일 시 : 2015년 6월 29일(월) 10:00~18:00

장 소 : 프레스센터 20층 국제회의장

주 최 : 고용노동부·한국노동연구원

언 어 : 한·영 동시통역

사회안전망과 적극적 노동시장정책Social Safety Net and Active Labor Market Policies

고용보험 20주년 기념 국제컨퍼런스International Conference in Commemoration of

the 20th Anniversary of Employment Insurance in Korea

순 서

10:00~10:15 등 록

10:15~10:30 개 회

◑ 개회사 : 방하남 (한국노동연구원 원장)

◑ 축회사 : 이기권 (고용노동부 장관)

세션Ⅰ 일자리 부족과 고용불안정의 현황 진단

◑ 사회 : 박능후 (경기대학교)

10:30~11:00 【발제 1】 기계의 대두: 컴퓨터는 노동을 어떻게 바꾸어 놓았나?

David Dorn (스위스 취리히대학교)

11:00~11:30 【발제 2】 비정규직, 일자리 양극화, 불평등

Michael Förster (OECD)11:30~12:00 【발제 3】 우리나라 노동시장에서의 일자리 양극화와 저임금노동

전병유 (한신대학교)

12:00~12:20 지정토론

김용성 (한국개발연구원)12:20~12:30 질의응답

12:30~13:40 오 찬

세션Ⅱ 사회적 보호 원리 변화의 다양한 경로

3:35 ◑ 사회 : 장지연 (한국노동연구원)

13:40~14:10 【발제 1】 유럽연합 사회투자 패러다임의 조용한 혁명

Anton Hemerijck (네덜란드 암스테르담자유대학교)

14:10~14:40 【발제 2】 국가성장전략과 복지국가의 개혁

Bruno Palier (프랑스 파리정치대학)

14:40~15:10 【발제 3】 한국 사회보장체계에서의 배제와 연대: 새로운 연대

패러다임을 찾아서

노대명 (한국보건사회연구원)

15:10~15:30 지정토론

최영준 (연세대학교)15:30~15:40 질의응답

15:40~16:00 휴 식

세션Ⅲ 실업자보호제도의 최근 변화와 적극적 노동시장정책

3:35 ◑ 사회 : 어수봉 (한국기술교육대학교)

16:00~16:30 【발제 1】 실업자 보호정책의 새로운 패러다임: 유럽의 경험

Daniel Clegg (영국 에든버러대학교)

16:30~17:00 【발제 2】 적극적 노동시장정책과 실업: OECD 회원국의 정책설계와

실행에서 얻은 교훈

Dan Finn (영국 포츠머스대학교)

17:00~17:30 【발제 3】 고용보험 20년의 평가와 과제: 사각지대와 실업급여를

중심으로

이병희 (한국노동연구원)

17:30~17:50 지정토론

이덕재 (한국고용정보원)17:50~18:00 질의응답

18:00 폐 회

Program

10:00~10:15 Registration

10:15~10:30 Opening SessionOpening Address by Hanam Phang (President of Korea Labor Institute) Congratulatory Remarks by Ki-Kweon Lee (Minister of Employment and Labor)

SessionⅠ. Changes in the Labor Market: Diagnosis of Job Scarcity and Insecurity

Moderator: Neung Hoo Park (Kyonggi University)

10:30~11:00 Presentation 1. The Rise of the Machines: How Computers Have Changed WorkDavid Dorn (University of Zurich, Switzerland)

11:00~11:30 Presentation 2. Non-standard Work, Job Polarisation and InequalityMichael Förster (OECD)

11:30~12:00 Presentation 3. Job Polarization and Low-paid Workers in KoreaByung You Cheon (Hanshin University)

12:00~12:20 Panel DiscussionYong Seong Kim (Korea Development Institute)

12:20~12:30 Q&A

12:30~13:40 Luncheon hosted by the KLI

SessionⅡ. Various Trajectories of Changes to Social Protections

Moderator: Jiyeun Chang (Korea Labor Institute)

13:40~14:10 Presentation 1. The Quiet Paradigm Revolution of Social Investment in the European UnionAnton Hemerijck (VU University Amsterdam, The Netherlands)

14:10~14:40 Presentation 2. National Growth Strategies and Welfare State ReformBruno Palier (Sciences Po., France)

14:40~15:10 Presentation 3. Solidarity and Exclusion in Korea's Social Security System: In Search of a New Paradigm of SolidarityDae Myung No (Korea Institute for Health and Social Affairs)

15:10~15:30 Panel DiscussionYoung Jun Choi (Yonsei University)

15:30~15:40 Q&A

15:40~16:00 Coffee Break

SessionⅢ. Protection for the Unemployed and Active Labor Market Policies

Moderator: SooBong Uh (Korea University of Technology and Education)

16:00~16:30 Presentation 1. Towards a New Unemployment Protection Paradigm: European ExperiencesDaniel Clegg (University of Edinburgh, UK)

16:30~17:00 Presentation 2. Activation and Unemployment: Design and Implementation Lessons from OECD CountriesDan Finn (University of Portsmouth, UK)

17:00~17:30 Presentation 3. Evaluation of Employment Insurance in Korea and the Challenges Ahead: Focused on its Non-registration for Employment Insurance and Unemployment Benefit SchemeByung-Hee Lee (Korea Labor Institute)

17:30~17:50 Panel DiscussionDeok-Jae Lee (Korea Employment Information Service)

17:50~18:00 Q&A

18:00 Closing

Contents

【SessionⅠ】

Presentation 1

The Rise of the Machines: How Computers Have Changed Work(David Dorn) ········································································································ 3

Presentation 2

Non-standard Work, Job Polarisation and Inequality (Michael Förster) ······· 31

Presentation 3

우리나라 노동시장에서의 일자리 양극화와 저임금노동 (전병유) ············ 125Job Polarization and Low-paid Workers in Korea [PPT] (Byung You Cheon) ························································································ 149

【SessionⅡ】

Presentation 1

The Quiet Paradigm Revolution of Social Investment in the European Union (Anton Hemerijck) ················································································ 167

Presentation 2

National Growth Strategies and Welfare State Reform (Anke Hassel / Bruno Palier) ··········································································· 203

Presentation 3

한국 사회보장체계에서의 배제와 연대: 새로운 연대 패러다임을 찾아서

(노대명) ··········································································································· 215Solidarity and Exclusion in Korea's Social Security System: In Search of a New Paradigm of Solidarity [PPT] (Dae Myung No) ······························ 243

【SessionⅢ】

Presentation 1

Towards a New Unemployment Protection Paradigm: European Experiences(Daniel Clegg) ·································································································· 263

Presentation 2

Activation and Unemployment: Design and Implementation Lessons from OECD Countries (Dan Finn) ·········································································· 291

Presentation 3

고용보험 20년의 평가와 과제: 사각지대와 실업급여를 중심으로

(이병희) ··········································································································· 319Evaluation of Employment Insurance in Korea and the Challenges Ahead: Focused on its Non-registration for Employment Insurance and Unemployment Benefit Scheme [PPT] (Byung-Hee Lee) ···························· 347

【SessionⅠ】

일자리 부족과 고용불안정의 현황 진단

【Presentation 1】

The Rise of the Machines - How Computers Have Changed Work -

David Dorn(University of Zurich)

The first fifteen years of the 21st Century have been a difficult period for workers in

developed economies. In many wealthy European, North American and East Asian

countries, the share of population who holds a job has declined, and wage growth for

the average worker has slowed or even turned negative. While the “Great Recession” of

the years 2007-2009 is to blame for an important part of this decline in workers’

fortunes, the employment rate in the United States was already falling prior to this crisis,

and labor markets in many countries have remained depressed for a remarkably long

time after the official end of the recession. New evidence also suggests that the fraction

of national income obtained by workers has already been declining for three decades in

developed countries.

It is therefore natural to hypothesize that labor markets are not just in a temporary

slump, but that they face a more fundamental force that increasingly imperils a worker’s

outlook for finding a job. Computer technology is an obvious candidate for that role.

