What this covers

Larry Summers sits down with Tyler Cowen to defend his diagnosis of the global economy: that it faces a prolonged era of secular stagnation, characterized by chronically high savings, weak investment demand, and a depressed neutral real interest rate. The conversation traces how this structural condition—driven by longer life expectancy, rising inequality, falling capital costs, and demographic shifts—explains phenomena that have puzzled policymakers: persistently low inflation despite easy monetary policy, interest rates that fail to stimulate as theory predicts, and eroding worker bargaining power. Summers argues that major institutions, from the Federal Reserve to antitrust authorities to universities, have not yet adapted their frameworks to this new regime.

The discussion ranges across economics, finance, and policy, with Summers applying his secular stagnation lens to specific domains. He examines how disruptive innovation fails in higher education because universities design online programs as complements rather than replacements; why monopoly power suppresses demand by concentrating profits among low-spending owners; and how the Phillips curve relationship between labor tightness and wage inflation has fractured as worker leverage collapsed. On cryptocurrency, he voices unease about Greater Fool Theory dynamics in assets bought primarily for resale rather than use. He defends unions as a corrective to imbalanced bargaining power and explores immigration policy through the lens of social cohesion. Throughout, Summers distinguishes between what can be measured (where measurability bias distorts judgment) and what cannot, argues for nominal GDP targeting over the current inflation framework, and acknowledges puzzles the profession still lacks good tools to solve, including the Dornbusch Overshooting Model's inadequacy for exchange rates.

Sharpest takeaway

Summers argues that the US and global economy is in a prolonged era of secular stagnation—chronically high savings, low investment, and depressed neutral interest rates—which explains low inflation, low rates, and weak worker bargaining power, and that policy (the Fed, antitrust, unions, higher education) has not fully adapted to this reality.

  • Secular stagnation means the neutral real interest rate is very low, so seemingly low rates aren't actually expansionary and the Fed keeps undershooting 2% inflation
  • Structural factors (longer life expectancy, inequality, cheap capital goods, monopoly profits accruing to low-spending owners) are semipermanent, not transient
  • Worker bargaining power has fallen, breaking the Phillips curve wage-spiral relationship

The argument · threads12 threads · 46 claims
0.85

Secular stagnation, rooted in high savings and low investment propensity, will persist due to structural factors.

8 pointscentrality 5/5
  • The Fed consistently undershoots its 2 percent inflation target primarily because it has not grasped secular stagnation: with a high savings propensity and low investment propensity, the neutral real interest rate consistent with full employment is very low, so rates that would historically have been highly expansionary are not, causing the Fed to be disappointed by how little inflation it generates.

    “the neutral real interest rate, the real interest rate that’s consistent with full employment, is very low. That means that interest rates, which would historically have been highly expansionary interest rates, are not highly expansionary. And the Fed has underestimated the extent to which that’s true.”

  • The structural factors driving low interest rates—longer life expectancy (which makes people save more), increased insecurity, more inequality—plus factors holding down investment such as the demographic factor and the falling cost of capital goods, are more likely to be semipermanent than transient, meaning low real interest rates and secular stagnation will persist for a long time.

    “the structural factors driving low interest rates, including longer life expectancy — which makes people save more — increased insecurity, more inequality, are more likely to be semipermanent than they are to prove transient”

  • Greater monopoly power contributes to demand shortfalls and secular stagnation because monopoly profits flow to firms and owners with a low propensity to spend—firms don't invest and owners tend to be rich or endowments—raising savings and reducing investment; additionally, markets that aren't growing rapidly see less competitive entry, so weak demand creates its own lack of supply.

    “if there’s more monopoly, there’s more money going to monopoly firms where there’s a low propensity to spend it, both because the firms don’t invest and because the owners of the firms tend to be rich or endowments that have a low propensity to spend”

  • Given that secular stagnation makes the zero lower bound a likely binding constraint in the next recession, the status quo inflation framework is unsatisfactory; Summers's current instinct favors nominal GDP targeting because it has the attractive attribute that the slower the underlying growth rate—pointing toward a more serious secular stagnation problem—the higher the inflation it tolerates, though he'd want much more study before a definitive conclusion.

