- Open Access
The anatomy of job polarisation in the UK
© The Author(s) 2018
- Received: 2 December 2016
- Accepted: 22 June 2018
- Published: 25 July 2018
This paper studies the contribution of different skill groups to the polarisation of the UK labour market. We show that the large increase in graduate numbers contributed to the substantial reallocation of employment from middling to top occupations which is the main feature of the polarisation process in the UK over the past three decades. The increase in the number of immigrants, on the other hand, does not account for any particular aspect of the polarisation in the UK. Changes in the skill mix of the workforce account for most of the decline in routine employment across the occupational distribution, but within-group changes account for most of the decline in routine occupations in middling occupations. In addition, there is no clear indication of polarisation within all skill groups—a fact that previous literature has cited as evidence that technology drives the decline of middling occupations. These findings differ substantially from previous evidence on the US and cast doubts on the role of technology as the main driver of polarisation in the UK.
- Job polarisation
- Occupational mobility
The increasing ability of technology to replace workers in performing easier-to-codify “routine” tasks has been singled out in the literature as the main driver of “job polarisation”, i.e. the decline in the share of mid-pay mid-skill jobs observed in several developed countries (Autor 2014; Goos et al. 2014). Recent contributions, however, have highlighted that the occupational wage patterns observed in many countries do not fit the predictions of the so-called routine-biased technology change (RBTC) hypothesis, igniting a debate among scholars and policy makers on the importance of technology in driving changes in the occupational structure.1
Goos and Manning (2007) used data for 1979–1999 and concluded that compositional changes could not explain the polarisation in the UK. This highly influential paper remains the only paper to have tackled this question [see McIntosh (2013) for a review]. This paper replicates and extends to more recent years the work of Goos and Manning (2007) and provides new insights on the polarisation process in the UK. In particular, the main contribution of this paper is to provide new evidence on the role of changes in the skill mix exploring for the first time the contribution of individual skill groups to the aggregate pattern of polarisation in general and to changes in the share of routine employment in particular. This new evidence provides a more nuanced picture of the implications of changes in the composition of labour supply for the occupational structure than in Goos and Manning (2007) and shows whether any particular feature of the polarisation process can be explained by changes in the relative size of particular skill groups.
As widely recognised in both the labour and trade literature, changes in the skill mix of labour supply can lead to changes in the occupational structure as firms switch to production methods making a more intensive use of the more abundant factor. Dustmann and Glitz (2015) and Lewis (2011) exploit exogenous variation in the supply of immigrants to provide causal evidence that firms adjust production methods to changes in the supply of skills, but isolating exogenous variations in the supply of graduates has generally proven more difficult.3 Recently, Blundell et al. (2016) argues that firms adjustment of their occupational structure to the largely exogenous increase in the supply of graduates over the same time period considered in this paper has led to stable educational premia in the UK. The literature on job polarisation has long acknowledged the potential role of changes in the skill mix as well. As mentioned above, Goos and Manning (2007) concluded that changes in the composition of the workforce cannot explain job polarisation in the UK between 1979 and 1999.4 It is therefore clear from the existing literature that changes in the skill mix of the workforce are a plausible driver of changes in the occupational structure that deserve empirical consideration.
The main evidence presented in this paper on the role of individual skill groups comes from a shift-share analysis which highlights the contribution to overall job polarisation of changes within and between gender-education-age-immigration cells from 1979 to 2009. Goos and Manning (2007) used a version of this approach to show that compositional changes could not explain the overall polarisation of the labour market between 1979 and 1999. They did not consider the role of immigrants (whose share had only increased marginally by 1999) and their results might not capture fully the cumulative effect of the strong growth in graduates that started a few years into the 1990s (Fig. 1). In addition, due to the data limitations of the time, Goos and Manning (2007) did not consider changes in routine employment within the UK at all.5 I use several alternative measures of routiness (Autor 2013) to provide the first evidence on the extent to which compositional changes can account for the decline in routine employment specifically.
Importantly, changes in the relative size of different groups could account for significant features of the polarisation process even if they do not explain the entire pattern as observed by Goos and Manning (2007). To investigate this hypothesis, I present a breakdown of the shift-share analysis which provides the first evidence for the UK on the contribution of different skill groups to changes in employment shares across the occupational skill distribution in each of the past three decades.
