The Future of Employment / 1
The Age of Digital Disruption: the Future of Employment III Jornadas de Postgrado en Iberoamérica "El Futuro del Empleo” Sevilla, January 24, 2019
The Future of Employment / 2
The consequences for human welfare involved in questions like these are simply staggering: once one starts to think about them,
it is hard to think about anything else
Robert E. Lucas, “On the Mechanics of Economic Development," 1988
In the same vein, the effects of the digital revolution on employment, productivity, inequality and welfare could be so
stunning that, once we start to think about them, it is hard to think about anything else
Economic progress and social welfare depend on technical progress in the long run. Technological and digital transformation represents an opportunity in the history of mankind, but also enormous challenges
The digital revolution is having disruptive effects on employment, occupations, required skills, the wage premium, inequality and polarization, although so far there are no grounds to claim that it affects unemployment at aggregate level
It is crucial for societies (public sector, firms and workers) to anticipate and manage actively the digital revolution, using a broad array of policies that
• ensure equal opportunities,
• enhance the long-term positive effects of an inclusive technical and digital progress to bring this age of new opportunities to everyone, and
• reduce transition costs in the short and medium term
Key messages
The Future of Employment / 4
Introduction 01
02 Overview and historical evidence of the effects of technical progress
The effects of the digital revolution Will this time be different?
03
Policies for the digital revolution 04
Contents
Conclusions 05
The Future of Employment / 5
01 Introduction
The Future of Employment / 6
Introduction: The conventional wisdom
Long-run progress and welfare in societies are determined by technological change, which boosts productivity, wages and income per capita
Over the last two centuries, technological change has allowed for increasing employment without raising unemployment rates
On an individual basis, workers with complementary skills to new technologies have traditionally benefited from better employment opportunities and higher wages (greater productivity and division of labor, Adam Smith)
On the contrary, workers with skills that are partial or perfect substitutes for automation have experienced wage losses and even unemployment during the transition to new occupations
The Future of Employment / 7
Introduction: Is the digital revolution different?
Is the current technological and digital transformation different from previous industrial revolutions?
To what extent will the automation of routine and non-routine tasks materialize?
How will the impact of technologies such as computerization, robotisation, big data and artificial intelligence differ from that of previous technologies?
We can attempt to forecast some trends by analyzing the immediate impact and scope of emerging technologies
Growing interest academia, think tanks, international institutions and many other: a new line of research at BBVA Research
The Future of Employment / 8
02 Overview and historical evidence
of the effects of technical progress
The Future of Employment / 9
Overview: The pessimistic vision of a world without labour
PESSIMISTS
Luddites, who destroyed industrial looms between 1811 and 1816, and the Swing Riots of 1830, where threshing machines were destroyed
Marx (1867): machines replace workers, whose value falls. This contradiction will mark the end of capitalism
Frey and Osborne (2013): 47% of employment in the US is under threat from computers
Brynjolsson and McAfee (2014): Digital innovations are contributing to the stagnation of average incomes in the United States and the disappearance of many middle-level jobs
Piketty (2014): Capital (in the hands of only a few) grows more than GDP and exacerbates inequality
Benzell et al (2015): A long-term decrease in labor's share in the redistribution of income from losers to winners. Smart machines can mean long-term misery for all
De Stefano (2016): risks for workers from new forms of work
