Book Summary: The Key Ideas
#1: Technological Trends in Blue-Collar Work. Robotisation and increased automation will have a bigger impact on manufacturing in developing countries, while service industries, such as retail, will feel the brunt in developed countries.
#2: The Technological Threat to White-Collar Employment. Big data and machine learning, cloud computing and advanced algorithms present a unique threat to knowledge work, not only automating specific tasks, but potentially replacing the predictive judgments of humans.
#3: Six Economic Reasons Why This Time Is Different. Stagnant wages, labour decreasing as a percentage of national income, diminishing job creation, rising inequality, graduate underemployment, and job polarisation all suggest the economics are different this time around.
#4: Why New Jobs Won’t Offset. As emerging industries leverage the increased cognitive capability of machines, it is unlikely that new jobs will offset jobs lost.
#5: The Consumption Paradox and Guaranteed Minimum Incomes. As unemployment increases, consumption would be suppressed, potentially stunting further innovation. A guaranteed minimum income would be one way to address this issue.
Book Notes: The Key Ideas in Detail
Premise of the Book
The world economy is entering a new era, which will be defined by a shift in the relationship between workers and machines. While in the past machines have been seen as productivity tools for workers, machines are now turning into the workers. This will challenge our ideas about labour and capital in the economic system.
In The Rise of the Robots, Martin Ford sets out why he believes new technologies will ultimately push us towards a world economy that is less labour-intensive. Without the right policy response, Ford suggests we will confront a grave economic and human cost.
The book provides an extensive investigation of the key economic trends to support this perspective, as well as an exploration of the emerging technologies and how their economic impacts can be countered by policymakers.
Key Idea #1: Technological Trends in Blue-Collar Work
Ford believes robotisation is at the same early stage that computer software was at in the 1980s. We are likely to see a similar kind of exponential rate of acceleration in the coming decades.
Automation and rising overseas labour costs are already driving a “reshoring” trend in manufacturing, with the advantage of reduced transportation costs and other benefits such as reduced lead times and responsiveness to customers.
In developing countries, where manufacturing makes up a much bigger proportion of the economy, this means the impact of manufacturing automation is likely to be felt the hardest. In developed countries, on the other hand, automation will have the greatest impact on services. Already, for example, we are seeing clear evidence of that in the retail industry.
One of the biggest contributors to the evolution in low-skilled work will be cloud robotics. This is drastically speeding up machine learning, and already impacting logistics and memory for visual perception tasks in factories.
The impacts of automation will extend across a range of low-skilled occupations. This reaches beyond factory manufacturing and across industries such as agriculture and retail.
But as Ford argues, these changes will extend beyond low-skilled work. Knowledge work is far from immune to technology replacing many aspects of it.
Key Idea #2: The Technology Threat to White-Collar Work
Ford points to three key areas which are likely to threaten white-collar employment as we know it:
- Big data and machine learning: Big data will lead to automation of some specific tasks and jobs. But more importantly, the predictions extracted from data could to some degree substitute human judgement. This will flatten organisations dramatically.
- Cloud computing: This is reducing physical server requirements, and the impact we are already seeing on IT employment is likely to be a precursor for wider impacts on jobs involving information processing on computers.
- Algorithms: Open-source algorithms are now available that are capable of curiosity and creativity. Algorithms are already playing a significant role in financial markets, and this is now extending to creative ventures.
Unlike the trend we are seeing in manufacturing, offshoring is now prevalent across services. Ford suggests that this, like in manufacturing, is often the prelude for more significant automation.
Two industries where employment has historically appeared robust in the face of technological change are education and healthcare. Ford suggests this is unlikely to continue to be the case.
Algorithmic marking systems, plagiarism detection, facial recognition for exams, and massive open online courses (MOOCs) present a unique threat to university employment. Indeed, adaptive learning systems will eventually pose a threat to the delivery of tuition itself.
Meanwhile, artificial intelligence will be applied to medicine, and robotisation is already impacting elderly care and drug dispensaries. Ultimately, technology will respond to reduce medical error rate and counter the increased cost burden of an aging population.
Key Idea #3: Six Economic Reasons Why This Time Is Different
Since the Luddites of the 1800s, many have highlighted concerns about the impact of technology on employment. But despite all the doomsayers, the economy has adjusted time and time again, and new jobs have emerged.
Ford believes this time is different, pointing to six “disturbing economic trends” to support this case.
- Stagnant wages: While productivity has sharply increased in the US, real wages have stagnated since the 1980s. This suggests the fruits of innovation are accruing to investors and owners, rather than workers.
- Labour declining as a proportion of national income: Conventional economic wisdom (Bowley’s law) suggests that the percentage of national income going to labour and capital should be constant over time. Indeed, this was the case until the mid-1970s. But since then, the percentage of labour has declined while corporate profits have rapidly increased.
- Diminishing job creation and growing long-term unemployment: The rate of job creation has slowed in every recent decade in the US, while time to recover employment has grown after each downturn.
- Rising inequality: The top 1% of wealth is earning an increasing portion of the national income.
- Underemployment for graduates: The return on investment for university education is falling and the rate of graduate unemployment is growing.
- Polarisation and part-time jobs: Jobs created from recessions are generally worse than jobs lost in recessions. Ford suggests that “mid-range” jobs are being flushed out by the business cycle as companies enhance their technological standing.
Ford recognises that technology cannot be the sole driver of these trends, but suggests it is a primary factor. Globalisation, financialisation and politics are other obvious contributory factors, but are unlikely to explain all these economic changes. Ford also believes that technology will play the biggest role in these areas going forward, as the other areas have “played out”.
Key Idea #4: Why New Jobs Won’t Offset
The new technological landscape is unlikely to present the same offsetting employment opportunities of the changes of the past. This will not be like the motor industry replacing horse-drawn carriage manufacturers.
The primary reason is that new emerging industries will utilise automations and AI and therefore be less labour-intensive. While Ford recognises that some jobs will remain as human-machine collaborations (e.g. supporting input to algorithms), he believes those will be far fewer than many economists suggest.
Economists have typically pushed back on this idea because of the theory of “comparative advantage”. But this misses the point that computer intelligence is capable of cloning itself and being in two places at once, eliminating opportunity costs.
The difference now is “cognitive capability”. Moore’s Law applied to the progress of the microchip, but the rate of improvement in algorithmic performance has far outstripped it. It’s the potential cognitive capability of this technology which poses a unique opportunity and threat.
The other problem is that the digital economy has a long-tailed distribution. In other words, a few companies and individuals take most of its income. As mid-range jobs are eliminated, it’s likely that individuals will turn to digital initiatives, but the data suggests making a living from this will be extremely challenging for most of us.
Ford attempts to illustrate this point by looking at the two emerging industries of 3D printing and autonomous cars, concluding that both are more likely to shed jobs than create them.
Key Idea #5: The Consumption Paradox and Guaranteed Minimum Incomes
Ford’s central concern is that widening income inequality will stunt economic growth. In other words, if people are pushed into unemployment, consumption will inevitably decrease overall, creating something of a paradox. The innovation that brought about that unemployment would indirectly stunt funding for research and further innovation.
One popular counterargument is that technology will drive down prices, offsetting this adverse impact. But the deflation argument ignores that many would be on zero income and that deflation creates a spiral of expectations and makes debt even more difficult to manage.
The reality is that automation and capitalism go hand in hand, so anything short of full-scale state control is unlikely to suppress the instinct of the private sector to drive down prices. But a draconian intervention like this would be economically ill-advised.
Instead, Ford believes that the most practical long-term solution would be a guaranteed minimum income. A means-tested package would need to be carefully designed to avoid perverse incentives, perhaps starting at a relatively high level and rewarding those with a better education more highly.
The central economic argument is that where machines have replaced a substantial portion of human labour, a “form of direct redistribution of purchasing power becomes essential if economic growth is to continue.” A guaranteed income would also encourage more entrepreneurialism due to the safety net – a kind of Peltzman effect.
That said, Ford recognises this policy is not without its potential drawbacks. It could act as a disincentive to work for some, but this shouldn’t be viewed in universally negative terms. The policy may also impact housing costs, though Ford suggests less demand in cities could offset this impact.