«[...] depressions are not simply evils, which we might attempt to suppress, but
[...] forms of something which has to be done, namely, adjustment to [...] change.»
It is obvious to state that, when an economy enters into recession, jobs are lost: some companies go bankrupt and others, suffering from a drop in sales, are forced to reduce their number of workers. In this article we will analyse whether there is a positive (or less negative) side to recessions. To be precise, we will investigate whether it is true, as hinted by Schumpeter during the Great Depression (1929-1939), that firms take advantage of periods of recession to adopt new technologies.
We will focus our attention on the last few recessions in the USA, which is where the best studies have been carried out, assuming that these conditions can be extrapolated to countries with similar political and economic characteristics. The first thing we observe is that the employment trend has been very different in recessions after 1990 (1991, 2001 and 2009) if we compare this with the three previous recessions (1970, 1975 and 1982). Jaimovich and Siu show that, while employment took no longer than 6 months to recover after the first three recessions, after the last three it needed at least 17 months.1 How can we explain this difference? Why, in these later recessions, has it taken longer to create employment?
One possible answer is that companies have taken advantage of the last three recessions to adjust their technology, thereby reducing the number of employees and increasing the productivity of each worker. This could explain why companies, having changed their production or type of good, now need fewer workers and why these, as they have had to move to other firms (or even change industry), take longer to find another job. Before confirming this hypothesis, we must verify two premises: (i) that this phenomenon has taken place particularly in the last three recessions, and (ii) that those workers more likely to be replaced by new technology are precisely those that have taken longer to find another job. To do so, we need to find a technology that was not present before the 1990s and check whether it is true that those workers carrying out tasks that are more easily replaced by this technology are the ones that lost their jobs.
The best candidate is the revolution in Information and Communication Technologies (ICTs), which started at the end of the 1980s. Think of the times you have taken money out of a cashpoint, bought a plane ticket or booked a hotel room online or marked a series of keys on your mobile phone to report an incident to your customer service. All these automated processes, which are now so familiar to us, have substantially transformed the work of employees at banks, travel agents and telephone operators, just to give a few everyday examples. Less visible but nevertheless just as notable is the restructuring, both internal and external, resulting from the adoption of these technologies in firms. Before the internet, companies would have to replicate information centres (such as data on sales, orders, stocks, etc.) in different departments because it was difficult to access the information required to take decisions. However, thanks to the internet, now they centralise the information in such a way that all departments can share it easily. One example of restructuring is the possibility of moving one or more departments to another country. In short, adopting ICTs has led to the loss of many jobs because they are carried out either in another country or by computers.
Having identified a suitable technology to verify our hypothesis, we have to determine the kind of worker who is more likely to lose their job. To do so, it is very useful to break down the content of each occupation into three types of tasks: routine, manual and abstract. The best way to understand this composition is by simplifying it through examples. Let us start with the job of a physician, whose work is highly abstract and not very routine or manual, since physicians need to study each case presented to them (one single rule cannot be applied in all cases) and, in general, they do not need to manipulate objects. The job of a taxi driver is not very routine, abstract or manual because they need to be at the wheel, monitoring the other cars, taking notice of traffic signs and changing the route if necessary. Another example is that of a supermarket checkout employee, a highly routine, manual and not very abstract job which is basically limited to passing the products in front of the barcode scanner, stating the total price and returning the change. Analysing this breakdown, it appears that the type of task that can best be carried out by ICTs is the routine one because it follows a highly standardised, unambiguous procedure, a conclusion confirmed by the studies of different economists for the USA. For example, Autor and Dorn document that the metropolitan areas most specialised in routine jobs in the 1950s are those that have most adopted ICTs and those that moved their workers on routine tasks to non-routine, low-skilled services.2 Note that, out of the three jobs we have examined, the one with the largest proportion of routine work and therefore the most likely to be carried out by a computer is that of a checkout employee, a characteristic shared with the rest of the examples from our everyday lives that are now carried out by computers (cashpoints, online travel agencies and telephone operators). In fact, in Spain there are now mass consumption outlets where customers can pay for their shopping without help from anyone else. Note also that, in all cases, we are referring to work that can be carried out today by computers, not to tasks that computers may carry out in the near or distant future, such as driver-less cars, robots carrying out medical diagnoses, computers writing economic reports...
Having identified the technological change (ICT revolution) and the type of task (routine) that is most exposed to new technologies, the next step is to check whether, in the last few recessions, companies have eliminated the most routine jobs and that these have not been recovered. We refer once again to Jaimovich and Siu, who have looked in detail at the employment trend over the last six recessions in the USA. The authors show that, in the recessions prior to 1990, routine jobs started to be recovered seven months after the recession ended. However, in the last three recessions, the routine jobs lost were not recovered but are still lost. In other words, while the economy picks up and employment in abstract and manual jobs increases, routine jobs continue to disappear. From this analysis we can deduce that there is a tendency to replace employees with ICTs to carry out routine tasks but it is during the last few recessions that this trend has intensified and the largest adjustment has been made.
To conclude, we would like to stress that we only have indirect evidence for the Schumpeterian theory that companies took advantage of the last few recessions to adopt ICTs and, consequently, this is only one of the possible explanations of the facts documented by Jaimovich and Siu. In any case, there is evidence that the adoption of ICTs is associated with the loss of routine jobs and that this loss has intensified in the last three recessions. As we explained in «Globalisation and the redistribution of wealth» in the Dossier of February's Monthly Report, the implications for society of this analysis revolve around the fact that, in order to minimise the negative effects of recessions and to ensure society is better prepared to take advantage of new technologies, it is necessary to invest in education. In particular, in an educational system that focuses on reinforcing more abstract abilities, the field in which we humans can still beat the machines.
European Unit, Research Department,
1. Jaimovich, N. and H. Siu (2013): «The Trend is the Cycle: Job Polarization and Jobless Recoveries», Duke Mimeograph.
2. Autor, D., Dorn, D. (2013): «The Growth of Low-Skill Service Jobs and the Polarization of the U.S. Labor Market» American Economic Review, 103(5), pp. 1553-1597.