Forecasts about future jobs.

Articles and videos about ‘new’ technologies and the future of work can be alarming at first impression – like “80 percent of operations planning jobs will be automated in 10 years”! Over the past 40 years, there have been forecasts, which have not eventuated, of change’ in industries and the workplace. Memorable instances were ‘computer integrated manufacturing (CIM)’, ‘RFID’ for inventory management and the ‘paperless office’. So, should we believe the current forecasts concerning job losses caused by ‘new’ technologies?

The prediction about future employment that has gained most publicity was written by Carl Frey and Michael Osborne from the Oxford Martin School for a Sustainable Future at Oxford University. Their prediction was that, due to automation, 47 per cent of American jobs would disappear in 20 years; that is by 2033.  This prediction was picked up by mainstream and social media and has now become a ‘truth’, often quoted by commentators. The ‘model’ was also used by academics in other developed countries to arrive at a similar conclusion. But, should we accept the forecast as accurate and authoritative?

Ross Gittins, the Australian economics journalist, has investigated the prediction and found the methodology used to have a questionable basis. The process undertaken by the researchers was to provide colleagues with descriptions of 70 US occupations and asked them to judge whether they were ‘automatable’ (capable of being automated) or not – which is a subjective guess. This sample was ‘analysed’ and the result used to classify all 702 recognised US occupations. An occupation with a predicted probability of automation of more than 70 per cent was classed as being at ‘high risk’ of automation.

This process is not a rigorous form of research and to support this view, Gittins quoted a critique of the methodology by Borland and Coelli at Melbourne University. Their main arguments were:

• Of the sample 70 occupations provided Frey and Osborne, 37 were identified as at risk from automation. Should these subjective assessments prove wrong, the whole exercise is wrong, as the sample was extrapolated to all 702 recognised occupations
• The modelling makes the assumption that if an occupation is automated then all jobs in that occupation are lost. For example, the mass adoption of ‘driverless’ vehicles is assumed to eliminate the jobs of all who drive for a living
• The modelling assumes that if it’s technically feasible to automate a job it will be done. Companies will not decide whether it is appropriate or profitable to implement the technology, whether they have access to the expenditure required or whether the money should be spent in other areas
• The model assumes there will be availability of skilled workers needed to develop, build and implement a technology
• The modelling does not account for the jobs created, directly and indirectly, by the introduction of automation technologies. It ignores an increase of jobs selling, installing or servicing the additional robots and other equipment
• The model does not consider that competition between the newly robotised businesses will oblige them to lower prices. The researchers assume that customers and consumers will continue to pay current prices for goods and services; therefore not have more to spend and so create new jobs in other parts of the economy

In 2016, the Organisation for Economic Co-operation and Development (OECD) commissioned a second opinion of the Frey and Osborne modelling. The review identified that occupations categorised with a high risk of automation often contain a substantial share of tasks that are hard to automate. Therefore, rather than assume that whole occupations are automated, it is more likely that particular tasks will be automated; so, employment in particular occupations could fall, but not be eliminated.

This change in the underlying model meant that, on average across 21 OECD countries, the proportion of jobs that are ‘automatable’ is not 47 percent, but 9 percent. This observation is supported by a December 2017 research report by the consulting firm McKinsey. It estimates that up to 30 percent of the hours worked globally may be automated by 2030, but this is not the same as saying 30 percent of jobs. The challenge is less about jobs disappearing than about preparing mid-career people for the automation of particular tasks that are part of their jobs.

To measure the historical relationship between technology and jobs, a study by economists at the UK office of the consultancy Deloitte used census data for England and Wales going back to 1871. Their analysis found that, over economic cycles, rather than destroying jobs, technology has been a “great job-creating machine” because technology reduced prices, which increased spending power, therefore creating new demand and new jobs.

Verify forecasts about technologies and jobs

So, we have a report stating that 47 percent of jobs in America (and other developed countries) will be automated by 2033. This has been countered by a report raising legitimate questions about the quality of the research; three reports indicating that it is tasks rather than jobs that will be automated and that implementing technologies can increase job opportunities. Yet it is the initial report that still commands the headlines – why?

Writing in the online website ‘The Conversation’, Borland notes that the mistaken view concerning technologies and jobs comes from:

• a human bias to believe that ‘we live in special times’ – it is substantially different to past times
• the greater intensity of feeling about events that we experience (or hear and read about) and
• an absence of knowledge about history of technologies and the effect on jobs and social order. The introduction of technologies that threaten a way of life are not new. The term ‘Luddite’ (to reject new technologies) comes from English hand-loom weavers who, in the early 19th Century, violently objected to being replaced by wide-framed mechanised looms that could be operated by workers who were relatively unskilled and low paid

If a prediction is negative and has big numbers, it is often repeated. We (and much of the media) accept these claims without challenge, because the prediction is based on ‘modelling’ with output from a computer, therefore it must be true. As a part of managing change in your lifetime of work, develop a scepticism about forecasts and ‘models’ that purport to tell us about the future of technologies and work, which are influenced by:

• Companies that have developed ‘new’ technologies must sell their products
  • A sales technique is to convince journalists and commentators to insert positive articles about the company and the technology in all parts of the media
• Commentators often assume that opinion is equal to fact
  • Typically, a measure of confidence about the forecast is not provided

An approach to predictions about the future of technologies and jobs should be to ‘read, hear and see, then verify’. Identify the facts that authors and commentators use, then check their source. Typically you will find there is a large amount of ‘spin’ and ‘puff’, but few hard facts.

A note to readers

It is summer in Australia, so I will at the beach over the holiday period. My next blog will be on Wednesday January 10. Thank you for reading my blogs and best wishes for 2018.