ILLUSTRATION: HANNAH ROSENBERG, LMIC
Table of Contents
Key Findings
- The identification of skills shortages requires measuring and comparing the unmet demand for skills with the supply of skills. The current suite of survey instruments used to capture demand and supply, however, are not designed to collect data related to skills or other work requirements such as language.
- Enhancements to existing surveys, subject to further examination and assessment, could improve our understanding of the unmet demand and supply of bilingual workers but are unlikely to yield much in the way of practical results.
- Online job postings, albeit limited to the demand side, offer an efficient alternative with some advantages:
Granularity
The number of online job postings listing bilingualism can be easily grouped by detailed occupation and location, as well as other work requirements. For instance, among the six bilingual corporate sales manager job postings in December 2020 in Saint John, interpersonal skills was the 2nd most sought after work requirement.
Timeliness
Data are collected weekly and released monthly. For instance, in February 2020, there were just over 13,000 online job postings asking for bilingualism.
Localness
Observations at the 4-digit NOC are available for small geographies.
Coverage
Data on public administration jobs are captured, which accounts for a large share of bilingual job offerings.
Trend analysis
Calculating shares within occupations at the 4-digit NOC level over time is possible. For instance, between 2018 and 2020, the share of job postings for Metallurgical and materials engineers asking for bilingualism increased by 14 percentage points.
Levels of bilingualism
This report documents and analyzes online job postings for which bilingualism is specified as a work requirement. Future work in this area could disentangle these with further refinements, such as whether being bilingual is a requirement or an asset.
- Despite these advantages, some considerable limitations also exist. In particular, online postings data only reflect demand and therefore can only signal the potential for shortages. Data also tend to be skewed towards occupations in certain industries and regions, as well as by firm size and educational requirements. In addition, it is not possible to know the extent to which certain requirements listed in a posting are truly required for the position. Finally, the data is proprietary and some knowledge of how to analyze the data is required.
Introduction
Skills shortages have long been a concern for policy makers1. The lack of certain skills among the working population reduces Canada’s competitiveness and hampers economic growth. Even amid recently elevated unemployment rates, employers report difficulty finding qualified skilled workers to fill key vacancies. Among these are positions requiring the capacity to communicate effectively in both official languages.
English and French provide a fundamental characteristic of the Canadian identity. The importance of language rights is clearly recognized in the Constitutional Framework. Unmet needs for bilingual workers can negatively affect the ability to offer essential public services, such as education and health, in an individual’s language of choice. In other cases, a lack of bilingual capacity can dampen the services provided by non-profit organizations and limit our access to foreign markets and capacity to cultivate important economic partnerships (see Box 1).
1 The report is in response to a request from the Intergovernmental Network on the Canadian Francophonie. In December 2020, the Council on the Canadian Francophonie Ministers mandated the Network to collaborate with expert organizations in the field to explore the tools needed to establish a national portrait of unmet needs for bilingual workers.
| Box 1: Census of population and bilingualism
According to data from the most recent census (2016), more than one fifth of Canadian workers were bilingual. This accords with a recent study conducted by the Conference Board of Canada on behalf of the Association des collèges et universités de la francophonie canadienne [Association of Colleges and Universities of the Canadian Francophonie]. Most bilingual Canadian workers live in Quebec (60%), but are also very present in New Brunswick, where they represented 38% of workers. In the other provinces, the rates vary between 4% and 12% of the workforce. In total, the 2016 census recorded 4.2 million bilingual workers. Outside of Quebec, this population was 1.6 million, or about 10% of the total workforce. Census data from 2016 also indicate the sectors of activity where bilingual workers are concentrated. Nationwide, bilingual workers were employed primarily in health care and social assistance (10.3%), retail trade (10.5%), education services (9.4%), public administration (8.8%) and professional, scientific and technical services (8.1%). |
Identifying the language requirements of jobs and the availability of workers who can meet those needs, however, is limited by the available data. As a result, it is challenging to estimate the true impact of language shortages in Canada, which in turn obscures the implementation of appropriate policy solutions. In this LMI Insight Report, we assess the current data and its capacity to identify potential shortages of bilingual workers. We then examine the feasibility of addressing these gaps through novel approaches.
First things first: Labour shortage or Skills shortage?
There is an important distinction to be made between a labour and skills shortage.
When deciding which data are needed to identify shortages of bilingual workers, it is important to distinguish between a labour shortage (inadequate supply of workers, regardless of whether they are bilingual) and a skills shortage (available workers are not bilingual).2 The two are distinct labour market phenomena with unique solutions and policy interventions.
In simple terms, a labour shortage exists when the number of workers that firms want to hire is greater than the number of qualified3 workers willing and able to work at the prevailing wage. Empirically, they are identified by measuring and comparing the unmet demand for labour (typically estimated through job vacancies) with the potential supply of labour (typically estimated through unemployment) (Green et al., 2001). A skill shortage exists if employers are unable to find workers with the specific skills and knowledge they seek among the pool of available candidates. They are identified by measuring and comparing the skill needs of a position relative to the skills acquired by an applicant.
Depending on the nature of the shortage, the policy response should differ. For instance, to address labour shortages, some industry leaders have called for renewed efforts to increase the supply of labour through immigration and by incentivizing labour market participation, notably of under-represented groups. To address skills shortages, others have called for renewed efforts to develop workers’ skills, notably social and emotional skills. Consequently, when discussing and examining shortages, it is important to bear these distinctions in mind, as the policy responses will vary according to underlying causes of the issue.
2Of course, the overall skills match between a job seeker and an open vacancy is likely to be determined by more than the language requirements.
3Qualified in this context refers to the minimum educational and/or licensing requirements that an individual must possess to gain entry to an occupation. There are no formal qualifications around language, which is classified under the knowledge domain.
Traditional survey data does not capture language requirements
Our ability to identify whether a skills shortage exists lies in our capacity to measure and compare the unmet demand for skills with the supply of skills. Unmet demand, typically measured as the number of job vacancies, is estimated via employer surveys. In Canada, the primary survey used for this purpose is Statistics Canada’s Job Vacancy and Wage Survey (JVWS). Potential supply of labour, usually measured as the number of unemployed persons, is estimated through household surveys. Unemployment levels by occupation are based on an individual’s last held job. In Canada, the primary survey used to measure unemployment is the Labour Force Survey (LFS), although employment data is also captured in the Census of Population and the Survey of Employment, Payrolls and Hours (SEPH).
Although the existing suite of data sources capture information on labour demand and supply, they were not designed to capture information on the skills or other work requirements of jobs (Table 1). The Census of Population is the only source to collect data related to language use. Specifically, respondents are asked to identify their knowledge of English and French; knowledge of non-official languages; Mother tongue; the language(s) spoken most often at home; and the language(s) used most often at work.
However, there are some limitations to using the census for this purpose. First, simply identifying persons who declare they use French and English at work does not imply that their position or occupation is necessarily bilingual. Second, census does not directly capture specific skills or other work requirements besides language. Data are available at the occupation level, and occupations can be linked to sets of skills in a static way (discussed below), but the census does not ask respondents to identify their skills. Finally, the census is conducted every five years and data come available approximately two years later. The inability to determine whether a position is truly bilingual, lack of other work requirements, reduced frequency and wide time lag make census data unsuitable for identifying shortages of bilingual workers. At best, census data can be used indirectly to infer the possibility of a shortage on an infrequent basis.
Table 1: Measuring Skills and Languages in Traditional Survey Data
| Lens | Data source | Are general work requirements (including skills) available? | Is language available among the list of work requirements? |
| Supply | Labour Force Survey | No | No |
| Census of Population | No | Yes4 | |
| Skills assessments (e.g., PIAAC) |
Yes | No | |
| Unmet Demand | Job Vacancy and Wage Survey 5 | No | No |
| Employer-oriented surveys (e.g., CFIB’s Help Wanted) | No | No |
4 Respondents indicate whether they use a language at work, but this does not imply that the language is required at work.
5 There are also provincial surveys of a similar nature — such as the Alberta Wage and Salary Survey and Quebec’s Enquête sur le recrutement, l’emploi et les besoins de formation dans les établissements du Québec (EREFQ) — neither of which capture any information on skills or other work requirements.
Adapting and amending current sources will yield limited results as it concerns bilingualism
One approach to address the shortfall of skills-related information within the current suite of surveys is to create a classification of skills (i.e., a skills taxonomy) and attach it at the occupational level. This is what the O*NET system in the United States does. It is an online database that provides standardized information about the characteristics of jobs — including skills — and workers in the US. Canada is working to create a similar system called OaSIS, which will associate a variety of descriptors, including skills from Employment and Social Development Canada’s (ESDC) Skills and Competencies Taxonomy, with each occupation (see Box 2).
When using a skills taxonomy, in-demand skills are identified as those associated with jobs where employment is growing. The drawback of this approach stems from the fact that the skills related to each occupation are static – within a given occupation, they do not vary by geography, industry, or even time. More fundamentally, these classification systems do not currently capture bilingualism.
| Box 2: The Occupational and Skills Information System (OaSIS)
ESDC is 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 (OaSIS). For those skills and other job descriptors identified as “important,” a level rating will be assessed on a scale from 1 to 5. OaSIS will contribute to meeting the needs of a diverse user base by presenting labour market and other occupational information that can help users make sound decisions related to career choices, curriculum planning and advising. Once completed, the system can serve to inform occupational similarity analysis, job task analysis and skills transferability analysis, as well as to provide more macro-level trends on skills demand. Census data from 2016 also indicate the sectors of activity where bilingual workers are concentrated. Nationwide, bilingual workers were employed primarily in health care and social assistance (10.3%), retail trade (10.5%), education services (9.4%), public administration (8.8%) and professional, scientific and technical services (8.1%). |
Another potential solution that could be explored is to add questions to the JVWS and the LFS to capture the language requirements of vacancies and the language knowledge of individuals (and what language is used at work), respectively. Doing so would, in theory, allow for the estimation of language requirements related to unmet labour demand and supply.6 The results that could be obtained, however, are unlikely to be sufficient. First, while there is some information in both surveys at the 4-digit occupation and sub-provincial levels, the relatively low prevalence and geographic concentration of minority-language populations and bilingual vacancies limit considerably the insights that could potentially be obtained with respect to bilingual shortages. Second, the JVWS does not measure vacancies in provincial, territorial and federal public administrations, which constitute a considerable share of bilingual jobs.
Nevertheless, it may be worthwhile exploring whether adding questions of this nature could help improve our overall understanding of this issue. This is particularly relevant given that the JVWS has the capacity to capture information on long-term job vacancies, defined as positions for which recruitment efforts had been ongoing for at least 90 days – a more refined measure of unmet demand. Adjusting the LFS or JVWS in this manner would require a more in-depth assessment of the possible outcomes and an examination into feasibility issues, such as sample size and response burden. It might be the case, for example, that a periodic supplement may be more beneficial than permanently lengthening the surveys.
6 There are other limitations to consider when using vacancies and unemployment as measures of unmet demand and supply, respectively. Not only are vacancies always present in the labour market, but when faced with difficulty hiring, employers may ask employees to work longer hours or they may recruit internally, rather than posting an open position. Similarly, when estimating supply, using an individual’s previous occupation may not accurately reflect the occupation for which they are trained or for which they are currently looking.
Can we do better? Insights from online job posting data
An alternative solution to estimate unmet demand is to use online job posting data. In the past few years, technological advances have made it possible to leverage data from online job postings to complement official statistics for labour market analysis. In addition to the large volume of data they provide, online job postings are available in real-time by detailed location and are collected from publicly available websites, which helps to improve our access to more local, granular, timely labour market information (LMI). Moreover, these data include detailed information on the skills and other work requirements, including language, demanded by employers, and these are not defined at the level of an occupation. As a data source, online job postings thus address many of the limitations associated with the other data sources previously discussed.7
Real-time data on the number and share of bilingual job postings
With the current data obtained from Vicinity Jobs,8 it is possible to estimate the counts of postings within an occupation and region that list “bilingualism” as a work requirement by different levels of aggregation. We observe, for example, that over two million job openings were posted online during 2020 across Canada (2,038,810). Of these, 110,296 listed “bilingualism” as a work requirement.
This can be further broken down by province and territory. In 2020, Nunavut9 and New Brunswick had the greatest share of postings listing bilingualism as a work requirement relative to the total number of postings within each province (24% each; see Table 2). On the other hand, only 1% of job openings posted online in British Columbia and Saskatchewan during this time asked for bilingualism.
7 There are caveats to proxying job vacancies with online job postings. See LMI Insight Report no. 32 and LMI Insight Report no. 36 for more details.
8 Vicinity Jobs is a Canadian company based in Vancouver that deals with big data analytics and internet search technologies. Each week, data are collected from thousands of French and English websites across Canada, yielding approximately 200,000 new, unique online job postings per month. LMIC has contracted with Vicinity Jobs to obtain this data and make it available through various public-facing tools and reports.
9 In most cases, we can interpret “bilingualism” in a job posting to mean French and English. This may not always be true, however. For job postings in Nunavut in particular, bilingualism may not strictly refer to French and English.
Table 2: Online Job Openings Listing Bilingualism as a Work Requirement in 2020, by Province
| Province or Territory | Total number of job postings | Number of postings with Bilingualism | Share of postings with Bilingualism |
| Nunavut | 992 | 236 | 24% |
| New Brunswick | 36,651 | 8,692 | 24% |
| Québec | 589,438 | 95,420 | 16% |
| Ontario | 638,047 | 45,342 | 7% |
| Manitoba | 39,384 | 2,645 | 7% |
| Prince Edward Island | 10,076 | 524 | 5% |
| Yukon | 1,726 | 85 | 5% |
| Nova Scotia | 51,837 | 2,180 | 4% |
| Northwest Territories | 2,838 | 115 | 4% |
| Newfoundland and Labrador | 15,980 | 637 | 4% |
| Alberta | 211,969 | 4,220 | 2% |
| British Columbia | 371,106 | 4,634 | 1% |
| Saskatchewan | 68,354 | 827 | 1% |
Analyzing trends over time – albeit with a rather short history
In Table 3, the 10 occupations with the largest total increase for postings requiring bilingualism from 2018 to 2020 are displayed.10 Of these, medical laboratory technicians and pathologists’ assistants top the list. In 2020, there were 424 total job postings for medical lab technicians and pathologists’ assistants listing bilingualism as a work requirement. In 2018, only 168 postings asked for bilingualism, representing an absolute increase of 256 job postings.
Table 3: Greatest increase in Postings With Bilingualism as a Work Requirement, 2020
| Occupation | Number of postings listing bilingualism as a work requirement in 2020 | Change in number of bilingual postings between 2018 and 2020 |
| 3212 – Medical laboratory technicians and pathologists' assistants | 424 | +256 |
| 7514 – Delivery and courier service drivers | 1,539 | +245 |
| 4411 – Home child care providers | 592 | +174 |
| 3411 – Dental assistants | 394 | +132 |
| 1513 – Couriers, messengers and door-to-door distributors | 187 | +97 |
| 3211 – Medical laboratory technologists | 167 | +83 |
| 4012 – Post-secondary teaching and research assistants | 475 | +79 |
| 6331 – Butchers, meat cutters and fishmongers – retail and wholesale | 146 | +69 |
| 9421 – Chemical plant machine operators | 85 | +63 |
| 4152 – Social workers | 544 | +61 |
10 It is important to be mindful of the short time period and the effects of the COVID-19 pandemic on job vacancies during 2020 when interpreting these data.
