Tariffs, Decline, and the Promise of AI | A Conversation with Larry Summers and Niall Ferguson
What this covers
Economist and former U.S. Secretary of the Treasury Larry Summers and historian Niall Ferguson discuss the Trump administration's economic policy, the promise of AI, and more. Their discussion is followed by a Q&A with University of Austin students.
This discussion occurred on April 1 during the University of Austin's first week of spring term.
Source description (no synthesized summary yet).
Economics as a discipline is not in decline but rather in ascendance, and despite legitimate critiques of specific models and forecasting failures, the field's expanding influence across policy and improved measurement of forecast accuracy demonstrate growing rather than diminishing intellectual authority.
- Economists have expanded from recession prediction into regulatory cost-benefit analysis, antitrust policy, and central bank leadership across the world
- Root mean square error in economic forecasts has trended downward through time, showing improved accuracy
- Other social sciences increasingly import economics' techniques and approaches rather than the reverse
This asset isn't compiled yet
You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.
Explicit mathematical and quantitative models have the advantage that when they fail, you can see the failure and fix them, whereas implicit subjective models fail invisibly; therefore the impulse to use mathematics and data more rather than less is proper.
“And it's not that non-economists have some alternative model. So, the question can be what are the virtues of explicit more mathematical models versus implicit more subjective models. And certainly quantitative models sometimes go wrong, but they have the feature that when they've gone wrong, you can see that they've gone wrong and you can change them. And you can uh fix you can fix them. So, I think the impulse to use mathematics, to use statistics, to use data more rather than less has to be a proper impulse.”
The world engages in too little experimentation because the experimenter does not capture all benefits (from copiers) and does not benefit from others learning from failures, creating an externality that reduces innovation.
“the world engages in much too little experimentation because the person who does the experiment doesn't get all the benefit from all the people who copy it and doesn't get all the benefits from the people who don't have to try it again after it fails. So there's just too little experimentation in the world in general”
Economies have capacity to work around blockages more than typically imagined, illustrated by fact that despite US strategic bombing of Germany in WWII it did remarkably little damage to German economic functioning
“I think that part of the right interpretation of that fact is that economists are more like frogs than they are like people. If you [54:35] cut off a frog's tail, it regrows the tail. It's able to reconnect all the systems and that economies probably have more capacity to work around blockages than we imagine. That's one aspect of it.”
One can improve car safety by studying automobile accidents and developing better brakes and tires, without being able to predict where and when the next accident will occur, illustrating that understanding mechanism doesn't require predictive power
“I can't predict with any great accuracy where and when the next automobile accidents will come. But by studying automobile accidents one can contribute to the development of better brakes and better tires and therefore make cars safer. And so [1:10:51] judging the ability to make something better by the ability to uh make uh predictions is I think uh not uh the right uh not the right criteria.”
All thought and judgment essentially takes the form of models; humans cannot think except with models, whether implicit subjective models or explicit mathematical ones, so the question is not whether to use models but which type of model is more useful.
“we all have models. How do you think about things except with a model? Why is it that when somebody sticks their hand out, I understand that I'm supposed to stick my hand out to shake hands? It's because my model of human behavior is that when somebody shakes their hands, that's what you do. You shake hands. So, all thought essentially takes the form of models. All judgments about, you know, is UATX going to be a success over the next 5 years? Is the economy going to grow fast next year? Reflect a kind of mental model of how the world operates. So, the question isn't should we have models or not have models. In the absence of models, you can't think.”
The University of Chicago's free tuition model is not 'unfair practice' that disadvantages competitors, but rather a good thing that benefits students; similarly, if China wants to sell low-cost goods to America, this is a beneficial trade that Americans should accept, not characterize as cheating.
“if China wants to sell us things at really low prices and the transaction is we get solar collectors that help make there be less global climate change or we get batteries that we can put in electric cars and we send them pieces of paper that we print. Think that's a good deal for us or a bad deal for us? Kind of think it's a good deal for us. So, the idea the way you frame the question which is that this is cheating and is there a way to deal with the problem of cheating? I think is largely wrong.”
The purpose of social science is not primarily to make predictions but to understand how social processes operate so one can contribute to their design and improve outcomes.
“It is to understand how social processes operate both because that's intellectually satisfying and because with a better understanding of how social processes uh operate, one has a better prospect of contributing to their design and contributing to the choices that people operating in social settings make so that there will be better uh so that there will be better outcomes.”
Even if new taxes are not enacted, a structural revenue problem remains because services (government's primary product) increase in price faster than goods, and existing debt requires more revenue just to avoid deficit expansion without policy changes.
“But I think the serious issue for the country that we haven't really reckoned with is that even if everybody does everything right because more of the population is aged because what government does is provide services and the price of services goes up relative to the price of physical uh uh goods because we've got a bunch more debt than we used to we just to even if we don't do anything new to avoid a big deficit we have to find a way to raise more revenue.”
The virtue of explicit quantitative models over implicit subjective models is that when they go wrong, you can see they've gone wrong and fix them, whereas implicit models conceal their failures
“And certainly quantitative models sometimes go wrong, but they have the feature that when they've gone wrong, you can see that they've gone wrong and you can change them. And you can uh fix you can fix them. So, I think the impulse to use mathematics, to use statistics, to use data more rather than less has to be a proper impulse.”
All thought necessarily takes the form of models—implicit mental models that everyone uses to understand the world, and the question is not whether to use models but whether explicit mathematical models are preferable to implicit subjective ones
“we all have models. How do you think about things except with a model? Why is it that when somebody sticks their hand out, I understand that I'm supposed to stick my hand out to shake hands? It's because my model of human behavior is that when somebody shakes their hands, that's what you do. You shake hands. So, all thought essentially takes the form of models. All judgments about, you know, is UATX going to be a success over the next 5 years? Is the economy going to grow fast next year? Reflect a kind of mental model of how the world operates. So, the question isn't should we have models or not have models. In the absence of models, you can't think.”
The inability to predict imminent events (like earthquake timing or hurricane location) is not a valid test of whether a discipline is in intellectual decline, because geologists and seismologists understand earthquakes better than anyone else does but still cannot predict when the next one will occur, and meteorologists understand hurricanes better than anyone else but don't know exactly when or where the next one will come.
“I don't think so. First of all, I'm not sure it's a proper test of a science to be able to predict imminent events. Geologists and seismologists know more about earthquakes than anybody else does, but they can't predict when the next one is uh going to come. Meteorologists understand hurricanes [0:30] better than anybody else does, but they don't know exactly when or where the next uh hurricane is going to come.”
Western sanctions on Russia have been less effective than hoped due to three factors: economies have resilience to work around blockages (like during WWII bombing), sanctions may not be complete due to corruption/leakage, and Europe has insufficient incentive (stagnant economy wanting to trade with Russia).
“I think that part of the right interpretation of that fact is that economists are more like frogs than they are like people. If you cut off a frog's tail, it regrows the tail. It's able to reconnect all the systems and that economies probably have more capacity to work around blockages than we imagine. That's one aspect of it. The other aspect of it is whether we were really sanctioning them or whether there was all kinds of corruption and leakage that caused the sanctions not really to happen and we had a big problem which was we had two big problems. One is that their big revenue source was the thing that we most wanted. That being oil. Because gasoline prices are hugely salient politically in the United States and around the world. So how good a job were we really going to do of cutting off their exports? Answer, not very. And that we had a whole continent with a relatively stagnant economy that really liked selling things to Russia.”
One problem in organizations where people do the same job for 15-20 years is that they tend to get attached to the way they've done things before and not update things in respect to ways the world has changed, which is a tendency at the Fed
“One of the problems in any [7:49] organization with having people do the same job for 15 or 20 years is that they tend to get a little attached to the way they've [7:49] done things before and not update uh things in respect to the ways the world has changed. Is there that kind of tendency at the Fed. So, yeah, I think I was right to criticize [8:20] the Fed's model”
45% of American imports are inputs to American exports, making supply chain integration a critical fact that tariff policy undermines by raising the cost of export production.
“Here's a number that when I first heard it, even though I've been studying this stuff for a long time, surprised me. 45% of American imports are inputs to American exports.”
National security dependence on China (particularly for sensitive technologies) could be a legitimate basis for policy, but general cheap imports are not a cheating problem—the solution is stockpiling, domestic capacity investment in semiconductors, not blanket tariffs.
