2.1.1. New technologies are rapidly permeating the world of work

Over the past two to three decades, the pace of technological progress and the speed of its diffusion across countries have been startling. For instance, while it took over seven decades for phone penetration to go from 10% to 90% in US households, it took only about fifteen years for mobile phones and just over 8 years for smartphones.3 Such technological leaps have had major impacts on the way people work and live.

The growth in information and communication technologies (ICT) use in the workplace provides a clear indication of how quickly new technologies permeate the workplace. From 1995 to 2007, the level of ICT capital services per hour worked more than doubled in every country analysed before growing at a slower pace (Figure 2.1). There are, however, substantial country differences in the pace of technology adoption. While in Hungary, Japan, and Slovenia, ICT levels increased by just over 150% over the period, the increase was as much as 300% in the Netherlands, the Czech Republic, Ireland, and Germany and above 350% in the United States, Belgium and the United Kingdom.

The diffusion of industrial robots perhaps best epitomises technological penetration and fears of job automation in the workplace. While robots have been on factory floors for decades, their diffusion has recently accelerated and spread beyond manufacturing. As one example, supermarkets have started to employ robots as shop assistants, and a number of companies are piloting cashier-less stores – e.g. Browne (2018[7]). The capabilities of robots are also expanding within the manufacturing sector. For example, certain robots are now able to move by themselves around the factory floor (Brynjolfsson and McAfee, 2014[8]). Data from the International Federation of Robotics show that orders of industrial robots have increased fivefold between 2001 and 2017, and such a trend is projected to accelerate further (Figure 2.2).4 Coupled with the increasing share of national income going to capital (as opposed to labour, as discussed below), such a trend directly fuels an important policy debate on the concentration of capital ownership.

These are only some examples of new technologies that have emerged and had an impact on the world of work. Going forward, further leaps in the development of artificial intelligence (AI) are likely to have applications in a broad range of domains, encroaching upon many more tasks that previously could only be performed by humans, with the potential to drive as-yet unforeseen changes in the world of work.

2.1.2. The world has become an increasingly integrated place

In conjunction with the diffusion of new technologies, the world economy has become increasingly integrated through international trade. As a share of GDP, international trade has risen across the OECD area in recent decades (Figure 2.3), and many emerging economies have become major players in the world market, both as exporters and importers. Industrial production has become increasingly integrated at the international level, with the world economy organised in global value chains (GVCs) whereby the different stages of the production process are distributed across countries and regions.

The integration of product, service, financial and technology markets fundamentally impacts labour markets around the world. It allows for and encourages greater specialisation in what gets produced and how it is produced with consequences for the skills workers require and the types of jobs that are created. Overall, more jobs are created through trade than are lost. For example, it has been estimated that, on average, 42% of business sector jobs in OECD countries were sustained by consumers in foreign markets in 2014 (OECD, 2017[9]). Yet the actual and potential negative effects of trade on certain occupations and local markets deserve careful scrutiny by policy makers as this constitutes the main cause of a growing discontent with globalisation worldwide.5 Such discontent is often intertwined with the fear of automation. Since technological progress and globalisation have historically progressed hand in hand and reinforced each other, it is difficult to isolate their individual effects (OECD, 2017[10]).

copy the linklink copied!

Figure 2.1. The rapid spread of information and communication technologies in the workplace

ICT capital services per hour worked, index (1995 = 100), 1995 to 2015

Note: ICT: information and communication technologies. ICT capital intensity per hours worked refer to the CAPIT_QPH variable in the EU KLEMS database. Data for Canada are taken from the World KLEMS database. Data series were extended using growth of the numerator and denominator of the ICT intensity ratio using the various releases of the EU KLEMS database (2009, 2013, and 2016). The 2009 EU KLEMS release covers the largest number of countries, covering the period from 1995 to 2007. Additional data were taken from later releases of EU KLEMS for the following countries: Austria, Belgium, the Czech Republic, Denmark, Finland, France, Germany, Italy, the Netherlands, Slovenia, Spain, Sweden, the United Kingdom and the United States. Values for Denmark have been adjusted to account for abnormally large increases in ICT intensity within the mining industry.

Source: EU KLEMS growth and productivity accounts, World KLEMS.

 StatLink https://doi.org/10.1787/888933965951

copy the linklink copied!

Figure 2.2. The march of the robots

Estimated worldwide annual supply of industrial robots, thousands of units

*: forecast

Source: International Federation of Robotics (IFR), https://ifr.org/.

 StatLink https://doi.org/10.1787/888933965970

copy the linklink copied!

Figure 2.3. International trade keeps rising

Trade in goods and services in selected OECD countries, 1975-2017

Source: OECD (2019[11]), Trade in goods and services (indicator), https://doi.org/10.1787/0fe445d9-en.

 StatLink https://doi.org/10.1787/888933965989

2.1.3. OECD countries are ageing

The transformation of the labour market is occurring against the backdrop of rapid population ageing in both advanced and some emerging economies. In 1980, there were 20 persons aged 65 and over for every 100 people of working age (20-64) on average across the OECD (Figure 2.4); by 2015 this number had risen to 28 and it is projected to almost double between 2015 and 2050 (OECD, 2017[12]). The challenge of rapid population ageing is particularly acute in Greece, Italy, Japan, Korea, Portugal and Spain, as well as in China. In contrast, emerging economies such as Indonesia, South Africa, and India will continue to face the demographic challenge of integrating large numbers of young people into the workforce. They will need to take advantage of the demographic dividend of a relatively young population to boost growth and prepare for the transition to a much older population.

The effects of technological progress and its global diffusion will further contribute to population ageing. Largely as a result of technological advances that increased productivity and living standards, as well as raising the quality and availability of health care, average life expectancy at birth increased across the OECD from 69 years in 1965 to 80 years half a century later.6 Going forward, scientists anticipate that new gene-editing technologies could lead to further improvements in the diagnosis and treatment of diseases, leading to longer life expectancies (Broad Institute, 2018[13]; Sanders, 2016[14]). Stronger research networks on a global scale and, more generally, the diffusion of knowledge across the world will allow these advances to reach an ever greater share of the global population, as incomes and access to health care increase in emerging economies.7 But such improvements are not inevitable, as some changes in lifestyle resulting in a rising incidence of obesity and overuse of opioids have slowed or even halted the rise in life expectancy in a few advanced economies (OECD, 2018[15]).8

These demographic trends affect the labour market in terms of technology adoption and consumption patterns. In countries with ageing populations, shortages of qualified labour may arise as the number of older workers retiring rises relative to the number of young people entering the labour market. These shortages may in turn lead to faster automation or stronger pressures to attract immigrant workers. Acemoglu and Restrepo (2017[16]) show that countries with the most rapidly ageing populations have also been among the fastest to adopt industrial robots (and consequently they suggest that an ageing population may not necessarily be a harbinger of slower economic growth). Ageing will also have a direct impact on consumption: demand is likely to shift from durable goods (such as cars) towards services (such as health care). As preferences adjust, so too will trade and the relative importance of different industries.9 All of these factors will have an impact on skill demands and the types of jobs that will be created.

copy the linklink copied!

Figure 2.4. Many countries are ageing rapidly

Projected change in the old-age dependency ratio, 1980-2050

Note: The old-age dependency ratio is defined as the number of people aged 65 and over per 100 people of working-age (20-64).

Source: United Nations World Population Prospects: The 2017 Revision, https://population.un.org/wpp/.

 StatLink https://doi.org/10.1787/888933966008

2.1.4. The global population will increase and migration pressures are likely to grow

As people live longer across the world while fertility rates remain high in a number of developing countries, the global population is expected to increase further. According to the United Nations’ 2017 World Population Prospects, the expected global population will be 9.7 billion in 2050, a 30% increase from 7.5 billion today.10 Whereas developing countries will account for the bulk of this increase, the population of OECD countries is expected to increase by less than 10%, from 1.3 billion to 1.4 billion people.

Thus, depending on infrastructure, economic opportunity, and policy choices, migration flows may radically change the makeup of the population in advanced economies. As one example, over half of workers in the Silicon Valley with a degree in science, technology, engineering or mathematics (STEM) are foreign born (Melville, Kaiser and Brown, 2017[17]). In 2017, about 258 million people around the world were living outside their country of birth, and about half of all these migrants were living in OECD countries (OECD, 2018_[14]). In 2017, more than 5 million people settled permanently in the OECD. In addition, more than 4 million temporary foreign workers were recorded in OECD countries in 2016 in order to fill skills shortages, and more than 3 million international students are enrolled in a higher education establishment in an OECD country. Given the widening demographic imbalances described above, migration flows may further intensify in the coming decades and pose fundamental policy challenges.

With respect to the issues addressed in this chapter, while migrants may help countries with ageing societies to overcome skill shortages, they are also heavily exposed to some key risks. First, in the majority of OECD countries migrants are more concentrated than natives in jobs at high risk of automation. In European OECD countries, for instance, 47% of foreign-born workers are in occupations that primarily involve routine tasks and most exposed to automation (OECD, 2017[18]). Second, migrants are more likely to be in low-skilled jobs, which are frequently of low-quality, despite their relatively high educational level (OECD, 2018[19]).

2.2.1. In spite of the continuous transformation of the labour markets, employment has historically been growing

Despite periodic waves of anxiety regarding labour displacement due to technological progress and globalisation, most OECD countries have seen their employment rates – the share of people of working age in employment – on an upward trajectory over past decades with the notable exception of the United States (OECD, 2018[22]) (Figure 2.5). In fact, labour demand rose strongly in line with the increase in labour supply as a result of a greater participation of women and older people. In the United States, the participation rate of women increased from 42% in 1960 to 68% in 2017. Across the OECD, female labour force participation grew by 10 percentage points since the early 1980s (from 54% in 1983 to to 64% in 2017). Countries such Spain and Ireland, where female labour force participation grew from less than 40% to more than 65% over this period, experienced the most striking results in this respect. On the other hand, a number of OECD countries there is still ample scope for further rise in women’s participation (in Turkey, for example, fewer than 4 in 10 women participate in the labour market, relative to about 8 in 10 men).

copy the linklink copied!

Figure 2.5. Employment rates have been rising in recent decades

Employment-to-population ratio, age 15/16-64

Note: Brazil data for 2000 and 2010 are from 2001 and 2011 respectively. Mexico data for 1990 are from 1991. South Africa data for 2000 are from 2001. The OECD average in the unweighted average of OECD member countries in the year indicated.

Source: OECD Employment Database, www.oecd.org/employment/database.

 StatLink https://doi.org/10.1787/888933966027

The increase in overall employment has occurred in parallel with rapid technological progress. The previous section offered an overview of the significant rise in ICT, in the use of robots at work, and the increasing deployment of artificial intelligence (AI). Such technologies have been directly responsible for substantial job destruction, sometimes contributing to significant employment decline in certain industries, ranging from textile production to complex electrical equipment manufacturing (OECD, 2017[10]). At the same time, by increasing productivity and raising incomes, they have generated additional demand for goods and services that has given rise to even more jobs (see Box 2.1 for a fuller discussion of the mechanisms through which technological progress destroys and creates jobs). Recent research indicates that the digital revolution has contributed significantly to job creation: 4 out of 10 jobs were created in digitally-intensive industries over the past decade (OECD, 2019[23]).

Technological progress has also contributed to higher female employment. Since women have historically borne the brunt of domestic work, increased productivity in home production (e.g. thanks to washing machines, dishwaters, etc.) is among the factors that may have contributed to the increase in their participation in the labour market. Moreover, in the past, automation has disproportionately impacted jobs typically held by men (e.g. factory workers, construction workers), while jobs in which women are overrepresented (e.g. health workers, service workers) have been buffered to a larger extent (OECD, 2017[24]). This trend, however, is changing. Recent OECD work shows that the expected risk of job displacement due to automation in the coming decades does not show significant differences by gender (OECD, 2017[24]).

Similarly, trade openness has historically gone hand-in-hand with increasing employment, despite the disruptive effects that import competition has had on specific industries. In a review of 14 multi-country econometric studies on the relationship between trade and economic performance, Newfarmer and Sztajerowska (2012[25]) find no negative impacts of trade on job quantity. On the contrary, greater openness to trade can play an important role in creating better jobs, increasing wages in both rich and poor countries, and improving working conditions. The risk of focusing on aggregate outcomes, however, is to overlook that technological progress and trade openness have not benefited all workers equally and have had strong negative impacts on certain industries and regions. This is a key challenge facing policy makers and represents one of the central issues highlighted in this report.

Another megatrend that is expected to affect jobs in the coming decades is the transition towards a low-carbon economy. In light of growing concerns about climate change and global warming, a number of countries have committed to strategies for limiting average global temperature increases to 1.5 degrees Celsius above pre-industrial levels (United Nations, 2016[40]). This will result in job losses in industries involving carbon-intensive emissions but create jobs in new forms of greener energy production and in energy conservation. Estimates of total job reallocation, however, suggest that the transition towards a green economy will have relatively low impacts on total job quantity — the difference between job creation and job destruction amounts to about 0.3% of employment in OECD countries and 0.8% in non-OECD countries (Château, J., Bibas and Lanzi, 2018[41]; Botta, 2018[42]; Château, Saint-Martin and Manfredi, 2011[43]).11 In fact, the overall impact on employment might be positive. The greenness of jobs index (goji) developed using German data, for instance, suggests that Germany’s transition to a greener economy has been correlated with higher employment growth and a slight increase in wages (Janser, 2018[44]). Yet, as for the impact of trade openness, the estimated job losses from green policies will be concentrated in specific industries and types of work, possibly fostering inequality (as discussed in the previous section).

2.2.2. Is this time different? The recent wave of anxiety regarding automation

While the historical evidence suggests that broad technological unemployment and a large negative impact of globalisation on overall employment are unlikely, the most recent wave of anxiety regarding automation is fuelled by the perception that technological change is faster paced and broader based than in the past, making more jobs automatable than previously thought (Brynjolfsson and McAfee, 2011[3]; Mokyr, Vickers and Ziebarth, 2015[4]). Some authors have even argued that in some cases automation may be excessive, with firm leaders inefficiently over-investing in adopting the latest technologies, and under-investing in preparing for the jobs of tomorrow and helping workers prepare for them, with the consequence of generating negative externalities for society at large (Acemoglu and Restrepo, 2017[45]; Acemoglu and Restrepo, 2018[32]).

In light of these concerns, several authors have attempted to predict what share of jobs may be automated as a result of new technologies permeating the workplace. A widely cited analysis in this field is the one by Frey and Osborne (2017[20]), who estimate that almost half of all jobs (47%) in the United States are at risk of being substituted by computers or algorithms within the next 10 to 20 years. These estimates are constructed using experts’ assessment of the probability that different occupations can be automated.12 Critics of these large estimates argue that occupations as a whole are unlikely to be automated, as not all workers in the same occupation perform the same tasks and hence face the same risk of their jobs being automated (Autor and Handel, 2013[46]). For example, the job of one worker may involve more face-to-face interaction or autonomy than the job of another worker in the same occupation. This may partly explain why the predictions of Frey and Osborne about the pattern and depth of job automation have not yet shown up in the labour market (Manning, forthcoming).13

2.2.3. The latest OECD results show that around 14% of jobs are at risk of complete automation but many more will be affected by deep changes

An alternative approach to estimate the number of jobs at risk of automation is to directly analyse the task content of individual jobs instead of the average task content within each occupation (Arntz, Gregory and Zierahn, 2016[47]; Nedelkoska and Quintini, 2018[21]).14 Using this approach, the OECD estimates that the share of jobs at high risk of automation (i.e. those with a probability of being automated of at least 70%) is around 14%, on average, across the OECD (Figure 2.6). The figures for individual countries range from 6% in Norway to 34% in the Slovak Republic. These figures, however, only capture potential job destruction and do not account for the (possibly larger) number of jobs that technology generates (see Box 2.1; and Box 2.2 for a focus of this discussion on emerging economies).

In addition, a large share of existing jobs may change substantially in the way they are carried out. The OECD estimates that 32% of jobs, on average across the OECD, may see a large share of their tasks be automated while entirely new tasks may emerge (Figure 2.6). The analysis also highlights that the risk of automation is higher among low-skilled workers, which may further increase disparities in the labour market (Nedelkoska and Quintini, 2018[21]).

While the risk of automation may not be as high as thought by some, it is essential to recognise that there is considerable uncertainty around these estimates. The consequence of such uncertainty is that responsible policy making should be prepared for a range of possible future outcomes and aim to increase the resilience of the labour market in the face of future transformations. In this regard, providing workers with adequate training opportunities throughout their careers will play a crucial role. According to the OECD Survey of Adult Skills (PIAAC), more than 50% of the adult population, on average in 28 OECD countries, can only carry out the simplest set of computer tasks, such as writing an email and browsing the web, or have no ICT skills at all (OECD, 2016[48]). Existing systems of adult education are often unable to bridge disparities among workers and may, in fact, contribute to widen them, as higher-skilled workers typically receive more training (OECD, 2013[49]). How to make adult learning systems more effective and inclusive is the subject of Chapter 6.

copy the linklink copied!

Figure 2.6. Jobs at risk of automation in OECD countries

Share of jobs which are at a high risk of automation or a risk of significant change (%)

Note: Jobs are at high risk of automation if the likelihood of their job being automated is at least 70%. Jobs at risk of significant change are those with the likelihood of their job being automated estimated at between 50 and 70%. Data for Belgium correspond to Flanders and data for the United Kingdom to England and Northern Ireland.

Source: OECD calculations based on the Survey of Adult Skills (PIAAC) (2012), http://www.oecd.org/skills/piaac/; and Nedelkoska, L. and G. Quintini (2018[21]), “Automation, skills use and training”, OECD Social, Employment and Migration Working Papers, No. 202, https://doi.org/10.1787/2e2f4eea-en.

 StatLink https://doi.org/10.1787/888933966046

copy the linklink copied!

Box 2.2. Technological unemployment in emerging economies: slow change but large risks looming on the horizon

The literature on the impact of automation on jobs is largely focused on advanced economies. Emerging economies, however, have very different initial conditions, including a different occupational mix, higher costs of information and communication technologies (ICT) capital, and greater skills shortages (Maloney and Molina, 2016[50]). The key question is whether, in such a context, the job opportunities created by new technologies will outweigh the loss of manufacturing jobs due to automation.

Based on their current stage of development, emerging economies face a higher predicted risk of automation. As economies develop, the industry mix of employment follows a predictable path, shifting labour from low-productivity activities, often in agriculture, to higher-productivity activities, mostly in the manufacturing and in the service sectors. In most emerging economies, agriculture and low value-added industries still make up a large share of employment. Hence, estimates based on occupations (The World Bank, 2016[51]) and more recent ones based on tasks (Nedelkoska and Quintini, 2018[21]) or work activities (McKinsey Global Institute, 2017[52])15 show that emerging economies face a higher risk of automation than more advanced ones. However, the picture is mixed and varies by income level, with countries like China, the Russian Federation, Turkey and Mexico facing a higher proportion of jobs that are potentially at risk of automation.

Nevertheless, while many jobs are “technically automatable”, automation may not be yet economically attractive in many emerging economies. As many emerging economies still have a productive structure biased towards small and medium-sized enterprises (SMEs), the resources that are necessary for costly investments in advanced technology are out of reach for most entrepreneurs. Furthermore, the incentive to innovate is dampened by skills shortages and by the relative abundance of cheap unskilled labour in young and rapidly expanding populations.

copy the linklink copied!

Figure 2.7. The potential cost savings from using robots are significant in some emerging economies

Projected labour-cost savings from adoption of advanced industrial robots (%, 2025)

Source: The Boston Consulting Group (2015[53]), The Shifting Economics of Global Manufacturing: How a Takeoff in Advanced Robotics Will Power the Next Productivity Surge, https://www.slideshare.net/TheBostonConsultingGroup/robotics-in-manufacturing.

 StatLink https://doi.org/10.1787/888933966065

The potential disruptions, however, could be significant. As the cost of industrial robots continues to decline and labour costs increase, the cost savings of using technology to replace labour are starting to become significant also in emerging economies. While the Boston Consulting Group predicts that countries with very low labour costs and expanding young workforces like India and Indonesia will not benefit from replacing humans with robots in the near future, such savings will amount to more than 5% in countries like the Russian Federation and Brazil, reaching up to 18% in China by 2025 (Figure 2.7). In addition, re-shoring of production to advanced economies may contribute to job losses in emerging markets. Although the evidence on re-shoring is still limited and mixed, some signs of this process are already visible, as a number of manufactures are choosing to relocate their production nearer their domestic markets (De Backer et al., 2016[54]). Increasing labour costs and the falling cost of technology may continue to fuel this process, potentially leading some emerging countries to experience premature deindustrialisation, which may leave them in a middle-income trap (Rodrik, 2016[55]). Depending on their occupational and industrial makeup (and on their stage of development), different countries will be impacted at different points in time. Policy makers in emerging economies should start preparing well in advance. Given the lack of adequate safety nets and retraining systems, the effects on workers’ welfare may be significant and foster increased social tensions.

Source: Alonso-Soto (forthcoming[56]), Technology and the future of work in emerging economies: What is different?

2.2.4. The mere existence of a new technology does not imply that it will become pervasive and replace humans at work

An important caveat that needs to be attached to any estimate on the risk of job loss due to automation is that technological diffusion depends on a host of factors that may speed it up or slow it down. Failing to recognise the impact of these different forces may mean falling prey to technological determinism, the idea that technology determines the development of society, its labour market, social structure, and cultural values. While this is certainly true to some extent, other factors, including active policy making and social preferences, play a crucial role. The fact that a technology exists does not necessarily imply that it will spread and change the way people live, and more specifically, the way people work. In fact, existing evidence reveals that the spread of technology is highly heterogeneous across countries, industries, and firms. Constraints to broader technology diffusion may help explaining why technological progress has not translated into productivity gains in recent decades (OECD, 2018[57]).

A number of factors may favour or hinder the spread of different technologies. Above all, market forces driving the relative prices of capital and labour play an important role in determining the profitability of investing in labour-replacing technologies. Countries with relatively low labour costs, for instance, have witnessed a slower process of automation and, also for that reason, do not display a similar pattern of job polarisation as higher industrialised countries (OECD, 2017[10]).

Institutional norms and regulations – for example, product and labour market regulations as well as safety, medical and ethical standards – may prevent certain technologies from becoming prominent in certain countries. Recent OECD evidence shows that certain labour market institutions, including the rate of unionisation and employment protection legislation (EPL), can mediate the effects of technology and globalisation on job polarisation – the fall in the share of middle-skilled jobs relative to low- and, most prominently, high-skilled jobs (OECD, 2017[10]).16

Finally, consumer and societal preferences, as well as ethical norms, will play a crucial role in determining the diffusion of labour-replacing technology. In this regard, an interesting example comes from Eurobarometer data on people’s preferences regarding the deployment of robots in different industries. While the majority of respondents would be happy for robots to be used in areas such as manufacturing and space exploration, the views are much more negative regarding the use of robots in health care and education.

These countervailing trends of market forces, institutional frameworks, and consumer preferences argue against technological determinism: the mere existence of a technology does not imply that it will necessarily become pervasive nor that it will be adopted to replace humans at work (rather than to complement them).

2.2.5. While the number of employed workers may not have fallen, an increasing number of them are under-employed

While overall employment might not be negatively affected by the megatrends, there has been a rise in under-employment.17 Much like employment, changes in under-employment are generally cyclical. However, the gradual post-industrial growth of industries facing volatile demand even over the very short term (such as accommodation and food services) has exposed more workers to the risk of fewer and more variable hours (see Chapter 3). There is some evidence that the global financial crisis has exacerbated this shift. Under-employment rose sharply in many countries hit hard by the crisis, and has been slow to return to pre-crisis levels.

The risk of under-employment has increased for all workers in recent times, but, on average across the OECD, the increase has been larger for the young and those with low or medium education (Chapter 3). Across countries, women remain at a much higher risk of under-employment than men, but men – and in particular those with less than tertiary education – have seen significant increases in the probability of being under-employed. Whereas trends in under-employment among women have varied across countries, men have experienced increases in almost all of the countries examined.

2.3.1. Wages have been stagnating for a large share of the population over the past decade

Both trade openness and technological progress have contributed to increase workers’ earnings and living standards, on average. However, for large segments of the labour force, earnings in recent years have been stagnating despite a recovery in employment after the global economic and financial crisis (OECD, 2018[22]). In OECD countries, annual growth in nominal hourly wages dropped from 4.8% on average in the pre-crisis period to 2.1% in recent years. Real wage growth decreased by 1 percentage point over the same period. Salary dynamics in low-pay jobs have been a key driver of the overall decline in wage growth. In particular, there has been a significant worsening in average earnings of part-time jobs relative to those of full-time jobs, which is associated with the rise of involuntary part-time employment in a number of countries, discussed below.

2.3.2. Jobs have become less stable

Another key dimension of job quality is labour market security, which is closely linked to job stability. Recent OECD work shows that over the past two decades job stability has decreased on average, although with considerable differences across countries – as discussed in Chapter 3 and in Falco, Green and MacDonald (forthcoming[59]). The evidence in this respect is nuanced, but clear. Average job tenure, a direct indicator of job stability, measuring the amount of time spent in one’s current job, has increased on average. This is, however, the result of an ageing population, as a larger share of older workers in the workforce is mechanically associated with higher average tenure-levels. Once ageing is accounted for, job stability has decreased in the majority of OECD countries. The trend is particularly evident among less educated workers and it is not exclusively concentrated among youth. Prime-age and older workers with lower levels of education have also experienced increased instability in their jobs. Chapter 3 discusses this trend in detail and investigates whether decreased job stability can be ascribed to increased risks for workers or better opportunities for mobility and career progression.

2.3.3. The impacts of globalisation on job quality are mixed

Turning to the link between trade and job quality, a number of competing factors are at play. On the one hand, trade openness may be conducive to higher earnings. Indeed, there is evidence that export-driven industries tend to pay higher wages.21 On the other hand, with regard to job security, greater openness to trade and integration in GVCs may lower it by increasing the risk of job displacement due to offshoring or outsourcing (Acemoglu and Autor, 2010[60]; OECD, 2017[10]). For example, when Chinese factories began undercutting production in the United States, workers in the affected industries faced a higher risk of job loss (Autor, Dorn and Hanson, 2013[30]), resulting in higher job insecurity and, therefore, lower overall job quality.

The megatrends can also impact job quality by directly influencing working conditions and the quality of the working environment. With regard to trade, the main risk is that firms use GVCs to jettison workers in countries with high labour standards and move production to areas where labour standards are lower. For example, if welders in Germany see their jobs go to emerging economies with lower health and safety standards, global job quality may fall. Such concerns find some support in the literature, but the existing evidence is still too limited to draw firm conclusions.22 On the other hand, globalisation and international trade can help “export” better working conditions, especially as multinational companies face increasing consumer pressure and closer international scrutiny of their work environment (OECD, 2008[61]). If the latter effect could be strengthened, international trade could effectively widen global access to good jobs. The OECD Guidelines for Multinational Enterprises and the OECD Due Diligence Guidance for Responsible Business Conduct are prime examples of instruments aimed at improving labour standards through global supply chains (OECD, 2011[62]; 2018[63]).

2.3.4. Technological progress has historically helped improve working conditions

Technological progress has considerable potential to improve working conditions. Across a number of industries, tasks have been automated that formerly required hard physical labour, were often performed in strenuous or even dangerous conditions, and could increase stress and alienation.

One clear illustration of how onerous tasks may disappear is the transformation of agriculture. Between 1991 and 2017, the share of global employment in this sector fell from 43.3% to 26.5% (ILO, 2018[64]), thanks to massive diffusion of productivity-enhancing technologies, ranging from tractors and combine harvesters to more recent innovations such as robo-pickers of fruits and vegetables (ILO, 2018[65]). Many agricultural jobs were of very low quality, involving physically onerous and repetitive tasks, sometimes combined with abusive working conditions (ILO, 2015[66]) and little access to social protection, training opportunities and collective representation. Similarly, technology currently helps workers to perform some of the most dangerous and hazardous tasks in the manufacturing and construction sectors. This welcome development directly contributes to improving working conditions and safety at work.

2.3.5. But greater use of technology can also have a negative impact on job quality in certain occupations

In some cases, however, technology in the workplace can reduce job quality. For instance, greater use of computers and digital technologies to standardise and monitor tasks may limit workers’ autonomy and independence, two key markers of high-quality employment (Weil, 2014[29]; OECD, 2014[58]), but the literature is not unanimous in finding evidence of these negative aspects. Menon et al. (2018[67]), for example, find evidence of a positive effect of computer use on autonomy in Europe. Some authors have been discussing new forms of “digital Taylorism” in which employees enjoy very limited control over their work – for a survey, see Gallie (2013[68]). Many cases of such developments exist throughout the economy, for example in industries like retail and logistics. Employees who work in the warehouses of large logistics companies can be micro-managed (e.g. receiving instructions via headphones) and their productivity can be closely monitored, raising pressure and stress. Looking ahead, some firms are assessing the possibility of introducing wearable devices that would allow close monitoring of workers’ movements on the company floor. Such tight control standards have generated much controversy as they can directly harm job quality.23

Perverse effects of technology on autonomy and discretion are not limited to low-skilled workers. Recent studies find that interconnected devices afford professionals greater control over the pace and organisation of their work, but also create the expectation of constant availability by colleagues and clients, reducing discretion (Mazmanian, Orlikowski and Yates, 2013[69]). Some countries have reacted against these changes. France, for instance, recently passed a law that requires companies with more than 50 employees to grant workers the “right to disconnect” (by not expecting them to respond to emails) outside of regular working hours (de Guigné, 2016[70]).24 At the same time, some executives have embraced new technologies for enabling “work-life integration” and improved flexibility (Lebowitz, 2018[71]). However, the literature is not unanimous in highlighting the negative aspects of technology on working conditions.

2.3.6. Platform work: greater flexibility vs digital Taylorism

The rise of platform work has thrown a spotlight on the impact of technological progress on job quality. Platform work encompasses a broad range of activities, which have in common the use of online platforms to connect the demand and supply of particular services.25 The services provided by digital labour platforms can be broadly distinguished as services performed digitally (i.e. micro tasks, clerical and data entry, etc.) or services performed on-location (i.e. transport, delivery, housekeeping, etc.), as outlined in a recent report by the Joint Research Centre of the European Commission (Biagi et al., 2018[72]). In some cases, the function of the platform goes beyond its mediating role and includes providing workers with an online environment and with the necessary tools to conduct their work.

One of the positive aspects of platform work is the increased efficiency of the matching process, which may help to alleviate problems such as frictional unemployment and skills mismatches. In many OECD countries, unemployment coexists with firms recurrently complaining about not being able to find workers to fill vacancies. Platforms can help employers find workers for tasks that their existing employees cannot perform (Manyika et al., 2015[73]). Another positive aspect of platform work, often cited by workers, is greater flexibility. In the EU Collaborative Economy and Employment (COLLEEM) survey, flexibility was the most cited motivation for engaging in platform work (Biagi et al., 2018[72]). In countries with a high incidence of informal employment, platform work can represent a route to formalisation (Box 2.4).

However, work through platforms can sometimes impose severe limitations to workers’ autonomy, which may have negative impacts on their job quality and well-being. While platform workers are often classified as self-employed and can in principle choose their own hours, demand may de facto be highly concentrated in certain parts of the day. Many workers cannot set their own pay rate, which is imposed by the platform, and face restrictions over other aspects of their work organisation, including the use of uniforms and stringent instructions regarding the way the work is carried out. Finally, platform work allows for close monitoring and levels of micro-management that would be difficult to attain in the absence of the new technologies (but which are by no means exclusive to platform work, as exemplified by the case of retail and logistics discussed above). For instance, employers can use monitoring software by companies like Crossover, which takes periodic photos through the user’s webcam to verify freelancer productivity (Solon, 2017[74]). And workers who do not perform well can be automatically excluded (see also Chapter 4).

While some of these factors may generate greater efficiency and productivity, benefiting consumers (chiefly through lower prices, as well as higher service quality and availability), the result is that some if not much platform work may in fact be far from flexible and may not provide workers with the autonomy and discretion they might wish for.

The potential downsides of certain types of platform work are not limited to the risk of job strain and poor working conditions but also include the risk of low (and uncertain) earnings. Some platforms, for instance, operate globally across very different labour markets. This might induce a race to the bottom in workers’ pay.26 Moreover, since platform workers are frequently classified as self-employed, they also face challenges with regard to the adequacy of social protection, collective representation, and employment protection. These problems are not unique to platform work and they may apply to a different degree to many non-standard workers (i.e. those with contracts that fall outside the “standard” of full-time permanent employment). As such, they will be discussed below. While the jury is still out on the potential advantages and drawbacks of platform work, it is important to underscore that the risks for job quality are not inevitable and can be overcome through careful policy action.

2.3.7. Work through platforms is still a limited phenomenon

How big is employment in the platform economy? Existing evidence in this respect is still scant and imprecise, largely because standard labour force surveys do not capture the phenomenon effectively. The available data, however, indicates that this segment of the labour market is still very small.

A recent survey of 14 European countries indicates that less than 2% of the entire labour force, on average, mentions platform work as their primary activity (Biagi et al., 2018[72]). Furthermore, this is likely to be an overestimate due to the features of the survey design, which is based on an online tool that tends to over-represent the most technologically savvy part of the population. Most of the other existing studies covering a range of countries have typically produced estimates that vary between 0.5% and 3% of the labour force – see OECD (2018[6]) for a survey of the literature). The most recent evidence from the United States, for example, indicates that platform workers accounted for 1% of total employment in May 2017 (BLS, 2018[75]).

copy the linklink copied!

Figure 2.8. The rapid growth of online work has recently slowed down

Index time series (May 2016 = 100; monthly average) of new vacancies posted on the five largest English-language online labour platforms

Source: http://ilabour.oii.ox.ac.uk/online-labour-index/. For further details, see Kässi, O. and V. Lehdonvirta (2016[81]), “Online Labour Index: Measuring the Online Gig Economy for Policy and Research”, Munich Personal RePEc Archive.

 StatLink https://doi.org/10.1787/888933966084

Also, while the platform economy may have grown fast, there are signs that its growth may already have started to slow down. Data produced by the Oxford Internet Institute provides an indication of the growing importance of online work (one type of platform work which is carried out entirely online). This Online Labour Index (OLI) is based on real-time data from five of the world’s largest English-language online labour platforms (Kässi and Lehdonvirta, 2016[81]). Despite its limitations and its focus on one particular kind of platform work, the indicator provides an indication of recent trends. Between May 2016 and May 2017 platform work grew by over a third. Since then, however, there was strong volatility and a flattening of the long-term trend (Figure 2.8).

Most vacancies are posted from OECD countries, particularly in the United States, but the majority of workers are based in non-OECD countries, with India being a particularly important player (OECD, 2018[82]). This global dimension of platform work and the risk of a race to the bottom in terms of labour standards for certain segments of this market indicate that coordinated action among countries is required.

2.3.8. More generally, non-standard work constitutes an important policy concern

The recent interest in the (still small) platform economy risks detracting from a more general and relevant issue: the significant (and in some countries growing) incidence of non-standard work more generally, and its potentially negative implications for job quality. Non-standard forms of employment encompass all forms of work that deviate from the “standard” of full-time, open-ended contracts with a single employer (see Box 2.3 for a detailed explanation). They include, therefore, workers with temporary jobs, part-time contracts, and those who are self-employed. Non-standard jobs are not necessarily of lower quality than standard jobs. The work of a high-skilled professional, for instance, may be non-standard since it falls into the category of self-employment, but might be characterised by high and stable earnings, as well as by good working conditions. Across countries, however, an association exists between many forms of non-standard work and poorer job quality, in the form of lower wages, less employment protection, reduced (or no) access to employer and social benefits, greater exposure to occupational safety and health risks, lower investments in lifelong learning, and low bargaining power of workers – e.g. OECD (2014[58]). For this reason, monitoring trends in non-standard work becomes crucial to assess developments in job quality. In the majority of OECD countries, non-standard work encompasses a significant share of the labour force (over a third), but recent trends have not been uniform.

2.3.9. Temporary employment has risen in one half of OECD countries, with a very marked upward trend in some of them

In around half of OECD countries, there has been a long-term upward trend in temporary employment. The growth of fixed-term employment has been particularly marked in countries like France, Italy, Luxembourg, the Netherlands, Poland, Portugal, the Slovak Republic, and Spain prior to the crisis (Figure 2.9). In the countries where the share of fixed-term contracts has fallen, the reduction has typically been small (with the exception of Greece, Japan, and Turkey). The share of contracts of very short duration (zero to three months) in fixed-term employment, a category that often concerns policy makers, shows a somewhat heterogeneous trend. In just over half of OECD countries, this share has increased. Yet, with the exception of the Baltic countries and Belgium, in the countries where it decreased, that trend was essentially due to the expansion of fixed-term contracts of longer duration.27 Finally, employment through temporary work agency (TWA) has grown in most OECD countries.28 Since the expansion of fixed-term employment has occurred in several countries prior to the 2000s, it is important to remark that it may only be partly attributable to the megatrends analysed in this report, and may in fact be the result of policy choices that facilitated the diffusion of temporary contracts.

copy the linklink copied!

