High long-end yields aren't policy weakness — AI data centers are bidding away the capital
At Jackson Hole, Kansas City Fed President Schmid said the 30-year Treasury climbing above 5% is the result of AI and data centers competing for capital; his personal judgment is that monetary conditions are still on the loose side, and inflation running up into the mid-3s has to be dealt with.
The video won't play here. Listen to the audio instead:
The argument · tap a timestamp to hear it
The endpoint of payments innovation is making payments boring
Schmid says the Fed's goal is to make payments boring: 5 to 10 trillion dollars settles every day across several sets of rails, and future technology will kill frictions like float and fees, delivering atomic settlement — money moves instantly and reconciles instantly. That is enormously innovative on one side and disruptive on the other, which is why the conversation has to be about how you preserve certainty on top of an instant system. This is the setup for the whole conference theme: payments innovation is not a technology problem, it is a settlement and liquidity problem.
— Jeff SchmidAn instant-settlement world has to bid for liquidity all over again
Asked how bond yields connect to the conference theme, Schmid pulled it back to two words: duration and liquidity. If payments are instant, there has to be verified liquidity sitting behind them to complete settlement. With the economy growing and inflation not yet back to 2%, the yield curve will re-price at more normal levels. Put differently, long-end rates are high not only as a consequence of monetary policy, but because the whole system is re-pricing the liquidity and the duration that an instant-settlement world demands.
— Jeff SchmidBoomer retirement drains experience, not headcount
Schmid says the baby boom cohort he belongs to is retiring at a pace of roughly 4 million a year, and that he has signed more retirement letters in the past three months than in the previous three years. That creates opportunity and risk at the same time: he worries about losing intellectual muscle, so he keeps asking the Fed economists and bankers who work for him how to use AI to translate the experience of a 65-year-old retiree to the people who are now 25, 35 and 45, rather than waiting until they are 55 or 65 to learn it. Immigration policy matters here too, but the process will run for a decade and ends up net positive for the economy.
— Jeff SchmidThe AI boom is showing up in the price of steel and copper
Asked whether businesses in his district feel a real crowding-out effect from data centers, Schmid answered that he hears about it every day. Data-center demand for machinery, steel, copper and other commodities transmits directly into other industries, and even agricultural futures have moved higher. His method is to peel it apart like an onion, working out what percentage of current growth comes from the AI/data-center boom effect. That is critical to his judgment on whether inflation can be forced back to 2% — not looking at GDP in the aggregate, but opening it up to see which sectors are running hot.
— Jeff SchmidEmployment is already fine, but inflation rules out pivoting now
Schmid says openly that the labor market is in a good place, but the inflation job is not finished. The closer you get to 2%, the harder the decisions become — everyone is worried about overshooting, and the potential errors are being too slow or too aggressive. Inflation has recently run up into the mid-3s, and that has to be dealt with. He calls the dissents at recent meetings thoughtful dissent, and he thinks that is precisely the atmosphere of debate Warsh wants: the Fed has an obligation to manage inflation to 2%, and should be capable of doing it.
— Jeff SchmidA dissenting vote isn't emotion, it's a different weighting of risk
Schmid explains the threshold behind his own past dissenting votes: the FOMC has 19 people, each with a team behind them, and his role is that of a communications transmitter — carrying the worries of business owners and local leaders across the Tenth District's seven states to the table, then carrying the FOMC's discussion back to the district. Dissent is not emotion, it is machinery — a way of saying that in his view the risk weighting between the two goals, full employment and inflation, is different from what the rest of them are estimating. He thinks that is exactly where the FOMC discussion is most valuable.
— Jeff SchmidThe floor under neutral is higher, so policy right now is loose
Schmid says that with the 08 cycle and the pandemic cycle both laid out in front of him, he would not second-guess the policy of the time; the more accurate framework now is that r-star is normalizing, but its base level may be higher than it was before 08. So his personal judgment is that the current level of rates sits in fairly accommodative territory, and the Fed needs to keep working through where r-star and the yield curve stand relative to each other. That line directly supports his position that inflation is unfinished business and monetary policy should not ease too early.
— Jeff SchmidThe Fed shouldn't make the front page; markets have to price risk themselves
Schmid cites Warsh's view: rather than appearing on page A1, the Fed would be better off back on page B12 — meaning a central bank should not crave the front page, but keep a low profile while making the payment system and the technology run soundly. What he wants from the task force is a fresh look at the data sets and at the way the Fed communicates, and clarity about which communication works and which does not; more importantly, the market itself has to carry out the function of pricing risk, and the Fed's communication must not leave markets fragile. He also disclosed that so far there has been no substantive conversation between the task force and the regional Fed presidents.
— Jeff SchmidIn their own words · checked verbatim
we move $ 5 to $ 10 trillion a day through the systems, through multiple different pipes, payment pipes.
Jeff Schmid2:06
if the payment is instant, then there's got to be proven liquidity behind it to settle it.
Jeff Schmid3:07
I think the labor force is in a pretty good place. We haven't done our job yet on inflation.
Jeff Schmid11:16
But we clearly have had a bit of a surge into the mid threes. It has to be addressed.
Jeff Schmid12:18
I'm a bit of a communication transmitter, right? I go around the district.
Jeff Schmid14:29
I think we're at a fairly accommodative place for rates right now.
Jeff Schmid21:34
Very rare. Yeah, I never heard.
Jeff Schmid22:37
Figures
| Pace of baby boomer retirement | about 4 million a year | 5:08 |
| Range of US growth (Schmid's framing) | 2% to 3.5% | 7:09 |
| Move in the unemployment rate (cited by the host) | from 4.3% down to 4.1% | 4:08 |
| Example of a move in the 10-year rate | 80 basis points | 10:14 |
| Height of this inflation rebound | mid-3s (mid threes) | 12:18 |
| 30-year Treasury yield | above 5% | 6:08 |
Glossary
- atomic settlement
- A form of settlement in which the transfer of funds and the reconciliation happen instantly and simultaneously, eliminating float and fee frictions.
- r-star
- The real rate of interest that neither stimulates nor restrains economic growth; the benchmark for judging whether monetary policy is loose or tight.
- reaction function
- The behavioral rule by which a central bank adjusts policy in response to economic data and fiscal/legislative outcomes.
- forward guidance
- Communication in which a central bank commits in advance to a future policy path; Warsh has said explicitly that he dislikes it.
- dual mandate
- The Fed's two statutory goals as assigned by Congress: stable prices and maximum employment.
- dissent
- An FOMC member voting against the majority decision, to express a different judgment about the risk weighting between the two goals.
How to listen
Macro traders and rates strategists tracking the Fed's path, and policy researchers studying how AI infrastructure transmits into inflation and the real economy.
You can skip the first minute of small talk and the last minute of social-media pointers.