Work is changing – as it always has

There is nothing new about anxieties created by job loss. History textbooks tell us about the Luddites in Great Britain, who protested against the mechanisation of the lace weaving industry and, particularly, the deterioration of the position of skilled workers in that process. The worries concerned, most of all, people’s own income, rather than the mechanisation itself. Others have considered that the growth of productivity thanks to machines is a desirable development.

In his work Economic Possibilities for Our Grandchildren (1930), the highly influential economist John Maynard Keynes estimated that rising work productivity would sooner or later enable significantly shorter working hours. The resources set free by increased work productivity could be used for producing altogether new goods and services. Indeed, we can largely thank technological advancement for the rise of living standards in the western world.

In the current debate, new digitalisation-related technologies – particularly those relying on artificial intelligence (AI) and robotics – are a factor crucially influencing change in work life. Robotics solutions are many times as efficient as human labour in mechanical routine tasks. AI-based solutions reduce costs, and ideally they liberate human labour for tasks where automation and robotics perform less well.

When discussing the impacts of technological advancement on the labour market, we often talk about skills bias. This refers to the tendency of technological advancement to favour tasks that require high skills. The proportion of tasks requiring high skills has grown considerably in recent decades.

For the most part, the tasks that technological advancement has enabled us to automate so far are routine tasks typical of average-wage jobs. This phenomenon is known as routine-biased technological change, and it has been seen as an explanation for why work in medium-paid occupations has diminished (cf. Oesch & Rodriguez Menes, 2010). Over time, as AI becomes more autonomous and better apt to learn, increasingly complicated tasks can be automated.

Apart from technological advance, other drivers in society also contribute to the transformation of work. One is the location of work. With urbanisation, a considerable share of new jobs are created in growth centres. In terms of occupational structure, too, new jobs are different from those that disappear.

The importance of the service sector as an employer has grown, and the proportion of service jobs is largest in big cities. Technological advancement in transport and communications has enabled a more intense economic integration between national economies. The global division of labour has deepened first through international trade in goods and services, and later with the unbundling, splitting up and offshoring of production processes (Baldwin, 2006).

From Western countries, substantial numbers of jobs in manufacturing, for example, have been offshored to countries with lower cost levels, typically in Asia. With advancement in communication technologies, production processes can more efficiently be split up and carried out where it is most profitable. This applies not only to manufacturing but increasingly also many other occupations, even such expert jobs that were earlier considered less prone to the globalisation of work.

Occupations at risk of disappearing – Frey and Osborne’s approach

Many assessments have been made of the impact of technology on jobs and occupations in the future. In 2013, Oxford University researchers Carl Benedikt Frey and Michael A. Osborne developed coefficients for the risk of automation by 2030 of different occupations. They applied these coefficients on the labour market in the USA and found that no less than 47 per cent of employment was at high risk of being replaced as a consequence of technological developments, particularly digitalisation.

Frey and Osborne analysed the occupations in terms of three dimensions. Social intelligence is manifested in human interaction as an ability to negotiate and persuade; also as care and attention. It is needed in many ways in the social and health care sector and in upbringing and education. Creativity is embodied by new inventions and valuable ideas as well as the ability to use concepts in a versatile way. Human perception and the ability to observe and identify objects in surprising situations is necessary, for example, in the handling and transport of goods in a changing environment. To date, people have been considered to be better than machines in these tasks.

The more these qualities are required in an occupation, the less prone it is to the effects of digitalisation in the near future, according to Frey and Osborne. And vice versa: the less social intelligence, creativity and perception are required, the more automatable a job is. For each occupation, a coefficient was calculated for its replacement risk. If the risk rate was over 70 per cent, the occupation was defined as running a high risk of being replaced by technology. Frey and Osborne found that jobs at high risk of disappearing due to automation were particularly common in the service sector, sales work, and administration and support tasks.

In 2014, researchers Mika Pajarinen and Petri Rouvinen at ETLA Economic Research converted the occupational titles in the calculation model to correspond to the classification used in Finland. As a reference level for the number of jobs they used occupational data in Statistics Finland’s register-based employment statistics of 2011. According to the calculation model, 36 per cent of jobs in Finland that year were at high risk of being replaced by 2030. A corresponding analysis was made for jobs in the city of Vantaa (Fröberg & Lönnqvist 2018).

In the following, we use the coefficients determined by the ETLA researchers to analyse how the advancement of digitalisation affects the number of jobs in the Helsinki Metropolitan Area.

Changes in job numbers by occupation in the Helsinki Metropolitan Area until 2030

According to Statistics Finland’s employment statistics, there were 634,700 jobs in the Helsinki Metropolitan Area at the end of 2017. Applying Frey and Osborne’s method, 23 per cent would be at high risk of being replaced by 2030. As stated above, high risk means that 70 per cent or more of jobs in an occupation could be replaced through automation. In theory, this equals around 140,000 jobs in the metropolitan area. This is a slightly lower proportion than in Finland as a whole, where 26 per cent of jobs would be in a high risk of being replaced, based on data from 2017.

The proportion of jobs running a high risk of being replaced varies somewhat between the big cities of the Helsinki Metropolitan Area: in Helsinki, it was 22 per cent, in Espoo 20 and in Vantaa 26 per cent. These differences are explained by the industrial structures of the cities. Although the total proportion of market services is rather similar in them all, differences between industries may be great. Espoo and Vantaa are clearly more trade-dominated than Helsinki.

Logistics-related industries stand out in Vantaa, while information and communication jobs are concentrated in Espoo and Helsinki. In the field of finance, Helsinki is the centre. The category professional, scientific and technical activities has clearly larger proportions of jobs in Helsinki and Espoo than in Vantaa.

In addition, automation already seems to have taken a part of those jobs that are easily replaceable. In Helsinki, for example, jobs at high replacement risk decreased by four percentage points between 2014 and 2017. Nevertheless, the total number of jobs grew by five per cent during that time.

Jobs at the highest risk of being replaced

An occupation-based analysis reveals that shop sales assistant jobs, in particular, would seem to diminish in near future. At present, this is the most common occupation in the Helsinki Metropolitan. In trade, the effect of digitalisation can be detected a growing proportion of shopping being done over the internet and of growing numbers of self-service tills in supermarkets.

At high risk of being replaced we find the categories of secretaries, accountants, accounting associate professionals, bank tellers and related clerks, as well as jobs in statistics, finance and insurance. IT software has replaced much of the computing work involved in these tasks. Part of the customer service work that requires reasoning can already be done by AI. Predictable inquiries in customer service are increasingly addressed by chatbots.

Many jobs in the sorting and delivery of post are disappearing. Much of this work has been automated, and the transformation in this field is often due to changing consumer behaviour: traditional greetings by mail have been replaced by text messages and social media.

In Frey and Osborne’s model, automation strikes hard on the restaurant business and institutional catering services. The role of self-service is growing, and some fast-food restaurants already have automated self-ordering. Many phases of the production process can also be automated. As growing numbers of consumers order their meals online in advance, possibly including home delivery, the need for waiters and sales staff in cafés and restaurants decreases.

Other large occupational categories threatened by replacement include contact centre salespersons, telecommunications engineering technicians, information and communications technology operations technicians, and cashiers and ticket clerks.

In Espoo, the occupations strongly threatened to be replaced are roughly the same as in Helsinki (Figure 2). One exception is chemical industry process workers, a group that is more concentrated in Espoo. Both Espoo and Helsinki have more high-skill specialist jobs that are not yet replaceable by digitalisation, compared to Vantaa.

Vantaa has the highest proportion of jobs in trade, storage and logistics, all of which are among the branches most touched by automation. Vantaa differs from its neighbours in the sense that office workers in transport, as also process workers in the food industry, are high on the list of the most threatened occupations. The need for security guards is likely to decrease as surveillance cameras increase in numbers.

Changes in job numbers by occupation in the Helsinki Metropolitan Area in 2010–2017

Frey and Osborne’s original analysis assessed the change in the numbers of jobs between 2010 and 2030. We now have the opportunity to make a “half-way report” about how the numbers of jobs in various occupations have changed.

The actual change seen in the numbers of jobs can be studied by analysing Statistics Finland’s employment statistics for 2010–2017. During this period, the number of jobs grew by a total of five per cent in the Helsinki Metropolitan Area. On the level of different occupations, however, the trends point in different directions. 

Our analysis does not allow us to easily determine which changes in the number of jobs are due to digitalisation and which are for other reasons. A major factor behind changes in job numbers is economic trends. Businesses are closed down primarily during economic downturns. At local level, the geographic location of businesses – and their relocations – may cause substantial changes in the number of jobs in cities. In the long term, the numbers of jobs are also influenced by consumer behaviour and cultural factors. The values that steer consumption tend to change relatively slowly and are harder to observe. Online shopping and the spread of ‘café culture’ can be seen as examples of such consumption-related change.

Figure 4 shows occupations that Frey and Osborne’s model identifies as being at high replacement risk due to digitalisation. In the category of administrative and support service activities, numbers of jobs have decreased strongly in many occupations. In the category general and keyboard clerks (secretaries, general office clerks, typists and word processing operators, data entry clerks) over 10,000 jobs disappeared in the Helsinki Metropolitan Area over 2010–2017, amounting to a 40 per cent decrease. One of the strongest relative decreases in jobs has been seen in the group statistical, finance and insurance clerks and for cashiers and ticket clerks.

2,400 bank clerk jobs disappeared in a six-year period, and almost all major banks have announced layoffs and staff reduction needs since 2017. At the same time, however, banks have announced large numbers of new vacancies due to digitalisation-related changes in job descriptions. Thus, digitalisation also creates new jobs and new tasks.

The number of jobs for postmen and mail sorters decreased by 1,500 (this figure does not yet include the much-publicised need for staff cuts in the Finnish postal service in 2019). Jobs for IT technicians and support staff decreased by 2,000 in the Helsinki Metropolitan Area, and this reduction is probably chiefly explained by the increased usability of IT software. 

As regards the administrational jobs described above, the Frey and Osborne model seems to work – the number of jobs in these occupations has decreased substantially. Another large occupational group diminishing due to digitalisation are service sector employees. For the latter, the actual change in job numbers in the Helsinki Metropolitan Area does not seem to match the model (see figure 4).  

According to Frey and Osborne’s model, a large proportion of shop sales assistant jobs would be disappearing. In fact, the number of shop (retail) sales assistants in the Helsinki Metropolitan Area remained the same in 2017 as in 2010. The area has many large shopping centres needing retail assistants. In addition, the Helsinki Metropolitan Area has seen strong population growth, which increases private demand. Yet the success of online shopping has posed difficulties for traditional trade. Nonetheless, the stagnancy in the numbers of sales assistant jobs is partly explained by the fact that staff demand is regulated through flexible work contracts. Sales staff increasingly work part-time. In addition, shops hire many s from staff agencies, and use people paid by the hour during rush hours and high season.

It is also possible that the terminology used in the model will influence our interpretations concerning certain occupations. For example, a cashier in English is typically someone working at a till, while the corresponding Finnish occupation may refer to a wider job description – including salespeople in specialist shops who also interact with customers and give advice on products. The classification used in Finland includes the small group cashiers and ticket clerks, and these jobs really have diminished rapidly: between 2010 and 2017 by 38 per cent.

Jobs in restaurants, institutional kitchens and cafés have increased in the Helsinki Metropolitan Area. The number of kitchen helpers grew by 1,000, and restaurant or institutional kitchen staff by over 1,400. Private consumption has increased demand as both residents and tourists spend more and more time in cafés. At the same time, these businesses have large numbers of vacancies and these are difficult to fill. This is probably due, in part, to atypical work contracts and low wage levels, and in future the recruitment troubles may increase businesses’ interest in automation.  

Criticism and alternative approaches

Frey and Osborne’s (2013) method has also been criticised. Although technological advancement reshapes tasks to a high degree, some of the tasks undergoing change will remain in future, albeit with a different content (Arntz, Gregory & Zierhan, 2016). The assumption about a rapid technological change contributing to the disappearance of jobs has also been put into question. It is possible that some opportunities opened up by new technologies are never actually taken, if the consequences of these changes are not considered acceptable by society (Arntz, Gregory & Zierhan, 2016). Whether or not jobs are destroyed apparently also depends in part on other solutions such as arrangements related to wage-setting (Dauth, Findelsen och Woessner 2017).

It is by no means self-evident that the coefficients for the disappearance of jobs presented in Frey and Osborne’s original study – based on US data – are applicable on a global scale. Even within a single occupational title, there is often significant variation in job descriptions and automatability between jobs in one country, on the one hand, and between countries on the other.

The study of Arntz et al (2016) used the so-called PIAAC (Programme for the International Assessment of Adult Competencies) data which describes the kind of tasks included in different occupations. According to the findings of Arntz et al, the proportion of tasks with a high probability of automatisation varies between six and 12 per cent in industrialised countries. The study gives Germany and Austria the highest rates for the disappearance of occupations. In both countries, over 12 per cent of jobs could be automated. For Finland, this figure is seven per cent. Significant changes, however, may be expected in many jobs in terms of their contents and qualifications requirements.