Looking at the largest changes in bilingual postings between 2018 and 2020 as shares of total job postings within each occupation, casino occupations has grown by the biggest margin — a 54 percentage point difference (see Table 4).
Table 4: Increase in Shares of Postings Listing Bilingualism as a Work Requirement, 2018 to 2020
| Occupation | Number of postings listing bilingualism as a work requirement in 2020 | Change in shares of bilingual postings between 2018 and 2020 (percentage points) |
| 6533 – Casino occupations | 30 | +54 |
| 9447 – Inspectors and graders, textile, fabric, fur and leather products manufacturing | 61 | +34 |
| 9212 – Supervisors, petroleum, gas and chemical processing and utilities | 27 | +14 |
| 2224 – Conservation and fishery officers | 31 | +14 |
| 2148 – Other professional engineers, n.e.c. | 27 | +13 |
| 2142 – Metallurgical and materials engineers | 30 | +12 |
| 7534 – Air transport ramp attendants | 33 | +12 |
| 9463 – Fish and seafood plant workers | 92 | +11 |
| 2153 – Urban and land use planners | 42 | +10 |
| 3211 – Medical laboratory technologists | 167 | +9 |
One key advantage to using online job posting data is the ability to analyze work requirements at greater levels of localness. We can, for example, identify the number of postings listing bilingualism in Ottawa during December 2020 (727) to determine that the greatest number (34) occurred for “Other customer and information services representatives” (see Table 5).
Table 5: Understanding Bilingualism Alongside Demands for other Work Requirements, select data for December 2020
| Geography | Top occupations for which postings listed bilingualism | Number of postings listing bilingualism as a work requirement (change from previous month) | Top 2 other work requirements (by frequency) | |
| Ottawa | 6552 – Other customer and information services representatives | 34 (+9) | Customer service
Communication skills |
|
| Edmonton | 6421 – Retail salespersons | 5 (+0) | Customer service
Sales |
|
| Saint John | 0601 – Corporate sales managers | 6 (+5) | Sales
Interpersonal skills |
|
Possible additional levels of specificity
While this approach offers many advantages, there are also some limitations. First, online job posting data provide estimates for unmet demand only. Without an equivalent measure of supply, it is difficult to conclude the existence of a shortage definitively. As with wage or employment growth, if demand for bilingual workers grows faster than expected, it is possible but not certain that supply will lag, making the case for a potential shortage.
Second, current data from Vicinity Jobs only provides information on whether a posting lists bilingualism as a work requirement. It is a binary association between an online job posting and its requirements. To address this, the Labour Market Information Council (LMIC), in collaboration with its partner Vicinity Jobs, could develop a level indicator to identify a posting’s requirement for an official language (i.e., French/English) according to three levels: “required,” “asset,” or “not required” (see Table 6). From this information, we would be able to determine whether a job requires French and English, only English, or only French. A job posting where both French and English are required can be inferred to be bilingual, whether bilingualism is a stated work requirement or not. Moreover, one could identify job postings in which one language is required and the other is considered an “asset,” which for certain applications might be a more informative definition of bilingual job requirements.
Including official language-specific requirement levels will allow for greater accuracy and insight into their demand across Canada. For example, one could determine the share of postings for mechanical assemblers and inspectors in Ontario where French and English both are required versus those positions where French is an asset but not required. Or, one could calculate the share of postings for retail salespersons in New Brunswick where French and only French is required. This greater level of detail will help better identify possible shortages with respect to the bilingual workforce.
