1.1 Introduction

The Skills Priority List (SPL) provides a detailed view of occupations in shortage, nationally and by state and territory, as well as the future demand for occupations in Australia. The list provides a single source of intelligence on occupations in shortage. While the SPL helps inform advice around the targeting of policy initiatives, it is important to note that it is not the only input into any such advice.

The list and occupation assessments are determined through extensive statistical analysis of the labour market, employer surveys, and extensive stakeholder engagement.

The National Skills Commission (NSC) engaged with a wide range of organisations as part of this year’s SPL consultation process. These included peak bodies, industry groups, professional associations, unions, regional representative bodies and major employers in the Australian labour market, combined with extensive consultations with federal and state and territory governments.

The use of various sources of evidence and stakeholder consultations ensures the SPL represents a comprehensive assessment of occupational shortages. The NSC thanks all participants for their engagement and their expertise. In particular, we thank state and territory agencies for their insights into the occupational shortages that are specific to their jurisdiction, including points of difference in their labour market when compared to the national picture.

This methodology will be revised and updated where required as further data sources and information become available. This will ensure the methodology continues to provide the most accurate assessment of occupations possible.

Scope and granularity

Occupations considered in scope for the SPL are:

• defined in the Australian and New Zealand Standard Classification of Occupations (ANZSCO) as Skill Level 1 - 4 occupations at the six-digit level, excluding occupations ‘not further defined’,¹ and
• occupations with an open and contestable labour market.²

Occupational assessments, where the data supports a robust assessment and there is evidence suggesting variation between metropolitan and regional locations, are made at a national,  state and territory level..

While data and stakeholder input on Skill Level 5 occupations are gathered by the NSC through employer surveys and can be provided during the SPL consultation process, these occupations are not considered in-scope for the SPL. This is due to these occupations having fewer barriers to entry and, unlike other in-scope occupations, they generally do not require significant post-school education and training.

The Australian Bureau of Statistics (ABS) defines Skill Level 5 occupations as having:

‘a level of skill commensurate with…AQF Certificate I or compulsory secondary education. For some occupations a short period of on-the-job training may be required in addition to or instead of the formal qualification. In some instances, no formal qualification or on-the-job training may be required’.

Shortage definition

A key component of the SPL is the assessment of occupational shortages. An occupation is considered to be in shortage when employers are unable to fill or have considerable difficulty filling vacancies. Or they cannot meet significant specialised skill needs within that occupation, at current levels of remuneration, conditions of employment and in reasonably accessible locations.

Based on this definition, the primary measure of an occupational shortage is the ability of employers to fill vacancies, i.e. the vacancy fill rate.

2.1 SPL Indicator Model

Overview

The SPL Indicator Model has been developed by the NSC to assist in the assessment of the current labour market for over 900 occupations at the Australian and New Zealand Standard Classification of Occupations (ANZSCO) six-digit level. It provides a valuable link to, and extension of, the targeted long-standing research undertaken on skill shortages and the Survey of Employers who have Recently Advertised (SERA).

The SERA employer survey undertaken by the NSC is unique as it can be taken as an estimate of current skill shortages and can be used as a direct measure of whether there is adequate supply to meet demand. The SERA provides a mechanism for testing the validity of secondary labour market indicators, and a way to sensibly combine them. If an indicator is predictive of SERA results, then it can be incorporated into a skills model directly.

By using SERA results as the benchmark for an occupation shortage, a robust model was produced that targets an explicit measure (the vacancy fill rate – that is, the percentage of employers who are able to fill vacancies for a particular occupation over the 12-month research period). It also allows precision about the expected accuracy of the indicator model and the relative strength of different variables in the model. The model produces an estimated fill rate, along with the level of certainty around the estimate, for all in-scope occupations at both the national and state and territory levels (where sufficient data are available). Lower fill rates imply a tighter labour market for that occupation and therefore more likelihood that the occupation will be in shortage.

The model will continue to be developed and tested further as additional or improved data sets become available. For example, the NSC will consider labour market nowcasting inputs or other additional data where available and relevant.

Method

The SPL Indicator Model is a binomial regression model that predicts vacancy fill rates.³

In the development of the model, many labour market indicators (around 190 variables overall) were tested to see if they were predictive of vacancy fill rates. Indicators were predominantly chosen for testing based on theoretical justifications and existing literature and research. A decision tree-based machine learning algorithm was applied to create a shortlist of labour market indicators that were most predictive of vacancy fill rates. This shortlist was used as the basis to determine the indicators that were included in the model. Through further testing, several indicators that were generally found to be predictive of fill rates in isolation but that were correlated with other indicators in the model were excluded. The final list of indicators used in the model includes those that are predictive of vacancy fill rates, with an interpretable modelled relationship with fill rates, that contribute most to improvement in model performance.⁴

The SPL Indicator Model uses the following variables to estimate the vacancy fill rate:

• Change in the number of average weeks spent looking for work for unemployed persons - An increase in the average number of weeks seeking work relative to the last three years is associated with higher fill rates.
• Internet Vacancy Index (IVI) - Occupations with a relatively large number of vacancies have lower fill rates. Additionally, occupations with a growing number of vacancies tend to have lower fill rates.
• Employment numbers - Smaller occupations tend to have lower fill rates. Additionally, occupations that are growing tend to have lower fill rates.
• Unemployment rates - Occupations where unemployment rates have increased relative to a year ago tend to have higher fill rates.
• Change in average salaries - Occupations with rising salaries tend to have lower fill rates.
• Job qualification requirements – Occupations that tend not to require at least a bachelor’s degree or above tend to have lower fill rates.
• Visas granted relative to occupation size - Occupations where the number of visas granted is large relative to the total number employed tend to have lower fill rates.
• Study diversity for occupation - Occupations where employees come from a more diverse range of fields tend to have higher fill rates.

Further detail on the indicators used in the model is included at Attachment B.

Where indicators are available at lower geographical levels, a credibility approach has been used.⁵ For example, the adopted state-level indicator is a weighted average of the state value and national value.

During development, performance of the indicator model was tested by fitting the model using a random 70% of the modelling data and comparing model predictions for the remaining 30% (‘test data’) to what was observed. Figure 1 shows the range of predicted values from the indicator model for observations corresponding to historical SERA estimates. The chart orders predictions from low to high and divides into 20 equally sized groups. It shows the 5% lowest-ranked occupations have a predicted fill rate of 43% and the highest 5% have a predicted rate of 83%, a significant spread. SERA results have then been overlaid to show the strong alignment with the models predicted results.

Figure 1 – Observed versus predicted fill rate, by predicted value (data as at early 2020)

3.1 Five year employment projections

Each year, the NSC produces employment projections by industry, occupation, skill level and region for the following five years. These projections are designed to provide a guide to the future direction of the labour market. However, like all such exercises, they are subject to an inherent degree of uncertainty.

The projections have been derived by using a blend of two techniques, namely:

• best practice time series models that summarise the information that is in a time series and converted into a forecast. The time series projections are made by combining forecasts from autoregressive integrated moving average (ARIMA) and exponential smoothing with damped trend (ESWDT) models, with some adjustments made to take account of research undertaken by the NSC and known future industry developments, and
• the Computable General Equilibrium modelling work the NSC undertook in 2022 to better understand how structural changes in the economy may affect employment growth. This work was undertaken with the Victoria University Centre of Policy Studies,.

The nature of both exercises means that there will be differences between them. The NSC believes that having a range of techniques assessing future skills trends is one way of mitigating the risks of error inherent in any forecasting exercise.

The Australian Government’s forecasts and projections for total employment growth from 2021-22 onwards, as published in the 2021-22 MYEFO is consistent with employment growth to November 2021.

3.2 Replacement rates

The NSC produces replacement rates by occupation at the national level, to provide an estimate of the total replacement demand resulting from flows of workers exiting a job irrespective of the inflows to employment over the same period. This information represents an indication of likely job opportunities in the occupation. This can occur when:

• people leave their occupation to take up a new job
• leave an occupation as a result of retrenchment
• move into retirement, or
• leave the labour force.

