How should I be positioned? with Joe Davis (Vanguard) and Jason Draho (UBS CIO)

Hi everyone, Brian Contreras here. Welcome back to How Should I Be Positioned on the UBS Market Moves podcast channel. On this podcast, we catch up with our industry colleagues and thought leaders to discuss the current market and macro environment along with thinking when it comes to asset allocation.

Joining us today here at the 1285 Broadcast Studio in New York, glad to welcome back from Vanguard Joe Davis. Joe Davis serves as Global Chief Economist and Global Head of the Investment Strategy Group. And joining us from the UBS Chief Investment Office, as always, Head of Asset Allocation Americas, Jason Drejo.

Jason, Joe, great to be at the table with you both. Welcome back and looking forward to hearing your insights today. Joe, great to have you.

This is the second time, but first in person. No, it's a privilege. Are you kidding me?

To be, for an economist, to be invited back is rare, so. No, no, no. We like the whole spectrum of people, don't we?

Great. So I want to start out with a question for both of you, starting with Joe. The geopolitical environment remains quite fluid as we're over two months into the U.S.-Iran war.

To what extent and how have geopolitics impacted your investment outlook? Yeah, I mean, I think, again, geopolitics are a tough risk factor to try to sort of forecast. So what we've generally done, I don't think we're alone, Jason, I encourage your thoughts.

We just think about what I call ceilings and floors with respect to what could really impact the economy and corporate profitability. So it generally takes you to oil prices, if you're talking about the Middle East. And so we never know if it's going to flare up or not.

And so we think of ceilings and floors of, if oil would breach, say, roughly $150 a barrel going into the year, we didn't know if it was going to test that, then that would start to materially affect our sort of outlook, both for the markets and the economy. Again, we haven't approached those levels. We've had, I would call, decimal point effects on our economic outlook, say, for the United States.

It's different in other markets. But that's how we generally approach it. We do not, I think we have an edge in some points of forecasting with respect to data, but geopolitics is one of them from our perspective.

But we have to think about that seriously. So that's how we think about it. Think about at what price would it be risk on or risk off in the market?

And so that would steal our nerve should we ever, say, breach $150 a barrel. And Jason? So to kind of borrow the floors and ceilings analogy, I guess the way I think of it is like, we're at a certain floor.

The question is like, how many floors do we drop in growth and how many floors do we go higher in terms of inflation? And so kind of take it as a baseline where we would have been a little over two months ago before this all began, you know, relatively constructive on the US economy this year, let's say 2.5% like level of growth. Inflation that would trend lower as the tariff-related inflation kind of wanes off.

I think the conflict is quite a situation where growth will be a little bit lower. So, you know, let's take it down to closer to 2%, you know, you can get overly precise in terms of t-sum. Yeah, sure.

But like, you know, there's some offset there. And then inflation will stay higher for longer, but ultimately the core elements still kind of go lower. So I think it's, you know, degrees of magnitude, not fundamentally kind of a different view.

But something that I've been kind of wrestling with is, and this is a question for you, Joe, to kind of follow up on this. For a month ago, end of March, like, well, the consensus view is like, well, if things don't open up by the end of April, it's kind of a problem for the economy because oil prices will go to 150 or higher. End of April, well, if things don't open up by the end of May, this will be a problem.

Here we are now, it's recorded on May 5th. How do you think about sort of this concept of non-linearity that we're good, but at some point, you know, supply won't be there, and the only way you, you know, kind of get back into equilibrium is demand has to be destroyed, prices have to go higher, and then it's a non-linear decline of potential to macro out. Like, do you believe that idea?

And if so, like, when would that actually be taken? I do believe it, Jason. I think we share a similar framework.

I think it's, at least with oil dynamics, what the research is pretty clear is that there are non-linear effects on both how high it has to go up, as well as the magnitude. So I think the duration of how long oil would stay, even at 125 hours a barrel, matters. It's not just how high it goes, how long it stays there, because that's how you start to really crimp business investment.

