The reason long-term interest rates keep rising, on Torsten Slok's account, is that the world ran out of spare savings. For a decade there was more money than there were projects. Now there are more projects than there is money.
The standard explanations for a 5% 10-year Treasury yield are inflation and deficits, and Slok agrees both are in the list. What he puts first is the mirror image of the global savings glut Ben Bernanke named after the financial crisis — data centers, the energy to run them, defense and government borrowing all bidding for the same pool of capital, and winning it by paying more.
"Today we now have a savings shortage."
Slok is Partner and Chief Economist at Apollo, spent 15 years on the sell side before that, and worked at the OECD in Paris and at the International Monetary Fund writing the World Economic Outlook — and he was at Princeton in 1995 and 1996, when Bernanke chaired the economics department.
The full episode is covered here so you can skip it.
Here are the 10 takeaways that matter.
👤 Guest: Torsten Slok, Apollo's Chief Economist, a Partner at the firm, previously at the OECD in Paris and at the International Monetary Fund
🎙️ Host: Jeffrey Roach, Chief Economist at LPL Financial
📰 Published: 15 September 2026 on YouTube · recorded 14 September 2026
🔴 YouTube | 🟣 Apple Podcasts | 🔗 Episode page | ⏱️ 33 min | ✅ Time saved: 19 min
Key Takeaways
AI will not cost jobs on net, because business creation is running at the highest level in US history
The displacement effect arrives slowly and is harder to implement than people assume
Entry-level hiring is the opposite of the scare story: the unemployment rate for 20-24 year olds has fallen more than the overall rate
Companies adopting AI show higher growth in entry-level jobs, not lower
The world has flipped from Bernanke's savings glut to a savings shortage, and that is why long rates keep rising
Data centers, energy, defense and government deficits are all competing for the same capital, and they compete by offering a higher yield
The list of reasons rates should fall is shorter but real: petrodollars, aging populations, and an AI disappointment that pushes money out of tech and into bonds
Inflation peaked at 10% in the pandemic and is now about 3.5%, and the last mile is the hard one
Consensus has it still at 3% in the middle of 2027, against a 2% target
The market has gone from pricing cuts in 2026 to pricing four hikes — September, December, March and June
If rates stay higher for longer, value beats growth, because growth's cash flows sit furthest out in the future
1. AI Is Not Coming for Jobs
Roach opened with the question he says advisers ask most: is AI coming for your job?
Slok's answer was flat. "No and the answer also is very important that is not going to displace a dramatic amount of workers."
He expects real automation inside financial services, consulting and legal work over the next five years, and says the job losses from it are small next to what is happening on the other side of the ledger.
The other side is new firms. Weekly census data on business formation shows more new businesses opening every week than at any point in US history: "In other words, business creation is at the highest level ever in US history."
If even a fraction of those new businesses hire, he said, that is likely why the labor market is still doing so well.
The displacement is also slower and harder than the headlines suggest — automating a process is more complex and more challenging for businesses to implement than deciding to do it.
His argument for why investment professionals specifically are safe is that the work is not one thing. "a lot of jobs are messy" — a person doing 20 or 30 different tasks a day may see some automated and still be left with talking to clients, to management and to investors.
"So for that reason, I'm very very optimistic that AI is actually both going to have a positive impact on productivity and it's also going to have a positive impact on employment because the displacement effect is being offset by business creation being so strong at the moment."
Roach tied that to the Federal Open Market Committee's new chair, Kevin Warsh, who has talked about the benefits of productivity and the boost the economy might get from AI, and pointed listeners to MIT's David Autor as the best writer on what the technology does to the labor force.
Roach's own summary of the advice: "So be more human is the answer I think to those nervous about the AI story."
2. Entry-Level Jobs Are Fine
Roach asked how AI hits someone with 15 to 20 years of experience against someone with two, and said he had watched people over 55 leave the labor force after the pandemic and go and hang a shingle.
Slok accepts the question is the right one — whether entry-level jobs are the ones being threatened — and says the data has so far said the opposite.
Spending data from Ramp shows that companies adopting AI tend to have higher growth in entry-level jobs. He was careful with the causation: it may be that AI adopters are simply higher-growth companies that grow faster anyway.
The decisive number is the age cohort. The unemployment rate for 20 to 24 year olds has dropped by more than the general unemployment rate.
"So yes it may be that young people who come out of college cannot get the preferred job that they want but they always actually end up getting a job and at the moment actually more of them have been getting jobs."
He still worries about what entry-level work looks like further out, and said so twice.
The mechanism he credits for the strength is that starting a business has never been easier: dorm-room entrepreneurs, people dropping out of college for AI jobs, and a graduate who can sit in a basement building agents rather than having to find a well-established firm in law, consulting or finance.
That dynamism, in his reading, is why the latest non-farm payrolls have stayed strong and why the unemployment rate is still 4.1%.
3. Why Generalists Win
The AI topic closed on a book recommendation aimed at anyone in college, or anyone who knows someone in college.
The book is Range, whose argument is that generalists do better in a specialized world, and it was recommended on the show as a fascinating read.
The reasoning given for why it applies now is occupational: the job most obviously threatened by AI is coding, and a software programmer is characterized by having one task and one task only.
Someone in sales, marketing, back office or legal services does many different things, so the more of a job that is a single task, the more exposed that job is.
The conclusion drawn from it is about training rather than employment: people going through their education should think across disciplines rather than only down into one.
4. Savings Glut to Shortage
The second topic was rates, and it started with Alan Greenspan and Ben Bernanke's mid-2000s phrase, the global savings glut. The argument is that we are now living through its opposite.
The setup was the decade from 2010 to 2020. Interest rates were zero because central banks were trying to boost the economy after the financial crisis, and cleaning up the housing market, the banks and household balance sheets took a long time.
On top of that came the recycling. China was exporting to the United States, accumulated dollars and had to put them somewhere; Europe, Japan, Australia and Canada all had savings too, and there were simply not enough projects to invest in.
Slok's flip: "Now we have an AI miracle, an AI boom. And now we have the total opposite. Now we can't invest fast enough in data centers. We can't invest fast enough in energy. We can't invest fast enough in defense." Nor, he added, in government bonds, because deficits are very high.
"Today we now have a savings shortage."
The mechanism is a competition for capital, and the currency of that competition is yield. Projects that need building now have to outbid each other for the money that exists, and they do it by offering more.
"So that's of course why yields on hyperscaler debt has been widening."
"That's why the yield on interest rates from the government has been going up because there's not enough money to buy all the Treasuries that are being issued."
The consequence he draws is directional rather than cyclical: as long as there are this many projects everyone wants to fund, rates stay higher for longer — and he thinks they are more likely to go higher still. The 10-year Treasury yield reaching 5% is the marker he used for where that has already got to.
5. The Greenspan Conundrum
Roach's contribution was to name the other half of the same coin, from the other end of the cycle.
Greenspan's conundrum was short and long rates staying surprisingly low in the early and mid-2000s, and the answer economists settled on was the savings glut — too much saving chasing too little investment.
Roach's reframing of today: "You could say where we are 2026, there's this massive amount of investment that dollars are wanting to flow to."
"Capital is going to flow to where it's used best returns." With that much investment demand, there is not enough saving to keep up.
He called the flip side of the Greenspan conundrum an underappreciated explanation for the rise in the term premium — the extra yield investors demand for lending long — alongside inflation and the rest.
6. The Case for Lower Rates
Roach then argued the other way, and Slok took the list on.
Roach's two candidates for forces still pushing rates down are the oil exporters, whose petrodollars have to be recycled while crude is where it is, and the aging of the developed economies, with a sharp fall in the prime-age workforce coming over the next 25 to 30 years.
Slok's list of why rates should rise ran to four, plus oil: the shift from a savings glut to a savings shortage; inflation, which is why the Fed is contemplating hikes; fiscal problems pushing up the term premium; and hyperscalers issuing so much debt for data centers that they crowd out other investment-grade borrowers, US Treasurys included.
"So the list of reasons why rates should be going up is relatively long."
He accepted both of Roach's offsets — higher oil means Middle Eastern countries have more dollars to recycle into Treasurys, and an aging population pushes the same way.
The third argument for lower rates is the one most people would file as an equity risk. If the hyperscalers cannot generate the revenue the market expects, technology valuations come down, and if open-source models take share from closed ones a correction in AI names follows — and some of the money leaving those stocks goes into bonds.
The summary position did not move: "We still think that rates in summary will stay higher for longer."
7. The Last Mile of Inflation
The third topic was inflation, and Roach said LPL's own projections do not have it getting materially weaker until late in the first quarter or early in the second quarter of next year, which is a forecast that has already been pushed out. He called the present an inflation fog.
Slok's starting point is how much has already been done. "So, if we just back up, remember as we all know on this conversation that in the pandemic inflation basically peaked at 10%." It is now about 3.5%.
"So, we've gone a long way, but the problem is that the last mile is going to be very, very complicated and going to be very sticky."
Three things are holding it up: oil prices driven by the conflict in the Middle East, the delayed effects of tariffs, and restrictions on immigration.
"Immigration restrictions putting some upward pressure on wage growth in agriculture, construction, hotels and restaurants." Those are the sectors where unauthorized immigrants normally work, and they have seen labor shortages.
His timing is later than the consensus discussion suggests: inflation does not begin moving meaningfully toward the 2% target until the early or middle part of 2027.
The number that follows from it is the problem. Consensus has inflation at 3% on both headline and core in the middle of 2027, against a target of 2%.
That gap is what the committee is arguing about meeting by meeting. Some members think rates should rise, some think they should rise a lot, some think not much and some think not at all.
