What this covers

Simon Johnson, former IMF chief economist, discusses technology's distributional consequences, financial fragility, and the century-long lag between innovation and broad prosperity. Across economics, finance, and history, he argues against technological determinism—the notion that new tools automatically raise living standards. With Daron Acemoglu, Johnson developed the concept of So-So Technology, exemplified by supermarket self-checkout kiosks that shift work to consumers without raising productivity or wages. This framework explains why decades of digital deployment coincided with weak productivity growth and widening inequality. The Industrial Revolution took roughly a century before ordinary workers in Manchester, children in coal mines, and factory hands saw real wage gains after the 1850s—a pattern Johnson sees risking repetition with artificial intelligence unless deliberate choices create Task Creation vs Automation mechanisms that distribute gains broadly, as Henry Ford-era automation did.

The conversation ranges across financial architecture and historical precedent. Johnson emphasizes that bank stability, whether achieved through deposit insurance design, reserve requirements, or the structure of the banking system itself, results from deliberate institutional choices rather than market forces or scale. He examines how uninsured deposits and Bank Run Signaling create systemic risk, why the post-2008 concentration of US banking differs sharply from the historically fragmented system that built American innovation, and why Central Bank Digital Currencies pose complications during crises. He also addresses how gains from basic science—the Human Genome Project generating 300,000 jobs despite venture capital's rejection—depend on capturing Positive Spillovers, contrasts with crypto stablecoins operating as unlicensed banks, and reflects on why English regional decline mirrors the footloose nature of prosperity when institutions weaken. Throughout, the dominant claim remains that outcomes depend on policy, not inevitability.

Sharpest takeaway

Johnson argues that technology does not automatically produce shared prosperity; whether new technology (including AI) benefits ordinary workers depends on social and policy choices about whether it creates new tasks or merely displaces workers, and that financial and institutional fragility requires careful, deliberate management rather than faith in markets or scale.

  • Historically, industrial technology took ~100 years to deliver broad gains because gains were captured by elites unless task creation and bargaining power existed.
  • 'So-so' automation displaces workers without raising productivity, consistent with rising inequality and weak TFP since the 1970s.
  • Bank stability and institution-building are matters of deliberate design and choice, not automatic outcomes of size or market forces.

The argument · threads9 threads · 36 claims
0.85

Uninsured deposits and bank-like services outside regulation create systemic instability through credibility collapse.

7 pointscentrality 5/5
  • Because roughly half of US bank deposits (about $8 trillion of $18 trillion) are uninsured, the entire deposit-insurance system runs on signaling: if depositors come to believe uninsured deposits are in jeopardy, even letting one small bank fail can trigger flight to safety regardless of bank size, making the credibility problem extremely difficult to manage.

    “Uninsured deposits were about $8 trillion at the end of last year; insured was about $10 trillion. Like you say, 50–50, and it’s all about the signal because if you believe that uninsured deposits are in jeopardy”

  • The bank-nonbank distinction is hard to manage because money market funds, allowed to appear in the 1970s as a reaction to high inflation and interest rate controls, let a genie out of the bottle that is very hard to put back; therefore one should be cautious about handing out banking licenses to nonbanks and equally cautious about nonbanks providing bank-like services.

    “it all goes back to allowing money market funds to appear in the 1970s, which, in itself, was a reaction to high inflation and controls and interest rates. Once that genie has been let out of the bottle, it’s very hard to put it back in.”

  • A shift of deposits away from regional and mid-size banks toward very large banks would be disruptive because large banks are not good at replacing what regional banks do—lending to the nonfinancial sector across the country—though community banks might step up to fill some of the gap.

    “if there is a shift in deposits away from, let’s say, regional banks or mid-size banks towards very large banks, for example, they’re not going to be particularly good at replacing the regional banks in what they do in terms of lending to the nonfinancial sector all across the country”

  • A central bank digital currency is far off in part because in a period of stress it would make bank runs worse: depositors could instantly migrate from vulnerable bank accounts into the CBDC, dramatically reducing banks' ability to fund their loan books, so the complications of a CBDC are more apparent during a banking crisis than before one.