The evidence on the contribution of different skill groups to polarisation is of interest for at least two reasons. First, in the absence of credible sources of exogenous variation on either side of the labour market, assessing the relative importance of within vs between group polarisation is crucial for a critical appraisal of the hypothesis that the process is driven by technology. This is why Goos and Manning (2007) use this method, and, more generally, evidence of pervasive polarisation within different skill groups is typically cited in support of a technology effect as it indicates that aggregate polarisation is not merely the result of changes in the relative size of groups specialising in different occupations (Spitz-Oener 2006; Acemoglu and Autor 2011). Second, the evidence on the contribution of individual groups to the aggregate polarisation enhances our understanding of the specific challenges faced by different skill groups in an increasingly polarising labour market.6 Evidence on the contribution of individual skill groups to aggregate polarisation is currently very limited for countries other than the US as I discuss in Sect. 2.
The key findings of the paper can be summarised as follows. The main feature of the polarisation process in the UK has been a shift of employment towards high-paid occupations, which have gained 80% of the employment shares lost by middling occupations. When occupations are ranked by education, it becomes apparent that it is those with the lowest initial level of education that have lost most employment shares. The results of the shift-share analysis suggest that the increase in the share of graduates has contributed significantly to this substantial reallocation of employment from middle-pay to high-pay occupations. The increase in immigrants, on the other hand, does not explain any particular feature of job polarisation in the UK. In addition, there is no clear indication of polarisation within all skill groups—a fact that previous literature has cited as evidence supporting the hypothesis that technology is the main driver of the process. Hence, this paper adds to the growing body of evidence from other countries (Autor 2015; Green and Sand 2015) that casts doubts on the extent to which technology can be seen as the primary driver of the polarisation process.
Goos and Manning (2007) were the first to propose a link between job polarisation and technological change. They argued that the hollowing out of the UK labour market is due to the concentration in middling occupations of “routine” tasks which are easier to automate because they can be executed following a precise set of standard instructions (Autor et al. 2003).
Later studies documented job polarisation in the US (Autor et al. 2006), Canada (Green and Sand 2015), Sweden (Adermon and Gustavsson 2015), Germany (Antonczyk et al. 2018; Kampelmann and Rycx 2011) and across Western Europe (Goos et al. 2009) and pointed to increasing global trade as a possible concurrent cause of the process. Studies that have compared the explanatory power of technology vs offshoring in explaining the decline of routine employment have generally concluded that the latter is more important than the former (Acemoglu and Autor 2011; Goos et al. 2014; Akçomak et al. 2016).7
Demand-based explanations for job polarisation imply that middling occupations should experience a decline in both employment shares and wages as a result of a fall in demand, generating both job and wage polarisation. This pattern, however, is only found in the US in the 1990s (Autor et al. 2006; Acemoglu and Autor 2011), but there is no evidence of the contemporaneous occurrence of job and wage polarisation in the 2000s in the US nor at any other time in any of the countries studied in the literature (Dustmann et al. 2009; Mishel et al. 2013; Autor 2015; Green and Sand 2015).8 For the UK in particular, the working paper version of this article provides evidence that occupational wage growth has not polarised over time (Salvatori 2015).9
Autor (2015) provides an in-depth discussion of this and other puzzles confronting the routine-biased technological change hypothesis (and demand-based explanations more broadly), but here suffice it to say that the heterogeneity of employment and wage patterns across countries and over time suggests that factors other than technology continue to play a significant role. For instance, Autor (2015) and Green and Sand (2015) have emphasised that recent developments in occupational wages at the bottom of the distribution (in the US and Canada respectively) appear consistent with significant shifts in labour supply as well.
More generally, the potential role of changes on the supply side of the labour market have long been acknowledged in this literature in light of the previous evidence from the trade and labour literature that firms adjust their production technologies to changes in the skill mix of the workforce.10 As discussed in the Introduction, Goos and Manning (2007) used a shift-share analysis to address this issue in their study on the UK for the years 1979–1999. Oesch (2013) offers an analysis similar to that of Goos and Manning (2007) for five European countries (including the UK) over the period 1993–2008 which shows that changes in the education and age mix of the workforce are associated with growth in high-skilled occupations. Neither of these studies includes immigrants in the compositional analysis, although Oesch (2013) documents the increase in their numbers and acknowledges their potential role.