“There's an argument you can make that is about dependence perhaps with respect to really sensitive technologies. We don't want to depend on China cuz then they might cut us off and their government might take control and that might be bad for our national security and there are a lot of things to do about that. We should stockpile more things than we do. There are probably some areas like semiconductors where we should be investing in developing domestic capacity. But that's not an argument that because they're making something available cheap it's cheating and you should have a policy.”
Tariff uncertainty induces people to delay purchasing decisions (automobiles, washing machines, factory construction) hoping the uncertainty will be resolved, creating additional economic damage beyond direct tariff effects
“you're inducing enormous amounts of uncertainty. And when there's uncertainty people delay hoping that the uncertainty will be resolved. So, you know, tariffs up for 3 days, tariffs down for 6 days. I mean, we, in addition to what the effect of it is, everybody thinks God, I [16:41] Why should I buy a washing machine till this is all clearer? Why should I buy a car till this is all clearer? Why should I input start make commit to building a certain kind of factory until this is all kind of clearer?”
China is raising quality of life for a billion people and may not pose a threat; hedging against Chinese hegemony without provoking them back into hegemonic buildup is the genuine strategic dilemma
“And so, the strategic dilemma with respect to China, and it's a very hard thing to manage, is how do you hedge against the possibility that they are going to aggressively seek hegemony in ways that deeply threaten us without your effort to hedge seeming a provocation to them to build up more strength because they fear you.”
Low-price Chinese imports benefit U.S. consumers by providing cheaper goods and technologies (solar panels, batteries), similar to how technological innovations like fluoridated water harmed dentists but benefited society overall
“if China wants to sell us things at really low prices and the transaction is we get solar collectors that help make there be less global climate change or we get batteries that we can put in electric cars and we send them pieces of paper that we print. Think that's a good deal for us or a bad deal for us? Kind of think it's a good deal for us. So, the idea the way you frame the question which is that this is cheating and is there a way to deal with the problem of cheating? I think is largely wrong. It's just like, you know, was it unfair to dentists when fluoride water got invented? It was terrible for the dental business. Terrible. Far fewer cavities. Far less for dentists to do. Was it terrible when electric got invented? Terrible for the candle makers.”
When a country has done terrible things, there are only three options: (1) do nothing, (2) economic sanctions, or (3) military action; given these are the only choices, sanctions are often chosen despite not being optimal
“usually there are three things you can do when a country has done terrible things. You got basically three options. Do nothing. Do economic sanctions. Or have like a shooting war. And given that those are the three choices that are exist, it's not that surprising that pretty often people go back to economic sanctions even if they're not the world's greatest policy cuz they're better than the alternatives.”
Even from a protectionist perspective seeking to create domestic employment, tariffing steel and aluminum is poorly designed because 60 times as many workers are employed using those materials as produce them, so tariffs primarily harm downstream users
“why is a centerpiece of the Trump policy tariffing steel and aluminum when there are 60, 6 0 times as many people who work in industries that use steel and aluminum as there are in the steel and aluminum industries? Why are we making automobile parts a central part of what we're putting tariffs on when the process of producing a vehicle involves it moving back and forth across the Canadian border or the Mexican border half a dozen times?”
Tariff uncertainty induces businesses to delay purchasing and investment decisions, amplifying the economic damage beyond what the tariffs themselves would cause by creating option value in waiting
“there's an additional problem in the way it's being done which is um that you're inducing enormous amounts of uncertainty. And when there's uncertainty people delay hoping that the uncertainty will be resolved. So, you know, tariffs up for 3 days, tariffs down for 6 days. I mean, we, in addition to what the effect of it is, everybody thinks God, I Why should I buy a washing machine till this is all clearer? Why should I buy a car till this is all clearer? Why should I input start make commit to building a certain kind of factory until this is all kind of clearer?”
Science sometimes predicts unpredictability (e.g., efficient markets hypothesis predicts that stock price movements will be unpredictable), making predictability itself a scientifically derived finding
“And so you can explain why things become unpredictable by applying various kinds of social scientific reasoning but that doesn't mean you can't think about how a stock market could be uh better or better organized and less prone to have uh crashes, for example.”
45% of American imports are inputs to American exports, making the relationship between import restrictions and export competitiveness far more complex than simple protectionist reasoning acknowledges
“Here's a number that when I first heard it, even though I've been studying this stuff for a long time, surprised me. 45% of American imports are inputs to American exports.”
The only legitimate argument for tariffs on sensitive technologies might be dependence concerns for national security, but this argues for stockpiling and domestic capacity investment, not for calling cheap goods 'cheating'
“There is an argument you can make that is about dependence perhaps with respect to really sensitive technologies. We don't want to depend on China cuz then they might cut us off and their government might take control and that might be bad for our national security and there are a lot of things to do about that. We should stockpile more things than we do. There are probably some areas like semiconductors where we should be investing in developing domestic capacity. But that's not an argument that because they're making something available cheap it's cheating and you should have a policy.”
Pre-WWI Germany-UK relations provide historical lessons relevant to understanding current China-U.S. dynamics; the Pearl Harbor precedent is cautionary regarding how containment policies can provoke attacks
“People often talk about the relationship between Germany and the UK before the First World War. I think there are things to be learned there. I have been influenced in your direction as I thought about these things by what I've learned about what provoked the Pearl Harbor attack. And that is I think cautionary with respect to some concepts that American policy makers have.”
Applied history—studying how human nature plays out in different contexts to help think about contemporary problems—is a valuable methodology that should be more widely adopted.
“Neil has been a progenitor of a whole movement towards what he calls applied history, which is basically the idea that we should as I as it has reached it has process been processed by me is basically the idea we should study history not just cuz we're kind of curious about what was going on during the Victorian period, but because human nature is kind of a constant and so studying how it plays out in all kinds of different contexts can help us think about the problems that we have today and I think absolutely”
Economic sanctions as a diplomatic tool are limited and ineffective, but they remain in use because the alternatives (doing nothing or military intervention) are worse, creating a comparative advantage for sanctions.
“every time I get too skeptical about economic sanctions as a diplomacy as a tool of diplomacy, I remind myself that usually there are three things you can do when a country has done terrible things. You got basically three options. Do nothing. Do economic sanctions. Or have like a shooting war. And given that those are the three choices that are exist, it's not that surprising that pretty often people go back to economic sanctions even if they're not the world's greatest policy cuz they're better than the alternatives.”
Interdependent economies like those of Germany and UK before WWI are not guaranteed to prevent war, as evidenced by their conflict despite economic ties, contrary to Norman Angell's thesis in 'The Great Illusion.'
“in 1910 Norman Angell, I think it was, wrote The Great Illusion in which he describes how impossible it is for Great Britain and Germany to go to war given their economic interdependence.”
History is essential to answering questions of strategy and human affairs because human nature is constant and history shows how it plays out in different contexts
“human nature is kind of a constant and so studying how it plays out in all kinds of different contexts can help us think about the problems that we have today and I think absolutely and considering the implications of China's rise, there are all kinds of historical episodes that are that are relevant.”
Cannot predict specific future automobile accidents but can improve cars' safety through studying accidents, suggesting predicting something is not the only valid purpose of science
“I can't predict with any great accuracy where and when the next automobile accidents will come. But by studying automobile accidents one can contribute to the development of better brakes and better tires and therefore make cars safer. And so judging the ability to make something better by the ability to uh make uh predictions is I think uh not uh the right uh not the right criteria.”
The impulse to use mathematics, statistics, and data more rather than less is a proper impulse in science, even though some people are overly mechanical in believing statistical results without applying judgment.
“I think the impulse to use mathematics, to use statistics, to use data more rather than less has to be a proper impulse. Do I think some people are overly mechanical about believing some statistical thing they estimated on a computer and not applying judgment? [7:18] Absolutely.”
Interdependence and economic integration do not prevent war, as shown by the failure of Norman Angell's 1910 argument that German-British economic interdependence made war impossible before World War I.
“in 1910 Norman Angell, I think it was, wrote The Great Illusion in which he describes how impossible it is for Great Britain and Germany to go to war given their economic interdependence.”