Figure 2.9. Temporary employment has risen in one half of OECD countries

Fixed-term employment as a share of dependent employment, all ages

Note: Data are for 1987 instead of 1986 for the Netherlands and Spain; 2001 instead of 2000 for Australia, Poland and the United States.

Source: OECD Employment Database, www.oecd.org/employment/database.

 StatLink https://doi.org/10.1787/888933966103

2.3.10. Part-time has grown and it is increasingly involuntary

Part-time employment has risen in most OECD countries over the past few decades, with a few notable exceptions including Iceland, Poland and Sweden (Figure 2.10). This is often viewed positively, especially since the rise in part-time employment has been associated with more women entering the labour market, and it has allowed individuals to find a better work-life balance. For some workers, however, part-time employment is involuntary and reflects the difficulty to find full-time jobs. Chapter 3 offers a discussion of this phenomenon within a broader analysis of under-employment.

The share of involuntary part-time in total part-time dependent employment has risen in two thirds of OECD countries for which data are available, although there have been declines in countries like Belgium, Poland and in Germany (since 2010). While in some countries this increase in involuntary part-time will have been partly crisis-related (e.g. Portugal, Spain, Italy and Greece), in most countries one can observe a longer-term trend increase.29

copy the linklink copied!

Figure 2.10. Part-time employment has generally increased

Part-time employment as a share of dependent employment, all ages

Note: Part-time employment is defined using a common cut-off of 30 hours per week usually worked in the main job. Data are for 1988 (instead of 1986) for Turkey; 1985 (instead of 1986) for Sweden, Spain and the Netherlands; and 2001 (instead of 2000) for Australia.

Source: OECD Employment Database, www.oecd.org/employment/database.

 StatLink https://doi.org/10.1787/888933966122

2.3.11. Short part-time and on-call labour have risen in many countries

In around one-half of OECD countries with available data, there has also been a rise in “short part-time” (i.e. individuals working 20 hours per week or less) (Figure 2.11).30 The share of short part-time is particularly high in the Netherlands (21% of dependent employment), Denmark (15%), Switzerland (13%) and Australia (13%). In some countries, there has been a fall in the share of short part-time, including Australia, Latvia, Poland, the United Kingdom, and the United States. When interpreting these trends, one should bear in mind that in some countries the rise of short part-time may be an enabling factor for some workers seeking greater flexibility (e.g. to cope with family responsibilities, combine work and study, etc.). The available data do not allow a clear distinction between these different interpretations.

This rise might be partly driven by increases in very atypical contracts (on-call and zero-hour work), but the evidence in this respect is mixed.31 Many countries have special forms of atypical part-time contracts which either involve very short part-time hours or no established minimum hours at all – such as “on-call” work and “zero-hour” contracts (Messenger and Wallot, 2015[83]) – and several of these have experienced rapid growth in recent years. In Australia, one in four workers is a casual worker, and over half of casual employees report having no guaranteed hours (Campbell, 2018[84]). In Italy, there were around 295 000 workers employed by means of an “on call” contract in 2016 (INPS, 2017[85]).32 In the Netherlands, according to a study commissioned by the ILO, on-call work is the fastest-growing type of flexible work arrangement. In 2016, there were 551 000 on-call workers in the Netherlands, making up about 8% of the workforce (Burri, Heeger-Hertter and Rossetti, 2018[86]).33 In the United Kingdom, nearly 3% of people in employment (about 900 000 people) said that they were on a zero-hour contract at the end of 2016. 34 This figure represents a 29% increase over that of 2014 (ONS, 2017[87]; Adams and Prassl, 2018[88]).35 In the Republic of Ireland, a 2015 study roughly approximates the employed population reporting variable hours at 5.3% – acknowledging that this population may include permanent and temporary workers whose hours vary (O’Sullivan et al., 2016[89]).36

copy the linklink copied!

Figure 2.11. Short part-time is on the rise in many countries

Short part-time employment as a share of dependent employment, all ages

Note: Data are for 1987 instead of 1986 for Spain, Sweden and the Netherlands; 1989 instead of 1986 for Norway; and 2001 instead of 2000 for Australia. Short part-time is defined as usually working 1-19 hours per week.

Source: OECD Employment Database, www.oecd.org/employment/database.

 StatLink https://doi.org/10.1787/888933966141

2.3.12. Self-employment is on a long-term downward trend, with some notable exceptions

There has been a long-term decline in self-employment as a share of total employment across the OECD over the past four decades, which can be observed in the majority of countries (Figure 2.12). This may be surprising and in contrast with the perception that new technologies and work models ought to facilitate the rise of independent work. Much of this trend, however, is related to the long-term decline in the agricultural sector, which predominantly occurred during the earlier part of the period. Since 2000, the incidence of self-employment has remained stable in the majority of countries.

Yet in some countries, there have been increases in self-employment, particularly in recent years. These countries include the Netherlands, the Slovak Republic and the United Kingdom.37 On the one hand, growing self-employment could be viewed as a sign of booming entrepreneurship. On the other hand, it can be linked to more precarious working conditions that may reduce job quality. This risk is particularly high for the self-employed without employees (also known as own-account workers or solo self-employed). There is no clear trend across OECD countries in the share of own-account workers in total employment in recent decades, but there have been substantial increases in countries like the Netherlands, the Czech Republic, the Slovak Republic and the United Kingdom (OECD, 2018[6]).

copy the linklink copied!

Figure 2.12. Self-employment is on a long-term downward trajectory

Self-employment as a share of total employment, all ages

Note: Data are for 1971 instead of 1970 for New Zealand and Greece; 2003 instead of 2000 for Luxembourg; and 2015 instead of 2017 for Latvia.

Source: OECD Employment Database, www.oecd.org/employment/database.

 StatLink https://doi.org/10.1787/888933966160

2.3.13. Dependent self-employment and false self-employment are becoming more common

Own-account work is a particular challenge in cases where individuals are financially dependent on a single employer. These are the so-called economically “dependent” self-employed who earn most of their income from just one client. The reason these are a group of interest to policy makers is that they tend to be in a vulnerable position vis-à-vis their client and may need special protections put in place (especially since they will not have access to the usual benefits and protections that employees do).38 Available data from the European Working Conditions Survey show that in around two thirds of countries, dependent self-employment rose between 2010 and 2015 (OECD, 2018[6]).

A closely related concept to dependent self-employment is that of false self-employment. This is a situation where a worker is not only financially dependent on one or several clients, but he/she is also in a situation of subordination with no or limited control over the work (e.g. in the form of mandated working hours, restrictions to the way work is carried out, the place of work, etc.). In other words, despite the worker is classified as self-employed by the parties (e.g. in a written contract between employers and workers), the characteristics of their relationship closely mirror those of an employment relationship (see Chapter 4). Obtaining an internationally comparable measure of false self-employment presents clear challenges due to differing statistical proxies and data availability. Despite the potential limitations, however, the evidence from the European Working Conditions Survey suggests that false self-employment has increased in the majority of EU countries between 2010 and 2015 (OECD, 2018[6]).39

2.3.14. While some forms of employment may be new, the key policy challenges are as old as non-standard work itself

In light of the evidence presented in this section, non-standard forms of work (especially very atypical ones such as on-call work), and low-quality work more generally, require further policy action. The objective, however, should not be to regulate them out of existence or to impose overly stringent regulation, since a diversity in employment contracts remains an important tool to allow firms to adapt to changing market conditions and to give workers greater flexibility in managing their work-life balance. The goal of policy action should be primarily to avoid abuse and to increase the quality of non-standard jobs.

2.3.15. The line between salaried work and self-employment is increasingly blurred, posing a key challenge for regulators

One issue which has received considerable public, policy and legal attention in recent years is the correct classification of workers who appear to fall somewhere in the grey zone between dependent and self-employment. Workers in the platform economy are a classic example of the potential ambiguity that may give rise to controversy. Platform workers are typically classified as own-account workers. However, like employees, they often have limited control over their work (for instance, in some cases they cannot fix prices, they are required to wear uniforms, they cannot choose the order of their tasks, etc.). The problem, however, is not limited to the platform economy – many hairdressers, plumbers, and gardeners have faced similar challenges in the past. In some cases, the issue may be that these workers are falsely classified as self-employed in order to avoid regulation, or to access preferential tax treatment. But this is not always the case. In many instances, employer-worker relationships are genuinely difficult to classify and may require a revision of the legislation and, in particular, of what it means to be “an employee”, “self-employed” and/or “an employer”. Even where individuals are correctly classified and genuinely self-employed, there may be a case for government intervention to improve their labour market outcomes, for example because these workers find themselves in a situation of monopsony (and are price takers) or are in a situation of economic dependency. Some countries have address this challenge by giving subgroups of these individuals access to some, though usually not all, of the rights and protections granted to employees (advantages and disadvantages of different approaches are discussed in Chapter 4). Aside from the need to resolve potential ambiguities in classification, governments should consider policy avenues to give non-standard workers greater access to collective representation, better training opportunities, and stronger social security, as well as adequate employment protection (as discussed in Chapters 4 to 7).