As noted by Autor and Dorn (2009), job polarisation might be affecting different skill groups in different ways. Most (if not all) of the direct evidence on this issue, however, comes from the US where the decline in middle skill employment of non-college workers has mostly been compensated by increases in bottom occupations, while for college workers both low skill and high skill occupations have grown (Autor and Dorn 2009). No comparable evidence is reported in existing studies on Germany (Antonczyk et al. 2018; Kampelmann and Rycx 2011), Sweden (Adermon and Gustavsson 2015) and Europe at large (Goos et al. 2009, 2014). For the UK, in particular, both Goos and Manning (2007) and Oesch (2013) focus on aggregate results providing no evidence on the changing occupational structure within individual skill groups or on the contribution of the individual groups to overall polarisation.
I use four different datasets covering the period 1979–2012. Data on occupational shares and socio-demographic characteristics come from the Labour Force Survey (LFS) which was carried out biannually from 1979 to 1983, then annually from 1984 to 1991, and finally quarterly from 1992 onwards. Between 1979 and 2012, the LFS uses four different occupational classifications (KOS, SOC90, SOC00, SOC10) (ONS 2002). There is some evidence from the US that different ways of bridging occupational classifications might lead to different results (Lefter and Sand 2011) but the issue has not been investigated with UK data. I use probabilistic matching and investigate the sensitivity of the conversion procedure to conditioning on different observable characteristics. To the extent that groups grow over time at the different rates, different ways of reallocating them across occupations might affect the estimates of changes in the size of occupations over time. Section 4 shows the results obtained converting across occupational classifications conditionally on education and unconditionally (as in Goos and Manning 2007) and the interested reader is referred to Additional file 1: Appendix S1 for more details. In addition, Additional file 1: Appendix S2 discusses the issues encountered in the LFS when measuring the share of graduates over time and the education of foreign-born workers. To maximise comparability with existing studies and facilitate the application of the shift-share analysis, I follow Goos and Manning (2007) and focus on changes in occupational shares by employment deciles at the beginning of the observation period.11
Because the LFS did not collect data on earnings until 1993, the wage data come from the panel dataset combining the New Earnings Survey (NES, 1979–2002) and its successor the Annual Survey of Hours and Earnings (ASHE), available from the UK Data Archive as NESPD. Wage information is provided by the employer and is therefore regarded as very reliable, but this dataset has no education variables and very limited information on workers’ characteristics in general.
Building on Goos and Manning (2007), this section documents the evolution of polarisation in the UK over the past three decades and offers new and important robustness checks that have been carried out for other countries but not for the UK. In particular, as discussed in Sect. 3, it considers the robustness of the results to different ways of bridging breaks in the occupational classifications and to ranking occupations by education rather than wages.
4.1 Changes in employment shares along the educational distribution
The difference in the results by wage and education rankings is in contrast with previous evidence for the US (Autor et al. 2006) and Canada (Green and Sand 2015). While the wider implications of the differences between the results by wage and education warrant further research, for the purpose of this paper I stress that the presence of a clear upgrading patter in occupations by education provides the first indication that increasing educational attainment might have contributed to shaping the changes in the occupational structure of the UK. The next section presents the results of a shift-share analysis to shed more light on this issue.
According to the RBTC hypothesis, the decline in middling occupations is due to the concentration in those jobs of routine tasks which are easier to automate. In fact, this notion has been so widely accepted that measures of routiness are often used as proxy for technology itself (Goos et al. 2014) and changes in routine employment are interpreted as driven by technology. In this section I investigate the extent to which changes in routine employment can be accounted for by compositional changes. I use different classifications of routine occupations based on all three main approaches described in Autor (2013) since no single classification is clearly superior to the others. Here, I offer a brief overview of these three approaches while the interested reader is referred to Additional file 1: Appendix S4 for a more in-depth discussion.
