This chapter discusses past, recent and prospective labour market trends for middle-class workers and their families. The chapter begins by analysing the job occupations of middle-class workers and its evolution in recent decades. It then examines how labour market polarisation and non-standard forms of work have affected the middle-income class in terms of jobs and wages. It also analyses the impact of labour market and demographic changes on the chances of households being middle-income. Finally, it assesses the risks of automation and employment prospects for the current occupations of middle-income workers.
Under Pressure: The Squeezed Middle Class
Chapter 3. Where have the middle class jobs gone?
Abstract
The statistical data for Israel are supplied by and under the responsibility of the relevant Israeli authorities. The use of such data by the OECD is without prejudice to the status of the Golan Heights, East Jerusalem and Israeli settlements in the West Bank under the terms of international law.
Introduction and key findings
Labour markets have been polarising, new technologies have replaced many middle-skilled1 routine jobs with new high- and low-skilled jobs. The fall in the relative number of middle-skilled jobs is often deemed to have contributed to the decline of the middle class in OECD countries. Nevertheless, changes in the distribution of skill groups across income classes may challenge the assumption a certain kind of job yields access to a certain income class. Such thwarted expectation may prompt social tensions.
The make-up and employment profiles of middle-income households have changed. In keeping with the overriding demographic trend, they are increasingly made up of single individuals. Although dual-earner couples remain the backbone of the middle-income class, they increasingly rely on at least one member working in a highly skilled job to maintain their income status.
At the same time as low- and high-skill relative to middle-skill jobs as shares of the labour market have increased, top wages rose faster than those in middle and at the low end of the scale, where the gap between middle and low wages has either widened or stayed unchanged. Unlike jobs, therefore, wages have not polarised (i.e. risen faster at the top and bottom), only become more unequal (i.e. grown faster solely at the top).
Rapid innovation in automation and digitalisation is fuelling fears that technology will generate massive unemployment in the future. Evidence suggests that people are increasingly anxious about the impact of technology on employment. This chapter also assesses the risk of automation and employment prospects of the current occupations of middle-income workers.
From the analysis and evidence in this chapter the following findings emerge:
Fewer working-age households are in the middle-income class than three decades ago. The probability of being middle-income fell both in households headed by workers (“working households”) and those headed by the out-of-work (“non-working households”). Among working households, the squeeze was moderate and led to rises both at lower and upper income classes. The squeeze on non-working middle-income households, however, was tighter forcing many of them into the lower-income class.
There is no archetypal middle-income job. While just a few occupations account for the jobs of most lower- and upper-income households, middle-income class earners work in a wide range of occupations. In the last two decades, the professionals have replaced crafts as the largest middle-income occupational group. Only in a few countries are middle-skill workers still the largest group in the middle-income class.
Job polarisation has had little overall effect on the size of the middle-income class because the decline in the number of middle-skill jobs has generally been offset by the increase in the number of highly skilled jobs.
The relationship between skill levels and income classes has changed. Skill levels are no longer passports to the income class once traditionally associated with them. Middle-skill workers are now more likely to be in the lower-income class and less likely to be middle income. Highly skilled workers are also less likely to make it to the higher-income class, while their low-skilled peers are increasingly concentrated in the lower-income class.
Households increasingly need not only two earners but require one or both of them to be highly skilled if they are to enter the middle-income class. Changes in the probability that a dual-earner household is in in the middle-income class vary depending on their skill levels. Working couples where one or both earners are highly skilled are now more likely to be middle-income and more unlikely to be upper-income. As for dual-earner couples who are not highly skilled, their chances of still being in the middle-income class are the same, while they have grown in the lower- and fallen in the upper-income class.
One-in-six middle-income workers are in occupations that are at high risk of automation. The risk is a little higher in occupations where lower-income workers are currently employed – one‑in‑five – but lower in jobs held by their upper‑income peers – one‑in‑nine.
3.1 The labour status of middle-income households
This section shows income-class profiles and trends over recent decades in households headed by people at work – henceforth referred to as “working households”.
3.1.1 Most working households remain middle income despite lower shares
Most working households are middle income, though to a lesser extent than three decades ago (Figure 3.1, Panel A). OECD-wide, two-thirds of working households were middle income in the mid‑2010s, slightly less than in the 1980s. Over the same period, by contrast, the chances of working households being in the lower and upper income classes increased. The result is a slight income polarisation among working households today.
There are fewer middle-income working households in 16 of the 20 OECD countries for which data are available (Figure 3.1, Panel B). In the 20 years between the mid‑1990s and 2010s the share of working households in the middle-income class grew in Chile, Italy and Mexico, though only in Chile was the rise considerable. In all other countries, it fell noticeably, particularly in Canada, Israel, Poland, the Slovak Republic and the United Kingdom. And in most countries, the growing share of working households in the lower and upper income classes went together with a falling share in the middle-income class. Exceptions were France, Hungary, Switzerland and the United Kingdom – where the rise of the working family came in the lower-income class – and the Czech Republic, Israel, Germany and the United States, where it was confined chiefly to the upper-income end of the scale.
