Overview
| Name | Canadian Occupational Projection System (COPS) |
| Author | Economic Policy Directorate (EPD) of Employment and Social Development Canada (ESDC) |
| Contact | SPPC-COPS@hrsdc-rhdcc.gc.ca |
| Timing | 10-year forecast, updated every 2 years |
| Outputs | COPS website
Datasets available on Open Government |
| Additional documentation | Information documents |
The Canadian Occupational Projection System (COPS) combines assessment of recent labour market conditions and projections of labour demand and supply to identify potential imbalances over the projection period. COPS is not a forecasting system but rather a method providing signals that, if trends continue as it has in the past, then current imbalances will continue, or new ones could emerge. Analysis is conducted for 293 occupational groups (since 2019), as a combination of the 500 4-digit National Occupation Classification (NOC) categories (details available here) and at the national level. Provincial estimates are partially developed for the demand side only and are shared with provincial governments on a confidential basis for further development.
Methodology
Recent Labour Market Conditions
For each of the 293 occupational groups and the 13 provinces, an analysis of labour market indicators is conducted. The process has two main steps: 1) data gathering and detailed analysis; 2) consultations with internal and external stakeholders.
Step 1: Data Analysis
Table 1. Indicators for Data Analysis
| Indicator | Details |
| Employment growth | Annual average growth over the previous 3 years |
| Average hourly wage growth | Annual average growth over the previous 3 years |
| Unemployment rate change | Change in the level of the unemployment rate over the previous 3 years |
| Unemployment rate level | Comparison of the unemployment rate of the previous 3 years with the 10-year average |
| Proportion of workers working overtime | Change in the proportion over the previous 3 years |
| Number of recipients of regular EI benefits | Helps validate the unemployment estimates |
| Ratio of regular EI beneficiaries to employment | Used to understand the level of “experienced” unemployed per actual employed |
| Number of job vacancies | Growth is preferred to level, which is not a proper indicator of labour market conditions |
| Job vacancies growth | |
| Unemployment-to-job vacancies ratio | Used to understand how many unemployed there are to fill a vacancy |
| EI regular beneficiaries-to-job vacancies ratio | Similar to the unemployment-to-job vacancies ratio but uses EI data; helps validate the unemployment data |
| Changes in the unemployment-to-job vacancies ratio | Used to understand if the ratio has declined (tighter conditions) or not |
| Change in the proportion of jobs that have been vacant for more than 90 days | An increase in this proportion should, in principle, be a signal of tighter conditions |
| Labour requirements | Defined as labour force (employment + unemployment) divided by total demand (employment + vacancies) = ration of the extra number of workers given total demand; estimate can be negative, signalling demand higher than supply or tighter conditions |
| Change in labour requirement estimate | Used to understand if labour market conditions are getting tighter |
| Level of new hires | Annualized value of workers employed for 3 months or fewer, extracted in the months of March, June, July and December |
| Change in level of new hires | Sometimes employment does not seem to grow, however, this does not mean that companies are not hiring; this measure is used to understand if labour demand is still there |
| Change in the average hourly wage of new hires | Used to capture if the hiring wages are growing stronger or weaker |
| Proportion of self-employment | Used to understand reliability of variables (e.g., hourly wages, proportion of overtime, new hires, unemployment rate, employment insurance and job vacancies); if the proportion of self-employment is high, then these variables are typically not reliable |
| Last historical year’s monthly unemployment rate | Used to understand if the unemployment estimate is volatile due to monthly variability and/or seasonality, or if the annual estimate is based on only a few months of observations |
| Employment growth of the industry where most workers in the occupation work (main industry for each occupation) | Used to understand if the labour market conditions are driven by a particular industry |
| Provincial distribution of employment in the last historical year | Used to understand which province or provinces drive the national condition |
The analysis mostly consists of a qualitative review of the indicators shown in Table 1 for each of the 293 occupations considered. The objective is to assess the most recent labour market conditions of each occupation. Signals of tightening or loosening labour markets in recent years are identified by comparing the indicator’s behaviour with the historical average as well as with averages from other occupations. For example, if the unemployment rate is “high,” the analysis consists of questioning what “high” means for that occupation. That is, is it high but still lower than in the past (signalling improving conditions)? To what extent is it lower/higher than in the past? Is it higher/lower than in other occupations? Is it high because it is a seasonal occupation? Given that the unemployment rate is high, do we have weak employment and wage growth, a lower number of new hires and vacancies, a higher number of EI beneficiaries, etc.?