Whether one enters an office building or a factory, the widespread use of personal

computers, communication devices, computer-guided machines and robots is a striking

feature of workplaces in the 21st Century. Computer technology often replaces work

tasks that were previously executed by humans, and one thus wonders whether

continued technology development will eventually lead to the obsolescence of most

human labor, as has been prominently suggested by Erik Brynjolfsson and Andrew

McAfee, two scholars from the Massachusetts Institute of Technology.

This essay discusses the impact of technological development on the labor markets of

developed countries. It argues that an imminent robot age is far from certain, and that

speculation about long-term future development is inherently difficult. However, there

already exist hundreds of years of historical experience regarding past technological

change, and several decades of experience with computer technology in the labor

market. I show that this evidence from the distant and recent past reveals some recurring

patterns, which may plausibly extend into the near future.

I. Is Technological Development Just Beginning or about to End?

Gordon E. Moore, co-founder of Fairchild Semiconductor and Intel Corporation,

predicted in 1975 that technological progress in the semiconductor industry would

allow a doubling in the number of electronic components per microchip every two years,

- 5 -

thus

becam

progr

in m

comp

comp

Figur

Sourc

Co

New

robot

wide

drivi

leading to

me known

ress in the

any electro

piled by th

puting opera

re 1. Evolutper se

ce: Nordhaus

omputers ar

w application

tics hold gr

er set of tas

ing car, wh

rapid prog

as “Moore

semiconduc

nic devices

he economi

ations has b

tion of costecond (MSO

(2001).

re not only b

ns in such a

reat promis

ks. One of

hich may ha

gress in com

e’s Law”, h

ctor industry

, and to rap

ist William

been falling

t per millionOPS)

becoming m

areas as ma

e for expan

the most re

ave profoun

mputer proc

has held up

y contribute

pidly falling

m Nordhaus

dramaticall

n standard

more powerf

achine learn

nding the u

emarkable i

nd impacts

cessing pow

remarkably

ed to great

g prices of c

s indicates

ly since the

ized compu

ful and chea

ning, artific

use of comp

innovations

on persona

wer. This pr

y well sinc

performanc

computing

that the c

1950s (Fig

uting opera

aper, but als

ial intellige

puter techno

s in recent y

al transport

rediction, w

ce. The dram

ce improvem

equipment.

cost per m

gure 1).

ations

so more ver

ence, and m

ology to an

years is the

tation in the

which

matic

ments

Data

million

rsatile.

mobile

n ever

e self-

e 21st

- 6 -

Century.

Erik Brynjolfsson and Andrew McAfee, two scholars at the Massachusetts Institute

of Technology, argue that the combination of a rapid fall in computer prices and a wider

applicability of computer technology are heralding a “Second Machine Age” that will

transform the economies of developed countries to an even greater degree than the

Industrial Revolution (Brynjolfsson and McAfee, 2011 and 2014). As ever cheaper

robots are able to execute an ever greater range of tasks, firms will only hold on to

workers as long as their jobs cannot be executed more cheaply by a machine. The result

is enormous downward pressure on workers’ wages, and as these wages fall below a

reservation level—the minimum level of wages that make employment worthwhile for

people--a rapidly growing pool of unemployed individuals will form.

A technology-driven disappearance of employment opportunities for most people will

profoundly change societies, and will require a new organization of many public

institutions. Despite the chilling outlook of mass unemployment, the Second Machine

Age does not only produce losers, however. The widespread use of cheap robots in

production will lower production costs, and the resulting productivity increase raises

aggregate societal wealth. However, these gains are likely to be concentrated among a

small group of owners of computer capital, while a much larger fraction of the society

suffers from the loss of gainful employment. It is an interesting intellectual exercise to

think about appropriate public policies that could successfully deal with a workless

society.

The jobless robot age predicted by Brynjolfsson and McAfee is however far from

certain. Quite to the opposite, Robert Gordon, an economic historian from Northwestern

University, posits that technological change is slowing rather than accelerating (Gordon,

2012 and 2014). He argues that the period of greatest technological advance in history is

not the past three decades when computer technology spread, but the century from the

1870s to the early 1970s. This century of “Great Innovation” saw the discovery of

electricity and the development of a panoply of electrical devices, the invention of the

internal combustion engine which revolutionized transportation, an improvement of

sanitary and living conditions due to running water, indoor plumbing and central

heating, the ability to rearrange molecules which permitted great progress in the

development of pharmaceuticals, plastics and other chemical products, and the

introduction of major communication and entertainment technology, following the

- 7 -

invention of the telephone, the phonograph, photography, radio and motion pictures

within a span of just fifteen years.

Annual growth of U.S. GDP per capita accelerated during the Great Innovation

period, and peaked at an average of 2.5% growth per year between 1950 and the onset

of the oil crisis in 1973. In the four decades since, average annual GDP growth has been

one third lower at 1.6% per year. While annual growth even exceeded 4% in ten years

during the 1950-1973 period, such a high growth rate has never again been attained

during the last thirty years. This slowdown in economic growth is not unique to the

United States, but it has been even more pronounced in Japan and in the major

European economies.

Gordon argues that the slowdown in economic growth results from a slowdown in

innovation. The Great Innovation century, which was characterized by major

development in multiple important technologies, has been followed by a period with a

comparatively one-dimensional development in computer technology only. Indeed,

Gordon makes the provocative prediction that innovation and economic growth will

continue to slow as it gets increasingly difficult to come up with fundamentally novel

discoveries.

Of course, path breaking future innovation is extremely difficult to predict. Who

would have foreseen today’s omnipresence of the internet-connected multi-purpose

smartphones just thirty years ago? But as much as it is problematic to extrapolate the

slowdown in economic growth into the future, one can also not confidently extrapolate

past trends in the development of computer technology and conclude that an age of

robots is immanent and inevitable.

While long-term predictions are notoriously difficult, this essay argues that one can

learn important lessons from past experience. Historical experience not only allows

assessing the impacts of past technological change, but these impacts can be contrasted

with past predictions about the transformative implications of technology. Indeed,

concerns about the replacement of workers by machines date back many centuries, and

there is mounting evidence on the impacts of computers and robots on the labor market

during the last three or four decades.

- 8 -

II. Drawing on Past Experience: A Short History of Technological Change in Textile Production

Since thousands of years, humans have been producing textile clothing. In this

production process, cotton, wool or other fibers are first converted to yarn, then yarn is

converted to cloth, and finally cloth is converted to clothing. While the basic sequence

of production steps remained unchanged over time, there were dramatic improvements

in the execution of these steps. During the Industrial Revolution, the textile industry

was at the forefront of a broader trend towards mechanization of production. Yet

already prior to that transformative period, the textile sector provides a useful example

for the study of technological change and its labor market implications.

A first major innovation in the textile sector affected the process of spinning, i.e., the

conversion of fiber to yarn. Historically, humans would use attach fibers to a spindle

and rotate that spindle by hand in order to twist fibers to yarn. From about the 13th

Century onwards, such hand spindles were gradually replaced by spinning wheels,

which took advantage of the rotational energy of a wheel, and greatly increased the

productivity of the spinning process.

The introduction of the spinning wheel met occasional resistance by the crafts guilds

that controlled production in many European territories. According to historical

documents, the city council of Cologne decided in the year 1412 that a local merchant

by the name of Walter Kesinger would not be allowed to construct a spinning wheel,

after he had seen such a machine during travels in Italy. The council argued that many

spinners would loose their livelihood if the use of the new, more productive technology

were permitted.

Mechanization also reached the second production step of the textile industry, the

weaving of yarn to cloth. As of 17th Century, a predecessor of the mechanized loom

permitted the simultaneous production of up to 24 woven ribbons, a dramatic

improvement over the classical loom that could only produce one ribbon at a time. To

prevent employment loss among ribbon weavers, the German emperor prohibited the

use of this mechanized ribbon loom in 1685, and regents of many other European

territories did the same.