    “If you had a central bank digital currency, I think a lot of people would’ve migrated... from the bank account into that central bank digital currency, and that would be a big shift in the ability of the banks to fund their loan book.”

  • Privately issued stablecoins are unstable and poorly regulated, and represent the main part of the crypto space with systemic implications, because they provide bank-like services without a banking license—and one should be very cautious of anything offering bank-like services without such a license.

    “I’d also be very careful and cautious of anything that provides bank-like services without a banking license, and that’s exactly what stablecoins are.”

  • A sensible deposit insurance reform would keep individual coverage at $250,000 but carve out and insure small-business transaction/operational accounts (payroll, working capital), because forcing a nine-person startup to spend time on bank financial management is an unfair distraction, while insuring all deposits should only be done in desperation.

    “if you have a nine-person startup, telling them to spend time on financial management when they’re trying to build a new hot sauce company — whatever it is — it’s a distraction and rather unfair.”

  • Crypto exchanges should be regulated as exchanges regardless of what they call themselves—'if it walks like an exchange and talks like an exchange, it's an exchange'—and there is a fairly clear definition of what counts as a security that regulators can apply.

    “if it walks like an exchange and talks like an exchange, it’s an exchange. Doesn’t matter what you call it.”

0.81

Technological progress historically takes decades to translate into shared prosperity for ordinary workers.

6 pointscentrality 5/5
  • Although entrepreneurs tinkering with technology drove the Industrial Revolution, working-class living conditions—Manchester's slums in the 1830s, children in coal mines in the 1840s—were worse than before the revolution, and ordinary people did not see real gains until after the 1850s; it took roughly a hundred years for technological progress to translate into shared prosperity.

    “if you look at the conditions of workers and children working in coal mines in the 1840s... it was also absolutely terrible and much worse than the conditions for children before the Industrial Revolution.”

  • The prevalent techno-optimist view—that technology simply happens, raises productivity, improves lives, and eventually benefits everyone—is too simple; looking back over 1,000 years, technology is sometimes massively beneficial but the Industrial Revolution took more than a hundred years to deliver broad gains, so the policy challenge is getting better outcomes for more people sooner.

    “We are challenging... some modern techno optimism view... that technology just kind of happens. It raises productivity. It improves how people live... and everyone benefits eventually. Well, when we look back over the past 1,000... we find it to be more complicated.”

  • The central worry about AI is that it may displace workers from previously well-paying jobs without creating new tasks, so worker productivity does not rise; while AI designers may be paid more, most people may end up paid less in real terms—breaking the historical mechanism (as under Henry Ford) where automation accompanied new tasks that made even low-skill workers more productive and whose gains were shared via wages and unions.

    “more workers may be displaced from previously well-paying jobs, and without creating new tasks. Therefore, worker productivity doesn’t go up. You may pay some people — maybe the designers of AI algorithms — more money, but most people may well be paid less in real terms.”

  • A large part of what went right in the Henry Ford era was the work of 'engineer managers' who created new tasks, designed companies, and built productive organizations; despite the later bad name of bureaucratic middle management, people who combine engineering aptitude with the desire to organize and improve productivity remain a very important group, which is why the future of the MBA may lie with engineering-minded managers attentive to task creation.

    “a big part of what went right in the era of Henry Ford was the so-called engineer managers, the people who created the new tasks, the people who designed companies, the people who built these organizations.”

  • OpenAI's relatively low ~$30 billion valuation despite ChatGPT being the most rapidly adopted application in history reflects the cliché—correct in this case—that we overestimate technology's immediate impact and underestimate its longer-term impact, with stock prices reflecting long-term views, leaving open whether AI value will accrue to foundational models or downstream applications.

