Home Alone: inflation and the new Fed chair; investing in China’s AI ecosystem; Prediction markets

Good morning, everybody. This is Michael Cembalest with the Memorial Day 2026 Eye on the Market podcast. This one’s called Home Alone. It’s about the new incoming Fed chair and inflation. And I thought of this mashup with Kevin Warsh in Home Alone because, like Kevin McCallister in Home Alone movie, Kevin Warsh faces the same kind of lonely vigil he has to survive until the adults get home. Also in this month’s Eye on the Market some comments on investing in China’s homegrown AI ecosystem and then a couple of closing comments on the predation and prediction markets. So just to get started here…

Warsh has highlighted something called trimmed PCE inflation. It’s, it’s, it’s sending the clearest signal that it would be okay to cut interest rates. The problem is that this particular calculation can have biases and lags. And, as you can see from the first chart in the piece, it did a terrible job highlighting the Biden inflation that the GOP is constantly and justifiably highlighting as a, as a big policy mistake.

So it’s just kind of odd only three years after trimmed PCE did such a bad job spotting the Biden inflation era that you would now highlight this because it happens to be showing you a clear, all clear signal that you can ease as opposed to some of the other inflation measures. I understand why he’s doing that.

Trump has already been commenting for months that the Fed should cut rates as fast as possible, that the U.S. should have the lowest policy rates in the world. And then last December, just, just four months ago when Trump was asked where rates should be a year from then, he said 1% and maybe lower than that.

So, you know, it’s don’t, don’t underestimate the amount of pressure that, that Warsh will be under here. The three big picture charts that Warsh is going to have to deal with, which is number one, the surge in commodity prices in the U.S. As we’ve discussed, the U.S. is energy independent, but that has not shielded the U.S. from 50, 60, 70, 80, a 100% increases on a year-to-day basis in a variety of energy, land, commodities, fertilizers and things like that.

Only natural gas prices have gone down. Everything else is up. Big picture chart number two: Over time, the sensitivity, inflation sensitivity of the business cycle is growing. So for given shifts up or down in the business cycle, the inflation has been responding more rather than less. And then the last big picture chart that you’ve seen many times from me before is this dreaded crossover point in 2031, when federal tax revenues are expected to be entirely needed to pay entitlements and interest on the debt. So, in other words, the scope for a monetary policy mistake by Warsh is getting narrower and narrower here as he comes into his new seat. And as we’ll discuss, increasing Treasury yields or shrinking the equity risk premium that gets earned by investors, which is another risk for him.

So that is a couple of charts in here that are a little wonky, but I think they’re worth looking at. And what we did was we picked four variables that relate to the business cycle that, that a Federal Reserve would be paying attention to. Employment conditions, Prices Paid Index and the manufacturing sector, which measures price pressures, the Supplier Deliveries Index, which is measuring how tight your supply chains are and something called the output gap, which estimates how much you’re growing relative to potential, and the more you’re growing faster that potential, the more potentially inflationary that is. And so we plotted all four of these variables and then colored them either red or green based on the intersection of these variables, whether the Fed was raising or cutting policy rates.

And as you can see from… and then the yellow dot in each chart here is where current values are. So the bottom line is the current values are much closer to conditions when the Fed has been raising rates than when the Fed has been cutting rates. So as it, as it relates to Warsh coming in and thinking that he might be having the scope to ease, he’d be going against the history of the Fed at least as it relates to how the Fed historically has responded to these kind of variables. I think it’s worth looking at these charts and understanding what we’re saying if you’re interested in this topic. Inflation expectations can be derived from a couple of different places. You can look at household surveys and consumer surveys.

Those have ticked up a little bit. Nothing catastrophic. Pardon me yet, but those have ticked up a little bit. Same with the inflation levels embedded in the, in the tips markets. You can derive what inflation expectations are there. Again, slightly higher this year, maybe by half a percent or so and consistently above the 2% threshold. The, the, I think, the clearest signals for Warsh are in the wage markets and in the labor markets.

Wage inflation numbers continue to come down. You can look at the Atlanta Fed wage tracker. You could look at the National Employment Index. Both of those look fine. Unit labor costs in the, in the business sector look pretty tame. They’re below 1%. I’m sorry, below 2%, so nothing bubbling up there. And there’s really very limited signs of a wage price spiral.

And we don’t even see wage growth that’s particularly high in immigration-sensitive sectors. And sometimes when I sit down and think about this, I wonder to myself, why aren’t we seeing, since and since immigration is down so much, why aren’t we seeing more wage inflation in things like childcare and cleaning and construction and food prep and home healthcare?

