LMIC economist Michael Willcox and principal researcher Lorena Camargo presented at the 2022 Canadian Economics Association conference, held at Carleton University. In their presentation, they analyzed the many difficulties in identifying and measuring shortages, including unclear definitions, data issues and missing dimensions (for example, the geographic scope of measures).

1. New approach to
measuring labour
market imbalances
2022.06.03 Michael Willcox and Lorena Camargo

2. 2
Outline
Labour Market
Imbalances:
• A bit of context
• Labour vs Skills
shortages
Common
Approaches and
Limitations
• What we’ve seen
thus far in our
research
LMIC’s Balance
Index
• What we’re
working on
• Aiming to resolve
some of the
limitations of
identified
approaches
Presentation

3. 3
Media outlets express concern over shortages in
Canadian Economy
• Historically low national unemployment rates
• Labour shortages major concern
• Businesses report difficulty finding workers

4. 4
Job vacancies
point to a tight
labour market at
the macro level
Job vacancies (bars – left axis) and vacancy/unemployment
ratio (line – right axis)
Canada, 2016-Q1 to 2021-Q4
Vacancies vs Unemployment Ratio

5. 5
A note on imbalances: Labour vs. Skills Shortages
Shortages: unfilled jobs/
persistent job vacancies
Labour
Shortages
Skills Shortages
Lack of candidates
Lack of candidates with
the right skills
nb. For a discussion on the difference between labour and skills
shortages see LMI Insight Report no. 3

6. 6
Common Types
of Approaches
COMPOSITE SURVEY-
BASED
RATIOS
TO IDENTIFY LABOUR
MARKET IMBALANCES
E.g. Recent
Labour Market
Asessment
Model (ESDC)
E.g. Employer
based surveys
E.g. Beveridge
Curve

7. 7
1 2 3
Unclear definitions
• No accepted definition
for shortage/surplus
• Difficult to identify
labour supply
Data Issues
• Hard to find local and
granular data
• Indicators disagree
• Arbitrary thresholds
contribute to difficult
to interpret signals
Missing dimensions
• Detection vs
diagnosis
• Severity
• Longevity
• Geographic scope
Common Limitations

8. 8
• 200,000 new, unique online
job postings per month
• 2,038,500 online job postings
in 2020
• Postings are matched to 4-digit NOC
and are available by P/T/ER
• Work requirements are extracted and
cleaned based on ESDC’s Skills and
Competencies Taxonomy
A COMPLIMENT: ONLINE JOB POSTING DATA
Skills and
other work
requirements
Vicinity Jobs Data

9. 9
9
Toward a proxy for excess labour demand
which is timely and granular
Ratio of Job Postings for Occupation i in t vs (t-1)
Ratio of Potentially Available Labour Supply for
Occupation i in t vs (t-1)
Balance
index_(i,t) =
=
Note:
• “Potentially available labour supply” is:
• ( Employment in Occupation i ) + ( Unemployment associated with Occupation i and
similar occupations )
• Unemployment counts are therefore duplicated across i
• t is a 3-month moving averages
• Growth is year-over-year change in 3mma

10. 10
1 2 3
Specific and detailed
• Uses 4-digit NOC
• OJP data
• Timely, granular
and local
• Linked to skills
Labour supply better
reflected
• Represents gross
supply
• Helps to avoid data
suppression
Highly generalizable
• Use of growth rates
normalizes level
effects from high
turnover
• Can be applied to
occupations,
industries and
geographic regions
Balance Index
Advantages

11. 11
2019: Potential labour supply and job posting growth relatively balanced

12. 12
2020: Job postings fall rapidly, potential labour supply increases rapidly

13. 13
2021 and 2022: Job postings increase rapidly, potential labour supply not keeping pace

14. 14
Job postings rising across all jurisdictions,
but JVWS-measured vacancies rising faster

15. 15
Balance Index Comparison: OJP vs Vacancy Data
How do OJP and Vacancies compare?
Lots of
missing data
from JVWS at
the 4-digit
NOC level
JVWS
distribution
more skewed to
the right –
towards tighter
labour market

16. 16
QUESTIONS REMAINING: NEXT STEPS:
REMAINING QUESTIONS AND NEXT STEPS RE:
BALANCE INDEX
• Why are we seeing differences
between OJP Data and JVWS
data?
• Are we properly accounting for
labour supply?
• Is there a way to make the index
more diagnostic?
• Adding wage element,
see if that changes the
story?
• Checking with OJP and
Employment/
Unemployment data to
gauge how detailed we
can get
• Strengthening the skills
element

17. 17
Questions?
Michael Willcox
Michael.Willcox@lmic-cimt.ca
Economist
CONTACT
CONTACT
Lorena Camargo
Principal Researcher
Lorena.Camargo@lmic-cimt.ca

18. 18
Appendix

19. 19
Data structure overview
• Employment and online job posting counts for detailed occupations (4-digit
NOC codes) from LFS and “Vicinity Jobs”
• Unemployment counts for detailed occupations at those in same
educational category and broad occupational category
Occupation Broad
Occupational
Group
Education
level
required
Employ-
ment
Unemploy-
ment in
group
Potential
Labour
Supply
Code Title
4411 Home child care providers Occupations in
education, law and
social, community
and government
services (group 4)
High school
and/or some
on the job
training (level C
or D)
23,666 5,250 27,750
4412
Home support workers, housekeepers
and related occupations
64,833 5,250 73,000
4413
Elementary and secondary school
teacher assistants
121,166 5,250 133,750
4423
By-law enforcement and other
regulatory officers, n.e.c.
9,083 5,250 15,500