To carry on the trend of amazing guests on the show, today we welcome Antti Ilmanen. Antti is the co-head of the Portfolio Solutions Group at AQR, and the author of a couple of really impactful books that we highly recommend. His most recent is Investing Amid Low Expected Returns, and in today’s chat, we get to hear all about the ideas contained in its pages, and the most important moments in his career that have shaped his philosophy. Antti received his PhD in Finance from the University of Chicago and has had a long career of working with high profile names that you might recognize from past episodes of this podcast. Our guest talks about the need for cultivating your patience in relation to your investment portfolio and shares many other great pieces of practical advice that you can implement today. We delve into some bad investment habits that Antti noted in his book and also touch on the usefulness of bonds, illiquid assets, trend following, and much more. Be sure to tune in today to hear all that this great mind has to offer.
Key Points From This Episode:
All right. So to kick it off with the first question, Antti, does long-term historical data contain information about expected returns?
Yeah, it does. If expected returns or some premia, if they’re constant over time, then long-run historical averages are our best estimate. But if they vary over time, this historical averages are less useful. And then it’s often better to study yield-based for wide looking estimates.
And under what circumstances can historical data be misleading?
So, even if these expected returns are constant, the sample average is a noisy estimate, and it’s moreover… It’s distorted if you’ve got big valuation changes within the sample. Fortunately, it is possible to adjust for this sample specific valuation changes. And let me make this concrete. If you think of last 10 years, S&P 500 realized 14% real return, while 10 years ago, the expected real return based on starting yields, was about four or 5%.
Well, we know that the Shiller price earnings ratio, or CAPE, almost doubled from 20 to 40, and the resulting windfall gains were almost 7% per year. So we could subtract that 7% from the 14% and we would sort of start to get to the right ballpark. But anyway, that’s sort of a bad example, because one 10-year history is just one observation really. So we need much longer histories to cancel sample specific noise.
So, you mentioned something important there, though. The earnings yield at the beginning of the period was very different from the realized return at the end of the period. Historically, in the full sample, how predictive are current yields of future realized returns?
Well the good news is that they get the sign right, so it’s better to look at that than the 10-year return and extrapolate from that. So low equity yields predict low future returns. The bad news is that this forecast is quite noisy for the next year. This is not very useful for market timing but even for a 10-year horizon, there’s a very large uncertainty band of several percentage points per annum on this point estimate that we all quote.
Actually, Cliff Asness, my boss, I and Tom Maloney, we wrote an article a few years ago where we showed how difficult it is to use equity yields for market timing. And the theme was, if market timing is a scene, then at most seen a little. And the intuition there was that, it’s very common to show some charts where there’s a nice relation between starting yields and next 10 years future performance. But the caveats now are that, there are very few independent observations. The statistical relationship is very weak. And then if you try any contrarian market timing rules using these signals, they have pretty underwhelming historical performance.
So, in practice, I think it’s good to think of this cliche, that if you are trading with contrarian signals, you too often tend to be early and early EQs are wrong. That just has happened too many times for good investors. Anyway, now I was focusing on the cautions. I would still say that these yield-based estimates are pretty much the best we have for next 10 years expected returns. So they are useful anchors and therefore today’s low expected returns are a great problem, especially for young retirement savers.
Interesting. So to follow on to Ben’s question, when you estimate expected returns, how much weight do you give to the historical data versus forward looking market implied expected returns?
So, no easy answer. The more evident it is that there is time bearing expected return in a way that you can measure, the more I lean on forward-looking data. That almost sounds like circular, but certainly for bond yields, you can see this. But I’d say, you can see this for equities and other long-lived assets, real estate, and perhaps for value stock selection strategy.
For some other factors, I think this would not be too helpful. And I’d be especially wary with using historical average returns in a situation where we know that current starting yields are far away from historical average yields. Then you sort of know that there’s something off in using that historical evidence.
Maybe I’ll ask the question this way. What we do for financial planning purposes is we combine the historical premium with the current market implied expected return, and weight them based on the historical predictive power of the CAPE. Does that make sense?
It does, with the only caveat that we have got very few observations, even of CAPE. Like if we talk… If you have 150 or so data, which sounds like you really trust that the world was stable in that time, that still would give you 15 independent observations, lots of noise. But still, again, we are in an imperfect world. Sounds good.
So, based on the current market implied, or the earnings yield, how low are expected returns?
So, I’ll focus on U.S. equities and ignore some sort of previously developed ways of doing this. But if we take now this Shiller price earnings ratio as a good way of, not just thinking of valuations, but flipping it around and thinking of that as an expected real return, then… I think the yesterday’s Shiller P/E I just checked, it was 33-ish, I think. Which conveniently, sort of says that flipping it around 100 over 33 gives 3%. That would be a simple answer.
By the way, then the important thing is that when you look at last decades, 40 years ago, it was 10%. And we have seen the decline to 3%. We’ve gotten these beautiful windfall gains for equities, and I call it, we have effectively borrowed returns from the future. We have capitalized the future of cash flows now at very expensive levels, and that’s sort of good for current asset holders, but it really, effectively, it locks in those low future returns. And locking is advisedly used because, I just told that we predicted 4%, then we saw 14%. So the forecast error is sometimes big, but it’s still… That’s the right base case.
And so, I think we have got these low expected returns that will materialize sometime, but we don’t know whether that’s going to come through, what I call slow pain or fast pain. And we are now having this, clearly a fast pain scenario. But the alternative would be that we would always have this very low starting yields and we would be so doomed to clip tiny coupons without any benefit of the windfall gains that we were used to in recent decades.
Are non U.S. equity markets comparatively cheaper?
They are. U.S. stands out, based on CAPE or any other metric, I think reach compared to other markets, both developed and emerging. Some of these reflect different sector decompositions. Maybe it doesn’t matter, but it’s sort of an interesting consideration. And partly it’s explained by that, but partly, it goes beyond that. So, for example, the larger tech sector and rich tech sector in the U.S.
You talked about market timing earlier, referencing your paper, which I love on that topic. It’s a fantastic paper. Does it make sense to tactically shift into higher expected return equity markets, like international markets as opposed to U.S. equity?
Yeah, it can. First maybe just clarify that we talked about market timing a little, and now what you are describing then is contrarian country allocation strategy. And it’s important. Sometimes we hear that people are not quite clear of the differences even between market timing and then value strategy cross section. So let’s just think that contrarian strategy can be done at market timing level on S&P 500 or global equities. Then we can do country allocation, what we are talking about now, or then we can go to stock picking and do value versus growth type of strategy using HML.
So, back to that question. I think in the long run, this kind of country allocation strategy has been moderately profitable, but it doesn’t have a great hit rate. And if you look at, again, last decade, U.S. has been looking expensive, and somewhat getting more expensive most of the time. So this is another example of the difficulty of using contrarian signals, much patience or lack is needed with them. One day it will work and so odds are in your favor, but the odds are not greatly different from 50/50.
So let’s shift to fixed income and a question that Ben and I get often. So with bond yields being lower negative, are bonds dead as an investment?
Well, I learned in my Fama class 32 years ago, that bonds can offer an interesting term premium as long as yield curves are upward sloping. And that they can be an interesting sort of negative beta asset, even with no premium. And you might want them if they are good recession hedges. So those lessons, they’ve been later refined, but the broad message I think still applies. Bonds could have a return generating role and they can have a very valuable diversification role, and they certainly had that in recent decades.
Now in the current situation, we’ve got heightened inflation situation, fair tightening, and we are starting from still, somewhat to a no longer wrestle with yields as a while ago, but still that yield levels’ where people see lots of upside potential. But such views, then that’s getting to duration timing. By the way, I did write my dissertation pretty much on duration timing in Chicago. And it’s funny, sort of every decade, I’m getting less confident about mine or anybody’s ability to do that well.
But on that, is it dead? On that mortality jargon, I’m just thinking, duration timing trades, where initially in Japan and later elsewhere, they were often called widow makers, because people were betting on the rising yields, and that never happened. And so, this year it is happening, and so great finally. And okay, eventually those rising yields will revive the corpse, and then we should be having more interest. Everybody should have more interest in bonds by that time.
That’s pretty funny. What people think is making bonds dead is actually reviving the corpse.
Indeed.
Interesting.
You mentioned term and credit. Is there an expected credit premium? Is there an expected premium for investing in corporate bonds over government bonds?
There is, on historical data. You can go somewhat further back, like even 1920s or further. But, I’d say you have got decent quality data on corporate since ’73, maybe. And really from late eighties. But over these periods… And certainly before, but I just don’t trust that data. There was a positive reward. And you can think that it’s a… So it’s partly a default premium, illiquidity premium, and there are often some tax rationales. And the further thing is, that it is quite highly correlated with equity premium.
Interesting. Okay. Because I’ve seen a few papers that kind of suggest if you duration match and option adjust and all that stuff, then the historical credit premium goes away. But you’re saying in the more recent data, it is there.
Yeah. So, my colleagues had a paper a few years ago that, which clearly… I mean, they made the right type of things, looked at regression. Whether when you put equities and treasuries and then corporate excess return there, whether it shows up as useful thing. Or another idea, you can put this into optimizer and ask what an optimizer says. So they found good role for credits.
And you can intuitively think that maybe they have got somehow the illiquidity premium, maybe different volatility exposure than equities that could justify it. But they did this analysis with data going back to, I don’t know, 1930s or so. And so, I’m a little cautious of that. But, I am thinking that there’s some role for credits beyond the base and equities. This is different actually than what I wrote in 2009, but I saw new research that made me change my mind.
Interesting. So something that I learned from your most recent book, which actually surprised me quite a bit is that there’s a long run commodity premium. And the reason that’s surprising is that, theoretically you’d expect low returns. Similar to what you’re talking about with bonds, they’ve got lower negative beta, therefore low expected returns. They’ve also had low realized returns. And then, the other thing I learned in your book is that individual commodities have had a 0% realized return on average, but commodities as a basket have had a positive premium, where does that premium come from?
Okay. Yeah, this does get very geeky. It’s sort of a lovely topic for niche audience. So, I hope you don’t lose too many people-
No, don’t worry. This is a very geeky audience. Don’t worry.
Okay. So, to understand it, you must be aware of the difference between arithmetic and geometric means. So, simple and compound returns. There’s always basically a positive gap, and that gap widens with volatility, it’s often called volatility drag or variance drain or something like this. Basically assets with high volatility will have a big gap. And the intuition is that large fluctuations eat up compound returns.
So sort of, if you start with a hundred dollars and you go up 50% and down 50%, you are not back at a hundred. You are down to 75, and this effect gets bigger with volatility. Okay. So, now in the case of commodities, single commodities have quite high volatility, typically near 30%. So their geometric mean is well below their arithmetic mean, and the geometric mean in our long data set is typically near zero.
So, you can actually think, if I can visualize it. So typical, some commodities have had a long run history compared to cash that they outperformed. Others had negative, but on average it was something like this. So that’s the case for individual commodities. But when you create a portfolio of them, you get this. And so, you get… Sorry, and for listeners, you get basically something upwards sloping. So there’s something like 3% commodity premium coming from the fact that you are reducing portfolio volatility from near 30% to below 20%, and you are reducing this volatility drag. And this weird technical feature, gives commodity portfolios this extra return.
And this mechanism is actually known for quite a long time. There was a paper by Fama and Booth talking about diversification return, others call it rebalancing bonus. And it is available in every asset class. Its just it’s much bigger in commodities with high volatility and local relation combination, which makes it so much bigger. And most memorably… So Cam Harvey dubbed this effect, turning water into wine.
And so, the bad news is that such a magic is sort of, I don’t know, it’s maybe hard to convince investors that this is a sustainable premium, the same way you would think of equity premium. But I do think as logics… So you mentioned that it’s a low, but it has got some market beta and growth beta. So those, I think are one argument. In addition, commodities have got lots of idiosyncratic risk and you can make a case for some hedging demands that would justify. Those would be sort of the rationals why this happens.