And the reason is, if we take a look, and this comes from Brookings… Brookings estimates a combination of both documented and undocumented in and out migration in the U.S., and it looks like this year that net flow is going to be minus a million people, which sounds like a lot. But at the peak of the Biden surge in 2023, it hit plus 3 million people, compared to the normal trend of around a million a year of net in migration.

So the bottom line is, while the immigration tides have shifted, the stock of both legal and illegal immigration that were taking place a few years ago, you still have a large stock of those workers, and it would take probably another two or three years of Trump policy before you eroded that. So that, to me, I think that’s the best explanation as to why we haven’t seen wage inflation yet in some of those immigration-sensitive sectors.

In terms of shelter inflation, which is an important component of all these inflation calculations, whether you’re, whether you’re looking at the rents from actual rental properties or the rent that the Fed looks at from this calculation that imputes rental growth from owner-, owner-occupied homes, nothing kind of remarkable way down from the levels in 2023 and hovering at, at roughly the same as pre-COVID levels.

And…

Now let’s get to PCI, PCE versus CPI. So there’s, there’s two big inflation measures that as many, as most of you know, there’s PCI, which is personal consumption expenditures. And then there’s the CPI, which is the Consumer Price Index. The primary difference between them is the weights that they ascribe to different categories. Right now the PCE looks more elevated than the CPI.

The Fed reportedly pays more attention historically to the PCE. The PCE has a much higher weight to software and certain business applications, where in the CPI, housing has a higher weight. That’s the primary reason right now that you’re getting higher PCE numbers than CPI. The bottom line is that, that both of them are still trending, on a core basis, over 2%.

And so this is something that the Fed would presumably be paying attention to. The real risk, I think, for Warsh and the FOMC is the inflation that’s bubbling up in producer prices. And whether you look at headline producer prices, core producer prices, intermediate goods producer prices, I mean there’s, there’s, there’s probably 30 or 40 different cuts on producer prices that you can find.

They’re all sending a pretty similar signal here, which is they’re rising, and they’re rising kind of sharply now. There’s several reasons for that. Transport costs in the U.S. are rising because of higher energy prices. There’s a lot of copper and aluminum demand for energy storage and solar panels and EV production. You’re having soaring costs for anything related to the AI boom: electronic components, memory chips, applications software, you name it.

We also have higher U.S. tariffs to deal with, and there’s new export controls from China on rare earths and medical equipment. So, lots of producer price inflation. And then the question is how much will that feed into consumer prices. I think the jury’s still out, but there’s the, the, these sharply rising producer prices are definitely warning shot for the Fed.

We have a couple of charts here that look at more specifically some of the AI-related electronics inflation and in, in components and accessories and batteries, memory prices, GPUs… these, these inflation numbers are all ticking up, pretty markedly.

Now, a lot of times in this inflation discussion, you’ll, you’ll have people raise the issue of productivity because the extent to which productivity is doing well that offset some of the inflationary pressures in the pipeline. And so, for example, I mentioned unit labor costs earlier. Unit labor costs are labor costs adjusted for productivity. And so if you get some kind of AI-related productivity boom, that would reduce the pressure on the Fed.

And Warsh seems to view things that way, which is that AI has the potential to be a supply shock, a deflationary supply shock, which would allow the Fed either to maintain steady rates in the face of some of these pressures or maybe even to cut them. He’s replacing Powell, who had a slightly different view, which is that in the short term anyway this AI boom is inflationary because of the pressure on capital spending and the ripple effect that has on the economy.

What do we know about productivity so far? Well, since the launch of GPT in the fall of 2022, and we have a couple of tables on here computed a couple of different ways. The productivity numbers have definitely picked up. Overall, overall corporate productivity, specifically, productivity in the information sector and information in the data processing subset of the information sector.

So it’s hard to ascribe definitively that this is AI related and things like that, but the evidence is pointing in that direction. And so it’ll be interesting to see how Warsh responds to these productivity numbers while the rest of the FOMC may still have that, that Powell bias that views them as inflationary rather than deflationary. Just one last comment on this.

That’s a little bit technical. J.P. Morgan’s economists do a lot of analysis on policy rates, and using a bunch of different monetary rules of thumb, Taylor Rules and alternative hours star and a bunch of different very technical rules, they would get today a Fed Funds Range that should be anywhere from 4 to 4.8%. That compares to the current range of three and a half to three and three quarters.

So that’s just another comment here that the, the technicians that are looking at this think rate should be higher or rather lower. Now, one of the consequences of rising actual bond yields is you get a, a shrinking equity risk premium. And what does that mean? The equity risk premium is a rough rule of thumb that, that market analysts often use to look at the return that equity investors earn over risk-free investments in the Treasury market. And for, for the bulk of the period post the financial crisis, that gap was pretty wide and which, which led to a lot of outperformance and a lot of flows and excitement about investing in equities relative to bonds. Well, the selloff in, in yields and more and more expensive equities has tightened that equity risk premium back to the lowest level that it’s been at probably since the late 90s, early 2000s. And so the bottom line is that equities are looking more expensive relative to bonds now that Treasuries are rising faster than inflation expectations, and that leaves equity markets a little more exposed than usual to, to an interest rate shock. And so that’s something that we have to continue to watch. And certainly if, if the fed, if the Fed’s monetary policy is perceived by the markets to be too weak and to accommodative, that could exacerbate this kind of selloff. So, my, my bottom line is like, like the Kevin McCallister character in Home Alone, you know, good luck to this Kevin as well. U.S. 30-year Treasury yields haven’t consistently exceeded 5% since, since the early 2000s but have just crossed that threshold. We’ll see how long they stay there. A lot is riding on normalization of energy prices and an end to the Iran war as we’ve discussed. And productivity gains, the next report comes out in early June… will be really important to track because Warsh might hang his hat on that as an argument to, to not raise rates and maybe keep them stable for a while, even though some of the economic pressures are showing some, some inflationary buildup.

And again, there’s also pressure that’s going to come from the White House to ease. And that might end up being the defining act of Warsh’s tenure at the Fed, which is how does he handle the intense pressure from the executive branch to cut policy rates. And let’s hope for Warsh’s sake that he doesn’t end up like Arthur Burns, for whom, a Fed chair for whom the adults did not return in time. And just as a brief time capsule on this, when he was running for reelection in 1972, Nixon wanted to bring unemployment back down to the 3.5% level that prevailed at the end of the 1960s during his first term. And he found a coconspirator here in Fed chair Arthur Burns, who resisted calls from the rest of the FOMC to raise policy rates.

Burns supported wage and price controls, which also ultimately were inflationary, and he oversaw a money supply expansion that peaked at 13%, which was the highest money supply growth on record and was until the Fed’s COVID bomb back in 2020. The, this Burns here at the Fed was kind of a disaster. It coincided with high inflation, high unemployment and a decade of roughly 0% real returns on both stocks and bonds.

And we have a chart in here that shows money supply growth by Fed chair era. And you can see the Burns era set, set some records for money supply growth until COVID came along. And the thing I wanted to highlight, aside from the striking similarity between our picture of Arthur Burns with his hands on his face and the iconic Macaulay Culkin fame picture from the Home Alone movie is it looks like Burns only agreed to this super easy monetary policy after really substantial pressure from the White House.

And if you’ve ever read some of the books about the Nixon dirty tricks campaigns that he would, he and his people would execute against political opponents, there’s some of them here. When Burns resisted pressure to guarantee full employment, the White House planted negative stories about him in the press. They falsely claimed that he was requesting a large pay raise when he actually had offered to take a pay cut. They floated stories in the press about diluting the Fed’s power by doubling the board members. And, and Nixon warned Burns on a recorded call that he could destroy the last conservative administration in Washington. And then he wrote to Burns separately that there’s no doubt in his mind that if the Fed keeps a lid on the money supply and the economy doesn’t expand, the blame will be placed squarely on the Fed.

And if by 1971, Haldeman, if you remember, Haldeman spoke about the effectiveness of this strategy, saying, we have Arthur Burns by the balls on the money supply. So I hope for Warsh’s sake that he handles this kind of pressure better than Burns did. Okay, another brief Home Alone-related topic… by both the Trump administration and the Biden administration imposing such a large array of sanctions on China related to everything, related to AI, to, to, I, to AI… The U.S. effectively compelled China to adopt a home alone strategy for their own AI ecosystem.

And in our Outlook earlier this year, the third section walked through China’s efforts to build a parallel system, from lithography machines to high bandwidth memory to semiconductor production fabrication. And I just wanted to give you an update because since that time, China has made a lot of progress, which the equity markets have been increasingly rewarding. And I just wanted to pass along how we view investing in that particular market dynamic.

