Projecting UK employment by
ethnic group between 2012
and 2022
Paper presented to the British Society for
Population Studies annual conference,
University of Leeds, 7th September 2015
David Owen, Anne Green,
Lynn Gambin and Yuxin Li
Institute for Employment Research
University of Warwick
D.W.Owen@warwick.ac.uk
Background
 Project funded under Joseph




Rowntree Foundation “Poverty
and Ethnicity” research
programme – report at:
http://www.jrf.org.uk/publications
/projecting-employment-ethnicgroup-2022
Access to employment is an
important factor underlying
poverty
The poorest ethnic groups also
have the lowest employment
rates.
Those in work are more likely to
be in low-status and low-paying
industries and occupations
How is this pattern likely to
change over the medium-term
future?
Challenge
 Projecting economic activity and
employment by ethnic group
 This has not previously been undertaken
 It becomes possible because:
o detailed employment projections are
available from Working Futures 5 (CE/IER for
UKCES)
o ETHPOP (Leeds University) – provide the
first detailed projections of population by
ethnic group
 The analyses brought these two sets of
projections together, and estimated future
trends in labour market participation and
employment rates
Aims of the research
 To identify patterns of change in economic activity and employment
by ethnic group from 1992 to 2012 for people aged 16 to 64
 To project employment by ethnic group from 2012 to 2022
 To examine the changing ethnic division of labour by sector and
occupation
 To demonstrate the historical and future impact of the changing
occupational composition of labour demand upon the probability of
employment of different ethnic groups, and identify the implications
for the incidence of poverty by ethnic group
 To identify gender contrasts within ethnic groups
 To identify the likely differences in employment trends within the UK
by ethnicity over the period 2012 to 2022
The research had three elements
1. Analysis of historical trends in economic activity and
employment by ethnic group – using Labour Force
Survey data to create time-series data on labour market
participation and employment by population group,
ethnic group, region/nation, gender and age, together
with the explanatory variables used to estimate
regression models
2. Creating projections - of participation rates and
employment by ethnic group, age and gender, for the
nations and regions of the UK
3. Interpretation and dissemination – summarising the
results of the projections and implications for poverty
Analysis of historical trends
 A time series of labour market participation data was created from LFS






data - variables extracted: ethnic group, nationality, age, gender, labour
market status, industry and occupation of employment
Quarterly LFS data was available from Spring 1992 to Winter 2013, but not
all variables are available for the first few years
A challenge for this exercise was the change in occupational and ethnic
group classifications which occurred in 2001 and 2011
Therefore, an important part of this work was creating measures of
occupation, industry and ethnic group which were consistent over time
and across the nations of the UK
Time-series models of labour market participation rates and employment
rates were estimated for ethnic group, age, gender and region
Alternative specifications of these models were experimented with.
independent variables include individual characteristics which are key
factors influencing participation in the labour market, time, and region
Occupational trends, 1992-2013 (LFS)
BME ethnic groups
25.0
25.0
20.0
20.0
% of employment
% of employment
White ethnic groups
15.0
10.0
15.0
10.0
5.0
5.0
0.0
0.0
1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012
Managers, directors and senior officials
Professional occupations
Associate professional and technical
Administrative and secretarial
Skilled trades occupations
Caring, leisure and other service
Sales and customer service
Process, plant and machine operatives
Elementary occupations
1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012
Managers, directors and senior officials
Professional occupations
Associate professional and technical
Administrative and secretarial
Skilled trades occupations
Caring, leisure and other service
Sales and customer service
Process, plant and machine operatives
Elementary occupations
Key patterns of occupational change, 1992-2013 for
white and ethnic minority people
 Growth in:
 Professionals
 Associate professionals
 Caring, leisure & other service
 Decline in:
 Administrative & secretarial
 Skilled manual trades
 Semi- and unskilled
occupations
 Stability in:
 Managers and directors
(white)
 Elementary occupations (BME)
 Broad occupational trends




common to white and ethnic
minorities
Trends exaggerated for ethnic
minorities
Relative loss of jobs in skilled
manual trades and semi-skilled
occupations greater for ethnic
minorities
Share of elementary jobs
declined for white people but
not for ethnic minorities
Relative growth of professional
jobs faster for ethnic minorities
Aggregate employment and population
projections which the research is based upon
Working Futures 5
 Projections of employment by industry,





occupation and region/nation (created for
UKCES by IER and CE)
A detailed set of projections of the labour
market for the UK and its constituent nations
and regions from 2012 to 2022
Based on a macroeconomic model - includes
estimates of labour market participation rates
by gender
Employment projections are detailed, broken
down by industry (22 categories), occupation
(25 categories) and level of qualification (9
levels)
Estimates of employment are made for each
year from 1992 to 2022
Employment is projected to grow by 5.8%t
between 2012 and 2022
ETHPOP
 Projections of population by ethnic group
from 2001 to 2050 were created by the
Geography Dept. , Leeds University as part
of the ESRC UPTAP research programme.
 ETHPOP provides projections of population
change by ethnic group, age, gender and
geography (for regions and local authority
districts in England, Wales, Scotland and
Northern Ireland)
 There is a ‘central’ projection representing
the most likely outcome for population and variants
 For the population of economically active
age over a 10-year period, the main
influence upon variations in projections is
assumptions made about international
immigration and emigration
Creating the population data






ETHPOP population projections underestimated the ethnic minority population in 2011, primarily because they
failed to replicate the high levels of international net migration experienced after 2001
The researchers who created these projections argue that the average of the TREND and UPTAP-ER variants
provide the best indication of the trend over time in the population of individual ethnic groups and this average
comes closest to the 2011 Census of Population. They suggest adjusting the projections to match the Census data
in 2011 (which is regarded as the most reliable estimate of the population by ethnic group) and then apply the
projected trends in population to this base year
A database of projected population by ethnic group, gender, region/nation and five-year age group (within the
age range 16 to 64) for each year from 2001 to 2022 was created from the average of the TREND and UPTAP-ER
projections by:
Calculating the ratio of the projected population for year 2011+t (where t runs from -10 to +11) to the projected
population in 2011
Multiplying this against the 2011 Census population gives an estimate of the population in year 2011+t (for each
age group, gender and ethnic group). This step is repeated for each year in order to create a time-series for 2001
to 2022: the population for 2011 is thus the Census estimate
Some manipulation and estimation of data was necessary in order to create comparable data for all four nations,
because the design of Census of Population 2011 Detailed Characteristics Table DC2101 differs between England
and Wales , Scotland and Northern Ireland, mainly due to differences in ethnic classification between the three
Censuses. In Scotland and Northern Ireland, there is only one group for people of mixed parentage, compared to
four in England and Wales. Additionally, the White British and White Other ethnic groups are not distinguished in
Northern Ireland. These population groups were estimated for Scotland and Northern Ireland and the white
group in Northern Ireland.
Trends in UK working age population,
2012-2022
 The working age population from ethnic






minorities is projected to grow by 22 per cent
between 2012 and 2022
The ethnic minority population is projected to
grow in all age groups
The white population aged 40-49 is projected to
decline
The slowest projected rates of increase amongst
ethnic minorities in the working age population
are projected for age groups from 16-34 years
The numbers in the immediate pre-retirement
age group are larger in 2022, and more of this
age group are from ethnic minorities
The share of people from ethnic minorities is
projected to be largest for people aged from 25
to 44 in 2022, compared with the 20 to 34 age
range in 2012
The working age population is projected to
increase fastest in southern and eastern regions
of England, but to decline in northern England
and the devolved nations
Working Futures 5 UK projections, 20122022
 Overall employment is projected to
increase by 5.8% between 2012 and 2022
 The fastest rates of growth are projected
for Construction and Business and other
services
 The Primary and Manufacturing sectors are
projected to lose 5.5% and 6.8% of
employment respectively over the decade
 Employment is projected to grow in Caring,
leisure & other service occupations,
Managerial, Professional and Associate
professional & technical occupations and
to decline in all other SOC major groups
 Employment is projected to decline fastest
in Administrative & secretarial and in
Process, plant & machine operative
occupations
 Employment in London is projected to
grow fastest, at 8%
Regression modelling of labour market
participation and employment rates (planned)







