From an economist's perspective, is there such a thing as the disinflationary boom, or the deflationary boom even?
So this is about, you know, making a distinction between now and what comes next.
No, it is about the high levels of investment, right? Clearly, right now it's pushing everything up. But let's imagine ten years from now and we've done it. We have this in credit, and somehow we've solved all the sci-fi scenarios so that the AI doesn't want to kill us all, etc. Let's just imagine the rosy scenario in which we have this. Look, there are some under-discussed risk factors. But right, like, let's just say the robots don't kill us. Assuming we solve that and the AI works on our behalf, and it does what we want it to do, and it can create incredible material gains while also delivering cheaply because it's just one AI, etc. Is that a contemplated scenario?
From an economist's perspective, there is absolutely a scenario where we could be in that wonderful place with high productivity growth. At the same time, we don't have civil unrest or, you know, rogue forces using AI for ill. That is a scenario. But I do not know a single person who will put a probability on that scenario and say that that's going to happen, you know, with a significant amount of certainty. There is a very high degree of uncertainty. And there are several who also believe that it's yet to be seen whether there is going to be any major productivity gain of the kind that, you know, there are analysts who believe the productivity could go up by two percentage points a year, you know, over and above where it is right now, which is around two percentage points a year. And that is huge. There is no evidence right now of that kind of productivity wave coming through. So it's early. But you know, I use the technology and I find it terrific. I mean, it's been really great for my own productivity. It's not affecting my wages or anything so far, but it is, you know, very valuable technology, but there is a lot of uncertainty.
Hello, and welcome to another episode of the Odd Lots podcast. I'm Tracy Alloway.
Joe, the big story in markets right now has to be one of the big stories: the bond market selloff. By the way, as I say that, I just got this massive feeling of deja vu because I'm pretty sure we've done a few episodes where I've started out saying the exact same line.
Well, I mean, one, obviously this is sort of one of the trends of our time, which is that after a decade pre-COVID where we just sort of assumed that rates were going to head lower, that is obviously, there's been a regime change, as economists sometimes like to say. And so now we have rates pushing higher again. They've come back a little bit in the last couple of days, but that's not really the point. The point is, it is this global phenomenon around the world. Rates going up. I would still say probably the big story in markets is AI and memory and chips. But if it weren't for that, everyone would be talking about interest rates higher in almost every country in the world. So here's the thing. I actually think AI and the rate sell-off is kind of connected. So, you know, we're talking about yields going up generally in developed markets. And we've seen that recently. I know we saw, for instance, the long end of the UK gilt market hit like the highest since 1998. The ten-year U.S. Treasury yield was kind of inching up towards 5%, but it's come down mostly. A lot of those yields have been moving in line with the oil price, right? So a lot of people will say that this is just because oil's going up. That's inflationary. Maybe we'll get higher rates. And so this is why yields have been backing up. However, there is an argument I'm seeing more and more people make, this one, that what's going on is actually a repricing of something, you know, less transitory. Am I allowed to say that word anymore? Less transitory and something more secular. In what's happening with the rates market, something that's more about the massive amounts of capital that AI is actually consuming and having a crowding-out effect on sovereign bonds, or maybe something that's more about, you know, the ability of the developed world to actually finance itself, you know, in the longer term. And so you're starting to see some of those bigger themes creep into the discussion about the bond market sell-off. This idea that it's something else is happening here, something more than the oil price.
Totally. Actually, just speaking of the nexus between interest rates and AI, Torsten Slok has a good chart out. Came out this morning pointing out essentially that one thing, what they are, is the sort of FOMO aspect, not among investors per se, but about companies not wanting to let their models be six months behind, and so they'll pay whatever the cost is to catch up. And therefore, he argues that perhaps higher rates do not have the slowing effect that they might have had in another cycle. Because it's like, well, yeah, it's no fun to finance this data center at higher rates. But if the alternative is being consigned to the permanent underclass when the other company builds the most advanced model, you're going to do it nonetheless. And so yes, between oil, between the AI boom, between demographics and the challenges of sort of resourcing for care of the elderly and the infirm, between all of these things, resource nationalism building up, thinking that we are in this real secular shift and we have to understand it better.
