Jeffrey Sherman and Jason Draho: How Should I Be Positioned? | UBS On-Air
What this covers
DoubleLine Deputy CIO Jeffrey Sherman joins Jason Draho, UBS CIO Head of Asset Allocation Americas and host Daniel Cassidy to exchange views on a range of topics, including how the current geopolitical landscape is impacting markets, the road ahead for monetary policy and implications to fixed income markets, the macroeconomic environment, along with how artificial intelligence is shaping innovation and the investment landscape. Recorded April 9, 2026, at the UBS 1285 podcast studio.
Source description (no synthesized summary yet).
Sherman and Draho argue that geopolitical shocks like the US-Iran conflict create inflation rather than growth concerns, justifying selective bond duration purchases while maintaining cautious positioning across credit markets until the bond market demonstrates discernment regarding AI-driven capex financing.
- War is inflationary and destructive but doesn't necessarily trigger recession; bond market repricing reflects inflation shock rather than growth shock
- Fed likely to cut rates back-end loaded (September-December) only if labor market weakens, making current rate levels quasi-neutral despite nominal GDP expectations
- AI capex financing through debt markets (corporate bonds, CMBS, ABS) represents growing leverage with unclear ROI; bond market's willingness to finance caps expansion remains the key constraint
This asset isn't compiled yet
You're seeing its claims, ranked. Compile it to build the argument threads, weight them, and check each claim against your library — the full view.
Break-even spreads on the front of the curve have elevated due to headline inflation from oil prices, but further-out curve break-evens (10-year, 30-year) haven't risen meaningfully, indicating the market views this as a temporary supply shock, not a fundamental inflation regime change.
“you've seen the elevation on the front of the curve as you'd expect because oil prices tend to dominate short-term inflation moves cuz remember the bond market trades headline. So, it trades the actual CPI. That's that's what's into the the TIPS market. And so, but you haven't really seen that respond further out the curve. And so, what I mean further out the curve is the break-even spreads haven't come up meaningfully when you look at like 10s or 30s.”
The Fed's break-even jobs growth number to maintain stable unemployment has fallen dramatically from ~150,000 pre-pandemic estimates to 70-80,000 in late 2024 to perhaps 20-30,000 currently, reflecting immigration's impact on labor supply and changing demographic needs.
“economists are trying to estimate what this break even number is, like how many jobs does the US economy need to produce every month given demographics, given immigration to keep the unemployment rate steady. You know, before it was like 150,000. Late last year like maybe 70 to 80,000. Now it's like, well, maybe it's 20 or 30.”
When Trump attempted to fire Powell in mid-2023, bond markets immediately spiked yields in protest, effectively forcing Trump to back down; this market reaction preserved Fed independence and should be replicated if future administrations threaten institutional autonomy.
“And that was the day that Trump tried to fire J. Powell. And they sleeked it out and they came out and bond yield spoke and spiked in the Treasury market. And to me, that's when they ran out business like, 'No, no, no, no, no, no, no, there's none of this.' and calmed the market down.”
Credit investors' core concern is getting repaid, not whether the debtor achieves high margins; therefore credit investors should focus on whether the AI capex will generate sufficient cash flow and returns to service debt, not whether it will be profitable.
“We think about getting our money back. We're not worried about your margin. We're worried about you paying us.”
The bond market has been consistently pricing Fed rate cuts since September 2022 (when Powell said 'we're done hiking'), but the Fed has repeatedly failed to deliver, creating a recurring pattern where geopolitical or economic shocks reset rate-cut expectations.
“if you go back and pull on your screens, you go back to September '22, the rates market has been pricing in Fed cuts since since J Powell said we're done hiking. The bond market heard we're cutting, right? And it's been that way for 3 and 1/2 years. And once again, there's something that derails that.”
Shorter-duration credit (12-24 month horizon) is preferable to longer-duration credit in current environment because it has less spread sensitivity, is more analyzable, and avoids long-term structural uncertainty.