Milanovic (2016): Technological progress, sectoral reallocation of labor, globalization and current policy are generating a second Kuznets curve that will not disappear any time soon
Avent (2016): New technologies will create new and good jobs, but they will not be enough to absorb the over-abundance of workers
The Future of Employment / 10
Overview: A more optimistic vision of technical progress
Technical progress has been a constant feature in history. At the height of the Luddite movement A. Lovelace, the creator of the first programming algorithm, was born in 1815
Progress in some sectors increase income growth, which raises the demand for production in other sectors, and prompts the appearance of new goods and services, which in turn increases employment
Moretti (2010): A job created in high-tech sectors creates further 4.9 jobs in non-trade Goods sectors
Mokyr (2014): the future holds occupations that will seem as strange to us as many of those today seem to our grandparents. Our lack of imagination is largely responsible for much of today’s pessimism
Arntz et al (2016): When one considers the various different tasks in each occupation, only 9% of employment is capable of being automated on average in 21 OECD countries, which is far below the number estimated by Frey and Osborne (2013)
Gregory et al (2016): Technical change that substitutes routine work has positive net effects on overall employment in a sample of 27 countries between 1999 and 2010, as the externalities predominate which compensate the substitution of certain jobs by capital
Conseil d’Orientation pour l’Emploi (2017): Estimates do not take account of the fact that current jobs will change or of direct and indirect job creation deriving from technological change
The “gig economy” can improve employment match-ups and labour market efficiency
OPTIMISTS
The Future of Employment / 11
Today’s unemployment in the US and the UK are at rates similar to those at the start of the 20th century
GDP per capita in the United States, United Kingdom and Spain 1900-2018 (1900 = 100)
Unemployment rate in the US, Spain and the United Kingdom, 1900-2018
Source: Andrés and Doménech (2019) from Lebergott (1957), BLS, BoE, OECD. INE, Alcaide (2007) and de la Fuente (2017)
Source: Andrés and Doménech (2019) based on The Maddison Project (2018) and OCDE. Data in logarithm form
For over a century, technical progress has not destroyed jobs in aggregate terms, despite population growth and the increase of women labor participation rates
85
231
628
1900 1915 1930 1945 1960 1975 1990 2005
USA x 8,8 UK x 5,1
Spain x 10,8
100 0
5
10
15
20
25
1900 1915 1930 1945 1960 1975 1990 2005
USA Spain
UK
The Future of Employment / 12
A negative correlation in the OECD between productivity growth and changes in the unemployment rate
In the last six decades there has been a negative correlation in the OECD between productivity growth, due to innovation, and changes in the unemployment rate
Nevertheless there are differences across countries and other factors explain these differences and the negative relationship between both variable (correlation is not causation)
Productivity growth and unemployment changes, OECD, 1960-2018
Source: Andrés and Doménech (2019) based on OECD.
AUS
AUT
BEL
CAN
CHE
DEU
DNK
ESP
FIN
FRA
GBR
GRE
IRL ISL
ITA
JAP
LUX
NLD
NOR
NZL PRT SWE
TUR
USA
-2
0
2
4
6
8
10
12
14
1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
Cha
nge
in th
e un
empl
oym
ent r
ate,
201
8-19
60
Annual rate of growth of GDP per hour worked
The Future of Employment / 13
A continuous process of creative destruction, new occupations, and sectoral reallocation
Sectoral distribution of full-time equivalent employment, Spain, 1900-2015
Employment in services and per capita income in OECD countries, 1840-2000
Source: Andrés and Doménech (2018) based on Herrendorf et al (2014) Source: Andrés and Doménech (2018) based on Prados de la Escosura (2017)
Sectoral development due to technical progress, increasing globalization and changes in consumer preferences (Baumol, 1967)
0% 10% 20% 30% 40% 50% 60% 70% 80% 90%
100%
1900 1915 1930 1945 1960 1975 1990 2005 Primary sector Manufacturing Construction Services
0%
20%
40%
60%
80%
6.5 7.0 7.5 8.0 8.5 9.0 9.5 10.0 10.5 GDP per capita (in logs.)
OECD (10) Spain USA UK
The Future of Employment / 14
Thanks to technical progress, working hours have voluntarily fallen over time, contradicting Keynes
Weekly hours worked in the United States, the United Kingdom and Spain, 1870-2000
Hours worked and productivity in the OECD, 2015
Source: Andrés and Doménech (2018) based on OECD Source: Andrés and Doménech (2018) based on Huberman and Minns (2007)
30
35
40
45
50
55
60
65
70
1860 1880 1900 1920 1940 1960 1980 2000
Wor
king
hou
rs p
er w
eek
Spain USA UK
IRL
MEX
NOR
ESP UK
USA
7.2
7.3
7.4
7.5
7.6
7.7
7.8
2.9 3.1 3.3 3.5 3.7 3.9 4.1 4.3 4.5
Wor
king
hou
rs p
er y
ear,
2015
(lo
gs)
GDP per hour, 2015 ($I and 2010 PPP, logs)
Keynes (1930) predicted that in the long run the working week would be 15 hours to keep employment stable, which has not been the case so far
The Future of Employment / 15
Technical progress in health has increased life expectancy
Since 1960 life expectancy has increased by 1.9 years per decade
Life expectancy at age 65 years is increasing more than one year per decade
The increase in life expectancy is one of the determinants of the great improvement in well-being
A challenge for the pension system. Retirement age (65 years) has remained virtually unchanged since 1916
Life expectancy at birth, 1900-2011
Source: Own work based on www.clio-below.eu EU5: Denmark, Finland, Germany, the Netherlands and Sweden
30
40
50
60
70
80
1900 1920 1940 1960 1980 2000
Year
s
USA
Spain
UK
EU5
The Future of Employment / 16
Inequality: shifting trends in the last century
Income share of the top 1%, 1900-2012
-0.5
0.6
1.7
2.7
3.8
4.9
6.0
4
8
12
16
20
24
28
1900 1920 1940 1960 1980 2000
0.1% share 1%
sha
re USA
UK Spain
(0,1%,right)
EU5
Spain (1%)
Source: Own work based on www.wid.world EU5: Denmark, Finland, Germany, the Netherlands and Sweden
Inequality increased at the end of the 19th century
The Great Levelling 1920-1970
Inequality (personal and functional) has increased since 1980, mainly in major Anglo-saxon countries … … affected by the interaction between technical progress, demography, educational policies, globalization, competition in product and labor markets, and the response of the welfare state
The Future of Employment / 17
03 The effects of the digital revolution
Will this time be different?
The Future of Employment / 18
Is automation destroying employment? An open debate
Correlation between exposure to robots and employment for metropolitan areas of the US
Source: Acemoglu y Restrepo (2017)
Detroit, MI
Lansing, MI
Muncie, IN Beaumont, TX
Cleveland, OH Toledo, OH
Exposure to robots between 1993 and 2007
Cha
nge
in e
mpl
oym
ent (
Cen
sus)
, 199
0-20
07
The race against the machine Acemoglu y Restrepo (2017):
One additional robot (autonomous machines with multiple reprogrammable applications) per thousand employees reduces the employment rate by 0.18 - 0.34pp and wages by 0.25 - 0.5pp
Greater effect among more exposed industries, manual occupations and workers without a university education
Graetz y Michaels (2016) In 14 industries in 17 countries
(1993-2007), robots boost productivity and wages, and reduce prices but not aggregate employment, although they do that among the least skilled
The Future of Employment / 19
Is automation destroying employment? An open debate
Digital intensity and unemployment in 40 countries, 2016-2017
Source: BBVA Research based on data from Eurostat (2018), European Commission (2017), International Telecommunication Union (2017), and the International Federation of Robotics (2017)
UE8
KOR
DNK
GBR
NLD
JAP
SWE
DEU USA
SGP
AUT
FIN
BEL
ESP GRE (21.5)
2
4
6
8
10
12
14
16
18
-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5
Une
mpl
oym
ent r
ate,
201
7
Digital intensity and robotics, 2016-2017
The race against the machine
There is a negative correlation between the digital and robotics intensity and unemployment is observed
The digital revolution is fostering new activities, while boosting demand for less innovative sectors
The Future of Employment / 20
Skill biased technological change
Relative supply of human capital and relative salary in 18 OECD countries, 2000-16
Source: Andrés and Doménech (2018) based on the OECD (2018)
1.81
1.83
1.85
1.87
1.89
1.91
1.93
1.95
0.8
1.0
1.2
1.4
1.6
1.8
2.0
2.2
2000 2004 2008 2012 2016 W
age
prem
ium
(ter
tiary
/low
sec
onda
ry)
Rel
ativ
e su
pply,
adu
lts w
ith te
rtiar
y ed
ucat
ion
Wage premium (right)
Relative supply (left)
The race against the machine
Goldin and Katz (2008) and Acemoglu and Autor (2011). Despite the increase of workers with higher education, their relative wage has risen compared to those with lower educational levels
Technical progress is complementary to skilled workers, increasing their demand faster than their supply
The Future of Employment / 21