    “it has the attractive attribute that the slower the underlying rate of growth — which would tend to point towards a more serious secular stagnation problem — the higher the level of inflation that you’re prepared to tolerate is”

  • Although higher asset prices operate in the direction of raising investment (as in Q theory), if there are limits on how far interest rates can fall—because safe rates can't go much below zero and other rates must hold spreads above safe rates—the system may not be able to equilibrate; moreover, awareness that at some future moment of falling demand rates cannot be reduced sufficiently itself acts as a deterrent that holds current asset prices and demand down.

    “if there are limits on how far interest rates can fall — because safe interest rates can’t fall much below zero — and other interest rates have to hold spreads relative to safe interest rates, you may not be able to get the system to equilibrate”

  • Market volatility moves with the level of the market because of leverage mechanics: if a firm has $100 debt and $100 equity it is 50/50 levered, but if the stock rises so equity is $200 against $100 debt it becomes only one-third levered, so as the stock market rises its volatility naturally falls—which, along with profits and macro data coming in unusually close to consensus, partly explains the low VIX.

    “if a company has $100 of debt and $100 of equity, and then the stock market goes up, it’s 50/50 levered. If the stock market goes up by $100, then it has $100 of debt and $200 of equity and it’s only one-third levered.”

  • An asset makes a careful observer nervous when most of its demanders are buying it in anticipation of selling to someone else at a higher price, rather than buying it for some use they'll put it to or cash flows they'll receive—which describes Summers's unease about Ethereum and cryptocurrencies despite his belief that blockchains are fundamental technology with eventual winners.

    “It makes me nervous whenever I see an asset where I think most of the demanders are buying it in the anticipation of selling it to somebody else at a higher price rather than buying it in the anticipation of some use they’ll put it to or some set of cash flows they’ll receive.”

  • Regulatory stress tests (CCAR) should put more emphasis on market values rather than over-relying on regulatory capital measures, because in spring 2008 banks whose stock prices were collapsing were declared splendidly capitalized; while markets can be wrong and miss problems, an alarmed market should make regulators alarmed too, and the fact that stress tests say no bank would need capital even in an event worse than 2008 says more about the tests than the banks.

    “It’s madness that in the spring of 2008, when the stock prices of all the banks was collapsing, everybody was saying they were totally, splendidly capitalized and all was well. I would put more reliance on market instruments.”

0.74

Low post-war growth was driven by a growing labor force; future trend growth will remain near two percent.

5 pointscentrality 5/5
  • Postwar 3 percent US growth was largely a product of a rapidly growing labor force—driven by population dynamics, immigration, and the tailwind of more women entering the workforce—so going forward, while productivity may accelerate a bit, the historical average growth rate is unlikely to return; roughly 2 percent is mostly the new normal.

    “Mostly new normal because the labor force was growing much more rapidly because of population dynamics, because of immigration, and because of the tailwind from more women working through most of the postwar period than it’s likely to be going forward.”

  • Western alarm about a sudden breakdown of Chinese finances is likely overstated because a large part of Chinese debt is explicitly or implicitly government-guaranteed and the fiscal capacity of the Chinese government is very large given rapid growth; applying Western thinking understates that these are really internal government debts, so a 1997-Asian-crisis-style event within five years is not something to predict, even as growth likely slows into a smooth glide path.

    “a large part of Chinese debt is explicitly or implicitly government guaranteed, and that the fiscal capacity of the Chinese government is very large, especially given rapid economic growth”

  • Post-Soviet Russia turned out far worse than hoped (versus the West Germany 26-years-after-1945 model) for several reasons: 60 years of communism with no clear military defeat made integration into the world harder than rebuilding Europe; in retrospect there was not enough respect shown to a proud nation that didn't see itself as defeated—manifested in pushing NATO toward its borders and in harsh conditionality from international financial institutions; and there was probably too little funding provided quickly enough at the moment of maximum malleability.

    “in retrospect, there was not enough respect shown to what once had been a proud nation that didn’t think of itself as having been defeated. That manifested itself in some of the efforts to push NATO very far towards its borders.”

