Regular readers will be well aware that there are a variety of ways that we can measure the strength of a Labour Market. A standard approach might be to track the unemployment rate (either seasonally adjusted or, preferably, Trend) but this measure can often hide nuances; the impact of increasing (or decreasing) rates of Participation can distort the unemployment rate as people join (or leave) the labour force (we call this the “Participation effect”).
We might decide instead to track employment, but this too comes with difficulties when we consider that anyone employed for an hour or more is counted as “employed”. Considering the split between full-time (35 or more hours a week) and part-time (1-34 hours a week) can help but still the “employed” number tells us nothing about whether, as an example, someone in the “gig economy” is working three different 10 hours-a-week jobs (one part-time employed doing 30 hours-a-week) or the same jobs are being done by three different people (three part-time employed doing 10 hours-a-week each).
One way around the participation problem can be to study the ratio of employment to working population (civilian population over 15 years); this removes the “Participation effect” but relies upon using “employment” as a genuine measure of labour market strength which, as we have seen, comes with its own issues.
The ABS provide us with monthly data (down to State level) on total hours worked in all jobs and this would appear to be a good place to start to look for a better indicator; as a labour market strengthens we would expect to see total hours worked increase as more people join the labour force, more jobs become full-time (rather than part-time) or people take on more available jobs. Nevertheless, the simple total hours worked data would also increase along with population growth; as our working population grows we need to create more jobs for people to do; an increase in hours worked per se does not equate to a “stronger” labour market. It could simply be attributable to stronger population growth while the labour market fails to keep pace. In such a case a declining Participation Rate could lead to a stable (or even falling) unemployment rate which would effectively mask the weakness at heart.
We therefore need to adjust the total hours worked data for changes in the size of the working population. A total hours worked per capita of working population should remove issues around increasing population levels, excludes the “Participation effect” and captures work done across all jobs whether those jobs are full-time, part-time or multiple. Changes in this measure will give us, in our opinion, a more robust measure of the actual strength of a labour market. So what do these various measures tell us?
The unemployment rate is Queensland (Trend) currently sits at 5.9%, where it has been for 5 months having slowly declined from a recent high (end 2014) of 6.7%. However, this story hides the fact that over that period we have seen Participation fall sharply until end 2016 before starting to recover; it has now returned to where it was at the end of 2014. The falls in the unemployment rate from end-2014 to early-2016 came primarily from increases in employment. However, the period through 2016 (when the unemployment rate remained relatively stable around 6.0-6.2%) was marked by a steady decline in Participation which masked the impact of declines in employment: During 2016 the state lost 29,500 employed while the working population grew by 60,100, surely a far weaker labour market than suggested by a relatively stable unemployment rate.
Since then we have seen the unemployment rate move lower, participation has increased (although it remains well below historical highs) and we have seen solid employment growth. But how does this measure up against the growth of the working population?
The chart below plots the same unemployment rate (inverted on the right hand axis) and the employment to working population measure. This confirms that during 2016 (when the unemployment rate remained steady) the “Participation effect” was masking a rapid, and dramatic, decline in the strength of the labour force as jobs created failed to keep up with population growth. Likewise it suggests that the recovery since 2016 has more than accounted for increases in population, although as we can see the measure remains well below pre-GFC levels. While this picture gives us a better insight than the simple unemployment rate it still does not account for the issue of the quality (number of hours work) over quantity (number of people employed). For that we must look further.
The chart below measures the total number of hours worked (per month) per capita of working population. Again it highlights the actual weakness in the market during 2016 as hours worked per capita fell sharply although the unemployment rate remained stable. It also shows a solid resurgence in total hours worked per head since the end of 2016 (as shown above); but the difference here is that the graph suggests that a significant gap has opened up between the level of the unemployment rate and the hours worked per capita. Why might this be? Quite simply it comes about from the fact that total hours worked have not increased as fast as the employment numbers would suggest they might have done. This could be a result of a large number of part-time jobs being created, or the fact that even those working full-time are doing fewer hours than previously. At this current level of unemployment (5.9%) history would suggest that total hours worked per capita would be around 1-2 hours per month (or about 1.7%) higher than they presently are. Or, to look at it another way, at these levels of hours worked per capita we might expect the unemployment rate to be closer to 6.1%.
In conclusion, total hours worked per capita of working population improves on the unemployment rate as a measure of labour market strength by seeing through the “Participation effect” but it also accounts for changes in the structure of employment (the mix between full-time and part-time and the number of hours worked). Studying the two together confirms the improvements in the Queensland labour market during 2017 but also highlights that the market remains somewhat weaker than other measures would suggest as total hours worked have failed to keep up with the “employment” data.
As an interesting aside, the second chart below demonstrates that this underlying weakness would appear to not exist at the national level.