When fluoride water was invented it was 'terrible' for the dental business because fewer cavities resulted in less work for dentists; when electricity was invented it was 'terrible' for candle makers; these are not examples of unfair competition but of technological progress that benefits consumers
“was it unfair to dentists when fluoride water got invented? It was terrible for the dental business. Terrible. Far fewer cavities. Far less for dentists to do. Was it terrible when electric got invented? Terrible for the candle makers. So, there's this thing that lets people [21:57] buy things for lower cost. Should you say that's terrible?”
Social science and social innovation represents an important domain for people seeking to impact the world, comparable in impact potential to hard science innovations like quantum computing or life sciences
“I'd also just encourage you to think about um social um social science and social innovation. You know, I learned when I was president of Harvard that the Harvard School of Public Health and some kind of social science person of some kind had germinated the whole designated driver idea. [1:08:14] And so I'd encourage you as you think about innovation not to confine it to hard science, but to think about social innovation uh on the Sesame Street uh [1:09:18] designated driver model as a type of area to work in.”
Economies have capacity to work around blockages and reconnect systems in response to disruptions, making sanctions less effective than strategists assume (frog regrowth analogy)
“I think that part of the right interpretation of that fact is that economists are more like frogs than they are like people. If you cut off a frog's tail, it regrows the tail. It's able to reconnect all the systems and that economies probably have more capacity to work around blockages than we imagine.”
Progress with AI has surprised Summers—most developments projected to take 3 years have already happened in the past year, contradicting his belief that progress takes longer than expected
“And I would say the thing that has surprised me most is that in the year and a half that I've been involved with AI that rule I said about things take longer to happen than you think they will hasn't happened. Most of the things that a year ago I was told would take 3 years to happen have already happened in terms of progress with uh these uh models.”
Predatory pricing (selling below cost to drive competitors out, then raising prices) is theoretically possible but has almost never been documented in the history of economics, making it a weak basis for policy.
“There is an issue. It's called uh it's economists call predatory pricing which is the idea that somebody will sell things at a really low price then they'll drive all their competitors out of business and then they'll jack the price way up. And so, if you thought that all of this was sort of setting us up for the kill in terms of total dependence on China that would be a an argument that might lead you to want to have a policy. It's pretty hard to make that case with respect to the vast number of things we're importing. China's only one source of them. If the price did get jacked up, our producers could reenter uh the market. In the entire history of economics there've been almost no examples documented of where predatory pricing actually happened.”
The vast majority of government spending is in the form of checks (transfers) rather than employment, so efficiency-based approaches focused on personnel are fundamentally misdirected.
“the vast majority of what the government does is write checks not employ people. So, the notion that you can get a bunch of efficiency experts and run around like you were a corporation taking costs out is just not the there's some waste to find and it's a good idea to try to find it but it's not a large-scale policy for uh reducing uh the deficit.”
Russia has performed better than expected in face of Western sanctions on its economy, revealing important questions about the effectiveness of economic statecraft
“Given that Russia and its war against Ukraine has done I think better than anyone would have expected in face of the sanctions Western countries have imposed on it. What do you think that reveals about Western economic statecraft?”
If a great power spends more on interest payments than on defense, it 'almost certainly has a problem coming'—a sign of unsustainable fiscal trajectory.
“if you're spending more on interest payments than on defense as a great power then you almost certainly have a problem coming.”
Chinese behavior has become less clearly benign over the past decade as evidenced by rhetoric of leaders, military spending on offensive weapons, interactions with neighbors, and efforts to sabotage the US economy.
“if one looks over the last decade, that it is harder to make the case that China is entirely benign than it would have been a decade ago. You simply look at the speeches of Chinese leaders, both what they say publicly and what it is reported that they have said privately. If you look at the extent of Chinese military spending and build-up of what of weapons that are really only with offensive use. If you look at their interactions with uh neighboring uh countries, if you look at efforts in various ways to sabotage and subvert uh things that are going on in our economy,”
The Pearl Harbor attack is a cautionary historical episode that informs thinking about current China relations, suggesting US security policies should avoid patterns that led to that conflict.
“I have been influenced in your direction as I thought about these things by what I've learned about what provoked the Pearl Harbor attack. And that is I think cautionary with respect to some concepts that American policy makers have.”
University of Texas at Austin's experiment with test-score-only admissions is one example of institutional experimentation, but many other experiments should be pursued in areas like AI pedagogy and moving away from teacher-centered classrooms.
“my instincts would be that experimental institution should be paying a lot of attention to AI and the use of AI in pedagogy and models that move away from uh teacher in front of room, students in back of room. Um I think there should be a lot of experimentation uh with that, whether this is the right venue and what form those experiments should take, uh I don't know.”
Designating drivers is a successful social innovation that may have saved as many lives as substantial biological research, and human augmentation from life sciences deserves more attention given AI excitement.
“the Harvard School of Public Health and some kind of social science person of some kind had germinated the whole designated driver idea. And my guess is that the designated driver idea saved as many lives as a huge amount of biological uh research.”
Sesame Street's impact may have justified 50 years of the Harvard Education School despite that school having aspects the speaker was skeptical of, demonstrating how social innovation can have outsized impact.
“The Harvard Education School did incubate Sesame Street. And I think you could make a case that the good that Sesame Street did was enough to justify 50 years of the Harvard Education School No. No. No. even if you didn't like some of the other things that happened in uh the Harvard uh Education School.”
The ability of geologists and meteorologists to predict earthquakes and hurricanes better than anyone does not require they predict exactly when the next one will occur, so the standard of predictability should not be the test of whether economics is in decline.
“Geologists and seismologists know more about earthquakes than anybody else does, but they can't predict when the next one is uh going to come. Meteorologists understand hurricanes better than anybody else does, but they don't know exactly when or where the next uh hurricane is going to come.”
Economic forecasts have trended downward in root mean square error through time, meaning economists have better understanding now than they used to.
“If you look at the quality of economic forecasts as measured by to use the fancy statistical term root mean square error, basically how big's your average mistake, that's trended downwards through time. So, by that standard, economists have a better understanding than uh they used to.”
The influence of economists has expanded dramatically since 1970: they now dominate cost-benefit analysis at EPA, antitrust analysis at Justice Department, and central bank leadership globally, which indicates the discipline is not in decline but has become more successful.
“when I started out, economists were mostly people who tried to guess whether there was going to be a recession or not. And now the EPA is filled with economists doing cost-benefit analyses of regulation. The Justice Department is filled with economists doing analyses of what will promote competition in the context of uh antitrust policy. The Treasury and the Fed used to all be people who were ex-bankers. Now there large numbers of people with the discipline of uh economics. If you look at what's happened in the cutting edge of the other social science disciplines, political science, sociology, basically a lot of it is importing techniques and approaches from uh economics. If you look at who gets appointed central bank heads around the world, it's in much larger number economists than it ever used to be.”
If protectionist policy's goal were to benefit domestic producers, tariffs on steel and aluminum are badly designed because there are 60 times as many workers in industries that use steel and aluminum as in the steel and aluminum industries themselves, so the policy harms far more workers than it helps.
“why is a centerpiece of the Trump policy tariffing steel and aluminum when there are 60, 6 0 times as many people who work in industries that use steel and aluminum as there are in the steel and aluminum industries? Why are we making automobile parts a central part of what we're putting tariffs on when the process of producing a vehicle involves it moving back and forth across the Canadian border or the Mexican border half a dozen times?”
Even if every federal employee were fired, the deficit would reduce from 6% to only 5.4% of GDP, showing that the problem cannot be solved through spending cuts on federal employment alone.
“I'd tell him that even if we fired every federal employee it would reduce the deficit perhaps from six to about 5.4. That the vast majority of what the government does is write checks not employ people.”
The IRS budget has been cut by approximately half relative to the scale of taxes since 2010, explaining the enforcement gaps.
“That's related to the fact that the IRS budget has been cut relative to the scale of taxes by about half since 2010.”
Economic history from the beginning of time through Pericles involved zero growth, then 2-3 basis points/year (2-3% per century) until ~1500, then jumped to 20 basis points/year (20% per century), then jumped again to 1-2% per year with the Industrial Revolution.