The analysis by Arntz et al (2016) is extended by Nedelkoska and Quintini (2018) who examine a total of 32 countries. They also include a larger number of occupations. According to their findings, 14 per cent of jobs in OECD countries can be automated, and another 32 per cent will face significant pressure to change as tasks become automated. Differences between countries are nevertheless great. According to Nedelkoska and Quintini, the impacts of automation are smallest in the Anglo-Saxon countries, Scandinavia and the Netherlands, and strongest in eastern Europe, Germany and Japan. In Finland, along with Norway and Sweden, the risk for jobs to be lost is smallest among all the countries in the study.

As we have already seen, job trends in many Western countries have been influenced not only by technological advance, but by economic globalisation – itself made possible by technological advancements. To give an example, Blinder (2009) has estimated that between 22 and 29 per cent of jobs in the US may be regarded as likely to be offshored. Tuhkuri (2016) applies the same method to a data material from Finland, and concludes that roughly one-quarter of jobs may be on the line due to globalisation in the next ten years. It is possible that globalisation has played an even bigger part than technology for developments in job markets. Using a research material collected in the USA, Acemoglu and Restrepo (2017) make the assessment that economic globalisation and offshoring of manufacturing to Asia – particularly to China – explains a considerably larger part of the decrease in manufacturing jobs in the US over 1990–2007 than robotics.

Conclusions

We can conclude, regarding the changes in job numbers up to this point, that the need for labour seems to already have diminished due to technological advancement – essentially digitalisation – in precisely those occupations where the replacement risk will also be greatest in future. At the same time, the high demand for services related to urbanisation and urban life, such as the coffeehouse culture, will be creating new jobs at least for a while. Of course, these jobs too have seen significant change of late. As an example, many fast food restaurants offer automated self-offering. Digitalisation spreads in phases and affects different occupations at different pace. 

Although new technologies and digitalisation cause significant change on the job market, the assumption sometimes made that work will disappear is unrealistic. The amounts of work and jobs in a society are no constants. Yet it is obvious that the new technologies affect, in many ways, the contents of tasks and the vacancies available. With new technologies, the productivity of work rises. Technological advancement creates jobs, too, both directly and indirectly (OECD, 2016). In the best cases, technology leads to the automation of routine tasks while more demanding tasks will remain for humans to handle. The question remains how many of us – and in what ways – will be able to master the transition to new tasks with many new skills requirements? Will there be sufficient opportunities for providing and acquiring re-education? As skills requirements change, the position of many occupational groups may change considerably on the job market. The coronavirus pandemic has, on its part, also rapidly contributed to change in our society, as telework and online shopping have quickly expanded everywhere. How permanent these changes will be – and what long-term effects the pandemic will have – remains to be seen.

Henrik Lönnqvist is Strategy and Research Director at the City of Vantaa. Minna Salorinne is Senior Statistician at Helsinki City Executive Office.

Literature:

Acemoglu, D. & Restrepo, P. (2017): Robots and Jobs: Evidence from US Labor Markets. NBER Working Paper 23285.

Arntz, M., Gregory T. & Zierahn, U. (2016). The Risk of Automation for Jobs in OECD Countries: A Comparative Analysis. OECD Social, Employment and Migration Working Papers, No. 189.

Baldwin, R. (2006). Globalization: the great unbundling(s). Prime Minister’s Office/Economic Council of Finland.

Blinder, A.S. (2009). How Many US Jobs Might Be Offshorable? World Economics, 10(2), 41–78.

Dauth, W., Findelsen, S. & Woessner, N. (2017). The rise of robots in the German labour market. https://voxeu.org/article/rise-robots-german-labour-market

Frey, C.B. & Osborne, M.A. (2013). The Future of Employment: How Susceptible Are Jobs to Computerisation? Oxford University.

Fröberg, W. & Lönnqvist, H. (2018): Teknologian kehitys, työn muutos ja uudet koulutustarpeet. City of Vantaa Information Services C4:2018. 

Keynes, J.M. (1930). Economic Possibilities of our Grandchildren. In: Keynes, J.M. (2010). Essays in Persuation. Palgrave Macmillan.

Nedelkoska, L. & Quintini, G. (2018). Automation, skills use and training. OECD Social, Employment and Migration Working Papers (202). Paris: OECD Publishing.

OECD (2016). Automation and Independent Work in the Digital Economy. Policy Brief on the Future of Work.

Oesch, D. & Rodriguez Menes, J. (2010). Upgrading or polarization? Occupational change in Britain, Germany, Spain and Switzerland, 1990–2008. Socio-Economic Review, 9 (3), 1-29.

Pajarinen, M. & Rouvinen, P. (2014). Computerization Threatens One Third of Finnish Employment. ETLA Memorandum (22).

Tuhkuri, J. (2016). Globalization Threatens One Quarter of Finnish Employment. ETLA Memorandum (22).