    “we overestimate the immediate impact of technology and underestimate the longer-term impact of technology. Presumably the stock price would reflect people’s views of the longer-term impact.”

  • Much digital technology since the 1970s was deployed not because it boosted productivity but because management thought it would displace workers; Acemoglu and Restrepo's 'so-so technology'—exemplified by supermarket self-checkout kiosks that shift work onto consumers without raising worker productivity or wages—explains why technological change has coincided with weak TFP growth and widening inequality.

    “Daron Acemoglu, my co-author, and Pascual Restrepo coined the term so-so technology, where you automate even though it doesn’t boost productivity... but you do displace workers. And their favorite example is self-checkout kiosks at supermarkets”

0.67

Banking crises arise from risk concentration in large institutions, not from housing bubbles alone.

6 pointscentrality 4/5
  • The IMF's pre-crisis view that derivatives had spread risk around the world was completely wrong; derivatives in fact concentrated risk in very large US and European financial institutions, and it was that concentration—not a classic housing bubble—that generated the panic and the downward pressure on house prices, so 2006–07 may have been less a housing bubble than a negative bubble in shadow banking.

    “we had a view that was completely wrong, that derivatives had spread the risk around the world. In fact, they concentrated the risk, and it was that concentration of risk in these very large financial institutions in the US and in Europe that caused a lot of the panic”

  • There no longer appears to be a clear minimum bank size at which a US bank can simply be allowed to fail, because the 2018 reforms set systemic attention at $250 billion in total assets, yet SVB and Signature—both smaller—were treated as systemic, suggesting the threshold for 'too big to fail' is unstable and shifting downward.

    “The view coming out of the reforms or changes of 2018 was that $250 billion total assets was where systemic attention should start to be paid. Silicon Valley Bank was smaller than that; Signature Bank was significantly smaller than that.”

  • In the Credit Suisse rescue, AT1 (CoCo) bondholders were taken to zero while equity holders received about $3 billion for a bank worth $8 billion the prior Friday, treating the convertible bonds as worse than equity, which will cause significant repricing in that market and effectively undermines CoCo bonds as an instrument.

    “the equity holders got some compensation in the purchase. They got $3 billion for a bank that was worth $8 billion on the Friday before this happened, but the AT1 bonds were taken to zero.”

  • Whether a country can host very large banks depends on the government's ability to stand behind them—including in foreign currency—so Switzerland, with around $1 trillion in central bank FX reserves backing a roughly half-trillion-dollar Credit Suisse balance sheet, had the capacity, whereas Iceland in 2008 could not back its outsized banks, which probably should not be allowed.

    “size relative to the ability of the government to stand behind its banks, including in foreign currency when they’re operating not in the domestic currency.”

  • A better resolution of SVB would have been to impose a haircut on its uninsured depositors—who likely would have recovered around 90 cents on the dollar given the high-quality but underwater government bonds, plus fast FDIC cash advances—while simultaneously insuring all deposits at other banks, thereby addressing moral hazard at SVB while preventing contagion elsewhere.

    “what you could have done was have the Silicon Valley Bank uninsured deposits take a haircut. The assets of that bank were mostly high-quality, long-term government bonds that were underwater, but only because of the interest rate increase. They probably would’ve got about 90 cents on the dollar”

  • Russia should be removed from the IMF board of directors because its conduct over the past year and longer—invading other countries and disrupting the world economy—is outside what is acceptable for nations, and autocrats who behave aggressively should not sit at the big table unless they reform.

    “I would remove the Russians from the board of directors... when autocrats behave badly and aggressively and invade other countries... you should not have them at the big table.”

0.58

Market concentration and firm size matter only if they block innovation, variety, and price competition.

4 pointscentrality 4/5
  • Even if a technology like AI initially looks distributive—as crypto and the early internet did—the US tends to recombine things, letting firms get big and powerful quickly; this happened with social media, which we came to regret, so there is no guarantee AI's gains will remain distributed rather than concentrated.

    “I think we tend to recombine things in this country in particular, Tyler. Again, we get big quickly, and we let people get powerful quickly. I think in the case of social media, we regretted that.”

  • A strength of the US financial system historically is its multitude of banks and suspicion of big banks; US economic growth, innovation, and productivity were not built on or associated with big banks, which only became large relative to GDP since the 1980s, unlike Europe's universal-banking model adopted in the late 19th century.

    “The US economy, US innovation, US productivity was not built on big banks. It was not associated with big banks. Large banks or banks that are large relative to our GDP are something that’s developed much more recently, since the 1980s.”

  • The universal banking model praised in 1980s-90s Japan and Germany was overrated: Japanese banks ran into big trouble at the end of the 1980s and have been rough since, and European banks like Deutsche Bank are widely regarded as problematic (surviving partly on German state backing), suggesting banking is a difficult, problem-prone business however it is organized.

    “The Japanese banks ran into big trouble end of the ’80s and the ’90s... the European banks are also regarded as an imperfect model. Deutsche Bank, for example, is often held out as being very problematic”

  • Market concentration is problematic only when it prevents innovation, prevents the development of varieties, or keeps prices high; if a concentrated sector fully satisfies consumers with adequate innovation and progress—as may be true where real prices fall and variety rises—there is much less to object to.

    “if concentration prevents innovation, if concentration prevents development of varieties, if concentration keeps prices high... then it’s problematic. If it’s fully satisfying the consumers... then yes, sure, there’s less to object to.”

0.46

Wage stagnation for non-college workers since the 1970s represents a reversible but difficult policy failure.

1 pointcentrality 4/5
  • The concentration of top wealth has never been Johnson's major concern; what matters is the stagnation since the 1970s of wages for people who did not complete or attend college—a sharp break from the broadly shared wage growth of the preceding 30-40 years—which is unnecessary and reversible in principle but very hard to reverse.

    “I’ve never been bothered by the top wealth, Tyler. It’s never been a major concern of mine. What bothers me more is lack of good wages, rising wages for people who haven’t completed college”

0.44

Strong institutions provide resilience against crisis and collapse, though they forecast near-term growth poorly.

7 pointscentrality 2/5
  • A strength of the American political system is the absence of a closed elite—you don't have to attend Oxford and study a particular degree to become a national leader—and the intense, open back-and-forth of American politics is valuable, contrary to fashionable disdain, with the country tending to do the right thing after exhausting all other possibilities.

    “We don’t have a closed elite. You didn’t have to go to Oxford and study a certain degree in order to become prime minister. I like the intensity of the back and forth in American politics.”

  • Countries with a credible reputation for fiscal discipline create space for extraordinary monetary operations during crises, as Indonesia's improved macro management over 20 years allowed money-financed deficits during COVID; but money-financed deficits are dangerous and should be used only when there is no alternative.

    “If you’re credible on the fiscal side, I think you do create space, Tyler, for extraordinary operations at a time of crisis.”

  • The real regret about social media is not market structure—the sector is competitive and has overexpanded—but its impact on political, economic, and social discourse, on civility, and on young people's mental wellness; companies could have been more responsible and the process of learning to use such technology responsibly is itself a matter of choices.

    “the regret actually there, Tyler, is much more about the impact of social media on political, economic, and social discourse, the way that people talk to each other... what has happened to civility.”

  • Institutions act like a positive hysteresis effect—once growth strengthens them they make relapse harder—but they have predictive value only over the long term, not over one, two, or five years; their real usefulness is in assessing whether a system is robust enough to handle crises and avoid hard-to-reverse collapse, not in forecasting near-term growth.

    “I think of institutions as being part of the hysteresis effect, if you can get it in a positive way, that if you grow and you strengthen institutions... it makes it much harder to relapse.”

  • Beyond intelligence and work ethic, the non-obvious quality to look for in promising students is the ability to communicate ideas—especially about technical topics—clearly and without condescension, in a way that invites further debate, and to listen to feedback and sharpen the message.

    “can people communicate ideas? Can they communicate with me?... about technical topics? And do it in a way that isn’t condescending... a way that invites further debate and conversation? And then, do they listen to the feedback they get and sharpen the messages?”

  • The British parliamentary system's instability stems partly from the fact that a party can win a majority with a minority of the vote and a relatively small shift in the vote can produce a large, unassailable majority; combined with long single-party runs that exhaust ideas, this is the unfortunate downside of an otherwise robust and entertaining system.

    “you can get a majority with a minority of the vote, and also, a relatively small shift in the vote can give you a big majority, and those big majorities are unassailable.”

  • Neal Stephenson's Snow Crash, written about 30 years ago, is the best science-fiction predictor of the future—anticipating the metaverse and various political-social developments including societal fragmentation—though Johnson believes our actual future will not end up quite as dystopian as the novel's political disunion.

    “Paul Krugman said in 1992 or something that it was the best possible predictor of the future, and it turns out to be pretty much right... Stephenson predicted the metaverse.”

0.40

Regional imbalance and stagnation require distributed science investment and entrepreneurship, not geographic concentration.

3 pointscentrality 3/5
  • England's urban geography is uniquely unbalanced—unlike France's Lyon or Germany's healthy second cities—because London has been an exceptionally dynamic, entrepreneurial trading center open to new ideas for over 500 years, which both enabled British industrialization and continues to swamp other regions, while the North's factory-based prosperity proved footloose as the empire declined.

    “it’s the appeal of London and the success of London over more than 500 years, Tyler. The North of England had a very good run... but it was a run that was based on factories. And factories and concentrated workforces... turned out to be fairly footloose, particularly after the empire declined”

  • Place-based science policy is now more encouraging because the US has a deep bench of talent in lower-cost regions outside Boston, New York, San Francisco, and Seattle, and since public basic-science investment is justified by spillovers, those spillovers should be sought more broadly to make other regions central—an approach aided by the pandemic's shrinking of distance.

    “All these other places where real estate prices are not so high — there are plenty of talented people who want to live there... why not look more broadly... and attempt to make those places more central, tap into the talent that already exists”

  • Escaping the middle-income trap requires an education system that encourages creativity and breaks from past strictures, more entrepreneurship, and deep integration with the world economy and exchange of ideas—as seen in South Korea and especially Ireland, whose globalized, English-speaking, adaptable people enabled it to break free with foreign investment and integration.

    “you need to spread an education system that really encourages creativity... You need more entrepreneurship. You need a lot of integration with the world economy, a lot of exchange of ideas.”

0.32

Government should invest in basic science with foreseeable applications that private markets reject.

1 pointcentrality 3/5
  • Government should invest in basic science where commercial applications are foreseeable but the private market won't act—as with the Human Genome Project, which venture capital rejected in the 1980s as unmonetizable yet later generated roughly 300,000 mostly good-paying jobs once downstream firms could appropriate the basic science—because the rationale is capturing large spillover effects rather than duplicating private financing.

    “The Human Genome Project is one of our favorites. That was actually turned down by private venture capital in the 1980s because they said... we don’t know how you can monetize it. That turned out to be a brilliant moneymaker... I think we got 300,000 jobs out of that”

0.18

Medieval and early-modern gains in productivity failed to create shared prosperity because surplus was extracted.

1 pointcentrality 2/5
  • There was substantial medieval innovation in agriculture, commerce, and early industry, but its gains did not filter down to ordinary people; instead surplus was squeezed from coerced, employer-restricted peasants who became more productive yet were squeezed harder, and the wealth flowed to the church and state who built cathedrals—symbols of medieval despair, rising productivity, and the failure to create shared prosperity.

    “you could see — because these were simple economies — how the money was squeezed out of peasants... They became more productive, but they were also squeezed harder.”