The prohibition of new technology in an attempt to prevent employment loss

- 9 -

however proved counterproductive in the long run. Right outside the borders of the

German empire, craftsmen in the Swiss city of Basel adopted the new ribbon-weaving

machine, and were henceforth able to produce at much lower cost than their German

counterparts. Competitive pressure from technology adopters eventually led to growing

opposition against the German empire’s machine ban, which was finally overturned in

the mid 18th Century.

The process of textile production soon changed even more dramatically as a series of

inventions in the late 18th Century, at the start of the Industrial Revolution. The spinning

wheel was replaced by the “spinning jenny”, a multi-spindle spinning frame that could

produce many yarns at a time. Weaving was revolutionized by the introduction of the

power loom, an automated loom that was powered by steam or water energy rather than

by human hand.

The new production technology required expensive machines and access to a central

power source such as a steam engine in order to operate power looms. Therefore, the

decentralized home production of yarn and cloth by spinning wheels and handlooms

was replaced by mass production in factories. In a mechanized factory, a single worker

was able to produce an output that would have required dozens of workers prior to the

Industrial Revolution.

The massively reduced need for human labor in textile production led to popular

unrest in England. Unemployed workers protested against the new system of factory

production, and in some cases attacked factories and smashed machines. The British

government reacted by making “machine breaking” a capital crime. The protesting

workers, who became known as Luddites after their alleged leader Ned Ludd, were

persecuted and their movement broken up.

A common theme during these centuries of technological progress in the textile

industry was the fear that new labor-saving technologies would lead to long-term mass

unemployment. Indeed, new technologies certainly disrupted the labor market, and

many workers lost their employment when their jobs became mechanized. While

concerns about technology-induced unemployment were well founded in the short run,

the predicted long-term decline in employment never materialized.

A simple, and yet profoundly mistaken view of the labor market is that there is a

fixed amount of work, which can either be done by humans or by machines. In this view,

an increasing use of machines in the production process necessarily reduces the work

- 10 -

that is available to humans. However, in reality the amount of work is not fixed. New

technology is often associated with the emergence of new types of jobs. The

development of spinning machines and power looms for instance led to the creation of

many jobs in a new machine-producing industry.

Less obvious, but probably more important is a beneficial effect of technological

change that operates via prices. Mass production in the textile industry led to a dramatic

drop in the price of clothing. This price decline allowed consumers to either buy a

greater quantity of clothing with the same amount of money, or to buy the same amount

of clothing for less money while adding purchases of other goods and services. The

resulting increase in demand for clothing and other outputs led to greater production and

rising employment in many sectors of the economy, particularly in those that were not

directly exposed to labor-saving technology.

Over time, technological change did not eliminate employment, but it strongly

changed its composition. From the 19th to the late 20th Century, a large part of

employment first moved from agriculture to the manufacturing sector, and later from

manufacturing to the service sector. It would have been unthinkable two centuries ago

that agriculture, which employed the large majority of workers at the time, would

account for a mere few percentage points of employment today, despite producing a

much greater output. Yet unemployment has not shown a pronounced upward trend

over time, as workers have found new employment opportunities in industries like

healthcare, finance, and entertainment, whose relevance for employment would have

been equally difficult to foresee. By ignoring the emergence of new employment

opportunities, many past observers of technological change have fallen victim to the

“Luddite fallacy” of wrongly predicting a rise of long-term unemployment.

III. The Computer Revolution: Renewed Fears about the Automation of Work

The last---and continuing---wave of technological change that reached the labor

market is the adoption of computer technology in the workplace. Its predecessors date

back at least to the Industrial Revolution, when the French inventor Joseph Marie

Jacquard developed a loom that was operated by replaceable punched cards, which

- 11 -

controlled a particular sequence of operations. The same principle of operating

machines with programs stored on punched cards was later used in mainframe

computers from the 1960s onwards.

The start of the computer revolution is however often dated to the late 1970s or early

1980s. Production in factories was already changing rapidly at the time as computer-

guided machines became more widely and cheaply available. These devices included

computer numerical control (CNC) machines that were operated by a computer program,

and industrial robots that could move a robot arm around multiple axes.

Computerization also started to affect office work, most notably due to the introduction

of personal computers in the early 1980s. The IBM Personal Computer was released in

1981, the downmarket Commodore 64, which became the best-selling computer of all

time, followed in 1982, and Apple introduced its iconic Macintosh in 1984. Continued

technological development has since led to ever more powerful and versatile computers,

production machines, and robots. And great advances in communication technology,

including the World Wide Web and wireless communication devices, have further

increased the reach of these machines.

The computer revolution quickly ignited fears about mass unemployment. As early as

in 1978, the German news magazine Der Spiegel titled: “The computer revolution:

progress creates unemployment.” It argued that computers differed from earlier

technological innovations, since they not only eliminated many jobs in automating

sectors, but also failed to create a meaningful number of new jobs in computer

production or elsewhere in the economy. The article cited a British union leader who

predicted that by the year 2000, most jobs would have been replaced by computers. This

pessimistic prediction turned out to be quite mistaken. The unemployment rate of the

United Kingdom, which stood at 6% in 1978, again hovered around 6% in the year

2000, when supposedly most work should already have been lost. At it remained at 6%

even in 2015, a full 37 years after Der Spiegel wrongly predicted a “social catastrophe”

of unprecedented mass unemployment.

The fact that computers failed to eliminate most human labor in the past four decades

despite predictions to the opposite invites a healthy skepticism about renewed claims

that robots are just about to take over from humans. More likely than not, today’s

technology enthusiasts will be seen as the next victims of the Luddite fallacy within a

few decades, thus joining the many previous pundits that predicted the end of human

- 12 -

work in the past.

Of course, the fact that computers did not create mass unemployment does not mean

that they lacked an impact on the labor markets of developed countries. Quite to the

contrary, many economists consider computerization, along with the globalization of

goods and worker flows, as the main driver of change in labor markets during the last

three or four decades. A closer look at computers’ effects on workers during that period

is thus warranted, and it may provide some guidance for thinking about computer

technology’s future impacts.

IV. Computers in Economic Theory: Skill-biased vs Task-biased Technological Change

Computers became an important topic on the research agenda of macroeconomists

and labor economists in the 1990s. Researchers observed that the wage differential

between university-educated workers and less educated workers had risen rapidly

during the previous decade, both in the United States and in several other developed

countries. A leading hypothesis to explain the growth of inequality was skill-biased

technological change (SBTC). It posits that new technology and machines augment the

productivity of all workers, but that productivity gains are larger for workers with a

higher education. The growing productivity advantage of highly educated workers in

turn increases firms’ demand for such workers.

The Harvard economists Claudia Goldin and Lawrence Katz (2008) have argued that

such SBTC has taken place throughout the 20th Century, thus continuously raising the

demand for skilled labor. That growing demand coincided with a rapidly growing

supply of skilled labor, as average education levels increased dramatically in all regions

of the world. The 1980s however marked a period where the growth in the supply of

skilled workers slowed in the United States, whereas demand for such workers

continued to grow or even accelerated. The excess growth of skill demand relative to

skill supply generated an increase in the relative wage of workers with higher education,

and thus greater wage inequality in the labor market.

Through the lens of the SBTC hypothesis, computers are seen as the continuation of a

sequence of technological innovations that favored more educated workers. But how

- 13 -

exactly could it be explained that computers were more beneficial for some workers

than for others? David Autor and Frank Levy from the Massachusetts Institute of

Technology and Richard Murnane of Harvard University conducted field studies that

analyzed the introduction of computer technology in firms, and they observed

computers’ impact on employment levels, wages, and job content of different types of

workers. Based on their findings, they formulated a refined theory for the impact of

computers on the labor market.

In the Autor, Levy and Murnane (2003) model, computers do not have a differential

impact on workers based on workers’ education levels, but based on the task content of

their occupations. This model draws on the key observation that computers have distinct

strengths and weaknesses when it comes to executing different work tasks. A personal

computer, CNC machine or robot is directed by software that was pre-specified by a

programmer. Computers are thus good at doing tasks that follow a well-defined

procedural sequence or routine. Such “routine tasks” are found in repetitive production

work that is often organized along assembly lines or conveyor belts. In mass production,

it is crucial to repeat the exact same work steps over and over, and the margin for error

is low in order to ensure high quality and replaceability of parts. Computer-guided

machines and robots are much better than humans at executing such high-precision

repetitive work.

Routine tasks also appear in a second area of the labor market. Many clerical

occupations deal with data work, including data processing, data storage, data retrieval,

and data transmission. Whenever such data tasks are executed according to clearly

specified rules, then they belong to the set of routine tasks that can readily be done by

computers. For instance, computers now execute many tasks that were previously done

by accountants, file clerks, or secretaries.

While computers are often better than humans at doing routine tasks, they face

important limitations in the execution of other tasks. First, computers and machines that

follow a pre-specified program have difficulties to come up with new ideas and

inventions, or to react to unforeseen influences on their work. And second, computers

and machines sometimes lack good interfaces for dealing with people and objects. This

includes limitations in verbal communication with humans, as well as difficulties in the

recognition and physical handling of objects---all tasks that require versatility and

adaptation to the environment.

- 14 -

The first type of tasks that is difficult to replace by computers has become known as

“abstract tasks”. These tasks include problem-solving, creativity, or managerial

leadership, which all draw on humans’ ability to react to new developments and

problems, and to come up with new ideas and solutions. Occupations that are rich in

these abstract tasks include managers, engineers, medical doctors, or researchers. A

common feature of these jobs is that they require a high level of cognitive skill, and

typically a university education. While computers tend to be poor substitutes for

humans in these jobs, they can instead be valuable complements. Many abstract task-

intensive occupations become more productive when computers allow a cheaper and

more rapid processing, storage and transmission of data. For instance, an engineer who

designs bridges benefits when computers permit a rapid calculation of a planned

object’s static properties, or a manager of a large firm benefits from observing a stream

of real-time data that indicate the state of operations at the firm’s multiple plants. Far

from being replaced by machines, these workers specializing in abstract tasks hence

stand to benefit from a more widespread use of computers.

A second type of tasks that is difficult to automate has become known as “manual

task”s, and consists in some combination of fine motoric movement, visual recognition,

and verbal communication. These tasks are important in personal service occupations

such as waiters, childcare workers or hairdressers, but also in a range of occupations

that provide transportation, repair and construction services. Workers in these

occupations typically require little schooling, since they use humans’ built-in abilities to

verbally communicate with other humans, to see and recognize persons and objects, and

to grab, hold and move many types of objects with the human hand. Computers have

little direct impact on these manual tasks-intensive jobs, which are not easily automated,

but also do not benefit much from an interaction with computers in the workplace.

Cleaners of hotel rooms are an excellent example not only for illustrating a manual

task-intensive job, but also for explaining the difference between repetitive and routine

work. The sequential cleaning of hotel rooms is certainly a repetitive chore. However,

this repetitiveness does not translate to routineness in the sense used here. For hotel

cleaning to be a routine job, it would be necessary that the cleaning of one hotel room

would encompass precisely the same physical work steps as the cleaning of the next

room. But in practice, every guest will leave her room in a slightly different state. Apart

from differences in cleanliness, guests can leave towels, pillows, toiletries, pens and

- 15 -

many other objects that belong to the hotel in different spots within the room. For a

robot, it would be very challenging to find and recognize all of the hotel’s objects,

assess their state of cleanliness, and take the appropriate measures of cleaning and

replacing objects. Compared to humans, robots are often very limited in the diversity of

their physical abilities, and cannot grip or clean many different types of objects. An

even greater challenge arises when hotel guests leave behind novel objects that they

brought into the room, like a pizza box or a jewel case. For a human, it’s easy to

recognize these objects, and to decide that the pizza box should be discarded while the

jewel case should be kept. The same task, however, presents a major obstacle for a

machine, and greatly complicates a replacement of human cleaners with robots.

V. Labor Market Polarization

The theory of task-biased technological change (TBTC) by Autor, Levy and Murnane

predicts that employment in routine task jobs should decline as cheaper and more

powerful computers become available that can substitute for workers in these jobs. This

prediction is supported by evidence from many developed countries in North America,

Europe and Asia, which are all exposed to computerization. For instance, my work with

David Autor observes that routine task-intensive occupation groups such as production

occupations, machine operators and clerical workers accounted for a roughly constant

37 to 38 percent of U.S. labor input from 1950 to 1980, before declining rapidly to just

28 percent of U.S. labor in 2005 (Autor and Dorn 2013). As routine occupations

declined, managerial and professional occupations that intensively use abstract tasks

grew rapidly. Also expanding since the 1980s are low-skilled service occupations such

as waiters, cleaners and childcare workers, whose work is rich in manual tasks.

Occupations in farming, mining, construction, repair and transportation, which also

mostly perform manual tasks, were declining rapidly until 1990, but have since

stabilized their employment share.

While the TBTC hypothesis provides clear predictions for the distinct effects of

computerization on occupations that use different job tasks, it also has indirect

implications for the inequality between workers of different education or income levels.

The European economists Maarten Goos and Alan Manning pointed out that the routine

- 16 -

occu

comp

distri

abstr

task j

routi

work

the e

show

Euro

occu

coun

relati

empl

so pe

Figu

Sourc

upations in

puters, ten

ibution (Go

ract task job

jobs such a

ine task-inte

kers become

economy. I

wed that su

opean count

upations wit

ntries, both

ive to the m

loyment po

ervasive de

re 2. Emplo

ce: Acemoglu

production

d to be c

oos and Ma

bs such as

as personal s

ensive occu

e increasing

In follow-u

uch polariz

tries they s

th intermed

employme

middle-wag

larization v

espite subst

oyment Pol

u and Autor (

and cleric

clustered to

ann ing, 200

managers a

service wor

pations tran

gly concentr

up research

ation is rem

studied as

diate wage

ent in low-w

ge jobs (Fig

varies by co

tantial inter

arization in

(2011) based

cal work, w

owards the

07). These

and profess

rkers. As a c

nslates to a

rated in the

h, Goos, M

markably p

well as in

levels has

wage and

gure 2). Wh

ountry, it is

rnational di

n Europe an

on Goos, Ma

which can m

e middle o

jobs typica

sionals, but

consequenc

pattern of e

highest and

Manning an

pervasive a

n the U.S.,

declined, a

in high-wa

hile the exa

striking th

fferences in

nd in the U

anning and Sa

most easily

of the occ

ally have lo

higher wag

ce, declining

employment

d lowest pai

nd Anna S

across coun

the emplo

and in all b

age occupat

act extent an

at the same

n industry

nited State

alomons (200

y be replace

cupational

ower wages

ges than m

g employme

t polarizatio

id occupatio

Salomons (2

ntries. In a

oyment sha

but one of

tions has g

and shape o

e basic patte

structures,

es

09).

ed by

wage

s than

manual

ent in

on, as

ons of

2009)

all 16

are of

these

grown

of this

ern is

labor

- 17 -

market regulations, and local economic growth. The average changes across European

countries are not just qualitatively, but even quantitatively very similar to the changes

observed in the United States.

The polarization of the occupational employment structure invites the question

whether similar polarization patterns can be observed for wages. In the United States,

this is indeed the case. At least since the 1990s, wage inequality has only increased in

the upper tail of the wage distribution. While the wages of high-income workers at the

90th percentile of the wage distribution have grown faster than the wage of the median

worker, the median wage has lost ground relative to the 10th percentile of the wage

distribution as the wage differential between workers in the middle and at the bottom of

the distribution has been narrowing.

This trend of wage polarization, whereby the median wage declines both relative to

the highest and lowest percentiles of the distribution, has a direct counterpart at the level

of occupations. Figure 3 orders the several hundred occupations that are observed in the

U.S. Census according to their average wage in 1980, and shows that during the next 25

years, both occupational employment and occupational wage growth were larger in the

highest paid occupations (on the right side of both panels) and in the lowest paid

occupations (on the left) relative to occupations with intermediate wages (towards the

center). Closer inspection of the data suggests that the very pronounced wage growth in

high-wage occupations was driven by managerial and professional occupations,

whereas wage growth in low-wage occupations stemmed largely from low-skilled

service jobs.

The international evidence on wage polarization is sparser and less homogeneous

than in case of employment polarization. The growth of employment in the highest paid

occupations has often been accompanied by substantial wage gains in these jobs,

suggesting that the growing supply of highly skilled workers for these jobs has not kept

up with growing demand in many countries. The expansion of employment in low-paid

occupations however has coincided with either rising, stagnant or falling wages,

depending on the country and time period.

Indeed, the U.S. experience of a combined employment and wage growth in low-

skilled service occupations is rather surprising in the context of the SBTC and TBTC

theories. These theories both provide the prediction that computers enhance the

- 18 -

Figur

Sourc

re 3. Emplo

ce: Autor and

oyment and

d Dorn (2013)

d Wage Pol

).

arization inn the Unitedd States

- 19 -

productivity of skilled workers in occupations with abstract tasks, and firms’ rising

demand for such increasingly productive workers can readily raise both their

employment and wages. The low-skilled workers in manual tasks however have little

direct benefits from computers according to both theories. In the task-based view, low-

skill service occupations and other jobs with manual tasks are better off than the routine

occupations, since manual tasks cannot be readily automated. This feature explains why

employment in manual occupations grows relative to employment in routine

occupations. However, as redundant clerical and production workers seek new

employment in manual task jobs such as cleaners and waiters, which are readily

accessible for workers with little formal education, one would expect wages in these

occupations to fall. The combined increase of employment and wages in low-skilled

service occupations suggests that a second force must be at work in addition to this

labor supply channel.

Autor and Dorn (2013) argue that this second force is a growing demand for low-skill

services. Technological change lowers the production cost for many manufactured

goods but not for low-skilled services whose production process is little affected by

computers. When consumers perceive goods and services to be poor substitutes in

consumption, then they react to the falling price of such goods as TVs not by

purchasing many more of those items, but they instead use some of the money that is

saved on cheaper goods in order to purchase more services. Increased spending on

restaurant meals, childcare services and so forth then contributes to raising both

employment and wages in the occupations that provide these services. In combination

with the labor supply effect of worker reallocation from routine to manual occupations,

one obtains employment growth in low-wage manual task occupations, which can be

accompanied by wage growth if the demand for their outputs grows sufficiently rapidly

relative to the increasing supply of workers.

The observation that the demand for low-skilled services is increasing is important

because such jobs can provide employment opportunities for workers who have little

education or job-specific training. It contradicts the notion that all low-skilled work is

rapidly becoming obsolete as a consequence of computerization.

- 20 -

VI. Technology versus Globalization

The task-based model of technological change provides a framework to understand

the use of computers and robots in the workplace, and it provides predictions about

occupational employment growth that are consistent with evidence from many

developed countries. However, the fact that computerization could be a plausible

explanation for labor market trends such as employment polarization does not imply

that it is the only or the main explanation for these trends. Concurrent with

computerization, developed economies have experienced other major economic changes,

with globalization being the most striking development.

Globalization is a process of international integration that encompasses rising trade in

goods and services, growth in international capital movements, migration of workers,

and dissemination of knowledge. Integration has deepened along all these dimensions in

recent decades, and progress in computer and communication technology may well

have acted as a catalyst for that integration. The use of foreign suppliers, for instance,

has become more attractive for firms as it has become easier and cheaper to

communicate over long distances, and to seamlessly track shipments.

Globalization provides an alternative narrative that could explain the decline of

middle-wage occupations in many developed economies. Workers in production and

clerical occupations may not only be replaced by computers and robots, but also by

workers in other countries where wage levels are lower. This spatial reorganization of

production can take the form of offshoring, where multinational firms shifts part of their

operations to another country, or of trade competition, where firms in developed

countries succumb to competitive pressure from cheap import goods and have to reduce

employment. These global shifts in production are particularly apparent in

manufacturing. Since the early 1990s, China evolved from being a minor player in

international trade to being the world’s leading exporter of goods, thus greatly

contributing to a dramatic rise in trade between developed economies and low-wage

countries. In addition to trade in goods, there is also an increasing trend towards

offshoring of clerical work.

The contemporaneous occurrence of computerization, globalization and other

economic changes makes it difficult to estimate their separate effects on employment

- 21 -

polarization and other aggregate labor market outcomes in a country. Autor and Dorn

(2013) proposed to study computerization within local labor markets in the United

States in order to overcome this dimensionality problem. The concept of local labor

markets builds on the empirical observation that workers’ job search is often confined to

jobs that are located within commutable distances from their home. As a consequence,

local labor supply and local labor demand combine to form local labor market equilibria,

and the real wages for a given type of work can quite persistently vary across cities.

Firms in different cities and rural areas of the United States should all have access to

the same technologies. Nonetheless, computerization will have spatial variation. This

variation stems from the fact that local labor markets vary in their industry, occupation

and task mix. The historical source of local specialization can be geographic proximity

to the raw materials that are required by an industry, or access to transportation

infrastructure, or even the serendipitous emergence of an important firm around which

other related businesses cluster. Importantly, local persistence is remarkably stable even

over long periods of time. For instance, local labor markets that made particularly large

use of routine labor in 1950 still had a disproportionately routine-intensive occupation

mix half a century later. This historical reliance on routine labor later created a large

potential for the replacement of workers by computers and robots, and thus a

particularly large exposure of these locations to computerization. An empirical analysis

of U.S. Census data indeed shows that historically routine task-intensive local labor

markets, which are dispersed across all regions of the U.S., adopted more computers

since the 1980s, and experienced greater declines in routine work. As a consequence,

local labor markets with a high initial employment share of routine occupations

experienced a stronger employment polarization than locations with comparatively little

reliance on routine work, and they also some evidence for greater wage polarization

(Figure 4).

The comparison between local labor markets not only allows establishing a more

direct link between computerization and labor market polarization, but also permits to

separate the effects of computerization from those of globalization and other economic

forces. In research with David Autor and Gordon Hanson, I show that the U.S. local

labor markets which are most exposed to computerization do only partially coincide

with the regions that are most affected by the dramatic rise of import competition from

China (Autor, Dorn and Hanson, 2013 and forthcoming). Econometric analysis can

- 22 -

Figur

Sourc

re 4. Emplovs Low

ce: Autor and

oyment andw Specializa

d Dorn (2013)

d Wage Poation in Rou

).

larization inutine Task-

n U.S. Loca-Intensive A

l Labor MaActivities

arkets with High

- 23 -

therefore identify the local labor market impacts of these technology and trade forces

separately. Empirical results show that the extremely rapid increase of Chinese import

competition since the 1990s has led to a substantial decline of employment in all

occupation categories of the manufacturing sector. Due to a slow reallocation of

manufacturing workers to other jobs, this import shock has also reduced overall

employment.

By contrast, routine task-intensive local labor markets have not experienced a

significant overall job loss, but their occupational employment structure has polarized in

each sector. Consistent with the timeline of technology development, the declines in

routine employment initially concentrated on production work in the manufacturing

sector during the 1980s, and later became larger in the service sector as computers

started to replace clerical work. The overall losses in routine employment have however

been largely balanced by offsetting job gains in manual task and abstract task-intensive

occupations within each sector.

Overall, both empirical studies from the United States and from Europe suggest that

the adoption of computers and robots has not caused major employment declines.

However, computerization has repeatedly been linked to labor market polarization, as

middle-wage routine occupations are being replaced with computers and machines. VII. Will the Lessons of the Past Remain Relevant in the Future?

The prediction that labor-saving technology inevitably causes a long-term rise in

unemployment has been proven wrong many times in history, and again during the first

few decades of computer adoption. Yet it remains a valid question whether the theory of

task-biased technological change, which currently guides many economists’

understanding of computers’ impact on the labor market, will remain useful in the

future as technology evolves further.

