Full Transcript: Rob Arnott on AI Stocks, Valuations, and Value

Why Small-Cap, International, and Mean Reversion Still Matter

Justin: Rob, welcome back to Excess Returns.

Rob: Glad to be here.

Justin: It’s great to have you on again. Our audience is very familiar with you — you’ve been on multiple times and we always appreciate having you on. You’re known in the investment world as one of the most influential speakers on quantitative investing, fundamental indexing, and really just a student of market history that we can always learn a lot from.

And so today we have a wide variety of topics. I think some may be more evergreen than others that we want to talk with you about today. That includes looking at some of the recent research that the firm has put out, talking about how you’re looking at market valuations and future returns. And then maybe to start, we’ll get into some of the things that are going on in the market that I think investors are thinking about, including what’s happening over in Iran and how investors should be thinking about that, AI, and some other stuff. So really appreciate you joining us. You can always learn more about Research Affiliates and access the research that these guys are putting out at researchaffiliates.com and we’ll put links to some of these research papers in the show notes as well.

So anyways, thanks Rob. Hopefully that was not too long of an intro, but I want to give you that because that’s important.

Rob: Yeah.

Justin: So to start, let’s talk about what’s on the news and what’s front and center with this Iran conflict. A lot of investors are asking themselves, how is this going to affect my portfolio? Should I be making adjustments? But what I wanted to ask you — what’s important is to try to look at what does the study of market history tell us about the impact of conflict and whether or not investors should be letting stuff like this influence their portfolio.

Rob: Historically, war has not been particularly damaging to stock and bond investments except in the losing country. The global markets shrugged off Ukraine and to this day, the tragedy of what happens in war notwithstanding, the markets focus on what’s happening in the global economy now. One of my favorite economists, Charles Gave, a brilliant French economist — sounds like an oxymoron, but it’s true — is fond of saying that GDP is energy transformed. That is to say, you take energy, usually oil, and you turn it into prosperity. And yes, over the course of decades, economies become more energy efficient — that is to say, more dollars of GDP per barrel of oil — but fossil fuels are still the dominant source of energy powering the entire planet.

And so a major oil producer that’s 5% of world supply that also has the potential to close off the Strait of Hormuz, through which 25% of world oil supply passes, has the potential to be massively disruptive. So this is not Ukraine. This is potentially a big deal, which is why very aggressive strikes very fast and obliteration of Iran’s navy so that they have less chance to close off the straits are important elements here.

The other issue that I think is equally important is, what’s the exit strategy? And if the exit strategy is to leave a collapsed economy run by unknown zealots or dictators or whomever, then what you’ve probably done is take much of 5% of world energy supply offline for a fairly long period of time.

So we don’t know what the end game is. We don’t know how it’ll play out. I do have a plaque — you were kind enough to share that question with me ahead of time. I do have a plaque that used to be on the desk of John Templeton. John Templeton was a legendary investor from the 1940s, and he had a plaque on his desk that said “Trouble is opportunity.” It was gifted to me about a decade ago, and it’s a useful reminder. People worry about trouble and they should, but they should also think in terms of what are the investment implications, because tumult creates opportunities. Oftentimes tumult creates market reactions that are sometimes overreactions, so you can have opportunities to concentrate on. Liberation Day, April 2nd, was a beautiful example of that. The market severely overreacted to tariffs that were of unknown eventual magnitude that turned out to be more of an economic event than anything tremendously influential.

Justin: Is the higher than, you know, average valuations here in the US — does that at all, or could that at all, play into maybe this being a little bit more of a delicate situation given what’s going on? I mean, how do valuations sort of factor into how you might kind of be looking at this?

Rob: Well, Ben Graham famously said that the markets in the short run are voting machines and in the long run are weighing machines. The voting is American exceptionalism is real. Let’s bid this thing up ‘cause it’s got nowhere to go but up. The weighing machine would say the US is at roughly twice the valuation multiples — whether you’re using CAPE ratios, dividend yields, or price-to-book value — roughly twice the valuation multiples of the rest of the world. And so I do view that as an extreme opportunity, not necessarily horrifically bearish for the US, but a very interesting opportunity to build positions in diversification and risk reduction by deploying money elsewhere in the world.

Elsewhere in the world is also likely to be more affected than US markets by the tumult in the Middle East, and we certainly saw that the last few days. I view expensive markets in the US as being vulnerable, but more from a perspective of opportunity — the best opportunities lie elsewhere.

Justin: And when you’re looking at expected returns, you guys have a tool on the site where you do asset class and market expected returns — I think seven to ten year returns. Mm-hmm. Are there any — where in the US looks attractive? I think like large-cap growth probably has long-term returns less than historical average, but are there any pockets in the US that do look more attractive to you based on your estimates?

Rob: Well, firstly, the spread in valuation between growth and value is historically extreme. It’s been wider than this only in the aftermath of the COVID value meltdown in the summer of 2020. The spread between growth and value in various valuation metrics — price to sales, price to book, dividend yield, price to cash flow — have the spread between growth and value roughly as wide today as it was at the peak of the dot-com bubble. Okay, well that’s a little alarming. The narrative is AI will change everything, and it will, but the narrative also says the companies leading the AI charge today are poised for extraordinary growth. And one or two of them might be, but there’s also the issue that transforming AI into profits is proving to be rather difficult.

The CapEx outlays are stupendous, estimated to be $600 billion next year. That’s a lot of money to spend when you don’t know how to get that money back. One of my colleagues, Chris Brightman, is writing a paper that’s coming out in a few days on exactly this topic. The spread between large cap and small cap is the widest ever — no exceptions, the widest in history. And we have a paper coming out. CFA Institute has a Research Foundation — like the Financial Analyst Journal that publishes academic articles, the Research Foundation publishes more in-depth monographs, longer than papers. We have a monograph coming out shortly, I think next month, that looks at the active side of indexing.

Indexes are thought to be passive. They’re not. They mostly are — if they have 5% turnover, you could think of it as 5% active, 95% passive, blissfully indifferent to what’s going on in the individual companies or the economy or the market, just along for the ride. But the 5% looks like a crazed growth investor, buying stocks that have soared at lofty multiples, selling stocks that have tanked at deep discounts. Now, why do I mention this in the context of small cap? Small cap isn’t in the big indexes, whether you’re looking at S&P 500 or Russell. And as money flows into index funds, it flows out of things that aren’t in index funds.

So the spread in valuation is now better than two-to-one between the S&P 500 and the Russell 2000 in terms of relative valuations. And here’s the fascinating thing: the companies that aren’t in the index have had 2% per annum faster growth in the underlying business over the last 30 years than the companies that are in the top indexes. 2% per annum faster growth, and they’re priced at a 50% discount today. That’s fascinating. So if anyone in your audience doesn’t have a toehold in small cap, put it in place. It should be a decent slug of your portfolio.

Jack: And it has been nice as a factor investor myself. For years we’ve been talking about value. We’ve been talking about international, we’ve been talking about small cap. And actually when I tell people about it now, they’re actually excited about it. In past years, for many years, they were like, you’re an idiot, what are you talking about? Now it’s actually working to some degree. So it is nice to at least see some results.

Rob: Yeah. Look for that monograph when it comes out, because there’s a wonderful graphic in it that shows the divergence between relative valuation and relative performance of the underlying businesses. It’s relentless.

Jack: So I want to switch and ask you about AI. Did you get an opportunity to read this Dario Amodei piece? It sort of took the world by storm.

Rob: Yeah. It’s a really interesting paper and I think there’s a lot of truth in it, but I think the truth is somewhat overstated.

Jack: Yeah. And it was meant as a thought exercise, which I kind of enjoyed it from that perspective. I think people who took it as “this is the way the world’s gonna be” and went crazy attacking it — for me, I feel like I know a lot more about AI because I read it and because I thought about the concepts within it.

And one of the things I thought when I read it is — we have Rob coming up and I want to ask Rob, because you’re really good at thinking about things from the long term. One of the things that struck me about the paper is we have this idea that in the long term, AI is going to create a lot of jobs, which I think is probably true. It’s going to create a lot of jobs, but we also have this push and pull with this idea that in the short term, it’s going to both maybe enhance us in our careers, but it also might destroy a bunch of jobs. And if it’s a more transformative technology, it might destroy more jobs than other technologies have. How do you think about that balance?

Rob: Every technological revolution in the history of mankind has killed millions of jobs. If they had job statistics back when fire and the wheel were invented, I bet you would’ve seen lots of jobs displaced. But whether it’s the steam engine, the cotton gin, railroads, telephone and telegraph, cars and trucks and airplanes, computers, the internet, or the AI revolution — every one of these killed millions of jobs. And that’s devastating to the people who are displaced and find, oh, I have skills that are no longer useful. I’ve got to find something totally different to do.

But a generation later, those jobs aren’t missed. “Computer” used to be a job description — somebody who’s very quick and facile with math and could do complex equations faster than anybody else in the office. That was a computer. Does anyone want that job now? I don’t think so. Computers can do all that stuff a billion times faster with absolute accuracy. When it comes to social media, lots of jobs were displaced, the newspaper industry upended. Did that create new jobs? You bet. And the same will hold true for AI. Anyone who’s not thinking about how they can use AI to do their work better and faster is an idiot. It is very powerful.

Fun little anecdote. I’m kind of old fashioned. I like to do initial research in Excel because it’s so quick and easy. I’ll have the research team put together data into an Excel file and then I’ll play around with it. I had them put together a file of all US companies for the last 70 years with their return for the year, their valuations ratios, and a construct for RAFI — the fundamental index — a cap-weighted index, a cap-weighted construct to compare attributes, and so forth. And I asked one of our programmers — doing this for 70 different pages on a monster spreadsheet, it was time consuming — can you set up a macro for me? And he said, I don’t know, I haven’t done macros. But let me take a look. He got back to me less than an hour later and said, it’s done. And I said, how did you do that? You learned how to do macros in less than an hour? He said, no. I asked ChatGPT to write it, and it put it together. I tested it, it works fine. So here you go. That would’ve been a day or two or three of his time if he had to learn how to do macros. It would’ve been three or four hours of my time even though I know how to do macros, and ChatGPT just flung it together and it worked. So cool.

Jack: It’s funny too because I just — I haven’t used it a ton yet, but Claude has a plugin that you just install straight into Excel. And I think that’s probably pretty promising too, is just to have these working for you right inside the spreadsheet if you prefer to continue using Excel.

Rob: Notwithstanding the other war going on right now between the administration and Anthropic, it bears mentioning that we polled our IT people and our programmers on which tool they most like to use — anyone relating to programming at all. It was unanimous: Claude. Cool.

Jack: Yeah, I’m kind of the same way. Like I use Claude Code all the time now to build interesting things. One of the interesting things you said to us last time you were on — or maybe the time before that — I think relates to what you were just saying: this idea that you had once called an all-hands meeting of Research Affiliates and basically told people, LLMs are not going to replace you, but people who learn to use them could. And I think that’s the idea. That stuck with me ever since. I’ve been thinking about everything I try to do in my life, like how can this enhance what I’m doing? How can this make me better? And I think that’s the attitude we all have to have with this.

Rob: Yeah, that’s exactly right. Think of it as a tool. It’s an extremely powerful tool. In fact, it’s an astonishing tool. And it interacts in the most complex programming language in the history of mankind — it’s called English. The notion that you can ask questions and get a reasonably reliable answer. Yeah, it still hallucinates. I asked it what the all-time high was for Cisco during the dot-com bubble, and it came back with a date in 1999, and I knew that was wrong. So I said, well, what was the high in 2000? And it came out with a totally different and higher number. So the first answer was a hallucination, which I knew enough to know was wrong.

But you know, you ask a human researcher, they’re going to get things wrong too. It’ll change everything. But one of the interesting challenges is how do you make money? If you’re creating AI hardware for the moment, that’s super easy because you’ve got people on waiting lists wanting to spend tens of thousands of dollars per chip to buy your AI hardware. But AI software is currently run at a massive loss by everyone who’s offering AI tools. And so one of the fascinating things is we’re going to get there — it’ll be profitable — the question is when and how.

Jack: On this Dario Amodei piece, one of the things I’ve been thinking about a lot is this — we can learn from all the other innovations of the past, but is this innovation on steroids because it’s intelligence? And do we magnify both the long-term value of this in terms of what it’s going to create, but also magnify the short-term pain? Because it is intelligence, it can replace more human jobs. Do you think that’s a fair way to look at it?

Rob: I think that is spot on. I think it will be more disruptive, perhaps, than any technological innovation since computers — since the railroad. Back in 1825, to get a message from Washington DC to New York, it had to be on horseback and it took two to three days, even if you were replacing horses every 25 miles and just kept going — a hundred miles a day was your maximum. Twenty years later it took one day because of the railroad. And ten years after that, it took milliseconds because of the telegraph. So there have been some humongous technological innovations. AI I think will be one of those.

Now, the leaders of AI today may not be the leaders of AI in 20 years. That’s why the massive CapEx spend — because they want to secure their place in the pantheon of leaders and feel that they have to spend hundreds of billions. I mean, Zuckerberg said as much. He said the cost of spending a quarter trillion on CapEx is horrific. The cost of not spending it may be much more horrific. And there’s a lot of truth in that. But the question of is it going to change our lives? Yeah. In more ways than we can possibly imagine. It’ll be massively disruptive — the most technologically disruptive disruption of my lifetime. And I’ve been around for a while.

Jack: What do you think the most important things are for investors to think about with this? Like, when I talk to clients, they’re always asking questions like, what is my exposure to AI? I want exposure to AI. But I think maybe the lessons from past booms may teach us that that may not be the right way to approach this. So how do you think people should think about investing in AI from the perspective of an investor?

