VoxEU Column Labour Markets

The who and how of disappearing routine jobs

As routine tasks are increasingly automated, middle-wage jobs are becoming rarer. This column documents the changes in labour-market dynamics behind this polarisation, and investigates which workers are affected by it. Flows into middle-wage routine jobs are declining (rather than flows out increasing). Interestingly, routine cognitive workers – who tend to be educated women – are benefiting from this hollowing-out by moving up the occupational ladder.

Labour markets around the world have experienced a profound polarisation in recent decades. The share of employment in middle-wage jobs has declined, while employment in high- and low-wage jobs has increased. In the US, this ‘hollowing out of the middle’ has been linked to declining per-capita employment in occupations with a high content of routine tasks – activities that can be performed by following a well-defined set of procedures and are therefore relatively easy to automate (Autor et al. 2006, Goos et al. 2014). To put this in perspective, as recently as the late-1980s more than one in three American adults was employed in a routine occupation; currently, that figure stands at about one in four.

To design appropriate policy responses, it is important to understand who the disappearance of routine employment is affecting the most, and how the process is playing out. For example, if the decline in routine employment is mainly accounted for by changes in the occupational choices of young workers entering the labour market, the appropriate response would be different than if the decline is due to increasing exit rates from the labour force of prime-aged workers.

Our current research investigates the labour-market changes that underlie the disappearance in middle-wage, routine jobs (Cortes et al. 2014).

  • We find that there has been a marked decline in the rate at which workers transition into routine employment. This change is particularly pronounced among the young.
  • Women and those with higher education levels have found it easier to adjust to these changes.

They have become more likely to move into higher-paying, non-routine cognitive (‘brain’) jobs.

  • This is not the case for men and those with less education, who tended to work in blue-collar occupations.

The loss of routine jobs among these individuals has been offset by transitions into other jobs that also have a high content of routine tasks (such as clerical and administrative work), or transitions into lower-paying, non-routine manual (‘brawn’) jobs.

Declining flows into routine employment

To better understand the polarisation phenomenon, we analyse the changes over time in the flows of workers into and out of routine occupations. The key factor accounting for the decline in routine employment is the fact that fewer workers are transitioning into routine jobs. Since the 1980s, unemployed workers – particularly those working in a routine occupation before becoming unemployed – have become less likely to find employment in a routine job. Similarly, the probability that an individual who is out of the labour force transitions into a routine job has also fallen.

On the other hand, changes in the rate at which workers quit or are dismissed from routine jobs account for very little of the decline in routine employment. There have been slight increases in the exit rates from routine employment to labour force non-participation. However, these changes are much less important than the fall in the inflow rates in understanding the overall decline, particularly in recent years.

  • Our results also show that these changes in transition patterns cannot be attributed to changes in the demographic composition of the US economy, such as population ageing or the increase in the number of people going to college.

Instead, they reflect behavioral changes for individuals with given demographic characteristics.

We also find that young workers – those aged between 16 and 34 – have experienced the largest decline in the likelihood of transitioning into routine jobs. Across all demographic groups, this change is the most important in accounting for the aggregate decline of routine, middle-wage employment. Figure 1 illustrates this. The blue solid line displays the evolution of per-capita employment in routine occupations from 1976 to the end of 2012. The green hatched line displays how this series would have evolved if the key transition rates that we identify had remained at their pre-polarisation levels for all demographic groups. More than half of the fall in routine employment is mitigated under this scenario. Finally, the red hatched line displays the evolution of routine employment when only the transition rates of the young are held constant to pre-polarisation levels. Much of the total mitigating effect is driven by the young.

Figure 1

Where do they go?

Given our finding that an important driver of polarisation is the fall in the rate at which unemployed routine workers return to routine jobs, we ask the natural question: where are these unemployed workers going instead? Are they switching into high- or low-paying non-routine occupations? Do they remain unemployed or leave the labour force?