Routine tasks and occupations
Share of routine tasks by occupations ranked by wages
Source: BBVA research based on Aum et al (2018), and Díaz, Doménech and Neut (2018)
0%
10%
20%
30%
40%
50%
60%
70%
0 20 40 60 80 100
Sha
re o
f rou
tine
task
s
Accumulated employment by occupations, sorted by average wage
Spain
USA
Technical progress biased against routine work
Autor, Katz y Kearney (2006) find that automation and computerization complement workers who perform non-routine and abstract tasks, substitute those who carry out routine work, and do not affect those who undertake manual and non-routine activities
Polarization has also occurred in Europe in the last two decades
The Future of Employment / 22
Job polarization
Change in employment between 1993 and 2010 in 16 European countries
Source: Andrés and Doménech (2018) based on data from Goos, Manning and Salomons (2014)
-15
-10
-5
0
5
10
15
NO
R
AU
T P
RT
DE
U
FRA
ITA
NLD
G
RE
G
BR
S
WE
D
NK
B
EL
ES
P IR
E
FIN
LU
X
NO
R
AU
T P
RT
DE
U
FRA
ITA
NLD
G
RE
G
BR
S
WE
D
NK
B
EL
ES
P IR
E
FIN
LU
X
NO
R
AU
T P
RT
DE
U
FRA
ITA
NLD
G
RE
G
BR
S
WE
D
NK
B
EL
ES
P IR
E
FIN
LU
X
Four occupations with lower salaries
Nine occupations with lower salaries
Eight occupations with lower salaries
Technical progress biased against routine work
Autor, Katz y Kearney (2006) find that automation and computerization complement workers who perform non-routine and abstract tasks, substitute those who carry out routine work, and do not affect those who undertake manual and non-routine activities
Polarization has also occurred in Europe in the last two decades
The Future of Employment / 23
Skill biased technological change and job polarization are having heterogeneous effects across countries
Per capita income, inequality and unemployment OECD, 2016
Source: BBVA Research based on the OECD and Eurostat. The size of each circle is in proportion to unemployment
AUS
AUT
BEL
CAN
CHE
DEU
DNK
ESP FIN FRA
GBR
GRE
IRL
ISL
ITA
JAP
NLD
NZL PRT
SWE
USA
40
50
60
70
80
90
100
22 24 26 28 30 32 34 36 38 40
Gro
ss N
atio
nal I
ncom
e (U
SA
)=10
0
Gini Index, 2015
An efficient and fair transition
There are big disparities among advanced economies as regards to per capita income, unemployment and inequality
The challenge is to manage the technological and digital transformation by spurring growth while reducing both inequality and the unemployment rate.
Certain countries, such as Spain, are far from the frontier, allowing for higher-growth policy strategies that do not raise unemployment and inequality
The Future of Employment / 24
The future of employment
It is much easier to estimate which jobs are going to be affected than to guess which ones are going to be created
The Baumol effect: Richer societies demand more labour-intensive services, increasing wages: health and personal care (ageing), education, leisure and tourism activities, personal services, …
The challenge is a fair transition from jobs destroyed to jobs created: Protecting people rather than jobs (Tirole, 2017)
The Future of Employment / 25
Workers affected by the digital revolution
Percentage of workers in occupations with high risk of automation in the U.S. and Spain*
Source: Andrés and Doménech (2018) based on data from Fey and Osborne (2017), Arntz et al (2016) and Domenech, García, Neut and Montañez (2018)
0
10
20
30
40
50
Frey y Osborne (2017)
Arntz et al (2016) Doménech et al (2018)
Arntz et al (2016)
EE.UU. España USA Spain
According to Frey and Osborne (2017), 47% of jobs in the U.S. face a high risk of automation
With the same methodology, in Spain the percentage is 36%
When you take into account the various tasks of each occupation the risk is around 10%
The Future of Employment / 26
Workers affected by the digital revolution
Characteristics of workers in jobs most at risk of being automated in Spain*
Source: Domenech, García, Neut and Montañez (2018)
Low level of education
Women
In the sector: • Primary • Manufacturing • Trade • Hotels, restaurants
and cafés • Finance • Property
Young people
Do not undertake non-formal education
In small firms
Not telecommuting
Without positions of responsibility
Seeking a new job
Have come out of unemployment
Private sector
With temporary contract
The probability of automation of individuals is determined by variables that differentiate between each employee with regard to personal characteristics, as well as employment and the firm
Technological transformation is both an opportunity and a challenge
Although the aggregate impact is positive, it may have very different effects for different groups of workers
It is essential to lead the change with policies that smooth the transition, cushion the costs and boost the benefits
The Future of Employment / 27
04 Policies for
the digital revolution