The first approach simply uses occupations as proxies for job tasks, classifying 1-digit occupations based on the task perceived as typical of that occupation. Following Acemoglu and Autor (2011), I classify as routine the following groups: (i) clerical and secretarial occupations, (ii) craft and related occupations, (iii) sales occupations and (iv) plant and machine operatives.
The second approach measures the relative importance of different task dimensions (e.g. routine, abstract, manual) using standardised job descriptors that provide information on the tasks performed in each occupation. I use the “routine task index” (RTI) constructed by David and Dorn (2013) using the 1977 Dictionary of Occupational Titles (DOT) and then mapped to the European ISCO88 (2-digit) classification in Goos et al. (2014, Table 1, henceforth GMS).
The third approach uses job task information collected directly in the British Skill Survey (BSS) to construct an RTI index in the same fashion as David and Dorn (2013) as described in Additional file 1: Appendix S4. To minimise the role played by my own subjective judgement in classifying the 36 available tasks as either routine or non-routine (Green 2012), I follow Akçomak et al. (2016) (henceforth AKR) who use a subset of the tasks split in groups intended to reflect those defined by the task measures in David and Dorn (2013).
Shift-share decomposition of changes in occupational shares (pp) by type of occupations using alternative routine classifications, 1979–2012
Non routine occupations
(A) Routine classification based on Acemoglu and Autor (2011)
(B) Routine classification based on RTI index from Goos et al. (2014)—top 30%
(C) Routine classification based on RTI index from Akçomak et al. (2016)—top 30%a
Routine occupations have declined relative to non-routine under all classifications. The comparison with the 1979 totals reported at the bottom of Table 3 indicates that the fall in routine-occupations is substantial as they have lost around 40% of their employment shares across classifications.14 In panels A and C, most of the decline in routine occupations is accounted for by compositional changes. In Panel B, within-group changes are more important, but the result is not robust to the use of the alternative classifications based on the same GMS RTI index. In all cases, routine occupations account for most of the decline in middling occupations and their contribution here is mostly from within-group changes.
Overall, therefore, two main conclusions can be drawn. First, routine occupations have declined relative to non-routine occupations and they account for most of the decline in middling occupations. Second, the overall decline in routine occupations is mostly accounted for by between-group changes, but the contribution to the decline in middling occupations is instead driven by within-group changes.
As discussed in Sect. 2, the US is the country on which most of the existing literature on job polarisation has focused. As it is customary in related papers (Green and Sand 2015, Antonczyk et al. 2018), it is therefore useful to refer to the US results when discussing those of other countries. This paper shows that the evolution of job polarisation in the UK appears substantially different from that documented for the US in previous literature. While employment growth in the US has progressively favoured bottom occupations and was only polarised in the 1990s, in the UK polarisation occurred in each of the last three decades and growth in high-skill occupations has always exceeded that in low skill ones. Overall, between 1979 and 2012, top occupations gained about 16 of the 19 pp of employment shares lost by middling occupations.
What exactly drives the patterns observed in the US is the subject of an on-going debate (Autor 2015; Beaudry et al. 2016) and beyond the scope of this paper, but the differences between two similarly developed countries suggest that factors other than (broadly similar) technological change might be at play (Green and Sand 2015; Antonczyk et al. 2018). The results of the shift-share analysis indicate that the increase in the educational attainment of the workforce is likely to have contributed significantly to the most prominent feature of the polarisation process in the UK, i.e. the substantial reshuffling of employment from middling to top occupations.
Unlike in the US (Autor and Dorn 2009; Autor 2014), graduates have played no role in the decline of middling occupations, but the increase in their numbers accounts for the entire growth in the top ones from the 1990s—when their growth accelerated dramatically. There is some polarisation within non-graduates (but heavily skewed towards the bottom), but no polarisation within graduates. This is a notable fact since pervasive polarisation within skill groups is cited in the literature as evidence of a technology effect (Spitz-Oener 2006; Acemoglu and Autor 2011). Overall, changes in the relative size of skill groups account for about a third of the decline in middling occupations and for most of the decline in routine occupations across the whole distribution. In addition, I find that the occupations that have lost the largest employment shares are those with the lowest initial level of education. This is a result in stark contrast with that for the US (Autor et al. 2006) and again points to the likely importance of changes in the education mix of the workforce. Moreover, the working paper version of this article (Salvatori 2015) finds no evidence of the polarisation of the occupational wage growth that has been interpreted as evidence in favour of the RBTC hypothesis in previous literature (Autor et al. 2006; David and Dorn 2013). This is in line with results from other countries such as Canada (Green and Sand 2015) and Germany (Dustmann et al. 2009; Antonczyk et al. 2018; Kampelmann and Rycx 2011).