His conclusion is that the assumption has to cover both ends of the curve: not only are long rates higher for longer, short rates are too.
8. Boomers Are Still Spending
Roach's addition to the inflation picture was the demand side, and specifically who is doing the spending.
"So, the baby boomers are traveling. They're going out to eat. You look at open table reservations up 10% from a year ago."
His point is that the sticky part of inflation is not only supply constraints. Discretionary demand is still strong, and it is contributing to the pressure.
The other real-time series he watches are Transportation Security Administration throughput, which is very high even against last year, and the Redbook weekly consumer spending data, which Slok also called very strong.
Taken together, both economists read the high-frequency data as saying the economy is in better shape than it was expected to be at the start of the year.
9. Four Hikes Are Priced In
Recording on Monday 14 September, two days before the decision, Roach asked whether a hike would be one and done, and when the repricing shows up — October, or December.
Slok said it is really unusual for the Fed to be one and done, and that the market has already moved a very long way.
"Let's not forget we all went into this year with the market pricing cuts the dot plot from the Fed meaning the expectations to short rates was that the Fed will be cutting rates in 2026 and here we are today literally pricing that we're now having four hikes priced in."
"A hike here in September, a hike in December, a hike in March and a hike again in June."
What the market is saying, in his reading, is that inflation is not under control — sticky prices plus a strong economy plus a consumer that is, in aggregate, fine.
"So the short answer to your question is I think the Fed will go several times and at the market currently pricing four hikes is a good guess and a good estimate at this point."
The investing consequence he draws is the same one as everywhere else in the conversation: rates higher for longer, and a Fed still actively trying to slow the economy and slow risk-taking.
10. Value Over Growth
The last question was the so-what: given higher-for-longer rates, low unemployment, a dynamic labor market and healthy households, where does the capital go?
The textbook answer is about debt service. If rates stay elevated, own companies that can actually pay the higher cost of servicing their debt.
That makes it a value-versus-growth question, and the reason is the timing of cash flows. Growth companies have their earnings furthest out in the future, which makes their valuations the most sensitive to today's rates.
"In other words, when interest rates begin to go up, valuations of growth companies tend to move much faster down. And therefore the short answer to your question is that value is preferred in a rising rates environment."
Slok extended the same logic into private markets, where he says it applies twice over — private equity in companies that have earnings today, and private credit and public credit where the yield level is now doing real work for an investor with money.
The counter-example he named is venture capital, which is characterized by having earnings and cash flows only far out in the future, and is therefore the most vulnerable if rates stay up.
"So that's why the short answer to your question is invest in firms, invest in assets that actually have cash flows."
Bonus Insights
The two have a Princeton connection they only discovered while preparing. Slok was there in 1995 and 1996, when Ben Bernanke chaired the economics department — the same Bernanke whose savings-glut speech the second half of the episode is built on.
Roach opened by promising not to get too heady "even though we are two PhD economists", which he said is not necessarily a good sale at this moment. Slok's reply was "No, I agree."
Roach brought back another Greenspan-era line to describe how markets handle a tightening cycle: "He also talked about partygoers complaining when the punch bowl is taken away."
LPL has measured how long the complaining lasts. "Perhaps they complain for about 2-3 months. By the time 4 months passed after the first hike markets have kind of get gotten over that. We show a chart in our weekly market commentary to that effect."
His aside on whether the complaint is even justified: households have healthy balance sheets and a high ratio of net wealth to disposable income, so the objection may be to the punch bowl going rather than to the economy.
The episode was structured as five "hot takes" drawn from questions advisers had been asking, and Roach kept the macro discussion to roughly 20 minutes so the allocation question at the end got real time.
Slok's bottom line is that the two stories are one story: AI is creating businesses faster than it is destroying jobs, and the capital those projects need — along with the energy, the defense spending and the deficits — is what has turned a decade-long savings glut into a shortage, which keeps rates high and makes cash flows today worth more than cash flows a decade out.
Products, Companies & Tools Mentioned
Apollo (Slok's firm, where he is Partner and Chief Economist)
LPL Financial (Roach's firm, which publishes the weekly market commentary carrying its chart on how long markets take to get over a first rate rise)
Ramp (The spend-management company whose data shows firms adopting AI growing entry-level hiring faster, not slower)
OpenTable (Restaurant reservations up 10% from a year ago, used as evidence that discretionary demand is still adding to inflation)
Redbook (The weekly consumer spending series both economists read as still very strong)
Books & Resources Mentioned
Range: Why Generalists Triumph in a Specialized World – David Epstein (Recommended on the show for anyone in college, on the argument that a job made of many tasks is harder to automate than a job made of one)
David Autor's work on AI and the labor force (The MIT economist Roach named as the best writer on what the technology does to productivity and to jobs)
Ben Bernanke's global savings glut (The 2005-era speech and idea the whole rates discussion is built on, now running in reverse)
Watch the full episode:
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: