Transcription
Welcome to Bogleheads on Investing, episode number 88. We'll be speaking with AQR's Antti Ilmanen, a principal and global co-head of portfolio solutions. We'll be discussing Antti's new series of papers titled, "How do investors form long-run expectations?"
Hi everyone, my name is Rick Ferry, and I am the co-host of Bogleheads-On Investing. John Luskin is the other host, and we're now switching back and forth, bringing you a greater variety of topics and guests.
This episode, as with all episodes, is brought to you by the John C. Bogle Center for Financial Literacy, a nonprofit organization that is building a world of well-informed, capable, and empowered investors. Visit the Bogle Center at boglecenter.net, where you will find a treasure trove of information, including transcripts of these podcasts.
Before we begin, I have one announcement. The Boglehead conference last month in October in San Antonio was a huge success. I don't know how it gets better and better every year, but it does. We had fabulous speakers, great sessions. We had record crowds, more than we thought we're going to attend, and this just keeps getting better. It's just phenomenal.
Every session was video-recorded. Those videos will be on boglecenter.net by the first of the year. We have a production crew that does a great job editing the videos and adding slides and so forth. So, look for those by the beginning of the year on boglecenter.net.
Today, I have a repeat guest, Antti Ilmanen. Antti earned his PhD from the University of Chicago. His PhD advisor was Nobel laureate Eugene Fama. He is a principal and global co-head of the portfolio solutions group at AQR Capital. He's published extensively and has received numerous awards. Antti is the author of two books: *Expected Returns*, published by Wiley in 2011, and *Investing Amid Low Expected Returns*, published by Wiley in 2022, which was the subject of podcast number 49.
Today, we're going to be talking about a series of papers that he recently published, 10 papers on "How do investors form long-run return expectations?" These papers are free on the AQR website. Each one is between 10 to 15 pages long, and I found these papers to be fascinating. There's a lot to be learned from understanding how we form expected returns of the stock market and how we form expected returns of interest rates because everybody has an opinion. And so this digs into those opinions, and it also digs into the academic side, which is not formed based on opinion but formed based on data. And there's a tug-of-war that goes on between the subjective opinions and the objective data, which is the crux of the study.
So, with no further ado, welcome back, Antti.
>> Thanks. It's great to be back.
>> You've been busy. You just finished a 10-part paper on how investors form expected returns, which is the topic of our conversation today. So, my first question is, what compelled you to look into this topic?
So, my own motivation why I decided to write it was twofold. One is I consider at least a little funny story from my past as a PhD student in '91. I was Eugene Fama's one of the PhD students he had, and I went to his office and asked about my dissertation topic. It would be about interest rates because that was my past specialty. So I asked him, could I also use some survey data to look at how investors are actually forecasting interest rates? And he took a piece of paper in his hand and gently looked at me and said, "Antti, do you want to know what I think about survey data?" He threw that piece of paper behind him into the wastepaper bin. This was very impressive, but this was also sort of heartbreaking for me because I knew that I was too much of a chicken to really then use that survey data in my dissertation. So I subsequently I've used some survey data, but I've always wanted to promote it more, and I've done it a little, but I thought that it's sort of at this advanced age, it's sort of nice to come back to it and really, really push it.
And apart from that, there's this idea that I have, a worry about rearview mirror perspective, which is people look too much at the last 5 to 15 years, what's been happening in the market, and extrapolating future returns based on that. And I think we have had such an exceptionally benign environment overall for equities, but especially US and tech. And I think that's really dangerous to use as an expectation for the future. So I'm trying to do this contrasting of objective and subjective expectations.
>> We have a lot to cover. So let's get into part one, basically the scope of the paper. You're describing what you're going to do in the next nine papers and looking at the difference between forward-looking expectations and rearview mirror expectations. So the forward-looking expectations are the objective expectations based upon interest rates, based upon earnings growth of the market, and so forth. And the rearview mirror expectations are the ones that are, "What has the market done for me lately?" and extrapolating that into the future.
Yeah, one could say that extrapolation and rearview mirror is sort of what comes easily to you, and sort of the human thing. And then there is a more rational and more deliberate who is doing this objective expectations. So I'm contrasting through the series this idea that let's look at market-based data, which because they have tended to on average forecast future returns in the right direction. Basically saying that if starting valuations are high, and therefore starting yields for bonds or stocks or whatever asset are low, high valuations mean low yields and therefore low expected returns. By the way, even without mean reversion, just the fact that you have got lower income coming through. So high valuations predict low future returns. That's empirically what we have tended to see in the long run, and sometimes there are huge exceptions to that. So US equities have obviously done very well, even though they looked expensive for a while.
And the contrast is then that subjective expectations, they are things that can be read off surveys. And investor surveys can be very different. Retail investors seem to have different patterns than institutions, and maybe equities different from bonds. We may come to that. But so I sort of highlighted many of those surveys have got this extra rearview mirror tendency, which is, I would argue, dangerous, even though it sometimes gets you right.