The replacement rate method is based on data from the ABS Participation, Job Search and Mobility Supplementary Survey, which is conducted throughout Australia in February of each year as a component of the monthly Labour Force Survey, and employment data from the Labour Force Survey. The survey provides a comprehensive and consistent dataset on a person’s experiences relating to job search, job change and labour market participation, and enables insights to be gained on detailed occupations in the ANZSCO classification of occupations.

When illustrating the growth in a particular occupation, all the people who leave a specific job need to be replaced. Replacement is equal to the number of people who were employed in that particular occupation at the end of the reference year, minus the number who did not change jobs , as well as the employment growth over the year.

The labour market is underpinned by significant dynamism. ABS Labour Force data show that over 4 million job vacancies are created through turnover each year (noting that this varied during years when COVID-19 lockdowns affected the labour market). Replacement rates are an indicative measure that should be used and interpreted with caution.

3.3 Future Demand Rating

Future demand for each occupation is based on combining the projected future employment growth (from the five-year employment projections produced by the NSC) and the replacement rate.

Each occupation receives a Future Demand Rating of either strong, moderate or soft. Occupations are categorised based on whether their future demand ratio (future demand divided by number employed in the occupation) is outside one standard deviation from the mean (weighted by occupation size, based on 2018 figures; a year that is judged to be a moderate labour market use for comparisons).

A proportional, rather than absolute measure of future demand is used to assign ratings to each occupation. Proportional measure is used as it puts all occupations on a comparable scale. A small occupation doubling in size should be classified as having ‘strong’ demand, even if a larger occupation with 10% growth will add more jobs in absolute terms.

The projected employment growth and replacement rate provide the Future Demand Rating at the 4-digit ANZSCO level, which is apportioned to the 6-digit ANZSCO level by allocating it proportionately to occupation size. This approach assumes the same Future Demand Rating ratio for all 6-digit level occupations within the same 4-digit level occupation. The size of the 6-digit occupation can be used to multiply the proportion to obtain the corresponding absolute predicted demand.

The Future Demand Rating is only available at the national level due to limitations in the availability of more granular data used in this assessment. Where required, national ratings can be used as a proxy for state/territory ratings.

Incorporating additional sources

In addition to the five-year employment projections and replacement rate, other sources of reliable and relevant information regarding future demand may be available. These sources are incorporated where appropriate by adjusting the final rating as needed. For example, the Future Demand Rating may suggest an occupation has moderate future demand, however, where high quality forecasts or workforce planning exist that suggest uncaptured high demand, the Future Demand Rating can be amended to reflect this.

As outlined in the previous sections of this paper, additional sources considered in the Future Demand Rating include:

• stakeholder input and advice from employers, peak, industry, government or regional bodies gathered through the stakeholder surveys or through submissions and/or meetings with the NSC
• occupation and industry assessments/information developed by the NSC (including the Reliance on Temporary Visa Holders and Training Funding Indicators)
• studies, assessments and reports by federal government departments
• state and territory government occupation and industry findings
• relevant skills needs or workforce planning information from government departments or industry groups
• industry activity statistics, projections and reports
• media articles, and
• presentations and speeches by industry groups, unions, workforce associations and government.

Where new and emerging data sets or methods become available to provide additional insight that are otherwise not captured, they are reviewed for incorporation into the future demand assessments.

Once all available sources of information have been considered, an initial determination is made regarding the Current Labour Market Rating and Future Demand Rating. Internally, quality checks are completed to ensure occupations have been assessed consistently.

The draft SPL is then tested with state/territory and federal agencies to verify the findings and to seek any additional context/evidence. State and territory governments are consulted on the findings relevant to their region. Similarly, federal government agencies are consulted on the occupations that are deemed relevant to their portfolio (e.g. Department of Health is consulted on health related occupations).

Where feedback does not align with the draft findings, the NSC conducts further consultation with the relevant agencies as needed. Any additional information obtained through this process is considered in the final SPL ratings.