Figure 3.1. Fewer working households are middle-income
Note: Working households are defined as households headed by people in employment. People are defined as employed based on the ILO definition.
Source: OECD calculations based on LIS and EU‑SILC (see Box 3.1 for details).
3.1.2 What are the jobs of middle-income workers?
There is no archetypal middle-income job. While lower- and upper-income heads of household fall into only a few occupational groups, the spread is wider in the middle-income class. Professionals, technicians, clerks, service workers, sales people and craft workers account for at least 10% of middle-income heads of household OECD-wide (Figure 3.2). When it comes to the lower-income class, heads of household work chiefly in services and sales, elementary occupations and crafts, while their upper-income peers are mostly professionals, technicians and managers.
Professionals have replaced craft workers as the largest middle-income occupational group. In the two decades to the mid‑2010s, the share of professionals grew in all income groups, as did that of technicians. Crafts were actually the biggest occupational group in both the lower- and middle-income classes in the 1990s. By the mid‑2010s, however, professionals had taken over that mantle among lower-middle-income workers, and service workers in the lower-income class. Professionals were also the largest occupational group in the upper-income class in the mid‑2010s, where they accounted for more than one-third of workers, compared to a quarter 20 years previously.
Figure 3.2. No job is archetypically middle income, but more and more middle-income workers are professionals
Note: Occupations are based on the ISCO‑88 version of the International Standard Classification of Occupations (ISCO).
Source: OECD calculations based on LIS and EU‑SILC (see Box 3.1).
3.2 Labour market polarisation
This section investigates the impact of job polarisation and changes in the labour status of household members on the middle-income class. Job polarisation (the relative decline in the number of middle-skill in comparison to low- and high-skilled jobs) is often deemed to have contributed to the decline of the middle class in OECD countries. The analysis below examines whether and how that perception is true.2
3.2.1 Polarisation has resulted in a net shift towards high-skill occupations
Labour markets across the OECD polarised in recent decades. While new technologies replaced workers in many middle-skill routine occupations between the 1990s and 2010s, they created new jobs at both the high and low ends of the skills spectrum (OECD, 2017[1]; Autor, 2015[2]). Recent work by the OECD confirms that in most countries the share of jobs in middle-skill occupations has declined relative to high-skill and low-skill occupations since the mid-1990s (OECD, 2017[1]). The OECD also finds that occupational polarisation is closely associated with changes in the distribution of occupations within sectors, although de‑industrialisation (the shift of employment from manufacturing to the services) also plays an important role. Furthermore, polarisation and de‑industrialisation both appear strongly related to technological change. Evidence of an association between polarisation and globalisation3 is weaker, however.
Job polarisation has resulted in a net shift of employment to high skill occupations in most OECD countries. On average across the 21 OECD countries for which data were available (see Box 3.1), middle-skill occupations have lost 8 percentage points in employment shares, while low skill occupations have lost about 2 percentage points and the high skill occupations have gained 10 percentage points (Figure 3.3). Indeed, there was a shift towards highly skilled employment in most countries, with the aggregate share of middle‑skill jobs declining in 19 countries, rising only in Mexico and the Slovak Republic. The increase in high-skill jobs offset the decline – except in Greece, Hungary and the United States. In those countries, the greatest climbs came in low-skill occupations, which nevertheless lost labour market shares in a number of other countries, though only in Belgium did they fare worse than middle-skill occupations. Overall, the most common pattern is one of a decline in middle-skill jobs relative to both high and low skill occupations, with most gains made by high-skill jobs.
Box 3.1. Measuring the impact of job polarisation and household working characteristics on the middle-income class
Middle-income class
A household is in the middle-income class when its disposable income is between 75% and 200% of the median household income in a given year and country (see Chapter 2.1.2).
Occupations
The skill content of a job held is defined in accordance with the third version of the International Standard Classification of Occupations – ISCO‑88. All countries have over time recoded their occupational classification systems, giving them 2‑digit codes that are consistent with ISCO.
Low-skill workers are those with jobs in sales and services and elementary occupations (ISCO 5 and 9).
Middle-skill workers hold jobs as clerks, craft workers, plant and machine operators and assemblers (ISCO 4, 7 and 8).
High-skill workers are those who have jobs in managerial, professional, technical and associated professional occupations (ISCO 1, 2 and 3).
Post‑2010 European employment data were mapped from ISCO‑08 to ISCO‑88 using the following methodology. In order to reconcile the ISCO‑2008 classification with ISCO‑88, information was collected from the two consecutive waves of the European Union Labour Force surveys between which classification changed. Using fuzzy logic to match individuals, this methodology allows to map the new occupation coding to the old one using a many-to-many mapping technique. Employment data for Canada and the United States were transposed from the respective occupational classifications (SOC 2000) into corresponding ISCO‑88 classifications.
Working adults
Working adults are consistently defined as individuals aged 16 to 64 years old who normally work as employees or are self-employed during the income reference period – the year to which the income data refer.
Working households
Working households are those that include at least one working adult and no retired person.
Sources: Estimates are based on data from the Luxembourg Income Study (LIS, www.lisdatacenter.org) for Australia (1995, 2010), Canada (1994, 2010), Czech Republic (1996, 2013), Germany (1994, 2013), Estonia (2000, 2013), Hungary (1994, 2012), Mexico (1994, 2012), Slovak Republic (1996, 2013), United Kingdom (1994, 2013) and the United States (1994, 2013). They are based on data from the European Community Household Panel (ECHP) and the EU Survey of Income and Living Conditions (EU‑SILC) for Austria, Belgium, Denmark, Finland, France, Greece, Ireland, Italy, Luxembourg, the Netherlands and Spain. Data for these countries refers to the weighted averages of the 1995‑2000 ECHP samples for the mid‑1990s, and the 2009‑14 EU‑SILC samples for the mid‑2010s.
Figure 3.3. Jobs polarised in OECD countries between the mid-1990s and mid-2010s
The net shift in employment to highly skilled jobs is why job polarisation has not led to a generalised decline in the share of workers in the middle-income class. In fact, while the decline in the share of middle-skill jobs has tended to reduce the number of workers in middle-income households, the increase in the share of high-skill jobs in the labour market has had the opposite effect. As a result, the main driver of both rises and falls in the overall share of workers in the middle-income class is not changes in the size of different occupations (i.e. job polarisation), but the fact that the distribution of these occupations across the income classes has changed – even holding their size constant (Manfredi and Salvatori, forthcoming[3]). The generally modest variations in the size of the middle-income class thus mask two very significant developments for policy:
1. The occupational make-up of the middle-income class has changed substantially.
2. The probability that different skill groups are in the middle class has changed over the past two decades.
The skills composition of middle-income workers
High-skilled workers now outnumber the middle-skilled in the middle-income class. Indeed, their share increased in every single country between the mid‑1990s and mid‑2010s from an OECD average of 35% to 47% (Figure 3.4). And although the rise was part of an across-the-board increase in shares of highly skilled workers, it was particularly pronounced in the middle-income class. One reason was that middle-skill workers struggled increasingly to secure their place in the middle-income class and their highly skilled peers to break into the upper income class. In fact, the share of middle-skilled workers in the middle-income class declined from 41% to 32%. Only in Mexico and the Slovak Republic was there no fall (Figure 3.5).
Figure 3.4. The middle-income class is increasingly highly skilled and less middle-skilled
Note: Results at individual level for working adults (see Box 3.1 for details). Results for lower- and upper-income classes are available in Manfredi and Salvatori (forthcoming[3]).
Source: OECD calculations based on LIS, ECHP and EU‑SILC (see Box 3.1 for details).
Link between skill levels and income classes are weakening
Skill levels are increasingly failing to yield entry into the income class with which they are traditionally associated. Highly skilled workers, for example, are less and less likely to belong to the upper income class in most countries (Figure 3.5). From one-quarter in the mid‑1990s, they accounted for only one-fifth of workers in the class 20 years later. Similarly, although most middle-skill workers live in middle-income households, the likelihood that they are in the lower-income class has increased in 14 countries. In 12 of them, there is a trend of both low- and middle-skill workers slipping into the lower-income class.4
Changes in the fortunes of the different skill groups may explain some of the social frustration that has been at the centre of the political debate in recent years. Jobs increasingly fail to yield the income status traditionally associated with their skill levels. In most countries, there are fewer prospects of high-skill workers being in the upper-income class, and of middle and low-skill workers in the middle-income class.
Figure 3.5. Changes in the probability that a skill group belongs to a given income class
3.2.2 Changing patterns in working households
Labour-market and demographic changes have significantly modified the occupational make-up and employment status of middle-income households.
More single adults and fewer single-earner couples in the middle-income class
Middle-income working households increasingly comprise single adults, while the share of couples with two earners remains much the same and that of single-earner couples has fallen. Those trends largely reflect changes in the composition of households in all income classes. Single-adult households make up a growing proportion of the middle-income class in most countries, with the average proportion rising from 30% to 37% between mid‑1990s and mid‑2010s (Figure 3.6).
The share of single-earner couples in the middle-income class has dwindled. From almost one-quarter it has dropped to less than one-fifth, with particularly steep declines in Luxembourg, Mexico and Spain. In fact, only in five countries do single-earner households account for more than one-fifth.
The share of dual-earner couples in the middle-income class has, on average, remained stable, though trends vary from country to country. In Denmark, the Czech Republic and Belgium, it fell considerably, but rose sharply in Austria, Mexico and Spain.
Skills and income levels in dual-earner households
Dual-earner households where one or both workers are highly skilled are more likely than two decades ago to be middle income – chiefly because they are increasingly less likely to be in the upper-income class. The probability of dual-earner households being in the middle-income class depends on skill level, but varies from country to country and over time. OECD-wide, those where one or both members are highly skilled have a greater chance of being middle income and less of belonging to the upper-income class than in the mid‑1990s (Figure 3.7, Panel A). However, some countries buck that trend. In Denmark and the United States, dual-earner couples where one or both workers is highly skilled are now more likely to be upper than low income. As for Canada and Estonia, the probability of their being middle income has fallen, as they stand a greater chance of being in the lower-income class.
