As part of our ‘Best of 2020 Acquirers Podcast Series’, earlier this year Tobias had a great interview with Cliff Asness, Co-Founder of AQR. During the interview Cliff provided some great insights into:
- Systematic Value Investing
- Do Fundamentals Still Matter?
- Does Price-To-Book Still Work?
- Is There A Correlation Between Value Investing And Interest Rates
- The Reason That Value Hasn’t Worked Recently Is Not Because It’s Too Well Known
- Investing Parallels Between Today and The Late 1990’s
- Value Works And Random Doesn’t
- Try To Stay Imperfectly Irrational
- I Get Excited At The 140th Percentile
- Market And Factor Timing Is An Investing Sin
- Sin Occasionally And A Little
- Codification Of The Momentum Strategy
- The MAGFAN’s
- The Valuesburg Address
- Never Has a Venial Sin Been Punished This Quickly and Violently!
- The Interaction of Value and Momentum Strategies
- Time for a Venial Value-Timing Sin
You can find out more about Tobias’ podcast here – The Acquirers Podcast. You can also listen to the podcast on your favorite podcast platforms here:
Breaker
Overcast
RadioPublic
Stitcher
Full Transcript
Tobias: Hi, I’m Tobias Carlisle. This is The Acquirers Podcast. My very special guest today is Cliff Asness, the cofounder of AQR. Cliff is the prototypical quant. He’s got an undergraduate degree in computer science and finance from the University of Pennsylvania, his PhD from Chicago in 1994, under the great Eugene Fama–
Cliff: I think it’s Fama, but I think he’ll answer to either.
Tobias: –who’s a Nobel laureate now, and it was two years after he had, along with Ken French published the seminal paper introducing the three factors. And then to Goldman Sachs, who established the quantitative research group. And then finally the founding of AQR in 1998, which is now $140 billion AUM firm. We’ll be talking to Cliff right after this.
Tobias: Hi Cliff, how are you?
Cliff: I’m good.
Investing Parallels Between Today and The Late 1990’s
Tobias: You published shortly after you established AQR. Long-term capital management collapses, Greenspan slashes rates, the dot-com boom takes off, and long short value gets crushed. And in that time, you wrote an unpublished paper, it’s now available around bubble logic. Can you draw any parallels from the late 1990s to today?
Cliff: Yeah, various episodes for value, both good and bad, but we remember the bad ones. They’re not the same, but they certainly rhyme. Let me take you to a little back to August of ‘98. We were launching our firm, as you said. Our first product was an aggressive market-neutral, long short fund. That is an interesting product to launch in the teeth of a very bad period. We still have a world where even if statistically you’re just as bad if they’re performing a bull market, as opposed to losing gobs of money at high vol even if that was intentional in design, that’s something I would probably go back and change the order of things we did things.
We never been pure value managers, and that’s true today. It’s always been at least value and momentum growing over time to quality, low-risk investing, various other things, but value has always been a big part of it. I’m Gene’s student and I have studied that a long time, and I do believe in it.
Our first month, August of ‘98, we were celebrating. I don’t to be mean maybe ghoulish. The S&P was down more than 20%. Long-term capital, as you pointed out, began its collapse that month. But we were up that month. As you’re lead-in indicated, was we really should not have gotten too excited about that because it’s certainly Greenspan’s main candidate, but there are many other possible explanations, but that was the last month before the 18, 19-month crescendo of the technology bubble.
A lot of parallels, value was obliterated, the world– Mr. [unintelligible [00:04:07] might think I don’t know this, but the world is not normally distributed. We know that going in, so we saw four or five standard deviation value realizations. You see that for virtually anything in investing over anything but the super long term. It happens occasionally. Interestingly, and this is a parallel to the last two years, and I make a real distinction. Eventually, of course, we’ll be talking about recent events. And it’s been like a 10-year bad period for value. I obviously have a quant perspective on this, but I think that’s shared by a lot of more traditional, Graham and Dodd kind of concentrated stock-pickers. I always feel like the interplay between those. What’s similar and what’s different is interesting also.
But for the first eight years of those 10 years, was actually a very good period for my firm. So, it does point out simply that we can’t simply be pure, systematic quantitative value investors for that would not have worked. The obvious answer is everything else worked real well. Profitability, earnings momentum, low-risk investing, all more than made up for value. In the last two plus years, both the value– and I think, again, value managers of all stripes, if you do this with simple value ETFs, not that’s how we implement, you absolutely see– it has accelerated, so values have done poorly and worse than– not the cumulative eight years, but worse than most of the periods. This is absolutely a parallel to 1999-2000 in the tech bubble.
The rest of our process that normally, even values doing poorly steps up, didn’t then and didn’t in the last couple years. This is my version of the story, I’m sure there could be other interpretations. But a lot of the rest of those things are meant to capture the non-value aspects of investing, our earnings getting better, is it a profitable company where you should pay a bigger multiple, sustained profitability, moats, whatnot.
So, it turns out– and this is a very self-serving way to phrase, I admit that though I do think it’s accurate, in what I would call a rational loss for value. We overload that word too much in our field, I don’t mean to give it a moral dimension, but in a loss for value that is deserved on the fundamentals. Actually, for at least our style, and I think for a lot of investors who aren’t just looking at price, that can be a pretty good time. Sure, pure value doesn’t work because it’s lost on the fundamentals, but much of the rest of the things you look at, have a chance to make up for that in the tech bubble, and notice I call it a bubble. As an ex-Gene Fama student, when I say bubble, I still feel–
Tobias: There’s no such thing.
Cliff: When I’m on the East or West Coast or outside of the US, I’m very comfortable saying bubble, but the closer I get to Chicago, the less comfortable I am saying bubble, but I do think that was a bubble and I think we’re seeing somewhat of a similar of a bubble within stocks now. But when value loses, what I’ll call irrational reasons, and that has been the case for the last two-plus years, and it was the case in the tech bubble. Not because they’ve massively underperformed on the fundamentals, but simply because multiples have done this. For the first eight years– and I know I’m blending 99-2000 with today, but for the first eight years and don’t hold me to the exact numbers. If you measure value slightly different than me, it’ll be nine years, it’ll be seven years. But for the first eight of these say 10 years of the value drawdown, value did not get super cheap on our measures.
It’s hard to remember, but three, four years ago, I was arguing and taking the other side from people who are a little more knee-jerk contrarians, looking only at returns, value has been destroyed. Therefore, it has to come back. But I like super simple examples. If you bought a stock [unintelligible [00:08:52] had a low P/E, and you shouldn’t buy a single stock based on this, and simple P/E is not a way to run your life. But if you bought a stock and it had a low P/E and the price drop 50% but the earnings drop 75%, a pure return contrarians says, “Hey, it’s got to be super cheap now.” But someone who’s looking at the most basic evaluation ratios and says, “No, it got more expensive.” That’s too extreme value did not get more expensive in those eight years. In fact, it cheapened a little bit, that was part of the loss, but a small part. Most of it was fundamentals. And again, it turns out that’s a pretty decent environment for us.
In the last two plus years, we’ve seen the factors that we believe in that are meant to pick up fundamentals not work, or at least not work nearly well enough. And the multiples smash out, the cheap getting cheaper, the expensive getting more expensive. Those, of course, go together. If value is losing not on the fundamentals, you see multiple expansion in the relative multiples of cheap versus expensive.
And that is precisely– I remember, back in the tech bubble, we measured momentum, two different ways, and pretty much always have, both price and various ones of fundamental momentum. And they are correlated, of course. Price tends to move with good fundamentals, but they’re different aspects. And we prefer both. As a momentum investor, we prefer both the need to one or the other. And in the tech bubble, we saw half of what we do in momentum, the fundamental side utterly failed. Price momentum held up.
Price momentum is about the only thing I know out there that’s your friend in an irrational bubble, or at least should be on average your friend in an irrational bubble. But fundamental momentum and other fundamental measures goes away and that’s a very strong parallel between those periods. I have often said even prior to this last two and a half years, I know this is a weakness for our process. Anything outside of the Jim Simons fund that they only do for themselves tends to have some period they don’t like. An irrational loss of value is what I think of as our very hard to hedge period.
Obviously, coming up with something that would make gangbusters of money in such a period and not lose money over the very long term. I know something that will make gangbusters of money in an irrational loss of value. It’s called short AQR.
Tobias: [laughs]
Cliff: I don’t think that’s a very good long-term strategy. And I wouldn’t add that to my process and timing that, of course, is extremely difficult. But if we ever found something on a more rational basis, outside of price momentum– price momentum has its own difficulties. It can help you in these periods, but it has a pretty bad short-term left tail itself. So, you’re always balancing risks. That would be the holy grail, would be to come up with something that was reasonable, that we think we understood why it worked, that we could tolerate a decent amount in the process and did great in irrational losses. for value. I would not hold my breath. We’ll never stop looking but that– I don’t see that on the horizon.
***
The Interaction of Value and Momentum Strategies
Tobias: Have you ever looked at Partha Mohanram’s G-score? Are you familiar with it?
Cliff: I have not. [crosstalk] response from my part.
Tobias: He’s an academic at the University of Toronto, a professor there.
Cliff: I do not look at anything by Canadians.
Tobias: Well, he’s studied at Harvard. But his G-score is—
Cliff: I was joking about the Canadian thing. You’re going to get me banned from Canada.
Tobias: He’s not Canadian, don’t worry. He has a G-score and it explicitly looks– so it’s somewhat akin to Piotroski’s F-score, except it looks explicitly in the most expensive 10%. And then, he tries to find the winners out of that 10%. And so that was actually going to be my last question for you. But it seems to have been the best performed strategy over the last two years at least.
I don’t know what the long-term track record is and that may be a failing of it, but he seems to be able to identify. He says, of that most expensive decile, you’re going to find that there are lots of losers and he finds that as a long-short strategy, it derives much of its return from being short to losers, but the last two years, it’s derived a lot of returned from being long the winners, which is unusual for– [crosstalk]
Cliff: I do this a lot. I answer a lot of questions with– I wrote a paper about that at some point. It’s partially a function of being at least mildly prolific. It’s mainly just a function of age, and I’ve done this for a long time. I wrote a paper called in Financial Analysts Journal called The Interaction of Value and Momentum. And it was really a very simple message. I wouldn’t hold through the whole process over this. Sometimes, people get overexcited on it. Any trading strategy that first cuts this way and then cuts another way, tends to be a pretty high turnover strategy because you can leave the preferred long or short portfolio for an intersection of reasons now and so that has its challenges.
But I found that momentum that was considerably better among the expensive stocks. There’s certainly quite a few places we incorporate that but we incorporate that at a modest level. I believe it has helped, though I don’t think– and again, I look at his stuff, I certainly wouldn’t turn the whole process over to that. But just hearing it from you, it sounds pretty reasonable. Of course, my definition of reasonable is I wrote something similar at some point. It’s a very self-serving definition of reasonable, but it does sound reasonable to me.
***
Codification Of The Momentum Strategy
Tobias: Your PhD thesis was on momentum. And you may have been perhaps codifying momentum. Momentum in some form has been around for quite a long time, but under Fama, who’s very well known for being efficient markets, and he’s your approach with your momentum idea. And he said, “If it’s in the data, write it up.” He and Ken French both say, you’re the smartest student they’ve ever had.
Cliff: Some people say they say that. I have never asked them for confirmation of that.
Tobias: It’s in the Wikipedia entry, so it must be true.
Cliff: It’s in my Wikipedia entry?
Tobias: Maybe it’s not, maybe I’ve read it somewhere. Sorry, I shouldn’t say that.
Cliff: I tried very hard to look at that extremely rarely only when I have to because, like everyone else, there’s always something that annoys me in it. Also, I will not be listening to your podcast. I don’t listen to myself afterwards because it always upsets me because I always sit there and go, “Oh, this is what I should have told, Toby.” And it’s too painful. But yeah, I did write a dissertation on momentum for Gene. It did some other things. There were some broader simulations that never saw the light of day and publication. But certainly, one of the big things in my dissertation was momentum.
I wasn’t even the first to codify it. I was just on the heels. I think if I was a professor, I would have been tied with them, but I had to get a dissertation approved. But Jegadeesh and Titman, I think deservedly get the first place in academia for studying this. Though to your point, momentum was not something that investors– investors have been looking at momentum, relative strength, trend, whatever you want to call it forever. Value was not new. I don’t think Fama or French would be insulted for me to say that that they didn’t invent value investing in the late 80s and early 90s. They codified, tested, which was huge. A lot of it was anecdotal, until then. The codification and the testing is not a small thing. But I think guys named Graham and Dodd roll over in their graves if we say, Fama and French invented value in 1986.
I think if you go back to the Roman Empire, somebody’s buying something because the price was II, but the fundamentals were V, and that was a low ratio, and someone was buying something because it started to go up recently and someone was doing both. I do think of what all of us did was actually the right thing to do. Take a lot of standard wisdom that’s out there. Often, standard wisdom is right. But often standard wisdom is absolutely not right. And it’s never been formally tested. So, I do think of us collectively, including very much myself, as not super original, but original in going to test it rigorously.
Where I added maybe to Jegadeesh and Titman is they had a– which is often the case for the first paper. A fairly convoluted way that was confusing to measure momentum that mixed some short-term contrarian signals with momentum– you got a weird result. I did a cleaner version that ended up working over the whole ‘26 through ’90 period. The last date in my dissertation was 1990. So, I am still successful, even given the last two and a half years, less successful after the last two and a half years, but still successful 30 year out of sample period, which is cool.
Tobias: Yeah, very much so.
Cliff: Unless, of course, [crosstalk] to be old, in which case it’s not cool.
***
Systematic Value Investing
Tobias: In November last year, you wrote your Time for a Venial Value-Timing Sin paper, inspired perhaps by– so we had a decade of underperformance, eight years of which was irrational– Sorry, pardon me, eight years of which was rational because the fundamentals were either– it was too expensive relative to its fundamentals, the spread was very tight. And then the spread widened through to last year. And so, then you said that the last two years up to November had been an irrational punishment of value to that point.
The two questions that you answered in it, how cheap was value at that point or how cheap is value now? So, you’ve done that on a variety of different bases. You use the academic version, which is price to book. The simple combo, which included forward P/E and TTMP or backwards-looking P/E, price to sales, price to book. Then the AQR version, which is 25 value signals, and perhaps some quality signals in there as well.
Cliff: No. It’s only price to something, but some proprietary measure, some price to free cash flow kind of members. There’s no magic. We don’t claim to know the precise way to value something. Again, an active manager is disadvantage. They have to take a more concentrated bet, and they better be right. But they have advantages too. They’re better than us to figure out what’s most relevant for each company. We try to take an approach of anything that’s reasonable, that’s not perfectly correlated to the other ways that we’re measuring it, we’d probably give some weight too, but yeah. We found late last year– again, remember, it did not get very cheap. Cheapened a little bit, but for the first eight years when you lose on the fundamentals, you don’t necessarily look cheaper. And that applies to individual stocks but it also applies to spreads.
One thing to take you back again to the tech bubble, in 99-2000, in round one, [unintelligible [00:21:22] for us, value is doing terrible, should we stick with it? We did invent this measure which the world that generally uses which– I don’t have a lot to be proud of return wise in the last couple years, I’m still proud of inventing this because I think it is useful, that’s called value spread. All it did was say almost all the work, all the work, that I had seen today, I don’t know what people did privately in investing firms, but all the academic work and all the published practitioner work were sorts. Go long, cheap, short, expensive, price to book being the most common, but whatever measure you like, and they ignored magnitude.
So if all of the price to books, price to sales, price to earnings were tight and clustered, you could still sort and go along, the cheap, and short the expensive, but you’re not getting a big difference. And sometimes, like in the tech bubble and like in today, that difference is relatively gigantic. We did show that that is like value for the stock market. Don’t expect to get the next month right, with this timing indicator. I think in that paper, it’s a little ancient history now. But we focused on three years horizons and it was not perfect, but a fairly decent predictor in the direction you’d imagine when it was cheaper. And I think those three-year results hold up. We’ve been tracking this ever since.
***
The Reason It’s So Hard For Investors To Follow Strategies
To be honest, we track it for both reasons. What if it gets super cheap or what if it gets super unprecedentedly tight? But I’ve always thought the second one, which is decidedly not the case now, was the most important. Does it get super tight? Do the differences in valuation get arbitraged away? Because this is hard to remember, but two and a half years ago when life was wonderful, at least for us, not for pure value managers but certainly for us, the most common question I probably got from investors was, “All right, but if you guys are right, maybe do some proprietary things that you hope are a little better but largely you publish on this and other people publish on it, how can it still work, if everyone knows about it?” And I would say, “Well, first of all, we’ve been through, even excluding these last two and a half years, some horrible periods for it. It’s actually much harder to do in real life.”
When you look at a back-test over 100 years and you see the drawdowns, you know where it ends, you know it ends well. So, you mentally look at that, and you go, “Oh, yeah, stick with that. Look, it’s a good strategy.” When you’re near the bottom of that drawdown and particularly if you are unlucky and started near the top, we do find this with investors and even employees, someone who’s been doing this with us for 20 years. No one’s happy about a bad period. But there really come.
Someone who happened to invest 2.4 years ago or a new hire that we made 2.4 years ago, we just never seen it work. But add all this up, we’ve measured this value spread, again, probably tracked it, if anything from both but more as an early indicator that if this was ever arbitraged away. We didn’t think it was easy to arbitrage it way, again, it’s hard to stick with these strategies, but it’s possible. What would that look like?
That would look like this price of the expensive divided by price of the cheap, that’s the order we did it, and so when it’s high value, it’s attractive. Expensive is more expensive than usual, against the cheap. I only memorize it for price to book and we’re certainly not a pure price to book shop.
But using Fama–French kind of data, the price to book of the top one, roughly one-third– most expensive on third, averages five, six times the price to book of cheap one-third. It’s gotten down three and a half, four. It’s gotten up to levels in the tech bubble. Today, parts of the GFC, anywhere between 10 to 13, so that’s the range we’re talking.
Try To Stay Imperfectly Irrational
But if we ever saw it smashed down to unprecedented low levels, one and a half, I’m making up a number, but a level well below what we’ve seen and stay there, you really might look at that and say, “Well, the fun is over.” Whatever imperfect rationality– I try to stay imperfectly rational as opposed to irrational, I think irrational is too pejorative and too extreme. I don’t think markets are wildly irrational, I think they’re imperfectly rational, which is a pointer in the spectrum, but I think an important one.
But we could wake up one day or probably more logically over the course of a few years and see the world has caught on to this and they’ve invested so much in it, you can’t make an infinite amount of money from any strategy, enough money in it closes the strategy. And ironically, we probably kept monitoring this again. Buy more in case that ever happens. We don’t need to do value, we have plenty of other things. If we thought value was arbitraged away, we would not shy away from telling people, :Look, they’ll always be cheap and expensive stocks, but the market is not pricing it to a gap.” And if you’re behavioralist, maybe they used to be making errors and now they’re getting it right. I will tell you, and this is the only way I’ll get political at all on this podcast, but I don’t look around the world today and wake up in terror that the world has gotten perfectly rational.
Be it in finance or outside of finance, that’s a tough story to tell. But if it did, we would change what we do. Again, being honest, I didn’t think– if you had asked me after the tech bubble, whether I’d see something as extreme as the tech bubble again in my career, I certainly would’ve said it’s possible, once you’ve seen something to say it’s impossible to see it again. I did think that you need a few more generations of memory loss before you have the same level of craziness and I’ve been proven wrong on that. To your point late last year, the spreads were quite wide, not as wide as today. But 90th and plus percentile versus history as opposed to today, which is in the best ways that we can measure it, the solid 100th percentile, at least against pre-2020. It’s bounced around the top.
Market And Factor Timing Is An Investing Sin
I wrote a piece saying, “We’ve all always called both market and factor timing an investing sin.” And we don’t mean that morally. We just mean that it’s really hard and you could screw up a good long term allocation on by getting in and out at the wrong times. My investing nightmare, and I think many investors’ nightmare, is they’re actually long term right, but they screw it up enough along the way that even though they were right, they failed. And if you hide things too much, you introduced that possibility.
Sin Occasionally And A Little
And timing is just hard. You do it empirically. It’s really hard to find timing strategies that you have a lot of faith that aren’t even data mined or strong enough to matter. So, we’re not big fans of timing, and repeatedly. Thank God, not just recently, repeatedly, we’ve said, timing is an investing sin and our recommendation is you sin occasionally and a little. At true extremes, after you’ve looked at it 100 ways to make sure that there’s not some easy explanation for why we’re at true extremes, and then do it in a modest way, don’t suddenly become a pure value investor. Stay on that, what we call a flat surface. It’s the optimal portfolio and we’re always guessing at the optimal portfolio, but if we think the optimal portfolio is this mix of factors, even when we think value is the beaten-up one, that is the best three-year horizon returns, we won’t add so much that history would say that’s a far worse combination. We’ll go to the edge of that mix.
So, sin a little, and sin only at true extremes. And then don’t expect to be rewarded immediately. We’re people who preach that the timing is hard. I did not expect to get slapped that hard, that quickly. In some sense, I’m glad that this was both intellectually honest and cover-your-behind kind of stuff. We spent a lot of time in that first piece saying, “Look, it could get a lot worse.”
Never Has a Venial Sin Been Punished This Quickly and Violently!
Going back to the tech bubble, NASDAQ 5000 could have gone to 6000. Once you accept that the world can be imperfectly rational or even in that kind of case, downright irrational, to say that you know exactly how irrational it can get, it’s madness. Once you’ve accepted irrationality, you do think the moral irrational it is, the better your odds get in the chance it gets crazier, at least gets smaller, but it doesn’t go away. So, we did talk a lot about this is a long-term bet. We’re putting our own money where our mouth is, both in the portfolios and even in some personal investments that I shared with clients. And then I had to write a piece early during the crazy first quarter, basically saying, “I’ve never written anything that’s been this wrong this quickly.”
Tobias: Never has a venial sin been punished so quickly and violently.
Cliff: Yeah. I think venial sin should be punished in a venial manner. But, again, it’s not fun. We all are susceptible to feeling once we put a position on the world supposed to go and it’s supposed to start to work. And all of us know that’s not true. That’s not how the actual world works. But we’re still, at least I am, and I don’t think I’m unique in this, a little mortally offended when it doesn’t work that way. We’ve updated that. We published that just to show the scale of the move. That was a short piece just saying wow. I switched temporarily, instead of using the AQR measures to using simple US value, a set of value and growth ETFs. I did that. It’s the exact same, the pattern has been very– I shouldn’t say exact, but it’s highly similar. You get the set, but I wanted to use some non-AQR sources. Every once in a while, if you use only someone else’s data and get the same result, it’s a little credibility that it’s not some odd, cute thing you’re doing that’s driving it.
And then I wrote an additional piece after that just updating the value spreads. And now they’re off the charts. Literally, they’ve created a new–
Tobias: [crosstalk] percentile.
I Get Excited At The 140th Percentile
Cliff: Yeah. I have a running joke for many years that I don’t get too excited at the hundredth percentile, I get excited at the 140th percentile. Obviously, it’s a joke. Your listeners are mathematically inclined. Hopefully every single person has said, “There’s no such thing bigger than hundredth percentile.” It’s a joke but there’s some meaning to it. I mean something well past prior extremes, not good at the hundredth percentile, something doesn’t get interesting and worth a very venial sin. But we are now– depends how you measure it, this is always a problem. You could measure value with the Fama–French way, I could take out the industry bet and use 25 measures of value. We will not come up with precisely the same results.
On your example, the tech bubble earlier when I said it’s similar but only rhymes, it’s not the same, the tech bubble was more of an industry bubble. It’s in the name. It was a tech bubble. It also was crazy within industries, even the way AQR does it. But it was a bigger bubble if you allowed industry bets, which is something we generally don’t take very big industry bets. Today, it’s bigger if you don’t allow industry bets. So, I just say that to say there’ll always be small differences and that’s not a gotcha. It’s just a fact. But the stuff that gets the most realistic, we find to be past prior hundredth percentiles. I will say that correctly, not whimsically. And again, no reason we’re going to get the timing precisely right. This time, I think the odds do get better, the crazier the world gets. And I think anybody who cares about price will say they prefer the odds. The more they’re being handed in terms misvaluation, they prefer their odds.