Table 6: Illustrative Examples of Data Possibilities
| Job ID | NOC | Location | Work requirements | English | French |
| 123456 | Plumber | Montreal, QC | Bilingualism | Required | Required |
| 123456 | Plumber | Montreal, QC | Oral Communication | Required | Required |
| 987654 | Software Engineer | Calgary, AB | Leadership | Required | Asset |
| 987654 | Software Engineer | Calgary, AB | SQL | Required | Asset |
Forecasting future skills shortages for bilingual workers
Identifying current shortages is helpful but crafting and implementing solutions takes time. Therefore, if we can predict the likelihood of a shortage developing in the future, appropriate measures can be taken to mitigate or prevent them from occurring. In general, there are three major approaches to forecasting the demand (not supply) of skills and other work requirements of jobs, with a possible fourth approach that is a hybrid of all three. The first and most common approach is to forecast the employment level of specific occupations (known as occupational outlooks) and link these to occupation–skill profiles (e.g., O*NET, OaSIS). This approach leverages existing employment level data (e.g., from the LFS) and applies one of several standard econometric techniques (e.g., ARIMA or macroeconomic structural models). Once generated, the forecasts are linked to the fixed set of skills associated with each occupation. Therefore, the skill composition of each occupation is fixed overtime. Lastly, as previously discussed, language is not among the skills currently available using occupational profiles, making this approach unsuitable.
A second approach relies on expert opinion to identify the future trajectory of key skills or occupations, which can then be used to predict the trajectory of a broader range of skills or occupations. The key drawback with this approach is that forecasts are only as good as the experts’ opinions. In addition, typical methods focus on a subset of skill drivers, such as automation, which do not include language. It also assumes that skills are statically tied to each occupation.
The third approach relies on forecasting future skill needs using online job postings. Traditional econometric forecasts can be applied directly to observed work requirements, so one could, for example, project if the demand for bilingualism within custom service occupations will increase based on historical data. Alternatively, machine learning techniques could be applied to predict future skills compositions. However, online job posting data has yet to be leveraged for robust skills forecasting largely due to several key data limitations. Some of these limitations could be addressed but the precise manner in which this could accomplished requires further research and consideration.
The Way Forward
The lack of an appropriately skilled bilingual workforce in Canada could have significant social and economic implications. Not only does it prevent minority language speakers from exercising their right to access services in the language of their choice, but it also hinders productivity by limiting competitiveness. Data limitations, however, prevent definitive identification and measurement of these shortages, necessitating the assessment of possible new solutions.
One potential approach would be to include questions on existing employer and household surveys to capture the language requirements of vacancies and the language knowledge of workers. And while a more in-depth sample size and response burden would be needed, these additions are unlikely to greatly improve our ability to understand the demand and supply for bilingual workers.
Online job postings may offer a robust, albeit limited, alternative. These data offer numerous advantages, including, but not limited to, the following: 1) timeliness: data are near real time; 2) localness: observations are available for small geographies; and 3) granularity: detailed categories are available by which data can be grouped, such as occupation, location and other work requirements. Indeed, multiple skill and other work requirements can also be assessed in addition to language. Additionally, online job postings could be analyzed in a manner to provide greater detail on whether bilingualism is required or an asset and eventually could potentially be used to forecast the demand for skills using standard econometric techniques.
Finally, it is important to be mindful that online jobs postings represent a measure of demand only and thus can only signal the potential for shortages. However, despite this and other limitations, online job posting data could help close the gap in information needed to better understand current and future workforce developments related to the demand for bilingual workers in Canada.
Acknowledgements
This edition of LMI Insights was prepared by Anthony Mantione of LMIC. We would like to acknowledge the valuable input of Julie L’Allier, Sami Bibi, Marc Delisle and Ingrid Ledrou (Employment and Social Development Canada); Steven Wald, Ray Gormley, and Helen Cranley (Ontario Ministry of Labour, Training and Skills Development); Jean-Pierre Voyer (NSAP); Syvlie Painchaud (CCFM); and Vincent Dale (Statistics Canada).
For more information about this report, please contact research@lmic-cimt.ca.