Given the number of occupations to analyze, work is divided among the COPS team analysts. Each analyst looks at two groups of occupations: one for which they have primary responsibility and the second to validate the work of another analyst. In other words, each occupation is assessed at least twice, by two different analysts. Then, the whole COPS team makes an in-depth analysis and final assessment of each occupation. During these sessions, the priority is to concentrate on those flagged as "unclear," as well as those with imbalanced conditions (shortage or surplus).
Step 2: Consultation
Once the COPS team has finalized their in-house analysis, the results are shared with ESDC’s central, regional and provincial analysts regarding their local labour markets. This process usually triggers several changes in the assessments at the provincial level and, depending on the number of changes, can impact the national assessment. The results are also shared with some external stakeholders (via publicly available reports and sometimes direct contact).
The final output of the current labour market condition analysis is as follows:
- An assessment (surplus/balance/shortage) for 293 occupational groups, at the national and provincial levels, for the whole 3-year period (national results available here) (A).
Former Methodology
Up until 2017, the recent labour market conditions assessment was done with a 3-step ranking approach. First, three core indicators were constructed using employment, wage and unemployment growth used to derive weighted measures and assess their evolution over the last three years compared to the average among all occupations. Then a similar exercise was conducted to assess the evolution of the three core indicators during the last ten years. A comparison of the averages would lead to a systematic assessment of imbalances. Finally, the outcomes of the preliminary and secondary assessment were validated by a qualitative review of the core indicators in conjunction with additional indicators.
Future Labour Market Conditions
For each of the 293 occupational groups and each year of the projection period, expansion and replacement demand (retirements, deaths and emigration) are estimated and summed to the total number of job openings. Both estimates are primarily informed by the following demographic, macroeconomic and industrial scenarios.
Demographic projections: Statistics Canada develops population projections by gender and age group:3
- Population projections by gender and age group (results available here) (B).
An ESDC in-house model (see Table 2 for ESDC models) uses a time-series approach to project participation rates for every combination of age, sex, education and student status:
- Participation rate projections by gender and age group (results available here) (C).
- Participation rate projections by education level (results available here) (D).
Combining population and participation rates, the following is developed:
- Labour force projections by gender and age group (results available here) (E).
- Labour force projections by education level (results available here) (F).
Macroeconomic projections: The Conference Board of Canada develops a custom scenario using ESDC’s assumptions and based on Statistics Canada public data for various macroeconomic variables at the national level, including investment, consumption, government spending, unemployment (Z), overall GDP and productivity growth.
Industrial projections: Based on the macroeconomic results, the Conference Board of Canada develops custom projections for GDP, employment and labour productivity, typically using the Cobb-Douglas approach for 42 North American Industry Classification System (NAICS) categories:
- GDP by industry at the national level (results available ) (I).
- Productivity estimates by industry (J).4
The main purpose of the industrial scenario is to estimate the number of workers needed (employment growth) for a given growth in the level of production (GDP):
- Number of workers for 42 NAICS groups at the national level (results available here) (K).
| Model | Author | Objective |
| Recent Labour Market Conditions at the national and provincial levels | Labour Market and Skills Research Division (Economic Policy Directorate) | Assess recent imbalances at national and provincial levels |
| Canadian Occupational Projection System (COPS) | Labour Market and Skills Research Division (Economic Policy Directorate) | Assess potential imbalances at the national level |
| Recent Labour Market Conditions at the regional level | Policy Research Analysis and Geomatics (Labour Market Information Directorate) | Assess potential imbalances at the regional level |
| 3-year Employment Outlooks | Labour Market Information Directorate | Identify location of job creation (demand driven) |
Projection of Job Openings
Expansion Demand
Using an in-house model, the share of occupational employment within each of the 42 industries is estimated using Labour Force Survey (LFS) data. These shares are projected using time series methodology and applied to the projected employment by industry. This would yield employment by occupation and industry. The sum of all these estimates results in the total projected employment by occupation:
Replacement Demand
An in-house model is used to estimate the number of job openings due to replacing workers who vacate positions because of retirement, in-service mortality or emigration.
Retirement: In the COPS model, retirement is defined as a "final withdrawal from paid employment" for those aged 50 and over. The measures are constructed specifically to capture both voluntary and involuntary retirements. Despite a number of existing surveys and databases containing retirement information, there is no comprehensive dataset in Canada capable of providing reliable detailed information using this definition at the occupational level.
Estimates of historical retirements using both the LFS and the Longitudinal Administrative Database (LAD) are developed. The number of voluntary retirees is estimated using the "reason why stopped working at previous job" question in the LFS. Then, involuntary retirement levels within LAD are estimated as the sum of individuals over the age of 50 who are separated from a job and have remained unemployed for at least three consecutive years.
Voluntary and involuntarily retirement rates are pooled and retirement probabilities by age and gender are projected using the LAD data. The projection is a weighted average of three methodologies. The first-order autoregressive model (AR(1)) smooths the data using an HP filter. The second one bases retirement rates on cohort movements rather than age. In the third method, retirement is a fixed share by age only. Retirement levels are obtained by applying the projected retirement probabilities to the projected employment levels (K). Voluntary and involuntary retirement estimates are examined, weighted and pooled to generate final estimates of future retirement rates by age and gender.
The age distribution methodology is used to project retirement by occupation; the average age of retirement at the national level is about 61 years old. Accordingly, all employed workers 56 years and older might potentially retire in the next five years. Two elements have an impact on the pressures associated with retirements: 1) the age profile of an occupation and 2) the median retirement age of an occupation.
A potential retirement pool of workers is then obtained by computing the distribution of the median retirement age in an occupation within five years. An unconstrained potential retirement series is produced and then constrained to the total derived from the aggregate retirement model. This represents the number of individuals susceptible to retirement within a given year, generated for all occupational groups. The retirement estimates are then distributed between the 293 hybrid NOCs based on the share of employment associated with each. This yields the following:
- Retirement projections by occupation (results available here).
- Retirement rate projections by occupation (results available here).
Since retirement rates are calculated at the occupational level, they are applied to employment projection estimates at the occupational and industrial level in order to obtain a sum of all retirements within an industry (results available here).
In-service mortality: Age-specific death probability is constructed as the ratio of deaths to population. These are projected using a simple time-series regression with a constant, a trend and a lag value. These age-specific death rates are then applied to single-age occupational employment projections (K).
Emigration: Initially, estimates of the number of emigrants are obtained from Statistics Canada population estimates (emigrants – returning emigrants net + temporary emigrants). Emigration projections are obtained from the demographic projection (B). Subsequently, labour force participation rates for emigrants are determined. After applying the labour force participation rates to the projected number of emigrants, the estimated number of active emigrants is allocated among occupations.
Combining estimates for in-service mortality and emigration provides the following:
- Number of job openings due to emigration and in-service mortality by occupation at the national level (results available here) (L)
The total number of job openings is equal to the sum of expansion and replacement demand:
- Total job openings for 293 occupations at the national level (results available here) (M).
Projection of Job Seekers
Three streams of job seekers are projected: 1) job seekers from immigration, 2) job seekers from the education system and 3) other job seekers. In addition to new job seekers, the composition of occupational labour markets through mobility is estimated.
Supply from Immigration
Job seekers from immigration enter Canada each year (under any immigration class) as permanent residents and participate in the labour market. The share of immigrants as a proportion of the Canadian population is estimated and projected.
Applying this proportion to the projected total population (B) provides the number of new immigrants both active and inactive in the labour market. An average participation rate based on LFS data for recent immigrants is applied to estimate the magnitude of new immigrants joining the labour market upon entering to Canada. The estimated new immigrant labour force is then allocated to the 293 occupational groups using trends of historical shares based on census data for recent immigrants:
- Number of new immigrants seeking jobs by occupation at the national level (results available here) (N).
Supply from School Leavers
Job seekers exiting the education system (also called school leavers) leave their full-time highest-level-of-education programs (either as graduates or dropouts) to participate in the labour force. The projection of new entrants is based on the education-level composition of the labour force below age 34 (i.e., in the age group most likely to exit the education system and enter the labour force).
First, the labour force population aged 15-34 is divided into students and non-students based on historical LFS data. Then, the education level composition (i.e., the ratio of individuals in each education level) of the non-student labour force is calculated. The non-student education level composition is then applied to the projected labour force (E) to generate a non-student labour force estimate by age and education for each year of the projection period.
The change in the labour force for each cohort (single-age) from one year to the next generates the number of school leavers by year and level of education. For example, for school leavers with only a high-school diploma, estimates are estimated as:
Where NS_LF stands for “non-student labour force, and HS refers to “high-school”. This exercise is repeated for all ages and years, and then aggregated by year. School leavers are then allocated to occupations based on the historical shares of school leavers by level of education to occupation (from the LFS). The National Graduate Survey, which provides information on both occupations and education programs of post-secondary graduates, is used to control the share of new graduates entering management occupations. The results are as follows:
- Number of job seekers from the education system by occupation at the national level (results available here) (P).