  • Mexico's surprisingly modest 2-2.5 percent growth despite NAFTA, reforms, and improved governance is a puzzle explained partly by serious rule-of-law and security problems (risk of becoming more like Colombia), and partly by a broader phenomenon where, in globalization, those who can build global supply chains thrive and the very poor are pulled up, but a middle group—including much of Latin America—can neither leverage the global economy nor wants to compete with Chinese labor, and is left behind and frustrated (the same problem as Michigan's).

    “Mexico has still some very serious rule-of-law issues, some very serious security issues associated with the risks of becoming more like Colombia than it wants to be”

  • Whether a rapidly growing country's infrastructure investment (an airport built where few people are) is brilliant long-term planning or foolish waste depends entirely on subsequent growth, so such investments can only be judged in the very long run—and even then, an ex post good or bad outcome doesn't settle whether the decision was ex ante sensible (as with the Japanese building ahead of overestimated growth, or Dulles Airport derided for decades before vindicating Northern Virginia's tech economy).

    “if you’ve grown fast, you’re a brilliant hero of masterful long-term planning, and if you’ve grown slow, you’re an idiot who’s building an airport where nobody wanted to go”

0.60

Wealth concentration and high savings by the wealthy drive down asset returns and neutral interest rates.

5 pointscentrality 4/5
  • Wealthy people do invest nearly all their money in financial assets rather than hoarding cash, but precisely because they pour it into financial assets they drive up asset prices, which drives down the return on those assets and thereby drives down the natural or neutral level of interest rates—so their investing does not eliminate an aggregate demand shortfall.

    “when they put it all in financial assets of one kind or another, they drive up the prices of those financial assets, which is driving down the return on those assets, which is driving down the natural or neutral level of interest rates”

  • The 1980s infinite-horizon models implying a zero capital income tax rate are now less applicable for two reasons: some of their mathematical results were shown to be wrong, and more consequentially their premise of infinitely elastic capital supply is a poor description of reality, since real interest rates fluctuate substantially without savings rising or falling correspondingly.

    “the premise of those models was essentially that the supply of capital was infinitely elastic. Whatever the tax rate, you would drive capital to the point where the after-tax rate of return was some fixed number. That now looks like a very poor description of reality.”

  • The optimal aggregate tax rate on capital income is closer to the rate on other income than to zero, because much capital income reflects economic rents, is held disproportionately by the wealthy, and includes unrealized capital gains that never get taxed—contradicting the view that capital income tax is merely a tax on future consumption that should be zero.

    “A fair amount of capital income reflects rents of one kind or another. Capital income is substantially held by those at the high end. There’s a fair amount of what’s really capital income in the form of unrealized capital gains that never gets taxed.”

  • It is an analytic conflation to equate the savings rate falling for various reasons with the savings rate becoming more elastic—these are separate issues, and there is no reason to think the savings rate will become more elastic even if it falls.

    “You conflated the idea that the savings rate would fall for a variety of reasons with the idea that the savings rate would become more elastic, which is a separate issue. I don’t see any reason to think the savings rate will become more elastic.”

  • Because corporations are creatures of their shareholders rather than separate entities, the way to get corporate America to fund infrastructure is to close corporate tax shelters—particularly addressing earnings stripping and transfer pricing, where profits that genuinely should be taxed in the United States are shifted to tax havens.

    “I would look, particularly, at issues relating toearnings strippingand relating totransfer pricing; where profits that genuinely should be taxed in the United States are transferred to tax havens”

0.56

Fed policy has undershooted its inflation target because it misunderstands secular stagnation dynamics.

4 pointscentrality 4/5
  • Because of reductions in workers' leverage and bargaining power, a degree of labor-market tightness that would earlier have set off a wage spiral no longer does so given how nervous workers are, so the Phillips curve relationship has either broken down or shifted, and the Fed has underestimated this.