Using the LFS data base, a set of time-series regression models were estimated of
the probability of being in the labour market and the probability of being in
employment
The dependent variable was labour market status: employed, unemployed or
economically inactive
Multinomial regression models were estimated, so the probabilities were of being
employed relative to being economically inactive, and the probability of being
unemployed relative to that of being economically inactive
Labour market participation and employment probabilities could be calculated
from the coefficients estimated
The independent variables were: Age, Age squared , Gender, Region, Highest
qualification, whether born in the UK, whether educated in the UK, No. of
dependent children and Marital status
Year dummies and interaction terms (e.g. gender*region) were also included.
The intention was that the regression coefficients would be used to project labour
market and employment rates forward for the years 2013 to 2022
Results of regression modelling









For females, the likelihood of being inactive is higher and the probability of being employed (or
unemployed) is lower than for males
The probability of an individual being employed (or unemployed) relative to being inactive
increases with increasing age and then decreases
The greater the number of dependent children, the greater the probability of being inactive and
the lower the probability of being employed (or unemployed)
People who were born in the UK have a significantly higher probability of employment (relative to
being inactive) for all ethnic groups except Asian-Bangladeshi, and a significantly higher probability
of unemployment (relative to being inactive) for all ethnicities except for Chinese and BlackCaribbean people
The degree of education (both within and outside UK) increases the chance of being employed but
does not necessarily lead to lower probability of unemployment
Being educated in the UK, compared to wholly non-UK education and partly non-UK education,
does not have any advantages in employment and in fact tend to be associated with a lower
probability of employment (or unemployment), except for the Chinese group
Married people are more likely to be employed compared to the unmarried, except for the
Bangladeshi group, but are also more likely to be inactive rather than being unemployed, except for
the Asian-Other and Black-African groups
The significance of other independent variables varies considerably between the “employed
relative to inactive” and “unemployed relative to inactive” models and across ethnic groups
Compared to the other ethnic groups, the White group presents the most significant results for the
variables considered in the multinomial regression, especially the region, qualification, year
dummies and the interaction terms between gender and region
Creating aggregate projections
 The research generated a time series of estimated total
labour force and employment, by ethnic group, age and
gender by multiplying the projected population in each
breakdown (ethnic group, age and gender) by the
corresponding:
 projected labour market participation rate
 projected employment rate
 These projections were then adjusted to sum to aggregate
projections from Working Futures 5
 The aim was to create projections for all nations of the UK
and regions of England
Main steps in producing projections of
employment by industry/occupation






Using LFS data from 2001 to 2013, the shares of employment within each industry
(occupation) for each gender by geographical area were calculated
A logarithmic function was fitted to the time-series and this was used to
extrapolate over the period 2014 to 2022
Using the total employment figures for each ethnic group by gender in each
geographical area, the actual and projected shares of employment for each
industry (occupation) was applied in order to produce a figure for employment for
each industry (occupation)
The industry (occupation) employment figures were then used to calculate each
ethnic group’s share of total employment in a particular industry (occupation) by
gender and geography. The sum of all ethnic group shares of employment in
industry (occupation) 1, for example, was equal to 100 per cent
These ethnic group shares were then applied to the employment figures for each
industry (occupation) (by gender and geography) contained in the Working
Futures projections. This provided an estimated figure for employment by industry
(occupation) for each ethnic group, by gender and geography
For the analysis of employment by industry, rather than occupation, a moving
average over 4 periods (3 in some cases where data did not permit use of four
years) was used in order to extrapolate the shares to 2022 as the data did not
exhibit any strong patterns of change over time
Labour market participation rates by gender and
ethnic group, UK 2001-2022