“that's why I like shorter duration credit, too. It has less spread sensitivity. It it's more analyzable and I don't have to worry about the next 5 to 10 years. I got to worry about the next 12 to 24 months.”
War is inherently inflationary because it is destructive and replacement costs a lot of money, creating an inflationary impulse that the bond market struggles to price because it doesn't know how to reconcile inflation signals from oil with growth implications.
“the thing is is that when I look at what's going on in the bond market, I think it's neurotic as well right now because it doesn't know how to think about this inflationary impulse, um you know, that comes from oil.”
The Fed operates as a consensus-building voting body with 12 governors, not a monolith with a single voice, though this reality is often obscured by the absence of public dissents under post-Greenspan norms; dissents during 2022 were notable precisely because they were rare.
“We think of the Fed as this this monolithic kind of uh vehicle that there's just one voice. But, there's these 12 governors that vote, as we all know, but we're so used to not seeing dissents that people forget that there is this consensus building that started with Greenspan's era when when Greenspan ran the Fed.”
Data-dependent Fed policy conducted by any chair, if properly executed, will naturally dampen both left-wing and right-wing criticism because it is faithful to its mandate (price stability and full employment) rather than political convenience.
“They're supposed to try to figure out what's for the good of our overall economy when it comes to price stability and full employment. So, do that. Don't worry about the externalities and if you're doing your job, yeah, you're going to have critics on both sides, you know, but if you're doing your job, that's what preserves it”
The supply-shock framework is preferable for investors because 'the cure for high prices is high prices'—meaning high oil prices naturally reduce consumption and restore supply-demand balance without requiring policy intervention.
“the nice thing about supply shocks is that the cure for high prices is high prices, right? And it does the behavior of slowing things down because it change your consumption basket.”
Corporate mega-caps (Microsoft, Apple) with massive free cash flow can finance AI capex through equity issuance or retained earnings without leveraging the bond market, giving them structural advantage over smaller players that must borrow.
“when we started really the big cap ex expansion. Most of it was done through equity issuance of that of names or the megas having just massive free cash flow, right? So, they've got money to burn. Let's call it. And so, what happened last year is that dynamic started to change. They started to come to the debt market, right?”
Tariff impacts on goods prices have been appearing in PPI data and duties collected at the border, with aluminum and copper specifically showing tariff-driven price increases in manufacturing; this creates an inflationary headwind the Fed must acknowledge.
“what you're seeing that is showing up in goods prices. So you're starting to see that. You see it with duties received at the border. You saw it in the last PPI data, right? You saw it specifically aluminum and copper was driving on the manufacturing side.”
Higher oil prices sustained at $80 or above for the rest of the year act as an impediment to growth, creating a trade opportunity in duration (bonds) because slower growth typically prompts Fed rate cuts.
“if we have ultimately a higher oil price for a longer period of time, and I'll just let's say we have $80 price for the rest of the year, that has to be an impediment to growth at this point.”
The bond market is currently pricing nominal GDP as unchanged post-conflict, meaning it expects real growth to be lower with higher inflation rather than a full economic collapse.
“the way I would say it is the bond market still thinks nominal GDP is the same. Right? So, and that's I you know, to kind of paraphrase you, it's a real growth rate that's a little bit lower with a little bit higher inflation.”
Positively sloped yield curve (long rates higher than short rates) is attractive for bond investors because it allows them to buy 2-year and 5-year instruments with better risk-adjusted returns.
“I do like having a positively sloped curve by the way cuz I like buying twos at times. You know, I like the front end of the curve when it comes to investment strategies today just because the analyzability is just so much greater with that shorter tenor to payback.”
In 2022, market participants needed to listen to Fed governors in real time (not just Powell) because governors were communicating that rate hikes were coming when Powell was still emphasizing transitory inflation; governors were signaling a more hawkish stance than the chair.
“during '22, I I 'You need to actually listen to the right now because everyone's telling you they're going to hike rates. You know, so forget what Powell's saying about transitory, it's coming. We didn't know the magnitude, of course.”