The Oxford University scholars Carl Benedikt Frey and Michael Osborne argue that

computers will soon be able to execute abstract cognitive tasks due to advances in

machine learning and artificial intelligence, while mobile robotics technology will allow

for the substitution of workers in manual task-intensive occupations (Frey and Osborne

2013). The driverless car is one of the most publicized technological breakthroughs of

- 24 -

the last few years, and a prime example for the expanding possibilities of technology.

While Autor, Levy and Murnane (2003) listed truck drivers as an example of a manual

task occupation that cannot readily be substituted for by technology, such automation

now seems immanent to many observers. And once it is possible to automate drivers,

then the replacement of other manual occupations by robots may not be far away.

It is certainly plausible to expect that the boundaries between automatable and non-

automatable tasks will continue to evolve as computers and robots become more

powerful and versatile. The fundamental insight that computers’ primary strength lies in

the execution of tasks that can be characterized by precisely defined routines

nevertheless remains valid. As a consequence, many of the occupations that are difficult

to substitute by technology now will remain hard to automate in the future unless their

work content can be translated to a well-defined routine that a machine can execute.

Moreover, enthusiastic predictions about rapid technology development and adoption

frequently underestimate challenges on the path from a technological prototype to a

large-scale market introduction of a new technology. Take the driverless car as an

example. The model developed by Google costs about ten times as much as a regular

car according to estimates in major news media. Moreover, it cannot drive in snow,

detect potholes, or deal with other unexpected and challenging environments. Of course,

it is very well possible that continued technology improvement will soon lead to the

construction of driverless cars whose price and capabilities make them clearly superior

to conventional cars. But it is far from certain that this price decline and technology

improvement will occur soon, or at all. Other transportation technologies that were once

announced as great breakthroughs, including supersonic airplanes, maglev trains or

solar vehicles, have never become cheap and powerful enough to be adopted widely in

the economy.

An additional and frequently underestimated obstacle to automation is that human

workers often execute a more diverse bundle of tasks than a computer or robot can

replace. A truck driver for instance does not only steer a truck through traffic, but also

loads, controls and unloads the cargo, deals with the accompanying paper work, and

does maintenance and repair work on the truck. The driverless technology alone will

thus not be able to substitute for all the work done by a truck driver. In some cases, it is

feasible to split up the task bundle of an occupation, and have some parts executed by

machines while other tasks remain in the hands of humans. One example is bank tellers.

- 25 -

The work that used to be done by a teller is now split into dispensation of cash, a

routine task that is handled by automated teller machines, and a large set of other

customer services, which are still executed by human employees. A counterexample to

this unbundling of tasks is airline pilots. In 1947, the first plane flown by an autopilot

crossed the Atlantic Ocean. But in the almost seven decades since, the job of the airline

pilot has not disappeared because pilots are still needed onboard a plane in order to react

to unforeseen conditions such as a failure of the aircraft’s engines or instruments.

VIII. How Can Workers Succeed in a Computerized Labor Market?

Computers are transforming the occupational composition of the labor market. Young

people who enter the labor market today face a very different set of job opportunities

than their parents a generation ago. Many middle-wage occupations in production and

clerical work hire fewer workers than they used to, and employment polarization is

hence particularly pronounced among the young, who are becoming disproportionately

concentrated in high-wage and low-wage jobs (Autor and Dorn, 2009). The continued

availability of jobs in low-skilled service occupations is generally good news, as it

offers employment opportunities even to workers who have little formal education. Yet

the workers who stand to gain most are those in the highly paid managerial and

professional occupations whose productivity has risen thanks to computer technology.

It is thus an easy policy recommendation that more schooling would help new

cohorts of young workers. After all, the rise of the wage differential between workers

with and without university education in many countries suggests that the growth in the

relative supply of highly educated workers has fallen short of the growth in relative

demand for these workers.

The solution should however not just be more education, but also different education.

Computers have changed, and will continue to change the demand for job tasks in the

labor market. Therefore, education should build skills in those areas where human

capabilities remain superior to machines, and not in dimensions where machines have

the edge. An education that emphasizes rote memorization and mental arithmetic is no

longer able to produce skilled workers who can hope to outdo computers in terms of

information storage or calculation. Victories of computers over extremely accomplished

- 26 -

humans in quiz shows and chess championships have impressively shown the

tremendous advantage that machines now have in such tasks.

Humans instead retain an advantage over machines when it comes to problem solving,

creativity, and interaction with other humans. An education that prepares young people

for the task demands of the 21st Century should thus seek to strengthen skills in these

areas, for instance by fostering problem solving skills and communication skills through

case study projects, group work, and other modern forms of teaching that provide

alternatives to lectures and memorization.

A greater focus on individualized customer interaction and on innovation and

problem solving also provides a perspective for some of the shrinking middle-wage

occupations. Workers in clerical and production jobs who combine the routine tasks

which are typical for these occupations with complementary non-routine tasks can

provide an attractive task bundle that cannot as readily be replaced by technology, at

least not without a substantial loss in quality. A machine operator who has a thorough

understanding of the machine’s operation, and the production process it is embedded in,

is more valuable and harder to replace than an operator who is just familiar with a few

buttons on the machine’s operating panel. The former will be able to quickly resolve

problems and may contribute to finding improvements that raise the efficiency of the

production process. Similarly, a salesperson who expertly advices customers and

responds to individual customer requests will not as easily lose her job to a machine as a

colleague who merely swipes credit cards at the cash register. Such virtuous bundles of

skills do not require a university education, but can be obtained from a high-quality

vocational training that combines hands-on experience on the job with schooling that is

tailored to the need of a specific occupation.

The rapid adoption of CNC machines, personal computers and robots have

transformed the organization of work in many workplaces during the last four decades,

and will likely continue doing so in the future. While automation has not caused the

large increase in long-term unemployment that some feared, it nevertheless presents

major challenges to society. Economic inequality has increased as the labor market has

polarized into a set of highly paid and a set of lowly paid occupations. Machines have

overtaken humans in their capability to cheaply and precisely execute routine tasks, and

jobs that specialize in these tasks have been irreversibly lost. Yet the employment

outlook will remain favorable for workers whose strengths in interpersonal interaction,

- 27 -

flexibility, and creativity give them uniquely human advantages.

References

Acemoglu, Daron and David Autor. 2010. “Skills, Tasks and Technologies:

Implications for Employment and Earnings.” In: Orley Ashenfelter and David

Card (eds.), Handbook of Labor Economics Volume 4, Amsterdam: Elsevier.

Autor, David and David Dorn. 2009. “This Job is “Getting Old”: Measuring Changes in

Job Opportunities Using Occupational Age Structure.” American Economic

Review Papers and Proceedings, 99(2): 45-51.

Autor, David and David Dorn. 2013. “The Growth of Low Skill Service Jobs and the

Polarization of the US Labor Market.” American Economic Review, 103(5):

1553-1597.

Autor, David, David Dorn and Gordon Hanson. 2013. “The Geography of Trade and

Technology Shocks in the United States.” American Economic Review Papers

and Proceedings, 103(3): 220-225.

Autor, David, David Dorn and Gordon Hanson. Forthcoming. “Untangling Trade and

Technology: Evidence from Local Labor Markets.” Economic Journal.

Autor, David, Frank Levy and Richard Murnane. 2003. “The Skill Content of Recent

Technological Change: An Empirical Exploration.” Quarterly Journal of

Economics, 118(4): 1279-1333.

Brynjolfsson, Erik and Andrew McAfee. 2011. “Race Against the Machine.” Lexington

MA: Digital Frontier Press.

Brynjolfsson, Erik and Andrew McAfee. 2014. “The Second Machine Age.” New York

NY: W.W. Norton & Company.

Frey, Carl Benedikt and Michael A. Osborne. 2013. “The Future of Employment: How

Susceptible are Jobs to Computerization?” Working Paper, Oxford University.

Goldin, Claudia and Lawrence F. Katz. 2008. “The Race Between Education and

Technology.” Cambridge MA: Belknap Press.

Goos, Maarten and Alan Manning. 2007. “Lousy and Lovely Jobs: The Rising

Polarization of Work in Britain.” Review of Economics and Statistics, 89(1), 113-

133.

- 28 -

Goos, Maarten, Alan Manning and Anna Salomons. 2009. “Job Polarization in Europe.”

American Economic Review Papers and Proceedings, 99(2): 58-63.

Gordon, Robert J. 2012. “Is U.S. Economic Growth Over? Faltering Innovation

Confronts the Six Headwinds.” NBER Working Paper no. 18315.

Gordon, Robert J. 2014. “The Demise of U.S. Economic Growth: Restatement, Rebuttal,

and Reflections.” NBER Working Paper no. 19895.

Nordhaus, William D. 2001. “The Progress of Computing.” Cowles Foundation

Working Paper no. 1324.

- 29 -

【Presentation 2】

Non-standard Work, Job Polarisation and Inequality

Michael Förster(OECD)

Abstract This paper provides evidence for the implication of trends in non-standard work for

individual and household earnings and income inequality. It first presents the socio-demographic characteristics of non-standard workers before discussing the contribution of non-standard work to overall changes in employment. It shows that, in a majority of OECD countries, standard jobs have disappeared in the middle of the distribution in terms of wages and skill, while non-standard jobs have contributed to an increase in jobs at both ends of the distribution. Non-standard jobs tend to pay lower wages than standard jobs, especially at the bottom of the earnings distribution, thereby raising earnings inequality. The paper then looks at the impact of non-standard work on household incomes and shows that non-standard workers living alone or with other non-standard workers suffer from higher chances of low income and poverty. Finally, the paper examines the work incentives and adequacy effects of tax and benefit rules. It finds that some non-standard workers, such as the self-employed, usually face different statutory rules and shows that taxes and benefits reduce poverty gaps for non-standard workers but create work disincentives for moving from inactivity to work. ∗

I. Introduction and Key Findings

Changes in earnings – which constitute three-quarters of household income – and in

labour market conditions have been identified as the most important direct driver of

rising income inequality. This concerns, in particular, changes in the distribution of

gross wages and salaries, which have become more dispersed in most OECD countries

in the past 25 years. But this is also linked to changes in employment patterns, working

conditions and labour market structures. For instance, growing levels of non-standard

work, such as part-time work, casual work and work on temporary contracts, may help

to explain the puzzle of increasing inequality despite aggregate employment growth

prior to the global economic crisis.

The effects of the rising share of employment in non-standard work (NSW)

** This paper is published as Chapter 4 of OECD (2015), In It Together: Why Less Inequality

Benefits All, OECD Publishing, Paris. http://dx.doi.org/10.1787/9789264235120-en ISBN 978-92-64-23266-2 (print), ISBN 978-92-64-23512-0 (PDF)

** The statistical data for Israel are supplied by and under the responsibility of the relevant Israeli authorities. The use of such data by the OECD is without prejudice to the status of the Golan Heights, East Jerusalem and Israeli settlements in the West Bank under the terms of international law.

- 33 -

arrangements have gained centre stage in policy debates in recent decades. Since the

1980s, labour markets in OECD countries have been subject to major structural changes.

The employment protection legislation (EPL) became less strict in countries where

protection had been relatively strong to start with, while countries where the strictness

of the EPL was below average in 1985 tended to stick with a similar policy in the late

2000s (OECD, 2011). Alongside these institutional changes, demographic and societal

developments – ageing and higher female labour market participation – have also

profoundly modified the labour force. Finally, structural changes in employment due to

growth in services and knowledge jobs, a greater use of ICTs and just-in-time delivery

have all had implications for the demand and supply drivers of atypical forms of work.

As NSW is often portrayed as being associated with lower earnings and with job

insecurity, this has drawn attention to its potentially adverse impact on the distribution

of individual earnings as well as of household income more generally.

Evidence from OECD (2011) has shown the impact of non-standard work on the

level of overall earnings inequality: adding the earnings of part-time workers to the

distribution of full-time employees increased earnings inequality by almost 20%, and

adding self-employed workers increased inequality by a further 5%. In addition, policy

reforms such as weaker employment protection for temporary contracts have tended to

increase employment opportunities but were associated with wider wage inequality.

There is however a lack of empirical evidence on the detailed channels through which

non-standard work may affect the distribution of individual and household income.

Non-standard employment might be associated with poorer labour conditions (wages,

working time, job security, leave entitlements, etc.), particularly in the case of dual or

segmented labour markets, if firms use such arrangements for cost or flexibility reasons

or as a probationary device. On the other hand, part-time, temporary and self-

employment arrangements may be attractive to certain workers, and workers might

choose this type of employment to achieve a better work-family life balance, higher life

satisfaction or, in the case of self-employment, a greater sense of control. The degree of

mobility between both segments is also likely to influence whether there are persistent

wage differentials between both sectors.

The chapter is organised as follows. Section II defines different forms of

non-standard work and the demographic composition of these workers. Section III

analyses the extent to which employment growth stems from non-standard work and

- 34 -

how NSW contributes to job polarisation. Section IV looks at the question of whether

non-standard jobs pay less and whether such jobs improve employment prospects. It

also discusses the implications for the distribution of earnings. The contribution of

NSW to household income inequality and poverty is discussed in Section V. Finally,

Section VI presents the impact of tax-benefit policies on income adequacy and on work

incentives for non-standard workers.

The key findings from this chapter are:

▪ Non-standard work (temporary, part-time and self-employment taken together)

represents one-third of total employment in the OECD, ranging from a low of

under 20% in the eastern European countries (except Poland) to 46% or more in

the Netherlands and Switzerland. Women (especially part-time), youth (especially

temporary jobs) and workers with lower level of education are over-represented in

NSW, as are workers in small firms.

▪ Close to half of employment growth since the 1990s and up to the global

economic crisis has been in the form of non-standard work; the share reaches

almost 60% of if the crisis years are included.

▪ Non-standard work contributes to job polarisation, i.e. to jobs disappearing in the

middle of the distribution relative to those at the bottom and at the top: nearly all

employment losses in middle-skill occupations were in standard work contracts,

while job gains in high- and low-skill jobs were mainly in NSW.

▪ Non-standard work is not always a stepping stone to stable employment. Temporary

contracts increase the chances of acquiring a standard job compared with remaining

unemployed, but a part-time job or self-employment does not increase the chances

of a transition to a standard job.

▪ Non-standard workers are worse off in terms of many aspects of job quality. They

tend to receive less training and, in addition, those on temporary contracts have

more job strain and have less job security than workers in standard jobs. Earnings

levels are also lower in terms of annual and hourly wages but, for part-timers,

once other demographic and job characteristics are taken into account, the

differences in hourly wages tend to disappear. On the other hand, compared with

permanent workers, temporary workers face substantial wage penalties, earnings

instability and slower wage growth.

▪ Non-standard work tends to lower wages at the bottom of the earnings distribution,

- 35 -

while the effect is often neutral at the top, thereby contributing to increased

individual earnings inequality.

▪ Adding earnings from non-standard work to households where standard work is

the norm increases household earnings inequality by three Gini points on average

and help explain about 20% of household income inequality.

▪ Slightly more than half of non-standard workers are the main breadwinners in

their household, and the great majority of them (80% or more) live in a household

with two persons or more, including children.

▪ While not all low-wage non-standard workers live in low-income households,

households with non-standard work arrangements are overrepresented at the lower

end of the household income distribution. But the household constellation matters:

low-income and poverty risks are five and ten times higher respectively if NSW is

the main source of earnings rather than if NSW live with a standard worker.

▪ About 60% of working poor households are households where the main source of

earnings is NSW.

