RDP 9215: The Evolution of Employment and Unemployment in Australia 1. Introduction

Between 1950 and 1974, the average unemployment rate in Australia was 2.0 per cent of the workforce. Since then, it has been 7.3 per cent. Even more strikingly, from 1984 to 1989, employment growth averaged 3.5 per cent per year, by far the fastest growth in the OECD area. Yet, at the end of this period, the unemployment rate stood at 6.2 per cent, higher than following the great economic shocks of the mid 1970s.

In this paper, we ask why labour market performance in Australia has deteriorated so much; i.e. why, the natural rate of unemployment, or alternatively Non Accelerating Inflation Rate of Unemployment (NAIRU), is about six per cent of the labour force, when, for many years, it was synonymous with less than two per cent unemployment.[1] We also ask why high unemployment is so persistent, i.e. why it rises very sharply during recessions, but falls only slowly during periods of strong economic growth.[2]

We examine in detail the recent evolution of employment and unemployment and offer a structural explanation for the deterioration in labour market performance evident since the mid 1970s. The data support the view that recessions are periods of accelerated structural change that impart asymmetric effects on the labour market, namely, the destruction of full-time jobs in manufacturing, mostly held by men, and the creation of employment in the service sectors of the economy. In the most recent period of sustained economic growth from 1984 to 1989, these new jobs were predominantly part-time and filled by women entering the labour force.[3]

A number of foreign studies, especially of the United States, have adopted structural explanations for a rising natural rate of unemployment. Summers (1986) finds that the secular increase in unemployment over the period 1965–1985 was concentrated among mature married men, job losers and long term unemployed; associated with this trend was a decline in employment in high wage sectors, such as manufacturing. Murphy and Topel (1987) examine this issue with an extensive micro data set which records employment, unemployment and earnings for 540,000 individuals over an 18 year period. They find that a decline in the inter-sectoral mobility of workers leads to higher unemployment rates and more frequent and longer spells of unemployment. In a follow-up paper, Juhn, Murphy and Topel (1991) find that the demand for less-skilled individuals in the United States has declined and that the secular increase in total unemployment is due to the rise in unemployment for this group.

In a recent exhaustive study, Layard, Nickell and Jackman (1991) examine unemployment in the OECD countries in terms of a macroeconomic model in which unemployment is determined, inter alia, by price and wage setting behaviour in imperfectly competitive goods and labour markets. They attribute differences in unemployment between countries to the extent of inter-firm co-ordination in price setting, and inter-union co-ordination in wage-setting (more co-ordination leads to less unemployment); to the design of systems of unemployment benefits and their effects on the job-search incentives of the unemployed; and to a host of other variables. Among other results, this study finds that Australia has a high degree of real wage rigidity, implying that adverse shocks (for example, to productivity) lead to large increases in the Australian NAIRU.

Our examination of the Australian data leads us to support the argument made by Gregory (1991), that a complete analysis of employment and unemployment requires separate scrutiny of four distinct labour markets; for men and women, and for full-time and part-time jobs[4]. We extend Gregory's study by examining these labour markets in terms of the gross flows into, out of, and within the labour force[5]; the duration of unemployment; and the evolution of employment, productivity and earnings in different sectors of the economy.

Briefly summarised, we find that the rise in the equilibrium unemployment rate that began in the mid 1970s was due to the economy's inability to adjust to the adverse shocks of the time. We also find that the sources of persistence in unemployment are different for men and women. Male unemployment is persistent because employment has mainly grown in sectors that employ women, not men. Female unemployment has been persistent because the growth in demand for female labour has been matched by the growth in its supply. Finally, we find that recessions appear to have permanent effects on the sectoral composition of the economy, i.e. recessions are periods of accelerated structural change.

The remainder of the paper is organised as follows. Section 2 presents some elementary facts on employment, unemployment and labour force participation. In Section 3 we examine the data on gross flows and duration of unemployment, while in Section 4 we look at disaggregated data on employment and unemployment by sector. An evaluation and summary of the paper are presented in Section 5.


The terms natural rate of unemployment and NAIRU are often used interchangeably. However, this usage is incorrect, as the two concepts have different microeconomic foundations. The natural rate of unemployment is based on models of competitive goods and factor markets and so it represents a Pareto-efficient allocation of resources, subject to the constraints imposed by labour-market regulations, principal-agent problems, and so on. (See Johnson and Layard (1986) for a survey of the determinants and policy implications of the natural rate.) The NAIRU, on the other hand, is based on models of imperfect competition and therefore does not imply Pareto-efficiency. Thus, although both the natural rate and the NAIRU are defined as the rate of unemployment consistent with constant inflation, their welfare properties are very different. In this paper we do not address the issue of whether the correct terminology is the natural rate or NAIRU and instead use the neutral term “equilibrium rate of unemployment”. [1]

For example, during the recession of 1982–83, the seasonally adjusted unemployment rate rose from 5.4 per cent in June 1981 to a peak of 10.4 per cent in July and September 1983. However, in the recovery and subsequent period of expansion the trough in unemployment, at 5.9 per cent of the workforce, did not occur until November 1989. Persistence is often used synonymously with the term “hysteresis”, but this is not, strictly speaking, correct. Hysteresis occurs when transitory shocks which lead to, say, an increase in the unemployment rate also increase the equilibrium rate. While hysteresis is often thought in practice to be a cause of persistence, unemployment can be persistent even in the absence of hysteretic effects, e.g. when unemployment has an autoregressive root which is slightly less than unity. Indeed, even in the presence of hysteresis, it is possible, though not very likely, that unemployment can quickly return to (just above) its pre-shock level, and so not be persistent, in the sense of being above its equilibrium rate for a long period of time.

The existence, or otherwise, of hysteresis is really just a technical issue regarding the time-series properties of the unemployment data. While important in some contexts, this issue is not our concern in this paper. The interesting economic issue, which we examine in detail, is the cause of persistence in unemployment. See Bean (1992) for a survey of some models of persistence. [2]

A part-time job is defined to be one where less than 35 hours per week are worked. [3]

Gregory, McMahon and Whittingham (1985) examine in detail developments in the female labour market in the period 1966 to 1982. [4]

See Blanchard and Diamond (1990,1992) for analyses of the labour market flows in the United States. [5]