So what effect do illiquid assets like private equity or direct real estate investments have on expected returns?
So, they are obviously super popular assets nowadays. Investors are sort of flocking into that area. And one of my key arguments is that illiquid private assets may not earn as big premia as is commonly thought. One intuition is that, for any amount of fair illiquidity premium that you should get for locking money away for 10 years, that may be offset partly, fully or maybe more. We just don’t know. But surely will be offset partly by another interesting feature of privates, what I call the smoothing service, lack of mark to market.
So we will accept not the full liquidity premium, but something much more modest. But the fact that there’s something very nicely smooth. Like I highlighted, that if you compare Cambridge buyout index with a public proxy, small cap value stocks during the GFC, you find it 100 draw down of 25%. Another one had draw down of 60%, which one would you prefer?
And then I’m saying that, then you should expect some pricing implications for that. And so, I think investors are very… I don’t know, they are always nodding when I tell this story, they sort of think of themselves in the mirror, and this is the way I view as well. So, that’s the sort of logical thing. And then when we go to data and ask how big illiquidity premia do we see there? We see actually very modest premia when we get to private assets. The longest history that we… Which is, I would argue this, is since 1978 we have got U.S. direct real estate compared to listed reads. And between those there’s been a negative premium. So, the more liquid, as it has outperformed. But there are better differences, leverage differences, sector differences. When you adjust for those, you may get closer to zero, but you never find a positive illiquidity premium.
And something similar, not as extreme, but with private equity, there is, I think, evidence that, especially in the last 15 years, there has been quite a modest premium between private and public equity. Again, I would link that then partly to the inflows into the space, but also just to this idea that investors really appreciate this other feature of private assets.
Yeah, I love the term smoothing services, that’s good. I haven’t heard that from you before.
Yeah. By the way, I just thought maybe for your mindset. The other interesting point with this inflows is, that the popularity of those assets’ means that we don’t get the fee pressure in private equity that we’ve had before with hedge funds. We are in a world where, with lower expected returns, but that’s a space where the fee pressure, I think, will come only after we… I don’t know, investors eventually learn that the return advantage is not as great as you see. You learn more slowly in such assets as private equity.
Let me make sure I understood what you’re saying. You’re saying that expected returns may be coming down because of inflows, and so valuations going up, but the fee pressure will come later once realized returns are low?
Exactly. Because in reality, those private equity manager who faces lots of institutional investors and maybe democratized participants, as well on the door, can say that, oh, you don’t like my high fees. I bet many other people here. So again, you have to have a different competitive balance between them than we see now.
That’s very interesting. I think… You guys had a paper on illiquid assets, and this is where you talked about this. And I think there was a relationship between valuation and realized returns as well, right?
Yeah, where we talk about… We have with more data when we talk of Shiller PE and S&P 500. Now we are talking of many fewer decades and very smooth data, you get, I don’t know, three observations were very useful to do any statistics on those.
Okay. To continue on real estate. We are in Canada, which is, as you may know, one of the hottest real estate markets in the world right now. Are the real estate returns of recent decades indicative of real estate’s expected returns?
No, they are not. Remember that what I was talking about returning valuations boosting realized returns of bond and equity returns in recent decades. So something similar has been happening in real estate globally. And I’m pretty sure, especially then in Canada, if you’ve seen this particular business. And I’ve seen in the economist, some of these signals there.
So instead of looking at the rear view mirror and extrapolating on past returns, which were boosted by windfall gains, here too, I would rather try to look at forward looking returns, some yield measures. And it’s hard, but they should be available somewhere. So, in general, I don’t know what the terminology there in Canada is, but I’ll call it rent to price ratio. So I would use rent to price ratio as current starting yield. Are they similar to longer and average? If they are not, and I bet they are lower, then don’t expect those historical average returns, which were anyway boosted by those windfall gains.
So, I am not an expert on this, but my best reading of the literature and thinking of what’s happening in other asset classes, is that a decent forecast for future real return for real estate is the rent to price ratio, which should be adjusted for the costs or expenses, which are often a third of the gross yield. Or, if you have got a real free cash flow yield, that would do the trick. If that’s the best estimate, this means that we assume an expected returns for equities also. You can think, yield and growth and expected change in valuations. Now I’m saying the same. I’m saying, let’s just look at the yield and ignore growth, and ignore expected change in valuations.
So, the book writes about literature, talking about historical real growth of rents, and this is mainly U.S. Real growth of rents or quality adjusted house prices. Some studies find small positive, sometimes some find small negative. I think, zero is a pretty decent base case estimate there. So, expect just zero growth. You just earn your yield, and then assume any changes in valuations. So we don’t extrapolate ever more rich earning, but we also don’t expect mere reverting valuations. Which I, if anything, I would say is more plausible now that we are in this rising yield environment, rising bond yield environment.
Right. So you mentioned investors are flocking to illiquid assets. Can you talk about how institutional investors might be responding to lower expected returns, or have been responding?
Yeah, it varies, but the most common approach is to take more risk. Investors want to keep earning the kind of returns they are used to. Instead of like, I use a serenity prayer in my intro. So, they don’t surreally accept lower prospective returns and tighten their spending plans. So this could be done through higher equity, higher equity rates, which Norwegians are very well funded. Or some allocations to style, or factor premium, or the most common choice, raising the allocations to alternative asset classes such as private equity, private credit, and real estate.
Does this cause you any reaction, any concern?
Well, I already told, I think there’s going to be an impact that you should expect lower returns in those places as well. And maybe the one concern is that, there’s often, I don’t know, apparent disconnect there that people don’t see these types of logics as well in those markets. But it is a world of where everything is expensive. So I think you can make fairly this choice that you take more risk as long as you do it with open eyes. And then you can debate which of these is the best one.
The smoothing feature that I said, I think there’s something you can criticize about it, but it does help investors somehow be more patient. And I have said sometime that thinking of Canada pension plan, that 85% equity risk. If that, even for them work, which are clearly would longer rise on investor, if all of that was marked to market, I don’t think the number would be 85%.
Yeah. That’s probably true. Now in equilibrium, taking more risk can’t be the answer, right? Because not everybody can take more risk. Is that the right way to think about that?
Yeah, well, it depends on of course what sort of issuers would do. In principle, it might be possible that there would be catering of that demand, but I think it’s fair. I think it’s typically, it’s a good way to think that you can’t have crowding at the aggregate level, valuations have to adjust. So actually, I’ll just tell this, that if you think of situation like October 87, or it could be this week, or it could be March 2020. If everybody becomes more risk adverse, we can’t all run to the exit at the same time, asset prices need to fall. And the flip side is that when there’s growing demand for something, valuations have to adjust all the other possibilities that the issuers come to the other side and refer you something.
How do you think the low expected return environment we’re in now will impact the required savings rate for individual investors?
Yes. So, we wrote a paper a few years ago where we tried to make tangible this impact of low expected returns for individual retirement savers. In that case, it was for U.S. DC savers, but the point I think generalizes. So your eventual pension pot depends on the savings contributions you make, and or somebody makes, might be employer. But savings, contributions, and on the investment returns on those savings. And if the market does less for you, you have to save more.
So in this paper we plugged in plausible assumptions. And we came up with estimates, that in the good old days, with sort of normal, longer and expected returns, it was enough to save, let’s say 8% from your paycheck. But if expected returns sustainably fall by 2%, the required savings rate to reach the same target doubles to nearly… Or almost doubles. So I think 15% was the number, eight to 15%. And again, details will, you can plug in a bit different numbers. But it’s really a first order impact on required savings rate.
So if we set theory and long term evidence aside, value’s been going through a pretty rough patch of late. So has value ever gone through a period this bad before?
Yeah. I do have that in the book as a small detail. The data before… You have to go before crisp data, if we are talking over the sort of Fama-French book to price, HML. So it didn’t happen since 1926, some big ones, but not quite this big one. But earlier studies show a pretty comparable value draw down between 1890, I think, and 1904. That was the time when there were big monopolies, standard oil and others in the U.S. And this period ended roughly with the pending Anti-trust legislation. So it is interesting to, I don’t know, consider the parallels now, which could be headwinds for the magnetite companies.
But I do want to stress that… So I think any version of value has had a huge draw down in recent years. But what press often refers to them, this 14, 15 years draw down. That’s pretty specific to looking at sort of the classic book to price Fama-French, without any industry adjustments. If you try to do more apples to apples within sector, within industry, type of value trading. Or in multiple signals, look at outside U.S., you often did okay, still in 2010s up to about 2017. And after that, that is very bad performance. But again, five years, as opposed to 15 years draw down seems more appropriate.
I think it’s unknowable, but I want to ask anyway. Do we know what caused value to perform so poorly in the last few years?
Yeah, we’ve studied so many different culprits. We have thought of some, and others have been offered to us, and that list is actually long. And I’d say many proposed courses are weak. One famous one is that it’s a low yield or polling yields. And so, there’s a little bit of empirical route for that, but it really, both logically and empirically that can’t explain much.
I think the best single explanation I’d pick then would be sort of loosely this structural change that we have seen from physical to digital, which has favored many young disruptive companies. And it probably started with some good fundamental roots 10 years ago or whenever. But by late 2010s, it had degenerated into an irrational but bubble… And I hope Fama isn’t listening to this one.
And so markets repriced, growth stocks, higher… Not just in tech, but in many industries, you could ask maybe that could be justified by fundamentals, better fun… So, typically growth stocks have better growth fundamentals and value stocks, but market discounts too much. But even that argument that they were exceptionally good environment now for growth stocks, doesn’t hold water. So we’ve tried it in so many different ways. And typically we find that it’s sort… With anything measurable, it was the other way. It really wasn’t possible to justify this repricing. That’s why we would call it a bubble. And of course we have seen recently some deflating of this bubble, but forward looking value spreads suggest that there’s more runway to go with that.
The post where you guys compared value and growth fundamentals over time and showed that wasn’t the explanation for the… That was very interesting.
Yeah, and there are many ways of doing that. Again, it’s every way we try… It is something, I don’t know. At some point then you become more confident. You rule out everything else and then you get to, okay, looks like a bubble.
On rates, you guys had a paper where you found that statistically the relationship just isn’t there. And in other samples, it just isn’t there. And so I kind of took that and said, okay, that’s fine. The relationship just kind of looks like it’s there, if you look at a pretty picture. But rates have been rising recently, and growth has been getting hammered, and value’s been doing okay. It’s interesting.
Yeah, so let me say that. So, it is also… It’s fair to say that the relationship between various rate measures… You can again, look at many different measures, but that has increased the correlation used to be close to zero. When now it might be 0.3 or something like that, but. So it has increased. And I do think that it has been such a media story, an investor story, that for some time markets can act as if this relation was deep and meaningful, and it’s going to be there always.
And so right now, I think that relationship is there until it somehow weakens one day, and we are not, we are going to have less. But basically, this common belief that there is this pattern, I think, does probably contribute to something. And I mean, that’s a weird way of thinking of markets, but I think that probably has been happening last couple of years.
That’s super interesting. So even if it’s not a real thing, it becomes real because people believe it is so. Wow.
Yeah.
So, is it possible to generalize when major value draw downs, like this tend to happen?
Well, you first… It’s pretty much in hindsight, but logically structural changes are bad. So, value and contrarian strategies sort of… You expect some normalization. If you get a structural change, that’s sort of bad news. And those, they’re often related to technological advances. And I think again, they can be things that start sort of rationale, and then they may degenerate into something less rational. Sort of, this time is different. And you take things too far. Like 50/50, phenomenon 50 years ago, Dot-Com One, 25 years ago, Dot-Com two recently.