    “a degree of tightness in labor markets that would in an earlier point have set off a wage spiral, no longer sets off a wage spiral because of how nervous workers are. So thePhillips curverelationship has either broken down or shifted”

  • There is a strong tendency in Japan and other countries, on seeing the first shoots of spring, to declare spring has arrived and prematurely tighten policy (the consumption tax hike analogy of taking down storm windows), which is usually a mistake; given this, Summers would defer Japan's planned consumption tax increase from 8 to 10 percent.

    “there’s a huge tendency in Japan and in other countries, when you see the first shoots of spring, to declare that spring is here and to take down the storm windows and put on the screens, and I think it’s usually a mistake”

  • The Fed claims a symmetric 2 percent inflation target, which logically requires being above 2 percent as often as below, yet inflation has been below 2 percent for nine years and not a single FOMC dot since the financial crisis has ever forecast above 2 percent—revealing a disconnect between what the Fed forecasts and its stated symmetric target; the remedy is to be genuinely symmetric and recognize policy is less expansionary than supposed, both counseling caution about tightening.

    “if you have a 2 percent inflation target and it is, as the Fed claims, symmetric, that means you should be above 2 percent as often as you’ve been below 2 percent. We’ve been below 2 percent for nine years now.”

  • The advice for filling Federal Reserve board seats should be to appoint serious, thoughtful people with real expertise who set monetary policy based on a pragmatic reading of data rather than a strong ideology or political orientation.

    “Serious, thoughtful people with real expertise who will set monetary policy based on a pragmatic reading of data rather than a strong ideology or a political orientation.”

0.50

Universities and philanthropists innovate too slowly and replicate success inadequately.

6 pointscentrality 3/5
  • A lot of successful innovation happens alongside big systems rather than by joining them or going to war with them, so in education the best philanthropic vehicle is acting constructively alongside the public schools—after-school, preschool, summer, or high-school-transition programs—rather than absorbing into or fighting the public school system.

    “a lot of the way successful innovation happens is alongside big systems”

  • Universities make Clay Christensen's elementary error by designing their MOOC and distance-learning efforts to be complementary to their on-campus education, judged by the standards of the existing product; genuine disruptive innovation requires separating the new effort from the original mission rather than keeping it 'within paradigm.'

    “the essence of Clay Christensen’s lessons aboutdisruptive innovationis if you want to do something all new you have to separate it from the original mission, not judge it by the standards of the original product, and let it be separate”

  • Unlike companies like General Electric that have radically transformed since 1975, elite universities (Harvard, Yale, Princeton, Stanford) look essentially the same—same size, same buildings, same calendar—so for an enterprise centered on ideas and young people the pace of innovation is stunningly slow, with the summer break being a vestige of agricultural planting schedules.

    “General Electric looksnothinglike it looked in 1975. Harvard, Yale, Princeton, or Stanford look a lot like they looked in 1975.”

  • Many education philanthropists set up charter schools that admit only highly motivated kids with highly motivated parents and pay 20 percent more to cherry-pick the best teachers from regular schools, then celebrate better achievement; but by the very nature of selecting students and cherry-picking teachers, such success is supremely non-replicable, so philanthropists should impose a replicability constraint on themselves.

    “their charter school is only going to admit highly motivated kids with highly motivated parents. Their charter school’s going to pay 20 percent more than the regular schools and cherry-pick the best teachers out of the regular schools.”

  • A philanthropist with limited resources relative to a city budget should resist the temptation to spread the money uniformly across many causes—if you do a little of everything you won't see any visible effect—and should instead concentrate on a couple of targeted, most-effective interventions.

    “your temptation is going to be to spread the butter uniformly across the bread and to try to do everything... If you try to do a little bit of everything, you won’t see anything you did.”

  • A philanthropist giving money to a specific government budget line, such as healthcare, must have a strategy for fungibility, because the city may simply reallocate its own money away from that area to other uses, so you will have demonstrated fungibility rather than spurred the intended outcome.

    “If you give more money to the health budget of the city and the city responds by reallocating its own money from healthcare to other things then you’ll have demonstratedfungibility; you won’t have spurred healthcare.”

0.49

Worker bargaining power has fallen due to weak organizing rights and labor-market leverage decline.

