Future uncertainties

There’s no shortage of predictions – ranging from the modest to the cataclysmic – about what automation means for the future of work. The Office for National Statistics (ONS) this week added its authoritative voice to the chorus, claiming that 1.5m jobs in England are at ‘high risk’ of automation. This means that, for these jobs, at least 70 per cent of the tasks involved could be performed by new technology.

This projection sits on the modest end of the spectrum, representing 7.4 per cent of the country’s jobs. The ONS have built on several previous attempts to quantify the impact of automating technologies – some of which have suggested as many as 35 per cent of jobs could be vulnerable. Such efforts have gradually established that determining which tasks are automatable, identifying which occupations involve those tasks, and then mapping those occupations onto real-life labour markets is the best way to arrive at an estimate. Where the ONS’s methodology sheds new light is in allowing us to break down the numbers across a range of demographic groups; it allows us to estimate not just ‘how many?’, but ‘who?’.

The potential for automation to deepen inequalities is the key concern behind the Commission on Workers and Technology, which was set up by the Fabian Society and Community trade union and is chaired by Yvette Cooper MP. The ONS data validates that concern, revealing that women, part-time workers, young people, those without higher education qualifications and workers in particular local areas and industries are disproportionately at risk.

Here are six key insights from our look at the numbers:

1. Women are more than twice as likely as men to have a job at high risk of automation. Over 10 per cent of women’s jobs are considered to be at high risk of automation, compared with 4 per cent for men. Women’s jobs make up 70 per cent of the total number at high risk of automation.
2. Part-time workers are much more likely to be at high risk of job automation. Overall, 21 per cent of part-time jobs are at high risk of automation, compared with 3 per cent of full-time jobs. There are interactions between working patterns and gender too. Women do 4 out of every 5 part-time jobs. Driving the increased vulnerability of women to automation could be the higher vulnerability of job types which tend to be done on a part-time basis – such as administrative and secretarial, and sales and customer services occupations.
3. Both younger and older workers are more likely to be at high risk of their jobs being automated than those in their 30s – young workers are especially vulnerable. Workers aged between 16 and 19 are a staggering 41 times more likely to be in a job at high risk of automation than workers aged 35 to 39, and over a quarter of jobs done by 16-24 year-olds are at high risk. This is perhaps less surprising when we consider the kinds of occupations that young workers do; ONS data on earnings tells us that almost a third of those aged 18-21 work in ‘elementary occupations’, which consist of mostly routine tasks – and these are the kinds of tasks that are most automatable. By contrast, only 1 in 10 workers aged 30-39 are in elementary occupations. It is true that, for most workers, the first few years of participation in the world of work are transitional – but policymakers should certainly contemplate the implications for the ability of young people to enter the labour market if typical starter jobs are at high risk of automation. Workers aged 60 to 65 are also more likely to be at high risk than those aged 35 to 39, and policy options like flexible retirement should be considered in order to address this generational effect.
4. You are very unlikely to be at high risk of job automation if you have a higher education qualification. Almost all jobs at high risk of automation (98.8 per cent) are done by people educated up to A level, GCSE or below GCSE level. Only 1.2 per cent of jobs at high risk belong to people who have gone through higher education. This is not to say those with higher education are totally secure; 29 per cent of jobs facing medium risk of automation – i.e. those with a probability of automation between 30 and 70 per cent– are done by people with degrees and other higher education qualifications.
5. The data reveals stark inequalities between local areas. The proportion of jobs at high risk of automation ranges from 2.4 per cent in Wokingham to 29.5 per cent in Boston (although it should be said that this data is missing for local areas where survey sample sizes get too small to make sound estimates). Another measure used is the ‘average probability of automation’ (the ONS’s methodology calculates a probability of automation based on the task composition of each individual’s job). Looking at these probabilities for each area, workers in Boston face a 57 per cent average likelihood of automation, while the risk is 33 per cent for a worker in Wandsworth. Overall, there are 26 local authority areas where a worker is more likely than not to see their job automated. Only one of these areas is within a city – all the others consist of towns, villages or smaller settlements (as defined by the House of Commons Library). None of these areas are in the regions of London or the South East.
6. There’s massive variation in the likelihood of job automation by industry. If you work in food and beverage service activities, you face on average a 63 per cent chance of your job being automated. If you work in scientific research and development, this probability is only 28 per cent. Industries where employees face a high average probability of their jobs being automated also include accommodation (60 per cent), retail (58 per cent), crop/animal production activities (57 per cent) and postal and courier activities (56 per cent). Industries with the lowest average risk of job automation include computer programming (31 per cent), programming and broadcasting (32 per cent), advertising and market research (34 per cent) and publishing activities (34 per cent).

The ONS, in their own write-up of the data, rightly stresses that we should view all of this in its correct context: technology change means many new kinds of jobs will be created, and the vast majority of existing jobs are more likely to change than disappear. There are also limitations inherent to such projection exercises. One is that the data only takes into account expert opinion on technological capability – meaning, simply, what technology could conceivably do. It does not consider possible real-world barriers to the actual development or adoption of that technology. The analysis also specifies no timescale. Even if all the technology developments do bear out, we don’t know when they will. We therefore can’t say whether the impacts summarised above will play out over the next 10 years or 30 years.

But despite these uncertainties, this analysis is important because of the huge inequalities in impacts it exposes. Whatever the pace of automation in the coming years, it is likely to affect some groups of people far more than others. It reminds us that we must pay very close attention to who is doing the jobs that are highly susceptible to automation – even if there is inherent uncertainty in identifying that ‘susceptibility’. This guide to who is most vulnerable will help the Commission on Workers and Technology to fulfil its central aim: to determine how to make automation an opportunity not a threat for all workers.