Tom Barkin, president of the Federal Reserve Bank of Richmond, joins Tracy Alloway and Joe Weisenthal at Jackson Hole to go through Kevin Warsh's speech, the reasons American consumers keep spending, and what he hears about AI at chamber of commerce meetings. They also get into the dot plot, tariff refunds, why a data center leaves no political constituency behind, and whether the Fed's inflation record is one long failure or two shorter episodes.
👤 Guest: Tom Barkin, president of the Federal Reserve Bank of Richmond, who fills the role by traveling his district and collecting on-the-ground color from businesses
🎙️ Hosts: Tracy Alloway and Joe Weisenthal, the Bloomberg journalists who host Odd Lots
📰 Published: 31 August 2026 on Bloomberg · recorded earlier
🔴 YouTube | 🟢 Spotify | 🟣 Apple Podcasts | 🔗 Episode page | ⏱️ 31 min | ✅ Time saved: 8 min
Key Takeaways
He was aligned with Warsh's speech, and says no September signal should be read out of it
"I thought he had a very accurate sense of the economy."
"So I don't think you should take any forward guidance from a speech from a guy who doesn't give forward guidance."
Inflation is not heading in the wrong direction; it is in the wrong place
"And I'm not saying inflation is definitely heading in the wrong direction. It's just not in the right place."
The job market is good "but it's not frothy", and the low-hire, low-fire pattern continues
Consumers are not richer, they are borrowing from the future
Auto borrowers run 60 days delinquent rather than 120 so they do not lose the car; households skip the gas bill in summer
Private label, Walmart and dollar stores are where the trading down shows up, with the savings rate falling
The construction cycle rotated between sectors rather than growing
Office buildings are not being built, home building is okay but not great, industrial is starting to come back
Multifamily developers blame interest rates; he points out rates were the same in 04 and 05 when they were building
The number one effect of AI in his district is political, not economic
Every chamber of commerce meeting and town hall brings questions about jobs, water and data centers
Substitution is real only where the structure allows it — call centers, programming, heavy documentation and compliance
Where AI does bite is hiring: fill the job with AI first, hire later
A world of repeated inflation shocks needs a different setting, not a different destination
Fracking, demographics, e-commerce and globalization pulled inflation down 10 years ago, and he says a comparable set of factors could push it up now
His sailing analogy: same destination, tighter sheet
He will keep producing his own forecast even if the Fed stops publishing one, because the dot plot swamps the narrative
A Greensboro audience hears that the Fed promised two more rate cuts this year, which is not what a set of independent forecasts says
The 65-month indictment has a defensible alternative, which he is careful not to endorse
"You can take a 65-month view or you can take a, I'll get the months right, a 47-month view followed by an 18-month view."
Tariffs went quiet because companies are collecting refunds rather than paying, and the refunds land in earnings rather than in price
"But I think it's a lot more marketing, store refits, staffing levels."
B2B companies say to a person that they are passing tariffs through, while the suppliers into the big box retailers cannot
A data center brings tax dollars and no political base
A manufacturing plant puts its workers' kids on the baseball team; a data center employs very few people once construction ends
His tease: name the elementary school after Microsoft or Google
Nobody knows how big the build-out gets, and the one certainty is that the number will be wrong
"And so we don't know whether we're going to get the number too many or too few."
The range of AI productivity outcomes is too wide to set policy against
"And there's 18 different versions in between."
What he takes to the chair is a synthesis, not an anecdote
"I think the anecdote doesn't, everyone likes a good story, but it doesn't really move policy."
Warsh's Speech Was Authentic and Accurate, and Carries No September Signal
Alloway opened on the speech the symposium had just heard. Her framing: "We just got the speech from Fed Chair Kevin Warsh, which I think most people would describe as hawkish, although maybe there's kind of a gap in between the hawk and the ish part." Weisenthal wanted the on-the-ground color that the theoretical questions at Jackson Hole — the neutral rate, AI and productivity, the term premium — do not supply. Barkin was there for the third year running, and noted the venue had improved: "We get the Tetons in the back, so this is great."
