Robots Don’t Get Sick: Accelerated Automation in the Post-Covid World

It is commonly said that Covid-19 will change everything. That we will learn the lessons of our vulnerability to extreme events, and accept more public investment to build resilience. That the shared experience of lockdown and admiration for health and care home staff will usher in an era of greater social solidarity. And that governments’ crucial role in response to both the health and economic emergencies will engender more support for action to address equally pressing long-term challenges such as climate change. 

Much of which should be true and might be. But we cannot be certain. Similar things were said amidst the global financial crisis of 2008, but in most countries the political winners of the subsequent decade were conservative nationalists, not social democrats or liberals. At this stage, we simply don’t know what social and political response this crisis will produce. 

But what is almost certain is that it will accelerate technological and economic trends already in place, with profound consequences to which societies will need to respond. 

Technological progress makes it possible to automate an ever-increasing proportion of all current work activity. As books such as Eric Brynjolfsohn and Andre McAfee’s The Second Machine Age or David Susskind’s A World without Work have described, the rapid progress of information technology means that all human work which involves repetitive physical tasks, and an increasing percentage of repetitive cognitive tasks, can be done as well or better by machines. The McKinsey Global Institute estimates that by 2060, technology will enable us to automate over 90% percent of all work currently performed in developed economies. 

Of course, as McKinsey also point out, the actual pace of automation will be determined as much by economics as by technology, and in particular by the cost of labour. If wages are low enough, many tasks which could in principle be automated will still be done by low-paid workers. 

Our current crisis, however, is increasing the complexity, risks and costs of employing labour: inevitably therefore it will accelerate progress towards a deeply automated world. 

The media is full of dire accounts and predictions of economic collapse: while the IMF has forecast only a 3% fall in global GDP this year, many experts foresee 10 or 15% declines in the US and much of Europe, at least for a six-month period. But suppose the fall were indeed 15%, and concentrate not on the economic glass 15% empty but 85% full. For in some ways it is striking that in a world where most people have been forced to stay at home, most meetings of more than a few people banned, and many businesses forcibly closed down, as much as 85% of our current economic activity can continue. 

Factories are slowly returning to life: online retailing is delivering a widening array of goods to people’s doors; entertainment is streamed direct to our houses; water, electricity and gas networks still operate: and the construction industry is gradually resuming work. Pubs, restaurants, hotels, theatres and museums will probably stay closed for many months further. But it turns out that the majority of the economy can operate without close human contact. 

Throughout the economic sectors which have either stayed open throughout the lockdowns, or which will soon start to reopen, the experience of the crisis will drive accelerated automation. 

Manufacturing is already a highly automated sector and becoming more so year by year. By 2050 we will need no more than 5% of the global workforce to produce all the physical goods consumers want. But progress towards that inevitable point will now accelerate because relative costs are changing. Keeping factory workers two metres apart adds complexity and cost: robots are as cheap as before and don’t catch viruses. Some Chinese companies already talk of achieving in the next 18 months steps towards automation previously planned for implementation over the next five years. 

Lockdown has accelerated the shift from bricks and mortar to online retailing. Even before the crisis, the British Retail Consortium forecast that British retailing jobs would fall from 2.9 million to 2.1 million in five years: but with many traditional retailers facing bankruptcy that fall could occur within a year. Meanwhile Amazon’s share price is at an all-time high, and while it is hiring 175,000 new workers, this is a small fraction of the workers it will replace, because the new way of doing business is massively more efficient. 

Even within online retailing and distribution indeed, the trend is towards the most automated operators. Companies like Amazon and Ocado, which have aggressively applied robot technology to warehouse operations, have been able to keep going throughout the crisis; smaller online operators, using more labour intensive processes, have found it difficult to keep operating safely: market share will shift to the most automated operators. 

Meanwhile the vast majority of all managerial and professional activity, and of data and information processing, have moved to remote working far more easily than anticipated. In developed economies, about 50% of all workers do a job which involves sitting in front of a computer screen, assessing and processing information, and communicating via email, with occasional meetings to interface with colleagues. Before the crisis, surveys of companies and employees suggested that only about 30% of such activity could be done effectively at home. 

But many companies whose whole business is about information — in financial services, accounting, law, human resources, consulting, or journalism — have moved smoothly to 100% remote operations: and across companies in all other sectors, information-processing activity has moved out of the office or call centre to be performed remotely in people’s homes. 

Some of that work will return to physical locations; some will remain done at home. But the successful experience of remote working will also stimulate many companies to consider more closely the potential for robotic process automation, which over the next few decades will do to repetitive information processing jobs what factory automation has done to manufacturing jobs over last 30. Those several decades may now become several years. 

The month of May is upon us and German thoughts turn to the Spargelzeit, the season of white asparagus. In Britain, many hope to eat strawberries at home, even if there will be no opportunity to do so at Wimbledon. But there are worries that a lack of migrant workers could leave produce rotting in the fields. Specially chartered flights, applying rigorous hygiene standards, have been used to fly in at least some Romanian workers to Britain. But the crisis will spur a shift away from migrant or any other form of labour. Already there are machines that can pick both asparagus and strawberries, but deployment has been slow because migrant labour is so cheap. Many firms will now start deploying it, and within five years, asparagus and strawberry supply will be highly resilient to the return of any future virus. 