Rob: Well, firstly, the narrative that drives the bubble is a narrative that says this is going to change everything. And that narrative goes on to say that the key players, the dominant players, have a moat that’ll protect them and it’ll be very difficult to displace them. Difficult doesn’t mean impossible, especially if you’re looking at a multi-year horizon. It also goes on to suggest that the change will happen very fast. I mean, AI that can relate to you in English has been with us since late 2022 when ChatGPT was introduced — what an innovation. But how many people spend more than a few minutes a day trying their hand at exploring what AI can do for them? I would venture to guess that at least 90% of the population doesn’t sit down and try using AI to explore what it can do for them more than a handful of times a week, if that. There’s 2% of the population that spends hours every day exploring it, and they’re going to own the future. So we all owe it to ourselves to learn.

Jack: You have a great practical definition of what a bubble is — one of the more practical ones I’ve seen. I’m just wondering if you could talk about what that is, but also how are you thinking about whether AI is a bubble based on your practical definition right now?

Rob: I don’t think AI is a bubble. I think AI stocks are a bubble. There’s a difference. Our definition of bubble is very simple: if you are using a discounted cash flow model — a Gordon equation type thing — to value an asset, you would have to use implausible growth assumptions to justify the current price. Not impossible, but implausible.

So Amazon as one example in the year 2000 was priced at levels that required what was then reasonably thought to be implausible growth assumptions. And sure enough, it was a disaster for the next decade. And then it got its mojo and it was no longer priced at levels that reflected implausible growth expectations. And sure enough, it became a wonderful stock — one of the most successful stocks of the last quarter century, but not in the first decade of that quarter.

So there are companies that go on to achieve growth greater than what you would need to justify the current price. Amazon and Apple are two vivid examples. The growth required to justify the price in 2000 has been exceeded for a quarter century. Cool. But those are the exceptions that prove the rule. The vast majority — we’ve talked about this in the past — of the 10 most valuable companies on the planet in the year 2000, only one, Microsoft, is still in the top 10. Only Microsoft has come anywhere near beating the S&P 500, and it’s only beat it by a couple percent a year. Of the 10 most valuable tech stocks in the world, the median result has been a negative return over the last quarter century. Over half of them have had negative returns. The ones that have been wildly successful — Qualcomm has seen 60-fold growth in sales in the last quarter century, 60-fold — and yet it’s behind the S&P 500. Why? Because it was priced to achieve that in 10 years, not 25.

Jack: I want to ask you about margins, ‘cause that’s something we’ve been talking about for a long time. Like, as a value guy, I’ve been one of these people who’s been saying margins have to mean revert, and they really haven’t been reverting for a long time. So I’m just wondering, like, first of all, if you have any thoughts on that idea in terms of why margins haven’t reverted. But I also wanted to bring it back to the idea of AI — do you think AI could lead margins to continue, maybe to go to higher levels than we think they could go to?

Rob: Mean reversion has been central to my career, central to everything I’ve done. Valuation multiples tend to mean revert — if they get stretched, they’re more likely to mean revert than to go further in the same direction. It doesn’t mean they can’t; it just means they’re not likely to go further in the same direction. One of the areas where mean reversion operates most powerfully is profit margins. Why? Because if you have a stupendous profit margin, you’re going to attract competitors galore. And it doesn’t matter how wide your moat is, somebody’s going to figure out a way to get past it.

Intel was the fourth most valuable company on the planet in the year 2000. It had a moat — nobody could touch them on making chips and doing it inexpensively with huge margins. Nobody could touch them until they did. Then Nvidia, TSMC, AMD, ASML — all of these wound up surpassing Intel over the subsequent 25 years. But it took time. I mean, Intel was still the dominant chip maker 10 years after, in 2010, and just slowly but surely frittered away its advantage.

Another interesting warning sign is the more dependent you are on government largesse, the more likely you are to not have a successful moat. When the CHIPS Act was passed, I thought, what horrible news for Intel — a $50 billion bailout from the government. What horrible news. And of course, they’ve gone on to achieve great disappointment.

Jack: It’s funny too, because thinking about these moats in real time is so hard to do. For instance, if you had asked me a few years ago about Google’s moat in search, I would’ve said there’s no chance — before 2022, like they’ve got such a strong moat. When you’re living through it, these companies seem so dominant. But now we’ve got ChatGPT. So now I’ve seen a way in which Google’s moat could be challenged. And I think that’s the case with a lot of these big companies in history — at the time it’s hard to see it, but then in the future something comes that you didn’t think of.

Rob: Yeah, we talked about this last time — using AI as a search engine. Google’s entire business model was predicated on sponsored links and sponsored pop-ups. And without those it has no profits. And then all of a sudden people realize, oh, I can use AI as my search engine, I’ll get no sponsored links and no pop-ups. How cool is that? So Google itself had to decide we’re going to disrupt our own business model and erode our own margins by introducing AI as part of our search engine, because if we don’t disrupt ourselves, we’re going to be disrupted out of business.