And when these things happen, they create major headwinds and losing faces for value. But on the other hand, it hasn’t stopped value from working over 200 years in stock selection or in other asset classes. Temporary headwinds, although they can be again, so painfully long that really matters for investors. So I think that’s the main answer, but maybe another interesting question, or way to address your question is, is there equity market better in value strategy? And the answer is that, not a consistent one. That if you think of recent period, or you think of post dot 20 years ago, value had a negative beta. It did very poorly in bull market, and then very well in the bare market, which is a lovely, lovely feature, of course.
But then when you think of 2007, eight, the beta was actually positive. Value strategies suffered when the market was falling. And over long histories, if I look at rolling correlation or beta, so market beta value strategy, there hasn’t been a consistent sign. There has been some of this sign flipping. So, I think it’s sort of… In the long run, I would think of it as pretty much market neutral, but maybe one can have some idea where… Clearly, since the playbook is so much like it was 20 years ago, I would expect value to continue to have that negative equity market beta for a while. But in five years time, I have no idea what kind of beta it’s going to have.
That’s really interesting. So we talked to Bill Janeway a while ago, and he talked about productive bubbles in technological revolution. So it sounds like we’re kind of relating that concept to value and growth.
Yeah. And again, rational underpinnings can be there.
Is the expected value premium constant or time varying?
Well, certainly, when you look at those value spreads, they vary a lot over time. And then if value then were a constant, it would sort of require that all of that fluctuation would be justified by those exceptional growth environments for growth stocks. And that doesn’t seem to be the case. So definitely I think it is time wearing, but it is hard to exploit it. Let’s go back to the things we talked earlier, contrarian timing is hard and early EQs are wrong. Happens time and again, even to us.
So how do the expected the returns of value today look compared to history?
Yeah, very attractive. So, but achieved through blood, sweat and tears. So lots of pain. So the painful draw down of 2018 to 2020 brought this expected returns to very attractive levels. So this value spread, you think… So, it’s a relative valuation of value stocks versus growth stocks. Well, we sign it so that positive number is attractive. That has been very high since 2020. Only really competing… The only historical competition is around 1999 there.
So again, that parallel seems important. And the interesting thing is that value spread has remained quite wide, even though value strategy has done very well in the last six months. And we just keep looking at, oh, we are like… Some payback for earlier troubles that you get good performance. And it looks like the value spread remains wide.
And the intuition, how it could remain wide is that, we are rotating into new value stocks and growth stocks over time. So that’s one explanation. It also could be that the fundamentals are moving sort of in our favor. That’s another way that can happen. So, beyond just looking at the value spread, if the momentum is now in the favor of value, which it is, that can give some good tailwinds to it. And maybe the rising rates as you mentioned, could do that as well.
But anyway, last thing I just say on this topic is that any answer to these types of questions is design specific. There are so many ways of doing, developing the value strategy. I mentioned the industry neutral or not, for example, which signals you use and so on. And also many ways of measuring the attractiveness of whatever strategy you chose. But the broad story, I think I said, looks robust to different ways. So, looking at value.
In your book you talk about a few different alternative style premia. One of them is value, which we’ve been talking about. Another one I want to ask about is trend following. Maybe before I ask the question, can you quickly tell the listeners what trend following is?
Sure. So trend following chase’s past returns, is sort of the insulting way. But again, in some way, the weird thing is, it’s surprisingly naive strategy, and it’s a close cousin to momentum. You can pretty much say both favor investments that have done well in the last few months up to one year, and sort of dislike, maybe underweight then, investments that have done poorly over this time.
The difference between trend following and momentum is that trend followers allow big market directionality. Momentum trading is often like value. Let’s say, it tries to be market neutral. Whereas trend followers say that, if the whole market went up last year, I’m going to go all beta long in a big way. And in a bare market, I go beta short in a big way. So, that’s I think… And by the way, that is implemented, not just in trading equity marketing assist, but also bonding assist, currencies and commodities. That’s typical for most… There could be other things, but those are other typical for liquid asset classes.
And the other thing is, that you can have pretty similar types of… You diversify by doing this, but directionally you may also have similar things. If there’s bearish environment, you may go short equities, long duration, anti-carrying currencies, and favor gold versus growth in commodities. And all of this are sort of risk of type of strategies.
Now, I understand that one of the benefits of trend following is supposed to be that it reduces left tail risk. But I also understand that in the COVID crash, it didn’t do a whole lot to help. Do we know why that happened?
Yeah, I think I can… But let me first tell the broader, so the more positive thing is that, trend following has had this weirdly good historical track record of combining positive long run reward with good performance in many equity market draw downs. And the intuition, how they tend to do well in equity market draw downs, is that trend followers tend to turn then, or can turn book from bullish to bare-ish and ride the bare market and be among the few investments that help you then when you most need it. But this of course works especially well if you have a slow moving protracted bare markets. And now getting to your question, the Achilles heel is, in sudden turnarounds, in practice those are pretty rare and the worst historical months tend to happen within bare market and not at the turn. But sometimes you get the sudden turn. And obviously with COVID, it was quite a sudden turn and that hurt.
But even in that case, trend strategies didn’t do too badly. Some lost a little, some made a little, because, of course equity markets turned pretty much only late February. And then it matters how fast your trend signals respond to that. But some other asset classes, or bonds, currencies, commodities may have already been tilted toward risk of positions. So it wasn’t actually too bad. And during this year’s equity market turmoil trend following has had a fantastic situation, because things have sort of gone sufficiently gradually. So, which also happened in 2008, can’t always work like that. But again, with this more gradual bare markets, trend following is a beauty.
Can you define the defensive style premium and how it relates to quality?
Yeah. So general let, I will… Let’s focus on stock selection mainly. So defensive and quality, they are sort of both umbrella concepts for this empirical finding that boring stocks perform surprisingly well. And the risk is stocks disappoint in the long run. Let’s just first recognize that this is a huge puzzle in finance. We learn, among the first things in finance, risks should be rewarded. And when you think of stocks versus bond or cash, risk does get rewarded. And that makes all the sense in the world.
And then when you go within equity market and you look at portfolios with different betas, you could do also different volatilities or some different measures of quality. You find that the riskiest stocks tend to have disappointing, actually worse performance than other stocks. And instead of having this positive relation, there’s inverse relation. And if anything, in last decade, for example, it was quite a strong inverse relation. By the way, so in Fama-French five factor model, you can think that the profitability and low investment, they are sort of in the camp of quality signals. So again, statistical low risk measures are one way of thinking of this, and fundamental low risk measures like quality, including those two, are other candidates for it.
But I’ll focus just with the story. I’ll just talk it through the beta effect, because that’s sort of the… Theoretically the… I don’t know, most important. So, it was found already by Fisher Black, the blacker source, back in early 1970s. That there was too little beta, reward for beta. And Fama and French found it again in their early nineties studies. And many other researchers, including my colleagues, Andrea Frazzini and Lasse Pedersen found it again in 2000, and took this research much further. And the empirical evidence is, it is super compelling because it works in most countries. You can look at it in, find it in within most industries. You can take it into other asset classes.
Boring stuff. Earns higher risk adjusted returns, often even higher raw returns. And then you can ask why. And I don’t go into details, but I just say, investors leverage aversion and lottery preferences are the main candidates. Those speculative stocks contain embedded leverage, which investors tend to like. And they are probably a lottery tickets in financial markets, investors tend to like them. Those sort of make them… It’s more expensive and imply then, lower, long run required returns.
Yeah, that’s very interesting. All right. I want to move to carry. It’s a tough… I have trouble thinking about carry, so I’m hoping you can describe it. And how is carry different from value? Because I see carry and it looks a whole lot like value to me.
Yeah. So carry is… We were talking how naive trend is, but carry is something, another embarrassingly naive type of strategy, favoring high yielders versus low yielders. And historically that has done surprisingly well in many different asset classes, just like value momentum and defensive strategies we described. And so to your question, in some cases, carry and value are very similar. I think you are mainly… I suspect you are thinking mainly of stocks, that if you think of dividend yield as a value signal, well, it is also, it’s carry signal, but it’s also value signal. And not the best, but it’s goes there with BP, EP, DP. So that’s fine.
But in some other cases they are quite different. So if you think of currency trading, high short rates are what we look in carry, that gives positive carry. And that’s not at all… At least it’s not mechanically related to undervalued currencies. Sometimes those high carry currencies achieve. Sometimes they are expensive based on things like purchasing power parity. So, the two will be similar if they share, what I call, the one over P effect. So, one over price effect. If both carry and value are inversely related to market price. And in many cases that’s true DP, but in other places, like currencies or commodities for example, that’s not.
Very interesting. Mm-hmm (affirmative). So if all these style premia are going to persist, who is persistently on the losing side of these trades?
Yeah. So I’ve had a 10 year project with coauthors and called, who is on the other side. And I have draft papers over the years. But it’s, so far they’ve always lost out to other priorities. And partly because, it’s a broad and difficult topic. But the main point is that, if you believe that the factor is rewarded in the long run, it’s sort of, you owe it to your… I don’t know, to yourself, but your clients also to think about who would be the one taking the other side sustainably.
And that question can be addressed either with theoretical arguments or empirically with some flow or holdings data, and both have been done in the literature. And let me just, I’ll take one example. There are no too easy answers to this one. But, so you should think certainly theoretically for each of these at least. So who would be there? And I take this defensive strategy that I just said. It has got very nice economic rational citing, with the leverage aversion and lottery preferences.
So the other side would be, leverage adverse investors and lottery loving investors. And then, when we go to flow and holdings data, we have what? We are limited by what kind of data is available. But when I sort of try to identify which investor groups might be on the other side, we have found that hedge funds are one group that systematically takes the other side. It’s sort of maybe surprising, but they are favoring the more speculative stocks.
And so, one can ask similar questions for other… At the sake of… And I’ve got the table, I think in the book where I show list of possible theoretical explanations’ rationale, or behavioral. And then separately, there’s this question of what does flow data tell? And for example, for value, flow data doesn’t identify easily a group. There’s an academic paper, which sort of even asks, to who is on the other side. But in general, I think we know that there are growth managers and value managers and that sort of. But we don’t find that all mutual funds all tend to be on one side, all hedge funds tend to be on another, or individuals. So it is within these broad groups, I think, where we find value and growth investors spreading themselves.
But last thing I say, that we did find in our research that there is persistence in investors, sort of factor kilts and that’s sort of good news. That’s suggesting that this other side, whatever it is, it tends to be persistent and arguably then sustainable.
We talked to Sebastian Betermier recently, who wrote the, Who Are the Value and Growth Investors’ paper. He’s got some interesting empirical work on those questions.
Oh, I’ve seen the value letter, and I went… I listened, and I asked my colleagues and… Any other data on this, because they’ve got the Swedish dataset. And so, I mean, this age story is so fantastic. Because again, I mean, it sounds right. And there is interesting literature. So just for readers, I think the main finding was that when you go older, you become more value oriented contrarian.
And there is outside finance, there are these things that young people tend to be more open to experience. And the older you get, the more you become stuck with your ways and comfortable. Sort of, from exploring we go into exploiting, and I find that when I’m aging. And so, it just sounds so interesting. And finding that in the case of value. But again, don’t know of any other study than that is. With this one Swedish database, I’d like to see something more before I sort of, I don’t know, start to do more on it. But I think it is a very interesting result, and hope there will be something more on that in the future.
He does have a working paper using Norwegian data, I believe. It’s a similarly interesting. Okay. So the style premia that we were talking about, they’re technically long short portfolios. How important do you think it is to actually be long short? And to add to the question, we’ve seen a paper suggesting from, a long only manager, to be fair, suggesting that tilting toward the long side of the premium in a long only portfolio is effectively equivalent to having a short position in the short leg. How do you articulate the importance of actually being long short to capturing the premium?
Yeah, I think partly that could depend on AC metrics between long and short. But, I think the more important thing I want to emphasize… So long short applications can allow you to do the purest most efficient way of capturing some style premia. And then they allow better risk diversification than long only applications, which are always dominated by market risk. I’ve heard you ask some visitors there are about diversification, and this is… It’s a what’s good diversification? Is market portfolio perfectly diversified?