2 pointscentrality 4/5
  • Looking only at the union wage premium (now perhaps 7-8 percent and shrinking) misses the local nonpecuniary benefits of unions—lower turnover indicating higher job satisfaction, more respectful handling of grievances and less discrimination, and better job safety—as well as the broader political effect of organized labor pushing for policies like Social Security preservation and healthcare expansion that benefit everyone.

    “If you look at turnover, which is a measure of job satisfaction, it’s much lower in union employers... If you look at job safety, it appears to be better in union workplaces. So looking at wages is a good way to miss the local benefits of unions.”

  • The playing field for union organizing is not remotely level because if a company fires an organizer the only remedy is years of litigation yielding at most back salary net of new earnings, so the cost-benefit makes firing the would-be organizer incredibly attractive to employers; combined with 16 percent corporate profit growth against stagnant wages, this means more worker bargaining power almost certainly has to be good even if its magnitude is uncertain.

    “The cost-benefit for the employer has to be that it’s incredibly attractive to fire the would-be organizer. We don’t have a remotely level playing field”

0.39

Mergers should be evaluated on long-term economic health, not employment preservation.

3 pointscentrality 3/5
  • We should probably pay some attention to long-term economic health when approving mergers, but using employment as a merger criterion is worrying because efficiency-improving mergers will compete with others and may reduce employment—and we do not want to become like France, where job preservation becomes a central goal at the expense of economic dynamism.

    “Employment gets me nervous because if you successfully do something that is much more efficient, the effect will be that you’ll compete with somebody, and that may lead to reduced employment.”

  • It is much more likely that antitrust enforcement has been insufficiently tough than too tough over the last decade, but one must be careful because being a successful business naturally tends to produce more profits, which is usually a good thing rather than a bad one.

    “It’s much more likely that antitrust has been insufficiently tough than it is that it’s been too tough over the last decade.”

  • Difficult issues with information-central firms like Facebook or Google—privacy, reliance on information, network effects—may not be best thought of through the prism of antitrust and monopoly, especially since their products are given away to consumers for free.

    “are those best thought of through the prism of antitrust and monopoly? I’m not at all sure that thatisthe best way to think about them. After all, in some sense, the products in those two examples are given away to consumers for free.”

0.35

Immigration policy should frame immigration as becoming American to gain public acceptance.

2 pointscentrality 3/5
  • The right immigration deal should embrace the melting-pot idea that immigrating to the United States means becoming American—including learning and living life in English—because greater acceptance that immigration is about becoming American would generate broader public acceptance of higher immigration levels; while countries should make policy in the interest of current citizens, framing immigration as a broad obligation to humanity predictably meets resistance.

    “If we had more acceptance of the idea that immigration was about becoming American, we would have more acceptance of higher levels of immigration than generate comfort right now.”

  • Trump's debt-ceiling extension done with Democrats is more sui generis than a template for future bipartisan cooperation; there is unlikely to be much agreement with Democrats on substantive issues like tax reform or healthcare reform through similar coalitional arrangements.

    “I think the debt ceiling is, in some ways, more sui generis than it is a template for things to come.”

0.29

Financial and policy analysis requires careful attention to measurement limits and unobservable progress.

4 pointscentrality 2/5
  • Literature, as in Melville's depiction, evokes the nonpecuniary aspects of work—a texture of alienation and depersonalization—that are lost when an economist reduces them to a mathematical utility function of consumption, leisure, and job attributes.

    “There was a texture of alienation and depersonalization that was provided by Melville that one missed if one thought about things in the economic paradigm”

  • It is a common mistake to acknowledge progress only where it can be readily and concretely measured and then doubt whether progress exists where it cannot be measured; the right first approximation is to assume there is progress in all spheres, supported by evidence like the Flynn Effect showing rising IQ-test averages worldwide.

    “I think it’s a mistake that people generally make, to acknowledge progress in the places where it can be readily and concretely measured, and then be in doubt about whether there is progress in the spheres where it can’t be measured. The right first approximation is to think that there’s progress in all spheres.”

  • It is an important and not fully understood puzzle that in many places the long-term interest rate is more responsive to what the central bank in Washington does than to what the local central bank does, and the economics profession lacks a framework profoundly better than the Dornbusch overshooting model for thinking about exchange rate fluctuations.

    “there are so many places where the long-term interest rate is more responsive to what the central bank in Washington does than what the central bank in their capital city does”

  • Economists have not adequately explored several important digital-economy issues—intellectual property and how it is marketed and enforced, market structure in digital industries and what distinguishes increased competition from competition-foreclosing measures, and intangible infrastructures that promote exchange (such as why three-day financial settlements still exist given today's technology)—leaving policymakers to make bad choices uninformed by economics.

    “I think there’s a lot of work to be done on market structure in digital industries, what constitutes increased competition, and what constitutes measures that foreclose competition.”

0.25

Trade disruption's real risk lies in psychological effects on the global system, not depression.

3 pointscentrality 2/5
  • Pulling out of NAFTA or the Korea Free Trade Agreement would not cause a depression because the extra trade involved is too small relative to a $17 trillion economy; the larger risk is psychological effects on the global trading system, and abandoning Korea on trade amid the Korean peninsula crisis would be a grave strategic error fostering doubt among allies—though economists err when they claim every mistake will cause a depression, as in the Brexit debate.

    “Neither the extra trade caused by Korea or NAFTA is large enough, relative to our $17 trillion economy, to mean that those direct impacts would cause a depression.”

  • A treasury official should define objectives in terms of the level of the economy rather than the level of the currency; for the UK the most important thing is to avoid Brexit, which is a costly unforced error that will make Britons poorer and less influential, but failing that, preserve as much common access to the European market as possible and create a business-hospitable environment to attract capital inflows—which would both spur growth and create demand for pounds.

    “I don’t think I’d define my objective in terms of the level of the pound. I think I would define my objectives in terms of the level of the British economy.”

  • With North Korea, the right instinct is to treat the rational actor model as probably correct (about 80 percent confident), but because the consequences of being wrong could be enormous, one should pursue that model while heavily hedging against the possibility that behavioral, non-rational aspects are at play.

    “my instinct is the rational actor model is probably right, but that means I’m 80 percent sure that that’s right. The consequence of getting it wrong could really be huge.”

0.25

Effective mentoring, athletic improvement, and career development require talent selection and honest feedback.

3 pointscentrality 2/5
  • Effective mentoring is not an act of generosity but the selection of highly talented people to work with, where if you make clear you value their talents and tell them the truth directly about what they did well and poorly, they welcome it and both parties benefit.

    “if you pick able people and you make clear that you think very highly of their talents, they will welcome it if you tell them the truth about what they’ve done better and what they have done worse”

  • In sports where performance can be measured, there is vast and rapid improvement over time—a good college athlete can now run a four-minute mile though no one had before 1954, and a 1964 world-record swim wouldn't qualify for the 2017 NCAA championships—driven by better technique, training, intensity, and equipment, so today's best players almost certainly far exceed past greats.

    “a good college track athlete can now run a four-minute mile, and nobody in the world had ever run a four-minute mile before 1954. That the world record in swimming events in 1964 wouldn’t qualify you for the NCAA championships in 2017.”

  • Continuing to improve one's answers and thinking late in a successful career comes from two sources: surrounding oneself with extraordinarily able young people from whom one learns through both their knowledge and their probing questions, and a self-applied opposition to complacency—never being satisfied with any institution, oneself, or anyone, and continually revisiting how one could have answered better.

    “I’ve always tried to surround myself and be around extraordinarily able young people. They probably do learn some things from me, but I learn a lot from them”

0.18

Online education can deliver both massive scale and deep personalization simultaneously.

1 pointcentrality 2/5
  • Online education has a unique capacity to deliver both huge economies of scale (once a lecture is filmed, 100,000 can watch it at the same cost as 100) and much more personalization (re-listening, diagnostic questions that branch the lecture), making possible the rare combination of more differentiation AND more economies of scale simultaneously.

    “Once the lecture’s filmed, 100,000 people can watch it at the same cost as 100 people watching it.”