And then ChatGPT itself was disrupted by DeepSeek, supposedly created on one-hundredth the resources of ChatGPT. And the initial version of DeepSeek was rated to be as powerful as ChatGPT-4o. The other thing that’s interesting is how fast AI is evolving. GPT-5 is phenomenally more powerful than 3.5, and the hallucinations are way down. It does feel like you’re talking to a person who has multiple PhDs and intimate knowledge of more or less all of human knowledge. It’s so cool.

Jack: Do you think on, on the idea of mean reversion, do you think there’s a case that it’s been slowing down? I mean, I think we — you and I both believe mean reversion is still a very powerful force in markets — but you could argue margins haven’t reverted like you’d think. You could argue some of these big tech companies have grown at rates that you thought weren’t possible. Valuations have taken longer to come back than people think. Do you think you could argue that the process of mean reversion has just slowed down for some reason?

Rob: I’m not so sure about that. I think mean reversion has always been choppy and sluggish and feels like a random walk, but it’s a mean-reverting random walk. When a company’s at frothy multiples, one of two things has to happen — the fundamentals have to catch up with the price, or the price has to revert back to the fundamentals. So you’re going to get mean reversion on the valuation multiples for sure, but it can happen one of two ways. And so the question on defining a bubble is, would I have to assume implausible growth for it to close with the fundamentals catching up?

That graph in the upcoming paper that looks at the relative valuation and relative growth of fundamentals for members of the indexes and non-members is just a fascinating case study. Does this mean that we’re going to see mean reversion in the next year, next five years, for small cap and value to mean revert towards large cap and growth? Not necessarily on a short horizon. I would say high odds on a long horizon. And that’s because we’ve seen it happen again and again and again.

Nvidia has vast margins. So does Palantir. So do a host of others. But to some extent they’re competing with one another — not so much Palantir and Nvidia, one is a user, the other’s a supplier — but they wind up having competition and competitors coming in. If your horizon is three to five years, there will be competitors. I don’t care what your moat is — within five years, if your margin is 50%, you’re going to have competitors. And on a 10 year horizon, the likelihood of your margin remaining as far away from industry norms as it is at the start of the decade is really remote.

So I do see mean reversion coming in the profit margins — very high odds. And the interesting thing is people think that I’m poo-pooing the AI revolution. I’m not. I think the AI revolution is very real and is going to continue to surprise us for years and years to come. I’m just questioning who’s going to come onto the scene that will have better ideas and displace some of the leaders, and who are the end users who are going to benefit in ways that can’t be anticipated.

Ultimately I think the big winners may be organizations way outside of the tech arena. I mean, if you’re running a trucking company and your dispatcher isn’t using AI to figure out what trucks to move where, you’re behind the curve. And things like that are coming.

Justin: You had mentioned international stocks before and maybe we’ve kind of hit on a little bit of this, but when we had you on previously and we were asking about how you were investing your personal portfolio and long-term assets, you know, you were pounding the table on emerging market value stocks and that ended up being a phenomenal call.

And so what I wanted to ask you related to international is — obviously the last couple days it’s been kind of painful if you’re new to allocating to international.

Rob: Yes.

Justin: Yeah, a little. But that’s kind of to your point — those international markets are going to be more impacted by things like this, maybe more than the US, at least initially. But what I want to sort of ask you is — if you were trying to convince someone, you know, if you look across most US investors’ portfolios today, strong home bias, very little international — if you were trying to make the case for allocating towards international, how would you do that? How would you emphasize sort of the long-term market history, the risk and diversification that international gets you? And then kind of going one level deeper — where would you be focusing that? Would it be broad-based international exposure? Would you be a little bit more developed-market focused?

Rob: Unless you want to take the time to dive in and study individual markets, broad-based exposure is I think the right answer. Fundamental index has been more powerful in emerging markets than in developed markets, I think because they’re less efficient.

RAFI — fundamental index, for those of your viewers who aren’t familiar with it — simply says, instead of creating a portfolio that looks like the stock market, let’s create a portfolio that looks like the publicly traded macro economy. Meaning that you choose companies based on how big they are and weight them based on how big they are — not on how expensive they are, not on how big their market cap is, but how big are their sales, how big are their profits, and so forth.

In any market, if you cap-weight stocks that are trading above their eventual fair value, they’re overweight in your portfolio. So you’re overweight the overvalued companies — you don’t necessarily know which ones they are, that’s the problem. But if you break the link with price, then companies that are overvalued might be overweight or they might be underweight; companies that are undervalued might be overweight or might be underweight. The errors cancel. And so there’s this rebalancing alpha that’s worth about 2% a year — and that’s not speculation or theory, that’s actual observed results. In emerging markets it’s more like 3 or 4% a year.

