Usefulness of the article: The rise of new technologies that began in the 1980s, with the rapid spread of robotics and information technology, has profoundly changed the way businesses operate. This article summarizes how these long-term trends have transformed the labor market in Western countries and reflects on the prospects for innovation related to artificial intelligence.
Summary:
- Starting in the 1980s, the wave of innovation linked to robotics and information technology significantly increased the remuneration of skilled workers and led to a decline in jobs in the middle of the wage distribution.
- This trend has contributed to widening the wage gap between graduates and non-graduates, but has had an ambiguous effect on career trajectories. The latter depends on each individual’s intrinsic ability to retrain, but also on the role played by companies in continuing education.
- Artificial intelligence could partially reverse the trend by increasingly replacing skilled jobs, but it is unlikely to affect the earnings of the richest 1% of workers.
- The COVID-19 crisis could provide new impetus for the automation of certain jobs with a high risk of viral contamination, affecting low-skilled occupations more, particularly those performed mainly by women.
Since the Industrial Revolution, production methods have continued to evolve and transform over time. However, what characterizes the period between the early 1980s and today is a significant acceleration in the pace of technological change, making adaptation increasingly difficult (Brynjolfsson & McAfee, 2012). This note summarizes how changes in production methods over the last four decades have affected the labor market in Western countries, and concludes with a reflection on what can be expected from the next wave of innovation linked to artificial intelligence. The effects of technological change are illustrated using the example of the United States.
Robotics and information technology are two of the major innovations of the late20th century (Figure 1). The blue curve shows that in the early 1980s, industrial robots were virtually non-existent. The period from 1980 to 2000 saw the first boom in robotics, which was mainly adopted in the automotive industry. From the early 2000s onwards, these technologies spread more widely in the manufacturing sector, reaching a density of 15 robots per 1,000 inhabitants in 2015. The development of information technology peaked between the 1980s and the 2000s, when investment in this technology rose from 1.6% to over 4% of GDP in the United States. This trend came to an abrupt halt when the internet bubble burst in 2000 and then stabilized.
These two technologies spread at around the same time in other Western countries. This led to a profound transformation in the way businesses operate, which in turn brought about major changes in the job market.
1. Increased demand for skilled workers
The most striking effect of the rise of new technologies has been a considerable increase in the demand for skilled labor. Katz & Murphy (1992) and Goldin & Katz (1996) were among the first to describe the phenomenon of « technological change biased in favor of skilled workers » and to outline its consequences for inequality in the labor market. Their analysis is based on the observation that the period from the 1980s to the 2000s was characterized by both strong growth in the number of graduates entering the labor market and a widening wage gap between graduates and non-graduates (Figure 2A). On the one hand, the level of education has increased: only 20% of workers had a post-secondary degree in 1970, compared to around 50% in 2012. On the other hand, the pay gap between different levels of qualification has widened: in 1970, a worker without a degree earned around two-thirds of the salary of a worker with a degree, whereas in 2012 they earned only half. These trends imply that during this period, the economy’s need for skilled workers must have increased considerably. Indeed, as Figure 2B shows, the increase in the population’s level of education (shift in labor supply) would have predicted a narrowing of wage gaps in the absence of an even greater increase in labor demand.
The widespread adoption of robots and computer tools during this period is often considered to be the main factor behind the increase in wage inequality between graduates and non-graduates. Nevertheless, Autor (2014) notes that, with the acceleration in university enrollment in the early 2000s, wage gaps began to stabilize in the United States. He therefore concludes that the increase in wage inequality is not inevitable, and that policies encouraging higher education can reverse the trend by leveraging the labor supply.
2. The decline of jobs in intermediate occupations
In addition to favoring workers with degrees, new production technologies have a second effect: the polarization of the labor market. Autor, Levy, & Murnane (2003) show that the main effect of robotics and information technology has been to replace routine tasks previously performed by employees in intermediate occupations. Table 1 summarizes their classification of the technological impact according to the type of tasks involved in each occupation. They confirm that skilled analytical professions such as engineers, doctors, lawyers, and managers are favored by these innovations because they complement them. However, they also show that the workers most negatively affected are not those with the lowest level of education, but rather those in intermediate jobs. Among service employees, these include accountants, secretaries, and customer service representatives—mainly replaced by IT—and among manufacturing workers, these include machine and assembly line operators—mainly replaced by robotics. Low-skilled manual jobs, such as delivery drivers, nursing assistants, and cleaners, are virtually unaffected by this phenomenon, as they involve tasks that are difficult to replace with machines.