Men
For the UK, there was convergence over the period
2001 to 2013 as male ethnic minority participation rates
increased, while those for White men stagnated or
declined slightly.
Indian participation rates are projected to be higher
than those for the White group by 2022.
For most other ethnic groups, a slow rate of increase is
projected.
The main difference is for the Chinese group, whose
participation rate is projected to decline (reflecting the
importance of student migration for this ethnic group),
more markedly for males than females. The
participation rate for people of Mixed parentage is
projected to fall slightly by 2022.
Women
White women are projected to have the highest
participation rates, with Black, Indian and women of
mixed parentage having the next highest rates.
The labour market participation rates of Bangladeshi
and Pakistani women increased considerably between
2001 and 2013. Some further convergence is projected
by 2022, but these rates are projected to remain much
lower than those of other ethnic groups.
Employment rates by gender and ethnic group, UK
2001-2022









Between 2001 and 2013, ethnic minority employment rates
increased, while White rates stagnated or declined slightly
(males).
Men
Indian employment rates overtook those of White men in
2012. The White employment rate fell in the recession and are
not projected to recover to pre-recession levels.: the gap with
the male Indian employment rate is projected to widen slightly.
For most other ethnic groups, a slow rate of increase is
projected over the period 2012 to 2022, with Bangladeshi men
experiencing the largest projected increase.
The relative position of ethnic groups is projected to remain
stable. Chinese men are projected to experience the largest
decline in employment rate (student migration).
Employment rates for men from the Black and Mixed
parentage ethnic groups are also projected to decline slightly
by 2022.
Women
The employment rate for White women is projected to only
increase slightly, but to remain highest. The Indian, mixed
parentage and Black ethnic groups display next highest rates.
The employment rates of Bangladeshi and Pakistani women
increased considerably between 2001 and 2013, and are
projected to increase more than for other ethnic minority
groups by 2022, but remain much lower than those of other
ethnic groups.
Method for projecting employment by ethnic group








Using annual data from the LFS, employment rates (the share of individuals in employment) for each ethnic
group (by gender and geography) were calculated. This can be denoted for a particular ethnic group (eth) and
gender (sex) in a particular region/area (geog) at time t as: 𝑒𝑚𝑝𝑟𝑎𝑡𝑒𝑒𝑡ℎ,𝑠𝑒𝑥,𝑔𝑒𝑜𝑔,𝑡 .
A logarithmic function was fitted to the time series of employment probabilities. This predicted the
employment rate for a particular ethnic group (eth) and gender (sex) at time t. In order to smooth the series,
the logarithm function was fitted for the period 2001 to 2022.
The appropriate employment rate for each ethnic group and gender in each geography was applied to the
projection of working age population for the particular group ( 𝑝𝑜𝑝𝑒𝑡ℎ,𝑠𝑒𝑥,𝑔𝑒𝑜𝑔,𝑡 for each period t) as below:
𝑝𝑜𝑝𝑒𝑡ℎ,𝑠𝑒𝑥,𝑡 × 𝑒𝑚𝑝𝑟𝑎𝑡𝑒𝑒𝑡ℎ,𝑠𝑒𝑥,𝑡 = 𝑒𝑚𝑝𝑒𝑡ℎ,𝑠𝑒𝑥,𝑡
Where 𝑝𝑜𝑝𝑒𝑡ℎ,𝑠𝑒𝑥,𝑡 represents the estimated working population of ethnic group eth and gender sex at time t
and 𝑒𝑚𝑝𝑒𝑡ℎ,𝑠𝑒𝑥,𝑡 is the estimated number of people between the ages of 16 and 64 years in employment within
ethnic group eth and of gender sex at time t.
In order to ensure the projections of employment by ethnic group are constrained within the overall projections
set out in Working Futures an adjustment was made to the employment figures generated in the previous step.
The total employment across all ethnic groups within a region and for each gender was calculated was
calculated as:
𝑗
for each gender within each geography at time t
𝑒𝑡ℎ=𝑖 𝑒𝑚𝑝𝑒𝑡ℎ
and the share of this total employment by ethnic group (𝑆𝑒𝑡ℎ,𝑡 ), by gender and ethnicity, for each year was
calculated as:
𝑒𝑚𝑝𝑒𝑡ℎ,𝑡
𝑗
𝑒𝑚𝑝𝑒𝑡ℎ,𝑡
𝑒𝑡ℎ=𝑖

𝑆𝑒𝑡ℎ,𝑡 =

These shares were then applied to the respective Working Futures total employment figure for each gender
within each geographical area in order to produce a figure of employment for each ethnic group (by gender and
geography) in each year from 2001 to 2022.
for each gender within each geography.
Headline results for employment






For the UK as a whole, males and females in all ethnic groups are projected to share in the
increase in employment over this period. the largest relative increases in projected
employment over the period from 2012 to 2022 are for people of Mixed parentage, followed
by people from an Other ethnic group.
For the UK, the ethnic minority groups with the smallest projected relative increases in
employment – the Indian, Chinese and Black groups – are all estimated to see an increase in
employment of between a fifth and a quarter, while a 3% increase in employment is
projected for the White group, which accounts for just over 40% of the growth in total
employment
The share of UK total employment accounted for by White people is projected to fall to 86%
in 2022, down from 89% in 2012
In England, a small decline in employment amongst the White British group is projected over
the period from 2012 to 2022, with this decline most evident in London (where employment
amongst the White British group is projected to decline by 6%); an increase in employment
of around a quarter is projected amongst Other White groups
Outside England the projected increase in employment for the White group is lower than for
ethnic minorities (defined here as all non-White ethnic groups)
In London, 40 per cent of individuals in employment in 2022 are projected to be from the
White British group (down from 46% in 2012) and over 21% are projected to be from the
Other White group (up from just under 19% in 2012), compared with just over 83% (down
from 86% in 2012) and over 6% (up from 5% in 2012), respectively, in England outside
London.
Geographical contrasts
White British
London
Other White
groups
Mixed parentage
Indian
(inner circle is 2012; outer circle is 2022)
40%
46%
Pakistani
Bangladeshi
Chinese
All UK
Any other Asian
Black Caribbean
Black African
Black Other
White
Mixed parentage
White British
Indian
Pakistani
Rest of England
Bangladeshi
Chinese
Pakistani
Any other Asian
89%
86%
Other White
groups
Mixed
parentage
Indian
Bangladeshi
Black
Chinese
Other ethnic group
86%
82%
Any other
Asian
Black
Caribbean
Black African
Black Other
Other ethnic
group
UK projected male employment by sector, 2022
 Ethnic minorities are over-represented
in (the relatively low paid) Trade,
accommodation & transport (most
notably those from the Bangladeshi,
Pakistani and Other Asian group)
sector
 These latter groups are underrepresented in (the highly-paid)
Business & other services (e.g.
financial services, legal services, etc),
but the Indian, Mixed parentage,
Chinese, Black and Other ethnic group
are over-represented in this broad
sector
 Black males are easily the most
disproportionately concentrated in
Non-market services of any ethnic
group. Other ethnic groups with a
higher than average share of
employment accounted for by this
broad sector are the Mixed parentage,
Chinese, Any other Asian and Other
ethnic groups.
UK projected female employment by sector, 2022





Relative to the average, White females are
slightly over-represented in the Primary
sector, Manufacturing, Construction and
Non-market services
Bangladeshi, Pakistani, Chinese and Any
other Asian females are markedly overrepresented in Trade, accommodation and
transport in 2022, which represents a slight
intensification in the position in 2012
In broad terms females from these ethnic
groups in this broad sector might be
expected to be at greater than average risk
of low earnings
In contrast, females from the Mixed
parentage, Indian and Chinese ethnic groups
are over-represented in the Business &
other services sector, as are (to a less
marked extent) females from Black and
Other ethnic groups
Black females are projected to maintain
their over-representation in Non-market
services in 2022
Occupational profile of UK male
employment, 2022