On average across OECD countries, couples without high-skilled workers are as likely as being middle-income as two decades ago, they more likely being lower income and less likely being upper income (Figure 3.7, Panel B). In five countries, they are more likely to be middle income, though in all five (save Italy) it is because fewer are in the upper-income class. Only in Austria and Belgium is there a higher probability than in the past that dual-earner couples who are not highly skilled belong to the upper-income class.
Figure 3.6. There are more single-adults and fewer single-earner couples in the middle-income class
Note: Results at household level for working households (see Box 3.1 for details). Results for lower- and upper-income classes are available in Manfredi and Salvatori (forthcoming[3]).
Source: OECD calculations based on LIS, ECHP and EU‑SILC (see Box 3.1 for details).
Figure 3.7. Dual-earner couples where one or both workers are highly skilled have fared better than those with lower skill levels
3.3 Wage distribution
This section considers how job polarisation has affected the distribution of wages. It looks into changes in the wage distribution, particularly how wages in the middle have evolved compared to those at the bottom and top of the distribution.
Job polarisation has led to wage inequality rather than wage polarisation. In most labour markets where polarisation has taken place, wages have increased fast at the top but not at the bottom of the distribution. In the majority of OECD countries, the gap between mid‑scale and low wages has either remained unchanged or widened (Figure 3.8). Wages, therefore, have not polarised – i.e. grown less in the middle than at the bottom and top of the scale. Instead, they have become more unequal – i.e. risen faster at top pay levels.5
In the United States, top pay levels have been increasing faster than at the middle since the 1980s, making the wage distribution more unequal (Spletzer and Handwerker, 2014[4]; Verdugo, 2014[5]). Only in the 1990s did wages polarise. In comparison to those in the middle of the distribution, wages at the bottom fell in the 1980s and remained flat in the 2000s (Acemoglu and Autor, 2011[6]; Autor, 2015[2]). Low productivity growth and very elastic labour supply could be one reason why, compared to middle-skill occupations, the number but not the wages of lower-skilled occupations has grown (Autor, 2015[2]). Evidence from OECD countries has also found more widespread and substantial rises in upper- than in lower-tail inequality (Atkinson, 2008[7]).
Recent evidence from European countries suggests that “occupational dynamics did not drive developments in wage inequality in the last decade”. Despite substantial changes in the occupational structure of employment in the wake of the global financial crisis, most changes in wage inequality between 2005 and 2014 stemmed from changes in the distribution of wages within rather than between occupations (Eurofound, 2017[8]).
Figure 3.8. Wages have become more dispersed than polarised
Note: “Top-middle wage ratio” is the ratio between upper wage in the ninth decile and the median wage.
“Middle-bottom wage ratio” is the ratio between the median wage and the upper wage in the first decile.
Source: OECD Earnings Database.
3.4 Non-standard work and middle-class security
The spread of non-standard forms of work in the labour market may increase the sense of insecurity and anxiety in the middle-income class. Non-standard work includes temporary, part-time and self-employment. Although it affords greater flexibility in working arrangements, it provide less employment protection, fewer social rights, less training opportunities and lower earnings predictability – all associated with the preferences of the middle class. This section looks into how trends in part-time work and self-employment have affected the middle-income class.6
Part-time workers are seldom the heads of middle-income households. Of 17 OECD countries with available data, only 8% of middle-income working households are headed by part-time workers (Figure 3.9). The share is greater in the poor and low-income classes and smaller among upper-income households. Part-time workers head a large share of middle-income households in the Netherlands, Ireland, Austria, Switzerland, Germany and Israel. And in the Netherlands, the share of households headed by part-time workers is greater in the upper-income class than in any other. In emerging economies, the incidence of poor households headed by part-time workers is particularly high – more than one in five in Russia and South Africa and one in three in Brazil.
Part-time workers head more middle-income households than three decades ago. The share of people living in middle-income households where the main earner is a part-time worker has more than doubled in the last three decades. In the 15 OECD countries with available data, the average share rose from 3% to 7.5% between the mid‑1980s and mid‑2010s. In the Netherlands, Switzerland and Ireland, the likelihood of part-time workers heading middle-income households climbed by 10 percentage points or more. By contrast, it barely changed in Canada, Czech Republic, Hungary, Mexico and the Slovak Republic.
Figure 3.9. The share of part-time workers in middle-income households has increased
Note: Poor: below 50% median income. Lower: between 50% and 75%. Middle: between 75% and 200%. Upper: above 200%.
Source: OECD calculations based on LIS, ECHP and EU‑SILC (see Box 3.1 for details).
The self-employed account for a sizeable share of working middle-income households, though shares are greater in poor and upper-income households. In the 21 OECD countries with available data, the self-employed head 9% of middle-income households (Figure 3.10), ranging from less than 4% in Estonia to almost 20% in Mexico. The share is higher in poor households at 12% and a particularly high at 21% in the upper-income class. As for lower-income households, 7% are headed by self-employed workers.
In some emerging economies, a considerable share of middle-income households are headed by the self-employed. In Brazil and China, for example, shares exceed least 20%, though they are lower in Russia and South Africa.
In most countries, fewer self-employed workers head middle-income households than three decades ago. In the 20 OECD countries with available data, the average share of middle-income households headed by self-employed workers fell by 2 percentage points between the mid‑1980s and mid‑2010s. The fall was particularly steep in Ireland, Italy, Mexico, Norway and Spain. Only in Austria did the share increase considerably.
Figure 3.10. Shares of the self-employed among heads of middle-income households have fallen
Note: Poor: below 50% of the median income. Lower: between 50% and 75% of the median income. Middle: between 75% and 200% of the median income. Upper: over 200% of the median income.
Source: OECD calculations based on LIS, ECHP and EU‑SILC (see Box 3.1 for details).
3.4 The future of middle-class jobs
Rapid innovation in automation and digitalisation is fuelling fears that technology will generate massive unemployment in the future. This section assesses the risks of automation and employment prospects for the current occupations of middle-income workers. While not forecasts,7 the assessment does seek to illustrate and provide pointers as to the potential level of labour market change in future decades among current middle-income workers.
3.4.1 Risk of automation for current middle-income jobs
People are increasingly anxious over the impact of technology on employment (Mokyr, Vickers and Ziebarth, 2015[9]). According to a national survey by the Pew Research Center (2016[10]), two-thirds of middle-income American families expect that, within 50 years, robots and computers will “definitely” or “probably” do much of the work currently performed by humans. In Europe, more than two-thirds of the population think that robots and artificial intelligence will destroy more jobs than they create (European Commission, 2017[10]).
Technological innovation has contributed to employment growth both historically and in recent times. One reason is that recent technological change has created new jobs, such as webmasters and software designers. A United States study found that, in the last three decades, employment growth has been greater in occupations with more new job titles (Acemoglu and Restrepo, 2016[11]). More importantly, and depending on the precise features of the labour and product markets affected, technological change tends to improve productivity, which spells higher incomes and lower prices (Bessen, 2018[12]; Acemoglu and Restrepo, 2018[13]). They, in turn, increase demand for products and services in the economy, ultimately generating greater labour demand even in sectors not directly affected by innovation (Autor and Salomons, 2018[14]).
The changes in the nature and organisation of work brought about by technology may, however, worsen inequalities. This happens if technology reduces the demand for skills that are abundant in the labour force and increases the demand for scarcer ones. Until the mismatch between the demand and supply of skills is solved, new jobs may widen the wage gap and squeeze middle-income workers (Zia, 2017[15]).
It is, therefore, important to understand the extent to which the jobs held by workers in the middle-income class are prone to shifts in demand stemming from future technological advancements. To assess the number and nature of jobs likely to be affected in coming decades, a number of country-specific studies have built on experts’ predictions as to what occupations could most easily be automated in the future (e.g. (Brzeski, Carsten; Fechner, 2018[16]; Frey and Osborne, 2017[17]) Of course, whether they actually will be automated will depend on a set of factors that go beyond technology alone, including some that are shaped by policy, such as the availability of skills and wage dynamics. Nevertheless, such studies do provide valuable insights into the extent to which future technological developments are likely to put pressure on the middle-income class.
Recent OECD estimates suggest that, OECD-wide, a typical job has a 47% chance of being automated (Nedelkoska and Quintini, 2018[18]). For 14% of jobs, the risk is high – over 70% – while for the other 33% it ranges between 50% and 70%.8 Those most prone to automation are routine jobs with low skill requirements and wages. The workers who are most at risk are the low-skilled and young, while those least at risk cover a broad spectrum – from professionals to social workers. As for countries, the share of jobs at high risk of automation ranges from 33% in Slovakia to around 6% in Norway. Generally speaking, jobs are less automatable in the Netherlands and English-speaking and Nordic countries than in Eastern and Southern Europe, Germany, Chile and Japan (Nedelkoska and Quintini, 2018[18]).
In the 18 OECD countries with available data, an average of 18% of workers currently in the middle-income class work in occupations at high risk of automation. That figure is somewhat less but rather close to the 22% in the low-income class but much higher than the 11% of upper-income earners (Figure 3.11). As for countries, Greece, Israel, Poland, the Slovak Republic, Slovenia and Spain have at least one in five middle-income workers in occupations at high risk of automation. In Finland, Korea and the United States, however, the ratio is lower than one in ten.
Figure 3.11. One in six current middle-income jobs are at high risk of automation
Note: The risk of automation is calculated as the average risk of automation by occupation, weighted by the share of each occupation in the income class.
Source: OECD calculations based on LIS and PIAAC.
3.4.2 Middle-income employment prospects
To complement of the risk of automation addressed above, this section describes employment growth prospects for the current occupations of middle-income workers. Projections by income group (measured using the current occupational structure) show that employment is likely to improve in some sectors, but fall or slow down in others.
Projections for European countries suggest that employment growth will differ considerably from occupation to occupation in the years to 2030.9 Employment will grow considerably in the more highly skilled occupations and fall in those where skill requirements are lower. About half of middle-income workers – technicians, managers and elementary workers – are in occupations likely to undergo strong employment growth (Figure 3.12). By contrast, a quarter of middle-income workers – such as plant and machine operators and service and sales workers – may look forward to limited growth only. Another quarter, who work in craft and related trades and clerical support, are in occupations that can expect large employment losses.
Figure 3.12. Middle-income workers look forward to different employment prospects
Source: OECD calculations based on LIS and Cedefop Employment Forecasts to 2030 (http://www.cedefop.europa.eu/en/publications-and-resources/data-visualisations/skills-forecast)
Overall, the current occupations of middle-income workers enjoy positive employment prospects. Projections point to their growing by an average of 7% in the years to 2030 (Figure 3.13). In Denmark, Ireland, Luxembourg, Spain and the United Kingdom, employment in the current occupations of middle-income workers could grow by 10% or more. Only in Estonia and Germany does it have negative prospects.
Figure 3.13. Employment prospects for the current occupations of middle-income workers are positive
Note: The employment prospects are calculated as the average employment prospect by occupation, weighted by the share of each occupation in the income class.
Source: OECD calculations based on LIS and Cedefop Employment Forecasts by 2030 (http://www.cedefop.europa.eu/en/publications-and-resources/data-visualisations/skills-forecast)
Upper-income occupations have even better employment prospects. Average projected growth is 11% and is particularly high in Denmark, Iceland, Slovenia and the Czech Republic. Employment prospects in the current occupations of low-income workers are also positive, although projections point to a lower OECD average growth rate of 6%. In Estonia, Germany and Poland, they actually suggest a fall in employment. The employment prospects for the current occupations of low-income workers are far dimmer than those of their middle-income peers in most countries. Only in France do low-income workers’ current occupations enjoy considerably brighter prospects than middle-income workers.
3.5 Conclusion
The relation between income class and work status has changed considerably in the last three decades. Fewer workers head households in the middle-income class, as more workers are either lower or upper income than before. Not working worsens income-class prospects far more than 30 years ago among people of working age. Considerably fewer working-age people who are out of work can now provide their households with middle-income living standards.
There is no archetypal middle-income occupation. While lower-income workers are employed chiefly in a few lower-skilled occupations and their upper-income peers in highly skilled ones, middle-income earners are spread across several occupations.
All income classes have experienced up-skilling, as job polarisation has resulted in a net increase in shares of high-skilled jobs in most countries. However, the share of high-skilled workers has grown more in the middle-income than in the other classes because middle-skill workers have become less likely to enter the middle-income class and highly skilled workers less likely to make it into the high-income class. Some jobs are increasingly failing to yield the income status with which they have traditionally been associated. In spite of such changing trends, however, workers in highly skilled jobs are still much more likely to enjoy higher incomes and standards of living.
Moreover, a dual-earner household is no longer in itself an effective way of bringing households into the middle-income class, as couples who are only low- and/or middle-skill workers are increasingly finding themselves in the low-income class.
When it comes job polarisation, most countries therefore grapple not so much with a “middle class” problem as with a “middle skill” problem, as workers with that skill level are now less likely to be in the middle class. Such a shift may help explain some of the social frustration at the centre of political debate – even in countries where the size of the middle class has remained stable.
Encouraging and supporting skills acquisition is still a policy that is fundamental to raising workers’ standards of living. Active labour market policies can play a crucial role to that effect, while ensuring young people are more fully equipped with the skills needed by employers and better able to learn new skills as their careers unfold. Redistribution policies, too, have a part to play in preventing further rises in the proportion of workers excluded from the ranks of the middle class.
Skills acquisition is also essential if workers are to be prepared for the upcoming changes and challenges in the labour market. Employment projections for the next years and estimates of the risks of automation in coming decades suggest that highly skilled occupations will do considerably better than those that require middle and low skills.
Fiscal policies need to adjust to the changing relationship between occupations and household income status. There is a particular need to adapt the design of taxes and social transfers, including in-work benefits, to give middle- and low-skilled workers incentive to upskill and avert the growing risk of slipping into the low- and very-low-income classes.
References
[6] Acemoglu, D. and D. Autor (2011), “Skills, Tasks and Technologies: Implications for Employment and Earnings”, Handbook of Labor Economics, Vol. 4, pp. 1043-1171, http://dx.doi.org/10.1016/S0169-7218(11)02410-5.
[13] Acemoglu, D. and P. Restrepo (2018), Artificial Intelligence, Automation and Work, National Bureau of Economic Research, Cambridge, MA, http://dx.doi.org/10.3386/w24196.
[11] Acemoglu, D. and P. Restrepo (2016), “The Race Between Machine and Man: Implications of Technology for Growth, Factor Shares and Employment”, http://www.nber.org/papers/w22252.pdf (accessed on 9 November 2017).
[30] Arntz, M., T. Gregory and U. Zierahn (2016), “The Risk of Automation for Jobs in OECD Countries: A Comparative Analysis”, OECD Social, Employment and Migration Working Papers, No. 189, OECD Publishing, Paris, http://dx.doi.org/10.1787/5jlz9h56dvq7-en.
[7] Atkinson, A. (2008), The changing distribution of earnings in OECD countries, Oxford University Press, https://books.google.fr/books?hl=en&lr=&id=KAIUDAAAQBAJ&oi=fnd&pg=PR11&dq=Atkinson,+Anthony+B.,+2008.+The+Changing+Distribution+of+Earnings+in+OECD+Countries&ots=QkJQrHskRY&sig=FXuUenQUi2wQV57cCE8H6Cx8daw#v=onepage&q&f=false (accessed on 30 November 2017).
[2] Autor, D. (2015), “Why Are There Still So Many Jobs? The History and Future of Workplace Automation”, Journal of Economic Perspectives, Vol. 29/3, pp. 3-30, http://dx.doi.org/10.1257/jep.29.3.3.
[14] Autor, D. and A. Salomons (2018), Is Automation Labor-Displacing? Productivity Growth, Employment, and the Labor Share, National Bureau of Economic Research, Cambridge, MA, http://dx.doi.org/10.3386/w24871.
[34] Autor, D. and A. Salomons (n.d.), Is automation labor-displacing? Productivity growth, employment, and the labor share, https://www.brookings.edu/wp-content/uploads/2018/03/1_autorsalomons.pdf (accessed on 20 November 2018).
[12] Bessen, J. (2018), “AI and Jobs: The Role of Demand”, https://www.nber.org/chapters/c14029 (accessed on 23 November 2018).
[27] Bowles, J. (2014), The computerisation of European jobs | Bruegel, Bruegel Blog Post, http://bruegel.org/2014/07/the-computerisation-of-european-jobs/ (accessed on 3 December 2017).
[16] Brzeski, Carsten; Fechner, I. (2018), “Die Roboter kommen (doch nicht?)”.
[26] Brzeski, C. and I. Burk (2015), “Die Roboter kommen. Folgen der Automatisierung für den deutschen Arbeitsmarkt [The Robots Come. Consequences of Automation for the German Labour Market]”, ING DiBa Economic Research, https://www.ing-diba.de/pdf/ueber-uns/presse/publikationen/ing-diba-economic-research-die-roboter-kommen.pdf (accessed on 3 December 2017).
[33] Cortes, G., N. Jaimovich and H. Siu (2017), “Disappearing routine jobs: Who, how, and why?”, Journal of Monetary Economics, Vol. 91, pp. 69-87, http://dx.doi.org/10.1016/J.JMONECO.2017.09.006.
[8] Eurofound (2017), Occupational change and wage inequality: European Jobs Monitor 2017, Publications Office of the European Union, Luxembourg, http://dx.doi.org/10.2806/332137.
[10] European Commission (2017), Attitudes towards the impact of digitisation and automation on daily life, Special Eurobarometer 460, http://dx.doi.org/10.2759/835661.
[25] Frey, C. and M. Osborne (2017), “The future of employment: how susceptible are jobs to computerisation?”, Technological Forecasting and Social Change, http://www.sciencedirect.com/science/article/pii/S0040162516302244 (accessed on 24 August 2017).
[17] Frey, C. and M. Osborne (2017), “The future of employment: How susceptible are jobs to computerisation?”, Technological Forecasting and Social Change, Vol. 114, pp. 254-280, http://dx.doi.org/10.1016/j.techfore.2016.08.019.
[23] Horemans, J. and I. Marx (2017), “Poverty and Material Deprivation among the Self-Employed in Europe: An Exploration of a Relatively Uncharted Landscape”, IZA Discussion Paper Series, No. 11007, http://www.iza.org (accessed on 3 December 2017).
[24] ILO (2017), World Employment and Social Outlook – Trends 2017, International Labour Organization, http://www.ilo.org/global/research/global-reports/weso/2017/lang--en/index.htm (accessed on 3 December 2017).
[32] Lordan, G. and D. Neumark (2018), “People versus machines: The impact of minimum wages on automatable jobs”, Labour Economics, Vol. 52, pp. 40-53, http://dx.doi.org/10.1016/J.LABECO.2018.03.006.
[3] Manfredi, T. and A. Salvatori (forthcoming), “Job polarisation and the changing work profile of the middle-income class”, ELS Working Paper, No. forthcoming, OECD.
[22] Matsaganis, M. et al. (2016), Non-standard employment and access to social security benefits, http://ec.europa.eu/social/BlobServlet?docId=15687&langId=en.
[9] Mokyr, J., C. Vickers and N. Ziebarth (2015), “The History of Technological Anxiety and the Future of Economic Growth: Is This Time Different?”, Journal of Economic Perspectives, Vol. 29/3, pp. 31-50, http://dx.doi.org/10.1257/jep.29.3.31.