The US federal budget deficit reached $1.05 trillion in the first 5 months of the year (before the war), implying an annualized $2.2 trillion deficit, creating structural pressure on long-end bond yields regardless of near-term Fed policy.
“the CBO came out and said, you know, in the first 5 months, it's a $1.05 trillion deficit. That's before the war. Right? I mean, you we're we're running at a $2.2 trillion rate.”
Software companies are inherently low-capex, high-margin businesses that have been market darlings for 6 years; concerns about AI margin compression represent repricing of this quality premium, not evidence the sector is obsolete.
“This has been the darling part of the market for the last six years. Software is a low cap ex kind of product, right? It has, you know, you can scale it very much and it has a very high margin.”
Current rate levels are approximately neutral or slightly tight for the economy, evidenced by real growth rates north of 3% in recent years despite the Fed having stopped hiking; this suggests the market was overestimating the need for rate cuts.
“This is probably the neutralish rate today. Maybe the Fed cuts, maybe they are a little tight, but, you know, the fact that we've been growing, you know, having a real growth rate north of three for the last few years tells me that the Fed wasn't that tight, right? They weren't choking off.”
AI models still have significant hallucination problems, which means they cannot be trusted for critical coding or data work without human verification; this limitation remains a barrier to full automation of complex analytical tasks.
“We're seeing it in coding. Uh the problem is is that it does still have some of the hallucinations, so that's the challenge. So, you still need sensibility about it”
Data centers have no readily apparent alternative use; repurposing a data center building in a remote location into residential or commercial space is economically unclear, creating high residual value risk for CMBS investors.
“And we're not even to the bottom in office yet, right? And repurposing buildings, what do you do with a data center that's in the middle of some place that people have never heard of, right? Does it turn into an apartment building? What do you do with it? Um and so what's the residual value is what you would say as a CMBS investor. And that has a big question mark.”
AI capex expansion is being financed increasingly through debt markets—corporate bonds for mega-caps, CMBS for real estate, ABS for revenue streams—creating exponential debt accumulation that will be constrained by bond market willingness to finance.
“Uh the debt level of AI is really growing exponentially right now. And I think it's going to be a function of the bond market will dictate how far that cap ex expansion goes now.”
Powell intentionally adopted a 'transitory' inflation framing in 2021-2022 as justification for the Fed's average inflation targeting framework, which allowed the Fed to tolerate overshooting without tightening policy.
“I think that led him to go that direction. But I think a Powell Fed has been pretty dang data dependent. You know, and you'll get political pundits that'll come in and say, 'Oh, you know, cut before an election.' Whatever. But that being said, um I I want to see how Kevin just operates in this environment. I think that you know, when we go back, there's a phrase you're never going to hear again. It's called average inflation targeting. Remember the reason that the Fed was probably harping on that transitory comment is that they had this new average inflationing targeting framework where they're saying we're going to make up for the ills of our sub 2% inflation.”
The February spike in spreads was driven by private credit concerns and AI-disruption fears in software companies, not fundamental credit deterioration; bonds repriced 2 points at worst but stocks stayed down, indicating repricing of margins rather than obsolescence.
“probably at their worst day they were down two points, right? But the stocks are still down, right? Is it because the stories at the time were AI is going to take over every software name, right? So there's going to be no more software, AI you're going to do it just in your garage and you're you're going to have the next Microsoft, right? That's was the story. But when you look at it, it was really a margin compression story.”
Data center commercial real estate financing through CMBS markets is carrying implicit leverage because these deals trade on the assumption that Microsoft, Oracle, or other tech giants will lease the asset, creating moral hazard where credit investors are exposed to tech credit risk without earning tech credit spreads.
“A lot of these deals that we see out there like in the CMBS market, ABS market, they trade because it's Oracle's going to lease it. Microsoft's going to lease it. It trades like there's this explicit covenant in this deal flow that they are going to back the deal. They're just saying, 'If you build the building, if you set it all up, sure, we'll lease it from you. Yeah, you got our word on it. You don't build it.' So, it's a different risk is the point.”
The break point in AI capex financing will occur when a major deal fails or a tech giant's financing arrangement collapses, proving bond market assumptions about AI ROI were unfounded.
“Once you get a hiccup, that's when we're going to see how discerning the bond market is. And the hiccup would be, you know, again, a a deal falling through, something like that.”
Whatever AI becomes in 5 years will be radically different from today's understanding; current assessments of AI's impact will retrospectively appear naive or vastly underestimated.
“If I had a final thought on AI, it would be this, is that whatever you think AI is, in 5 years' time, you will have no clue what that actually is. In 5 years' time, when you look back, you'll say, 'I missed this so bad.' Right? I just think it's going to be completely different, but it's going to be amazing, probably, right?”
Kevin Warsh, the incoming Fed chair, was historically known as a hawk during his prior Fed service, but he likely adjusted his public positioning to signal dovishness to secure presidential support; his actual policy direction is unknowable until observable.
“the Kevin Warsh that I would know from being in the Fed is a hawk. But, we know that the president wanted someone who's going or willing to cut rates. And so, what did he say in his interviews to make that happen? I don't know and I don't care. I want to see him in action because I can't extrapolate his history because in a Trump orbit, you know, people are influenced by him and I want to see how he behaves.”
DoubleLine Capital uses AI primarily for data scraping (quarterly reports, earnings transcripts) to extract structured data faster; report writing and coding are applied with caution due to hallucination risks.
“Yeah, yeah. I mean, it's mainly doing a lot of the scraping of data. Um especially when it comes to quarterly reports, you know, when we get earnings and everything, the scraping is it's massive. That that's been like the huge efficiency there.”
DoubleLine Capital believes equity is the appropriate vehicle for AI risk, not debt; residential mortgages (unexposed to AI capex cycle) are preferred for credit allocation.
“that's why when I come back down to buying credit products today, we're buying things like residential mortgages. They're not really exposed to that. You know, yeah, we could get some, you know, prepayment changes potentially because of some automation, but I would I joke about that with the resi guys cuz I'm like, remember we had the robo mortgage?”
Tariff announcements have been recurring for a year with 'fits and starts,' suggesting policy uncertainty rather than permanent trade regime change; investors should monitor whether tariffs actually translate into sustained price pressures.
“we've been talking about tariffs now for a year, right? And there's been a fits and starts going off and on, but what you're seeing that is showing up in goods prices.”
Bond market movements suggest that people are actively looking for reasons to buy bonds and haven't seen meaningful selling pressure, indicating that when corrections occur (like after ceasefire announcements), buying demand is robust.
“people are looking for a reason to buy, and there hasn't been meaningful sellers. And so, I think the the selling's been in the rates market”
Prior to the geopolitical crisis, on February 27th, the UBS CIO outlook was for 2.5% growth and declining inflation, with the Fed expected to cut rates a couple times, creating a constructive environment for risk assets.
“we do think back on February 27th, the Friday before? What was our outlook? Our outlook then was you know, growth this year of let's say 2 and 1/2% and inflation would come down. The Fed's going to cut a couple of times. It's pretty constructive environment for risk assets.”
Despite neural networks and linear program models being pursued as 'holy grails' in mathematics for decades, their actual limits may have been reached; Sherman is skeptical about near-term breakthroughs in AI's fundamental capabilities.
“And like I maybe maybe I'm a little cynical because, you know, it's always been neural networks. The the these, you know, linear program models, like this has been the holy grail in mathematics for, you know, my my entire lifetime. And so, maybe we're getting there, but maybe we've hit some of the limits”
Residential mortgage assets offer attractive returns because they are not exposed to AI-driven disruption risks, despite the possibility of margin changes from automation or AI-driven robo-mortgage origination, which Sherman views as overstated given past hype around robo-mortgages in 2006.