These results indicate that the increase in the share of graduates might help explain why the UK continued to see growth in top occupations in the 2000s when the evidence from the US suggests that technology-driven demand for cognitive skills slowed down (Beaudry et al. 2016). I also find that in the last decade the employment structure of graduates in the UK shifted towards the bottom, a result consistent with the hypothesis that the supply of high skills might have outgrown the demand arising in top occupations. Similar results have been found for Germany for the latest decade (Reinhold and Thomsen 2017). In addition, two recent papers focusing on the UK have also provided evidence that the market for graduates in the UK might be near saturation. Green et al. (2016) report on the rising prevalence of over education and Blundell et al. (2016) argue that a decline in the graduate wage premium since 2010 might be indicative that firms’ ability to absorb the increasing supply by adapting their organisational structure might be exhausting.15
While the growth of graduates can account for the shift of employment from middling to top occupations, it cannot explain the entire polarisation process. In particular, the employment growth in bottom occupations has occurred in spite of the increase in education. Immigrants account for a substantial fraction of net growth in these occupations (mostly in the 2000s), but the most significant change offsetting the downward pressure arising from the increase in education is, by far, the reallocation of native workers with intermediate qualifications from middling occupations into service occupations. Wage growth for these occupations was robust over the past 30 years and the highest across all occupational groups in the 2000s (Salvatori 2015).
The presence of both wage and employment growth in bottom occupations is consistent with the explanations that emphasise increased product demand either from high-skill workers or arising from complementarity in consumption between services and the goods made cheaper by technology (David and Dorn 2013; Mazzolari and Ragusa 2013).
The methods employed in this paper are very closely related to those of Goos and Manning (2007) whose conclusion that compositional changes cannot explain the broad pattern of polarisation still informs the understanding of the process in the UK. While this paper also finds that changes in the skill mix alone cannot explain the entire polarisation process, the results indicate that the increase in the share of graduates in the past 30 years has contributed significantly to what is clearly the distinctive feature of the polarisation process in the UK in comparison to the US, namely the substantial reallocation of employment from middling to top occupations. This paper therefore adds to the growing body of empirical evidence from different countries that cast doubts on technology as the dominant driver of polarisation (Autor 2015; Green and Sand 2015).
Admittedly, a convincing causal analysis is impeded by the lack of credible sources of exogenous variations in the supply of graduates that characterises this literature more widely. Changes in the supply of graduates are typically treated as exogenous in related literature (Autor 2014; Card and Lemieux 2001). More specifically, there are strong reasons to doubt that the surge in the share of graduates in the UK was an endogenous response to changes in technology. In fact, this was in large part a stepwise change following the reforms of the late 1980s16 which led to a 93% increase in participation in higher education between 1988 and 1996, as opposed to only 15% in the US—arguably the technology-leader (OECD 2007). Blundell et al. (2016) note that the increase in graduates in the UK is unmatched in most developed economies and argue that the puzzling fact that the graduate wage premium has not fallen can be explained by changes in the organisation of work implemented by firms to take advantage of the increasing number of graduates.
In conclusion, while the results of this paper suggest that changes in the skill mix have contributed significantly to re-shaping the occupational structure of the UK in recent decades, a difficult but important task for future research is to overcome data limitations and devise empirical strategies to formally assess the relative contribution of demand and supply factors to the polarisation of the UK labour market.
Additional file 1: Appendix S2 discusses data issues affecting the Labour Force Survey (LFS) in 1992.
Beaudry et al. (2010) report evidence that areas with initial higher shares of graduates adopted computers more quickly in the US, and Moretti (2004) uses the lagged city demographic structure and the presence of land-grant colleges to identify the spill-over effects on wages of the supply of graduates across local labour markets in the US.
