Aussie Labour Market

Trying to explain the strength in jobs growth

Economists have a habit of only talking and writing about things that they can explain with confidence.¹

This piece is about our struggles to come close to fully explaining the strength in Australia’s labour market.²

On the one hand, it is clear that the recent composition of GDP growth - led by robust growth in government spending - has supported overall jobs growth (and indirectly exacerbated skills shortages in some industries).

But trying to quantify this compositional effect on employment growth is another matter.

We ran several different employment models and the results were mostly underwhelming. We provide a few reasons for what we might be missing.

Following a period of robust jobs growth, and weak labour productivity growth, we remain wary that jobs growth may eventually underwhelm despite the expected cyclical pick-up in (private) demand growth.

Non-market industries as a key driver of jobs growth

The data paint a clear picture on the sources of jobs growth.

It is now well-understood that so-called non-market - or government-aligned - industries have contributed significantly to growth in both jobs and hours worked in recent years, with health care & social assistance the standout.³

These are not all public-sector jobs per se but these industries are heavily influenced by government spending and decisions.

Since Q1 2018, non-market industries have accounted for more than half of total jobs growth despite comprising ~30% of jobs in Australia. It’s a very similar story for growth in hours worked over this period.

It is undeniable that strength in government spending - including on aged, disability and other health care - has contributed to robust non-market sector jobs growth.

Since 2015-16, spending on social protection - including disability and aged care spending - rose from 11.5% of total government consumption to 17.7% in 2023-24 and has risen noticeably as a share of GDP.

But the increase in government spending has also supported labour demand in the market sector (despite recent sluggishness in market-sector jobs growth).

For example, while a significant share of NDIS spending is on disability carers and other health care workers, it also contributes to construction and home modification activity, demand for transport, retail and recreation activity, the bureaucracy, and so on.

Strong government demand is also likely to have contributed to labour shortages in some market-sector industries by increasing overall demand for labour. Anecdotal evidence strongly suggests that this has been the case in the construction industry, with robust infrastructure demand sucking labour away from other types of construction activity.

Consequently, the job vacancy rate in market-sector industries remains well above pre-pandemic levels despite weak jobs growth overall in the market sector.

A standout is the hospitality sector where the job vacancy rate remains more than twice as high as pre-pandemic rates despite no growth in real spending on eating out & takeaway for more than 2 years!

It is highly plausible that very strong demand for relatively well-paid caring roles has made it more difficult for the hospitality industry to retain and attract workers (despite very strong net immigration).

In summary, there is ample partial and circumstantial evidence that robust growth in government spending has supported jobs growth in some industries and is likely to have exacerbated skills shortages in other industries (i.e. ‘crowded out’ some jobs growth).

But this doesn’t help us quantify the aggregate employment effect of strong government spending. It also doesn’t get us close to accurately gauging how important changes in the composition of GDP growth have been for the strength in jobs growth.

So we tried a more structured approach using some (relatively simple) modelling.

Can modelling help us understand the resilience of jobs growth?

Standard employment models attempt to explain jobs growth using a combination of output and real wages growth as explanatory variables. We attempted several variations.