But the problem with that logic is that your portfolio is driven fully by… Not fully, but very strongly by equity market directional risk. If you allow this long short, if you allow shorting to reduce the impact of market, and you allow leverage to make some other things… Could be value, could be many styles. Could be other asset classes also. If you let them matter as much as your equity risk, you are letting your portfolio be driven by many other things. So that to me, that allows clearly better risk diversification. There are problems. I promise I’ll get to that in a moment.
But so again, my sort of core point here is, then that when you allow sorting and leverage, you can get much better risk diversification. If you combine for uncorrelated strategies, you can double your sharp ratio. And with the long only world you just can’t do that. Equity risk will dominate. You are sort of only getting small incremental improvements with the carbon equity market risk. There are downsides. The downsides are, that the very features of shorting and leverage are uncomfortable for many investors. And there are limits to how much investors want to do this.
The unconventionality of this is such a challenge that I think it can hurt investor patience in bad times, especially. And then I think this type of logic makes me understand why most investors choose to not take this path of better risk diversification, but focus on equity concentrated portfolios with un-levered log only approaches.
And they can still use this style premia, blending market risk premia with much lower tracking error, or some style premia. So sort of silver light or sliver over the market exposure. But again, I do love diversification, and to some extent, I said that if you really want to go that path, then shorting and leverage are needed to take full advantage of it. But I know that this is for, I would say, only a small subset of investors. You really have to be a believer and you have to be really able to live with this for a very long time.
Okay. Interesting. Let me put that into geek language for a second and see if I get it right. If we looked at regression loadings on a long only value tilted portfolio, you’re going to have a market beta of one, and maybe a value of 0.2 or something like that. But you’re never going to get to a value loading of one, unless you go long short, is that a good way to think about it?
Yes, that is good. And again, there are some other… That was just now two things. If you start to add some other things, it’s again, the equity risk there. Again, I think, I mentioned earlier, equities are forgiven that bad decay and there are good reasons for letting that dominate. But, it is to me, this is not the best possible diversification. I can get more portfolio volatility reduction through combining lots of uncorrelated return sources.
So where is the line between alternative risk premia and alpha?
So, there used to be this alpha and beta distinction, and then people added alternative risk premia in the middle. So both alternative risk premia and alpha are, they are uncorrelated from the market exposure. So, somewhat uncorrelated anyway. But the difference between those two is that, these are pure alphas’ proprietary and manager specific. And this alternative risk premia, they are widely known and systematic.
So prime examples would be the kind of styles we talked about. Long short value, momentum carry defensive. So I think that widely known systematic are the key points. And this also means that this alternative risk premia, they are more evidence based. There’s lot of historical support on them in many countries and asset classes. I think manager specific alpha is more beliefs based. You rely on particular manager and your own ability to have picked a good manager, and often that’s influenced by recent track record. And so, you hear where my leanings lie.
You very briefly touched on this, but I want to dig into it a little bit more. Do you think investors are as patient waiting for alternative risk premia as they are waiting for the equity risk premium?
Yeah, okay. So, that’s… Equities are the only risk which are forgiven a bad decade. I think that’s… It’s a cliche, but it’s important. There’s greater patience by investors, for good reasons. Good and bad, I would say. So great empirical evidence, good theoretical justifications, bad returns in bad times. It just makes sense that this is a meaningful risk.
Conventionally helps, you can lose money sometime a lot, but you are in the same boat with others. Embedded leverage there. So this all help, they give… I think that even in my mind now, like this justification for holding those equity dominated portfolios, unless you have got such conviction on the diversification benefits of some lonely correlated style premia that you want to sort of go that way.
I have a practical question for you. What are some of the things that investors can do to stick with investment strategies during periods of poor performance?
Yeah. So, in the background there is this issue of, I don’t know, investor impatience. Investors require more consistent performance than is feasible in competitive markets. And this is, again, even with equities, but certainly with anything. Anything less conventional then we get even more this problem. Can be a factor, can a manager, can be anything. And we know that we need very long data to truly distinguish lack from skill, or whether factor is positively rewarded, but few investors have the patience for this. But strategy is good, only if you can stick with it.
And in that spirit, I emphasize the importance of identifying the strategies that are consistent with your beliefs and that you can then stick with. And for many that answer may be that, okay, really let’s let equity conventional risk dominate. By the way, it’s sounded like it’s a binary choice. Then I think, one can do some of these other interesting things, but size them modestly so that you don’t then capitulate too quickly.
But what I also do in the book is, suspect that most investors look in the mirror and like to see they’re a patient long horizon investor. And I try to then cultivate that patience. And so, I give some ideas. So, I think being educated about both the empirical evidence and theoretical rationales for anything, they can help sustain patience in bad times. And I suggest some practical tips, such as viewing your portfolio broadly and rarely. So broadly means that you don’t fall to the line item thinking, or try to reduce the line item thinking type of things. And there are tricks for that. Look at return contributions as opposed to… So, that’s return of the asset times portfolio weight, which is a smaller number, that can be helpful.
Or when you look at performance over time, you have to probably tell last quarters’ performance, but put it last as opposed to the first in the list. Start with some long horizon estimate. So it’s… And again, this doesn’t have to be… This shouldn’t be done without explaining this to, let’s say your client. The client should know that we are doing this together so that we are cultivating better long run behavior. And then, I think it makes all the sense in the world.
Now on sticking with a strategy, something that we heard a lot through the value draw down was, okay, the evidence is there, fine, but the world has changed. And so value’s not going to work anymore. And you touched on this earlier. What would you have said to that?
I think it’s a fair concern when a strategy has a particularly bad performance. But it’s also predictable overreaction by inpatient doubting investors. So one has to, not totally given… No equities were dead in 1979, value in 2020. We should worry about the structural changes, but that we should also recall that the value strategy tended to work 200 years through lots of structural changes, technological breakthroughs, and so on.
