In his 2019 book. The Technology Trap: Capital, Labour and Power in the Age of Automation, Carl Benedikt Frey covers the history of the Industrial Revolution and how it affected society:
Frey’s argument rests on the distinction between labour-enabling technologies and labour-replacing technologies. As the names suggest, labour-enabling technologies complement workers, boosting productivity and opening up avenues for new employment, while labour-replacing technologies boot workers out of the labour market entirely, forcing them to re-skill or search for other opportunities.
The Technology Trap serves as a timely reminder that while technological change may benefit everyone over the long run, ‘short run’ adjustment costs can represent a lifetime for some workers. In detailing the history of the British Industrial Revolution, Frey highlights that ‘in the period 1780-1840, output per worker grew by 46 percent. Real weekly wages, in contrast, rose by a mere 12 percent’. Income inequality grew and workers generally saw no improvement in their standard of living for decades. This period of history became known as ‘Engels’ Pause’ and ultimately led to the Luddite uprisings of the early nineteenth century.
The rise of the personal computer and the internet was largely a labour-enabling technology.
On the other hand, Artificial Intelligence (AI) is unequivocally a “labour-replacing” technology and the political economy disruptions associated with that are far more significant and dangerous. As the OECD has said:
While firms’ adoption of AI is still relatively low, rapid progress including with generative AI (e.g. ChatGPT), falling costs and the increasing availability of workers with AI skills suggest that OECD countries may be on the brink of an AI revolution. It is vital to gather new and better data on AI uptake and use in the workplace, including which jobs will change, be created or disappear, and how skills needs are shifting. When considering all automation technologies including AI, 27% of jobs are in occupations at high-risk of automation. Initial findings from a new OECD survey of AI’s impact in the manufacturing and finance sectors of seven countries highlight both the opportunities and risks that AI brings.
Once AI deployment is mastered, many existing service firms will become little more than modules bolted onto an ephemeral AI spine. The rollout could be swift and will be unstoppable, given that it operates from a cloud with no borders.
This is troublesome enough for traditional economic models in which business investment leads growth. It will be a policy challenge to ensure the rising income of higher productivity accrues fairly to labour.
Fiscal redistribution is the most likely tool to make it happen via much higher unemployment benefits, retraining subsidies, increased taxation of AI entities, income tax cuts, or some forms of universal basic income.
But for Australia, with a growth model led by immigration and labour market expansion, the challenge looks more likely to be catastrophic.
Made considerably worse by the timing of it coinciding with the end of Chinese growth.
There are three legs to the current Australian economic model:
- Commodity exports support national income and tax revenues.
- Waves of mass immigration suppress wages to keep inflation and interest rates low.
- House price inflation is guaranteed by semi-nationalised banks that borrow from offshore, and the wealth effect drives consumption growth.
It is a slow-moving train wreck that shallows capital and is disproductive amid relentless falls in living standards.
But living standards fall relatively slowly, so underemployment not unemployment becomes the endemic maladjustment (which is fully intended).
This makes the model politically sustainable under a regime of progressive tyranny that suppresses all dissent as “racist” or xenophobic”.
But add AI and a step down in Chinese growth to this mix; disaster is what happens next:
- Commodity prices collapse, hammering national income, dumping wages and triggering tax hikes.
- AI delivers a relentless unemployment shock that dumps wages again and causes interest rates to crash.
- Unemployment not underemployment becomes endemic. House prices struggle to grow, and consumption stalls.
The critical calculation is the speed of the transition. So, it is impossible to say whether it would lead to a house price crash.
We can only say that property might unravel if the labour-led economic model of mass immigration becomes politically untenable overnight.
Or, the AI transition is so fast that unemployment spikes.
Or, the China shock prevents the fiscal redistribution of AI income benefits to workers.
Or, all three at once.
On the other hand, the probable inhibition of fiscal easing places more focus on monetary policy. Interest rates will crater and, like the end of the mining boom and capex cliff from 2012, property could boom in the short term.
Only to bust in the medium term as AI and unemployment keep pounding away at the model’s foundations.
There is one final alarming observation.
AI is just one unfolding labour shock. Peak Fat and autonomous vehicles are also coming to boost participation and further shock labour demand.
The lessons of the 19th century are not reassuring for this transition in the Australian economy. They suggest Australia’s macro institutions will be tested in extremis over the next decade.
Yet the entire political economy is in a ragged state. The progressive tyranny that enables the economic labour importation growth model mires politics and media in culture wars.
Neither has the older wisdom of the battle of capital versus labour which will describe the critical polarities in managing the AI et al transition.
At the policy level, I’m unsure if we can even discuss what needs to be done, let alone do what needs to be done, to prevent a catastrophic outcome.
