Many commentators have written about movements in the Participation Rate (PR) and the impact this can have on unemployment rates (one such recent example from Nick Behrens at QEAS). The discussion is certainly a valid one, but in this post we want to try and dig a little deeper into what is actually driving the changes in PR.
Firstly, some definitions. PR is the percentage of the working age population (those 15 years and older) who are in the Labour Force (those either employed or counted as unemployed). Changes in the PR tell us something about the proportion of those able to be in the Labour Force who actually are, but the number itself tells us nothing about why the changes might be occurring. If we wish to understand more fully the structure of the Labour Force we need to look at the two factors at work determining changes in the net PR.
- Changes caused by shifts in PR within the working population age groups
- Changes caused by shifts in the demographic make-up of the Labour Force
Fortunately there is a way of splitting these two effects out from the data. The method was described and detailed in 2009 in a paper entitled Decomposing Changes in the Aggregate Labor Force Participation Rate written for the Federal Reserve Bank of Atlanta by Julie L. Hotchkiss.
The paper specifies the aggregate PR to be the population weighted average of the PR for different age groups. The difference in PR between different time periods (t and t-1) is therefore;
Where LFPR is the Labour Force PR; i are the age groups and p is the percentage of working age population.
The formula tells us that changes in PR between periods can be deconstructed into the change caused by shifts in PR within age groups (let us call this the “Propensity effect“) and change caused by shifts in the demographic make-up (which we shall call the “Demographic effect“).
The Propensity effect can be caused by a range of issues. The most often cited is the impact that a weak labour market will have in discouraging people from entering the labour force; the reverse is true when a strong labour market encourages people to return. However, this is not the only thing causing the Propensity effect. More young people opting to go into tertiary education, or staying longer in secondary education, would also shift the PR within their age group. Likewise more older workers opting for early retirement, or being forced to delay retirement, would impact on PR in that age demographic.
The Demographic effect is a result of shifts in the age make-up of the population. An aging population, as we are currently witnessing across much of the developed world, will shift aggregate PR as a higher proportion of the population moves into the older age groups where PR (naturally) is lower than in middle-age.
In order to deconstruct changes in PR we need to have data for PRs and population proportions for the various age groups considered. While the ABS provides us with aggregate Trend PR levels at a National and State level we have had to construct our own Conus Trend PRs for the chosen age cohorts^ (youth, 15-24 years; middle 25-44 years; older 45+ years) and utilise the Conus Trend Regional Jobs database to make comparisons at a regional level.
^ ABS Trend PR for the 15-24 age group is available at National level and used here.
We have opted to consider the period of a decade from April 2007 – April 2017. We need to look at relatively long periods of time if we are to see meaningful changes caused by demographic shifts.
In Australia over that decade Trend PR fell from 65.0% to 64.8%. The table below details the elements of interest within the relevant age groups.
Applying the formula above to this data we see that the Propensity effect actually increased aggregate PR by 0.3 ppts while the Demographic effect subtracted 0.5 ppts; the decline in PR between April ’07 and April ’17 was caused exclusively by demographic shifts.
We carried out the same process looking at Queensland and the results were quite different. Over the decade Trend PR in QLD fell from 67.4% to 64.6%. The Demographic effect accounted for just 0.7 ppts (or 26% of the total) of that decline while the Propensity effect subtracted the remaining 2.0 ppts. In Queensland it would appear that a weak labour force (perhaps with other factors at play…see above) has been the main driver to lower PR.
We also consider the Cairns region and found that Trend PR had fallen over the decade by 5.9% to just 61.6%. In Cairns the Demographic effect caused 1.7 ppts (29% of the total) of the fall while the Propensity effect caused the bulk (4.3 ppts) of that decline. Although the nominal impacts to PR are greater in Cairns it is worth noting that, as a proportion, the two effects had similar impacts at both State and regional levels.
When looking at Townsville, where the labour market has been particularly weak in recent times, we see Trend PR fell from 70.3% to 61.6% over the decade. However, here we see the Propensity effect having caused 6.9 ppts of the fall while the Demographic effect is responsible for just 1.8 ppts (20% of the total) decline. It would appear that the relative impact of the Propensity effect (perhaps caused by a weak labour market discouraging workers) in Townsville has been greater than at a State level.
We see that at a National level declines in Trend PR over the past decade have been caused exclusively by changes in the demographic make-up of the working population with PR shifts within age groups actually increasing PR. However, in Queensland it has been the Propensity effect that has been the major driver of lower aggregate PR levels. These shifts vary across regions but in the North we see a similar pattern at play, although on a greater nominal scale (perhaps not surprising given the tendency for regional data to be more volatile that aggregate State data).