Oesch (2013) offers an analysis similar to that of Goos and Manning (2007) for the period 1993–2008 and considers education-age cells focusing exclusively on the aggregate results for the compositional effects.
Instead, Goos and Manning (2007) use US data to show that routine occupations are concentrated in the middle of the occupational distribution in that country.
See Sparreboom and Tarvid (2016) for a discussion and analysis of the link between polarisation skill mismatch that also illustrate the importance of understanding whether changing in the occupational structure result from changes in the demand for certain skills by firms or from changes in the supply of such skills.
Recent studies have pointed to the importance of international trade more widely. In particular, Autor et al. (2015) and Keller and Utar (2016) have shown that increasing import penetration from China (to the US and Denmark respectively) has contributed to job polarisation by accelerating the decline of manufacturing employment.
For Sweden, Adermon and Gustavsson (2015) conclude that the RBTC cannot explain between-occupation changes in wages, but can account for some of the changes within occupations.
Goos and Manning (2007) discuss the link between job polarisation and overall wage inequality, but do not report evidence on changes in occupational wages which have been the focus of later papers on other countries.
For a theoretical discussion of changes in the supply of different skill groups in the context of a task-based model see Acemoglu and Autor (2011).
For a discussion of alternative approaches to measuring polarisation see Sparreboom and Tarvid (2016).
I discuss the stark change in the slope of the series in 1992 in Additional file 1: Appendix S2 and conclude data issues likely exaggerate the expansion of top occupations between 1991 and 1992, but the slowdown in their growth in the early 1990 s is likely genuine as also found in other datasets.
In the more conservative classifications in panels B and C, some non-routine occupations also appear to have lost shares, but this result generally does not hold when the top employment-weighted 50% occupations by RTI are considered or when all occupations with an RTI above average are identified as routine.
For the point about the falling wage premium in recent years see: https://www.ifs.org.uk/uploads/publications/bns/bn185.pdf.
See Bolton (2012), Blanden and Machin (2004), Walker and Zhu (2008), Devereux and Fan (2011). More specifically, following a reform of the age 16 examination system in 1988, the share of 17 year old in education climbed from under 30% in 1988 to more than 50% in 1993 greatly expanding the pool of potential university applicants in subsequent years. This change and the expansion in the number of university places available from the early 1990 s also led to a sharp increase in the participation rate in higher education which rose from 19.3% in 1990 to 33% in 2000 (Bolton 2012).
AS is the sole author of this paper and the only one involved in carrying out the underlying research. The author read and approved the final manuscript.
The views expressed are those of the authors only and should not be taken to represent the views of the OECD or its member countries. This research was supported by funding from the UK Economic and Social Research Council (ESRC) under the grant “Job Polarisation and Job Quality” (Grant ES/K001116/1). I thank David Autor, Mike Brewer, Mark Bryan, Tomaz Cajner, Matias Cortes, Ben Etheridge, David Green, Alan Manning, Lawrence Mishel, Matthias Parey, John Schmitt and participants of the 2014 EEA conference in Toulouse, the 2015 RES conference in Manchester, and the 2015 SOLE/EALE conference in Montreal for comments and useful discussions. The usual disclaimer applies.