One of the most important charts to look at is, is at the time that GPT was launched, China was only 20% GPU self-sufficient. That number is now around 40% and projected to go up to 80% plus by the end of the decade. So China is pushing really hard on its national champions domestically to use more and more Chinese chips.

And China’s chips are narrowing the performance gap. They haven’t closed it, but they’re narrowing the performance back gap with Nvidia in ways that allow the, the Chinese private sector to be increasingly using this Chinese ecosystem. And there’s lots of different ways that people’s core models, the, the frontier model scores from China are tracking the improvements in the U.S. almost on a one-to-one basis over the last two years.

Now there’s still a gap, but the gap is getting smaller and smaller and smaller. And then this is the other thing too, that, that has been interesting to me. As the Chinese private sector thinks about adopting Chinese GPUs, TPUs, XPUs, which… whichever version you’re talking about, it’s not just about the raw power of the chip, right? Because Nvidia still leads there. It’s, it’s total cost of ownership.

And it’s a combination of power and installation and hardware and software, and here when we look at Huawei and Capricorn, the total cost of ownership of a lot of those chips is considerably lower than the Nvidia suite of H200, H100 and H20 chips. And so I think it’s important that people remember it’s total cost of ownership that matters and, and cost per token, which are roughly similar across both Nvidia and these two large Chinese companies.

So that’s one of the other reasons why uptake in China, it’s not simply because they’re being forced to do it, it’s because it’s cost effective for them to do it. And you can see this explosion in AI activity. We took the top nine models from OpenRouter that does the, the token API calls. And a year ago the Chinese models were almost invisible, and now they’re, they’re, they’re higher in aggregate than the top U.S. models with the top nine. And so, that’s kind of a remarkable shift that’s taking place in just 12 months. And so I reached out to our China portfolio managers at J.P. Morgan Asset Management, and I asked them to help us look at a basket of the, of the companies that are leading this AI effort in China.

They identified 32 of them. And, and those stocks have been roughly keeping pace with the 42 AI stocks in the U.S. AI basket that are in the S&P 500. And so this, this looks increasingly like an interesting investment opportunity and, and one that you can speak to your J.P. Morgan money managers about accessing.

Okay, just a couple of final questions here… that I thought was, was interesting. On prediction markets… you can go ahead and participate in these markets. Just be aware that the algorithms are probably going to beat you to it. I thought this was remarkable. There’s a whole bunch of studies that have just come out that look at the share of profits that are earned by a very small share of participants.

One example, 1% of all the participants on Polymarket earn between 75 and 80% of all the gains, and 0.1% of the accounts on Polymarket earn 67%. So 0.1% of the accounts earn two thirds of the profits. This is due, at least in part, to very high-speed algorithms that are used by this small subset of accounts to reprice these prediction market contracts based on breaking information much faster than the average person.

And one way to think about how absurdly high that profit concentration is, is to compare it to the profit concentration of the top 1% in online poker, retail options trading, equity day trading, horse racing, sportsbooks, and obviously the Polymarket number is much, much higher. So again you can place your bets, but be aware that the algorithms are probably going to beat you to it.

Then one last thing to keep in mind. Suppose there’s a dispute about the outcome… A lot of people on this call are familiar with the, with the ISDA arrangements between professional counterparties and the process by which ISDA disputes are resolved by committees as it relates to… was a certain credit default swap contract triggered or not?

Just where do you see the Polymarket version? If a Polymarket bet does not have a clear outcome, there’s a third party token-based voting system that gets used, and any individual, not necessarily the Polymarket… you don’t even have to be Polymarket user, can purchase tokens from crypto exchanges that get used to cast votes in these resolutions.

So Polymarket bettors can essentially purchase a whole bunch of tokens to manipulate the outcome of resolutions that they’re actively betting on. This is kind of amazing and not in a good way. So, for example, there was a Polymarket bet on whether Ukraine would agree to a critical mineral deal with the U.S. before April 2025. And there was a dispute about that because it wasn’t resolved until the end of April and for the end of April, it’s not the same as before April. Yet a single voter cast 25% of the votes and swayed the decision to Yes. And that ended up completely changing the outcome, for the, for the, for the participants in that contract. So, anyway, just read the fine print as you get involved in these prediction markets. That is it for today’s Eye on the Market.

The next one we’re doing is a 250th anniversary of United States piece on the U.S. grip on global financial markets and the world reserve currency and things like that. So stay tuned for that 250th anniversary piece in June, and we’ll see you soon. Thank you for listening. Bye.