The fact that new technologies are mainly replacing jobs in the middle of the wage distribution has led to polarization in the labor market. Autor et al. (2006) show that between 1990 and 2000, employment in intermediate occupations declined, while employment at both ends of the distribution grew (Figure 3). The latter was more significant in skilled occupations due to their complementarity with machines, but also affected lower-paid occupations. This phenomenon is not unique to the United States but can be observed across all industrialized countries. Goos, Manning, & Salomons (2014) show that the same trends are visible in 16 Western European countries, and Harrigan, Reshef, & Toubal (2016) document this for France. Nevertheless, this phenomenon seems to have faded since the 2000s, probably due to a slowdown in the demand for skills, while the supply of skilled workers continues to increase (Beaudry, Green, & Sand, 2016).
3. An ambiguous effect on career trajectories
Empirical evidence has reached a consensus on the diagnosis that new production technologies related to robotics and information technology are destroying jobs in the middle of the wage distribution and creating jobs at the top of the distribution, without changing much for occupations at the bottom of the distribution. Nevertheless, the effect on total employment and career trajectories remains ambiguous and seems to depend on the specific context under consideration. Bessen (2017) explains how the effect of new technologies on total employment also depends on the impact on demand for goods produced with these innovations. The latter is set to grow due to lower prices associated with lower production costs (Graetz & Michaels, 2018). Dauth et al. (2019) find that the rise of robotics in Germany is destroying jobs in industry but creating just as many in services, and show that employees tend to retrain for better-quality jobs while remaining with the same employer. Similarly, Battisti, Dustmann, & Schönberg (2017) show that routine workers in Germany do not necessarily experience a decline in wages or an increased probability of unemployment. They recognize that companies play a crucial role in retraining their employees by training them in analytical tasks that complement technology. Cortes (2016) points out that in the United States, the consequences vary according to individuals’ intrinsic abilities. The most capable workers succeed in retraining for more skilled analytical jobs and see an increase in their wages. Conversely, workers with fewer skills face an increased risk of downgrading. The effect on job loss, however, appears to be fairly small.
In a recent contribution based on Swedish data, Edin et al. (2019) compare career trajectories over 30 years according to whether or not the jobs individuals held in the mid-1980s—at the very beginning of the technological shock—were routine in nature. They find that workers in routine jobs experienced a decline in their cumulative earnings of around 2% to 5% over the period from 1986 to 2013. This can be explained by a combination of factors, including lower wage growth in these occupations and an increased risk of unemployment and downgrading to lower-paid jobs. Furthermore, they agree with Cortes (2016) in showing that individuals who were already lower paid among routine workers are those who suffer the sharpest decline in income over the period. In a work in progress, Le Moigne (2021) shows that the disappearance of mid-level jobs within companies significantly reduces promotion opportunities for workers at the bottom of the ladder. This suggests that pathways to upward mobility may deteriorate as a result of the disappearance of intermediate jobs.
4. Artificial Intelligence: A reversal of the trend?
In recent years, we have seen the introduction of the next wave of technology linked to artificial intelligence. This recent innovation is driven by the increasingly widespread availability of huge databases— BigData— andby the achievement of the computing power needed to analyze them. While it is still too early to measure the impact of its development on the labor market, a few prospective studies are beginning to emerge and suggest that this technology could, at least in part, reshuffle the deck. Webb (2019) takes the same approach as Autor et al. (2006) by classifying tasks that can be performed by artificial intelligence and identifying which occupations are therefore at risk of being replaced by machines. The results are shown in Figure 4.
Unlike the previous wave of technological change, this one seems to affect the most skilled jobs more. The ability of artificial intelligence to quickly analyze large volumes of data to detect trends, make diagnoses, and make predictions could in the future replace certain tasks currently performed by doctors, lawyers, laboratory analysts, etc. We could therefore expect a downward leveling of the wage distribution. However, this would not necessarily imply a reduction in inequality in our societies. Workers in STEM (Science, Technology, Engineering, and Mathematics) professions will remain highly complementary to artificial intelligence, and capital owners will be even more favored—two groups that already represent the wealthiest segments of society today. Webb (2019) estimates that artificial intelligence will reduce wage inequality between the top 10% and the bottom 10%, but will not affect the profits of the top 1%.