“here's an economic history of humanity in 90 seconds. From the beginning of time until roughly the time of Pericles there was no growth. Sometimes there was a lot of food and people ate better, sometimes there was less food and people ate worse but basically there was no progress. From the time of Pericles till kind of after the Renaissance maybe 1500 there was growth at about two or three basis points a year. That means it was two or three percent a century. Then nobody can measure these things exactly, but around 1500 there was a seismic change. The growth rate went from two basis points a year, 2% a century to 20 basis points a year 20% a century. And so it was very clear that in England in 1800 things were better than they had been in 1500. But the growth wasn't fast enough that anybody could see between the beginning of their life and their end of their life that standards of living had systematically improved. And then there was this extraordinary epic event the Industrial Revolution. And for the first time in human history you could expect that living standards would be better at the end of your life than they were at the beginning of your life and growth rose, depends on exactly the period exactly the place to 1 and 1/2 or 2% a year.”
Strategic dilemma with China involves hedging against possibility of aggressive hegemonic behavior without appearing provocative, which is a difficult task of statecraft requiring balance between deterrence and reassurance.
“the strategic dilemma with respect to China, and it's a very hard thing to manage, is how do you hedge against the possibility that they are going to aggressively seek hegemony in ways that deeply threaten us without your effort to hedge seeming a provocation to them to build up more strength because they fear you.”
In democracies, fear does the work of reason, so serious fiscal adjustment is likely to occur only after interest rates spike, not through rational advance planning.
“it's been said that in democracy fear does the work of reason. And my guess is sooner or later we'll do something about the deficit, but it will take spiking interest rates before we do.”
Trade is a self-imposed supply shock—tariffs make all the things we buy more expensive, raising the price level, which makes people poorer, reduces demand, and increases unemployment by combining inflation and unemployment effects
“I think the simplest way to think about this is as a self-imposed supply shock. We're going to take all the things we buy and we're going to make them more expensive. And when we do that the price level's going to be higher. That's bad. People are going to have to pay those higher prices. So, the same income is going to go less far. So, they're going to be poorer. And they're going to be able to buy less. And that's going to mean there's less demand. And that's going to make unemployment go up. So, doing something that increases both inflation and unemployment kind of seems like it's not a very good idea.”
Other social science disciplines (political science, sociology) are increasingly importing techniques and approaches from economics rather than the reverse, indicating economics' intellectual dominance among social sciences
“If you look at what's happened in the cutting edge of the other social science disciplines, political science, sociology, basically a lot of it is importing techniques and approaches from uh economics.”
A problem at the Federal Reserve is that people doing the same job for 15-20 years become attached to how they've done things before and fail to update methods to reflect how the world has changed
“One of the problems in any organization with having people do the same job for 15 or 20 years is that they tend to get a little attached to the way they've done things before and not update uh things in respect to the ways the world has changed. Is there that kind of tendency at the Fed. So, yeah, I think I was right to criticize the Fed's model”
If Chinese prices are jacked up after driving competitors out, U.S. producers could re-enter the market, preventing monopoly extraction because there are multiple sources of most imports
“It's pretty hard to make that case with respect to the vast number of things we're importing. China's only one source of them. If the price did get jacked up, our producers could reenter uh the market.”
The U.S. cannot indefinitely run fiscal deficits above 6% of GDP without eventually facing real trouble; when spending on interest payments exceeds defense spending, a great power almost certainly has a problem coming
“The United States cannot run a fiscal deficit north of 6% of GDP indefinitely. And that leads at some point uh to to real trouble. The way I've tried to formulate it is that if you're spending more on interest payments than on defense as a great power then you almost certainly have a problem coming.”
Even firing every federal employee would only reduce the deficit from 6% to 5.4% of GDP, proving that cost-cutting in government employment is not a large-scale solution to deficit problems
“I'd tell him that even if we fired every federal employee it would reduce the deficit perhaps from six to about 5.4. That the vast majority of what the government does is write checks not employ people.”
Over the next 10 years, $8 trillion in taxes that are legally owed will go unpaid, representing a massive enforcement gap
“The projections are that over the next 10 years there will be 8 trillion that's trillion with a T R dollars in taxes that are owed but not paid.”
The IRS budget has been cut by approximately 50% relative to the scale of taxes since 2010, creating the enforcement capacity crisis
“That's related to the fact that the IRS budget has been cut relative to the scale of taxes by about half since 2010.”
The first fiscal policy priority should be enforcing existing tax law, which could raise hundreds of billions annually without new legislation
“So, the first thing we should do is we should enforce the tax law and we could raise hundreds of billions of dollars a year just by enforcing the law we have.”
The U.S. has shifted from 4-5 working-age people per retiree to 2.5 per retiree, requiring either substantially higher tax rates or significantly reduced benefits to maintain current benefit levels relative to payroll
“it used to be that we had four or five working-age people for every person over 65. And now we have two and a half people for every person over 65. And so, if you want to provide the same benefits that we did before you need to have relative to the payroll that working people are earning you've got to have a much higher tax rate than you did before.”
The quality of economic forecasts, as measured by root mean square error (average magnitude of mistakes), has trended downwards through time, indicating economists have better understanding than they used to
“If you look at the quality of economic forecasts as measured by to use the fancy statistical term root mean square error, basically how big's your average mistake, that's trended downwards through time. So, by that standard, economists have a better understanding than uh they used to.”
Liberals have opposed value-added tax (VAT) for 35+ years because they view it as regressive; conservatives oppose it because they view it as invisible/enabling larger government; both must flip positions for VAT to be politically viable
“I think look, I've said something 35 years ago and it was true then and it's I think still true today. Which is that liberals oppose a value-added tax, which is basically a kind of national sales tax because they think it's regressive. And conservatives oppose it because they think it's invisible and therefore a money machine that will enable a larger government. And we'll get it only when liberals decide it's a money machine for government and conservatives decide it's regressive.”
Even if government does everything right, higher tax revenues will be necessary because of aging population, rising service costs relative to goods prices, and higher debt levels
“even if everybody does everything right because more of the population is aged because what government does is provide services and the price of services goes up relative to the price of physical uh uh goods because we've got a bunch more debt than we used to we just to even if we don't do anything new to avoid a big deficit we have to find a way to raise more revenue.”
Higher interest rates due to deficits will make it harder for young people to buy houses and for businesses to make new investments.
“It's going to be harder for you guys to buy houses cuz of higher interest rates cuz of all the debt that we have. It's going to be harder for businesses to make new investments because of the debt that uh we have.”
The world undertakes too little experimentation overall because individual experimenters don't capture the full benefits of success (which are distributed) and don't fully avoid the costs of others' failed experiments
“there's just too little experimentation in the world in general and that's why this the kind of experimentation that you represent is so positive. There's just too little experimentation in the world in general and that's why this the kind of experimentation that you represent is so positive because the person who does the experiment doesn't get all the benefit from all the people who copy it and doesn't get all the benefits from the people who don't have to try it again after it fails.”
The designated driver social innovation, incubated at Harvard School of Public Health, likely saved as many lives as massive amounts of biological research
“I learned when I was president of Harvard that the Harvard School of Public Health and some kind of social science person of some kind had germinated the whole designated driver idea. And my guess is that the designated driver idea saved as many lives as a huge amount of biological uh research.”
History is essential for understanding problems that require human affairs knowledge; there is no substitute for historical analysis when dealing with complex strategic questions.
“I don't know what we can learn from in human affairs other than history. So, I very much share uh your your sort of trying to provoke me into doing an advertisement for you.”
Tariffs function as a self-imposed supply shock: they make all things Americans buy more expensive, raising the price level and making people poorer in real terms, reducing demand and increasing unemployment.
“I think the simplest way to think about this is as a self-imposed supply shock. We're going to take all the things we buy and we're going to make them more expensive. And when we do that the price level's going to be higher. That's bad. People are going to have to pay those higher prices. So, the same income is going to go less far. So, they're going to be poorer. And they're going to be able to buy less. And that's going to mean there's less demand. And that's going to make unemployment go up. So, doing something that increases both inflation and unemployment kind of seems like it's not a very good idea.”
The right way to think about economics is in terms of promoting exchange and division of labor and gains from trade, not in terms of making domestic producers more competitive through tariffs.
“I think the right way to think about economics is in terms of promoting exchange and division of labor and gains from trade. But even if one thought the whole idea was about making yourself more competitive somehow, these policies are terribly designed with respect to that objective.”
The fundamental fiscal challenge is demographic: the ratio of working-age people to people over 65 has fallen from 4-5 to 2.5, meaning taxes must increase relative to the past to maintain the same benefit levels.
“it used to be that we had four or five working-age people for every person over 65. And now we have two and a half people for every person over 65. And so, if you want to provide the same benefits that we did before you need to have relative to the payroll that working people are earning you've got to have a much higher tax rate than you did before.”