He put himself alongside the chairman without reservation. "He's a great speaker, and I thought it was a very authentic speech." The people he had talked to appreciated the clarity, and "I thought he had a very accurate sense of the economy. So I was very much aligned with what he said, and I thought he said it well."
On the economy, he stacked up the strength. GDP and consumer spending are solid, and he called two things crazy: "I mean, it's been crazy this year that gas prices went up and consumer spending accelerated." and "It's been crazy that you have all this uncertainty and artificial intelligence spending has led business investment to almost double versus this historic thing."
On inflation he drew a distinction the whole conversation runs on: "And I'm not saying inflation is definitely heading in the wrong direction. It's just not in the right place."
The jobs market has stabilized and unemployment is low, but "I also would say the job market's good, but it's not frothy. I mean, this low-hire, low-fire thing continues."
The read-across to September is the one thing he refused to give. Weisenthal said the first question everyone asked him after the speech was what it meant for September. Barkin's answer: "I said, well, I listened to his speech and I'm pretty sure he doesn't like forward guidance. So I don't think you should take any forward guidance from a speech from a guy who doesn't give forward guidance."
Alloway's objection was that "We can't help ourselves is the problem."
His reply was that the two things separate cleanly — talk about the economy, then decide whether to give guidance. "If you choose not to do forward guidance, you don't do forward guidance."
Why Consumers Keep Spending: a Creative Consumer Borrowing From the Future
Alloway asked for his working theory of the resilience, given high prices, rising gas and persistent uncertainty. Barkin answered with a comparison between the two most recent recoveries.
Coming out of the Great Recession, he said, people had lost jobs, houses, cars and savings and had retirement to rebuild, which produced "We had five, six, seven years, the secular stagnation where people weren't spending the way you thought they would." Coming out of COVID was the opposite: money in pockets, stimulus, unspent savings, equity values up, home values up. "And I think you had a mindset that just says, I'm bound and determined to spend."
The wealthy are spending their wealth, but the interesting behavior is below that. "But even those with less wealth, what I hear is a very creative consumer figuring out ways to find money, to spend money, to borrow from the future."
Where the money is being found: the growth of private label and the move to Walmart and the dollar stores, visible in those companies' results, and people dropping insurance cover
Two anecdotes from lenders and utilities carry the point. "And I've talked to auto lenders who talk about people being 60 days delinquent, not 120 days delinquent, because they need to find the money, but they don't want to lose their car."
"I've talked to gas providers who say people aren't paying the gas bill during the summer because no one's going to have a problem with gas in the summer. It's the winter that matters."
The conclusion is a conditional one, and the condition is jobs and markets. With the savings rate down and the spending impulse intact, "And I really do think as long as the markets are healthy and people have jobs, they're going to keep finding a way to spend."
The Construction Boom Rotated Between Sectors Rather Than Growing
Weisenthal went back to the road trip the show made with Barkin through North Carolina in 2024, where the scarcity of skilled trades had come up, and asked whether the AI build-out is now crowding out everything else. He noted that Austan Goolsbee had told the show a couple of days earlier that he hears the same complaints but was not sure how much of it traces to data centers.
Barkin agreed the constraint is real. "Yeah, so it's been a monumental construction investment cycle. $700 billion announced in one week, I think, at the beginning of February alone." — and "And for sure, if you're trying to find switchgears or transformers or electricians, they are very hot and in very short supply."
The aggregate picture is a rotation, not an expansion. Office buildings are not being built, industrial is starting to come back, and home building is "okay, not great", so the money has moved from one part of the construction market into another
He is skeptical of the interest-rate explanation developers give him. When multifamily builders tell him a project will not pencil, they want to talk about rates, and his answer is "Because we had the same interest rates in 04 and 05 and you're building lots of buildings."
Pressed, they acknowledge construction costs are up, labor costs are up and more equity has to go into projects
Whether that counts as crowding out is genuinely unsettled in his mind. Data center construction makes other construction more expensive, which means the other construction does not happen because it is more expensive. "So there's a chicken and egg question in there."
In His District, AI's Number One Impact Is Political
Alloway asked what he actually sees from AI inside the Fifth District — in the labor market, in the low-hire, low-fire pattern, or in prices.
His first answer is not an economic one. "I mean, the number one impact is a political impact." Every chamber of commerce meeting and every town hall brings questions about jobs, water and data centers
"And you can see when you travel the issues on people's minds by the questions that they ask."
The productivity effect is real but narrow so far. He does not see a massive productivity impact yet outside places with a structure that lets an agent be substituted in — call centers, programming, heavy documentation and compliance documents
The productivity boom that is showing up is significant, and he traces it to an earlier round of investment, when firms were short of workers and responded with automation, new staffing models and different operating processes, and are reaping the benefits today
Owners and executives are enthusiastic about AI and workers are less so, and it is still mostly used as an add-on to get a job done better and faster
The place it does bite is hiring. Employers tell him they do not know what the future looks like, and: "I wonder whether AI can do that job. And so why don't you leverage and see if you can't fill the job using AI first and then we'll hire later."
Housing Is Helping on Inflation Now and Is Set Up to Stop
Alloway put it to him that a soft housing market is working in the Fed's favor on the dual mandate, and asked whether that persists given weak construction.
He expects the tightness to come back. He called it "highly likely" that it returns: a whole generation wants houses at prices that are not quite right for them, and at some point they have a second child and need a bigger one
"And then if you don't have the housing inventory, whether it's rental or single family, then the prices will go back up."
He argued against picking the inflation data apart, using his own inbox as the example. "It's very tempting to do that, to torture it and get the answer you want." When market rents were falling faster than the CPI measure, real-estate developers emailed to tell him his numbers were wrong; he is not getting those emails now
The frame he prefers is the whole basket. Housing prices may come down while something else goes up: "And you're looking at inflation as a total basket. That's how I like to think about it."
Normal Today May Not Be the Normal of 10 Years Ago
Alloway noted Warsh had talked about the breadth of inflation and about one-off shocks such as the Iran war and its effect on gas prices, and asked whether the environment is simply more inflationary when the shocks keep arriving.
His answer reframes the question as a question about the baseline. "And I think the question is, is normal today or was normal 10 years ago? And I think normal may be a lot more like today than it was 10 years ago."
The disinflation of a decade ago had identifiable causes, and he listed them: fracking bringing energy prices down, demographics that supplied workers and kept wage costs under control, e-commerce cutting prices on what people bought online, and globalization
"I don't think it's ridiculous to imagine that 10 years later you might have a bunch of factors that would bring inflation up."
The policy conclusion is about the setting, not the destination. "You know, you just sail differently if you've got the wind behind you than if you're sailing into the wind."
"You just have to lean against that wind."
The realized inflation still depends on what the Fed does about it, which is the point of the analogy
Two Thoughts on Forward Guidance, Held at the Same Time
Weisenthal raised the task forces that may revisit the Fed's communications, and asked whether everything should be on the table — the dots, the press conferences, even this interview. Alloway backed him up.
Barkin's answer was explicitly two-sided, on the model of holding two opposed ideas at once. The first: "One is, if we're relatively transparent about how we think about things, that helps build credibility with the public. It helps build trust in the institution. And it helps markets do some of the work for you. That's the famous Bernanke theory."
The second is Warsh's case, which he endorsed in full. "I also 100% agree with Chairman Warsh when he says, sometimes if you give too much forward guidance, you get stuck in it and you end up, having to make a suboptimal decision because you've misguided."
He named the episode: "And I think it's fair to say that's part of the 2021, 2022, story, which is we had very strong forward guidance in place and it was very hard to get your way out of it."
On whether he should be talking to the show at all, he was unambiguous: "And I hope to keep coming and talking to you as long as you'll have me."
The Dot Plot Is a Picture That Overwhelms the Story
Weisenthal pushed on the specific techniques rather than the principle.
He wants to keep making his own forecast whatever the Fed publishes. He likes negotiating and arguing with his team over the projections, setting his policy bias against his economy bias, and says the process sharpens his thinking. "So regardless of whether we release it or not, I plan to keep doing an SEP because I think it, having your own forecast and working against the forecast is a pretty healthy thing."
