Job polarisation – growth in the shares of high-wage and low-wage jobs at the expense of middle-wage jobs – has contributed to the increase in income inequality that has taken place in many countries. Sometimes referred to as the ‘vanishing middle class’, job polarisation has implications for a wide range of economic policies including employment, education, trade, and taxation.
Job polarisation has been documented in the US (Autor et al. 2006), the UK (Goos and Manning 2007), and in many countries in Europe, including France (Goos et al. 2007, 2014). A leading explanation for job polarisation is the ‘routinisation hypothesis’ – information and communication technology (ICT) does routine tasks that were formerly done by middle-paid workers, while having less of a direct impact on higher and lower paid occupations. A second explanation for job polarisation is offshoring, where firms replace goods and services produced by middle-paid workers with imports.
Our research sheds new light on the process and extent of job polarisation by using firm-level data from France between 1994 and 2007 (Harrigan et al. 2016). Like many countries during this period, France deepened its integration in the global economy and saw widespread adoption of ICT. Firm-level data allow us to show precisely how polarisation occurs over time, both within and between firms. This is important, since it is firms who make employment decisions. Previous research on job polarisation has invariably used labour force surveys, which are useful for learning about workers’ jobs and wages but are silent about the firms where workers are employed.
Our use of firm-level data also allows us to study the firm-level causes of polarisation. We introduce a novel measure of firm-level technological sophistication – the share of workers in occupations related to the development, management, installation, and maintenance of technology. We call these workers ‘techies’, and one of our key findings is the sustained growth in the share of techies in employment. We show that this ‘March of the Techies’ is broad based – firms on average increase their share of techies over time, and firms with more techies grow faster than less techie-intensive firms. Techies are both part of job polarisation (since they are relatively well paid) and, as we show in our econometric analysis, a causal force behind polarisation.
We also match workers with imports and exports through the firms where they work. This allows us to see which occupations are most exposed to globalisation and also to show how firm-level trade affects polarisation. Our econometric analysis finds that firm-level imports contributed to job polarisation in the manufacturing sector.
We analyse data on hours paid by occupation and firm, which comes from mandatory reports by all French employers to the French national statistical agency, INSEE.
The key to our research is the occupational classification used by all firms. This classification, called the PCS, has 22 occupational categories. The two PCS occupations that we call techies are Technical Mangers and Engineers and Technicians. These two occupations include the following job descriptions:
Technical Mangers and Engineers (PCS 38)
- Engineers and R&D managers
- Information technology R&D engineers and managers
- Information technology support engineers and managers
- Telecommunications engineers and specialists
Technicians (PCS 47)
- Designers of electrical, electronic, and mechanical equipment
- R&D technicians
- Installation and maintenance of information technology equipment
- Telecommunications and computer network technicians
- Computer operation, installation and maintenance technicians
These job descriptions make it clear that these are the workers in a firm who facilitate new technology adoption.
Documenting job polarisation
Figure 1 illustrates how the French labour market has polarised. The figure plots the change in an occupation's share of aggregate hours from 1994 to 2007 against the occupation's rank in the wage distribution in 2002. The circles are proportional to the average size of occupations, and the curve is a weighted quadratic regression line. The pattern is clear: the two large, highly-paid occupations on the right – PCS 37 (managers) and PCS 38 (technical managers) – grew, as did three large low-wage occupations on the left – PCS 68 (Low-skilled manual labourers), PCS 56 (personal service workers), and PCS 55 (retail workers). The middle-wage occupations that shrank over the period include skilled industrial workers and manual labourers (PCS 62 and 63), unskilled industrial workers (PCS 67), and clerical and middle-management workers (PCS 54 and 46). Exceptions to this pattern in the middle of the wage distribution include drivers (PCS 64), an occupation that can be neither offshored nor automated, and technicians (PCS 47). To summarise, polarisation and the March of the Techies proceeded together from 1994 to 2007.
Figure 1. Change in occupations share of hours, 1994-2007
An occupation’s share of hours worked in the economy can increase because the average firm added more workers in that occupation (a ‘within-firm’ effect), or because firms with many of those workers grew faster than average (a ‘between-firm’ effect). Figure 2 illustrates this within-between composition for the 14 largest occupations in the economy. The overall change in each occupation’s share of hours is shown in blue, with the between and within components in red and green respectively.
