Larry Summers: we haven’t nailed the landing yet

Good morning. Yesterday the 10-year Treasury yield fell below 4 per cent for the first time since July, as the market continues to process Wednesday’s pivot by the Federal Reserve. On the equity side, small-cap stocks continue to be the story: while the S&P 500 has shrugged, the Russell 2000 is up 6 per cent since Jay Powell started talking. Send your reading of the post-Fed moves: robert.armstrong@ft.com and ethan.wu@ft.com.

Friday interview: Larry Summers

Larry Summers — former Harvard president and Treasury secretary, and new OpenAI board member — is one of the most influential economists of our time. Below, he talks to Unhedged about the state of the US economy; the right approach to monetary policy; the future of the Phillips curve; the productivity impact of AI; and the economic implications of a second Trump presidency (the interview was recorded early in the week, before Wednesday’s Fed press conference).

Unhedged: Have we landed softly?

Larry Summers: I think it’s premature to judge that we have landed softly, because I think that if you look at underlying inflation rates, depending upon your measures, some of them are still running well above 2 per cent. If inflation is currently at 2 per cent, it’s not clear that it won’t go back up again. And it isn’t clear that the landing has been soft in the sense that there are a variety of problems — declining flows of credit, inverted yield curves, aspects of consumer behaviour, rising evidence of credit strains — that raise the possibility that the landing won’t be soft, if there is one. So at this point, we may soft land on the aircraft carrier, but the landing may be hard, and we may overfly. 

That said, if by a soft landing one means a period when you have inflation above 4 per cent and unemployment below 4 per cent, and you extricate from that situation without a recession — that’s something that’s never happened before in the United States and for which there’s very little precedent in the industrial world. And it certainly looks in play as a possibility, though I think it’s a long way from assured.

Unhedged: What do you think is unique about this period in history, such that it is at least possible that such a thing might happen?

Summers: It’s probably a little bit premature to start speculating about what aspect of history makes this happen, when it’s far from clear that it has happened. But if events have been more favourable than what I might have expected a year or two ago, I think primary credit should be given to the Fed for having acted relatively rapidly to correct its earlier errors. Early on I compared the excessive Covid stimulus to the experience of the Vietnam war period, and talked about how inflation had ratcheted up between 1966 and 1969, and then became entrenched. Here, the fiscal expansion was scaled very substantially back between 2021 and 2022. And the Fed in 2022 raised rates very sharply in a way that did not take place during the Vietnam period.

So I think that, ironically, if team transitory proves to be vindicated, it will only be because their policy advice was not taken. It will be because the Fed moved strongly enough that [inflation] expectations never became unanchored. And I think that will be the lesson: that strong Fed action that asserts credibility can be surprisingly effective in containing inflation through its impact on inflation expectations, rather than by generating economic slack. If we are successful, it will be a tribute to the fact that the Fed’s actions in the two years after 2021 were a less than 1 per cent probability set of actions relative to what the market expected. 

Unhedged: If you had the job of setting the economic parameters around easing or normalising policy, how would you do it?

Summers: I think the Fed has done itself considerable damage by putting as much emphasis on forward guidance and transparency as it has. I [prefer] the Volcker/Greenspan approach, which is to recognise that the Fed is a little bit like the Delphic oracles. People regarded them as omniscient and omnipotent, but they were in fact neither. So the oracles kept their pronouncements vague and oracular, not concrete and specific, because it was impossible to be concrete and specific without being wrong frequently and undercutting credibility. So I think that the Fed has had a problematic approach to financial communications.

The two most successful bits of financial communications in the last generation, I would argue, were both of an entirely non-dot-plot or forward guidance variety. They were Mario Draghi’s statements about “whatever it takes”, and Bob Rubin’s “the strong dollar is in the national interest”, both of which qualified as general and oracular.

So I would be endeavouring to not constrain myself substantially with any set of predictions or attempt to lay out my reaction function, because I would recognise that events would come in ways that I wouldn’t anticipate, and that I would run the risk of trouble. Another way to say it is: forward guidance is a bit of a fool’s game. The market doesn’t especially believe it and the Fed feels constrained by it down the road.

I think I would be looking for very clear evidence that inflation was durably put down, because I would be very concerned that we would confuse touching 2 per cent with achieving 2 per cent, and even more concerned with touching 2.7 per cent and regarding that as a basis for easing . . . I would be very much aware that if the transmission from monetary policy to inflation when there had been tightening was more direct, and involved output less than one might have expected, there was the risk of something parallel on the loosening side.

So I would be in less of a hurry. The current market view that we will have a soft landing and the Fed will be able to cut rates by 100 basis points over the next year strikes me as possible, but it is not at the centre of my expectations. We may have no recession, in which case I rather doubt that the Fed will be able to cut rates by 100bp, or we may have a recession, in which case the Fed will cut rates by somewhat more than 100bp.

Unhedged: How do you think the Phillips curve emerges from the experience of the past few years?