Yes. So I am very happy to say we do, in fact, have the perfect guest to talk about all of this. We're going to be speaking with Gita Gopinath. She is, of course, a professor of economics at Harvard University, but also famously the first deputy managing director of the IMF. So truly the perfect guest to speak to someone who's been talking about, you know, a change in the bond market for quite some time. Gita, thank you so much for coming on Odd Lots.
It's a pleasure, Tracy. And, Joe, great to be on your show.
So what's your take when you're staring, presumably, you know, on a minute-by-minute basis at a chart of the U.S. ten-year yield? What are you thinking?
I mean, firstly, I think it's absolutely right to start with the conversation about what's happening in bond markets, because frankly, you know, despite all the many different shocks going around in the world, I actually do think the one that's most worrisome is what we see with public debt levels everywhere in the world, in the U.S. We've seen yields go up. It's a combination of things. You just talked about all of them, which is one is the fact that inflation is now expected to be higher. And there is a sense that the real rate at which the economy will stay, you know, at a somewhat stable level of inflation is higher. So the kind of the real interest rate has drifted up. The R-star has drifted up from pre-pandemic when it was like half a percentage point. Now it's at one percentage point. But on top of that, you have the premia that's coming from the risk of inflation, from very importantly the large amount of fiscal, you know, the large fiscal deficits that the U.S. is running and is projected to continue to run into the foreseeable future. And, of course, the third element, which is the AI boom and the expenditure of the capital expenditure that's being undertaken for that, is also shifting the R-star up to maybe even higher than one percentage point. So because of all these reasons, we've suddenly moved away from the pre-pandemic period of low-for-long interest rates. And what we were talking about, the, you know, the end of, I think we have the end of secular stagnation at this point. Secular stagnation was about the fact that there was not enough investment happening, especially in the private sector. That is no longer an issue anymore. So because of the combination of inflation, AI boom, fiscal deficits all over the world, high public debt everywhere, you know, we are seeing yields go up. And that's true in the U.S. too.
What changed? I mean, what I want to drill into all of these specific things. But, you know, let's start with like the high level of public debt. That was the thing that people were talking about quite a bit prior to the pandemic as well, and rates just kept going lower and lower, including famously in Japan, where debt-to-GDP levels are even much higher than they are in the Western world. And that was sort of famously known as the widowmaker trade because rates kept going lower. What's changed between 2020 and 2023 such that this suddenly, in your view, and perhaps the market's view, this became an important thing that was not perceived by the market as being important pre-COVID?
So a few things changed. One, the AI boom was unexpected. That was not something that was being priced in markets pre-pandemic for sure. That big increase in demand for capital coming from the private sector is one, a big change. The other big change is the fact that fiscal deficits are now projected to stay at levels that nobody was expecting the U.S. to run. So some 7% fiscal deficits for the foreseeable future. That is another important factor. And the third is the composition of who's the marginal buyer of this debt. So we had a period when central banks everywhere were buying government debt. And that also helped to keep interest rates low. In fact, that was part of the strategy of how to, you know, strengthen the economy. Quantitative easing was part of the toolkit. And so that helped keep interest rates low. But that's changed. And now we have the central banks everywhere who have either stopped buying or they're winding it down, like it's happening in Japan. And the marginal buyers are the more volatile, you know, investors, hedge funds in the U.S. are the U.S. market makers over here. And so whenever there are any shifts in global market conditions, you see a lot more sensitivity than you would have seen if it was mainly official credit flows. And by the way, that's also true about capital flows coming into the U.S. Previously, the buyers of U.S. Treasuries used to be foreign central banks. They're not doing as much anymore. It's mainly coming from non-bank financial institutions from the rest of the world. And so they're also much more volatile. And you're going to see just generally high volatility in the yield curve.
Can you talk a little bit more about the AI boom? Because we hear people talk about a crowding-out effect. And I think this is actually like something that is just starting to get a lot of attention. But the proportion of issuance in the corporate bond market that's coming from AI companies or AI-related, you know, investment right now is just insane. And you mentioned Torsten's chart. Torsten is going to be at our upcoming... That's right. We're recording this on May 27th, our upcoming live show in New York. And so I've previewed some of the charts he's going to be sharing there. There's a chart there that shows basically the proportion of AI in the corporate bond market. It's now 50% of all investment-grade issuance year to date. And in even junk-rated debt, it's creeping up to like almost 40%. So this is a significant amount of debt that's being issued into the market. Is it reasonable to think that investors are maybe going to think like, well, I'll buy some big tech mega-cap IG debt versus a U.S. Treasury?
At least when you look at the pricing in markets, that seems to be the case, especially when it comes to equities. Everybody wants to have a piece of the AI boom. And yes, I think there is that sense that, well, this is a sector where we could really see real gains, especially, you know, in terms of productivity increases and profitability. And that's going to help. That's going to be something that they want to be a part of. So there is that demand for corporate bonds and for U.S. equity, which is coming again both from domestic investors but also from international investors, which, you know, where AI is the trade. I mean, that's where all the dynamism is and that's where people want to put their money.
Now, we've pointed at all the reasons why rates are likely to stay high. But I just want to point out that we still have the ongoing Iran conflict and we still have the Strait of Hormuz closed. If that is not resolved anytime soon, like, you know, the next month or so, and you see an even steeper increase in oil prices and crude prices going up to, say, $160 a barrel, which is what some of the projections would be in that case, then we could see much more demand destruction than we have seen today. And we could be back in that space where at least the short end, interest rates are being cut pretty rapidly.
I want to, on this question, going back to the effect that the AI build-out is having across rates and bonds and so forth. I want to sort of get some clarification here of what either you or what economists mean when they talk about, say, like crowding out. Because there's one version of it that is like, okay, there is a lot of debt being issued right now by very highly rated companies that probably yields a little bit more than U.S. government bonds. That is attractive for investors. Perhaps there's, and maybe that has some sort of crowding out in the financial markets. The other way that I tend to think of crowding out is that these AI, the AI build-out, it's like they're taking up all of the wind turbines. They're taking up all of the trucking capacity to get the goods to the data centers. They're taking up all the skilled contractors and laborers within the regions that these data centers are being built. And that creates inflationary pressure that adds to the strain and therefore, all things being equal, that says higher inflation and therefore higher rates for longer to maintain that. Which of those two models, whether it's the sort of financial markets version or the sort of real economy version, is a more useful way of thinking about that linkage between private and public sector spending slash debt?
So both of those are in play right now. So that's the difference between what's happening to real rates versus what's happening to nominal rates. And what's happening to the pricing of the Fed rate path. So the first channel that you mentioned, which is just the fact that there is so much demand from the private markets, from the AI investors' forecast that, you know, companies for capital is going to raise real rates. You know, even if there is no effect on inflation or inflation expectations are not moving, we should expect to see real rates rise. And that's certainly, we're seeing some of that. And then the other is the effect that's working through the demand for the different inputs going into AI. And that's creating an inflationary pressure which would then need higher nominal rates. And that is also, you know, playing out. I think right now, I suspect that the real rate piece is more important. The inflationary part is being driven a lot by what's happening with energy prices and pass-through from energy prices into also core inflation. So I think that's more of the, you know, the higher inflation, higher rate path story is coming from other forces on inflation as opposed to what's coming from AI itself. And then you have the real rate path, which is going up also because of the general risky environment that we are in, but also because of what's coming from this increase in, you know, capital demand coming from the AI sector.
So one of the reasons we wanted to speak with you is because you do have that very significant policy experience. And so when faced with, you know, potentially rising real rates because of an AI boom, what should policymakers be doing here? Because, you know, certainly in the U.S., we've already seen some Fed officials or outgoing Fed officials start to argue that they can look through the AI boom and its impact on inflation. But if real rates are, you know, structurally rising, if R-star is structurally higher than it was before, is that the right move?