So you can buy fundamental index-based ETFs and mutual funds that will give you broad diversification that won’t pull you into putting most of your money in the countries that are the most expensive. It’ll put you into the countries and companies that are most important to the global macro economy. And if you do that, you get broad diversification, you get risk that’s not dissimilar to conventional cap-weighted indexes, but you capture that mean reversion alpha. If a stock soars — or a country’s stock market soars — and its underlying fundamentals don’t, a RAFI-based portfolio will say, thanks for the nice gain, I didn’t see any fundamental validation for it, I’m going to trim it. And that rebalancing alpha is just really powerful.

Justin: Is there anything with the long-term trend in your opinion on the strength or weakness of the US dollar and how that might play into that? I know most of these strategies don’t hedge currency or anything like that, but do you have any feelings on that?

Rob: It’s interesting. I was at a conference just a few days ago where there was a panel discussion on the dollar. When I went to this meeting a couple years ago, the question was, why invest anywhere else when the dollar’s clearly the strong currency and going to continue to soar? And then this go-around, it’s why would you invest in the US when the dollar’s going to continue to tank.

Mean reversion happens in currencies too — they can get ahead of themselves. I think the dollar looks a little weak. One of the things I like to do is use a CPI-adjusted currency, and then you can compare them around the world. The US dollar’s a little on the cheap side relative to where it’s been in the past. So that’s one very minor argument in favor of maintaining that US focus. But part of the tremendous strength of international and emerging stocks in the last few months has been dollar weakness.

Jack: You wrote a great paper recently — Trifecta: A Fundamental Revolution in Indexing. And you just talked about the first leg of the trifecta with RAFI here. But before we talk about the different legs, I just wanted to maybe see if you could talk at a high level — like, what were you trying to accomplish with this paper?

Rob: When we developed fundamental index back in 2004, we knew we were onto something important. And it’s become — there’s almost $200 billion invested in fundamental indexes around the world, and probably a similar amount invested in imitating and otherwise similar products. So that’s a lot of money. It’s probably the most successful non-cap-weighted index in the world.

When we developed that, we knew it was going to be important. We thought if you choose companies based on how big they are and weight them based on how big they are, you’re going to get a rebalancing alpha, you’re going to get a value tilt. I thought this will probably add 50 or 100 basis points a year. No — in back-testing it added over 200. Live, it’s added over 200 when compared with an appropriate value-oriented cap-weighted index. The outperformance has been relentless.

What I didn’t realize at the time was that one of the core attributes of RAFI involves fundamental selection — not choosing companies based on how frothy and expensive they are, but based on how big their business is. And fundamental weighting — weighting companies based on how big their business is.

Fundamental selection also works for cap-weighted indexing. Imagine I said to you, I’ve got a great idea, been working on this, and I’m really excited. We will buy a stock — any stock that has soared past a certain threshold. On average it’ll be priced at twice the market multiple. On average it will have beaten the market by about 75% in the last year before we buy it, but it’s clearly onto something and it’s headed for great things. Now some of them don’t work out. My sell discipline is the opposite — we’ll sell it if it tanks below that threshold. Could be top 500, could be top thousand. On average we’ll sell it for half the market multiple, and on average we’ll sell it after it’s underperformed by 7,000 basis points. But you know, most of these companies do go on to good things.

What do you think — you wanna invest with me? Described that way, it sounds like an absolute dithering idiot. But that’s the way indexes trade. And it’s because they buy into this mantra that markets are efficient, prices are efficient. And so yeah, if you want representative stocks that mirror the look of the stock market, you’re going to cap-weight because the market’s cap-weighted. And you might as well cap-select, because otherwise you’re including companies that are too small in market cap to matter.

Well, it turns out that you can use fundamental selection, not just for creating RAFI, but for creating a better cap-weighted index. And for creating a better growth index. So the essence of Trifecta is we decided to launch a better cap-weighted index back in 2021. We did the work in 2020 and ‘21, we launched in ‘21. The ETF was launched last year. We started working on whether this could also be adapted to growth investing and did that back in 2023 to ‘24. The index was launched in early ‘25. And I’m not allowed to speculate on whether ETFs might be in the pipeline or not, but it wouldn’t be shocking.

So you can use fundamental selection to choose better growth companies, to choose more representative companies for a cap-weighted index, and to have a better value strategy. That’s a trifecta. If you have a world-class value strategy — RAFI — a world-class cap-weighted index, and a world-class fundamentally selected and fundamentally weighted growth strategy — RAFI Growth — and it works everywhere in the world, and they add value with statistical significance everywhere in the world, that’s a trifecta. I’m excited. I feel like a septuagenarian getting ready to launch another revolution. It’s just a hoot.

