RBA Annual Conference – 2011 Discussion

1. Judith Sloan


Jeff Borland's paper represents a comprehensive and thorough analysis of the high-level developments in the Australian labour market in the first decade of the twenty-first century. He describes the decade as quiet, perhaps suggesting that it was boring. My take on the decade was that labour market developments in Australia during the 2000s should properly be assessed as ‘exciting’ – in the sense that the average rate of unemployment was much lower than the previous decade, the labour force participation rate increased overall and for particular groups, and the employment-to-population ratio reached historic highs.

While unemployment had averaged 8.8 per cent in the 1990s, the figure for the 2000s was 5.5 per cent. In fact, unemployment got down to 4 per cent at one point (in 2007) – a figure which had not been seen for several decades. And the number of employed persons grew at an average annual rate of 2.2 per cent in the first decade of the twenty-first century compared with 1.2 per cent in the 1990s.

Perhaps the most exciting development of all was the resilience of the Australian labour market to the economic downturn associated with the global financial crisis (in 2008–2009, in particular). Adjustment occurred almost entirely through average hours worked rather than the number of persons employed – a combination which was not paralleled in the sharp downturn that occurred in the early 1990s.

Unit-record data

At the beginning of the paper, Borland refers to the growing importance of unit-record data sources, including longitudinal ones, as an important development of the 2000s. Unfortunately, Borland's analysis does not make use of any of these sources. Lest it be thought that mining unit-record data has little relevance to policy issues, there are in fact many topics that can be helpfully considered with these data.

Take the issue of the impact of minimum wages. If it is the case that low-paid workers stay in low-paid jobs year after year, then the equity argument for minimum wages is very different from the case where low-paid employment is a mere stepping stone to higher-paid jobs for many workers. Longitudinal unit-record data enable the researcher to estimate the fate of low-paid workers over time and the characteristics of the worker and other variables that determine these outcomes.

Similarly, the discussion of working hours is much enhanced by the use of unit-record data. There has been a noticeable increase in the extent of underemployment – which actually dates back to the late 1980s. When assessing the state of the labour market, it is appropriate to add the rate of underemployment (weighted) to the rate of unemployment. But it is interesting to know whether or not workers remain in a state of wanting more working hours or whether the desire for more working hours is generally met within a relatively short time. The analysis of working-time patterns using HILDA data points to a lack of persistence in underemployment. By contrast, overemployment – where workers say they would prefer to work fewer hours – does seem to persist.

Does labour market regulation matter?

The results of one of the important analytical exercises undertaken by Borland are presented towards the end of the paper. Using the rate of unemployment as the independent variable, Borland estimates the Phillips curve (curiously using CPI inflation rather than wage inflation) to test whether or not any structural break was evident in the 2000s. (The data span the period 1978:Q2 to 2011:Q2.) He concludes that ‘there has been an inward shift and substantial flattening of the Phillips curve relation after the latter part of the 1990s’ (p 207). He then goes on to ponder whether this result is due to reforms to the wage-setting system that had occurred in the 1990s.

Borland argues that the rate of unemployment is not a good proxy for the state of the labour market; he prefers an augmented measure which combines the rate of underemployment with the rate of unemployment. However, he is forced to use an unweighted version of underemployment as the data only provide annual observations on weighted underemployment. His results point to the rate of labour underutilisation having a significant negative effect on inflation but ‘the structural change variables [are] no longer significant’ (p 207). On this basis, at the Conference he concluded that there was little evidence of change in the nature of wage setting in the past decade.

There are a number of points that could be made about this analysis. The first is the decision to model the rate of consumer price inflation rather than the rate of wage inflation; the latter is more likely to be sensitive to the changing state of the labour market than the former. The second is that unweighted underemployment may be a misleading indicator if the proportion of additional preferred hours to actual hours changes over time. A final point is that irrespective of the modelling exercise, it is clear from the lower panel of Figure 31 that the 2000s experienced much lower levels of inflation for a given rate of labour underutilisation.

One of the reasons why this type of analysis is interesting is that the question is begged: do labour market regulations matter? Economists should be able to contribute to this debate. If they do not matter – say, in terms of unemployment and inflation – then there are few reasons why minimum wages should not be doubled, for instance, or employment protection laws further strengthened. Of course, most economists would baulk at these propositions; rather, the argument is that there are limits beyond which labour market regulations will drive outcomes that are quite divorced from what otherwise would be market-driven outcomes. Hence, there would be economic losses associated with the impact of regulation. But where to draw the line in terms of defining ‘sensible’ labour market regulations? Indeed, how should labour market regulations be conceptualised and measured for the purpose of economic analysis?

These questions were beyond the scope of Borland's paper. However, they are ones that should interest central bankers because the functioning of the labour market is one of the aspects of the economy on which they focus.

2. General Discussion

The performance of the Australian labour market during the global financial crisis was a key issue taken up during the discussion. One participant made the point that, unlike during the downturns of the early 1980s and early 1990s, the fall in labour demand during the GFC was largely reflected in a decline in hours worked, rather than a large number of jobs disappearing. One view was that the expected length of a downturn was an important consideration, and since the slowdown towards the end of the 2000s was relatively short-lived, it was in the interests of both workers and employers not to break ties and shed jobs. This point was contrasted with the US experience, where the downturn during the GFC was more drawn out and there has been a larger shedding of jobs, despite the highly deregulated labour market. However, one participant noted that the downturn was not forecast to be short and another mentioned that labour shortages in Australia prior to the financial crisis could have made employers reluctant to break ties with existing employees.

A different participant disagreed, stating that evidence from panel data (HILDA) does not support the argument that existing employees had their hours reduced, showing no evidence of a direct substitution between heads employment and hours worked. They said that according to those data, there was a significant change in the composition of employment, suggesting that many jobs were lost during the crisis period but this was offset by jobs being created in industries that hired a large share of part-time workers.

It was also mentioned by one participant that various concerns relating to the changing labour market over the past 20 years have all surprised on the upside. First, worries that decreasing tariffs and increasing free trade would lower employment had not eventuated. Second, fears surrounding labour market regulation had not come to fruition. The participant stated that Australia had largely followed its own path in terms of labour market regulation (somewhere in between the US and European models) and that this had delivered positive outcomes. Third, increased immigration has had the opposite effect on unemployment to what was originally feared by some, with a large number of high human capital workers arriving in Australia and generating jobs for other workers (‘we have been importing employers’). Fourth, it was mentioned that concerns over introducing equal rights and equal pay for females had been misplaced.

There was also substantial discussion on the prospects for the Australian labour market over the coming decade. It was mentioned by one participant that when unemployment fell below 5 per cent in the latter half of the 2000s, there was a noticeable pick-up in wage growth, suggesting that an unemployment rate of 4½ to 5½ per cent was perhaps the lowest rate that was sustainable. It was noted that for wage growth of around 4 per cent to be consistent with the inflation target, productivity growth would have to increase. One participant pondered what the current level of the non-accelerating inflation rate of unemployment (NAIRU) in Australia might be. One view was that over the 2000s it had probably fallen by around ¾ percentage point, suggesting that the unemployment rate could possibly decline without inflation increasing. However, if a labour underutilisation measure of demand was used in this calculation instead (to account for underemployment) there was less evidence of a fall in the NAIRU. An analysis of the Beveridge curve does not suggest there has been an increase in jobs mismatching as a result of the mining boom, with jobs better able to be matched over the 2000s due to a fall in the long-term rate of unemployment. Compositional change across states and between industries was similar between the 1990s and 2000s, and while wages did increase in Western Australia and Queensland as a result of the mining boom, overall wage increases were not transmitted to other states. This was seen as evidence that Australia has so far been able to accommodate the mining boom without a wage breakout (which was not the case following past booms in the terms of trade – a point also made in the paper by Connolly and Orsmond).

It was also highlighted during the discussion that active labour market policies (e.g. targeting local areas) were not very successful and that a combination of deregulation and policies that promoted growth were likely to deliver the best outcomes.