Again, suffering in those times, but they didn’t stop value from working in the long run. And very often markets took things too far. So, new age thinking, this time is different type of mentality took valuations, I would argue, to irrational levels sometime. That’s often the time when such discussions are. But that’s maybe too self-serving, sometime that concern is fair, and it should be addressed sort of seriously. But I do think that in the current situation we have that type of irrational behavior and the current opportunity is quite attractive and the value is not dead.
Should investors be more fee conscious in a low expected return world?
Yes. Fees are mechanically a larger part of the equation in this situation. And I write again, like chapter on trading cost and fees, and it is important to be cost conscious. But I also say just sort of, maybe I stop with just one lighter. Cost efficient investing is not minimizing costs. It’s still maximizing risk adjusted expecting net returns. And so, one has to be thoughtful about it, like there can be a role… For example, for those private assets in some cases. But I think one has to be skeptical. The more expensive products are… Cliff has got this lovely saying, the Asness law, that there is no investment that is so good, that high enough fees can’t make it a bad investment experience.
That’s a pretty good law. Can you talk about the premier bad habit, as you called in the book, that investors should avoid?
Yeah, I think chasing returns at multi-year horizons, when it’s typical, that mean reversion is more likely. I think that’s another Cliff quote here, that investors act like momentum investors at reversal horizons, sort of capitulating after three to five, four years. And that’s probably pronounced… So for institutions, by consultant aid in manager selection process.
And so, it’s related to this investor impatience theme that I’ve highlighted, and there are many studies that suggest that multi-year return chasing is unhelpful, but it is just human. And I think we just have to try to do better against that challenge. But if the message from statisticians, or textbooks, or wherever it is. That you have to wait 10, 20 years or so. And I do understand that investors roll their eyes. And if it’s an underperforming manager that tells that, not a helpful conversation.
Can you say that Cliff line again?
That investors often act like momentum investors at reverse self horizons.
That’s a very funny geeky joke.
Yeah, because again, just to… If somebody doesn’t know, you want to be a momentum investor at momentum horizon. So, momentum tends to work. And I’ll add to this, not for you. This is for the audience, momentum works up to one year when you look at historical data in many asset classes. But after that, you start to get gradually mean reversion tendencies over multi-year horizons. So you don’t want to be a momentum investor at multi-year horizons. And that’s that.
Yup. That’s very, very funny. Very geeky and very funny.
And Cliff doesn’t like explaining jokes, it’s sort of spoils them. So I’ve just done that.
What are some of the other bad habits that investors should be cautious of?
I think I list things like, under diversification, cycles of fear and greed. But the one I emphasize most, and after this, is overconfidence sort of. Because, obviously the direct impact is sort over trading, but there are many other problems, sort of bad behaviors, bad habits that are exacerbated by that. You don’t learn as well, for example, any mistakes when you are over confident.
Maybe I’ll say here that… So many of these ideas, they are coming from behavioral finance literature. And I think that can be very helpful. But I would say that they’re often used to motivate strategies, how you can exploit other investors mistakes. But the best role really, should be self improvement. Good long run performance requires good investments and good investors. So let’s just start with that. Let’s try to improve our patience, consistency, discipline. I say stable confidence, these types of considerations.
We understand that Ken French, who was our guest episode 100, and who we know indirectly at least, through our relationship with Dimensional, that he’s had a big impact on your life. Can you tell us about that?
Yeah, I guess in most people’s lives, or many people’s lives, there’s a sliding doors’ moment. This kind of crossroads where your future takes one path. And with me, that was 1989 when I met early in the year, I met Ken French. I was a young portfolio manager in Phoenix Central Bank. And he liked me enough that he encouraged me to come and study finance PhD at the university of Chicago. And he asked me how old I am, and I was getting 28 soon. And he’s… You are pretty old. And we’ve got lots of young people there and you might want to come soon. And so, I did, and I applied and likely aided by his good words, I began my studies that same fall. And that first year it really, it changed the rest of my life in three wonderful ways.
So the education… I mean, the first, the learning curve in the first year is still, I don’t know, it’s sort of almost like a drug thrill. And it was the basis of my finance career, subsequently. I also met my German wife who was doing MBA there. A year later she would’ve been gone, we wouldn’t have ever met. So that’s there. And then I met my fellow students Cliff Asness, John Liew. I love them. And they later co-founded AQR, which I joined then much later, a decade ago, finding my natural home. So, I’m thinking, Ken French has a big impact on many lives, but I think few as dramatic as mine. So, I’m forever grateful. Actually, I did feel very lucky, like if Fama had the Nobel Prize. I got myself a way to get in Sweden, myself that Nobel Prize event, and met Ken then. And I told him this. And I was just glad that I had that chance.
Wow. That’s an incredible story, Antti. Thanks for sharing. Our final question, how do you define success in your life?
I think I’m a diversifier. So, balance is key. Balance. It’s work and family, that’s balancing my soft Nordic values with, I don’t know, harder, more rational university of Chicago and finance career values. So those are important when I think of success. And then, on this giving something back. On that front, I don’t know, maybe it’s a lazy way of thinking about it. But I do think that my best contributions are related to this financial education, these books. I am proud of them and hope they’ll help investors. And that’s something which has given me big joy, not quite as much joy as my two children, but I show my picture of my two books almost as if my two children.
Well, congratulations on this book. I’m holding it up for those on YouTube. It’s just a wonderful book. Thank you so much for joining us. This has been a great conversation.
Good. Thanks guys. Great show. I’m a real regular on it. And so, on the audience side. But, wonderful job there.
Thank you very much.
Extra: When I’m in Chicago, when I went to the school, in the first class that I had… The first year’s big class is the Fama class. And I go there, I’m a little older. Cliff is a teaching assistant there, and he’s five years younger than me, which is awkward. But he’s also super smart, and so also this is… But, sort of for my future, it is important. I got sort of best both from the whatever exams and some paper, I did very well. It was good for my reputation then with Fama. And I think it was good for French, who had sort of helped me in. But also, I think Cliff has since then, sort of thought that, oh, he’s a somewhat smart guy. So he got good impression of me. And when I think of the career implications of that one, it has been something. So anyway, that part I left out now, so.