Jack: Would there be a way to combine all of these three things into one index?

Rob: Absolutely. But let’s start with your core index fund. If your core index fund chooses companies based on their market cap, based on crossing that threshold, you’re buying stocks that are frothy and expensive after they’ve soared. And by the way, you missed it. If it underperforms badly enough, you’re going to drop it after it’s underperformed. And by the way, you rode it all the way down and you’re selling cheap.

If you can mitigate that by choosing companies where the fundamentals are now big enough to matter, regardless of price level, then you have a more powerful core cap-weighted index. And it has 99.9% correlation with the S&P or the Russell 1000, or for global investors, the ACWI. 99.9% correlation — that’s pretty cool. And ballpark of 50 to 100 basis points incremental performance — that’s pretty cool. So that’s your better core.

Now, what’s another way to do things? We have a paper coming out in the next issue of the Financial Analysts Journal, I think also next month, called “Fundamental Growth,” that looks at our research in applying fundamental index principles to choosing growth stocks. And what we find is if you choose stocks based on how fast they’re growing, percentage growth, and then weight them based on the dollar magnitude of that growth — so you’re weighting them in proportion to their contribution to the growth in sales and profits of the total economy — you wind up with something that over the last 30 years beats the Russell Growth index by about four and a half percent a year. Very cool.

RAFI beats the value indexes by two to two-and-a-half percent. RAFI Growth wins historically — it is a back-test, but it’s a very simple back-test, it’s not data-mined. Four and a half percent — let’s haircut that and say maybe it’s going to be two and a half, the same as RAFI in the future. You could put those two together and say, instead of buying RAFI alone, I’m going to put half my money in RAFI and half in RAFI Growth. Instead of adding 50 to 100 basis points, I’m going to add two to 2.5%. That’s really cool. But it’s not an inclusive index — it doesn’t span the market, it leaves big chunks of the market out.

So you have companies that are cheap and growing slowly — those would be considered standard value stocks. Well, we want cheap, but we don’t want sluggish growth. If you have stocks that are expensive and growing fast, that’s a growth index. It’s treated as a simple, linear, binary choice: it’s either growth or value. Throw that out the window and say, if it’s cheap, it’s value. If it’s growing fast, it’s growth. Now you’re leaving out companies that are expensive and growing slowly. That fourth of the market historically underperforms the market going back — and this we’re able to test way back, we’ve tested it over 50 years — that quadrant of the market underperforms by 2% a year. If you’re lagging by 2% a year for 55 years, you end up less than a third as wealthy as you could have been. So leave those out and you wind up with a strategy — I’ll call it a strategy rather than an index — that captures the best of value and the best of growth. It’s not a core index in the sense that it leaves a fourth of the market out.

Justin: What I love about the research that you guys put out, Rob, is that there’s always these little — like, I’m gonna use the word nuances, it’s probably not the right word — but like this conglomerate piece that your colleagues put out. Like, somebody had to think about, okay, what did conglomerates look like in the past and what type of valuation did they command? Or maybe not command, and how does that stack up relative to today’s conglomerates that investors sort of see in the index. And so I just think it was a great paper. And I think let’s just kind of talk through some of the things that they found and maybe some of the possible takeaways for investors. I mean, to start, based on the historical data, it looks like conglomerates have had this diversification sort of discount, as the paper called it. Can you kind of help us work through what that actually means?

Rob: Back in the sixties, conglomerates were priced at a substantial premium. And that was because, gosh, they can put their money anywhere. The opportunity set is everywhere. ITT — not AT&T, ITT — was a famous conglomerate that was involved in telecom, transportation, retail, and everything else under the sun. And the narrative was the management can steer company resources to wherever the growth is. Boy, isn’t that a powerful tool? So we found that at the time they were priced at a premium.

Then it turned out that running too many disparate businesses, you’re not liable to be very good at some of them. And you’re liable to be reactive — chasing what’s been successful rather than anticipatory, putting money into businesses that are poised to take off. So we went from a conglomerate premium to a conglomerate discount.

Now you’ve got big businesses. Several of the Magnificent Seven are diversified into a wide array of businesses, and the narrative is they’ve got it covered. So we’re back to a premium. And the question is, are some of them going to stub their toes and turn out to be not very good at choosing where to allocate their resources? For example, that $650 billion of anticipated CapEx next year on data centers and other infrastructure for AI — they’re all chipping in a hundred billion here or 150 billion there, adding up to those lofty numbers. Are some of them going to turn out to have wasted their money? Probably.

And then the conglomerate premium can go back to a discount. That’s not a projection — it’s a scenario that is possible that gets very little attention today.