A report by France Stratégie on this subject points out that, as with previous revolutions, the effect on total employment is ambiguous and could even be positive due to the productivity gains enabled by artificial intelligence (Benhamou & Janin, 2018). This report maintains that the challenges raised by AI are similar to those described for previous innovations: the challenge will be to support professional retraining towards tasks where human contact remains crucial, such as supervision or reception. Nevertheless, it also highlights an increasingly significant risk of « a loss of employee autonomy, subject to increasingly insidious automated control, with the associated psychosocial risks. »
5. Conclusion
The rise of robotics and computerization technologies has contributed to increasing wage inequality by increasing the demand for skilled workers and destroying jobs with average pay levels. These structural changes have sometimes polarized individuals’ career paths by inhibiting social mobility, but in other contexts they have enabled workers to retrain for more interesting and better-paid jobs. Continuing education and support during career transitions play a crucial role in helping workers already in the labor market to adapt. In addition, access to higher education should be encouraged by increasing the number of places available in technical fields, in order to adapt the skills of new generations to the needs of the future. Artificial intelligence could reduce the wage advantage in certain skilled occupations, but it is not expected to reduce the level of inequality in our societies on its own. In addition, the COVID-19 health crisis could provide new impetus for the automation of certain jobs with a high risk of viral contamination, affecting low-skilled occupations more, particularly those performed mainly by women (Chernoff & Warman, 2020).
References
Autor, D. H. (2014). Skills, education, and the rise of earnings inequality among the “other 99 percent.” Science. https://doi.org/10.1126/science.1251868
Autor, D. H., Katz, L. F., Kearney, M. S., Berman, E., & Chandra, A. (2006). The polarization of the U.S. labor market. In American Economic Review (Vol. 96, pp. 189–194). https://doi.org/10.1257/000282806777212620
Autor, D. H., Levy, F., & Murnane, R. J. (2003). The skill content of recent technological change: An empirical exploration. Quarterly Journal of Economics. https://doi.org/10.1162/003355303322552801
Battisti, M., Dustmann, C., & Schönberg, U. (2017). Technological and Organizational Change and the Careers of Workers.
Beaudry, P., Green, D. A., & Sand, B. M. (2016). The Great Reversal in the Demand for Skill and Cognitive Tasks. Journal of Labor Economics, 34(1), S199–S247. https://doi.org/10.1086/682347
Benhamou, S., & Janin, L. (2018). Artificial Intelligence and Work. Report to the Minister of Labor.
Bessen, J. E. (2017). AI and Jobs: The Role of Demand. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.3078715
Brynjolfsson, E., & McAfee, A. (2012). The Race Against the Machine. Digital Frontier Press.
Chernoff, A., & Warman, C. (2020). COVID-19 and Implications for Automation. National Bureau of Economic Research. https://doi.org/10.3386/w27249
Cortes, G. M. (2016). Where Have the Middle-Wage Workers Gone? A Study of Polarization Using Panel Data. Journal of Labor Economics, 34(1), 63–105. https://doi.org/10.1086/682289
Dauth, W., Findeisen, S., Suedekum, J., & Woessner, N. (2019). The Adjustment of Labor Markets to Robots. IAB-Discussion Paper, No. 30.
Edin, P.-A., Evans, T., Graetz, G., Hernnäs, S., & Michaels, G. (2019). Individual Consequences of Occupational Decline. CEP Discussion Papers.
Goldin, C., & Katz, L. F. (1996). Technology, Skill, and the Wage Structure: Insights from the Past. American Economic Review. https://doi.org/10.2307/2118132
Goos, M., Manning, A., & Salomons, A. (2014). Explaining job polarization: Routine-biased technological change and offshoring. American Economic Review. https://doi.org/10.1257/aer.104.8.2509
Graetz, G., & Michaels, G. (2018). Robots at work. Review of Economics and Statistics. https://doi.org/10.1162/rest_a_00754
Harrigan, J., Reshef, A., & Toubal, F. (2016). The March of the Techies: Technology, Trade, and Job Polarization in France, 1994-2007. NBER Working Paper Series, 1994–2007. https://doi.org/10.1017/CBO9781107415324.004
Katz, L. F., & Murphy, K. M. (1992). Changes in Relative Wages, 1963-1987: Supply and Demand Factors. The Quarterly Journal of Economics, 107(1), 35–78. https://doi.org/10.2307/2118323
Le Moigne, M. (2021). Exploring the “Fissured Workplace”: Internal Job ladders’ Fragmentation and its effects on plants and workers. Unpublished Manuscript.
Webb, M. (2019). The Impact of Artificial Intelligence on the Labor Market. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.3482150