Occupations were grouped into high, intermediate and
low pay occupations.
White males are over-represented in High pay and
Intermediate pay occupations.
Indian males, Chinese males and males of Mixed
parentage are also over-represented amongst High pay
occupations.
Pakistani and Bangladeshi males are over-represented
amongst Intermediate pay occupations.
Males from the Any other Asian, Bangladeshi and Black
groups are particularly likely to be employed in Low pay
occupations.
Males in nearly all ethnic groups are expected to share in
the projected increase in employment in High pay
occupations.
A larger share of projected total employment for males
in all ethnic groups is accounted for by Low pay
occupations in 2022 than in 2012, with the exception of
the Chinese group, where the share of male employment
in such occupations is projected to decrease very slightly.
Males from the Black group and the White group display
amongst the smallest projected increases in the share of
employment in Low pay occupations, while males from
the Bangladeshi, Mixed parentage and Other ethnic
groups display the largest projected increases in the
share of employment in Low pay occupations
Occupational profile of UK female
employment, 2022








Low pay occupations account for a greater share of female
employment than of male employment.
With the exception of the Other ethnic group, females in all
ethnic groups are expected to share in the projected increase
in employment in High pay occupations.
The increase in the share of employment accounted for by
High pay occupations is greater for females than males in all
ethnic groups.
The Chinese, Indian and Bangladeshi groups have the largest
increases in the share of female employment in High pay
occupations, followed by the White group.
The ethnic groups with the highest shares of female
employment in High Pay occupations are the Chinese and
Indian groups.
There is a decrease in the projected share of employment in
Intermediate pay occupations for females in all ethnic groups.
The projected decreases in employment shares are most
marked for the Indian, Mixed parentage and Pakistani groups.
There is a projected reduction in the share of female
employment accounted for by Low pay occupations over the
period from 2012 to 2022. The White, Indian, Bangladeshi,
Chinese and Black groups all share in this reduction.
For the Mixed parentage, Pakistani and Any other Asian
groups polarisation in employment is evident with an increase
in the share of employment for both Low pay and High pay
occupations.
Implications of projections





Many of the existing ethnic inequalities in the profile of male employment are
likely to persist over the medium term
In the UK, ethnic minorities are projected to account for 14%, 12% and 22% of
total male employment in high, intermediate and low pay occupations respectively
in 2022. Indian males, Chinese males and males of Mixed parentage are also
disproportionately concentrated amongst High pay occupations. Pakistani and
Bangladeshi males are over-represented amongst Intermediate pay occupations.
Males from all ethnic minority groups other than the Chinese group are projected
to be concentrated in Low pay occupations in 2022, particularly those from the
Other Asian, Bangladeshi and Black groups.
Although females are more concentrated in Low pay occupations (especially those
from the Bangladeshi, Any other Asian and Pakistani groups) than males they are
also set to benefit more than males from the projected increase in High pay
occupations between 2012 and 2022, particularly in London. This is the case for
most ethnic minorities.
Polarisation of employment structures is more evident in the Rest of England and
here the maintenance of the advantage of the White British group remains more
apparent. Polarisation of employment is also more evident for males than for
females. Relative to other ethnic groups, the White group tends to be more
concentrated in Intermediate occupations.
Challenges encountered








The research was quite ambitious, being the first known attempt to produce
detailed employment projections by ethnic group in the UK; the availability of
relatively robust projections of the population by ethnic group made this possible
The database for estimation of labour market participation and employment rates
proved not be adequate for the initial ambitions of the project to be met
Changes in occupation, industry and ethnic group classifications over time and
differences between the nations of the UK meant that it proved necessary to
aggregate ethnic groups
The sample size available for regression modelling proved to be too small for
statistically significant results to be obtained for all nations and regions of the UK
Because of this, the original methodology was substituted by a simpler and less
data-intensive method for projecting labour market participation and employment
rates
Consequently, it proved necessary to aggregate regions
In England, it was possible to distinguish London from the rest of England, but
projections could not be made for individual regions
For Wales, Scotland and Northern Ireland, only projections for white people and
minority ethnic groups as a whole could be produced