The author declares that he has no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
- Acemoglu, D., Autor, D.: Chapter 12—Skills, tasks and technologies: implications for employment and earnings. In: Card, D., Ashenfelter, O. (eds.) Handbook of Labor Economics, vol. 4, Part B, pp. 1043–1171. Elsevier, New York (2011)Google Scholar
- Adermon, A., Gustavsson, M.: Job polarization and task-biased technological change: evidence from Sweden, 1975–2005: job polarization and task-biased technological change. Scand. J. Econ. 117(3), 878–917 (2015). https://doi.org/10.1111/sjoe.12109 View ArticleGoogle Scholar
- Akçomak, S., Kok, S., Rojas-Romagosa, H.: Technology, offshoring and the task content of occupations in the United Kingdom. Int. Labour Rev. 155(2), 201–230 (2016). https://doi.org/10.1111/j.1564-913X.2015.00051.x View ArticleGoogle Scholar
- Antonczyk, D., DeLeire, T., Fitzenberger, B.: Polarization and Rising Wage Inequality: comparing the U.S. and Germany. Econometrics 6(2), 20 (2018). https://doi.org/10.3390/econometrics6020020 View ArticleGoogle Scholar
- Autor, D.H.: The ‘task approach’ to labor markets: an overview. J. Labour Mark. Res. 46(3), 185–199 (2013). https://doi.org/10.1007/s12651-013-0128-z View ArticleGoogle Scholar
- Autor, D.: Polanyi’s Paradox and the Shape of Employment Growth. Working Paper 20485. National Bureau of Economic Research (2014)Google Scholar
- Autor, D.H.: Why are there still so many jobs? The history and future of workplace automation. J. Econ. Perspect. 29(3), 3–30 (2015). https://doi.org/10.1257/jep.29.3.3 View ArticleGoogle Scholar
- Autor, D., Dorn, D.: This job is ‘getting old’: measuring changes in job opportunities using occupational age structure. Am. Econ. Rev. 99(2), 45–51 (2009). https://doi.org/10.1257/aer.99.2.45 View ArticleGoogle Scholar
- Autor, D.H., Levy, F., Murnane, R.J.: The skill content of recent technological change: an empirical exploration. Q. J. Econ. 118(4), 1279–1333 (2003). https://doi.org/10.1162/003355303322552801 View ArticleGoogle Scholar
- Autor, D.H., Katz, L.F., Kearney, M.S.: The polarization of the US labor market. Am. Econ. Rev. 96(2), 189–194 (2006). https://doi.org/10.1257/000282806777212620 View ArticleGoogle Scholar
- Autor, D.H., Dorn, D., Hanson, G.H.: Untangling trade and technology: evidence from local labour markets. Econ. J. 125(584), 621–646 (2015). https://doi.org/10.1111/ecoj.12245 View ArticleGoogle Scholar
- Beaudry, P., Doms, M., Lewis, E.: Should the personal computer be considered a technological revolution? Evidence from U.S. metropolitan areas. J. Political Econ. 118(5), 988–1036 (2010)View ArticleGoogle Scholar
- Beaudry, P., Green, D.A., Sand, B.M.: The great reversal in the demand for skill and cognitive tasks. J. Labor Econ. 34(S1), S199–S247 (2016). https://doi.org/10.1086/682347 View ArticleGoogle Scholar
- Blanden, J., Machin, S.: Educational inequality and the expansion of UK higher education. Scott. J. Political Econ. 51(2), 230–249 (2004). https://doi.org/10.1111/j.0036-9292.2004.00304.x View ArticleGoogle Scholar
- Blundell, R., Green, D.A., Jin, W.: The UK Wage Premium Puzzle: How Did a Large Increase in University Graduates Leave the Education Premium Unchanged? IFS Working Paper W16/01. Institute for Fiscal Studies, London (2016)Google Scholar
- Bolton, P.: Education: historical statistics. House of Commons Library (2012)Google Scholar
- Card, D., Lemieux, T.: Can falling supply explain the rising return to college for younger men? A cohort-based analysis. Q. J. Econ. 116(2), 705–746 (2001). https://doi.org/10.1162/00335530151144140 View ArticleGoogle Scholar
- David, H., Dorn, D.: The growth of low-skill service jobs and the polarization of the US labor market. Am. Econ. Rev. 103(5), 1553–1597 (2013)View ArticleGoogle Scholar
- Devereux, P.J., Fan, W.: Earnings returns to the British education expansion. Econ. Educ. Rev. Special Issue Econ. Returns Educ. 30(6), 1153–1166 (2011). https://doi.org/10.1016/j.econedurev.2011.03.004 Google Scholar