So, I mean, what matters crucially is what we believe are the main drivers of R-star at this moment. Is it coming because of higher productivity growth which is then leading to higher investment and therefore demand for capital? You know, all of that is good. Is it kind of the good kind of increase in R-star? Because that's an economy that is projected to grow at a faster rate. And that helps. That helps on many fronts, including in terms of bringing debt-to-GDP down. The other reason R-star is going up is because of the increase in fiscal deficits and just general high levels of government borrowing in the U.S. That is less appealing because that tends to be not necessarily growth-enhancing. That, you know, the money that's being raised is not for, you know, necessarily productive infrastructure investment that's going to generate enough growth. So that's more problematic because it's just generating R-star without generating the higher growth that should come with it. And that can be a problem from a policymaker's perspective. Of course, you have to firstly be able to tell what is driving R-star. Is it a good kind or is it the bad kind? But regardless, if R-star is drifting up and you have an inflation target of 2%, you are looking at higher nominal interest rates, right? So right now the Fed has an R-star of about 1%. If you put 2% inflation on top of that as their target, we're looking at 3% nominal rates, which is a clear shift away from what it used to be pre-pandemic. Then you have to check to see whether the R-star increase is actually slowing the economy down, or the increase in your nominal rates are slowing the economy down. And how much higher than that do you have to be to be able to bring inflation down? Because there is obviously above-target inflation in the U.S. at this moment. That is now the big question: whether the productivity boom is going to mean that you don't need that much of an above-R-star interest rate, or do we have any other forces coming from energy prices passing through into core inflation, the lower levels of immigration in the country, just general trade disruptions, supply chain disruptions? And those are the main drivers in which case maybe you need to keep interest rates even higher. So being able to tease that apart is where, you know, I think that's where the tough decisions are. But what is squarely the case is that we are looking at higher nominal interest rates.
I mean, regardless of where the R-star is coming from, the good kind or the bad kind, you know, there is this fantasy, and hopefully it comes true, but there is certainly this fantasy of a lot of people who are into AI, I would think it is called, they would call like the disinflationary boom, right? So let's just imagine we have extremely powerful artificial intelligence that is capable of delivering, you know, incredible material gains for people. It makes health care really easy and quick. It could power robots that care for us. It can build things, etc. And meanwhile, the cost of commodities collapses. Maybe the cost of labor collapses. Is that a scenario in which it's worth contemplating and thinking about? Well, so everything gets really cheap because it all gets super automated. But also our standards of living rise dramatically because the AI takes care of it for us. Is that conceivable? Like, from an economist's perspective, is there such a thing as the disinflationary boom, or the deflationary boom even?
So this is about, you know, making a distinction between now and what comes next. No, it is about the high levels of investment, right? Like, clearly right now it's pushing everything up. But let's imagine ten years from now and we've done it. We have this in credit, and somehow we've solved all the sci-fi scenarios so that the AI doesn't want to kill us all, etc. Let's just imagine the rosy scenario in which we have this. Look, there are some under-discussed risk factors. But right, like, let's just say the robots don't kill us. Assuming we solve that and the AI works on our behalf, and it does what we want it to do, and it can create incredible material gains while also delivering cheaply because it's just one AI, etc. Is that a contemplated scenario? From an economist's perspective, there is absolutely a scenario where we could be in that wonderful place with high productivity growth. At the same time, we don't have civil unrest or, you know, rogue forces using AI for ill. That is a scenario. But I do not know a single person who will put a probability on that scenario and say that that's going to happen, you know, with a significant amount of certainty. There is a very high degree of uncertainty. And there are several who also believe that it's yet to be seen whether there is going to be any major productivity gain of the kind that, you know, there are analysts who believe the productivity could go up by two percentage points a year, you know, over and above where it is right now, which is around two percentage points a year. And that is huge. There is no evidence right now of that kind of productivity wave coming through. So it's early. But you know, I use the technology and I find it terrific. I mean, it's been really great for my own productivity. It's not affecting my wages or anything so far. But it is, you know, very valuable technology, but there is a lot of uncertainty.
And which is what is very curious about the markets, right? Because on the one hand, it is impressive where the stock markets are again at close to a record high. And maybe one can explain that by saying that, well, there is a scenario where everything goes perfectly well, but there are so many other scenarios that could play out between now and next year or even two years from now, and you barely see that priced in markets. So I think that's frankly more surprising than just looking at what's happening with just the level of, you know, of the stock market.
Okay, we keep talking about the stock market and, you know, debt issuance in the corporate bond market and how everyone wants a piece of AI. Does that basically mean that we're seeing, I guess maybe scarcity of capital versus the global... We used to call it a global savings glut in the early 2000s, right? Which ended up, per Bernanke, pushing yields lower. Does anyone still talk about a savings glut or should we all be talking about, you know, like capital scarcity?