The bottom line is, unless we're talking about 100, and we remain of this view ever since we saw the headlines in the Middle East, we have to be talking about closer to 150 hours a barrel, well through the end of the summer, so you start to seriously push down economic growth. I mean, meaningfully, like, say, 1% or so, and I think it's the strength of the other factors going into year that mattered, and, you know, in our thesis, and I have yours as well, Jason, at UBS, is around AI is such an important risk factor that it, that strength. We were, three years ago, I think we'd be talking about weaker growth numbers, but you're talking about a trillion hours plus, and hyperscalers alone, doesn't mean the U.S. economy's immune to oil prices, it just means we're not surprised by the resilience we're seeing, and I think we will continue to see that.

Oil has to go up another 30 or 40% from here, and stay there well through the summer, before, at least from our perspective, we would be maturely knocking down growth expectations. So the timeline that some would have, like, if this doesn't, the straight-to-four moves doesn't open up, or oil can flow through, you know, by June, this is a real problem, which if you're an investor, the sell-in may go away. Yeah, yeah, yeah, yeah.

But you're kind of saying, like, no, like, it's, you know, it's usually six months. It's not, like, the magic rule, but that's usually where you have to start really talking about it. So given that, so, you know, you know, the growth impact, you know, some, but not dramatic unless this gets really bad, which then kind of going to the Fed, you know, kind of dealing with the same issue we are, of trying to assess what are the economic implications, how much could grow slow, but also how long inflation could stay high.

We had an FOMC meeting last week, stayed on hold, which was expected, three dissents by people who thought the statement should not be kind of as, like, an easing bias, more balanced. Yeah. Kevin Warsh now looks very likely to get approved by the Senate, would be the Fed chair for the next FOMC meeting in June.

You know, a cut then seems incredibly unlikely, but the thinking is that Warsh's bias towards, you know, being cutting rates. You add all this together, what is your thought in terms of, like, where does monetary policy go given the challenge of trying to think through this? Sure.

I mean, that's, you know, I understand the bias. I think, you know, the Fed has generally been of the view that when inflation shocks hit, they'll be pretty temporary. That said, I think our view has been and remains that they'll be hard-pressed to cut.

The labor market is a little bit stronger than people think. It has clearly slowed, Jason, right? I'm not going to dismiss it, but given the analytics we have in the labor market, it's a pretty unique view, which we could get into.

It's a little bit stronger than people think, and so I think there's that horse race. You have inflation that says they shouldn't be cutting at all, but it's temporary. I think, you know, my only counsel to Fed Reserve policymakers, and I've shared this with them repeatedly, is don't think that the slowdown, whether it's in labor market or GDP, is simply just because of coin and demand.

There has been significant movements in the supply of labor, some of the immigration, which is masking what you think of as a material slowdown in hiring. That's not entirely the case, and so that's one of the reasons why we don't have the Federal Reserve cutting materially. We could get another negative job growth numbers on some Friday, first Friday of the month.

They'd sneak one in, but we don't have material cuts. I think you have to be talking about recession. So the labor market, I think, will hold up, and the GDP numbers say they shouldn't be cutting clearly at all.

So we are officially still at two cuts, September and December. I think the risk certainly is that it gets pushed back. To me, the markets, whether they cut September, December, Q1, doesn't make a big difference.

No, it doesn't. What matters to me is, of the markets at least, is there's an easing bias. But stepping back from, will they or won't they cut in the next six, nine months, bigger picture, thinking about transitioning some of the conversation to more structural, secular trends.

But on the Fed, we do have a new chair, let's assume Warshel will be in place, who during his testimony talked about much more focus on the balance sheet, shrinking the balance sheet, whether he can or not, that's something different. Change of communication. I think he's of the view that the Fed over communicates, which, good luck trying to get some of the Fed presidents to stop talking.

So I guess, and maybe other sort of policy rules he has. At the same time, change is difficult, the Fed is a relatively conservative institution. I would think that tweaks at the margin.

But if you were to think under a Warsh-led Fed, relative, say, a Powell Fed, in terms of the conduct of monetary policy, what are things that you think structurally or approach policy rules, whatever it might be, would actually materialize in some, maybe sort of material way? Yeah, I would say, well, at least I don't foresee, Jason, I would say radical change. I mean, oh, they're going to change some inflation target, they're going to do something contrary to their dual mandate.

What I do see, though, and I would encourage the Federal Reserve to consider it, one would be in terms of how they think about the economic outlook. I think, as we do in the asset management industry, I think scenario analysis is really important. The other thing that our research has shown at the margin is that they have to respect what economists call the supply side.

I would say things such as technology, demographics, just as much as so-called consumer spending and demand, right? Demand matters, but, hey, listen, if we're sitting here in 2026 and it's much like 1997, you can have GDP growing up and it does not mean the economy is overheating. That's what I mean by considering the supply side.

And so it's opened our eyes to some of the work we've done. And I think that that would be good for at least the Federal Reserve to start challenging those assumptions that we all have to do, right? If you're talking about productivity, that's another percentage point higher than it is.

You're not going to have an inflation problem, even if GDP goes up for a time. And I think that would be good, because that's where you get, I think, policy mistakes. I don't see the Federal Reserve raising rates any time soon, but that's the two things I think we'll see Chairman Walsh's comment officially.

I think that's where we'll see them. It's more behind the scenes, but that'll matter for what the market is digesting from prognostications from the Fed. So Walsh has, even last week, commented that he thinks AI would lift productivity, ultimately going to be somewhat disinflationary.

He said this before. It's nothing new. I know you've done a lot on megatrends, AI as a part of that.

And you probably don't remember this, but I attended an event where you were speaking. This was in December of 2024. Oh, here we go.

What did I say then? No, no. But I remember the question was, you were talking about AI, and I think I kind of asked you at the time, everyone was asking, when is this AI impact going to be in terms of really materially impacting productivity?

And I think the general sense from you back then, and this is now 16 months ago, was like, it's probably not going to really materialize until the late 2020s, if not the 2030s. Fast forward 16 months, when we look at some of the developments on these various agent tools, it seems like the capabilities are happening faster. They're getting deployed maybe more quickly.

Have you sort of altered, maybe you're thinking of the timeline of when it could happen, how big the impact could be as we go? I appreciate you bringing that up, Jason. I think it's a little bit earlier.

So just as context, we have a 3% or so growth number for the United States for the next several years as our baseline, which is, we're pretty, at least in the economist community, we're putting our neck out there, but it's all in the work we've done on AI in particular. Just on that point, there's an element of like lifting, investment is so large, that sort of cyclical growth is a little bit, versus like trend growth. Yeah, yeah, yeah.

This is a trend. So just to be very clear, given all the data we've collected, because we're trying to be data driven, we're not trying to tell stories, right? That's why we want to take this really seriously.

We've been looking at AI for a decade. We have a 65% probability that AI emerges as general purpose technology and trend growth in the United States is 3% in three years. So that's a very specific horizon.

The confidence interval, yes, the use of some of it would be delayed, 2030 could be some of our simulations. It is coalescing. Some wanted to go even sooner, but it's coalescing around 2029 when it really starts to feel it.

Now, before that though, if we're talking the trend, we have to have actual growth picking up in the investment cycle. I think that's not news, right? We're already seeing that.

That's like a necessary condition. That doesn't guarantee that we're going to be right in our forecast, but that's a necessary condition. And now we clearly are seeing some of those tools, the ability for automation and some augmentation is there.

But although we're all starting to experiment, they are not streamlined in production processes. So when it manifests itself in the data, I think, and across all industries, I'm not talking within IT, I think we still feel, I feel confident that, I feel good about the 2028, 2029 is like, it feels like 1997, 1998 if we were there with the personal computer and the internet, right? It wasn't 1999, but you could really start to see it and feel it in real life.

I think we're starting to experiment. But outside of Silicon Valley, the average person, the average firm has not completely operationalized it in their business. But that day is coming.

