Searching for an OaSiS in the world of skills and occupation mapping
LMI Insight Report no. 43
June 2021
Table of Contents
Key Findings
- Employment and Social Development Canada, with support from Statistics Canada and the Labour Market Information Council is currently developing a database that maps skills to occupations, much like O*NET in the US, the UK Skills Taxonomy and Europe’s ESCO, to name a few.
- The development of this system draws on best practices from the different approaches assessed in previous LMI Insight Reports (e.g., job analyst approach, web scraping) and the lessons from various international examples.
- Focusing on the needs and requirements of a Canadian system, several features from worldwide examples stand out. First, recognition that using multiple lines of evidence (e.g., O*NET data, web scraped data and expert validation), enabled by comparisons of their complementarity, would improve the timely dissemination of occupational profiles. This would maintain the quality of information provided in the system as well as increasing its relevance to the Canadian context. Second, a series of “tagging” and “filtering” features would help Canadians navigate large data sets and improve the user experience.
- The Occupational Skills and Information System (OaSIS) will provide a framework of occupations and skills in the Canadian context. OaSIS will be constructed in two phases. The first phase will develop and validate profiles for almost 900 occupations. The second phase will see dissemination of information through a bilingual, searchable user interface.
Introduction
The Labour Market Information Council (LMIC) has been working jointly with Employment and Social Development Canada (ESDC) and Statistics Canada (STC) to describe jobs in terms of their skills requirements and other characteristics. First, a concept note was presented, outlining the anticipated approach towards mapping ESDC’s new Skills and Competencies Taxonomy to the National Occupational Classification (NOC) system. Next, certain key features and requirements for a Canadian system were explored. This included a review of the US O*NET system, focusing on the skills domains and the manual rating of skills by occupational analysts. Also included was an assessment of modern technologies for generating and/or updating occupational profiles — specifically, the large-scale collection of data from online job postings.
As Canada looks to develop its own occupational information system, it has the unique opportunity to study other international models for inspiration. In this LMI Insight Report, five such models are reviewed: O*NET (US), the UK Skills Taxonomy (UK), ESCO (Europe), JEDI (Australia) and Skills Framework (Singapore) to highlight their real-world data collection and methods for identifying the skills requirements of jobs.
Applications in Other Systems Worldwide
In previous work, three approaches for identifying skills requirements were noted: 1) surveys, 2) manual assessments by trained analysts, and 3) partially or completely automated techniques that rely on artificial intelligence. In practice, one or more of these techniques is used by five of the most widely known occupational data systems worldwide (see Box 1).
| Box 1: Five international occupational information and skills systems | |
|---|---|
 |
Occupational Information Network (O*NET)
The US O*NET database is an open access, widely used public repository providing worker and occupational information — such as skills requirements, work activities and job tasks — for 923 occupations. Data is organized according to a six-tiered classification system called the O*NET Content Model. Users can view data online from within searchable occupational profiles or download the full data in one of five file formats. |
 |
Classification of European Skills/Competencies, Qualifications and Occupations (ESCO)
ESCO is a system that classifies occupation information across EU Member States. It includes 13,485 descriptors for skills, competencies, tasks and knowledge requirements, with definitions for each. The purpose of ESCO is to improve communication between the education and training sectors in the EU labour market, as well as to facilitate data exchange between employers, educational providers and jobseekers, irrespective of language or country. |
 |
UK Skills Taxonomy
This collaboration between NESTA and the Economic Statistics Centre of Excellence (ESCoE), groups multiple occupations into the same profile to create skill clusters. There are 143 clusters, each containing information on the broad skills required to work in any of those occupations and includes the demand for each cluster. The UK Skills Taxonomy is an open access tool available for researchers; however, it is not yet used by the UK government to develop or provide a skills framework. |
 |
Jobs and Education Data Infrastructure (JEDI)
Designed for Australia’s National Skills Commission (NSC), JEDI is a data engine. Using machine learning and other techniques, it brings together multiple sources of traditional and real-time data on the supply and demand for skills and can help identify the transferable skills that workers have for other jobs. As part of the JEDI project, the NSC developed a data-driven Australian Skills Classification. Approximately 600 skills profiles currently exist for occupations in the Australian labour market, categorized according to the Australian and New Zealand Standard Classification of Occupation (ANZSCO) taxonomy. This information is accessible to users via Australia’s Job Outlook website. |
 |
Singapore’s Skills Framework
Under Singapore’s SkillsFuture initiative, the Skills Framework was developed to help improve information and therefore decision-making around education and training, career development and skills upgrading. The framework itself was developed by employers’ groups, industry associations, education institutions, unions and the government. It includes information on different sectors, occupation/job roles, skills and training information. The Skills Framework is tied to the government’s Industry Transformation Maps, developed for key industries to help address needs and improve relations between workers, employers and the government. Accordingly, it allows the framework to be informed by these key industries, including enterprise information gathered via survey templates on job roles and requirements. |
Building occupational profiles with survey data
UK Employer Skills Surveys, implemented by the UK government on a periodic basis, are used to develop the UK Skills Taxonomy. The surveys help identify challenges faced by employers in a given economic climate and can help identify skills needs, recruitment processes and training practices. As a result, they are valuable sources of information for gauging labour demand. A 2017 review of the UK Employer Skills Surveys found that they were valuable for identifying skills shortages, gaps and mismatches at a granular level; however, the costs involved in implementing the surveys were a notable downside.
As detailed in LMI Insight Report no. 31, O*NET also uses surveys as part of its data generating process. It does this using questionnaires sent to employers and job incumbents to measure and assess trends in skills demand to supplement the dataset. This helps expand information on an occupation’s tasks, knowledge domain, work activities and work context — information used by occupational analysts in the skill-ranking process. This information also identifies in-demand occupations so that occupational experts can update these profiles.
Getting to skills through job analysts
O*NET uses a predetermined set of 35 skill descriptors, identified by the O*NET prototype development team in the mid-1990s. These are then aggregated into seven mutually exclusive sub-categories. Occupational analysts assign skills to occupations using a rating system for the importance and associated level of skills and abilities. Ratings are driven by reviewing the occupation information from the O*NET system drawn from surveys of job incumbents in that occupation. This also reflects the primary reliance on surveys for O*NET’s development.
The information provided through ESCO is collected from a range of sources such as national classification systems, reports, research papers and job vacancies. Experts are then engaged to identify occupations sector by sector and analyze the knowledge, skills and competences relevant for them. A major challenge for ESCO is that there is little standardization in the development of skills clusters. A hierarchy system helps with navigating the large amount of information included.
Web scraping
Unlike ESCO and O*NET, the UK Skills Taxonomy does not rely on analysts to validate occupational profiles, for reasons of both cost and timeliness of updates. Instead, the system relies on AI technology to scrape job postings to gather skills information, although not all jobs are advertised online. LMI Insight Report no. 32 also highlights that online job posting data can be skewed towards certain industries, regions, firm sizes and educational requirements. It is also unclear how the UK Skills Taxonomy captures/categorizes unfamiliar language or new skills/competency entries. This can be a shortcoming of formula-based AI models. The UK Skills Taxonomy does, however, highlight that one key benefit of drawing from job postings is that skills are described in the language of employers and not academics and policymakers.
Meanwhile, JEDI utilizes web-scraped Australian job postings purchased from a data brokerage firm to supplement transposed US O*NET data and ensure that the Australian occupational context is reflected. One of the primary uses of job posting data is to ensure that Australian qualifications, credentials and educational requirements for occupations are captured. Further, near real-time scraped job posting data offers workers the possibility of updating their skills and occupational information as occupations evolve.
O*NET updates specific in-demand occupations on a constant basis. To do this, it relies on a mix of sources, including survey data from job incumbents, as well as occupational experts and web research. However, trawled employer job postings are analyzed and used to identify which occupations are in-demand and need to be updated first. This provides an automated indication of changing labour demand — albeit skewed towards online postings — that can be used in conjunction with, and to triangulate, manually collected information.
Knowing your user base
Users vary from workers looking for career transitions, students looking to enter the labour market, employment counsellors, vocational specialists, researchers and policymakers. All these users have different needs and preferences and require different types of information. For instance, the UK Skills Taxonomy provides information on the average salary of a skill cluster — including wage data, which is particularly important in career decision-making.
Australia’s Job Outlook website targets those looking to make a career transition. Every tool on the website is designed to help this user base, from career quizzes meant to provide ideas for potential careers, to the skills match tool that helps workers find potential careers that use the skills they already possess.
Complementary information in these data sets can include knowledge requirements as well as analytical information. ESCO, for instance, provides information on knowledge requirements to accompany skills for most occupations within a given industry.
Meanwhile, O*NET tags occupations expected to grow in the future (using a “bright outlook” tag), as well as those associated with the green economy (using a “green jobs” tag). These satisfy one of the primary goals of such an information system, namely to identify potential future skills demands and/or looming skills shortages.
Introducing OaSIS
ESDC, in collaboration with STC and LMIC, has been developing a Canadian system designed to house information on skills, occupations and other relevant sources of LMI under one roof. This system will be known as the Occupational and Skills Information System. Considering Canadian needs, OaSIS draws on the following key takeaways from the international systems above:
Recognition that using multiple lines of evidence (e.g., O*NET data, web scraped data, expert validation), enabled by comparisons of their complementarity, would improve the timely dissemination of occupational profiles while maintaining the quality of information in the system. It would also increase its relevance to the Canadian context.
A series of “tagging” and “filtering” features would help users navigate large data sets and improve the user experience.
Eventually linking skills and occupational information with other sources of labour market information (e.g., wages, outlooks, sectoral and regional considerations) would strengthen the system’s value to its diverse user base.
OaSIS will contribute to meeting the needs of Canadians by presenting labour market and occupational information that can help different users make sound decisions about career choices and curriculum planning. It will also lead to more informed advice on integration/reintegration in the labour market, shifting to new occupations, the right training for an individual’s skills gap, and whether a specific educational/training program will lead to gainful employment, amongst other outcomes. The information found in OaSIS will also serve as a rich source of data for the study of a broad range of labour market issues and interventions. For instance, once completed, the system could serve to inform occupational similarity analysis, job task analysis, skills transferability analysis, as well as to provide more macro-level trends on skills demand.
The Way Forward
Launched in March 2020 by assembling previously accomplished initiatives and steps, the OaSIS project should be completed by end of 2022. Currently the project is in the database development phase. Profiles for almost 900 different occupations are being constructed using detailed information from reliable sources. These will then be validated by expert stakeholders to ensure relevance to the current Canadian context. The following phase of the project will disseminate information through a bilingual, searchable user interface.
Acknowledgements
This LMI insight report Insight Reportwas prepared jointly by the staff of the Labour Market Information Council, Statistics Canada (Centre for Labour Market Information) and Employment and Social Development Canada (Labour Market Information Directorate).
For more information about this report, please contact research@lmic-cimt.ca.