I picture subjective and objective returns as two lines on a graph. One of them, as the stock market return goes higher, the objective academic expectations go lower because the earnings yield goes lower, interest rates go lower. But as the market goes higher, the subjective returns or the expectations of investors or some investors in the stock market go higher. And vice versa. So if we have bare markets like in 2007-2008, the subjective expectations of return for primarily retail stock investors go lower because the markets have not done well lately. But because earnings yields are higher, the expectations of the academics who are following the data and creating objective returns go higher. So you've got this cross in the middle of the chart, and that there is an equilibrium that seems to be passed through as investors are going one direction and academics are going the other. Is that a fair assumption?
>> So in equities, it works like you said. In 2008, it was like that. My favorite is really thinking about the dot-com bubble, '99-2000, and actually, like sort of the little sisters of that happened in 2021, and again, '24-'25, we had these very high CAPE ratios approaching 40. So we had this high valuations, low objective expected returns, and very bullish growth estimates, for example, both from retail investors and CFOs. And my favorite is to look at equity analysts' long-term earnings growth forecasts. They were super high. And it is interesting to think that so we see objective low just when subjective is very high. And that's totally inconsistent with what I sort of learned in Chicago with, for example, Professor Fama, would have told that the rational reason why we had very low expected returns or required returns on equities in 2000 was that investors were feeling very wealthy, they had very low risk aversion, they required a lower equity premium. So that was a nice rational story, but that wasn't at all consistent with the finding that those surveys were telling that there's very high expected return by actual investors or equity analysts. And that story is better told through an irrational lens. And it's basically saying this kind of extrapolation that after very strong years in the 1990s, equity analysts and investors were over-extrapolating what had been happening in the past and getting overly bullish. So typically, this happens after we have had many strong years in the market, or if there's just some really strong story like internet and AI is recently. So in those situations, you tend to get this excessive growth optimism causing very high valuations, and those high valuations mean low expected returns. And because it's very difficult to satisfy those high valuations.
>> I like the quote from Alan Greenspan back in 1996. And I know I'm jumping ahead a little bit here.
>> Yeah. But how do we know when irrational exuberance has unduly escalated asset values, which then become subject to unexpected and prolonged contractions? Now, after Alan Greenspan said that in 1996, the market sold off a little bit because people took that as meaning Alan Greenspan thinks the stock market is overvalued. But from then to 2000, the S&P doubled. As the market goes higher, and the academics and the objective analysts like yourself believe that the expectations are lower, this can go on for a while.
>> Yeah. Yeah.
>> Keynes said, "The market can remain irrational longer than you can remain solvent."
>> And that's very relevant in today's environment because again, valuations are very high. And I would definitely lean on, and I do lean on, being cautious here. Not surprising, but I'm very humble about that because of the very story that you tell that there was three and a half years of continued bull market after that Greenspan quote. And Shiller himself had been, and me as a young PhD student, had been sort of cautious on equity markets many years before that because any measures had already been telling that markets are pretty expensive, even before '96 again. So that's just telling how difficult it is. But I think the flip side is that I think people really don't easily get this idea that when we get lower required returns, so the sort of this Fama logic, time-varying required returns, when people get more bullish and demand ever lower equity premium, like happened in 1999. Part of that story then is, you are quite right, that those situations can last uncomfortably long if you are a contrarian. And the worst thing is that, you know, you are sort of betting on some mean-reverting valuations, and then if there is a structural change, as there was with equity valuations, we never got full mean reversion, really. So like the CAPE ratio, which was averaging 15 through the 1900s, really came back to these under-20 levels for a during the Global Financial Crisis. Then it gave a good signal. But somehow, anybody who had been sort of as a contrarian investor had been waiting for this normalization, that let's go back to 15, has been probably underweighting US equities too much in the last 20, 30 years because we just went from 15 average to 25 average or 30 average CAPE ratio. So those are challenges for these types of contrarian approaches, apart from the smaller problem of being three years too early. Yet, despite all of that, on average, these things have worked. When we look at, so, 5,000 years of data, and they gave good warning signals in the 2000s and good buying signals since 2009 or so. But since then, we have had basically pretty much re-rating US equity markets without yet getting the payback time for that, and we'll see whether it comes.
Just a quick point of clarification, the CAPE is cyclically adjusted PE from Robert Shiller. All it is is looking back 10 years at the earnings of the S&P each year and taking the average of that 10-year period and dividing that by price to come up with a long-term valuation metric of whether the market is overvalued or undervalued. It doesn't give more weight to current earnings, which are much higher than they were 10 years ago. They give equal weight to all years. So it's a very blunt instrument. We'll be discussing Shiller's CAPE ratio. You go more in-depth into it in parts of this analysis.
>> The second paper is US versus international stocks.
>> Yeah.
>> You know, we know that most US investors have a US bias. We like to invest at home. The US economy is 25% of the global economy, but the stock market is 70% of the international market as far as valuation. So, I hear this from investors a lot. They say, "I'm really not hot on international stocks because they have performed so badly." It takes me back a little bit because when I look at the 10-year return of the Vanguard Total International Fund, which is equity market, including developed and emerging markets, it has annualized at 8.3%. And you know, I mean, if you're looking at an equity risk premium for the international market, you'd give it maybe 4% over 10-year Treasuries, and that would bring you to about 8%. And so the international market, in my view, has done just fine.
>> Agreed.
>> Done well in the last 10 years. The problem is the US market has done much better. The US market has done 14.6%.
>> Shame on it.
>> So we tend to look at the US market and compare everything to that, when in fact, we're not looking at the returns of these other markets.
>> Yeah. US has been a tough benchmark to beat. But it is also true that when we look at the last 50, 60 years, we've seen US equities underperforming in the decade of the 2000s, in the 1980s, 1970s, like three out of six of the last decades or so. If we include this one, it's not that US always outperforms. But I think again, for people who have got short memories, rearview mirror of young people especially, they only recall that post-GFC era, and they think that US always outperforms. And I think that's dangerous.
To be fair, though, there are some older gentlemen who were favoring US, John Bogle, Warren Buffett. They were correct. So I think I think there is something. If I cannot talk both about the valuations of US versus non-US and then the positive exceptionalism and the growth advantage, I think it is important that people recognize that even when we look at the data in the last 15 years, when US pretty much outperformed all the time, much of that outperformance came from re-rating. So compared to the other markets. And so, yes, US outperformed, but it wasn't that most of it was due to some fundamental growth advantage. There was a decent amount of that, but most of the return outperformance was coming from US getting relatively richer. So again, the CAPE ratio rose to 40 recently. But what I look at especially is the relative CAPE, CAPE of US compared to the CAPE ratio of the other markets, where I use pretty much like the eight largest non-US markets to proxy the ex-US on that one. It's interesting that we saw roughly similar valuations in much of '95 to 2010 period, and since 2010, US went on a tear and it hit almost double valuations at the end of last year. And I love to then sort of go back further in history and show that actually there's sort of a mirror pattern that in 1990, US had half the valuation of the rest of the world, and that was a time of Japanese Nikkei bubble, when Nikkei was a very big part of non-US and very, very expensive. So that's how US was up. And actually, when we look at historical data, how well did this valuation ratio predict future relative performance? Beautifully in these times of the '80s, '90s, 2000s, and very poorly since GFC, because it was always telling that you should expect US to underperform, but US kept doing well. But I think it is important for people somehow to remember that much of the US outperformance came from valuations. But that's not the only thing. Again, there is something from the growth story. And so there's it's a nuance story in the sense that when we look at, let's say, 50 or 100 years of historical data, we do find that US has had a fundamental growth advantage. So earnings per share growth has been about 1% higher in US than in the rest of the world. And that's great, and that probably does reflect things like entrepreneurial culture, more capital-friendly environment, maybe bigger markets, and tech edge, and so on. There are different stories to US exceptionalism. But I would say that 1% is sort of the norm there. However, when we look at the period since 2010 or so, there's been a two and a half, 3% edge for us, which is max tech-related. And I think people are now thinking that this is now going to be with us forever, and that I think is unlikely. We've never seen something like that to work for multiple decades. Now it has been reality for 10, 15 years, but to assume that that persists till kingdom come, I think is too much. But that logic has, I think, pushed those relative valuations to this double US valuation, which means that US is now, I think, pretty vulnerable in the future. You really need some great things to happen. Of course, the great thing could be exactly the AI story, which is right now making people so excited, might turn out to be true. But I guess I am cautious on that front.
>> So your estimates for real return on international stocks are about 6%. And your estimate for real return on US stocks are about 4%. So you're pricing in a repricing of US stocks into your forecast.
>> Yeah, that involves already then some repricing. Normally, we just say yield plus growth. So when you have got these higher valuations in the US, that means lower starting yields. And since the growth differences, we don't see them as big. We see not quite a 2% but I think, yeah, if you add some mean reversion there, then you are going to get something like the 2% advantage. But I think anybody who uses yield-based or mean-reverting valuations-based estimates on thinking about future returns will predict low and has predicted lower expected returns for US. And that, of course, has turned out to be wrong. And I think that's sort of given capital market assumptions a bit of a bad name now in the last decade being so off with the US. We may come back to that, but I'm just going to say it's just one decade. It's of course, that's not going to make me many friends in investment circles. 10 years is an eternity, and I'm saying one observation, but that's what it is. And if you want to look at seriously these things, you have to look at multiple decades, and they then start to tell that these things tend to work, and sometimes they get this exceptional decade against them, which has been just behind us. I think now we have got such extreme valuations that it's really difficult for us to keep outperforming on a multi-year basis from here.
>> Well, this is the difference between objective and subjective. I mean, you're objective. You are looking at the data. You're looking at the yields. You're looking at the interest rates. They're looking at growth to come up with your reasonable expectation of return going forward, where subjective investors are just looking at what happened in the past and extrapolating that into the future and thinking it's going to continue. And they may have good stories. They may have, like anybody who really got this idea that we are moving from this physical world to digital world and rode that well, good for them. But I think again, from this point, saying that this is just going to keep going like that and exponentially further and so on, that is a very aggressive view. And I'm sort of happy to lean against that one. Historical long-run data agrees with me that that's sort of what has happened in historical technological revolutions, that you get these types of episodes. But again, back to your Greenspan comment, nobody knows whether we are on the fifth inning or the ninth inning in this kind of bull market. I am definitely leaning more towards the ninth to 10th inning.
>> Well, let's switch from stock investors to bond investors and go on to part three of your study. Bond investors, you found, are different than stock investors. They form their expectations of future return differently. So, could you explain that?
>> Yeah. Yeah. Yeah. I really like this for a couple of reasons. It is sort of an original finding that I have got here, and I'll already tell the bottom line that while equity, many equity investors tend to be extrapolative, bond investors tend to be contrarian, having mean-reverting ideas. But let's just tell that you may have seen some pictures of yields falling in the 2000s, 2010s. Every year, we get lower yields, and the expectations of either economists or market expectations were just for higher yields. So year after year, they were wrong forecasts, and they were sort of laughing off of the silliness or the bad track record, and "Don't they ever learn?" type of thinking. The interesting thing is that newest research by academics says that those expectations for rising rates were quite rational because we got really an exceptional outcome in yields. When we look at the last 50-year history, first we went from this normal maximum 6% yields or something like that, that was broken in the late 1960s, and we went to double-digit yields. And then from the early '80s onwards, we came down from those double digits back to six, and to three, and to one and a half, and in Europe, even to negative levels. And so that type of persistent decline is something which was truly surprising for virtually everybody. And if you get those types of persistent structural shifts like you did there, sort of the rational right thing to do was to predict normalization, mean-reverting towards higher yields. It just didn't happen until it happened in '22. That's sort of a description of the history. But it is also hinting at this idea that there was mean-reverting expectations. That was a trend story. But there was a cyclical variation in those forecasts that whenever we had very low rates, then you had upward-sloping curves. But whenever we had a bit higher rates, such as after '22, '22, '23, the expectations turned to lower rates and inverted yield curves. This happened for the last 20, 30, 40 years that there was a correct expectation from the market in the cyclical sense. Today, abnormally low rates, you should expect rates to go higher, and from high rates, you should expect things to go lower. So that's that's contrarian, mean-reverting expectation. That to me was an interesting result, but it was especially interesting since I had just been reporting lots of extra behavior by equity investors. So why do the two groups behave so differently? And my best explanation is related to something what I call salience, sort of easy to access, what comes to mind easily. And so bond investors tend to quote things in yields, tend to talk about yields, they are naturally forward-looking, which gives them this sort of mean-reverting spirit. If bond yields have fallen to near zero, that doesn't make bond investors excited, bullish. But if you get something similar happening in equity markets, people don't worry so much about valuations. They are excited about the rearview mirror, perhaps because we tend to talk about prices in equities, and we talk about past performance, last year or last five years, whatever. And that that makes people, whatever that's the natural way of thinking for them, that gives naturally this kind of extra mindset for equity investors. And again, that has tended to lead them sometime to bad places, such as in 2000 and 2009, sort of getting just the wrong way around.
>> Well, I do find that fascinating that maybe because we quote stock prices, you know, "What is the stock market doing? How much percentage has it gained? How much is it up?" That's how we quote stocks. But with bonds, we don't quote what the value of the 10-year Treasury is. We quote what the yield is.
>> You know, it's 4%, it's 3.9, it's 4.4. Yeah. And because we quote these things differently, it changes perhaps people's perspective of this. I think that's a fascinating insight into why this might take place.
>> Yeah. And seems seems like some people have noted that real estate is a bit like equities then here, that people again, maybe price, but certainly people talk more about what has happened recently and extrapolate that, then somehow think about the fair value of the market or what's the some kind of forward-looking estimate that just doesn't happen so naturally for lay investors, anyway. It's really a kind of a dual component between interest rates and price because you need to borrow money to buy real estate. So you don't need to borrow money to buy stock. Most people are not doing that. But most people borrow money. So the interest rate component is there to buy real estate, which makes it a price component. So it seems like it may be a combination of the two. Logically, certainly. But again, like let's say some academic studies suggest that it is it is more the extra speculative part that drives things. And I did think exactly, you know, when we were talking early about what happened in 2022, 2023, when when real yields went from minus 1 to plus 2, I think, you know, for many real assets, including real estate, there had been all this talk of how you can justify the high valuations because low real yields, real bond yields, real interest rates. And yet, when when when real interest rates rose sharply, that type of logic was sort of pushed aside pretty pretty quickly by by marketers and so on. And I and I think that led to basically frozen markets where we have been now for for quite a while.
>> I'd like to go on to part four. Now, here it's going to get a little geeky for listeners. If it hasn't been already, this is going to get even worse. Okay, this is where you lay out how practitioners like yourself come up with their capital market assumptions, like what they believe the future of the market will be. And there are four different basic formulas for coming up with objective expectation of return will be for the market. So, could you go over those four?
>> Yeah. Yeah. Yeah. In the 1900s up to really 2000 and early 2000s, almost everybody when they were asked about expected long-run equity returns, they just went to the so-called Ibbotson Yearbook, which showed basically historical returns since 1926, and they read it from that. It really meant that they were assuming there is a constant equity real return or equity premium or whatever it is that they were quoting from that book, and they were just taking a long-run average. At least, by the way, I'm glad it was a long-run average, that's better than taking a last three to five-year average or something like that. So that was the standard thinking about constant expected returns. And somewhere in the 1990s, it was sort of weird that when we got these four very strong years, '96, '97, '98, '99 for S&P 500, every year that historical average return was getting higher. And you are supposed to be more bullish at that time. When when this other school of thought was saying that high valuation should make you more worried, and you got people like Fama, French, Campbell, Shiller, Campbell, John Bogle, actually Rob Arnott, they were all telling a story there that those high valuations can be a problem. So they were turning to this forward-looking approach, which was basically saying that don't look at just historical averages, there is some kind of time-varying expected return, and you should look at those starting yields or valuations. And so there was sort of a tug-of-war by the time we were in 2000 between these two ways of thinking about this.
>> Is that when Shiller came out with this CAPE ratio?
>> A bit earlier, he really presented that idea to Greenspan just a few weeks before Greenspan's comment. There these ideas had been coming up during those years, and they were sort of heating up by the time we were in 2000. Cliff also had a great paper, my my Cliff's bubble logic around that time, and all of these, and the really great paper by Rob Arnott and Peter Bernstein emphasizing that these forward-looking estimates are the better way to assess future equity returns. And again, basically, those forward-looking estimates were right in the 2000s. So equities had a horrible decade, and anybody who was looking just at the historical average had had been wrong. And so I would say at that point, these capital market assumptions, this both academic but especially institutional ways of thinking about expected returns shifted from looking at the long-run average return into a forward-looking approach. And then there are just nuances then that with the historical averages, you could have done an adjustment for valuation changes. And likewise with this forward-looking approach, it can be just thinking about equities, you know, like we'll later come to the yield plus growth and valuation change. So just yield plus growth, that already gives you low expected return when valuations are high. But if you add the last term, expected change in valuation, and you assume mean-reverting valuations, you are going to be even more bearish at times of 2000, and then it was right to be, or in the 2010s when it was not right to be bearish because of the high valuations. It turned out that it was a it the markets could go still richer from that point. Anyway, so I would say we have been in this world of most investors using this forward-looking estimates, at least on the institutional side, for the last 20 plus years. But there's maybe some doubt about it because they just got so wrong the US equity markets in the last decade. So I'm going to just go through the list and talk about what you wrote in the paper, the estimates that you would be using today if you use these four different models. And for people who are interested in these models or how to create these models, we're on part four of the study. So the constant return model, going all the way back, Ibbotson studied back to 1926, and just using the average return would be about 7% over the inflation rate. That's what you would use as an expected return regardless of what the value of the market was.
>> If you're going to make an adjustment for valuation using Shiller's CAPE ratio, just using that. So a constant but unadjusted to bring CAPE towards average, you're down around 6% real return on US equities. If you approach objective returns from a different angle, where you're bringing in fundamental data such as earnings yield and earnings growth, you're down to about 3%.
>> Yes, but that's a constant. If you're adding the fourth one, which is mean reversion of the valuation on top of that, you're looking at a real expected return of US equities of about 1.6%. Which may be 1.5% because you wrote this particular paper a few months ago.
>> So that's the difference. Even with the objective return models, there's there's a range in these expected returns. As you put more layers and get more sophisticated with this, and at least right now, as you get more sophisticated, the expected return comes down.
>> Yeah. Now, contrast that to subjective returns, which are going out and surveying investors and asking them what they think the stock market is going to do over the next 12 months, and you get a completely different picture.
>> I would say out of those four, the first one, like you said, was just, let's look at historical average. And that historical experience includes the fact that US equities went from very cheap levels to very expensive levels. And the second point is to say that, hey, maybe we shouldn't think that we should extrapolate that now that we've seen CAPE ratio go from 15 to 40, we really should assume that it's going to go to 50 or 60 in the future. So taking just that away gets you from that 1 to 2, from 7% to 6%. But that is still taking historical averages, not recognizing there that those historical returns were delivered from levels when starting yields were much higher. If we start from today's earning yields, then you should expect something much less. And I so I do think that something in the 3% range is the base case. And then if I want to be really bearish, I say that well, actually, those valuations with with CAPE 40, they are so high that we should expect that 40 to go to 30. And then then you get something like 1.5% real expected return. And there I'm a little humble saying, let's let's not assume that mean reversion, even though historical data shows some mean reversion, but there also have been structural changes such as this general re-rating of CAPE ratios. Let's just take the current CAPE and reasonable earning yield from here, and that gives us something like the 3% anchor. So that's where I sort of come from. And again, most institutional investors are in this considering between these these last two approaches. So there are some studies which just recently have come out which show that these capital market assumptions by institutions or consultants and so on, they are in line with the last two approaches, and there is some variation between them because some assume mean reversion and others don't. But it's really, really those two. People haven't yet gone back to those historical averages. Whereas I think if we look at equity analysts and we look at retail investors, they are more looking at the rearview mirror, and they might be even worse than looking at the last 100 years of data. They might be looking at just the last 10 years of data, and then they get really excited and I would argue misled after a very bullish decade.
>> You did the correlation analysis between subjective return expectations of the stock market and objective return expectations.
>> Yes. And you found that subjective return expectations, meaning this idea that taking surveys and people extrapolating from the past into the future, has basically no correlation with what actually happens in the stock market, or negative. If anything, it's a negative correlation, which is saying that don't do or don't use those things. And if you find yourself being this kind of rearview mirror thinker, you should know that even though it worked in the last decade, historically it's it's more often led you wrong than right. Whereas those yields, sometimes they fail, like they did now, but on average, they are positively correlated, like 0.5 correlated or whatever. Yeah. So that's my humbly positive story for this contrarian thinking. Yeah.
>> The correlation with your estimates is quite high, like 0.5, if I'm not mistaken, over time, is that correct?
>> Yeah, so there's some tendency to predict future returns in the right direction with occasional misses. But the baseline is not that you are going to get it wrong, like it is with rearview mirror.
>> So if you're going to make a bet, you bet with the objective case rather than the subjective case. But it's very hard, right? Because you want to be with the crowd, right?
>> Yeah. So difficult.
>> Yeah. Yeah. And it's painful to be wrong for even for a decade. Even though I'm then saying that that decade is just one observation, but it's still painful that.
>> Yeah. But difficult to be wrong for one week if you're betting money on this stuff.
>> Yes. Yes.
>> Okay. Part five, you get into decomposing S&P 500 returns. And here is where you pull out the analysis that what has been going on with the S&P. There's been some acceleration of earnings growth which has helped to increase the value of the S&P, but mostly it's about an increase in the pricing, an increase in valuation, be the CAPE ratio.
>> Yeah. And this for Bogleheads, you know, you should feel a warm fuzzy feeling here because John Bogle was writing quite a bit about this three-part decomposition of equity yield, growth, and valuation change. And I think he was also using it primarily to explain what has been happening in the market. And there again, it's true that most of the explanatory power in the past comes from changing valuations. So growth and very little is coming from this slow-moving yield part. But when it comes to predicting, then it's that slow-moving yield part which has got most power in it. And again, bad news is here that the strong performance of markets in the last 15 years or so have come just from those places that shouldn't make you feel bullish in the future. And the starting yields are typically lower now, and that means that we should be more cautious rather than more bullish here. The thing that I was really then getting into in that part five was focusing on that growth stories. So what's a G number? What real growth we should expect?
>> In earnings. Correct.
>> Yes. In earnings per share. Indeed. Indeed. And just to be clear, part five is about understanding historical returns on S&P 500. Then six is about forward-looking estimates like CAPE ratio. And seven is about surveys. So, so there basically I'm really drilling into into S&P 500 as a case in point. And so, so thinking of the what's what's the earnings earnings per share growth? What's a fair number to assume there? I basically say 2% real is a good anchor for people to think about. And if you want to think after inflation, then it's four or five. But so, if you look at historical data, it has been rising over time. So for long histories, the number is roughly 2%. But if we look at data in the 2000s, this is 4%. Before I talk about the two versus four, let's be very clear that I am talking of real compound return. So, and compound, so real is after inflation, and inflation tends to be two and a half percent. And compound is basically it's not a simple return. So this is geometric rather than arithmetic average, and that difference can be one and a half, 2% when you look at data. So why did US real earnings per share growth double in recent decades? And there are alternative or complementary explanations. So why one is just that the share buybacks became much, much bigger. And when there's a lot of share buybacks, that means that the number of shares goes down, and so earnings per share actually goes up through the denominator effect. So that's one story, and that's fair. But then there are other stories where which tell you can explain much of this increase in earnings per share growth by the fall of corporate tax rates or interest rates. And if those, if those hadn't fallen so dramatically in recent decades, you would have gotten again another 2% rather than 4%. So I think that's interesting. And then there are the loser stories that it's been capital-friendly environment, globalization, loose monetary fiscal policy combination in US, or the tech bias. So there are lots of possible stories. We don't have enough data to disentangle which of these has been the most powerful story. But I think when when people look at the 4% recently versus 2%, I think many, many people extrapolate that recent performance. And I'm saying that pretty good arguments against extrapolating that and anchoring rather to that 2% longer number where where I am.
>> And what would you say earnings per share growth for international stocks would be? 2% or do you think it might be a little higher, a little lower?
>> I think for developed markets, probably a little lower, emerging a little higher. You know, the valuations, US has got valuation advantage partly because the higher growth expectations. And I think it's fair. And historically again, let's say European market, they have not, I don't know the sigma stories or so on, but or the European museum type of stories. So somewhat slower growth, I think is fair to assume. But then maybe a little more for emerging markets.
>> So getting into the next part, which is number six. So, the second one of this three-part portion of the series, objective expected returns. And you talk in here about the CAPE ratio, very popular with the media, easy to understand, basically 10 years, average 10 years earnings of the S&P 500 and using that, which might be a little off when you have this high earnings growth period. You might have to make an adjustment to that. And then you talk about another valuation method, which is a dividend discount model, which is similar to the Bogle model, if you will, the dividend yield, earnings per share growth, valuation change. Can we use any of this in the short run to time markets? And I found this section to be interesting because you say no, you can't use any of this in the short run.
>> Yeah. Uh, only over decades or when the markets are at extremes can you really use this to make any kind of a shorter-term prediction.
>> Yeah, if you use it, rather use it for longer horizons and as at extremes, like you said. And, you know, I'm a pretty poor advocate for it because I'm very much a two-handed economist here, that I am telling that, yeah, it has worked historically on average and so on. But then I'm showing quite strongly then this that actually these types of structural changes, such as the CAPE ratio rising historically from 15 average in 1900s to to 30 average more recently, it's a terrible trap for any investor who is using such measures because you are somehow expecting the normalization that never happens. So, so there are good counter-examples which tell why you might not want to use it. But still, I'm saying that I think using it humbly for long horizon and maybe more for cross-country opportunities than directional opportunities, though that's debatable. I think there are some ways of using it. But but again, at longer horizons. And then I then I talk about some improvements that you could do. But but I don't think any of the improvements that I mentioned are a game changer. So, so you said certainly one can adjust the CAPE. We do something, we adjust for the recent earnings growth to make it sort of more timely than the 10-year number. Or there's a Jeremy Siegel story that you really shouldn't use those reported earnings because they have become too conservative in this new world where there's a lot of intangibles which those reported earnings don't capture. So all of those are plausible things. But but I don't see any of them really changing the fact that these things tend to work in the long run, and then they can really disappoint for 10 years or 15 years, especially if there are structural changes. I keep trying, and everybody keeps trying, but it's so damn difficult.
>> Part seven. Now we get into the other side of this, which is subjective returns. And I love to quote Benjamin Graham: "In the short run, the market is a voting machine, but in the long run, it's a weighing machine." Seems to fit really well into your paper. I mean, you can almost put it as a subtitle.
>> Yeah. Yeah. Yeah.
>> What was interesting in this is separating out the equity investors between your individual investors and your institutional investors. And we find that the institutional investors tend to have more objective views, like yourself, on valuation, whereas the individual investors, with the exception of Bogleheads, if you will, Vanguard investors,
>> Yes.
>> have more subjective views. You know, they extrapolate into the future, you know, what just happened in the past. And then you go on to say, no, it depends on who is in the market at the time, which could affect valuation. So it would lead one to believe that right now, with at least up until a few days ago, with the market hitting all-time highs, that there were a lot of individual investors out there that were just putting a lot of money into the stock market because they were the ones who were the most had the highest subjective views of what might. And that was driven by the equity analysts themselves, who also, unlike institutional investors, the analysts or the sell-side people are always increasing of earnings growth and so forth. So please.
>> Yeah, I'll bring it back to that rearview mirror and sort of this idea that
People over-extrapolate, and we can think of it as over-extrapolating past growth or return. Let's just focus on growth rather than return. So, again, it's really difficult to forecast growth, like company-specific growth, more than a couple of years ahead, two, three years ahead. And the same is true for the overall economy. Maybe there is some predictability for the next couple of years, but people like to think that there is a longer crystal ball, especially when it comes to company-specific things.
Or now we can think of the Mag Seven source value strategies, whether it's market timing or single-stock selection or sector or country. It sort of works because people over-extrapolate too much. They think that there is persistent, more persistent growth than there is. They think that we can predict 10 years out because whatever was growing fast recently is going to continue to do that. And there are, of course, prominent examples, but there are many, many counter-examples. And on average, it hasn't worked, but sometimes it does work.
And then there is a nice quote by Larry Summers that he says that the real danger for the markets happens when these over-extrapolative investors get it right for a few years, and then they become overconfident. And they assume even longer growth persistence, and then you get even higher valuations. And then you start to get bubbles, either in the overall market or for certain sectors, or let's say value versus growth of single stocks. And we've seen plenty of that happening now.
In the 20th century, then, but you were asked who is in the market. So that's another thing which is interesting. There was an old paper by William Sharpe, and their idea was that the majority in the market tends to evolve with whatever was working in the last few years. So, the rearview mirror sort of guides us there. And then they say that the minority probably benefits then from that point. So, I think they have got this contrarian leaning spirit.
But to understand how markets are evolving, there is always this balance where we go a little too far one way, and then there's going to be some reward for disagreeing with that. And I think today we might be in one of those situations when it really is right to be not joining the crowd.
Your next two papers were on understanding treasury yields and understanding the yield curve. Yield curve expectations, which are further out you go, you get a higher rate. And here in the first part, you decomposed the return of 10-year treasuries, and you said that the 10-year treasury is sometimes called the most important asset in the world for what takes up its yield. I believe, and because it anchors so many other things, it's sort of the riskless asset underneath so many other assets, that's why.
So you wrote that the composition of the 10-year is number one, inflation expectation. I mean, what are people expecting from inflation? And second, a real yield, which is a return that we get over inflation. And then third, the term premium, which is you get an extra little bit by putting your money into longer-term assets, only because you're locking your money up.
Long term. So there's a little bit of like an illiquidity premium there for putting your money away. And that you should get paid something on. And you decomposed the Treasury market and looked at these three things, and you came up with some interesting analysis that, oh, I would have never guessed, having to do with inflation as part of that.
Yeah. So first, first is that that decomposition. It's really there. The survey data is beautiful because you only need to have the 10-year yield itself, and then you need survey data. Let's say I'm using economist surveys because it's available. Economist survey of next 10-year average inflation and next 10-year average cash rate or T-bill rate. Once you have got that, you have got a three-part decomposition. Those three pieces fall out, and then you can analyze them.
And when you look at them, you see that in the last 40 years, all went from quite high levels to very low levels. Especially the expected real short rate went to negative levels, as did the term premium. Inflation expectations came down from something like six, 7% to 2% or so, but no lower than that. But so all of them came down and contributed to this overall decline in yields. And we had pretty much a low point in all of them somewhere in 2021, after the COVID episode. And since '22, we have seen big increases in bond yields, but they came only from the latter two, real yield and term premium. They have both gone from negative to positive.