So our conviction in our so-called non-consensus growth outlook is growing, despite the oil headlines and some of these other things, so. I want to come back and maybe we'll end the question regarding like the actual, how you guys use AI, but just sticking more with like the implications for AI on the economy. So right now, an investment boom, at least in AI, that's helping to lift growth, all is equal.

That should drive us into higher productivity growth. I think that timeline is reasonable. A concern is AI impact on the labor market.

And there's a whole range of outcomes and debates. It's clearly going to disrupt some jobs, it will create new jobs. Some workers will be enhanced, some will be substituted.

In terms of the growth outlook, one of the questions would be, does this disruption happen before the creation? Meaning, will nothing new jobs kind of be created such that, you know, because in order for the economy to grow at 3%, you do need consumers to have income to spend. And if like, I mean, hypothetically, if 20% of the workforce gets laid off, well, you're not getting that 3% growth, right, you're getting a recession.

So how do you think about maybe the sequence of the- You have the destruction without the creative destruction part. Yeah. I mean, in most of the work we do, we're going to, I mean, if history is any guide, we should get both at the same time.

Our analytics support the fact that we'll see some of the automation first, because AI has a more automation dimension than augmentation, meaning complement versus substitute, you mentioned, right, Jason? So bottom line, though, is we still have a fairly tight labor market, given 3% growth. That said, 20% of the occupations will see so much disruption that we'll start to see the job curve bend negative.

However, that sounds really bad. The fact is, is that the lack of automation in the U.S. economy is subtracting the most from economic growth in our history. And so we need automation, given the lack of demographics, new workers coming into workforce, retirements are the baby boom.

And so I know it's unnerving from the headlines, but that's the only reason, that's the only way we're getting to 3% growth, is because we have a significant automation front. But it's not that. So for every job that's going to lose more because of automation, that's roughly 16% to 20%.

And it's going to happen fast. Personal computer affected, had a 16% effect on time savings in 10 years. We're anticipating that for 35% of the occupations in five years.

So twice the effect at twice the speed. That said, three more occupations will benefit from augmentation, as well as automation. So they'll save some time and move up the value stack, as the automation just hits some jobs.

Some jobs, it's going to be pure replacement. We identified those occupations in our work seven, seven, eight years ago. It's computer programming, it's parts of finance, it's IT, it's part of marketing.

So they're, they're pretty intensive, but, but we are hard pressed to get dystopian things that we have unemployment of seven, 8%. That's the numbers you'd have to hit before you really start to destroy demand, given, given what we see on the investment build out. So it's possible that you could, you could destroy your own demand by firing enough people.

It's just that it would have to be three times the magnitude of the assembly line in the 1910s and 20s. The AI will first be the assembly line for the office. We just, in our math, and we've looked at every job description in the world.

We just can't get the magnitude of the numbers. We can get the direction, but we can't get those magnitudes that would destroy demand. So you mentioned like the 1910s, there was Henry Ford who said like, I have to pay my workers more because if they don't make the money, they can't buy my car.

So what's, what's, you know, something, which was very nice of them. Yeah. I don't think it was altruistic on his part, sticking with kind of this labor piece and like tying in with this longer term, another megatrend, like the whole demographics and aging workforce.

And like, especially when you think about immigration in the country currently, it's, you know, it's low and probably won't change, you know, in at least the next few years. This gets into a little bit back to the Fed policy, like what is sort of a break even amount of jobs that needs to be created every month in order for the unemployment rate to stay stable? Which, you know, historically it would have been like, say 150,000 jobs you need to create based on immigration demographics, you know, aging population, lower immigration, 60 to 80.

But Jay Powell, I think last, earlier this year, last fall said, you know, maybe as low as like 10,000 or even zero to keep the unemployment rate. When you talk about negative job growth, given those demographic trends, do you think that's a situation where it's, we need the automation because like, we actually don't even need to create new jobs. We don't have enough workers.

And so the unemployment rate stays in the four and a half range. Would actually rise in the five or six percent range in your scenario? Well, we don't have the unemployment rate going materially higher.

I could see it inching up to the high fours, as you mentioned, it's four and a half. That's why it's probably the best statistic if I was, if I could only pick one statistics, statistic today, I would pick the unemployment because it's balancing that supply, how many workers we have and then how much hiring do we have. So it's a nice sort of summary statistic.

And the fact is, is that you're right. I mean, it's in a bizarro world where we have flat sort of potential for labor force contribution, which is one sense why people should be cheering for automation. Given our fiscal commitments in the United States, other markets, why do you think China's investing so much?

They have negative population growth as far as I can see. That said, you can go too far on the automation front and then the balance tips and you have unemployment rate go up. But in most of the projections we see, unless we are talking about a two sigma, a two standard deviation move on automation, and we have data going back 150 years, so this would be worse than the assembly line.

If we can generate that sort of automation shock with no benefit to workers, it's only substitution, then you can generate unemployment rates that are north of 6%. We're talking about finance, IT, marketing, business, professional, legal services, they start to get hit in middle to higher income jobs. That is a scenario that we don't completely dismiss out of hand.

It's just that we are also anticipating complementaries, co-pilot effects of AI, and then the most important one, which is the sort of new industries that are created. That's the moneymaker, by the way. It's funny, when I've been talking about this forecast that we have, because we're kind of out there a little bit, Jason.

Not as much as perhaps in the past when we first set it, but 3% growth with no demographic contribution, just mentioned. That's a massive productivity spike, and I either get two reactions to the forecast. If I'm Silicon Valley and I say, that's our most likely outcome, it's two and three, it's 65% odds that we have general purpose technology with effects that are beyond the personal computer, and they say, why is it only 65?

We know it's 100%, but then it goes to other audiences to say, wait a minute. We have low demographics. How can it be that high?

I feel on both sides that this tension, but we haven't seen those other two effects that are going to really carry the investment boom, which really will get into investment theses, because the equity market is assuming 100% probability of our baseline, which AI becomes a general purpose technology. It meets the demographic headwinds we face, and it completely balances the job loss with the jobs needed for AI to fill in a perfect balance, which we can generate those numbers. Our projections are 100% odds, and that's the conflict we feel in our investment strategy.

How can we play offense and defense in this bifurcated outlook we have? 2027 is fairly easy. We get the same path regardless. It starts to get messier if AI only delivers automation, and we don't get smarter as workers as a result.

That's where it gets a little bit uneven and rocky. Well, in terms of your growth forecast being on the optimistic end, except for Silicon Valley, where it sounds like you're a debbie downer. Oh, we're definitely at the low end, right?

Anthropic's been on the tape, right? It's been even higher, right? Yes, yes.

I know. There was an Economist article last time. I'm like, oh, 10% GDP growth.

Oh, okay. That is the case where you're saying, well, that's the thing, is that we could very well see an environment, though, unemployment rate rising modestly for a time in 2027, 2028. Modestly.

I'm talking like five, low fives, yet you have GDP growth of three. That would be very unusual. We should have unemployment rate going low, but we could see a little bit ingestion of We're not seeing the data, by the way.

With 401k records, Vanguard has a fairly unique lens, not the only lens on the labor market, but we know how many people are being hired at every moment in time, because if you join a company, if we manage your 401k plan, you get auto-enrolled, and so we have hiring rates immediately. We have 50 million clients, so it's a deep data set, and what I can tell you, though, is I get a lot of questions, particularly for entry-level workers. AI can do the easy stuff, and so the college graduate sort of labor market.

We're not seeing any evidence of it deteriorating at all. In fact, we see the same hiring rates this year as three years ago, as five years ago, as 10 years ago. No change across industries.

So that narrative is just not in the data. I'm not saying we could see some of those headwinds. I think it's more a 2027 story, but we're clearly not seeing that in the analytics.

There is a scenario, from the market's perspective, that you could get this productivity lift even sooner, so you're going to 3%. Job growth is very low, so unemployment actually ticks higher. The Federal Reserve is like, well, what do we do, because our dual mandate dictates we have to support full employment, so they actually cut rates, which if you want to think about Roaring 20's boom-bust scenario, from a market's perspective, there is a scenario where you get this weird kind of dynamic, like you're growing and you're cutting at the same time.

It is. I would worry about that environment, because I think if we're talking about accelerating economic growth, but job displacement, that's what economists would call structural unemployment, which we know by definition, the Federal Reserve, not a criticism, they can't affect. Cutting rates in that environment would really concern me, because I would imagine asset prices are already doing fairly well, and that's where you can, at the margin, contribute to financial bubbles.

I'm not calling one now. I think that would have to be thought through. I think you could, though, let the economy run hot, for sure, because it's really being affected through the productivity side.

It's not overheating, and I think that would be enough to carry the day for the financial markets. So I've been kind of, you're talking about being optimistic, I wrote, first started talking about like a Roaring 20 scenario in 2021, and it's, you know ... Well, kudos to you, because that's definitely ahead of the curve.

You know, for a lot of people in the country, it probably doesn't feel like a Roaring 20. In the stock market, yes, there's this bifurcation, but if you think it is the Roaring 20, well, that story ends with like a bubble and a crash. I'm not forecasting it.

I know, I know, I know. And then to use the parallel, I think we're probably like 96, 97. I'm of the same mind, Jason, and you know, apologies, because I'm just, you know, supporting what you're saying.

I'm being a good guest, but I genuinely believe our research is similar. You'll be back. No, no, but I think it's similar, and it's like, what year are we at?

I've been hard pressed. I said, I don't know. No one knows for certain.

I come back to 96, 97. So we're getting close to the end of time here. Just one final question.

We've talked a lot about AI. Just curious how you are now actually incorporated into your investment process, research process. What tasks are you automating versus augmenting?

Well, you know, we talked to the team, so let's get out the stuff we don't like doing the most, but I'm not doing it. We have a 100-person research organization. I'm telling my team, don't just focus on saving time.

So yes, do that because it's an annoyance. We are high-valued employees. Focus on augmentation.

Get me at the new answers for clients, or insights for our investors that we could not get at before because the math was too difficult, the coding was too hard, we didn't have the data sets available, or we couldn't say yes. We declined two projects for everyone, and we say yes to. So it includes the throughput.

So rather than reducing FTEs for the same output, I want to do the top line up 2x. I'm counseling CEOs when I'm asked. There's a lot of focus on the automation.

I'm not interested in cutting 5% of my work for 5% savings. That's not going to be game-changing. I hope the CFO's not listening here.

No, because I'm- Actually, I hope he is. But increase the top line revenue 20x. That's what the computer did, and that's what I at least want us to try for because if we're just talking about the 5%, I don't think that's going to move the dial.

I'd agree. There's things that I have a list of projects I've wanted to work on for a while. If this can help me do that, I think that's beneficial.

And that's what we get at, right? At least that's worth exploring as much as, hey, how much time can I take out of the stack? Well, thank you, Jason and Joe, for spending time with our listeners and clients here on How Should I Be Positioned, and we look forward to our next conversation.

Thank you. Thank you for tuning in. Be sure to visit UBS.com slash studios to view the entire UBS Studios suite of podcast channels along with our video offerings, such as UBS Trending.

You can also follow us on Instagram for content highlights at UBS Trending. UBS Studios is part of the UBS Chief Investment Office within UBS Global Wealth Management. Visit UBS.com slash CIO to view the latest research.

UBS Chief Investment Office's investment views are prepared and published by the Global Wealth Management business of UBS AG or its affiliates. The views and opinions expressed in this material by external guest speakers are those of the author, speaker, and are not those of UBS, its subsidiaries, or affiliates. Accordingly, UBS does not accept any liability over the content of this material or any claims, losses, or damages arising from the use or reliance of all or any part thereof.

This material has no regard to the specific investment objectives, financial situation, or particular needs of any specific recipient and is published for informational purposes only. For a full legal disclaimer applicable to the independent investment views produced by UBS, please visit our website at UBS.com forward slash CIO dash disclaimer.