Across multiple sectors of the economy automation will accelerate because the relative cost of employing labour has increased. Redesigning work processes to reduce human proximity; paying workers during sickness and enforced quarantine; monitoring workers heath; all these will add to cost in the short term. And even if by next year we have a vaccine and effective medicines for COVID 19, fear of future viruses will make companies worry that these costs will return. Substituting machines for labour will look like the less risky strategy. 

That will intensify striking recent shifts in both equity and job markets. 

Microsoft has a market capitalisation of over $1.3 trillion but employs only 150 000 people. Alphabet and Apple are worth $1.3tr and $ 1.2tr but employ only 120,000 and 140,000 respectively. Alibaba, Tencent, Bytedance and Facebook appear similarly huge in equity values but all their employees combined are a drop in the ocean of the global labour market. 

Such huge wealth but minimal employment creation is a new phenomenon: when General Motors was the world’s most valuable company, it employed almost a million people. The crisis has driven the relative value of these employment-light companies higher still. Netflix and Zoom are booming, and while the stock price of oil companies, airlines and hoteliers has collapsed, the NASDAQ is roughly where it was a year ago and up 90% on its 2016 level. In a world where the cost of computing and communication hardware keeps falling, and where software is infinitely replicable at zero marginal cost, very small numbers of highly skilled people can create products and services which deliver high consumer value with trivial employment. The crisis will only intensify that effect. 

Meanwhile, automation is destroying many jobs. Of course, that does not make permanent mass unemployment inevitable: we can always find more goods and in particular more services for workers to provide. Over the last 30 years, Europe and the US have seen relentless declines in manufacturing employment, but on the eve of the crisis employment rates in many countries were close to all-time highs as new jobs were created in face-to-face services — in restaurants, hotels, care homes and healthcare, in gyms and personal trainers, dog walkers and personal delivery services. 

The world doesn’t need many workers to make and move all the physical goods we can want: but so far it is shown a limitless capacity to create service jobs. A container ship carrying 20,000 containers from Pearl River Delta to Singapore and on to Rotterdam typically carries a crew of only 20 to 30 people, and port handling operations can be almost entirely automated: by contrast a typical cruise ship with 3000 passengers will have over 2000 more crew members to clean and cook, serve and entertain. 

But that comparison highlights a severe short-term problem. It is quite easy to design work processes that can make 25 crew members on a container ship safe from cross infection: but preventing infection from spreading in a ship carrying 5000 people is an almost impossibly difficult task. In all the sectors of the economy which can operate with reasonable safety — such as manufacturing, distribution, and construction, the need for tight health related controls will drive accelerated job reduction: but the hospitality and leisure sectors which have in the past absorbed the resulting surplus labour cannot for now perform that role. 

In the short term that means significant unemployment, and in the long run a further impetus to rising inequality. For the faster jobs are destroyed in the rapidly automating sectors, the lower will be free market wages for many workers in sectors which at least for now defy automation. Across the world, health and social care assistants are being applauded for their care of Covid-19 patients, but if we continue to buy their services at the lowest price which competitive markets can deliver, their wages will remain pitifully low. We should hope that the crisis does engender a greater sense of social solidarity; without it and the policy interventions it could inspire, the post Covid-19 world could see inequality continuing to rise. 

Economic policy in the post-crisis world, not only over the next six months but in subsequent years, should therefore include strong focus on job creation to offset the impact of accelerated automation. Public expenditures should support more spare capacity and higher wages in health and social care, improving resilience to any future health shock as well as treating workers more fairly. Policy should also be designed to increase resilience in the face of the equally dangerous long-term challenge of climate change. Maintaining the pace of renewable energy investment and technological innovation despite lower fossil fuel prices is essential: and major programs should be launched to improve the energy efficiency of commercial and residential buildings, for instance via better insulation, grid improvements to support the use hydrogen and the installation heat pumps, activities which are likely to remain labour intensive for many years to come. 

For developed economies therefore, the crisis will accentuate already existing challenges, but there are obvious policy responses. By contrast, low income countries with still rapidly growing populations will face more intractable problems. Many such countries hope to emulate the East Asian path to middle income levels, initially absorbing a rapidly growing workforce in low-wage manufacturing. But in a world where lengthy supply chains seem vulnerable to future health shocks, and where radical automation is possible, manufacturing activity may return to developed nations, but with few jobs attached. Remittances from migrant workers too may be threatened; and international tourism is potentially vulnerable to a resurgence of health-related fears. Countries with rapidly growing and youthful populations, above all in Africa, already faced a severe job creation challenge. Covid-19 will make it still more severe. 

Our current crisis may not produce the dramatic changes in direction which some commentators suggest: rather it will intensify trends already at work. But the challenges that creates will be no less profound.