- Dustmann, C., Glitz, A.: How do industries and firms respond to changes in local labor supply? J. Labor Econ. 33(3), 711–750 (2015)View ArticleGoogle Scholar
- Dustmann, C., Ludsteck, J., Schonberg, U.: Revisiting the German wage structure. Q. J. Econ. 124(2), 843–881 (2009)View ArticleGoogle Scholar
- Goos, M., Manning, A.: Lousy and lovely jobs: the rising polarization of work in Britain. Rev. Econ. Stat. 89(1), 118–133 (2007)View ArticleGoogle Scholar
- Goos, M., Manning, A., Salomons, A.: Job polarization in Europe. Am. Econ. Rev. 99(2), 58–63 (2009). https://doi.org/10.1257/aer.99.2.58 View ArticleGoogle Scholar
- Goos, M., Manning, A., Salomons, A.: Explaining job polarization: routine-biased technological change and offshoring†. Am. Econ. Rev. 104(8), 2509–2526 (2014). https://doi.org/10.1257/aer.104.8.2509 View ArticleGoogle Scholar
- Green, F.: Employee involvement, technology and evolution in job skills: a task-based analysis. ILR Rev. 65(1), 36–67 (2012)View ArticleGoogle Scholar
- Green, D.A., Sand, B.M.: Has the canadian labour market polarized? Can. J. Econ. Revue Canadienne D’économique (2015). https://doi.org/10.1111/caje.12145 Google Scholar
- Green, F., Felstead, A., Gallie, D., Henseke, G.: Skills and work organisation in Britain: a quarter century of change. J. Labour Market Res (2016). https://doi.org/10.1007/s12651-016-0197-x Google Scholar
- Kampelmann, S., Rycx, F.: Task-Biased Changes of Employment and Remuneration: The Case of Occupations. (2011)Google Scholar
- Keller, W., Utar, H.: International Trade and Job Polarization: Evidence at the Worker-Level. National Bureau of Economic Research, (2016)Google Scholar
- Lefter, A., Sand, B.M.: Job Polarization in the U.S.: A Reassessment of the Evidence from the 1980s and 1990s. Economics Working Paper Series 1103. University of St. Gallen, School of Economics and Political Science, (2011)Google Scholar
- Lewis, E.: Immigration, skill mix, and capital skill complementarity. Q. J. Econ. 126(2), 1029–1069 (2011). https://doi.org/10.1093/qje/qjr011 View ArticleGoogle Scholar
- Mazzolari, F., Ragusa, G.: Spillovers from high-skill consumption to low-skill labor markets. Rev. Econ. Stat. 95(1), 74–86 (2013)View ArticleGoogle Scholar
- McIntosh, S.: Hollowing Out and the Future of the Labour Market. 134. BIS Research Paper. Department for Business, Innovation and Skills, London (2013)Google Scholar
- Mishel, L., Shierholz, H., Schmitt, J.: Don’t Blame the Robots: Assessing the Job Polarization Explanation of Growing Wage Inequality. Working Paper. Economic Policy Institute, (2013)Google Scholar
- Moretti, E.: Estimating the social return to higher education: evidence from longitudinal and repeated cross-sectional data. J. Econom. High. Educ. 121(1–2), 175–212 (2004). https://doi.org/10.1016/j.jeconom.2003.10.015 Google Scholar
- OECD: Education At a Glance 2007: Oecd Indicators. OECD, Paris (2007)Google Scholar
- Oesch, D.: Occupational Change in Europe: How Technology and Education Transform the Job Structure. Oxford University Press, Oxford (2013)Google Scholar
- ONS.: Relationship Between Standard Occupational Classification 2000 (SOC2000) and Standard Occupational Classification 1990 (SOC90) (2002)Google Scholar
- Reinhold, M., Thomsen, S.: The changing situation of labor market Entrants in Germany. J. Labour Market Res. (2017). https://doi.org/10.1007/s12651-017-0227-3 Google Scholar
- Salvatori, A.: The Anatomy of Job Polarisation in the UK.” IZA Discussion Papers. (2015)Google Scholar
- Sparreboom, T., Tarvid, A.: Imbalanced job polarization and skills mismatch in Europe. J. Labour Market Res. 49(1), 15–42 (2016). https://doi.org/10.1007/s12651-016-0196-y View ArticleGoogle Scholar
- Spitz-Oener, A.: Technical change, job tasks, and rising educational demands: looking outside the wage structure. J. Labor Econ. 24(2), 235–270 (2006). https://doi.org/10.1086/499972 View ArticleGoogle Scholar
- Walker, I., Zhu, Y.: The college wage premium and the expansion of higher education in the UK*. Scand. J. Econ. 110(4), 695–709 (2008). https://doi.org/10.1111/j.1467-9442.2008.00557.x View ArticleGoogle Scholar