The employment data in Australia has been lacklustre for over a year now. This characterization is based on below average growth in employment and persistent growth in unemployed persons. These measures have taken a turn for the worst over the last few months together with a continued decline in the participation rate. The importance of the participation rate is as a measure of labour supply – either a fall or an increase in the share of workers in the labour force will impact economic output, spending and government outlays and receipts.
The aim of this post is to ‘decompose’ the drivers behind the recent large decline in the Australian participation rate by looking at the contribution of changes in population share and the contribution of changes in participation. One is a broad measure of behaviour (participation within age groups) and the other is a broad measure of structural demographic change (the change in population share across age groups).
Using a simple methodology to measure the contribution of both changes in participation and changes in population share since the peak of the participation rate in late 2010, it’s clear that the decline in participation is not just due to one age group or to an ageing population. In fact, most age groups for males recorded a decline in participation, except for those aged 65yrs and over. But the participation rate of those 65yrs+ only peaked recently in May 2013, so if you look at this more recent time period, then the ageing population is also likely affecting the participation rate. There was also an equal contribution to the decline by both changes in population and changes in participation across age and gender groups. But there was a far greater contribution to the decline by males than females overall.
Does this decline reflect such a poor labour market that workers are discouraged from continuing to look for work? The popular belief (at Feb 2014) is that our ageing population is partly behind the recent declines in the participation rate in Australia, with some influence from cyclical factors.
The aggregate participation rate in Australia peaked at 65.2% in November 2010 and currently sits at 64.6% in January 2014 (trend data series). Using original data (which is the only data available for the age/gender analysis), the 3 month moving average participation rate peaks at a similar time – Dec 2010 at 65.6% and is currently 64.5%.
The long-term trend in the participation rate highlights that there have been several periods where the Labour Force Participation Rate (LFPR) has declined – during the early 80’s and the early 90’s are the two most notable periods. These two periods also coincided with recessions in Australia. There was also a decline in participation from the mid-90’s until the early 2000’s.
The most recent down trend represents thirty (30) consecutive months of either decline or no change in the participation rate (year on year basis). This is not the longest consecutive period of declines. The decline in the LFPR during the early 80’s lasted 34 months and 32 months during the early 90’s. The current economic climate in Australia can hardly be described as recessionary, and while it seems unlikely that poor job prospects are keeping people out of the workforce, the labour market has been weakening.
Changes in participation by gender reflect some broad cultural changes over time. The main one is increased female participation in the workforce. At the same time, male participation has declined. Note that both male and female participation rates have declined recently:-
The long-term charts are great, but honing in on the last 14 years shows a most interesting trend for male participation.
There was a period of time – from approx. mid 2003 until early 2008 when the male participation rate bucked its long-term trend and started to grow. Since then though, male participation has declined.
From an age perspective, the decline in participation has not been limited to one age group. Below is a comparison of participation within each age group since the peak in the aggregate LFPR in Nov 2010 and the current period.
Source: ABS. Note that the data is ‘original’. The ABS recommends not using original data due to the volatility. Given this is the only source of data, I’ve used a 3 month moving average for Dec 2010 and Dec 2013 to smooth out any volatility, rather than seasonally adjust the data.
The most affected group has been 15-19 year olds – whilst this has been part of longer term trend (participation in the labour force for this age group peaked back in Dec 1980 at 70.3%), the decline appears to have accelerated somewhat since 2008.
In fact, most age groups experienced a decline in participation during this period of time. There were three (3) exceptions – the participation rate for those 25-34yrs, 60-64yrs and 65yrs+ all increased.
At the same time there has been a significant shift in the population share among age groups – the most significant is the increase in the population share of those 65yrs+ – which grew by 1% point.
Two age groups have increased in population share during the Dec ’10 – Dec ’13 period – those 65yrs+ and those 25-34yrs. These two age groups are now the largest in the population. There is no doubt that ours is an ageing population – the share of 65yrs+ has been growing, whilst 25-34yrs has been declining until very recently.
There are a number of different ways to break down the contribution to the change in the LFPR. The methodologies use slightly different approaches, but directionally, result in the same conclusions. I’ve listed a range of different approaches in the sources section at the end.
I’ve selected a straight forward methodology, sourced from the Atlanta Federal Reserve:-
The two time periods that are the subject of this analysis are Dec 2010, when the aggregate LFPR peaked and the current data at Dec 2013. The objective of using these two dates is to isolate the factors that are contributing (participation and/or demographics) to the current decline in the aggregate participation rate.
As mentioned, this analysis uses original data sourced from ABS 6291.0.55.001. I’ve taken a 3 month moving average for Dec ’10 and Dec ’13 data using identical time periods in order to account for seasonality.