2.4.1. While overall employment has been growing, some industries have dramatically declined

While overall employment has continued to grow in the face of major structural transformations, as documented above, entire sectors of the economy have declined as a result of the megatrends, leading to considerable job losses and to major disruption in the lives of many workers. Recent work by the OECD documents trends in employment across broad economic industries (OECD, 2017[10]). It shows a clear imbalance: over the past two decades, with new employment predominantly created in service industries, while manufacturing has typically shrunk (Figure 2.13). This trend has contributed to increase disparities between different groups of workers, as it has been partly responsible for the polarisation of the labour market.40

The transition towards a low-carbon economy is exacerbating some of these trends. Primary industries, such as mining, and high-emission industries in manufacturing are the ones that will be most negatively affected by the transition. The estimated impacts of a relatively moderate carbon tax policy is the elimination of 8% of employment related to fossil fuels mining and electricity generation by 2035 (Château, J., Bibas and Lanzi, 2018[41]). While new jobs will also be created as a result of the greening of the economy, they will typically be in different industries, often in different regions, and they may frequently require different skill sets compared to the jobs that were lost. In the absence of adequate re-training programmes, the new jobs may be simply out of reach for displaced workers.

copy the linklink copied!

Figure 2.13. The decline of the manufacturing sector

Percentage change in total employment within industry for selected OECD countries, 1995 to 2015

Note: The figure depicts the percentage changes in total employment by industry (by two-digit ISIC Rev.3 classification). The results are obtained by pooling together employment in each industry across all the countries analysed. The average industry growth (red bar) is a simple unweighted average of changes in total employment across industries.

Source: OECD (2017[90]), “How technology and globalisation are transforming the labour market”, in OECD Employment Outlook 2017, https://doi.org/10.1787/empl_outlook-2017-7-en.

 StatLink https://doi.org/10.1787/888933966179

An ageing population has made the prospect of displaced workers an even more significant concern. In the short-term, this may lead to an increased risk of long-term unemployment among specialised older workers, who have a more difficult time retraining and finding work with comparable wages (OECD, 2005[91]; OECD, 2018[22]). As life expectancy increases and the retirement age has been rising in many OECD countries, workers who are dismissed from manufacturing jobs in their 40s or 50s may still have two or more decades of active working life ahead of them. One potential reason for optimism, however, is that educational attainment in the workforce has been growing over time and older workers are increasingly well-educated, which should make them better equipped for career changes.

The need to help displaced workers through difficult transitions is a pressing policy issue. Effective safety nets in the face of job displacement are a key piece of the policy response. Activation measures that intervene early and possibly prior to dismissal are equally important (OECD, 2018[22]). Timely and effective action requires identifying the workers who are most in need of support and design tailored assistance programmes (see Chapter 7). Another crucial element is how to help workers to update their skills and acquire new competences (see Chapter 6).

2.4.2. The labour market has become more polarised

Another transformation that is upending the labour market of advanced economies is job polarisation. Over the past decades, the share of middle-skilled jobs has decreased relative to the share of workers in high- and low-skilled occupations (Autor, Katz and Kearney, 2006[92]; Goos and Manning, 2007[93]; Goos, Manning and Salomons, 2009[94]; OECD, 2017[10]).41 In almost all countries for which data are available, this process has resulted in an overall shift of employment towards high-skilled occupations (Figure 2.14).42

copy the linklink copied!

Figure 2.14. The labour market is polarising

Percentage point change in share of total employment, 1995 to 2015

Note: High-skilled occupations include jobs classified under the ISCO-88 major groups 1, 2, and 3., that is, legislators, senior officials, and managers (group 1), professionals (group 2), and technicians and associate professionals (group 3). Middle-skilled occupations include jobs classified under the ISCO-88 major groups 4, 7, and 8, that is, clerks (group 4), craft and related trades workers (group 7), and plant and machine operators and assemblers (group 8). Low-skilled occupations include jobs classified under the ISCO-88 major groups 5 and 9, that is, service workers and shop and market sales workers (group 5), and elementary occupations (group 9).

Source: OECD (2017[90]), “How technology and globalisation are transforming the labour market”, in OECD Employment Outlook 2017, https://doi.org/10.1787/empl_outlook-2017-7-en.

 StatLink https://doi.org/10.1787/888933966198

What are the drivers of job polarisation? The decline of the manufacturing sector has been partly responsible, since many manufacturing jobs are also in the middle of the pay distribution, but it does not account for the entire change. In fact, the majority of polarisation is due to the loss of middle-skilled jobs within industries (OECD, 2017[10]). The forces of technological change and globalisation have both played a major role in fostering polarisation. Middle-skilled jobs have been the most prone to automation and offshoring, due to their highly routine nature, which makes them relatively easy to codify into a set of instructions that could either be carried out by a machine or by a worker abroad.

The relative importance of technological progress and globalisation in driving job polarisation is the subject of a lively debate. It is a difficult question to answer, since the two megatrends complement and reinforce each other. Recent OECD work attempts to address the issue. It shows that the penetration of ICT in a given industry bears the strongest correlation with its polarisation, while the role of globalisation is less clear-cut (OECD, 2017[10]; Breemersch, Damijan and Konings, 2017[95]). This implies that raising barriers to trade may have limited effects in reducing job losses in declining industries when the effects of automation are also at play. It is important to remark, however, that the evidence varies across countries, and a growing literature documents the adverse impacts of import competition from countries like China on local labour markets, which should not be downplayed (Autor, Dorn and Hanson, 2013[30]). It is also important to note that technology-induced job-polarisation does not necessarily imply that workers will be displaced. Rather, it may occur as the result of a lower share of young labour market entrants taking middle-skilled jobs and older workers in these jobs retiring (Dauth et al., 2017[96]; Green, forthcoming[97]).

The impacts that globalisation had on the labour market of OECD countries over the past decades were closely linked to the rapid integration of major global players, most notably China, in GVCs. This led to what some commentators have called the “great doubling” of the world labour supply (Freeman, 2007[98]). As the success of China was largely driven by low-cost, labour-intensive manufacturing, the rise of China’s exports put middle and low-skilled labour in many OECD countries under pressure in terms of their jobs and wages – see e.g. Autor, Dorn and Hanson (2013[30]) – and fuelled their perception of being left behind by globalisation. Going forward, the integration of more countries with expanding populations in GVCs will continue to have important consequences for the labour market, but the effects may not be as dramatic and will likely be different compared to the past. Emerging economies nowadays are indeed also producing a growing pool of high-skilled workers who are competing in the global labour market.

In the policy debate it is often assumed that the decline in the share of middle-skilled occupations has led to a decline in the share of middle-pay jobs. However, this is not the case for two main reasons (Chapter 3). First, many of the high-skilled jobs (whose share has increased in all countries) also pay mid-level wages. Second, there have been changes in the propensity of all occupations (including both high- and low-skilled ones) to pay middle-level wages, which, overall, have tended to increase the share of middle-pay jobs.

The complex interaction of these transformations has affected the fortunes of different workers differently (see Chapter 3). In particular, on average across the OECD, young people without tertiary education have seen increases in the probability of being neither in education nor employment as well as increases in the probability of being in low-pay jobs for those who do find employment. In addition, while women remain at much higher risk of non-employment, men have seen increases in non-employment in most countries. Women also remain more likely to be in low-pay jobs and less likely to be in high-pay ones, despite an improvement in the probability of being in middle-pay jobs.

2.4.3. Labour market changes may contribute to the growing sense of frustration and discontent among the middle class

Job polarisation is frequently associated with the perception that the middle class in advanced economies is being squeezed (OECD, 2016[99]; Manfredi and Salvatori, forthcoming[100]; OECD, 2019[101]). That is because middle-skilled jobs have been traditionally associated with middle-class households and the relative decline of those jobs has sparked concerns that an important source of income for the middle-class may be drying up.43 A squeezed middle class is a key policy concern because it directly implies that economic opportunities are less equally shared and that opportunities for social mobility may have decreased.

Recent OECD work shows that job polarisation per se has not resulted in a decline in the share of workers who are in middle-income households (Manfredi and Salvatori, forthcoming[100]). In fact, the share of workers in middle-income households has not changed significantly over the past 20 years, although there are differences across countries.44 This is because the decline of middle-skilled jobs (plant and machine operators, assemblers, clerical, and craft occupations) has been mostly compensated by an increase in high-skilled jobs (technicians and associate professionals, managerial and professional occupations) which are also commonly found in middle-income households.45 In addition, the fraction of high-skilled workers who are in middle-income household has also increased. The combination of these trends implies that the skill composition of the middle class has changed considerably, as the share of high-skilled workers has increased more in middle-income households than in the economy as a whole (OECD, 2019[101])

2.4.4. A tale of broken promises?

The overall conclusion is that some jobs are increasingly failing to deliver the same income status and the same labour market security as in the past. Middle-skilled occupations no longer guarantee middle-class status, and high-skilled jobs no longer give workers automatic access to the higher echelons of the income distribution (OECD, 2019[101]). This may be a cause of significant frustration especially for workers who made their occupational choices at a time when these trends were not yet clear, and found their labour market outcomes falling short of their expectations.

This type of phenomenon may help to explain the growing sense of preoccupation and discontent registered in many OECD countries, which spans well beyond people in the lowest tiers of the income distribution and increasingly encompasses middle-class households. A recent report by the UK Resolution Foundation, for instance, paints a detailed picture across occupations and shows that some public-sector jobs such as teachers, lecturers, the police, and armed forces, which used to be typical middle-class occupations, were among those recording the largest decline in relative position in the income distribution (Corlett, 2016[102]).