To answer this, we distinguish between two subsets of routine occupations that differ markedly in terms of their demographic composition. This first subset, routine manual jobs, includes occupations such as machine operators and other blue-collar jobs that tend to employ men with relatively low education levels. The second subset, routine cognitive jobs, includes occupations such as clerical and administrative support jobs that are generally female-dominated with higher levels of education.

Unemployed workers from both of these subsets have experienced falls in the rates at which they return to their previous occupation group. However, the way in which workers in these categories have responded is markedly different.

The top panel of Figure 2 displays the change in the job finding rate during each expansionary or ‘boom’ period since the 1980s for unemployed workers who used to hold a routine manual job. The first four clusters of bars illustrate the change in the rate at which they find employment in routine manual, routine cognitive, non-routine cognitive, and non-routine manual occupations, respectively. The rightmost cluster illustrates the change in the rate at which they find employment in any job.

Relative to the pre-polarisation period, the leftmost cluster shows the fall in the rate at which unemployed routine workers return to a routine manual job. As we will discuss further, this fall has been particularly strong since the Great Recession. Prior to this latest episode, the rate at which they switched into other occupational groups increased. However, this occupational switching was not directed towards high-paying non-routine cognitive jobs, but rather towards middle-wage, routine cognitive jobs (which are also in decline) and low-paying non-routine manual jobs (such as food-preparation workers and home-care aides).

Figure 2a. Change in transition rates from URM: 12−month horizon (Relative to 1976−79 expansion)

Figure 2b. Change in transition rates from URC: 12−month horizon (Relative to 1983−90 expansion)

On the other hand, the bottom panel of Figure 2 displays analogous changes for unemployed routine cognitive workers. These workers have also become less likely to return to their old occupation. However, here we see an increase over time – even after the Great Recession – in the rate at which these workers transition into high-paying non-routine cognitive jobs (such as managers and financial analysts). Thus, workers in these occupations (who tend to be women and have higher levels of education) have been better able to cope with the disappearance of routine employment – by moving up the occupational ladder.

Changes since the Great Recession

The period since the Great Recession has been characterised by a precipitous fall in the rate at which unemployed routine workers return to routine employment (what we refer to as the ‘return job finding rate’). Interestingly, as Table 1 shows, the analogous changes for the unemployed from non-routine occupations since the Great Recession were much less pronounced. Whereas the return job finding rate was nearly halved for unemployed routine workers between the pre-polarisation era and the post-Great Recession era, the fall for non-routine workers was relatively mild, at around three percentage points.

Table 1

Pre-polarisation expansion 23.12% 15.80% 13.98% 14.80%
2009-2012 expansion 13.87% 7.91% 11.30% 11.68%
Difference -9.25% 7.89% -2.68% -3.12%


To fully understand the phenomenon of employment polarisation and design appropriate policy responses, it is crucial to understand the changing labour market transition patterns of different demographic groups in the economy. Our findings indicate that an important factor behind the decline in employment in middle-wage, routine jobs is the substantial fall in the transition rates of young workers into such jobs. In addition, while routine cognitive workers (who tend to be women with more education) have had some success in moving up the occupational ladder, routine manual workers have not.

Editors’ note: The analysis and conclusions set forth in this article are those of the authors and do not indicate concurrence by other members of the research staff or the Board of Governors of the Federal Reserve System


Autor, D H, L F Katz, and M S Kearney (2006), “The Polarization of the US Labour Market.” American Economic Review, 96, 189-194.

Cortes, G M, N Jaimovich, C J Nekarda, and H E Siu (2014), “The Micro and Macro of Disappearing Routine Jobs: A Flows Approach.” NBER Working Paper No. 20307.

Goos, M, A Manning, and A Salomons (2014), “Explaining Job Polarization: Routine-Biased Technological Change and Offshoring.” American Economic Review, 104, 2509-2526.

7,349 Reads