2. Agenda
09:30 – 10:00 Registration with tea and coffee
10:00 – 10:05 Welcome and introduction – Grant Fitzner, Chief Economist
10:05 – 10:20 Over education and hourly wages in the UK labour market, 2006-2017 – Maja Savic
10:20 – 10:40 ESCoE Reearch: Leveraging big data to make sense of labour markets – Jyldyz Djumalieva (Nesta)
10:40 – 10:55 Long term trends in employment – Blessing Chiripanhura
10:55 – 11:05 Question and answer session
11:05 – 11:20 Refreshment break
11:20 – 11:30 Blue Book 2019 update – Sumit Dey-Chowdhury
11:30 – 11:45 Faster indicators of economic activity – Dr. Louisa Nolan
11:45 – 11:55 Questions and answers session
11:55 – 12:00 Round-up and closing remarks – Grant Fitzner, Chief Economist
4. Overeducation and
Hourly Wages in the UK:
2006-2017
Economic Advisor
Economic Advice and Analysis
Maja Savic
29 April 2019
5. Background
• Being overeducated means having more education than
required for a job.
• Persistent overeducation is a form of resource
underutilisation and/or underemployment.
• Overeducation has been associated with lower
productivity.
6. Research questions
1. What is the incidence and persistence of overeducation in the UK labour
market by sex, age, region and for graduates?
2. What is the relationship between overeducation and wages? Do results for
women and men differ?
3. Are younger (recent) overeducated graduates earning lower wages,
compared to older (non-recent) overeducated graduates?
7. Overeducation for men and women
converged during the latest periods
Source: Annual Population Survey, 2006-2017, Office for National Statistics
12.5
13.0
13.5
14.0
14.5
15.0
15.5
16.0
16.5
17.0
2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
Male Over-educated (%) Female Over-educated (%)
8. Overeducation is persistent for 25-
34 & 35-49 age groups
Source: Annual Population Survey, 2006-2017, Office for National Statistics
0.0
5.0
10.0
15.0
20.0
25.0
2006 2008 2010 2012 2014 2016
16-24 Over-educated (%) 25-34 Over-educated (%) 35-49 Over-educated (%) 50-64 Over-educated (%)
9. Overeducation rate was highest in London
Source: Annual Population Survey, 2017, Office for National Statistics
0.0
5.0
10.0
15.0
20.0
25.0
30.0
North East North West Yorkshire and
the Humber
East Midlands West Midlands East of england London South East South West Wales Scotland Northern Ireland
10. Graduate overeducation is
persistent
Source: Annual Population Survey, 2017, Office for National Statistics
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
Percentages of overeducated graduates with first degree or equivalent
qualification
Recent graduates Non-recent graduates
11. Source: Annual Population Survey, 2017, Office for National Statistics
Overeducation was higher for non-
STEM degree subjects
12. Main findings and future analysis
• In 2017, around 16% of all those in employment aged 16-64 were overeducated. The
corresponding figure for graduates was around 31%.
• The incidence of overeducation is higher for those aged 25-34 and 35-49, it is generally
lower for graduates with STEM degrees
• There is a wage penalty associated with overeducation (around 8%).
• In 2017, the overeducation rate was similar for women and for men, however the wage
penalty for overeducation was somewhat higher for men than for women.
• Recent graduates experience lower pay penalty on overeducation compared with non-
recent graduates.
We propose to extend our analysis to investigate the effect of overeducation on productivity
directly, measured in terms of output per worker or total factor productivity.
13. Leveraging big data to make sense of labour markets
Jyldyz Djumalieva
April 2019
14. • ESCoE project 3.2
• Leveraging big data to improve understanding of the labour
market
About our project
15. The high costs of
opaque labour
markets and poor
skills matching
Current and future workers lack support
and guidance on how to develop skills that
meet employer demand. Many face
stagnant pay and low social mobility.
Businesses are unable to find workers with
the right skills. The Open University
estimates that skill shortages cost the UK
£2bn a year in higher salaries, recruitment
costs and temporary staffing bills.
At a national and regional level, the lack
of alignment between supply and demand of
skills contributes to poor productivity
growth. This has adverse effect on living
standards and wellbeing.