“we're buying things like residential mortgages. They're not really exposed to that. You know, yeah, we could get some, you know, prepayment changes potentially because of some automation, but I would I joke about that with the resi guys cuz I'm like, remember we had the robo mortgage? Everyone was up in arms about that in 06, but now it's AI. But now it's cool, right?”
The worst version of AI that will ever exist is the one today; whatever AI becomes will be strictly better than current capabilities, making the current state the floor for performance.
“we can say for sure that this is the worst AI will ever be is right here today and whatever it is will be will be better in 5 years.”
Microsoft can sustain $200 billion annual capex spending without materially impacting its viability, meaning credit risk is primarily about the business cycle, not capex affordability for mega-caps.
“Microsoft is an AI name or not we could argue you that all day long. Um but you know what? Microsoft can blow 200 billion dollars a year and it doesn't matter, right? They're going to have revenues.”
Petrochemical companies benefited from rising oil prices due to the US-Iran conflict because they are net long energy commodity exposure; this created a winner-and-loser dynamic across credit sectors.
“Some may have gotten bailed out with like petrochemicals that have been a really bad part of the of the credit markets for a while. Obviously, they're being helped out with with these new prices out there.”
Double-B rated credit currently trades inside 200 basis points spread; while this is tight and doesn't excite from a risk-reward perspective for bonds, it reflects reasonable pricing given the quality of the asset.
“Um let's use software as the example because you mentioned kind of some of those three private credit and it's mainly software related. Look at how much the loans traded down. Um let's use software as the example because you mentioned kind of some of those three private credit and it's mainly software related. Look at how much the loans traded down. when I look at like double B's they're inside of 200 doesn't get me excited on corporate bonds, but also it's a pretty good quality asset and so I understand why people are gravitating to that.”
As data professionals, the primary value Sherman and his team see in AI is the ability to ingest and process data faster; any tool that delivers more data more quickly aligns with their workflow priorities.
“we definitely see it from a process stuff. Like, we're data people. At the end of the day, any way you can get us more data faster, that's what we all want.”
The ceasefire that was negotiated between the US and Iran appeared to be violated within minutes of being signed (or possibly wasn't properly signed), creating uncertainty about the durability of the agreement and the likelihood of continued escalation.
“we have a truce as of last night, looks like we violated within a couple of minutes of signing it or if it wasn't even signed, who knows.”
AI is useful for report writing but Sherman finds it most valuable as a tool for editing written content rather than generating initial drafts, as a mathematician he prefers to write content in bullet points and use AI to structure it into prose.
“definitely report writing is the worst part of the job, Jason. As as a mathematician, I am uh it's just I I can't stand it. I write in bullet points on my make it sound good, you know, um you know, and I'll edit it, you know, I I feel like I'm a better editor, but um from that standpoint, you're seeing a lot of that.”
Advanced AI models are starting to show promise in building infrastructure (like quick spreadsheets) from raw data, but Sherman is reserving full enthusiasm until he sees what the 'next level' of capability looks like.
“I will say that we've we've started to use more of these more advanced models and they're starting to show some promise of a point in data being able just to kind of build the infrastructure of a quick spreadsheet. And so, I can see why people get super excited about it, but I want to see that next level still, you know”
The US government was promised a reduction in the deficit approximately one year ago but deficit spending has not declined, suggesting that fiscal discipline is not improving.
“And I just ask everybody, where's dosh? You know, we were promised that a year ago.”
The recording is happening on Thursday, April 9th, 2024, providing a dated context for evaluating the geopolitical and market developments discussed.
“we're recording here on Thursday, April 9th, to give a bit of context around timing.”
The new UBS studio has improved broadcast facilities compared to previous setups, making this a better production environment than phone-based recordings.
“glad to welcome back from our partners at DoubleLine Capital, Jeffrey Sherman. Uh Jeffrey Sherman serves as deputy chief investment officer and portfolio manager at DoubleLine Capital. Also joining us today from the UBS Chief Investment Office, glad to welcome back head of asset allocation for the Americas, Jason Draho. Jason, great to have you back as well, and Jeff, I'm glad you can join us here in person at the new studio.”