▪ Non-standard workers face different statutory and effective entitlements to taxes

and benefits in comparison to workers in standard jobs. For the self-employed,

this is due to structurally different policy rules, while for part-timers it is the

particular circumstances of these jobs that lead to different outcomes in terms of

adequacy and incentives. In most countries, taxes and benefits significantly

reduce in-work poverty gaps for NSW, though they are more effective for part-

time than for self-employed workers.

II. A Snapshot of Non-standard Work

There is no universally accepted definition of non-standard work arrangements. In its

broadest sense, NSW may be defined as all employment relationships that do not

conform to the “norm” of full-time, regular, open-ended employment with a single

employer (as opposed to multiple employers) over a long time span. Such a broad

definition of non-standard employment includes three partly overlapping types: a) self-

employment (own-account workers1); b) temporary or fixed-term contracts; and c) part-

1 Employers are excluded from the analysis since transitions between employers and standard

- 36 -

time work.2 It is clear that such a definition comprises very different groups of workers:

for some (e.g. involuntary part-timers), this employment may have job characteristics

associated with precariousness (low pay, instability); for others (e.g. voluntary part-

timers with long tenure), such a job may actually be a desired outcome. Furthermore,

transforming this definition into comparable cross-country statistics is not without

problems, and the process is constrained by data availability (Box 1).

Box 1. Defining non-standard forms of employment

Figures on non-standard employment are not easily comparable across countries because

of national differences in definition and measurement. The difficulties in defining non-

standard work on a comparable basis are accentuated if attempts are made to link non-

standard forms of employment with wages and household earnings, as few data sources

contain information on both employment and wages over time. Labour force surveys or

household surveys typically ask respondents first, to classify themselves as employees or

self-employed according to their status in their main job, and then ask employees to report

on their type of contract and their working hours. Self-reporting errors may be present in

such information, and figures should be used to indicate broad levels and trends across

countries.

In its broadest sense, NSW arrangements are defined by what they are not: full-time

dependent employment with a contract of indefinite duration, or what is generally considered

the “standard” work arrangement. This definition generally implies that self-employed own-

account workers and all part-time workers fall under “non-standard workers”. While

problematic – as this lumps together precarious and non-precarious forms of work – this

convention is followed by a large part of academic international and national research (e.g.

Houseman and Osawa, 2003; Wenger, 2003; Görg et al., 1998; Kalleberg et al., 1997;

Kalleberg, 2000; Leschke, 2011), as well as by international organisations (e.g. International

Labour Organisation, World Bank, Eurofound).

As noted above, this chapter breaks down non-standard employment into three separate

categories: 1) self-employed (own-account), 2) temporary full-time employees and 3) part-

workers are likely to be small and employers differ from other workers in their remuneration (receiving earnings as well as business income). In the OECD they represent an average of 4% of total employment for the working-age population.

2 Student workers and apprentices are excluded from the analysis, as they may increase the share of part-time workers and temporary workers. They represent on average 2% of total employment.

- 37 -

time employees (including permanent and temporary contracts). Unpaid family workers are

excluded from the analysis. Where possible, a distinction is made to break down the

category of part-time employees into voluntary and non-voluntary part-timers, as well as

part-timers on temporary and permanent contracts.

The distinction between different forms of employment has become increasingly blurred.

There is a growing grey area, for instance between self-employment and wage employment

(OECD, 2000). The growth in the numbers of self-employed contractors working for just

one company or franchisees constitute groups on the borders of dependent and self-

employment.

Temporary jobs for the purpose of this analysis are defined as dependent employment of

limited duration, including temporary work agency, casual, seasonal or on-call work.

Definitions across countries outside the European Union are not harmonised and are based

on different approaches. For Korea, workers in temporary jobs include fixed-term jobs or

jobs of a limited duration, which is close to so-called contingent workers, as well as other

atypical workers, i.e. temporary agency workers, individual contract workers, at-home

workers, on-call workers and others. In the case of Australia, a broad definition of

temporary work includes jobs of fixed-term duration, those employed through a labour hire

or a temporary work agency as well as casual workers. Casual workers may lack

entitlements to key fringe benefits such as paid vacation or sick leave or may not be

protected by legislation against unfair dismissal, but might otherwise have continuous and

stable employment, and are therefore one form of atypical or NSW. In this respect, this

definition follows the work undertaken by the Australia Productivity Commission (2006) in

classifying casual work as one form (and the most sizeable one) of non-standard work.

Part-time employees are defined based on their weekly working hours, namely working

less than 30 hours per week. This may differ from national definitions which use different

hours thresholds. Part-time work is also further disaggregated into part-time temporary and

part-time permanent jobs when the data is available.

Employment in NSW arrangements in the OECD today is sizeable, comprising on

average one-third of total employment (Figure 1). Permanent full-time employment

remains nonetheless the norm in a majority of OECD countries, although there is

substantial diversity across countries. In the Netherlands, more than one job in two is

non-standard (though more than half of these are permanent part-time jobs), while in

some eastern European countries the share is less than one in four jobs.

- 38 -

1. Different forms of non-standard work and their prevalence across the OECD

The three main forms of non-standard work, i.e. self-employment, temporary

employment and part-time work, account for fairly similar shares on average in the

OECD, but they differ greatly by country (Figure 1, Panel A). For instance, self-

employment is the most prevalent form of non-standard work in Greece, Turkey and the

Czech Republic. On the other hand, part-time employment represents close to or over

60% of total non-standard employment in the Netherlands, the Nordic countries (except

Finland), Belgium, Luxembourg and Switzerland, while it is only 12% in Korea and

Poland. In Australia, where a broad definition of temporary employment also includes

casual workers (Box 1), this type of work accounts for 85% (43%) of part-time

(full-time) workers with a temporary employment contract.

Part-time workers are a very heterogeneous group with very different labour supply

patterns. Some people work part-time because they wish to do so and would not take on

full-time employment, while others do so because there is no full-time employment

available. On average, involuntary part-time accounts for close to 30% of total part-time

employment, with just under half of this associated with a temporary contract (Figure 1,

Panel B). There are, however, large variations across countries. In Greece, Spain and

Italy, over 60% of part-timers want to work more hours but could not find full-time jobs.

In contrast, in Austria, Luxembourg, the Netherlands, Belgium and Switzerland, part-

time work is predominantly voluntary and is associated with a permanent contract.

The characteristics and preferences of workers, as well as institutional factors and the

sectoral composition of employment, all play a role in explaining cross-country

differences in the share of non-standard workers. The tax wedge, product market

regulations, employment protection legislation and the size of the public sector have been

found to have an impact on the incidence of different categories of non-standard work.

For instance, there is a well-established negative relationship between the level of GDP

and the self-employment rate (Acs et al., 1994). In addition, self-employment rates tend to

be high in countries where the public sector is small, taxation levels are high, product

market regulation (PMR) is tight3 and the rule of law is weakly enforced (OECD, 1999;

3 While high levels of PMR could be detrimental to business activities, regulations can be used to

protect small-sized firms from large-sized competitors (Torrini, 2005).

- 39 -

Schuetze, 2000; Torrini, 2005). Temporary employment tends to be higher in countries

with stricter employment protection legislation for regular workers (OECD, 2014; Chen et

al., 2015, forthcoming). One explanation put forward is that the employment protection of

permanent jobs has a minor impact on total employment, but leads to a stronger

substitution of temporary jobs for permanent jobs (Cahuc et al., 2012).

Figure 1. Share of non-standard employment by type, 2013

Panel A. Non-standard forms of employment as a percentage of total employment

Panel B. Part-time employment by type

Note: Sample restricted to paid and self-employed (own account) workers aged 15-64, excluding

employers, student workers and apprentices. Breakdown of part-time employment by voluntary/involuntary is not possible for non-European countries. Panel A. For Australia, 42.6% of full-time temporary contract are casual; and 85.2% of part-time temporary employees are casual.

Source: European Union Labour Force Survey (EU-LFS, 2013), Household, Income and Labour Dynamics in Australia (HILDA, 2012), Japan Labour Force Survey “Basic Tabulation” (2012), Korean Labor & Income Panel Study (KLIPS, 2009) an