16. Big data and data science could enable innovation in how individuals,
organisations and governments make labour market decisions
Use web data and
text mining to
classify, track and
identify new skill
sets and jobs
Use data linking to
map outcomes and
career transitions
Use granular data and
interactive
visualisations to
communicate findings
at relevant levels of
detail
Leverage new
datasets to provide
insights on the
labour market of
today and
tomorrow
The opportunity
17. Advantages:
• Online job adverts provide near real-time data on skill demands
• Data can be collected at scale
• Greater geographic granularity
• Adverts use employers’ language and capture more detailed requirements
Limitations
• Imperfect coverage
• Bias towards high-skilled occupations
Big data in the form of online job adverts
18. Online job advert dataset
• 41 million adverts collected by Burning Glass Technologies in 2012 - 2017
• Over 11,000 unique ‘skills’
• Variables on position, geographic location, offered salary and requirements
19. Research outputs
• Deliver data-driven frameworks to link skills, jobs and education
• Develop methodology to analyse evolution in skill requirements
• Evaluate novel earnings indicator
19
21. Why?
• Skill shortages are costly
• And skill needs are changing
• But we don’t measure skill demand or
supply in a detailed or timely way
• First step to fixing this: create an open
classification of skills
How?
• Take the skills mentioned in online job
adverts
• Cluster the skills based on co-
occurrence in the same advert
• Pros:
• Objective
• Can be updated frequently
• Cons:
• Skills not mentioned
• Skill categories underrepresented
22.
23.
24. Research outputs
• Deliver data-driven frameworks to link skills, jobs and education
• Develop methodology to analyse evolution in skill requirements
• Evaluate novel earnings indicator
24
25. • Measure the overall rate of change in skill sets across domains
• Study the evolution of job profiles over time using phylomemetic reconstruction
How are jobs changing?
26. • Identify skills that retain importance over time
How are jobs changing?
27. • Identify skills that retain importance over time
• As well as skills growing rapidly in demand
How are jobs changing?
28. Research outputs
• Deliver data-driven frameworks to link skills, jobs and education
• Develop methodology to analyse evolution in skill requirements
• Evaluate novel earnings indicator
28
29. • Analysed variation between offered salaries in job adverts and ASHE over 5
years at the most granular occupation level
• Using an LSTM neural network model we found that salaries from adverts can
improve accuracy of forecasts for:
﹣ 3 out of 13 industries
﹣ 4 out of 6 major occupation groups
• Provided a methodology for assessing a new data source to improve an existing
statistic
Can online job adverts help improve existing earnings
statistics?
30. How does our research add value?
• Help produce improved and new indicators (skills mismatch, earnings)
• Contribute to more responsive and evidence-based policy making
(prioritise investment in skill development, identify career transitions)
• Detect change over time (support ONS SOC revision, identify emergence of
new skill sets)
30
31. • Apply developed frameworks to measure skill demand and supply at regional
level
• Explore local labour markets (geographic isolation, London premium)
• Incorporate education in the skills taxonomy
• Continue bringing together pieces of the labour market puzzle (Open Jobs
initiative at Nesta)
What’s next
32. Leveraging big data to make sense of labour markets
Dr Cath Sleeman: cath.sleeman@nesta.org.uk, @CathSleeman
Jyldyz Djumalieva: jyldyz.djumalieva@nesta.org.uk, @d_jyldyz
33. Economic Adviser, Economic Advice and Analysis Division
ONS
Blessing Chiripanhura
Long-term trends in
employment, 1861 to 2018
34. • Data sources:
• Bank of England’s Millennium of Macroeconomic Data [MMD] (under Research
Datasets)
• For the long series (employment, private/public; employee/self-employed) [MMD]
• For shorter series – ONS sources, especially from 1971 onwards [ONS]
• Graphical presentation
• Employment rate and several disaggregations
• Implications and conclusions
Introduction
34
39. 0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
1984
1987
1990
1993
1996
1999
2002
2005
2008
2011
2014
2017
(a) The ratio of female to male
full-time workers
0.0
1.0
2.0
3.0
4.0
5.0
6.0
7.0
8.0
1984
1986
1988
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
2010
2012
2014
2016
2018
(b) The ratio of female to male part-
time workers
6.9
2.7
• The ratio of female to male full-time workers in the UK increased between 1984 and 2018
• The ratio of female to male part-time workers in the UK decreased between 1984 and 2018
0.4
0.6
UK, seasonally adjusted, 1984 to 2018 [ONS]39
41. The decline in public sector employment throughout the 1980s and 1990s is
due to a decrease in public corporations employment
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1949
1951
1953
1955
1957
1959
1961
1963
1965
1967
1969
1971
1973
1975
1977
1979
1981
1983
1985
1987
1989
1991
1993
1995
1997
1999
2001
2003
2005
2007
2009
2011
2013
2015
2017
Local government
43.3%
32.6%
38.5%
58.2%
Public corporations
Central government
24.1%
3.4%
public sector employment, 1949 to 2018 [MMD]41
43. The employment rate is at levels last experienced more than 75 years ago;
recent shocks had less severe impact on employment than earlier in the 20th
century
The sectoral composition of employment changed significantly from the
1960s onwards – fall in manufacturing employment
The participation of women in employment has increased significantly since
the 1970s (from 55.5% in 1971 to 74.2% in 2018)
Women’s participation in full-time employment has been increasing; men’s
participation in part-time employment has been increasing
Self-employment has increased significantly since the 1980s
Implications and conclusions
43
47. Nominal GDP Output
Input (of goods
& services)
output / production
approach
Real GDP Nominal GDP Prices
The ONS will be making three major changes to GDP in Blue Book ‘19:
1 2
3
1
2
3
New data on the output of goods and services across the economy;
unprecedented detail on services
New data on the goods and services businesses use
New approaches to deflation (accounting for price change): we are
going to improve the consistency between real and nominal GDP by
doing both stages of the calculation at the same time.
Improving the
production
approach
Improving
Deflation
Reminder: A Framework Fit for the Future
48. Updated Communication Plans
Publication Date
Economic Forum – Communication plan update Monday 29th April
Annual indicative GDP estimates, 1997 – 2017 June
Quarterly indicative GDP estimates, 1997 – 2017 mid-August
Sector and Financial Accounts & Balance of Payments indicative estimates, 1997-2017 end-August
Publish Quarterly National Accounts Monday 30th September
Publish Blue Book and Pink Book 2019 Thursday 31st October
49. FASTER INDICATORS OF UK
ECONOMIC ACTIVITY
29 APRIL 2019
Louisa Nolan, Jeremy Rowe,
Alex Noyvirt, Edward
Rowland, Stephen Campbell,
Daniel Ollerenshaw, Luke
Shaw, Ioannis Kaloskampis,
Andrew Sutton, Arthur
Eidukas
50. FASTER INDICATORS OF UK
ECONOMIC ACTIVITY
Identify close-to-real-time data which represents useful economic conceptsReal-time
Create early-warning indicators of potentially large economic impactsLarge changes
Provide insights into economic activity, especially increased granularityInsights
Improve power of nowcasting / forecasting modelsNowcasting
51. • HMRC value added tax
returns
• Expenditure and turnover
diffusion indices
• Reporting behaviour
• Up to 1 month before
GDP
52. • Highways England sensor data
• road traffic counts
• average speeds
• all-England and English ports
• by vehicle length
• 2 months before GDP
53. • Marine and Coastguard Agency,
ORBCOMM, UN Global Platform
• Automated Information System ship
tracking data
• port traffic frequency
• time in port
• real time
55. level > 1.5 s.d. above mean
(s)
0.5 s < level < 1.5 s
0.5 s < level < 0.5 s
1.5 s < level < 0.5 s
level < s
VAT HEATMAP
Key
Note: reporting type colours are reversed
R = Repayment claim
Re
p = Replacements (tax due & repayment)
RI
R = Re-input repayment claim
RI
T = Re-input tax due
58. Is it useful?
Faster.Indicators@ons.gov.uk
May 17 – next publication
datasciencecamp
us@ons.gov.uk
@DataSciCampus
www.ons.gov.uk/
datasciencecamp
us
FASTER INDICATORS OF UK
ECONOMIC ACTIVITY
WHAT NEXT?