Supply from Other Job Seekers
Other job seekers include the following:
- Net re-entrants: those who previously left the labour force and are returning
- Working students: those who look for work while going to school
It is assumed that the share of re-entrants and working students is fairly stable across time so it is used to balance the unemployment rate estimated by the macroeconomic scenario (Z) and the unemployment rate that will stem from the projection of job openings and job seekers. To obtain the unemployment rate projected by the components of the model (Q), the previous year’s (t-1) labour force is added to the current year’s (t) number of job seekers, minus retirements and deaths, to calculate the labour force for the current year. Similarly, employment for each year is obtained as the sum of the previous year’s employment (t-1) and expansion demand for the current year (t). An unemployment rate can then be calculated for each year. The other job seekers (Q) must be added to make this unemployment rate match the one from the macro scenario (Z).
Other job seekers are then allocated to occupations using projections based on the LFS:
- Number of other job seekers by occupation at the national level (results available here) (Q).
Net Occupational Mobility
As a final step, the composition of occupational labour supply (i.e., job seekers (R)) is adjusted to account for occupational mobility. For each occupational group, other groups are identified as being part of the same career path (i.e., typically an individual within an occupation has probably also worked in other related occupations throughout their career). The probability of moving to each occupational group is calculated using longitudinal occupational data from the Survey of Labour and Income Dynamics (SLID; this calculation may be redeveloped using the Longitudinal and International Study of Adults [LISA]).
For each occupational group, job seekers are then assumed to move to other occupations within their career path aiming at reaching a targeted unemployment rate for each occupation. The targeted unemployment rate is estimated using historical LFS data trends.
For each occupational group, the number of job seekers is equal to the sum of new immigrants, school leavers and other job seekers entering the labour force after adjusting for occupational mobility over the projection period (not by year):
Summing job seekers from immigration (N), from school leavers (O),from other job seekers (Q) as well as the net occupational mobility provides the total number of job seekers:
- Total number of job seekers by occupation at the national level (results available here) (R).
Identification of Imbalances
The final assessment, to identify imbalances, is done based on recent labour market conditions and projections of job openings and job seekers. This includes assessing recent labour market conditions and occupations estimated to have experienced an imbalance in recent years (surplus/balance/shortage) for the 293 occupational groups (A).
Future labour market flows are then analyzed. For each of the 293 groups, the projected annual average of job openings (M) and job seekers (R) as a percentage of their respective employment levels in the base year (e.g., 2018 for the 2019 COPS projections) are compared:
 |
Openings significantly higher than seekers |
 |
Openings similar to seekers |
 |
Openings significantly lower than seekers |
The final assessment combines both recent and future labour market conditions. A shortage or surplus is determined to exist only if a recent imbalance is expected to persist or a new one will develop.
| Future Labour Market Conditions | ||||
| Recent Labour Market Conditions (Z) | Openings significantly higher than seekers | Openings similar to seekers | Openings significantly lower than seekers | |
| Recently in shortage | Shortage | Shortage | N/A | |
| Recently in balance | Shortage | Balance | Surplus | |
| Recently in surplus | N/A | Surplus | Surplus | |
Final results are available for the overall projection period, not for each year:
-
- Total job seekers for 293 occupational groups at the national level for the entire projection period (results available here) (S).
Applications
COPS results are distributed and used in many ways; for example:
Individual Canadians
- Results are available through the COPS website for individual Canadians to make education, training, career and immigration decisions.
- Results from COPS are used to develop the ESDC’s 3-year outlooks, which in turn inform the Job Bank.
Employment Policies and Programs
- Results inform ESDC’s different policies, such as the Temporary Foreign Worker program and the Youth Employment and Skills Strategy program.
- Results also inform other federal department activities, including international trade agreement negotiations and Correctional Service Canada.
Immigration Policies and Programs
- Results are used by employers to inform the Labour Market Impact Assessment in Temporary Foreign Worker program requests.
- Results are also used to inform Immigration, Refugee and Citizenship Canada activities.
Education
Results are used by career counsellors.
Economic Development
Results are used by industry and professional associations, as well as other labour market institutions (unions, non-for-profit organizations, etc.).
Footnotes
1 These measures are estimated using the Labour Force Survey (LFS), the Job Vacancy and Wage Survey (JVWS) and the ESDC internal Employment Insurance (EI) dataset.
2 Annualized estimates are the monthly averages in one year of the indicators. However, this might hide certain characteristics of the occupations, such as seasonality. For example, fishermen tend to have high annualized unemployment rates since they are employed for half the year and unemployed for the other half.
3 Source: Statistics Canada, Projected population, by projection scenario, age and sex, as of July 1 (x 1,000), https://www150.statcan.gc.ca/t1/tbl1/en/tv.action?pid=1710005701
4 The results of this estimate are not fully available to the public, but can be reproduced by dividing the GDP by the employment projected estimates, which are made available to the public.