The Demographic Liability
AI, Inequality, and the Myopic Lens of Nationalism
“When, if ever, do you think our world will be post nation-state? If so, what form would it take?”
After Ant’s S1 dive last week, we are switching things up and getting a bit more philosophical. Bear with me as I bounce around a bit… these are important themes which I have yet to fully flesh out. It will not be the last time I write about them. As always, sharing is appreciated - my marketing regimen is rather… organic.
Re-diagnosing inequality after 2008: from Reaganomics to Trump, from Piketty to Lee, from policy to tech
Nationalism, AI and the coming economic “underclass”
Why accelerating tech + democratic governance = socialism
Emerging market “cyber-protectionism” - cheap labor is out, market access is in; why today’s tech chess match will evolve into a game of Go
The evolution of national identities, decreasing returns to violence in the digital age, and challenging the “national” lens of inequality in the AI age
In the debate on inequality, I find myself shifting increasingly from Thomas Picketty towards Martin Ford and Kai-Fu Lee. Picketty’s core argument is inequality stems from social & political factors while Ford and Lee point to technology and automation. Obviously, the answer is an interplay between the two, but the relative weightings are shifting rapidly.
I think 2008 marked the turning point. The global financial crisis raging amidst the launch of the iOS and Android operating systems. A computer in each pocket and a disenchantment with globalized supply chains and financial markets.
The march towards greater inequality continues. However, the baton has been passed. Unfortunately, our politics haven’t caught up.
Diagnosing the Problem
In hindsight, the “decoupling” of wages and productivity clearly started in the 70s and 80s.
The decoupling coincided with the dismissal of the gold standard (financialization) & the rise of the Laffer Curve (tax cuts), followed by 1980s Reaganomics to break out of the brutal stagflation of the 1970s. The prioritization was clear. Growth over equality. Fiscal stimulus over social programs. Private over public. Milton over Keynes. The Chicago School.
The school’s preeminence turned bi-partisan, crossing from the Reagan era into the Clinton administration where the twin pillars of globalization and ICT proliferation joined forces under the banner of the post-cold war free market expansion. Outsourced labor, multi-national corporations, optimized productivity, and global capital markets all fueling cheap consumer goods and high profits. The name of the game was growth - which, in absolute terms, proved fairly effective.
While the absolute gains were substantial, they were not evenly distributed. Inequality is now one of the dominant issues of our time; the under-current of the social unrest, extreme politics, rioting, and identity politics bubbling to the surface in 2020.
However, inequality depends very much on the lens from which you view. Picketty, a leftist French economist, focuses on wealth inequality in western democracies which is a very different lens from a Chinese technocrat. The same social, political and ideological factors driving “globalization” and “de-regulation” causing wage stagnation in the west clearly benefited large swaths of the emerging world. The arrangement - cheap labor fueling the American consumption engine - single-handedly lifted hundreds of millions of Chinese out of poverty. From a humanist perspective, liberalism, technology, and globalization have clearly been net positives for global equality.
However, nationalism still reigns in the 21st century. States maintain a monopoly on violence. Politicians are accountable to their citizens. And much of the citizenry in the west is unhappy with the status quo. Brexit, Trump, Bernie, Five Star, Le Pen - are all symptoms of the same underlying discontent. However, there seems to be a lag in our diagnostics.
The shift from an industrial economy to a services economy clearly had winners and losers. The shift from a services economy to a digital economy will concentrate the gains further still. Our world is moving on. The sad reality is many of the outsourced “manufacturing jobs” which Trump rants about will never come back to the US. If they do, most will be in the form of bits and robotics. Policy and globalization may have been the 20th century culprit; automation will clearly be the 21st’s. Unfortunately, the recent surge in nationalism seems targeted at the former.
We are fighting with the past while the future presses on. Exponentially.
Nationalism, AI & the “Economic Under-class”
As I pointed out, the shift appeared to happen in ~2008. Disillusionment with global markets coinciding with the rise of the smart phone. The seeds of both nationalism and AI growing in tandem over the past decade. While populist politicians scapegoated immigrants, outsourcing and financiers, fuel to the AI engine - mobile phones, cloud infrastructure, and digital engagement - went vertical.
In his book AI Superpowers, Kai-fu Lee - a leading AI researcher at Apple, Microsoft and then Google - outlines recent studies on the impact of AI on jobs. I’m genuinely skeptical of the luddite AI scare-mongering but hearing one of the field’s pioneers dissect why the studies forecasting an impact to ~20 - 40% of US jobs by 2030 is probably an under-estimate is… worrisome. The percentage is even higher in emerging markets with a greater skew towards unskilled labor. The coming automation onslaught will not be kind to the average Joe.
As we have explored in prior posts, AI is a centralizing technology. The concentration of market cap within FAAMG and BAT are harbingers of the world to come.
“As a technology and an industry, AI naturally gravitates toward monopolies. Its reliance on data for improvement creates a self-perpetuating cycle: better products lead to more users, those users lead to more data, and that data leads to even better products, and those more users and data. Once a company has jumped out to an early lead, this kind of ongoing repeating cycle can turn that lead into an insurmountable barrier to entry”
The same framework can be applied at the national level. Countries benefitting from AI network-effects today are likely to become increasingly dominant within the global economy. PWC estimates ~70% of the ~US$16t that AI will add to the global economy by 2030 will go to China and the US. Rising nationalism will ensure those gains stay at home. The closing international divide spawned by globalization will begin to yawn again. Pro-nationalist politics will fan the now-reversing international divide.