The Future of Employment / 28
Policies for managing technological change: Education
Investing in human capital is crucial to achieving skills complementary to technical progress, even for less-skilled tasks
01 Raising the quality of jobs in the services sector and improving society’s views towards many of them
02 Improving professional (e.g. languages), social, managerial and personal skills which are required to meet society’s growing needs
03
Continuous training and flexibility to change occupations over working life. New tasks for public employment services and in collective bargaining
04 Adapt the educational system and training to new demands tanking into account skill biased technological progress
05 The educational system must evolve in step with society, fomenting creativity, boosting non-cognitive skills and improving social intelligence
06
The Future of Employment / 29
Policies for managing technological change: Education
Percentage of the population aged between 25 and 34 years by education level in 2016
Source: Doménech, García, Neut and Montañez (2018) based on data from the OECD (2017)
Skill-biased technical progress and new jobs that might be hard to imagine at the present time not only require more but, above all, better and more flexible training
The dual distribution of educational levels in Spain means that approximately one third of the younger population may lack the skills needed for the digital transformation
KOR POL
DEU
FRA
ITA POR COL ARG
ESP BRA
UE8
0
10
20
30
40
50
60
70
10 20 30 40 50 60 70
Bel
ow u
pper
sec
onda
ry e
duca
tion
(%)
With tertiary education (%)
The Future of Employment / 30
Labor market policies
Removing barriers to job creation, investment and firm growth: finance, start-ups, taxes, red tape, regulations in product markets, quality of institutions, etc,
01 Better labor market regulations, and active and passive policies: A more efficient and equitable
02 Improving the matching process between vacancies and the unemployed using big data + AI. Additional information to improve skills
03
A better tax structure that allows for redistribution without harming employment or investment in new technologies
04 Adapt labor regulations to the gig economy and the needs of independent workers and new employers in the labor market
05
The Future of Employment / 31
Policies for inclusive growth and redistribution
Education and employment policies are necessary conditions, but may be not enough
01 Greater welfare for all in the long run, but substantial transition costs for many workers in the medium term
02 First, ensure equality of opportunities and later insure individuals against adverse situations (ex-post redistribution)
03
Efficiency of public policies: ensure inclusive growth at the lowest possible cost in terms of employment and investment in innovation
04 The challenge is to distribute wealth and not to curb its creation, with taxes on automation (what is a robot?), artificial intelligence or big data
05 The welfare state still has a lot of scope for improvement in boosting employment, income and equality before proposing a basic universal income
06
The Future of Employment / 32
Multiple transitions: Winners or losers?
Low productivity and high unemployment
(Inequality with low redistribution)
Low productivity and low unemployment
(Ageing problem)
High productivity and high unemployment
(Inequality with redistribution)
High productivity and low unemployment
(A new Great Levelling)
Employment creation
Pro
duct
ivity
Countries can end up in very different equilibria, depending on how they manage the digital revolution. Better policies will allow more successful outcomes.
A new international division of labor and wealth (strategic map)
+
+
The Future of Employment / 33
05 Conclusions
Key messages
Economic progress and social welfare depend on technical progress in the long run. Technological and digital transformation represents an opportunity in the history of mankind, but also enormous challenges
The digital revolution is having disruptive effects on employment, occupations, required skills, the wage premium, inequality and polarization, although so far there are no grounds to claim that it affects unemployment at aggregate level
It is crucial for societies (public sector, firms and workers) to anticipate and manage actively the digital revolution, using a broad array of policies that
• ensure equal opportunities,
• enhance the long-term positive effects of an inclusive technical and digital progress to bring this age of new opportunities to everyone, and
• reduce transition costs in the short and medium term
The Future of Employment / 35
The Age of Digital Disruption: the Future of Employment III Jornadas de Postgrado en Iberoamérica "El Futuro del Empleo” Sevilla, January 24, 2019