Returning to the gold standard would be unwise because it would make monetary arrangements hostage to the 'accident of how much gold miners find and decide to get out of the ground,' which is extremely unlikely to produce stable monetary conditions.
“having it depend on the accident of how much gold gold miners find and decide to get out of the ground seems a an approach that's extremely unlikely to generate monetary arrangements that are conducive to a well-functioning economy.”
When extreme values are fed into the Federal Reserve's model (particularly on fiscal policy), its results are bizarrely clustered together in ways that seem intuitively wrong, indicating a problem with the model.
“when you fed into the model [4:37] uh very extreme values, I think on fiscal policy, its results were bizarrely uh bizarrely clustered together in ways that seemed intuitively wrong.”
Two major problems limited Russian sanctions effectiveness: (1) oil was the main revenue source and gasoline prices are politically salient, so governments couldn't effectively cut off oil sales, and (2) Europe has a stagnant economy and continued selling to Russia despite sanctions attempts
“We had a big problem which was we had two big problems. One is that their big revenue source was the thing that we most wanted. That being oil. Because gasoline prices are hugely salient politically in the United States and around the world. So how good a job were we really going to do of cutting off their exports? Answer, not very. And that we had a whole continent with a relatively stagnant economy that really liked selling things to Russia. And so how good a job were we going to do of stopping selling things to Russia?”
Artificial intelligence may lead to a discontinuity in growth similar to the Industrial Revolution, potentially increasing growth rates by 10x from current levels over coming decades
“I think it's possible that artificial intelligence will lead to another discontinuity in the growth process that will lead to very substantial acceleration in the rate of growth.”
Balancing the federal budget without higher taxes or lower benefits for the aged is mathematically impossible, so any claim that the budget can be balanced without one of these two things is false
“When you have to decide to have one of those two things but if you say we can balance the budget without having one of those two things you're not speaking the truth.”
Economists have massively expanded their influence in government from mostly doing recession prediction in the 1970s-80s to now filling EPA, Justice Department, Treasury, and Federal Reserve with economists doing cost-benefit analysis, antitrust analysis, and monetary policy
“when I started out, economists were mostly people who tried to guess whether there was going to be a recession or not. And now the EPA is filled with economists doing cost-benefit analyses of regulation. The Justice Department is filled with economists doing analyses of what will promote competition in the context of uh antitrust policy. The Treasury and the Fed used to all be people who were ex-bankers. Now there large numbers of people with the discipline of uh economics.”
There is no constituency in the business community for tariffs designed in the current way, suggesting the policy is driven by something other than economic logic or business interests.
“There's no constituency in the business community for doing this in this way. There's an additional problem in the way it's being done which is um that you're inducing enormous amounts of uncertainty.”
Projected federal tax collections will fall short of taxes owed by $8 trillion over the next 10 years, representing a massive tax enforcement gap that could be recovered through better IRS enforcement without raising tax rates.
“The projections are that over the next 10 years there will be 8 trillion that's trillion with a T R dollars in taxes that are owed but not paid.”
In 2018, 100 individuals with incomes over $10 million did not file tax returns at all, and the IRS statute of limitations ran without any audit attempt, indicating severe enforcement capacity gaps.
“In 2018, which is one of the most recent years for which we have uh the data there were 100 people who paid who had incomes over $10 million dollars who didn't file any tax return at all. And the statute of limitations for 3 years ran without anybody at the IRS noticing and trying to audit them.”
Artificial intelligence will likely lead to another discontinuity in the growth process and very substantial acceleration in economic growth, driven by three mechanisms: productivity from AI capabilities, compressed scientific progress, and improved coordination and scaling.
“I think it's possible that artificial intelligence will lead to another discontinuity in the growth process that will lead to very substantial acceleration in the rate of growth. Part of that is because of what everybody focuses on which is that gosh, an AI can do this and an AI can do that and it can do this job and it can do that job and it can do the other job, so isn't that going to make things much more productive? That's aspect one. Aspect two is it might mean we have 50 years of scientific progress in seven. And ultimately it's scientific progress that drives growth. And aspect three is it permits a degree of coordination that and replication and scaling that isn't possible with people.”
President Biden's statement that sanctions would reduce the ruble 'to rubble' failed, and sanctions were largely ineffective, reinforcing skepticism about sanctions as a major diplomatic tool
“President Biden famously said we were going to reduce the ruble to rubble. We didn't. And why the sanctions were so ineffective I think is a hugely important question for study and it reinforces my skepticism about economic sanctions as a huge tool of diplomacy”
This is the second worst performance of the stock market in the first two months to end of March since inauguration day among 20 presidential terms since 1948, indicating massive market distress from tariff policy
“I was just looking at some data. There have been 20 presidential terms since 1948. This is the second worst in terms of the performance of the stock market in uh two and a two months to the end of March since inauguration day.”
Consensus forecasters at the beginning of January estimated 10-15% probability of recession; by the time of speaking they estimate 40-50%, representing a massive shift caused by tariff policy
“If you ask all the consensus forecasters at the beginning of January what the odds were on a recession this year it'd be they would have said 10 to 15%. You ask people today, they'd kind of say 40 to 50%.”
Debt crowding out will make it harder for households to buy houses and harder for businesses to invest due to higher interest rates caused by government debt, creating significant economic drag.
“It's going to be harder for you guys to buy houses cuz of higher interest rates cuz of all the debt that we have. It's going to be harder for businesses to make new investments because of the debt that uh we have.”
Science sometimes predicts unpredictability; efficient markets theory predicts that stock price movements are unpredictable because all predictable patterns have been arbitraged away.
“science predicts unpredictability. If you've taken I don't know how many people here have taken a finance course but one of the two or three most important ideas you'd be exposed to if you took a finance course is what could be called the efficient markets or the martingale theory which in essence is the following idea. Think about IBM stock. If there was a good reason to think that it would go up people would buy it and then it would have already gone up. There was a good reason to think it would go down, people would sell it. And so almost the fundamental logic of the situation is that the price goes to a level where there's equipoise between opinion dollar weighted that it's going to go up and that it's going to go down and it follows that the movements will be completely unpredictable.”
China has cheated in multiple ways since admission to WTO in 2001 by subsidizing industries, engaging in unfair competition and dumping for nearly a quarter century, but tariffs are not the appropriate policy response.
“after it was admitted to the World Trade Organization, China cheated in multiple ways. China continues to subsidize its industries in all kinds of ways. It engages in unfair competition. It does things that look to me, at least, like dumping. And it's done this uh for a quarter of a century, nearly.”
The United States cannot run a fiscal deficit above 6% of GDP indefinitely; this represents a fundamental fiscal constraint that eventually requires policy adjustment.
“The United States cannot run a fiscal deficit north of 6% of GDP indefinitely. And that leads at some point uh to to real trouble.”
Strategic bombing of Germany in WWII caused remarkably little damage to the German economy despite theoretical expectations, suggesting blockades and sanctions are poorly understood
“the first significant thing he did in his career was he was in charge of the review of Allied strategic bombing of Germany after during World War II. And the Allies dropped a ton of bombs on Germany and they were doing what they thought was strategic bombing. There were all kinds of good ideas like if we there are only two factories that make ball bearings and if we knock out the two factories, then they won't have any ball bearings and then all their stuff won't work. And what appears to have been the case is that our strategic bombing did remarkably little damage to the functioning of the German economy.”
In a democracy, fear does the work of reason, so spiking interest rates will eventually force action on the deficit even though fear would be preferable.
“it's been said that in democracy fear does the work of reason. And my guess is sooner or later we'll do something about the deficit, but it will take spiking interest rates before we do.”
Marxists have been critiquing orthodox economics for decades but have produced no improvement in prediction or explanatory power, and modern monetary theory represents a challenge to orthodoxy that hasn't gotten anywhere
“You know, the Marxists have been yelling for decades, I think, to know to produce no improvement in prediction or explanatory power. The so-called modern monetary theorists represent a challenge to the orthodoxy that I don't think has gotten anywhere.”
Cutting defense budgets in the face of Chinese military buildup would be a 'Neville Chamberlain level error'—a catastrophic strategic failure.
“The idea that has been brooded several times though not implemented by the Trump administration that in the face of what's happening right now, we should substantially cut our defense budget seems to me to be a Neville Chamberlain level error. That's a very big error.”
Evidence that China has become less benign over the past decade includes: Chinese leaders' public and private speeches about expansion, expanded military spending on offensive weapons, military interactions with neighboring countries, and efforts to sabotage and subvert the US economy.
“You simply look at the speeches of Chinese leaders, both what they say publicly and what it is reported that they have said privately. [47:42] If you look at the extent of Chinese military spending and build-up of what of weapons that are really only with offensive use. If you look at their interactions with uh neighboring uh countries, if you look at efforts in various ways to sabotage and subvert uh things that are going on in our [48:14] economy, it is harder to see them as entirely benign and to think that it's responsible to not hedge than was the case a decade ago.”
Predatory pricing—where firms sell at low prices to drive competitors out then raise prices—has almost no documented examples in the entire history of economics
“In the entire history of economics there've been almost no examples documented of where predatory pricing actually happened.”
The speaker hasn't studied quantum computing deeply enough to have nuanced opinions on it, but believes the life sciences and human augmentation are receiving less attention than they should relative to current AI excitement
“I don't have nuanced intelligent opinions on different aspects of quantum computing. I suspect that with all the excitement about AI at this instant, the various revolutions in the life [1:07:13] sciences and their capacity for human augmentation and all the ethical implications that follow from that are probably receiving less attention uh than uh they should.”
The US should apply less economic pressure to China except regarding national security technologies, and should substantially increase military capacity in the Pacific region rather than reducing defense spending.
“My instincts would be towards significantly less economic suppression of China except in with respect to national security technologies and significantly more strengthening of our military capacities with respect to the Pacific region than appears to be the policy that we are pursuing.”
The Industrial Revolution was extraordinary because it was the first time in human history that people could expect living standards to improve within their own lifetime, marking a fundamental transformation.
“And then there was this extraordinary epic event the Industrial Revolution. And for the first time in human history you could expect that living standards would be better at the end of your life than they were at the beginning of your life”
The right way to think about economics is in terms of promoting exchange, division of labor, and gains from trade, not in terms of making producers more competitive through tariffs.
“I don't think that's the right way to think about economics. I think the right way to think about economics is in terms of promoting exchange and division of labor and gains from trade.”
The speaker and colleagues (Phil Zelikow, Bob Zelik) have advocated for tougher measures on frozen Russian reserves, but Europeans have resisted confiscation; their attitude may change given European fiscal pressures and the need for funding.
“you've been consistently arguing that there should be much tougher action on those frozen reserves than has so far been been envisioned. Do you think that might change under the new dispensation? The Europeans were the people who resisted confiscation or some other tougher measures. I I have a sense that their attitude has changed quite significantly since January the 20th. Do you think there's going to be more action on on that front? I am uh guardedly optimistic that there may be that there may be more action cuz Europe's going to have to come up with a lot more money and while they didn't much like seizing the Russian assets as between seizing the Russian assets and taxing their own citizens, seizing the Russian assets is likely to look a bit better.”
In 90 seconds of economic history: zero growth until Pericles; 2% per century growth from Pericles to ~1500; then 10x acceleration to 20% per century; then Industrial Revolution with 1.5-2% annual growth, allowing lifetime improvement in living standards
“From the beginning of time until roughly the time of Pericles there was no growth. From the time of Pericles till kind of after the Renaissance maybe 1500 there was growth at about two or three basis points a year. That means it was two or three percent a century. Then nobody can measure these things exactly, but around 1500 there was a seismic change. The growth rate went from two basis points a year, 2% a century to 20 basis points a year 20% a century. And then there was this extraordinary epic event the Industrial Revolution. And for the first time in human history you could expect that living standards would be better at the end of your life than they were at the beginning of your life and growth rose, depends on exactly the period exactly the place to 1 and 1/2 or 2% a year.”
The most surprising aspect of AI progress is that most advances happen faster than predicted; things said to take 3 years happened within a year, violating the speaker's prior rule that 'things take longer than you think they will'.
“the thing that has surprised me most is that in the year and a half that I've been involved with AI that rule I said about things take longer to happen than you think they will hasn't happened. Most of the things that a year ago I was told would take 3 years to happen have already happened in terms of progress with uh these uh models.”
The Fed is uncertain about policy direction and about whether interest rate increases or decreases are appropriate given conflicting inflation pressures from tariffs
“the Fed is as uncertain as every place else is. Uh you would have said a little while ago that the issues were mostly about [39:55] uh reducing interest rates. Now that we're getting all this inflation pressure from the tariffs, I don't think it's 100% clear which way interest rates uh will uh which way interest rates will move.”
The neutral interest rate is substantially higher than market consensus believes, so interest rates are unlikely to fall significantly unless a recession occurs.
“My view is that the neutral interest rate is much higher than the market consensus thinks. And so, I don't think interest rates will come down substantially unless we have a recession.”
Cutting US defense spending is a 'Neville Chamberlain level error' when facing Chinese military buildup and potential Taiwan conflict, indicating strong opposition to defense cuts in current environment.
“the idea that has been brooded several times though not implemented by the Trump administration that in the face of what's happening right now, we should substantially cut our defense budget seems to me to be a Neville Chamberlain level error. That's a very big error.”
There is a reasonably good chance (in the neighborhood of 50%) that the US will have a recession.
“But I think there's a reasonably good chance that we'll have a recession. Reasonably good is still what, 50%? In the neighborhood of 50%.”
The neutral interest rate is much higher than market consensus believes, so interest rates are unlikely to decline substantially unless there is a recession
“My view is that the neutral interest rate is much higher than the market consensus thinks. And so, I don't think interest rates will come down substantially unless we have a recession.”
The Fed is uncertain like everywhere else; issues that were previously about reducing interest rates are now complicated by inflation pressure from tariffs, making the direction of interest rate movement unclear.
“Look, I think the Fed is as uncertain as every place else is. Uh you would have said a little while ago that the issues were mostly about [39:55] uh reducing interest rates. Now that we're getting all this inflation pressure from the tariffs, I don't think it's 100% clear which way interest rates uh will uh which way interest rates will move.”
Even if one accepted protectionist philosophy that claims we cannot find enough work for everyone in the country and so restricting imports would put more people to work, the premise is wrong because the unemployment rate is 4%, indicating we can find enough work for everybody.
“That has a certain logic to it. I think [12:33] it's basically more wrong than right because the premise is wrong. We can find enough work for everybody to do. The unemployment rate is 4%.”
Even if one believes in protectionist philosophy that restricting imports increases domestic employment, the unemployment rate is currently 4%, so the premise that 'we can't find enough work' is wrong, undermining protectionist justification.
“That has a certain logic to it. I think it's basically more wrong than right because the premise is wrong. We can find enough work for everybody to do. The unemployment rate is 4%.”
Mark Zuckerberg has a net worth of $200 billion but pays far less than $1 billion in taxes annually, so a tax rate increase on the wealthy would represent a minimal burden compared to the magnitude of wealth.
“take like a economical member of uh the generation that I would imagine is between us, Mark Zuckerberg. He's got $200 billion. If his tax rate his tax rate went up a little bit and $200 billion pays pays Well, who knows exactly what it pays, but let's say it pays 6%. That's $12 billion a year. I promise you he's not paying $1 billion a year in taxes. Um he paid more taxes, be okay with me.”
Long-term interest rates were expected to rise substantially but have not in the past 3 months due to recession concerns overshadowing deficit concerns, causing the speaker to have less confidence in previous forecasts.
“I have kind of thought that long-term interest rates were likely to rise substantially. I've been wrong for the last 3 months. I think that is because it's looking more likely that we're going to have a recession and that's a bigger deal than what's happening with uh the deficit. But I have been wrong, so it gives me a little bit more humility in making forecasts”
Stock market performance in the first two months of the Trump administration ranks as the second worst among the 20 presidential terms since 1948, indicating market concern about economic policy.
“There have been 20 presidential terms since 1948. This is the second worst in terms of the performance of the stock market in uh two and a two months to the end of March since inauguration day.”
Consensus forecasters estimated 10-15% probability of recession at the beginning of January, but by the time of speaking they estimate 40-50%, a massive shift indicating tariff policy has substantially increased recession risk.
“If you ask all the consensus forecasters at the beginning of January what the odds were on a recession this year it'd be they would have said 10 to 15%. You ask people today, they'd kind of say 40 to 50%.”
The Biden administration attempted to address tax enforcement deficits, but the Trump administration has reversed those improvements and made additional cuts, potentially causing a deterioration in tax compliance.
“The Biden administration was trying to fix all of that. The Trump administration has taken away what the Biden administration did and then done a bunch of other cutting besides. And that's why there are people at the IRS, and we don't yet know whether they're right, who think that we're going to have a they think that tax collections are running way behind schedule cuz people are deciding that nobody's going to do anything to them if they don't pay taxes.”
A value-added tax would likely be adopted if liberals recognized it enables larger government (contradicting their progressivity concern) and conservatives recognized it is regressive (contradicting their invisibility concern).
“I've said something 35 years ago and it was true then and it's I think still true today. Which is that liberals oppose a value-added tax, which is basically a kind of national sales tax because they think it's regressive. And conservatives oppose it because they think it's invisible and therefore a money machine that will enable a larger government. And we'll get it only when liberals decide it's a money machine for government and conservatives decide it's regressive.”
Summers' instincts favor substantially less economic suppression of China except for national security technologies, combined with significantly more military strengthening in the Pacific region.
“My instincts would be towards significantly less economic suppression of China except in with respect to national security technologies and significantly more strengthening of our military capacities with respect to the Pacific region than appears to be the policy that we are pursuing.”
University of Texas at Austin's free tuition model is not unfair competition to other universities; it is a good thing that more universities could emulate
“Do you think that if the University of Texas or Harvard or Rice called this an unfair practice because it put them at a competitive disadvantage? That that would be a reasonable thing? Or do you think that it's a good thing that you're able to provide free college educations? It would be even better if more people could provide free college educations? That's I kind of have the latter view.”
Social science and social innovation should not be confined to hard science; social innovations modeled on Sesame Street and designated driver have transformative potential
“So I'd encourage you as you think about innovation not to confine it to hard science, but to think about social innovation uh on the Sesame Street uh designated driver uh model as a type of area to work in.”
The speaker got inflation predictions right while others did not, which is more a problem of political incentives of forecasters than a fundamental intellectual failure of the economics discipline.
“Yeah, I did get it right on uh inflation, and I think it's a bit of a puzzle why others didn't. I think it probably has more to do with the political incentives of people making forecasts than it does some fundamental intellectual failure of uh the discipline.”
The Fed model produces bizarrely clustered results when fed extreme fiscal policy values, which reveals a deep problem with how the model represents economic relationships, though this is a critique of a specific tool rather than a fundamental critique of the economics discipline.
“when you fed into the model uh very extreme values, I think on fiscal policy, its results were bizarrely uh bizarrely clustered together in ways that seemed intuitively wrong.”
Trump administration might be stumbling into a war with China through aggressive tariff policy that damages the Chinese economy without US adequate military buildup, potentially provoking security channel retaliation over Taiwan.
“Do you worry that we are stumbling through an aggressive trade policy towards a war with China?”
No U.S. president has stated they might have a recession as part of accepting policy costs; instead presidents typically assert recessions won't happen to boost confidence, making the Trump administration's casual recession acceptance unprecedented
“I've been doing this stuff for 40 years. I've never heard a president say, 'Well, we might have a recession. It'll be okay.' I've heard presidents say we won't have a recession when they're probably wrong and they're trying to add confidence to the situation. But saying like, 'We might have a recession. It'll be okay.' I've never heard uh that particular play before.”
Summers should pay attention to life sciences' capacity for human augmentation and its ethical implications as likely to receive less attention than AI currently does
“I suspect that with all the excitement about AI at this instant, the various revolutions in the life sciences and their capacity for human augmentation and all the ethical implications that follow from that are probably receiving less attention uh than uh they should. And so I think that that would be a thing That would be an area that uh I would be paying uh attention to.”
Some people are overly mechanical about believing statistical results from a computer without applying judgment, though this is something the speaker tried to avoid as a policymaker, and there is a particular model used more by Fed staff than Fed policymakers that the speaker does not think is a good model
“Do I think some people are overly mechanical about believing some statistical thing they estimated on a computer and not applying judgment? [7:18] Absolutely. Is that something I've tried to avoid when I was a policymaker? Absolutely. Is there a particular tool that is used more by the Fed staff than by the Fed uh policymakers that isn't the tool that I would use if I was the Fed staff? Yeah. They They use a model that I don't think is a very good uh model.”
The speaker does not favor cutting Social Security benefits because the maximum possible benefit (~$45,000/year) is modest, making benefit cuts inequitable compared to tax increases on higher earners who have been fortunate.
“My personal view would be when the maximum social security benefit in the country that a person can get if they've worked for their whole life and earned about $125,000 their maximum possible social security benefit that they can get is $45,000. I don't tend to find it all that attractive for people like me and you who've been fortunate to talk about how we need to find the courage to bite the bullet to cut the benefits of a bunch of people who are at most going to get $45,000.”
Kissinger's formulation of policy options as: one, surrender; two, what bureaucracy recommends; three, World War III—is an effective way to present foreign policy choices to the president
“the Kissinger point that this shows that you've been in government that recommendations to the president should always be three. In in the case of a foreign [57:12] policy problem, the choices are one, surrender. Two, what the bureaucracy recommends. Three, World War III. And this is the way that you should always present any decision to the president.”
Summers does not advocate cutting social security benefits for lower-income retirees because the maximum possible benefit ($45,000) is modest; wealthy individuals should shoulder the adjustment burden instead
“My personal view would be when the maximum social security benefit in the country that a person can get if they've worked for their whole life and earned about $125,000 their maximum possible social security benefit that they can get is $45,000. I don't tend to find it all that attractive for people like me and you who've been fortunate to talk about how we need to find the courage to bite the bullet to cut the benefits of a bunch of people who are at most going to get $45,000.”
The purpose of social sciences is to understand how social processes operate both for intellectual satisfaction and to improve design and outcomes of those processes
“It is to understand how social processes operate both because that's intellectually satisfying and because with a better understanding of how social processes uh operate, one has a better prospect of contributing to their design and contributing to the choices that people operating in social settings make so that there will be better uh so that there will be better outcomes.”
Political incentives of forecast-makers rather than fundamental intellectual failure of the economics discipline explain why many economists missed the 2021 inflation forecast that Summers got right
“Yeah, I did get it right on uh inflation, and I think it's a bit of a puzzle why others didn't. I think it probably has more to do with the political incentives of people making forecasts than it does some fundamental intellectual failure of uh the discipline.”
There is a 'reasonably good chance' of a recession, which the speaker characterizes as 'in the neighborhood of 50%' probability
“But I think there's a reasonably good chance that we'll have a recession. Reasonably good is still what, 50%? In the neighborhood of 50%.”
China's rise has improved quality of life for a billion people in 50 years and does not necessarily pose a security threat, suggesting US concerns may be overstated.
“China has raised the quality of life immensely for a billion people in just 50 years, and they don't really seem to pose a threat to like us that much at all and don't seem to in the future.”
Critiques and improvements to economics have come more from within the discipline than from outside it; Marxist and Modern Monetary Theory critiques have not produced improvements in predictive or explanatory power.
“It seems to me most of the interesting and significant criticisms and improvements that have come to economics have come more from within economics than from the outside. You know, the Marxists have been yelling for decades, I think, to know to produce no improvement in prediction or explanatory power. The so-called modern monetary theorists represent a challenge to the orthodoxy that I don't think has gotten anywhere.”
AI is likely to be the first story in future historical accounts of this period (more important than Trump or Xi), as it will be transformative of how people live and how things are produced.
“When your intellectual great-grandchild writes the history of this period in uh 22 21 25 my guess is that stuff about Donald Trump will be the second or third story and stuff about Xi will be the second or third story and the first story will be that we had this transformative change in the ways in which um the things ways in which people lived and the ways in which the things they enjoyed were produced.”
Young people will benefit from AI-enabled improvements in living standards that will far exceed the modest tax increase burden from supporting an aging population, making generational fairness arguments in favor of current young people overstated.
“If I'm even close to right, if 6% of what I said about AI is right, your generation is going to have opportunities to work with far more ease, to enjoy opportunities and forms of entertainment and forms of spending that go beyond anything that my generation uh enjoyed, that the notion that the tax rate would be a little bit higher for the most fortunate members of your generation than it was of my generation.”
The Trump administration is implementing tariffs that represent a return to the average US tariff rate of 1937 if implemented fully, constituting one of the biggest policy-induced economic shocks in recent lifetimes.
“we're on the eve of one of the biggest policy induced shocks that there have been, I suspect, in our lifetimes in that the president intends tomorrow to increase tariffs on a substantial number of US trade partners, some on the reciprocal basis, and more perhaps to come on the sectoral basis. I've seen estimates that make it a return to the average US tariff rate of 1937 if he goes through with all of the things that he's talked about.”
The Biden administration invested in IRS enforcement, but the Trump administration has cut it further and reversed Biden's improvements, causing IRS staff to believe tax collections will fall as people recognize there's no enforcement
“The Biden administration was trying to fix all of that. The Trump administration has taken away what the Biden administration did and then done a bunch of other cutting besides. And that's why there are people at the IRS, and we don't yet know whether they're right, who think that we're going to have a they think that tax collections are running way behind schedule cuz people are deciding that nobody's going to do anything to them if they don't pay taxes.”
If even 6% of Summers' AI optimism is correct, younger generations will have far more opportunities and ease of work than his generation, making concerns about modest tax increases seem minor
“If I'm even close to right, if 6% of what I said about AI is right, your generation is going to have opportunities to work with far more ease, to enjoy opportunities and forms of entertainment and forms of spending that go beyond anything that my generation uh enjoyed, that the notion that the tax rate would be a little bit higher for the most fortunate members of your generation than it was of my generation.”
Future historians in 2125 will likely view Trump and Xi as second or third stories; the first story will be the transformative change in how people lived and how things were produced caused by AI
“When your intellectual great-grandchild writes the history of this period in uh 22 21 25 my guess is that stuff about Donald Trump will be the second or third story and stuff about Xi will be the second or third story and the first story will be that we had this transformative change in the ways in which um the things ways in which people lived and the ways in which the things they enjoyed were produced.”
Europeans have likely shifted their position on confiscating Russian assets since January 20th because they now must finance Ukraine defense and will prefer seizing Russian assets to taxing their own citizens
“I am uh guardedly optimistic that there may be that there may be more action cuz Europe's going to have to come up with a lot more money and while they didn't much like seizing the Russian assets as between seizing the Russian assets and taxing their own citizens, seizing the Russian assets is likely to look a bit better.”
Many economists are too complacent about current models, but there is no evidence that the profession is in intellectual decline when judged by comparative standards against other disciplines
“are there a ton of economists who are much too complacent about the current models? Yeah, I I certainly do think uh that that's true, but thinking of it, I think there's no evidence for the proposition that it's a profession or a discipline in decline. And you have to judge things relative to some kind of comparative standard.”
Forecasts are inherently probabilistic, so if an economist says a recession is more likely than not but it doesn't occur, this is not a hugely consequential error since they weren't claiming certainty
“Forecasts are probabilistic. I don't know of anybody who said they thought there was an 80% chance of a recession. So, if you say you think it's more likely than not that there's going to be a recession and then there isn't, I don't think that you've made a hugely consequential error.”
UATX should experiment significantly with AI and pedagogy, moving away from traditional teacher-in-front-classroom model, because experiments that succeed can be scaled and those that fail provide valuable negative information
“my instincts would be that experimental institution should be paying a lot of attention to AI and the use of AI in pedagogy and models that move away from uh teacher in front of room, students in back of room. Um I think there should be a lot of experimentation uh with that, whether this is the right venue and what form those experiments should take, uh I don't I don't know. I mean, the great thing about experiments is that if they succeed, then they can be multiplied a hundredfold. If they fail, then we've learned that something we might have tried doesn't work and we don't have to try it again.”
In 40 years of observing presidential behavior, the speaker has never heard a president say 'We might have a recession. It'll be okay,' suggesting the Trump administration's acceptance of recession risk is unprecedented and concerning.
“I've been doing this stuff for 40 years. I've never heard a president say, 'Well, we might have a recession. It'll be okay.' I've heard presidents say we won't have a recession when they're probably wrong and they're trying to add confidence to the situation. But saying like, 'We might have a recession. It'll be okay.' I've never heard uh that particular play before.”
China has engaged in unfair competition practices including subsidies, dumping, and gaming the free trade system for nearly 25 years since joining the WTO
“after it was admitted to the World Trade Organization, China cheated in multiple ways. China continues to subsidize its industries in all kinds of ways. It engages in unfair competition. It does things that look to me, at least, like dumping. And it's done this uh for a quarter of a century, nearly.”
Approximately half the federal budget is spending on the aged and another sixth is spending on interest payments, leaving little flexibility for other spending
“that that's the basic logic of what happens when half your federal budget is spending on the aged and another quarter of another sixth of your federal budget is spending on interest.”
Moving from the gold standard to central bank control of money was clearly wise because gold mining accidents should not determine monetary supply rather than actual economic conditions
“There lots of arguments you can make about exactly how you should determine how much money you want to have to be in circulation. But having it depend on the accident of how much gold gold miners find and decide to get out of the ground seems a an approach that's extremely unlikely to generate monetary arrangements that are conducive to a well-functioning economy.”
A significant part of fiscal adjustment must occur on the tax side, implying the speaker favors progressive tax increases rather than benefit cuts as the primary fiscal adjustment mechanism.
“So, I think a significant part of the adjustment has to take place on the tax side.”
Kissinger's framework for presenting foreign policy options to the president: (1) surrender, (2) what the bureaucracy recommends, (3) World War III—shows that the middle option always appears reasonable by contrast
“the Kissinger point that this shows that you've been in government that recommendations to the president should always be three. In in the case of a foreign policy problem, the choices are one, surrender. Two, what the bureaucracy recommends. Three, World War III. And this is the way that you should always present any decision to the president.”
Mark Zuckerberg has $200 billion; if his tax rate increased to pay ~$12 billion annually, he would not pay his current ~$1 billion/year and could easily absorb the increase without hardship
“take like a economical member of uh the generation that I would imagine is between us, Mark Zuckerberg. He's got $200 billion. If his tax rate his tax rate went up a little bit and $200 billion pays pays Well, who knows exactly what it pays, but let's say it pays 6%. That's $12 billion a year. I promise you he's not paying $1 billion a year in taxes. Um he paid more taxes, be okay with me.”
The statistical methods used to measure economic performance were invented and developed by economists, giving them authority in defining success metrics for their own discipline
“and you know, it's all the statistics we used to measure all these things were invented and developed and refined by econom- by economists.”
There is 'very little support among people who think about these things' for returning to gold standard, indicating consensus among experts.
“I think there's very little support among people who think about these things for the idea that a that there should be some kind of return to the gold standard.”
The most important idea you'd encounter in a finance course is efficient markets or martingale theory
“one of the two or three most important ideas you'd be exposed to if you took a finance course is what could be called the efficient markets or the martingale theory”
There is very little support among people who study these issues for the idea that the gold standard should return; the abandonment of gold was clearly correct
“I think there's very little support among people who think about these things for the idea that a that there should be some kind of return to the gold standard.”
Many economists are too complacent about current models, but the right conclusion is not that the discipline is in decline but rather that there is room for improvement within economics
“Yeah, are there a ton of economists who are much too complacent about the current models? Yeah, I I certainly do think uh that that's true, but thinking of it, I think there's no evidence for the proposition that it's a profession or a discipline in decline.”
Harvard Education School produced Sesame Street, whose benefits arguably justify 50 years of the school's existence even if other aspects were problematic
“And so there's all kinds of things that happened at the Harvard Education School that I was skeptical of. But the Harvard Education School did incubate Sesame Street. And I think you could make a case that the good that Sesame Street did was enough to justify 50 years of the Harvard Education School No. No. No. even if you didn't like some of the other things that happened in uh the Harvard uh Education School.”
Tariff policy that increases both inflation and unemployment is 'really a quite terrible idea' and represents poorly designed economic policy.
“And so, I think this is not a very good idea. It's really a quite terrible idea.”
Summers co-authored important work with Phil Zelikow and Bob Zelik on what should be done with frozen Russian reserve assets, consistently arguing for tougher action than has been undertaken
“your disclaimer that you hadn't looked into this too deeply was a little insincere given that you've been one of the co-authors of a very important piece of work on how we should deal with the Russian reserve assets that were frozen when the war broke out.”