[18] Nedelkoska, L. and G. Quintini (2018), “Automation, skills use and training”, OECD Social, Employment and Migration Working Papers, No. 202, OECD Publishing, Paris, http://dx.doi.org/10.1787/2e2f4eea-en.
[36] Nedelkoska, L. and G. Quintini (2018), “Automation, skills use and training”, OECD Social, Employment and Migration Working Papers, No. 202, OECD Publishing, Paris, https://dx.doi.org/10.1787/2e2f4eea-en.
[35] OECD (2019), OECD Employment Outlook 2019.
[1] OECD (2017), “How technology and globalisation are transforming the labour market”, in OECD Employment Outlook 2017, OECD Publishing, Paris, http://dx.doi.org/10.1787/empl_outlook-2017-7-en.
[20] OECD (2015), In It Together: Why Less Inequality Benefits All, OECD Publishing, Paris, http://dx.doi.org/10.1787/9789264235120-en.
[21] OECD (2015), “Non-standard work, job polarisation and inequality”, in In It Together: Why Less Inequality Benefits All, OECD Publishing, Paris, http://dx.doi.org/10.1787/9789264235120-7-en.
[31] OECD (2014), OECD Employment Outlook 2014, OECD Publishing, Paris, http://dx.doi.org/10.1787/empl_outlook-2014-en.
[29] Pew Research Center (2016), Public Predictions for the Future of Workforce Automation, Pew Research Center, http://www.pewresearch.org (accessed on 9 November 2017).
[4] Spletzer, J. and E. Handwerker (2014), “Measuring the distribution of wages in the United States from 1996 through 2010 using the Occupational Employment Survey”, Monthly Labor Review, http://dx.doi.org/10.21916/mlr.2014.18.
[19] Vaughan-Whitehead, D., R. Vazquez-Alvarez and N. Maître (2016), “Is the world of work behind middle-class erosion?”, Chapters, pp. 1-61, https://ideas.repec.org/h/elg/eechap/17301_1.html (accessed on 31 January 2018).
[5] Verdugo, G. (2014), “The great compression of the French wage structure, 1969–2008”, Labour Economics, Vol. 28, pp. 131-144, http://dx.doi.org/10.1016/J.LABECO.2014.04.009.
[28] World Bank (2016), World Development Report 2016: Digital Dividends, The World Bank, http://dx.doi.org/10.1596/978-1-4648-0671-1.
[15] Zia, Q. (2017), “The Not-So-Dire Future of Work”, Project Syndicate, https://www.project-syndicate.org/commentary/future-of-work-labor-market-reform-by-zia-qureshi-2017-10 (accessed on 9 November 2017).
Notes
← 2. This section focuses on the implications of changes in the demand for different types of jobs (i.e. job polarisation) for the middle-income class. See OECD (2019[35]), instead, for an analysis of how the chances of different workers (by gender, age and education) to access jobs of different pay levels have been changing as labour markets continue to polarise. The main conclusion in OECD (2019[35]) is that the young with less than tertiary education emerge as a group that has seen a considerable deterioration in their labour market outcomes in recent years (2006‑16). In particular, across the OECD, they have seen an increase in the probability of being neither employment nor in education as well as an increase in the probability of low-paid employment.
← 3. The OECD study measures globalisation against the yardsticks of involvement in global value chains and penetration of Chinese imports (OECD, 2017[1]).
← 4. Further analysis zooming into the years of the recession and its immediate aftermath (2007‑13) shows that the shift of high-skill workers away from the upper-income class was not particularly pronounced. By contrast, middle-skill workers saw a more rapid decline in the probability of being in the middle-class. Finally, the recession also accelerated the shift of low-skill workers into the lower-income class (Manfredi and Salvatori, forthcoming[3]).
← 5. While wages have not polarised, job polarisation has seen middle-wage workers increasingly transitions into non-employment and low-skilled employment (Lordan and Neumark, 2018[32]; Cortes, Jaimovich and Siu, 2017[33]).
← 6. Due to lack of data, temporary contracts are not analysed.
← 7. Projections for current middle-income occupations are not forecasts of the future employment prospects of middle-income workers, as they do not account, among other things, for two important dimensions. First, middle-income workers may change their current jobs and, in doing so, switch occupations. Second, workers profiles change with differences in the skill levels and occupations of those entering and leaving the labour market.
← 8. These figures are computed following an approach based on characteristics of occupations (Frey and Osborne, 2017[17]), which takes into account the heterogeneity of tasks within occupations. For each occupation, a score that measures the likelihood of automation is computed from a set of required skills. For instance, the inclusion of dexterity increases the automation score. The ability to advise others or plan the work of others reduces the automation score. Studies that take a similar approach have estimated automation scores for Germany (Brzeski and Burk, 2015[26]), European countries (Bowles, 2014[27]) and developing countries (World Bank, 2016[28]).
← 9. For European projections, see www.cedefop.europa.eu/en/publications-and-resources/data-visualisations/skills-forecast.