RDP 2010-04: Employment Composition: A Study of Australian Employment Growth, 2002–2006 2. A Look at the Literature

The literature suggests that a cyclical increase in the employment rate will be particularly beneficial for those groups with low productivity or a low propensity to supply labour – that is, people with relatively low employment rates.

There are several reasons why both the demand for and supply of labour by low-skilled workers tend to fall disproportionately during a downturn (and vice versa during an expansion). First, in a downturn employers may choose to hoard workers with more firm-specific human capital – that is, those who are likely to be more highly skilled. Second, a flat unemployment benefit creates a higher relative replacement ratio for low-productivity (and hence low-wage) workers (Borland 2000), reducing their incentive to participate if expected wage growth slows. Third, higher-productivity workers have an incentive to remain employed, owing to their comparative advantage in market work, and during a downturn they are more likely to gain a lower-paid job following separation, thereby crowding out low-productivity workers during an economic slowdown (Okun 1973; Blank 2000).[1] Concurrently, workers with a comparative advantage in non-market work are relatively more inclined to leave, or stay out of, the labour force when wage growth slows in a downturn (Bils et al 2007). Finally, if wage growth slows, the minimum wage rate can become binding for low-productivity workers, encouraging employers to lay off such workers.

These hypotheses are supported by empirical studies, which generally find that employment becomes more broadly distributed in periods of high employment growth. The main body of empirical studies use time series data and focus on the relative cyclicality of unemployment rates by skill level, industry, average wage or education. Borland (2000) presents data for Australia that show that employment rates increase by more for low-skilled cohorts (proxied by age, education and occupation) when the aggregate employment rate rises. Using a long panel dataset from the United Kingdom, Bils et al (2007) also find low-productivity workers have much more cyclical unemployment rates. Gray, Heath and Hunter (2002) adopt a cross-sectional approach using Australian data. They attempt to identify a causal effect of the local unemployment rate on the transition probabilities into employment of two groups with different propensities to supply labour. They find only tentative signs that a higher unemployment rate has a different effect on the transition rates of men and women, but note that this result may be because the survey occurred during a period of volatile employment growth.

However, these studies do not establish whether there is a change in the relative rates of unemployment, marginal attachment or participation across individuals with different personal characteristics. This is an important focus of this study.

Our study does not probe a causal relationship between changes in the aggregate employment rate and the distribution of employment. This would be difficult because over the episode of rising employment relevant to our study, structural or long-term influences are also likely to have affected the relative employment rates of different groups. First, there may have been a relative shift in labour demand owing to changing industry composition or other factors.[2] Second, the deregulation of labour markets over recent decades may have increased flexibility in minimum employment conditions and wages, increasing the relative demand for lower-skilled labour. Finally, the demand for, and supply of, labour by different cohorts may have been influenced by demographic factors and changes in income tax and welfare eligibility. Nevertheless, the period provides a useful case study.


During the period of rising employment over the past decade, Australian Bureau of Statistics (ABS) data suggest that job-to-job churn and the rate of voluntary separation both increased. This is consistent with the idea that higher-skilled or more productive workers take low-productivity jobs when employment rates are low, and then ‘trade up’ as new high-productivity jobs are created and employment is expanding, thereby creating vacancies for those with lower skills to fill. [1]

Nickell and Bell (1995) examine data for a range of countries including Australia and conclude, in contrast to other studies, that the relative unemployment rate of the unskilled generally fell while aggregate unemployment rates rose during the early 1990s recession. They suggest a change in relative labour demand favouring the less-skilled caused this shift. [2]