

Kenyon, P. (1997), ‘Infrastructure Spending and Unemployment: Government Responsibility for Growth and Jobs’, Australian Economic Review, 30(4), pp. 421–432.

Mackay, H. (1997), Generations, Pan Macmillan, Sydney.

Mitchell, W. and M. Watts (1997), ‘The Path to Full Employment’, Australian Economic Review, 30(4), pp. 436–444.

Pusey, M. (1991), Economic Rationalism in Canberra: A Nation Building State Changes Its Mind, Cambridge University Press, Cambridge, England.

Quiggin, J. (1997), ‘Estimating the Benefits of Hilmer and Related Reforms’, Australian Economic Review, 30(3), pp. 256–272.

Sen, A.K. (1970), Collective Choice and Social Welfare, Holden-Day, San Francisco.

Whitwell, G. (1986), The Treasury Line, Allen and Unwin, Sydney.

2. General Discussion

The general discussion was centred on two main issues:

  • the apparently greater attention paid by public policy-makers to microeconomic reform than to the reduction of unemployment; and
  • whether hysteresis exists in the unemployment rate.

It was argued that policy-makers have been more concerned about microeconomic reform than addressing unemployment. Why has micro reform had greater prominence in public policy debate? The welfare gains of microeconomic reform, expressed in terms of per cent per annum contributions to economic growth, are estimated to be between 1 per cent and 5 per cent. In contrast, GDP growth lost per year through unemployment is likely to be considerably higher. Some argued that this raises important questions about whose preferences influence the policy process and why. Others suggested that the public might have become immune to policy failure with respect to reducing unemployment, so that the persistence of high unemployment rates becomes tolerated. This, in turn, influences public choice (or acceptance) of microeconomic reform policies for which, at the level of an enterprise or industry, some unambiguous measure of success can usually be identified.

The efficacy of particular policies for the reduction of unemployment will, however, depend on whether there is hysteresis in unemployment. Here, there was much debate about whether the efficiency of labour markets had deteriorated in a structural sense. In their paper, Borland and Kennedy argued that there was no clear evidence of hysteresis in the period since the early 1980s, either in terms of a shift in the ratio of long-term unemployed to total unemployment or a tendency for regional unemployment rates to increase with the aggregate unemployment rate. However, this was questioned by a number of participants. Some appealed to estimates of the upward drift in the natural unemployment rate. Others appealed to estimates of the Beveridge curve; even simple Beveridge curves have shifted outward, as have sophisticated versions which map the increase in the unemployment rate for an equilibrium rate of vacancies. Some participants countered that while these shifts can be observed, it is not clear that they are statistically significant for any period since the 1970s.

Given the possibility of hysteresis in unemployment, some argued that the profession needs to offer a clearer exposition of the determinants of the NAIRU, rather than simply attributing it to a one-off increase in the level of real wages that occurred in the 1970s. Indeed, with this lack of exposition by the profession, it is unsurprising that politicians fail to commit to a target for unemployment. Such lack of commitment, in turn, conditions the public's preferences and their influence on the policy process.

Finally, it was noted that Borland and Kennedy's paper was arguably the first review article conducted by an economist that paid attention to the social costs of unemployment. This was generally applauded. However, some participants noted that solutions to unemployment, particularly those involving further increases in wage dispersion, also have social costs. Thus, in contemplating solutions to unemployment, one needs to consider both the social costs of unemployment and the social implications of any solution to it.