In the west, a combination centralizing profit pools (i.e. data / network effects) and democratic governance foreshadow a tilt towards socialism and populism. As relative wealth gaps increase, the growing number of disenfranchised laborers will inevitably vote for redistribution. In a top-down economy with means, like China, the stark inequality will also be assuaged by redistribution. In much of the rest of the world, there will be less wealth to redistribute. The manufacturing-led path to a middle class will become increasingly fleeting. Unemployment rampant. The welfare state non-existent. Political institutions fragile. Large swaths of unemployed youths have never been a promising formula for political stability. The demographic dividend giving way to a demographic liability.
Will countries earlier in development continue to grow faster? Yes.
Will the shift of supply-chains out of China benefit certain countries? Yes.
Will digital businesses in emerging markets thrive in the gap left by lower physical investment? Yes.
However, at the fragile inflection point for many of these nations, the tailwinds of globalization are now giving way to the headwinds of automation. From 2020 onwards, the average worker in Indonesia, South Africa, India, Turkey, Brazil - will be swimming against the current.
The “economic under-class” per Yuval Noah Harari.
Cyber-Protectionism, Market Access, and Weiqi
The job prospects in many of these countries may look grim. However, the desire for market access does not. Large populations = large data sets = training ground for AI / data = runaway flywheels. As laborers, large emerging market citizens will lose value. As digital netizens, their value is enhanced. Market access. As leverage decreases in certain areas (low-cost labor), governments will follow China’s example - using access to their large populations as a geopolitical and economic tool.
Techno-nationalism is on the rise as we have seen with China, India, and the US. However, it’s economic cousin - cyber-protectionism - will follow suite. In the AI-age, more and more digital sectors will likely come under the umbrella of “national security,” subject to thresholds for foreign shareholders. Under this framework, market entry will increasingly take the form of minority investments as opposed to unfettered market access. Backing local champions - by supplying capital and technology - will become the new normal. In a cyber-protectionist world, the tech-battle between the US and China is less likely to be a chess board, and more likely to resemble China’s Weiqi or Japan’s Go. A spattering of minority investments from both US and Chinese tech champions - using tech and capital - to back local proxies in an effort for profits, data, and influence. In short, the playbook of Chinese tech champions in India and Southeast Asia. Facebook’s investments in Jio and Gojek foreshadow the new reality for US titans.
Many more industries will be considered “strategic” like payments with numerous licenses needed to operate. Local partners will be more mandatory. In the current world of power-laws, and the coming AI-charged world of “super-power” laws, it may be the best tool emerging market governments have in their tool kit. Cyber-protectionism will allow large emerging economies to capture a small portion of the AI riches.
It is unlikely to be enough.
Myopia & National Lens
It’s a recurring theme. Many of the world’s most pressing issues - climate change, nuclear proliferation, AGI development, inequality - would be handled very differently by a global political body vs. national bodies. On the one hand, national-level power distribution fosters more dynamism - a diversity of experiments, a Darwinian jungle of organization and ideologies which fosters cut-throat competition and innovation in the name of… survival (see - atomic bomb). On the other hand, clearly different groups vying for resources, power, and influence will run into conflicts that a single, unified group would not (see also - atomic bomb).
However, it’s interesting we have arrived at the current status quo. I tend to side with Hobbes over Rousseau. Looking at expanding quality of life and declining violence, a monopoly on violence is the more necessary evil. But how do we decide where the monopoly on violence sits? The barriers of topography and climate combatted by human coordination and innovation which gave way to agrarian societies which gave way to cities which fostered unique pockets of language, culture, and organization which began clashing with other unique pockets of language, culture and organization has rendered the modern map of nation-states. A relatively modern phenomenon compared to the long history of human allegiances to say the tribe? The feudal lord? The Church? The King? The nation-state rose to prominence because it coincided with the industrial revolution - and the returns to large scale coordination, mass production, and violence increased dramatically in an industrial era.
However, the industrial age is giving way to the digital age.
Supply-chains are global. Our economies are intertwined. Our assets are intangible. Bits and brain cells are worth more than physical resources. The returns to physical violence in a digitizing world are plummeting. The mass coordination of the nation state for protection seems less relevant in the 21st Century.
However, the nation-state is as strong as ever.
Perhaps it is not the returns to violence which forecast our organizational structures, but the beliefs and values shared by the group. A communal series of ideas woven together. A national identity. Looking at history, the group-identity has generally grown in size to confront ever more complex challenges ahead. From tribes, to feudal peasants, to national laborers. Most of humanity’s challenges are now global in nature.
But it takes time.
Cultural change moves slower than technological progress. National loyalties outweigh humanitarian concerns. In a perfect world, the benefits of AI abundance would be a tide that raises all boats - from Silicon Valley to Surabaya. In reality, the coming AI-world is shaping up to be much more Machiavellian. Those with material comforts and less time will get more comforts and even less time. Those with little material comforts and plenty of time will get fewer comforts and more unproductive time. It seems the combination of nationalism and AI will get the distribution of the world’s resources exactly backwards.
Inequality is a matter of lens. The past 500 years have taught us to look through the nationalist lens. Looking through this lens, we realize inequality is an issue, but we are stuck in the 1990s diagnosing its drivers. The more interesting question though…
Is this even the right lens to be using?