Figure 2. Change in hours share, 1994-2007
Focusing first on the fortunes of high- and low-skill industrial workers (PCS 62 and PCS 67), both saw big overall declines. For the high-skill industrial workers in PCS 62, the overall decline was entirely due to within-firm shedding of these workers. The story is exactly the opposite for the low-skill industrial workers in PCS 67 – the collapse in the share of these jobs is due to the decline of firms that rely heavily on these workers. Putting these two facts together, the implication is that firms that are intensive in skilled industrial workers grew, but within these firms there was substitution away from skilled industrial workers. Firms that are intensive in low-skill industrial workers simply shrank in relative terms.
Next, consider the skilled and unskilled manual labour occupations, PCS 63 and PCS 68. These jobs are probably less subject to both automation and offshoring than the similarly skilled, but better paid, industrial jobs. Firms that were intensive in these occupations shrank, but within firms the importance of these jobs actually increased substantially.
Turning to clerical workers, PCS 54, the collapse in office jobs was more than accounted for by the between-firm component – firms that had a lot of office workers shrank substantially, even as the within-firm component was slightly positive. This within-between split is not consistent with a simple story of replacing clerical workers with computers; rather, it is suggestive of a heavy reliance on office workers being associated with slower firm employment growth.
The accompanying boom in lower-paid retail (PCS 55) and personal service (PCS 56) jobs was fairly evenly split across the within and between components. Thus, firms heavy in retail and/or personal service jobs expanded, and increased the share of these jobs within their firms as they did so.
The March of the Techies was broad based, with both technical managers (PCS 38) and technicians (PCS 47) growing rapidly. This growth was mainly accounted for by between-firm changes (techie-intensive firms grew faster), but firms additionally shifted hours towards techies.
Causes of polarisation
What caused the polarisation of the French job market? Was it globalisation, technological change, or other factors? In the econometric analysis reported in our paper, we use levels of techies and trade in 2002 to explain both within- and between-firm changes from 2002 to 2007. We use levels of techies and trade between 1994 and 1998 as instrumental variables for the levels in 2002.
We find that employment growth was much faster for firms with more techies in 2002. Manufacturing firms that imported from lower-income countries saw slower employment growth, which is consistent with the hypothesis that firm-level offshoring replaces jobs within the firm with imported inputs.
Among non-manufacturing firms, techies caused skill upgrading, with higher-wage managerial jobs expanding significantly. In contrast to non-manufacturing, techies in manufacturing caused polarisation, with high- and low-wage occupations growing at the expense of middle-wage occupations. Among manufacturing firms, importing caused skill upgrading within the blue-collar occupations, which suggests that offshoring replaces less-skilled production workers. Manufacturing firms that exported saw much faster growth in the share of managers.
Our research is the first to analyse firm-level data on employment by occupation. We find that the French job market polarised between 1994 and 2007, with middle-wage jobs shrinking as a share of employment as both high- and low-wage jobs grew. These changes occurred both within and between firms, with the latter effect being more important. Our econometric analysis finds an important role for trade and especially for techies in causing these dramatic changes.
Autor, D H, L F Katz and M S Kearney (2006) “The polarization of the US labor market”, American Economic Review, AEA Papers and Proceedings, 96(2): 189–194.
Goos, M and A Manning (2007) “Lousy and lovely jobs: The rising polarization of work in Britain”, The Review of Economics and Statistics, 89(1): 118–133.
Goos, M, A Manning and A Salomons (2009) “Job polarization in Europe”, American Economic Review, AEA Papers and Proceedings, 99(2): 58–63.
Goos, M, A Manning and A Salomons (2014) “Explaining job polarization: Routine-biased technological change and offshoring”, American Economic Review, 104(8): 2509–2526.
Harrigan, J, A Reshef and F Toubal (2016) “The march of the techies: Technology, trade, and job polarization in France, 1994-2007”, NBER Working Paper 22110.