The second last two columns on the right (table 1 below) reflect equation (1) above – merely the population share multiplied by the participation rate for each age/gender group for each of the two time periods. The sum of which equals the aggregate LFPR.
The difference between Dec ’13 and Dec ’10 is -1.13% points – with males contributing -0.8% points and female -0.33% points.
Table 1 – Conversion to Weighted LFPR by Age
Source: ABS 6291.0.55.001 and The Macroeconomic Project
There are some important insights here:
- There has been a significant increase in the population share of those 65yrs+ over the last three (3) years of +1.1%pts, which was also matched by an increase in participation for both men & women 65yrs+
- There were declines in both participation and population share for men 35-44yrs and 45-54yrs – these two age groups had the largest negative impact on the aggregate participation rate
- There were also declines in participation and population share of those 15-19yrs and 20-24yrs (male and female)
- The groups that contributed to any growth in participation were 25-34yrs and 65yrs+ (male and female), as well as females 55-59yrs and 60-64yrs
This alone does not tell us the contribution of changes in participation and population share. Which brings us to the second part of the analysis.
Equation (2) is split into two parts because the movement in LFPR’s are influenced by changes in participation and in population share:-
- The change in the LFPR by each age group, multiplied by that age/gender group’s current population share – in other words, the contribution of changes in age/gender specific LFPR’s
- The change in the population share of each age group, multiplied by that age/gender group’s previous LFPR – in other words, the contribution of changes in population share
These are the results:-
Table 2 – Contribution to Changes in Aggregate LFPR
Source: ABS 6291.0.55.001 and The Macroeconomic Project
What does it mean?
- Over two thirds of the decline in the LFPR is due to 1) decreased participation by males (-0.5pt contribution) and 2) a decline in the male population share (-0.304pt contribution)
- Lower male participation in all age groups except 60-64 and 65yrs+ is contributing the single biggest negative impact to the participation rate – what is causing this?
- Males 45-54yrs had the single biggest negative impact on the LFPR of all age/gender groups, driven mostly by the change in population share, but also in lower participation
- Females 35-44yrs and 45-54yrs also had a negative influence, contributed mostly by a change in population share
- Males & females 65yrs+ had a positive contribution to the aggregate LFPR. Although population share grew, which would have a negative impact, the participation rate within the group during this time also increased, neutralizing the somewhat large increase in population share. This is confirmed by looking at ‘those not in the labour force 65yrs+’ which actually declined as a share of the 65yrs+ civilian population (between the same two periods).
A NOTE ON TRENDS IN LFPR’S
There is one important feature, or short-coming, of this analysis. It takes two discrete points in time to calculate the contributions to the change in LFPR’s, which tells you nothing of the trend before or in between those two points. The trend in the movement of LFPR’s provides a further layer of insight.
Of most interest to this analysis is the 65yrs+ age group. The participation rate for this age group actually peaked very recently, in May 2013 at 12.6% (this was the highest point since records started in Feb 1978).
Since May 2013, the participation rate for 65yrs+ has started declining (down -1.1%pts), but the important point is that the current participation rate is still above the starting point for this analysis – which is why this age group still shows a positive contribution to the aggregate LFPR. But the downward trend is now a caution. So since May ’13, the lower participation rate within this age group, together with its higher population share is now likely applying downward pressure on the LFPR.
The group with the largest negative impact on the LFPR was/is males 45-54yrs. Whilst participation in this group has been declining (the peak in participation was actually in April 1978 at 92.7%!), it did stabilize between Mar ’12 and Apr ’13. Since then there has been quite a dramatic decline (-1%pt) – is this pointing to greater cyclical influence now?
The second largest negative contribution to the decline in the LFPR was males 15-19yrs. There is a clear seasonal trend in this data, so I’ve been sure to reflect identical time periods in the pre/post comparison.
On the positive side, females 25-34yrs had the largest positive (upward) influence on the LFPR, due to both growth in population share and participation. The trend for participation (behaviour) appears to be growing, which will continue to apply upward influence on the participation rate.
CHANGES IN EMPLOYMENT BY INDUSTRY & GENDER
Given that males, generally, are having a greater negative impact on the LFPR, I wanted to dig a little deeper and see whether the greater contribution to a lower LFPR was due to changes at an industry level, specifically for example, manufacturing. This level of analysis isn’t definitive in the sense that a decline in employment equals a decline in the LFPR – it doesn’t. Certainly a large proportion of people who are no longer counted as “employed” will then be counted as “unemployed” (and therefore still counted in the labour force and by definition in the participation rate). The idea here is to merely identify any broad changes in employment by industry and gender to see if it helps to explain the underlying trends in LFPR’s.
