A former president of the Federal Reserve Bank of Dallas put the neutral level for US interest rates at 3.75% to 4%, and said the Federal Reserve is within about 50 basis points of it in either direction.
The argument about the September meeting is usually about whether the Fed cuts. Rob Kaplan's answer is that he would raise, and that the part of the bond market investors care about most is not the Fed's to set in the first place.
"Assuming there isn't a surprise, soft surprise on either, I would be going into the September meeting leaning into the thought of raising interest rates in the meeting."
Kaplan sat on the committee that makes this decision, as head of the Dallas Fed. He is now a vice chairman of Goldman Sachs, and ran the firm's Japanese business earlier in his career, which is where the last question of the interview went.
I listened to the full segment so you can skip it.
Here are the 9 takeaways that matter.
👤 Guest: Rob Kaplan, vice chairman of Goldman Sachs and a former president of the Federal Reserve Bank of Dallas, who ran Goldman's business in Japan earlier in his career
🎙️ Host: David Ingles, a Bloomberg Television anchor in Hong Kong, who conducted this interview at the conference
👥 Also on: Doug Krizner, who anchors the Daybreak Asia podcast and introduced the recording
📰 Published: 1 September 2026 on the Bloomberg Daybreak: Asia Edition feed
🔴 YouTube | 🟣 Apple Podcasts | ⏱️ 10 min
Key Takeaways
He would go into the September meeting leaning toward a rate increase Two data points come first: one jobs report that week and one inflation report
Nominal neutral is 3.75% to 4%, which puts the Fed within 50 basis points of it His path is one move in September, nothing in October because of the election, and a fresh look in December
The long end of the Treasury curve is priced off the deficit, not off the Fed Investors hoped the deficit would calm this year; official projections went the other way
Treasuries have stopped rallying when the news is bad Leverage is high enough that money no longer moves into government bonds on a growth scare or a geopolitical event
The AI build-out is inflationary and AI adoption is disinflationary, and adoption is the part that has barely started
China is several innings ahead of the United States in using AI, not in building it Goldman Sachs is studying how McDonald's China and Starbucks China do it
Whether AI reaches the economy depends on where the displaced worker goes Company margins improve either way; national productivity only improves if that worker is redeployed
Japan's problem is not policy, it is that the workforce is shrinking
1. He Would Hike in September
Krizner set the scene from the studio: Goldman Sachs was holding its Asia Leaders Conference in Hong Kong that day, and Bloomberg Television's David Ingles caught Kaplan there. The opening question was what he would do at the Federal Reserve meeting two weeks away.
Kaplan's first answer was a refusal to decide early. "I would strive to keep an open mind between now and the meeting." Two releases come before the meeting, and he named them: "We've got one jobs report later this week and one CPI report."
Absent a soft reading in either, he would arrive at the meeting inclined to tighten. "Assuming there isn't a surprise, soft surprise on either, I would be going into the September meeting leaning into the thought of raising interest rates in the meeting."
2. A Two-Speed Economy
Ingles pushed past timing to size: how many rate increases does the Fed have in it? Kaplan asked to lay one piece of background first, and the background is that the aggregate numbers hide two different economies.
Growth is being driven by capital spending on artificial-intelligence infrastructure, which is where he started the description of the US economy
The split runs along the line of what a business sells. "If you're related to AI or AI adoption or defense, the economy is very strong. If you're related to housing, autos, and the low moderate income consumer, it's very sluggish."
That division is the reason the central bank has waited, and he said so directly: patience is the answer to an economy running at two speeds at once
3. Where Neutral Actually Is
Kaplan then did the arithmetic out loud rather than giving a number of hikes. Neutral is the level of interest rates that neither speeds the economy up nor slows it down, and the real neutral rate is that level after inflation is taken out.
He put the real neutral rate at 0.75% to 1%, with the error on the high side. "If the natural neutral rate, the real neutral rate is approximately three quarters to one percent. It's not less than that. Might be a little more."
Adding the current inflation rate gives the nominal figure. "If inflation is running three-ish, nominal neutral is probably three and three quarters to four."
On that math the Fed is close to neutral, not far from it, and he said policy is either at neutral or slightly stimulative right now: "I think they're right now either neutral or slightly accommodative."
The path he described has one move in it, then a pause. "And so if you're going to take action, I'd move once in September, see how the economy evolves." "And then I would guess they won't act in October. It's right before the election. And then revisit this again in December."
4. Deficits Own the Long End
Ingles put the apparent contradiction to him: the US Treasury is trying to hold down long-term borrowing costs at the same time the Fed may be raising short-term rates, and asked how a market is supposed to read two policies pointing in opposite directions.
Kaplan's answer divides the yield curve by who actually sets it. "So I think the market recognized that the Fed has enormous influence on the front end of the curve." "The back end, I think, is much more driven by deficits, expected deficits."
Investors had expected the deficit to improve this year and it did not. He listed what should have helped — solid if unspectacular growth, nominal income running a little higher — against a war in Iran, and said official budget projections instead point to bigger deficits, possibly because the war is costing more
The extra yield on long-dated bonds is a price for doubt about the fiscal path. "And because of the skepticism, it's requiring more compensation."
Government bond buybacks do not touch the cause. Kaplan said the problem "really is an issue that needs to be addressed with fundamentals" Ingles: "That doesn't sound like a quick fix." Kaplan: "It's not."
5. Leverage Is Global
The debt problem is not a US peculiarity. "And it's not just a U.S. issue." He named the others in a row: "Japan has got a significant leverage issue. Europe does. China does."
What makes it harder in the United States is demographics. He said labor force growth there is very sluggish, which leaves productivity as the only route to the growth that would improve the ratio of debt to output
The fix he named has two halves and no shortcut. "And so you need productivity growth and higher GDP from productivity to help bend this curve and you need to control spending." He said it could take a number of years
6. No More Flight to Quality
Ingles asked whether a long-term interest rate of 5.5% is now a permanent feature: "Should we call it higher for longer? Should we get used to the long at 5.5%"
Kaplan would not forecast the level. "So, I don't know is the answer."
What he said has changed is the behavior of Treasuries in a crisis. Investors used to buy government bonds when growth weakened or a geopolitical event hit, and he said that reflex has weakened "This level of leverage is meaning people are not jumping into Treasuries when growth weakens or there's a geopolitical event."
The market would pay up for evidence rather than announcements, and he listed what would count: changes to entitlement programs, faster labor force growth, visible productivity gains and demonstrated spending discipline
Until then he expects long-term yields to stay where they are. He said the back end of the curve will be sticky for that reason
His inflation view cuts both ways on AI. "I still believe that Chinese overcapacity and over the horizon AI adoption, the infrastructure push is probably inflationary. The adoption push, which we're early in, should be disinflationary." "So I do think that there's still a good prospect over the next few years for inflation to trend down toward 2%." Either that shows up as a steeper curve, he said, or investors conclude they are already being paid well to own long-dated bonds
7. China Leads on AI Adoption
Ingles turned to China, noting that Kaplan had been there when the economy was opening up and that it is a different country now — "Let's call it 35, 40 years after the 90s" — and that Kaplan had said before the interview began that China is ahead on AI, which might explain why its interest rates are low.
Kaplan separated building the technology from using it. "So I think the U.S. has got enormous advantages in semiconductors, a lot of the infrastructure capability, memory that powers AI. But China is a couple innings ahead in adopting AI. I would say the U.S. is in the first or second inning."
He expects broad implementation and better productivity in the United States to follow, as companies there develop and learn from more uses of the technology
The learning runs the other way for once, and he named where his own firm is looking. "China's a few innings ahead of that, and I think there's a lot to be learned from companies in China with how they're using AI, and I think U.S. companies, including Goldman Sachs, we're studying some of the things we're seeing in McDonald's China, Starbucks China, other businesses that are further along."
8. AI Needs Redeployed Workers
Asked directly whether AI adoption is a deflationary force, Kaplan gave a two-part answer: yes in principle, and not yet in the data.
"It should be disinflationary." What he sees at this stage is the gain landing in company results rather than in prices
He expects a multi-year improvement in S&P 500 earnings from both the infrastructure spending and the adoption
The open question is whether that reaches the wider economy at all. "The key to AI in the economy is, where's the worker go?" A company whose driver loses his job is more productive and earns more either way "For the economy, that worker needs to get redeployed in another productive job."
His prescription is a national retraining effort, and he was specific about it: early childhood literacy, secondary education, skills training, and a focus on redeploying people displaced by AI
He is optimistic but not complacent about it. "I'm confident that we can do it, but it probably needs a real hard look."
9. Japan Is Still Shrinking
The closing question came from Kaplan's own past running Goldman Sachs's Japanese business: does he remember the exchange rate from then?
He could not, and said so rather than producing a number. He guessed that "it was probably in the neighborhood of if we're 160 now, we might have been, I'm guessing, 100, but I have to go back and look." Ingles's point was that Japan is now far cheaper in dollar terms than at any time he remembers
Kaplan listed the genuine improvements first. "So Japan's got a number of very positive developments. They're redeploying savings into more risk assets. There are governance reforms."
Then the constraint, and it is the same one he applied to the United States, only worse. "The challenge they have, same, they're highly leveraged at the government level and their workforce is not just decelerated, it's shrinking."
Bonus Insights
This was the second annual running of the event. Bloomberg's own notes for the episode say the Goldman Sachs Asia Leaders Conference followed a first edition last year that drew major regional and international firms
This conversation is the second half of the episode. The first half is Doug Krizner's interview with Adam Coons of Winthrop Capital Management on oil, the Bank of Japan and food inflation, and it is written up on its own
Kaplan's bottom line is that the Fed should raise rates once more in September and then wait, and that nothing the central bank or the Treasury does will fix long-term borrowing costs, because those are priced off a deficit that only slower spending and faster productivity can bring down.
Products, Companies & Tools Mentioned
Goldman Sachs (Kaplan's employer, host of the Hong Kong conference where this was recorded, and, he said, a firm now studying how Chinese companies use AI)
McDonald's China and Starbucks China (The two businesses he named as further along in AI adoption than their US equivalents)
S&P 500 (Where he expects a multi-year improvement in earnings from AI — and the gap he says still has to be translated into economy-wide productivity)
If this was worth your time, send it to someone who has to have a view on this.
Get the latest market chatter as it happens: