Craig Holmes

1
Polarisation, mobility and segmentation in the
labour market

• Craig Holmes
• ESRC Centre on Skills, Knowledge and
  Organisational Performance,
• University of Oxford

2
Labour market segmentation

• Labour market segmentation theory developed as a
  departure from traditional models of labour
  supply and demand in the 1960s and 1970s
• LMS suggests it is possible to identify parts of
  the labour market between which mobility is
  severely or entirely restricted
• These restrictions are related to factors other
  than individual skills or abilities
• Dual market primary and secondary sector
  distinguished by wages, security, prospects for
  promotion and training investment
• Initial employment matters ? workers becoming
  trapped

3
Labour market segmentation

• The initial literature did not find significant
  empirical support.
• Mayhew and Roswell (1976) looking at mobility
  between three labour market segments in the UK
  over the past working life of employees in 1972.
• Segments were defined by occupation and status
  within their jobs.
• Allocation of each occupation-status pair was
  based on the authors' own judgment, intention of
  creating segments of the lowest possible
  mobility.
• Mobility matrices derived from this method show
  significant mobility between segments for many
  individuals.

4
Job polarisation in the UK

• Polarisation hypothesis (Goos and Manning 2007)
• Price of computer capital has fallen since late
  1970s
• Computer capital replaces labour engaged in
  routine tasks
• Non-routine tasks may be complementary to
  computer capital (e.g. management, skilled
  professionals)
• Result growth in non-routine occupations due to
  changes in demand (complementarities) and supply
  (displaced routine workers)
• Routine occupations found in middle of income
  distribution
• Non-routine occupations found at top and bottom
  of distribution
• Managers, skilled professionals at the top
• Non-routine service occupations at the bottom
  e.g. hairdressers, cleaners

5
Polarisation and segmentation

• Obvious overlap between the primary and secondary
  segments and growth occupations
• Individuals tend to move short distances within
  the labour market in terms of job quality.
  Declining middling occupations reduces options
  for transitory upward steps to better
  occupations.
• Hence, a hollowed-out labour market could create
  two segments with limited mobility between them.

6
The labour market and segmentation

• Jobs defined by occupation, industry and skill

Industries
Occupational groups increasing in quality
Levels of skill or expertise within each
industry-occupation
7
The labour market and segmentation

• Example 1 craft union model or occupational
  labour markets

Industries
Primary segment
Occupational groups
Secondary segment
8
The labour market and segmentation

• Example 2 internal labour markets

Industries
Primary segment
Occupational groups
Secondary segment
9
The labour market and segmentation

• Example 3 polarisation

Industries
Primary segment
Occupational groups
Middling occupations
Secondary segment
10
The labour market and segmentation

• A hollowed-out labour market has
• Fewer middle jobs for low wage workers to move
  into
• Increasing competition for those that remain.
• Significant upward mobility may either be slower,
  or require much more difficult and sizeable
  leaps.
• Before embarking on a study of mobility using
  longitudinal analysis, it is important to
  understand the ways the polarisation phenomena
  has or has not manifested within a dataset that
  can be used for analysing working life mobility

11
Job polarisation in the UK an assessment

• Holmes, (2010), SKOPE research paper No. 90
• Looks at single cohort from National Child
  Development Study between 1981 (aged 23) and 2004
  (aged 46).
• Replicates the Goos and Manning methodology for
  our NCDS dataset
• Finds growth in high wage and low wage
  occupations, decline in mid-range occupations,
  proxied by 1981 wage
• Evidence of routinisation driven employment
  changes

12
Job polarisation in the UK an assessment

• However, wage distributions exhibit little
  evidence of polarisation
• Most jobs still fall in the middle of wage
  distribution
• How can these two observations be reconciled?
• Existing evidence relies on a strong assumption
  that wage structures have remained constant over
  the past three decades
• Changing wage structures, due to the associated
  changes in supply and demand of different
  workers, may have led to a new type of middling
  occupation

13
Job polarisation in the UK an assessment

• Change in employment share of wage deciles.
• Initially highest and lowest paid occupations
  grew more than the middle earning occupations

14
Job polarisation in the UK an assessment

• Resulting wage distributions are important
• Absent of other effects, a polarising labour
  force should be observed as in the diagram below

15
Job polarisation in the UK an assessment

• Changing distributions from NCDS cohort (hourly
  and weekly, full-time workers)

16
Job polarisation in the UK an assessment

• Econometric methods for analysing changes
  accurately
• Descriptive method (see Holmes, 2010) change in
  employment at each (log) wage percentile
• Polarisation illustrative example

17
Job polarisation in the UK an assessment
18
Further research

• This suggests polarisation may not be as
  clear-cut a mechanism for creating labour market
  segmentation
• However, still suggests several issues for
  further mobility analysis
• Destinations of displaced routine occupation
  workers
• Are they able to move upwards?
• Difference between occupational and wage mobility
• Experience of new entrants compared to existing
  workers
• Are new entrants more polarised?
• Do they experience different patterns of
  occupational and wage mobility
• Role of skills in both cases

19
Further research

• Destination of displaced routine workers
• Model four occupational categories
  (professional, managerial, routine and service)
• Separate out transitions from routine occupations
  caused by routinisation from those caused by
  career advancement
• Similarly for transitions between routine and
  service occupations
• Empirical strategy
• Define four occupational categories (SOC, SEG)
• Calculate transition probabilities in NCDS (1981
  2004)
• Apply to larger sample of 23 year olds in 1981
  (e.g. LFS) and compare to a counterfactual
  occupational structure (e.g. 46 year olds in LFS
  1981)

20
Further research

• Destination of displaced routine workers
• Initial results
• Probability of transition to professional depends
  on having a degree, rather than starting
  occupation
• Need to breakdown upward movers by upskilled and
  non-upskilled

Initial structure, 1981, 23 year olds Pre-routinised structure, 46 year olds Post-routinised structure, 46 years olds
Professional 8.4 9.5 13.2
Managerial 10.1 12.6 25.6
Routine 69.3 66.6 41.1
Service 12.2 11.3 20.1
21
Further research

• Different experiences of later cohorts
• Autor and Dorn (2009) declining occupations are
  getting older
• Continue using cohort approach - make use of
  later cohort study (1970 British Cohort Study)
  for comparison
• Differences at entry more polarised as older
  workers hold on to positions in declining
  occupations. New entrants driving cross sectional
  polarisation?
• Different mobility opportunities compared to NCDS
  cohort?