On the whole, several key industries that have a high % of males employed have seen no, or relatively lower employment growth (except for mining) and industries that have a more equal gender balance or are dominated by female employed persons have seen higher employment growth, and more likely to be PT in nature.
There are a quite a few industries where the number of employed males has declined. These include Construction, Manufacturing, Info Media and Technology, Agriculture and Accommodation & Food Services. Some of these are also large employers of males (Construction, Manufacturing & Agriculture). At this stage though, these declines have been more than offset by growth in employment for males in industries such as Mining, Professional, Scientific & Technical and Public Admin & Safety. The degree to which there is a skills and geographic match between the decline and the growth industries will partly determine whether or not labour can/will move between these industries. That is unclear from this analysis.
The growth in males employed was much lower than the growth in female employed persons (over the same time period Nov ’10 v Nov 13) of +1.5% and +6% respectively. In other words, 2/3rds of the growth in employed persons were female. In the most recent employment data (Jan 2014), females now represent 90% of the latest annual growth in employed persons – this suggests that the situation has continued to deteriorate for male employed persons.
Data from the chart below highlights the change in employment by gender in major industries:-
Source: ABS 6291.05 and 6291.06
Industries where male employment has grown:
- Mining: most of the growth in mining was male employed persons of +72k and most of that growth was FT employment – this has clearly been the single most important industry for growth in employed males. The trend in employment in mining appears to be holding up (as of Nov 2013).
Professional Scientific & Tech Services: the growth was mostly all male employed persons of +36k and most of that growth was FT – again another important industry for male employment
Industries where male employment has declined:
- Construction; the largest employer of FT persons, of which 89% are male, declined by -16k employed persons, mostly FT and -14k of the -16k decline were males employed
- Manufacturing: the entire decline in manufacturing was male employed persons -47k persons, again, all FT. There has been no swap of FT for PT jobs in this industry.
Significant growth in employment has taken place in the Health Care and Social Assistance industry. Growth in employed persons in this industry was just over +129pk employed persons. Of that growth, +100k were female employed persons, and +80k were PT roles. This is now the single largest industry by employed persons in Australia.
CHANGES IN LONG-TERM UNEMPLOYED PERSONS
The weaker labour market conditions are showing up in a higher proportion of long-term unemployed persons. That is, a higher proportion of a growing pool of people. This is most evident for males, where the growth in the proportion of long-term unemployed (+52weeks since last FT job) has accelerated since April 2013.
This chart highlights a worrying problem in our labour market. Whether by choice, or by force, there is a growing proportion of people who have been out of FT work for over a year – and the total number of unemployed persons has been growing at the same time. It’s not unusual for males to represent the larger proportion, but the recent spike higher is a worrying feature. An identical trend is evident for those without FT work +104wks.
There are several concerning trends highlighted in this analysis. Firstly, lower male participation and change in male population share are the largest contributors to the decline in the aggregate LFPR. Data suggests that since early 2008, there have been shifts in industries such as Manufacturing and Construction that have likely affected male employment. These trends haven’t reversed and the weakness in the labour market, especially for males has persisted since early/mid 2008. At the same time, employment for males in industries such as Mining and Professional, Scientific and Technical Services has grown strongly – and whilst they might represent a smaller proportion of the labour force, the growth in the number of employed males in these two industries has outweighed the declines in other bigger industries. But given that LT unemployment for males has grown since mid/late 2008, it suggests that growth in other industries hasn’t been enough to reverse these trends in LT unemployment (or there is some kind of mismatch in skills/geography), especially recently. For the moment, Mining employment continues to be reasonably strong, contrary to popular belief. As the investment phase of the mining boom slows down over the next year, the extent of changes in employment growth/decline will become evident.
The other defining feature of the Australian labour force is the ageing population. This has not had the negative impact on labour force participation that most people think. But that is in the process of turning around and as the population share of those 65+yrs continues to grow and the participation rate for this age group now starts to decline, the 65+yrs age group will likely add significant downward pressure on the aggregate participation rate. As these trends continue, it will likely weigh more heavily on economic growth.
Hotchkiss, Julie L, Changes in the Aggregate Labour Force Participation Rate, Federal Reserve Bank of Atlanta, 4th November 2009 https://www.frbatlanta.org/documents/pubs/economicreview/er09no4_Hotchkiss.pdf
Fallick, Bruce & Pingle, Jonathan, A Cohort Based Model of Labour Force Participation, Federal Reserve Board, Washington DC 2007-09 http://www.federalreserve.gov/pubs/feds/2007/200709/200709pap.pdf
Van Zandweghe, Willem, Interpreting the Recent Decline in Labour Force Participation, Kansas City Fed, Economic Review Q1, 2012 http://www.kc.frb.org/publicat/econrev/pdf/12q1VanZandweghe.pdf