Justin: Yeah. And one of the charts you had in the article — for Amazon, Apple, Alphabet, and Microsoft — you kind of show the revenue makeup starting in 2015 and how these businesses have gotten more diversified, gotten into more non-core services over the last 10 years as they’ve looked to expand their business model into some of these areas that they may not be as good at, that may not be as profitable as what the core businesses are.

Rob: Yeah. If you’re Apple, how are you going to move the needle on a business the size of Apple’s with a new iPhone? Come on.

Justin: Right.

Rob: If you’re Microsoft with a new software package, no. And so the presumption is, well, we’ve got to find something to do. You don’t have to find something to do. You can return the money to the shareholders. You can say, we’ve done a great job for you, here’s a bunch of money for you, you figure it out. Now, companies don’t like to do that because they have the hubris that we know way more than our shareholders do. And they do in their core businesses, but they don’t when looking at the aggregate opportunity set in the macro economy.

Justin: One of the charts in here shows that this conglomerate premium — right now, when you take those four mega-cap tech companies, it looks like on average they’re trading at a 70% conglomerate premium.

Justin: What explains that premium?

Rob: The narrative. Narrative sets prices. The narrative that these companies know what they’re doing and they’re building our future, and you better get on board or you’re going to miss it.

Justin: What does that imply for market-cap-weighted investors, given these are the largest companies in the market today?

Rob: I think we are likely to see a pivot back to value. Value is very nearly the cheapest it’s ever been. You’d have to see value stocks beat growth stocks by a hundred percent — they’d have to double relative to growth stocks — just to get back to historic norms for relative valuations. Small would have to double relative to large cap in order to be back to historic norms of relative valuation. Now, I’m not saying small cap is going to double or value is going to double, but some sort of mean reversion where they move in that direction is certainly possible.

If you look back at the dot-com bubble as a wonderful example: the narrative coming into the dot-com bubble was get on board, this internet thing is huge, and it was. And these internet companies are going to be stupendously successful. And some of them were, but they were also priced as if they were going to be even more successful. The narrative was these companies are where you’ve got to invest ‘cause that’s the future.

Well, the first two years after the bubble burst — let’s choose March of 2000 as the bubble bursting — the first two years after that, NASDAQ was down a little over 50% by March of 2002, on its way to a drop of just under 80%. The S&P was down 27%, on its way to a 46% drop. The Russell Value was down four. The Russell 2000 was up four. And the Russell 2000 Value was up 53%. So you literally tripled your money if you pivoted from Nasdaq into Russell value at that moment, if you had the prescience or the luck to choose that moment.

On a 10 year horizon, our work suggests that small-cap value will beat large-cap growth by on the order of 700 basis points a year on a 10 year horizon. That’s enough to double your money relative to sticking with growth. I’m not saying get out of growth — partly because who knows what the right timing is on this, but also because people will get cold feet if they make the move at the wrong time. What I am saying is fade some of your winners, buy into what’s out of favor and cheap, and just lightly average in to increased exposure to what’s newly cheap.

Justin: Just in closing on this — like the article, you know, started with the story of GE and it kind of came all the way to the late nineties when Jack Welch was, you know, doing all these roll-ups and making GE into this like massive conglomerate. And I just remember in the late nineties, early two thousands — I think my timing’s right here — GE was top of the food chain. It was the preeminent company, maybe the most valuable company in the S&P 500, and then it sort of all came somewhat crashing down. The company has turned around now, they split off different units, and it’s been much better as they’ve kind of realized the value of the different parts of the business.

But the point that stood out to me as I was reading this is you see those lists of, for each decade, the most valuable stocks in the S&P 500, and for all of market history that we have, they’ve always been different and they’ve always been changing. And so that’s kind of what really made me think when I was reading this — I’m like, you know, when we look out 10 to 20 years from now, the most valuable companies today are probably not going to be at the top of the list. I don’t know if you agree with that or not, but that’s kind of what it made me think of.

Rob: That’s a reliable pattern. The companies in the top 10 — on average, seven or eight of them are gone within 10 years, no longer in the top 10. And on average, eight or nine of the 10 underperform the market over the next 10 years.

Now sometimes — the 2010s being a vivid example — there were only two winners out of the top 10: Apple and Microsoft. But they beat the market in the 2010s by enough that the top 10 list actually did fine. But in most decades, no, the top dogs — the very business practices that propel you to being a top dog are suddenly decried as predatory. So regulators are all over you. Your competitors are gunning for you. Your customers are no longer fans of you because you’re too big for your britches and you’re too arrogant. Staying a top dog is really hard.

Justin: Rob, thank you very much. We’re looking forward to following all the great research that you guys are putting out. Please don’t be a stranger — you’re welcome back anytime. Thank you.

Rob: Thank you so much. This has been great fun.

Jack: Thanks, Rob.