This was the fascinating part. I would have guessed that it was due to inflation expectation. Inflation went up, so therefore the yield on the 10-year went up. But your research shows us is no, that's not what happened.
Exactly. Yeah. Maybe those two other things, they really went, in some way, out of whack or so on. In the QE era, they went to such low levels that they needed normalization. And that normalization certainly has happened. But it is also weird that we didn't see more happening to longer-run inflation expectations in this time. I think people probably remember that for a long time, yield, sorry, inflation rate, CPI inflation was near 2%. And then in '22, we got this sudden rise to 9%. And then we came down gradually, sort of painfully gradually, maybe, but we got down to near 3%, and that's where we are nowadays.
But so, while this 2 to 9 to 3 happened, long-run inflation expectations went from 2-plus to 3-minus. They never exceeded 3% when you look at economist or market forecasts. So, somehow something was anchoring them. And that something was, I think, the Fed's credibility. So, somehow people just trusted that when inflation goes up too much, the Fed will come and sort of save the day and not let it go too far. And so, we really can give some credit to this Fed credibility and maybe Fed independence for these stable, well-anchored inflation expectations.
There is a risk when you are playing with Fed independence and credibility that next time there is any even mild inflation problem coming, maybe from tariffs or maybe from fiscal deficits or whatever hits us, we don't get that anchoring back effect because people don't trust that the Fed will come to help anymore. And in that situation, I think we are at bigger risk of getting de-anchored inflation expectations.
Part nine is the yield curve. Expectations again. So, how much higher would rates should rates be in the future than they are now? So, what does the yield curve look like? And historically, the 1 to 10-year increase premia is about 1.4. That's the historic number. Not quite there yet right now, but sort of heading that direction. So, you looked at how do you form expectations about yield curves. So, this is related to what I was telling of how bond investors seem to be contrarian, but just more directly.
So, yield curve steepness, if the yield curve is very steep, it either tells that investors are expecting rising rates, or it's telling that there's a particularly high term premium in the market. So, it must be one of those two things. And I think it is more the expectations part, based on historical analysis, that I show that really moves yield curves from upward sloping to inverted. And so, when we shifted in '22, '23 from upward sloping curves to inverted yield curves, it just happened when interest rates were rising.
So, I think the idea that there is mean-reverting rate expectations is pretty clear. But looking at this recent example where we saw markets reacting to Fed rate cuts in a way that short rates fell but long rates rose, that's a pretty extreme thing. Normally, we see when there is some soft news and monetary policy easing, we see short rates falling, but long rates falling a little less. So, you get a little bit of steepening. Right. But recently, we got some examples where you got even the long rates rising at the same time. And that's sort of abnormally strong reaction. And that could be related to fiscal concerns or it could be related to Fed independence, Fed credibility concerns in the market.
Part 10, you deviate and you go to alternatives. And here we're talking about real estate, which we've already discussed a little bit, factor investing, private equity, private debt, and you know, what are the expectations for these? Unfortunately, we're running low on time, but for those who are interested, you can download the paper on AQR's website. I think one big distinction is that people have really added a lot of allocations to those illiquid alternatives, partly because they perceive that it has sort of better rearview mirror performance, which is not actually so clear when the data is carefully studied. But that certainly is a perception that when you can quote things in IRS and you don't get any mark-to-market, and whatever, and so you can tell pretty wonderful stories of past performance, and nobody's checking it so carefully.
I do think that when it comes to alternatives being diversifiers, there is an interesting thing that liquid alternatives are much better as diversifiers. They really can be uncorrelated or even negatively correlated with the market. Whereas illiquid alternatives, private equities, and so on, they give diversification pretty much only through the fact that they have got smooth or stale pricing. And so, but they have lots of what I call slow beta. They really move closely over three-year windows with the market. Not much diversification.
In part 10, you wrote something in your paper which I was so happy to see because I've been saying this over and over again, and it has to do with portfolio complexity. When individual investors start getting into things that maybe they don't really understand, there is a higher probability that they will throw in the towel on it when it's not meeting their expectations, and that lowers return and increases risk.
Yeah, I said that that I think it's fair, but of course, I take a more positive spin on that. That if you are among those who can study these things, take the long horizon, and so on, basically improving your diversification and getting your rewards from multiple sources is really valuable. And another somewhat related idea is almost everybody's portfolio is dominated by equity risk. So, if there are some strategies that can help when equities have a few bad years, those are really valuable strategies to complement the portfolio, and you should really try to stick with them.
But again, it is hard for investors to stick with them, and that maybe sustains the opportunities for the future for the minority who can do it. But I think anybody should look in the mirror before you take that path. If you don't think that you can stick with it, just don't even try, because then you are going to leave it just at the wrong time.
And that concludes all 10 parts. Fascinating study. Anybody can download this at AQR and read the details. Auntie, thank you so much for being our guest here on Bogleheads on Investing.
Thanks, Rick. This was a great discussion. Thanks again.
This concludes this episode of Bogleheads on Investing. Join us each month as we interview a new guest on a new topic. In the meantime, visit booglecenter.net, bogleheads.org, the Bogleheads Wiki, Bogleheads Twitter, the Bogleheads YouTube channel, Bogleheads Facebook, Bogleheads Reddit. Join one of your local Bogleheads chapters and get others to join. Thanks for listening.