Summers: First of all, one should always have been aware that a substantial part of the increase in inflation was transitory. So no one should have thought that most of the route from 7 per cent to 2 per cent needed to be achieved in ways that were correlated with increases in unemployment. I think there’s a question as to how much disinflation has been achieved in terms of core-type measures of inflation already. I think it is fair to say that it looks like there’s more susceptibility to the inflation expectation term. And I think it is true that the Phillips curve coefficient is looking small.

So it certainly hasn’t been a glorious period for the Phillips curve theory in any of its forms. But I’m not sure we have a satisfactory alternative theory. The theory to which many economists are gravitating to is that the Phillips curve is basically flat, inflation is set by inflation expectations, and inflation expectations are set by the people who form inflation expectations. And that’s a little bit like the theory that the planets go around the universe because of the orbital force. It’s kind of a naming theory rather than an actual theory. So I think inflation theory is in very substantial disarray, both because of the Phillips curve problems and because we don’t have a hugely convincing successor to monetarist-type theory.

Monetarist theory had an idea that the price level had to do with the quantity of paper versus the quantity of goods. But now that money pays interest, what the nominal quantity is, that is divided by a real quantity and sets the price level, is unclear. We know from extreme examples, like the monetary history of Argentina, that in some contexts a theory about the price level and nominal quantity of money becomes the right theory for thinking about inflation. But how one thinks about that in the context of relatively low inflation environments, I think economics is embarrassingly short on clear, operational theories. While I have Keynesian policy instincts, I have long been pretty unconvinced by new Keynesian models as an intricate structure.

Unhedged: Pivoting to your new role as a board member of OpenAI, is there any disciplined way to think about the potential effect of an innovation like generative AI or large language models on productivity? Are there historical parallels we can turn to?

Summers: First of all, I think that there are those who know they don’t know, and those who don’t know they don’t know. I’m one of those who knows that he doesn’t know. But it is a useful generalisation that fundamental technologies often take longer to happen than you think they will, and then happen faster than you thought they could. If you think about electrification, that was the story. And the first time I heard that shale oil was close I was 12-years-old. But in the last 10 years, the fracking revolution and US energy independence has been much larger than anybody would have expected. The first time one heard that computers were going to be transformative I was a graduate student, and the impact in terms of big productivity acceleration took place 15 years later, in the 1990s. It’s important to remember with respect to technology that there’s a kind of productivity J-curve. Think about autonomous vehicles. Tens of thousands of people are working on them. And so far not a single driver has been replaced.

I’m struck that most of the discussion of artificial intelligence that I encounter involves thinking about how I can do tasks I recognise, better. And my guess is that when the history of this is written several decades from now, a large part of it will involve tasks of which we currently can’t conceive. It remains to be seen just how transformative artificial intelligence will be for the macroeconomy in the next several years. I do think the ultimate long run impact, whether it’s once a decade, once a generation, once a century, or once a millennium, is going to be very profound.

Unhedged: If Donald Trump is our next president, what are the key macroeconomic implications, as against a second Joe Biden term?

Summers: If you look at valuation multiples associated with any asset, they have a lot to do with the basic security of property rights and the ability to rely on the rule of law. And that is a fundamental feature of an economy that we tend to take for granted in the United States, the way we take the success of anaesthesia for granted during a surgical operation. But when you have a president who challenges the results of elections and brags about what he could do in one day as a dictator, it is not something that can be completely relied on. That is a profound threat to our long-run prosperity, and therefore short-run asset prices, economic behaviour, hiring, investment and everything else.

Paul Samuelson wrote a famous essay after the second world war in which he considered a group of countries marked by strong natural resources and educated population. It had four countries in it: Canada, New Zealand, Australia and Argentina. No one would group those countries together today. And so if one asks: “What did the coming of Peronism do to a potentially successful economy?”, I think that gives a frame for thinking about the risks associated with a Trump presidency. A Peron-ish, Mussolini-ish type leader can have positive impacts on markets and some economic variables for some intervals. Mussolini ultimately was said to have made the trains run on time; Germany had some economic success during the 1930s. So I think it’s a mistake to confidently predict recession or depression in the short run. But I think it’s a mistake not to be extraordinarily alarmed about the medium-term economic prospects of a government of that kind.

And I say that as somebody who, unlike many Democrats, had very strong views and did not get my way in the elections of 1980, 1984, 1988, 2000 and 2004. None of those elections went the way I wanted them to, but in none of them did it occur to me to express alarm about the situation of the United States, the basic capacity of the United States as the world’s most successful economy and a linchpin of the international system. I always thought the system was robust and that pendulums swing. I do not feel that way about the prospect of a Trump presidency. I think there is a substantial risk that it would be immensely destructive because it moves the conversation into realms that we don’t usually think about. Its destructiveness won’t take the form of interest rates being 100bp lower or higher than they should have been, or the budget deficit being one-and-a-half percentage points of GDP higher or lower than it should have been. But it will take the form of destruction of the fabric of rule of law, which is the air that successful capitalism breathes.

One good read

The shrinking equity risk premium.

Source: Economy - ft.com