His objection is to the chart, not the exercise. "The one thing I don't like about the SEP is I think the dot plot itself is a picture that overwhelms the story."
The test he applies is what a business audience takes away. If someone at a chamber of commerce in Greensboro tells him the Fed "promised two more rate cuts this year", he counts that as a communication failure, because the projections are individual forecasts rather than a promise: "We've done a set of forecasts independently."
"And if what's happening is that picture is swamping the narrative, then we're going to think differently about the picture."
One 65-Month Failure, or Two Shorter Episodes
Weisenthal noted that Warsh had said the responsibility for inflation running above target — he put it at 65 months — sits squarely with the Fed, and asked Barkin, who was in the room for that stretch, for his reaction.
He accepted the standard. "Well, I'm definitely 100% insistent on getting inflation under control. And I think it's fair to hold that mirror against what we're doing and ask ourselves the question of whether we're doing it the right way."
Then he laid out an alternative accounting, and was careful to say he is not the one making it. "You can take a 65-month view or you can take a, I'll get the months right, a 47-month view followed by an 18-month view." — "And if you do the second, which I'm not arguing for, I'm just saying it's a perfectly defensible way to do it"
The story in that version: inflation happened, the Fed was maybe a little slow, rates went up, inflation came down. "If you go to March of 2025, you've got 2.3%, 2.4% inflation, and everything seems to be headed in the right direction."
"The plane is going to land." Then a series of external shocks — AI, tariffs, oil price increases — took inflation back up
The counter-case he expects to argue with colleagues is blunter. "On the other hand, you say, don't give me your excuses. It's been 65 months and it's been over. And, maybe rates aren't that restrictive and maybe you have to think about those. And that's the argument I'm sure we're going to have."
Tariffs Went Quiet Because the Refunds Are Rolling In
Alloway asked what manufacturers in his district are saying about tariffs, which have dropped behind AI and the oil shock in the conversation.
The effect splits by where a company sits. "I mean, if you're a steel or aluminum manufacturer, that's created a price umbrella that's helped your industry." For a company moving operations into the country there is a real argument for them too
The most poignant cases, in his word, are foreign manufacturers with assembly operations, many of them in South and North Carolina. They moved manufacturing to the United States, but the components are made in Europe and shipped in for assembly, so they are still getting tariffed
The quiet has a mechanical explanation. The headline numbers went up, the collections were never as big as the numbers, the Supreme Court ruled and refunds followed. "So one of the reasons you're not hearing a lot about it is for the last three or four months, people have been collecting refunds as opposed to paying more tariffs. And when you're collecting, you're not talking about it quite as much when it's working to your advantage."
The Refunds Are Stimulative, But Mostly Not Through Price
Weisenthal asked whether the refunds are stimulative, since the money has to go somewhere.
"They're very positive for earnings." Read through the earnings reports of companies that have collected, he said, and you hear a lot about reinvesting it in the customer
What that reinvestment mostly looks like is not a lower price: "But I think it's a lot more marketing, store refits, staffing levels."
Stronger earnings make companies more likely to hire and more likely to invest, so the effect is stimulative
On whether it reaches price, his answer is narrow: in very targeted ways, but not in a massive system-wide way
Pass-Through Works Business to Business and Stalls at the Big Box
Alloway noted the original debate about whether stretched consumers would stop companies passing tariffs on, and asked how he sees it now that spending has held up.
The B2B answer is unanimous in his sample. "So the B2B companies I talk to, to a person, they're convinced they're passing it through." Costs came in, they had to pass them on, they know the other side does not like it, and they say they have been pretty successful
The B2C answer has changed since the spring of 2025, when companies assumed they would pass it through. Many had trouble. It is easier serving wealthy customers and harder serving less wealthy ones
Suppliers into the big retailers are where it jams. "But those people who sell into the major big box retailers, they tell me they're having a devil of a time trying to pass it through."
"And the mindset of those retailers is I need to find some price to give to my customers."
