Minerals Council of Australia Speech by the National Skills Commissioner

Image
NSC Speech

Introduction

I wish to acknowledge the traditional custodians of the land we are meeting on, the Ngunnawal people and pay my respects to elders past, present and emerging.

Nigel mentioned in that introduction that I’ll be addressing the role that data plays in evidence-based policy making. Before I do that, I want to start with a sense of the volatility seen in the labour market over the past year or so.

To give that volatility some context: in the early 1990s recession, the unemployment rate rose from a pre-recession trough of 5.8 per cent in December 1989 to a peak of 11.2 per cent in December 1992.

The unemployment rate didn’t return to that pre‑recession low until August 2003.

And it took around four years for employment (which peaked at 7,900,500 in June 1990) to return to its pre-recession level. 

Compare that to today, where the unemployment rate, at 5.5 per cent in April 2021, is now just a whisker above the pre-COVID rate of 5.3 per cent in March 2020, after reaching a peak of 7.4 per cent in July 2020.

Similarly, while the onset of the pandemic initially resulted in a dramatic fall in employment from March 2020 to May 2020 of 856,600; it has rebounded strongly with employment now 45,900 above the pre-pandemic level.

Of course, the volatility in the labour market has been accompanied by volatility and adjustment, in business models, supply chains and customer behaviours and preferences as well as closed borders.

It’s against that backdrop that I’d like to discuss the National Skills Commission’s (NSC) approach to assessing the economy’s skills needs, both today and into the future.

In doing so, this is less of a discussion on our forecasts per se, but more a discussion on:

  • how we can best bring together a range of data, tools and techniques
  • how to best use forecasts and mitigate the risks inherent in forecasts, and
  • the opportunities afforded by big data and nowcasting.

Weather forecasts

I suspect all of us today are users of forecasts. Indeed, I think I can go so far as to say almost all Australians use forecasts.

Perhaps not labour market forecasts, but weather forecasts.

Starting with a discussion on weather forecasts might be apt, given the storm that engulfed us all a year or so ago.

In 1861 the world’s first public weather forecast was published in The Times; a simple estimate of fine, fair, rainy or stormy weather for two days ahead.

Over the following 160 years weather forecasting evolved from a simple ‘single-answer’ prediction, to a more nuanced chance of what could occur (‘a 70 per cent chance of rain’).

When I prepared the final text for this presentation a few days ago – the Bureau of Meteorology was forecasting a 50% chance of rain today, with rainfall in the range of 0 to 4 mm.

As the Bureau notes in a blog on their website [Right as rain: How to interpret the daily rainfall forecast - Social Media Blog - Bureau of Meteorology (bom.gov.au)] this rain forecast reflects two pieces of information.

The first is that there is a 50% change of more than 0.2mm of rain over the 24 hours from midnight to midnight.

The second piece of information is that there is a 50% chance of receiving 0 mm or more, and a 25% chance of receiving 4 mm or more.

So not only do we have a probably of rain, there is also a range that encompasses the most likely set of outcomes.

Without wishing to push an analogy between economic and weather forecasts too far – nor invite comments on relative accuracy – there are of course some similarities.

The extent of precision, the degree of uncertainty, and the question of how to mitigate the risks inherent in using any set of forecasts.

And with economic forecasts, there is the question of knowing exactly where you are at any one point in time.

Current economic and labour market data – our starting point for a forecast – can be revised over time. Or we can get different observations for what might be the same variable – the census, as an example, does not necessarily show the same data as the labour force survey around the same time.

This should not surprise – the two are conceptually different. Still it points to a ’starting’ point challenge in granular labour market forecasting: where exactly am I right now?

At least when I get home today, I’ll know if I got rained on or not.

 

The NSC’s labour market forecasts

With that in mind, let me outline our thinking on how to approach the task of skills forecasting – noting that a global pandemic is likely to make that task more difficult than it might have already been.

The NSC has undertaken modelling to help understand the nature of the jobs and skills recovery from COVID-19. The purpose of this modelling is to examine the impact on occupations, industries and skills as we move further through recovery. By examining a range of scenarios, we can see what might be common across different recovery paths and where the differences might lie.

Importantly, this modelling also helped inform some of our thinking about the sorts of structural change we might expect to see – and have already seen – as a result of COVID-19. The strongest conclusion we drew around that question of structural change was that trends that had been evident prior to the pandemic could be accelerated, rather than seeing whole new trends emerge.

We released the results of that modelling in December last year in a report called: The shape of Australia’s post COVID-19 workforce.

And with that key conclusion about structural change in mind, we have been able to recently release five‑year industry and occupational based employment projections. These projections show anticipated growth in employment across around 250 occupations over the period to November 2025.

The objective in releasing such projections is not to proclaim with absolute certainty that employment of:

  • aged and disabled carers will increase by 54,700 or 24.7%
  • that employment of registered nurses will increase by 46,500 (15.6%), or
  • that software and applications programmers will see their ranks swell by 46,100 or 30%.

That said, those are our best estimates. However, the use of such forecasts ought to go beyond such a literal interpretation and focus on the big shifts and dynamics at play.

In that regard:

  • Our projections also suggest that employment growth will be strongest across skill level 1 occupations typically requiring a bachelor degree or higher (here we project employment to rise by 523,100), followed by skill level 4 occupations (typically commensurate with a certificate level 2 or 3) where we expect an increase in employment of 233,700.
  • We also find STEM-related occupations are likely to see percentage growth in employment of around double non-STEM occupations.

As former RBA Governor Glenn Stevens put it in a 2015 speech[1] around economic forecasts and the keen interest displayed in changes in the RBA’s own forecasts:

“the far more important question is whether we have recognised and understood the big forces at work. Even if we cannot predict the outcomes with great accuracy, an understanding of these forces ought to help us get policy responses roughly right.”

With that thought in mind, some of the big picture dynamics we can infer from our own five‑year employment projections are:

  • the ongoing shift toward services industries and sectors (and within that the care sector in particular)
  • an ongoing shift toward higher skill level occupations (underscoring the importance of a post-secondary education)
  • the ongoing robust percentage growth in employment across STEM occupations, and
  • the challenging employment outlook for routine manual occupations.

 

The Australian Skills Classification

Traditionally, when we talk about skills shortages or skills forecasts, the focus is on occupations, rather than skills themselves. An example of that are our own five‑year employment projections and the numbers I just cited.

This is partly because detailed labour market data released each quarter by the ABS examines employment by industry and employment by occupation. To be sure, it’s a good proxy for skills.

But there is no regular set of data that examines skills per se.

The recent release by the NSC of the Australian Skills Classification has the potential to change that.

It means for the first time we can start to think about not just the range of jobs across the economy, but the skills embedded in those jobs.

The beta release of the Australian Skills Classification includes skills profiles for some 600 occupations.

It includes 10 core competencies that are used in all occupations, 1,925 specialist tasks required by particular occupations, and 88 technology tools.

Those 1,925 specialist tasks can be grouped up into 279 skills clusters, which can in turn be grouped up into 29 skills families.

Grouping individual skills into clusters gives us a new and unique way of looking at the labour market. Skills that are like one another are clustered together – if you can do one task in a skills cluster, you can likely do the others.

When it comes to assessing skills needs across the economy, the skills classification can make it possible to view the portfolio of skills at an economy wide level.

As an example, our five‑year employment projections tell us that the occupation to see the greatest increase in numbers employed will be aged and disabled carers.

Our mapping of those occupational based projections across to the skills classification tells us that of the 279 skills clusters the largest increases in hours worked will be:

  • undertake food service activities (reflecting in part a bounce-back from still depressed levels on account of the pandemic)
  • communicate and collaborate
  • provide customer service and communicate information.

The clusters with the fastest growth are likely to be:

  • test computer or software performance
  • resolve computer application or system issues
  • develop or administer testing routines or procedures.

One does wonder if a little more “testing of computer or software performance” might result in a little less “resolve computer application or system issues”, but that might be a question for a different speech.

Returning to the question at hand, it is this skills focus – and the ability to think about the portfolio of skills the economy might need into the future – that offers another opportunity to mitigate the risks inherent in occupational based skills forecasts.

It does this, in part, by drawing out linkages and implications that might not be readily apparent from occupational based forecasts.

An example of this is the importance of ‘communicate and collaborate’. This broad skills cluster – based on our projections – is expected to see both rapid growth in percentage terms and also a robust increase in the number of hours. That in turn reflects the widespread importance of these skills in a range of occupations – something an occupational based lens doesn’t easily draw out.

We’ll be releasing more work on this in the coming months.

 

Where am I?

I mentioned earlier that one of the challenges associated with economic and labour market forecasting is the challenge of knowing exactly where we are right now.

Economic data, such as the national accounts, can be revised years after the original event.

Detailed labour market data, such as employment by occupation and region, are only available from the five‑yearly census.

We simply don’t know outside of the census how many electricians are employed in the Shoalhaven – Illawarra region, for example.

To help address that – and to also provide a better understanding of where we are right now – the NSC is in the process of developing monthly estimates of employment by occupation and region.

These data will provide estimates for over 350 occupations across more than 80 regions – over 30,000 individual series.

This is the first step in using big data and nowcasting to provide a clearer and more detailed assessment of labour market trends and, over time, indicators of skill shortages.

We hope to release our first experimental employment by region and occupation estimates in the next month or so.

Glenn Stevens observed in 2015 that:

“In my 35 years as a maker, observer and user of forecasts, I think forecasts have improved. Part of that has come from learning more about how economies work. But a lot of it, I suspect, has come from what could be described as improved ‘now-casting’: finding ways of assimilating a host of disparate pieces of information to judge more accurately what the economy has been doing in the very recent past. (That's probably where ‘big data’ potentially has some use.)”

 

While I think it’s highly unlikely that we’ll be forecasting in excess of 30,000 individual data series – this is not an exercise in claiming we will know exactly how many plumbers Bega will need in 10 years’ time – understanding developments and trends at disaggregated level and more frequently than that provided by the five-yearly census will be a major step forward.

I should also acknowledge here the former Governor opened a speech in 2011 with a discussion on the use of weather forecasts in aviation – I am indebted to the use of that device in my talk here today.

 

The Skills Priority List

It’s important to acknowledge, however, that we cannot be solely and exclusively driven by just data. Data can be noisy, and different sets of data can from time to time conflict.

As the Treasury noted in last year’s budget: “The Government’s macroeconomic forecasts are prepared using a range of modelling techniques including macroeconomic models, spreadsheet analysis and accounting frameworks. These are augmented by survey data, business liaison, professional opinion and judgment.”[2]

From the perspective of the work the NSC is doing on skill shortages, survey data, business liaison and judgement – as well as forecasts and projections – are all part of the mix.

As a specific example, the NSC is in the process of developing a Skills Priority List (SPL). The list will outline the occupations that are currently in shortage as well as their expected future demand.

Evidence informing the composition of this list will encompass:

  • labour market data – including forward looking elements and forecasts
  • employer surveys – which we are in the process of expanding
  • stakeholder consultation with representative bodies and federal and state government, and
  • a qualitative assessment where we will seek to bring other information to the table – such as workforce development plans or government priorities.

The SPL will form the backbone of the NSC’s labour market advice around skilled migration, and training and employer incentives, for example. The first list will be ready and published around the middle of this year.

 

Measuring outcomes

Of course outcomes matter. And by outcomes I’m not just thinking about forecast or skills list accuracy.

I’m thinking about the outcomes from training.

Knowing the outcomes is a crucial piece of the skills puzzle. And it enables us to consider the question (for example) of ‘what qualification should someone enrol in’ from a very different perspective.

The VET National Data Asset (“VNDA”) that the NSC is developing in conjunction with the ABS will provide a robust evidence base on the employment and further study outcomes from VET through linking VET activity data with a range of outcomes-related data sets also held by government.

The VNDA will be developed in a staged approach:

  • The first stage of VNDA is now in place. Total VET activity data has been linked with taxation, social security, higher education and census data to assess economic outcomes for specific qualifications.
  • The quality of data available through the VNDA will allow outcomes measures to be risk-adjusted, taking into account the diversity of students, courses and providers within the sector.

As just the first stage of output, knowing the labour market outcomes achieved by qualifications over time – taking into account the nature of the student cohort – will be very powerful information in helping to assess the performance of qualifications in the labour market.

Do graduates get jobs; do they get an income uplift following the completion of training; and are they getting jobs in the sectors most in need of VET graduates are just a few of the questions we will be able to answer using these linked datasets.

It will be incredibly useful information that complements the labour market and skills needs analysis work of the NSC.

 

Data and responding to skills shortages

While we can bring a strong data lens to the question of skill shortages and the economy’s current, future and emerging workforce skills needs; data cannot – of itself – resolve all the economy’s skills needs or shortages.

Indeed, as the Productivity Commission (PC) noted in their review of the National Agreement for Skills and Workforce Development, data from the predecessors to the NSC “suggest highly persistent skill shortages in a range of occupations”.

At this point it’s probably worth explaining how we define a skills shortage.

Our working definition is that skill shortages exist when employers are unable to fill or have considerable difficulty filling vacancies for an occupation, or significant specialised skill needs within that occupation, at current levels of remuneration and conditions of employment, and in reasonably accessible locations.

There are therefore a range of factors, beyond the provision of formal training, that might result in skill shortages, and that might see those persist.

Some examples, drawing on both the PC’s review and my own thinking as well as NSC data are:

  • employers wanting more advanced soft skills or on the job experience, things that are difficult to provide through formal training alone
  • skill mismatches may occur geographically with there being a mismatch between where those seeking work are located and the work itself
  • the labour market itself may be too rigid
  • an employer might have a need for a highly technical or specialised skill which is emerging and hence might not yet be reflected in the training system, or
  • there may mismatches between the preferences of employers and potential employees.

These highlight, to my mind, the importance of flexibility in labour markets, labour mobility, jobs themselves and the training system – as well as the importance of the insights offered by the NSC – if we are to effectively address skill shortages and prepare our workforce for the future.

 

Conclusion

My aim today was to provide some insights into how the NSC approaches the task of assessing the economy’s skills needs, both today and into the future.

I hope I have also given you some insight into how we intend to use not just a traditional occupational based lens, but to also transform that into a skills-based lens. 

That is, what are the skills that sit within the occupations we expect to grow – either in absolute or percentage terms – and what does that imply for the portfolio of skills the economy might need? Thinking about the portfolio of skills an economy might need provides, in my view, a practical way to mitigate the risks involved in forecasting the outlook for hundreds of individual occupations.

I’ve given a brief taste of our soon to be released estimate of employment by region and occupation – over 30,000 series – as an example of nowcasting and big data in action.

In doing all that, though, I hope I’ve also left you with the thought that the key – especially when it comes to looking forward – isn’t a rigid focus on a specific forecast or number, but a sense of what the big picture dynamics at work are.

It’s by understand these bigger picture dynamics, and by focussing on the portfolio of skills the economy might need in response to those bigger picture dynamics, that will enable us to best assess the skills needs of today and tomorrow.  

 

[1] ‘The Long Run’ speech – Address to the Australian Business Economists (ABE) Annual Dinner, 24 November 2015.

[2] Budget Strategy and Outlook Budget Paper No.1 2020-21.