2.4.5. A shrinking share of national income is going to workers

A trend that is related to increasing economic inequality and to growing discontent in many OECD countries is the falling share of national income that goes to workers in the form of labour earnings, while the share that goes to capital owners has been increasing. Over the past two decades, the aggregate labour share fell by 3.5 percentage points (from around 71.5% to 68%) in the 24 countries covered by a recent OECD study (OECD, 2018[22]). Within the same period, the economy witnessed a decoupling between real median wages and productivity, with the latter growing much faster than the former.46 If real median earnings had perfectly tracked productivity over the period, they would have been 13% higher at the end of it (Figure 2.15). In other words, contrary to previous decades, the productivity gains generated by the economy have not resulted in broadly shared wage gains for all workers (Schwellnus, Kappeler and Pionnier, 2017[103]).

The labour share, however, has not been falling uniformly across countries. While it fell by about 8 percentage points in the United States and by nearly 6 percentage points in Japan, it remained broadly constant or increased in about half of the covered OECD countries, including France, Italy and the United Kingdom.47 These differences partly reflect cross-country differences in business cycle developments. Schwellnus et al. (2018[104]) show that an increase in the output gap of 1% reduces the labour share by 0.5 percentage points. However, structural reforms in a number of areas, including product and labour market institutions, as well as collective bargaining, also emerge as significant determinants of labour share developments and may partly explain cross-country differences (Schwellnus et al., 2018[104]).

Technological progress and (to a lesser extent) globalisation can explain most of the contraction in the labour share (OECD, 2018[22]). Capital-augmenting technological progress or technology-driven declines in relative investment prices reduce the labour share by fostering labour-capital substitution and increasing overall capital intensity. Globalisation can have similar effects. Offshoring and import competition typically lead to job displacement in relatively labour-intensive tasks and hence increase the capital intensity of the production process, but such impacts have been less marked. Furthermore, these dynamics do not impact all industries equally and low-skilled workers are more likely to be negatively affected. In industries with a predominance of routine tasks, the substitution of capital for labour in response to declines in relative investment prices is particularly pronounced. A higher share of high-skilled workers, on the other hand, reduces the substitution of capital for labour even in industries with a high incidence of routine tasks.48

copy the linklink copied!

Figure 2.15. Real median wages have decoupled from labour productivity

Indices, 1995 = 100

Note: Gross domestic product (GDP) weighted average of 24 countries (two year moving averages ending in the indicated years). 1995-2013 for Finland, Germany, Japan, Korea and the United States; 1995-2012 for France, Italy and Sweden; 1996-2013 for Austria, Belgium and the United Kingdom; 1996-2012 for Australia and Spain; 1997-2013 for the Czech Republic, Denmark and Hungary; 1997-2012 for Poland; 1996-2010 for the Netherlands; 1998-2013 for Norway; 1998-2012 for Canada and New Zealand; 1999-2013 for Ireland; 2002-11 for Israel; 2003-13 for the Slovak Republic. In Panel A, all series are deflated by the total economy value added price index. In Panel B, all series are deflated by the value added price index excluding the primary, housing and non-market industries. The industries excluded in Panel B are the following (International Standard Industry Classification – ISIC – rev. 4 classification): (1) Agriculture, Forestry and Fishing (A), (2) Mining and quarrying (B), (3) Real estate activities (L), (4) Public administration and defence, compulsory social security (O), (5) Education (P), (6) Human health and social work activities (Q), (7) Activities of households as employers (T), and (8) Activities of extraterritorial organisations and bodies (U).

Source: OECD (2018[22]), OECD Employment Outlook 2018, Fig. 2.1, https://doi.org/10.1787/empl_outlook-2018-en. Based on OECD National Accounts Database, https://doi.org/10.1787/data-00727-en; OECD Earnings Distribution Database, https://doi.org/10.1787/data-00302-en.

 StatLink https://doi.org/10.1787/888933966217

2.4.6. Winner-takes-most dynamics, superstar firms, and the falling labour share

The fall in the labour share is also connected to the phenomenon of “winner-takes-most” – the process through which the most productive firms in the economy capture an overwhelming share of the market (Rosen, 1981[105]; Frank and Cook, 1995[106]; Autor et al., 2017[28]). This is itself a result of technological progress and globalisation. Thanks to a fall in ICT costs and transport costs, increased access to consumer data, as well as a reduction in tariffs, firms have been increasingly able to access global markets. This has increased enormously the potential for economies of scale. The result is that the most productive firms in the economy (the so-called “superstar” firms) are now significantly larger than the most productive firms were decades ago, which implies that their labour share declines (as the value added share of fixed labour overhead costs declines and/or their mark-up increases). As a consequence, a significant increase in industry concentration has been recorded in both Europe and North America between 2000 and 2014 (Bajgar et al., forthcoming[107]). Furthermore, as the market share of the best firms increases, production is reallocated towards them (and, therefore, towards a production process with a lower labour share). Recent OECD work supports this view. Winner-takes-most dynamics have contributed to the fall of the labour share both through a decline in labour shares among leading firms and through the allocation of market shares towards those firms (OECD, 2018[22]). The most important implication for policy makers is that they should maintain a strong focus on competition policy to ensure that superstar firms do not engage in anti-competitive behaviours and that the markets they operate in remain contestable. Doing so will not only benefit consumers and workers. It will also be beneficial for small businesses, for whom accessing highly concentrated markets may become prohibitive.

2.4.7. Increasing market concentration in certain industries sparks new worries of growing monopsony power

Rising concentration in the product market, which partly results from winner-takes-most dynamics, is becoming an increasingly important policy concern. De Loecker and Eeckhout (2017[108]) show that company mark-ups are increasing. Calligaris et al. (2018[109]) demonstrate that this phenomenon is particularly evident in the digitally intensive industries, where winner-takes-most dynamics are most prominent. High mark-ups are a sign of greater market power as firms are able to charge higher prices (or to offer lower wages) the higher their share of the total market is.

Increasing market concentration is also bringing back concerns about possible monopsony in the labour market, a situation where a company dominates the market and can keep wages low since it faces little (or no) competition for workers. A classic case of monopsony from the past is the “company town”, such as coal mining communities in rural areas. More recent research and the development of search-and-matching models have shown that different kinds of frictions may give rise to monopsony. For instance, limited information about available jobs, constraints on geographical mobility of workers, and skills mismatches, may be contributing factors. Regulations can also play a perverse role. Non-compete covenants and occupational licenses, as well as health and pension benefits that are linked to specific jobs, for instance, can contribute to lock workers in and lock other workers out of better remunerated jobs (see Chapter 4). In addition, declining trade union membership and weaker collective bargaining institutions can further reduce workers’ bargaining power and increase monopsony power (see Chapter 5).

The policy discussion on monopsony, however, is still relatively limited. This is due partly to difficulties in documenting the phenomenon and partly to the modus operandi of competition policy. Despite the measurement challenges, some recent papers on the United States document growing labour market concentration. Azar, Marinescu and Steinbaum (2017[110]) use data from a large online job board and show that higher concentration is associated with lower posted wages.49 Benmelech, Bergman and Kim (2018[111]) measure employment concentration and its effect on wages using Census data for manufacturing industries over a long time horizon. They show that, there is a negative relation between local-level employer concentration and wages, which is more pronounced at high levels of concentration. They also find that exposure to greater import competition from China is associated with more concentrated labour markets. The open question is to what extent such trends are also visible in other OECD countries (see Chapter 4 for a more extensive discussion of evidence and policy issues).50

2.4.8. The effects of the megatrends are geographically concentrated and contribute to regional disparities

While inequality between countries in per-capita GDP and labour productivity have decreased over the past two decades (especially in Europe), within-country inequality (i.e. inequality between different regions in the same country) remain large and have even grown (OECD, 2018[112]).

Geographical disparities are particularly clear between rural and urban areas. While average business creation rates are 13% (of the total number of existing firms) in predominantly urban regions, it is only 10.9% in predominantly rural ones.51 More importantly, urban and rural areas display very different sectoral composition and characteristics of new firms. Urban areas tend to attract more knowledge-intensive firms, which are likely to have the best future prospects (OECD, 2018[112]).

The megatrends have contributed to growing regional imbalances (OECD, 2018[113]). The adverse effects of import competition and offshoring, as well as the labour-displacing impacts of new technologies, are particular strong in regions with the highest concentration of firms in routine-intensive industries. A classic example is the Midwest and the Great Lakes region of the United States (nowadays often referred to as the “rust belt”), where the decline of previously very prominent manufacturing industries (such as the automotive industry), led to the disruption of the regional economy. Similar trends are ongoing across the OECD, with countries facing a geographically unequal distribution of jobs at risk of automation. A common pattern is that capital-city regions tend to face the lowest risk of automation, while peripheral regions (often characterised by a stronger presence of mature manufacturing industries) have a higher share of automatable jobs (OECD, 2018[113]).The transition towards a greener economy amplifies these trends. Population ageing and outmigration exacerbate the economic challenges faced by such regions and further reduce their productive potential.

A growing literature documents the impact of the megatrends on local labour markets. Autor, Dorn and Hanson (2013[30]), for instance, show in the context of the United States that when a local labour market was more exposed to import competition from China (because it accounted for a larger share of national employment in industries that faced heavy import competition), it experienced a 4.5% fall in manufacturing employment and a decline in the employment rate by 0.8 percentage points, relative to a less exposed local labour market. A number of other studies show similar findings in different countries and reveal that when the industries most affected by import competition are clustered in specific regions, employment losses in those regions can be significant (Dauth, Findeisen and Suedekum, 2014[114]; Balsvik, Jensen and Salvanes, 2015[115]; Donoso, Martín and Minondo, 2015[116]).

A key problem is the speed at which job losses occur. If jobs were lost gradually and the phenomenon was spread over a wide geographical area, workers could more easily find new opportunities and might even benefit from job turnover as new jobs may be created in more productive firms (OECD, 2018[112]). However, a growing literature indicates that job losses due to the megatrends, and particularly to trade, are highly concentrated and take a long time to be offset by local job growth in other firms or industries (OECD, 2017[117]).

The very unequal effects of the megatrends across regions has contributed to a marked “geography of discontent,” with an increasing concentration in specific regions of feelings of dissatisfaction with trade, immigration, and economic inequality (OECD, 2018[112]; OECD, 2017[118]) . The failure of economists and policy makers to acknowledge the pitfalls of globalisation for certain regions and communities has contributed further to international scepticism of international trade, and – more holistically – policy advice from elites (Krugman, 2018[119]).

In light of this evidence, policy makers face a difficult dilemma. Trade is beneficial to the national economy (as discussed above), but can have long-lasting negative effects in some regions (and for specific groups of workers). In order to generate shared prosperity, trade integration needs to be accompanied by timely policy action to help the areas, industries, and workers that risk falling behind. Protectionist policies that aim to restrict trade to protect specific industries or regions, however, risk having detrimental effects for the rest of the economy. Reducing trade would reduce living standards in the long run by limiting productivity gains from specialisation, slowing down innovation, and leading to higher prices for consumers – for an extensive discussion of specific policies to help reducing regional imbalances, see the policy discussion in OECD (2018[112]).

2.4.9. The megatrends may contribute to further inequality in the labour market without opportune policy action

The transformations brought about by technological progress, globalisation, and demographic change have been accompanied by a worrying trend across many OECD countries: a rise in income inequality. Today, across OECD countries, the top 10% of adults by income have incomes that are 9.4 times the amount of the poorest 10% (OECD, 2018[120]). Only one generation ago, the ratio was seven to one (Figure 2.16 shows the divergence of bottom and top incomes over the past three decades). Wealth distribution figures are even starker, with the top 10% holding the same amount of wealth as the bottom 90% combined—and only 3% of wealth held by 40% of the population. Such inequalities in wealth and income translate into other forms of inequality of opportunity, including in the domain of education and health (Andersen, 2015[121]; Chetty et al., 2016[122]). Ultimately, these large inequalities lead to lower mobility for individuals and lower productivity for economies (OECD, 2015[39]; OECD, 2018[120]).

copy the linklink copied!

Figure 2.16. Income inequality is growing rapidly

Real income trends at the bottom, middle and top of the income distribution since the 1980s, OECD-17

Note: Income refers to real household disposable income. OECD-17 refers to the unweighted average of the 17 OECD countries for which data are available: Canada, Denmark, Finland, France, Germany, Greece, Israel, Italy, Japan, Luxembourg, Mexico, the Netherlands, New Zealand, Norway, Sweden, the United Kingdom and the United States. Some data points have been interpolated or use the value from the closest available year.

Source: OECD Income Distribution Database, http://oe.cd/idd.

 StatLink https://doi.org/10.1787/888933966236

Policy makers and experts increasingly agree that inequality constitutes a key policy challenge that can severely hamper the functioning of economies and societies. Income inequality has been at the top of the policy agenda at the OECD with its Inclusive Growth initiative, as well as in other International Organisations.52 A Pew Research Center survey of general publics across the world found that at least half of the respondents in each European country surveyed had serious concerns about income inequality (Pew Research Center, 2013[123]). In addition, a survey of over 10 000 opinion leaders in developing countries found that over 50% ranked income inequality as “a very big problem.” Within the countries surveyed, Mexican and Colombian leaders were the most concerned (Guo, 2017[124]).

These findings mirror the conclusions of a growing literature showing that high inequality can harm productivity and reduce social mobility, hampering growth and fostering discontent. Recent OECD work shows that by preventing large segments of society from investing in human capital, inequality may reduce productivity and growth (OECD, 2015[39]). Furthermore, while one often-cited rationale for tolerating inequality is to motivate workers to reap the rewards of their achievements, the chances of someone doing better than one’s parents are lower in more unequal societies. There is not a single example of a country with high economic inequality and high mobility in the OECD (OECD, 2018[120]). Instead, inequality seems to snowball across generations, as the wealthiest create and sustain both literal and figurative gated communities around high-quality education, health care, and political influence (OECD, 2016[125]; OECD, 2015[39]; Epp and Borghetto, 2018[126]).

The ongoing labour market transformations are linked to the deepening of market income inequalities and, without significant policy changes, the trend is likely to continue. Skill-biased technological progress may continue to increase the earnings of the top earners, who possess the necessary skills and capital, widening the gap with the most disadvantaged. Furthermore, new technologies and access to the global market facilitate the rise of a few superstar firms with increasing market power and growing profits. At the same time, workers’ bargaining power is weakening (see also Chapter 5) and new forms of precarious employment are expanding (also due, in many industries, to an increasingly “fissured production process”, whereby tasks are farmed out to contractors and sub-contractors, as opposed to employees, as discussed by Weil and Goldman (2016[127]) and Weil (2014[29])). As a result, there is a risk that incomes and wealth will become even more concentrated, and social mobility may well fall further (OECD, 2018[120]). Demographic changes can further exacerbate this gap. Without effective policy action, increasing income inequality may cumulate over the course of longer lifespans to create an elderly underclass (OECD, 2017[12]).

However, the cross-country evidence on rising inequality also shows that there is nothing inevitable about its rise. Policies and institutions matter and can play and important role in mitigating the impact of new technologies, globalisation, and population ageing on inequality.

copy the linklink copied!

Box 2.5. Structural transformation and the challenge of growing disparities in emerging economies

Structural transformation has supported economic growth and reduced poverty in emerging economies. The move from low productivity, labour-intensive activities to higher productivity, capital- and skill-intensive ones is at the heart of economic development.

However, structural transformation has been associated with job polarisation, though the process has not been uniform across countries (Figure 2.17). India, the Russian Federation and Brazil are experiencing a dominant shift in employment to more skilled occupations (upskilling), while other countries like China, Mexico, South Africa and Turkey are undergoing a relative growth in low-skilled occupations.53 The data also show that in some countries polarisation has not occurred (e.g. Argentina and Peru have experienced an increase in the share of middle-skilled jobs relative to the share of workers in high- and low-skilled occupations). Moreover, unlike in more advanced economies (OECD, 2017[10]), job polarisation in emerging economies is predominantly the result of shifting employment from less polarised industries (agriculture, but also manufacturing in some countries) to more polarised service industries, with polarisation within industries playing a less important role (Alonso-Soto, forthcoming[56]).

copy the linklink copied!

Figure 2.17. Labour markets are polarising in many emerging economies too

Percentage points change in share of working adults in each skill group, mid-1990s to mid-2010s1

Note: High-skilled occupations include jobs classified under the ISCO-88 major groups 1, 2, and 3, that is, legislators, senior officials, and managers (group 1), professionals (group 2), and technicians and associate professionals (group 3). Middle-skilled occupations include jobs classified under the ISCO-88 major groups 4, 6, 7, and 8, that is, clerks (group 4), skill agricultural and fisheries workers (group 6), craft and related trades workers (group 7), and plant and machine operators and assemblers (group 8). Low-skilled occupations include jobs classified under the ISCO-88 major groups 5 and 9, that is, service workers and shop and market sales workers (group 5), and elementary occupations (group 9).

← 1. 2004-17 for Argentina, 1995-2015 for Brazil, 1996-2009 for Chile, 2000-10 for China, 1997-2010 for Costa Rica, 1994-2012 for India, 2007-15 for Indonesia, 1995-2017 for Mexico, 2002-17 for Peru, 1997-2017 for Russian Federation, 1996-2007 for South Africa and 2001-10 for Turkey.

Source: ILO KILM, www.ilo.org/kilm, except China (Chinese Census, http://www.stats.gov.cn/english/statisticaldata/censusdata/).

 StatLink https://doi.org/10.1787/888933966255

From an historical perspective, structural transformation helped reduce poverty in emerging economies as a large part of employment shifted from low-pay low-productivity agricultural jobs to better-paid jobs in the manufacturing and service sectors (Baymul and Sen, 2017[128]). Yet, automation and job polarisation may now contribute to further increase inequality in emerging economies. First, certain groups of workers are increasingly exposed to the risk of job displacement due to automation. This is likely to be higher for low-skilled workers in manufacturing sectors where jobs with a high routine content are prevalent (Alonso-Soto, forthcoming[56]). Second, the ongoing transformation is linked with two specific drivers of inequality in emerging economies: informal employment and the persistently large geographical differences in economic performances (OECD, 2011[129]). On the one hand, the growing share of service-sector jobs may contribute to increase informality, as it is associated with the rise of non-standard forms of work. On the other hand, new jobs tend to be created in urban areas and in different regions compared to those that are disappearing, fostering the already high regional divides and the urban-rural gaps in emerging economies (OECD, 2018[112]). These changes are occurring against the backdrop of the already high levels of inequality that characterise emerging economies. The lower coverage and generosity of social protection systems, along with a tax system that delivers only modest redistribution makes these challenges even harder to tackle.

Source: Alonso-Soto (forthcoming[56]), Technology and the future of work in emerging economies: What is different?