So the consumer pushback is real, and it is concentrated: most real in B2C, and most real in low- to moderate-income B2C
The reconciliation with good earnings is productivity. There has been very little wage pressure, and companies are driving productivity through the set of things he says they launched three or four years ago
Weisenthal's gloss: "Those big box retailers, they're truly our strongest soldiers in the fight against inflation, holding the line on prices on behalf of the customer." Barkin tied it back to the 2010s, when the rise of the big box retailers and of private label brands helped keep prices under control and spending okay — prompting Weisenthal's "You're saying we didn't know how good we had it."
A Data Center Brings Tax Dollars and No Political Base
Weisenthal described the picture he carries of data center politics — anxious residents, a mayor or town manager explaining that it will bring the property tax bill down — and asked whether that is the shape of it.
Barkin's explanation is about who ends up in the town. If there is a manufacturing plant in the town, the workers' children are on the baseball team, the football team and the hockey team. "And if you have a data center, they don't have very many employees."
The two sides talk past each other. "You've got the economic developers talking about how great this is for the tax base, and the citizens saying whatever version of, I don't like what it's doing to water, I don't like how it looks, I don't trust AI. It's been politicized in that way."
The structural point: it brings tax dollars, but not enough workers after the construction phase for citizens to have friends who work there, so there is no political base for it
His half-joking remedy is a marketing one. "I tease sometimes that they ought to name an elementary school after Microsoft or Google or whoever and say, here's Google, they just brought you this." — and he added that this is not how people think about it
The One Thing the Fed Knows Is That It Will Get the Number Wrong
Alloway asked whether political pushback against data centers becomes big enough for the Fed to have to pay attention, given how much of both the price pressure and the growth runs through the build-out.
He framed the whole thing as a forecasting problem with an admitted error term. "We're going to grow the data center footprint massively. We have no idea how massively." The energy footprint grows too, and "And the one thing we know is we're going to get those numbers wrong."
"And so we don't know whether we're going to get the number too many or too few." A backlash might turn out to be economically useful, by preventing an overbuild that would otherwise have happened, or it might tip the country into underbuilding
"So, yes, it could have a big impact, but I need AI forecasting skills to help me figure out how big this thing's going to be because you're trying to meet a very significant moving target."
Weisenthal offered a source of those forecasting skills: "Have you tried asking ChatGPT what it thinks? Just type it in. Just ask."
Barkin had already tried it on a different task: "I've also asked ChatGPT to try to write a speech for me, and I didn't think it did a very good job."
Alloway's verdict: "It could never capture your voice."
Another Road Trip, and What They Would Ask This Time
Barkin has driven through Mount Airy since the 2024 trip without stopping, and had been in Greensboro that week
Alloway proposed a return, on the grounds that the questions themselves have changed. She wants to know about the state of housing and how the manufacturer making synthetic textiles is doing; Weisenthal added the carport companies
Barkin offered West Virginia the following week, which Weisenthal was up for and Alloway was not, being on vacation
The Range on AI Productivity Is Too Wide to Set Policy Against
Alloway asked how far ahead the Fed should be looking for the productivity boom that would give it room on R-star and the neutral rate, given that the most tangible effect of AI right now is higher prices.
"I think it's really hard to make a forecast of what it's going to do a year, two, three years from now. Like I say, the range of possible outcomes is very wide."
The channel matters as much as the size: how the outcome plays out in prices, or in borrowing rates because of crowding out of capital, or in labor employment
He cited the chairman's version of the same point from the day before — that there are models of AI that end up looking very inflationary and require moving one way, and others that look very distressing for the economy and require moving the other. "And there's 18 different versions in between."
The conclusion is a bar on acting early: "And so we can argue about the direction, but I think you've got to get you know, some more confidence before you could make too much policy based on an assumption."
Warsh's Meetings Versus Powell's: 'A Family Fight'
Weisenthal prefaced his last question by conceding it would not work — "I just have one last question and I know you're going to play it coy, but so I will not ask you, you know. What are we doing at the next meeting?" — and switched to whether the two meetings under Warsh have felt substantively different from Powell's.
The meetings themselves look much as they did. They are relatively structured, they will be reviewed along with everything else by the task forces, and "And the chair has shown up and, led very well."
He named the chairman's own phrase for the style he encourages: "And I think, he does like this idea of a family fight is a phrase he likes to use." and "We certainly have that."
Asked by Alloway whether Powell meetings were also a family fight, he gave the line of the episode: "There are a lot of different families."
Weisenthal reached for the Tolstoy line about unhappy families and could not retrieve it — "I'm thinking of that. What's that Russian quote?" — and Alloway got as far as "Yeah, all unhappy with something."
What He Takes to the Chair Is a Synthesis, Not an Anecdote
Alloway's closing question was about his own method: whether Warsh is interested in different anecdotes or data series than Powell was.
He does not bring the anecdote. "Well, so what I try to do is I try to come up with a synthesis that backs off from individual anecdotes. I think the anecdote doesn't, everyone likes a good story, but it doesn't really move policy."
What travels is the synthesis — the consumer spending picture, and the difference between what B2C and B2B companies are reporting. Both chairmen, in his impression, have been very attentive to it
The reason the job exists in that form is access. "You know, they live more in a cloister. I get to go out a lot more than they get to go out because the press follows them everywhere."
The discipline he sets himself is to tie the ground-level material to the open questions: productivity has expanded, what is driving it, how long it lasts, underlying inflation, what is happening in the consumer world
The Hosts Afterward: the De-saving Story Is Underrated
Alloway's takeaway was the consumer. "You know what I thought was really interesting is his description of the creative ways consumers are continuing to spend, which actually I don't think has gotten the attention probably it's deserved, right?" The talk is all about the inflationary impulse from business investment and about oil, not about de-saving and borrowing to keep consuming
Weisenthal has been carrying a different theory and let it go. "I always thought the strong consumption was like maybe sort of like a nihilistic response to the pandemic experience. But actually you don't see it that much in the savings rate."
The savings rate has gone down, both agreed, but it has not plummeted
Alloway's test for panic buying is a specific one, and the spending does not pass it: "No, it doesn't feel like the sort of spending that you would expect before the upcoming hyperinflation where everyone's going out and buying silver candlesticks."
Weisenthal, unprompted: "I do have silver candlesticks."
They made the case for the return trip on how much the questions have moved. "And I do think going back to Mount Airy next year would be a really good idea because if you just think back to 2024, I mean, we didn't ask a single AI-related question, right?"
"If we would have asked the textile company, are you using ChatGPT in your daily operations, I don't think they would have had a response."
Barkin's bottom line is that the economy is strong for reasons that do not settle the policy question — a consumer who keeps finding ways to spend, a construction boom that has moved from one sector to another rather than grown, and an AI effect that so far shows up in town hall meetings and hiring plans rather than in measured productivity — and that the argument about whether rates are restrictive enough is one the Fed has yet to have.
Products, Companies & Tools Mentioned
Walmart and the dollar stores (Where he sees the trading-down showing up, alongside the growth of private label, as consumers find money to keep spending)
ChatGPT (Weisenthal's suggested source of AI forecasting skills; Barkin had asked it to write a speech for him and "didn't think it did a very good job")
Microsoft and Google (The names in his tease that a town ought to put on an elementary school, since a data center brings tax dollars but almost no local workforce)
Books & Resources Mentioned
Kevin Warsh's 2026 Jackson Hole keynote (The speech the episode opens on, including the line that responsibility for 65 months of elevated inflation sits with the Fed)
The FOMC's Summary of Economic Projections and the dot plot (The exercise he says he will keep doing whether or not it is published, and the chart he says overwhelms the story)
Odd Lots: On the Road with Richmond Fed's Tom Barkin (The 2024 Mount Airy road trip the hosts want to repeat, with no AI questions in it)
Austan Goolsbee Is Worried the Economy Is Overheating (The Odd Lots interview from days earlier that Weisenthal quotes back on skilled-trades scarcity and the AI build-out)
Get the latest market chatter as it happens:
If this saved you the listen, send it to one person who would rather have the time back.