But this is where long- term investors– not just in value, but any kind of long-term investor. This is where you make your bones. You do a strategy that long term works. You spend a ton of time, which we really haven’t talked much about. But we’ve spent a ton of time trying, I hope, in a very honest way to say isn’t broken. Is there something we’re missing? There are many– Whenever strategy doesn’t work, it becomes an industry of people to explain why of course it doesn’t work anymore and of course, it will never work again. You can’t be too quick to believe that or you will always abandon a good strategy, usually three days from the bottom [crosstalk]
Tobias: That’s next point that I was going to make. I thought the timing of the first paper was excellent because I felt that the bottom of that spread was around late August and then it had been closing slightly. And so, I thought you’d done very well to get that paper out, unfortunately then–
Cliff: We did have a small victory party that was then subsequently well more than eradicated.
The Valuesburg Address
Tobias: And then you’ve had to write the second paper, reading The Valuesburg Address, which is hilarious too. And then–
Cliff: [crosstalk] there was no content to that.
Tobias: Your colleagues have written is systematic value dead and you’ve attached a blog post to that where you’ve brute-forced some ideas in that looking at the spread. And from what I could see, those papers were just trying to deal with many of the narratives that have a reason for why values not working and which I think is done very effectively in there.
The most interesting thing I thought from it was just showing what side of the spread is driving the valuation, so you compare the middle to the cheap side and the middle to the expensive side in an effort to see is this a phenomenon of the stocks being very expensive as they were in the tech bubble or is this a– although they were cheap in the tech bubble too? Or is it a cheap to the middle phenomenon? I think that you’ve said that this is more cheap being very cheap than expensive being very expensive, although it is both.
Cliff: Yeah. You got that exactly right. It is both. But I think most of the stories you hear– and we try to create our own stories too, but certainly anything that’s become popular, we get asked about and a lot of them are reasonable. And we go to try to test them. Most of the stories I’ve heard for values defeat has been some acronym, the FANGs, the FANMAGs, I call them the MAGFANs.
Tobias: FANMAG.
The MAGFAN’s
Cliff: Well, here is the story. I came up with my own because everyone does. The MAGFANs, Microsoft, Apple, Google, Facebook, Amazon, Netflix, Tesla. But I always joke– I joked in the piece that I use Google, not Alphabet because it was hard to make an acronym with three As. And one of my employees sent me an email that was quite amusing. He said actually if you use A for alphabet, it spells a fat man and you could make it autobiographical. That was not the last [crosstalk]. That person is still employed. But the story has been more of those stocks.
We show just in a mechanical way, like you said, instead of comparing cheap to expensive, comparing both to middling stocks. The expensive or more expensive than they normally are, the cheap or cheaper than they normally are, but not historically unprecedented, but more of it is coming from the cheap side, which may have its own stories, and maybe I’d be encouraging more exposed stories that industry can produce stories, “Oh, yeah, we meant that was really the problem.”
But clearly, that doesn’t mean value is going to work tomorrow again, but it does mean that the most common stories seem to be a stretch. Most of the common stories you hear are really only explaining one side. And it’s that cheap side is, if anything, contributing at least somewhat more to the spread than the expensive side, not that we wouldn’t do both in a value strategy.
Does Price-To-Book Still Work?
Tobias: I wonder if I could just take you through, from your colleague’s paper Israel, Lawson, and Richardson. They identify five of these arguments and then they deal with each one in turn. The first one is that book doesn’t work anymore and book is the value factor. When anytime anybody hears systematic value, that’s the first thing they point out. Nobody uses price to book to invest. And I thought that of all of the parts that they deal with in there, I couldn’t quite– they sort of seem to agree with that sentiment.
Cliff: Let me think about that.
Tobias: You would never use price to book in isolation, you would always use some– [crosstalk]
Cliff: Oh, well, that I absolutely agree with. Price to book, first of all, we think it’s somewhat less broken and they go through that, then other people do. One dirty secret about value investing is price divided by anything gets you at least somewhat of a value strategy. The famous equal weight portfolio is essentially a value in a small cap till that’s just one over market cap, is just moving away from that. If you divide price by anything– if you divide it by something smarter, you get a less noisy strategy, divided by– I don’t think the equal weight version is close to a particularly good value strategy–
Tobias: But better than market cap.
Cliff: And I think they and I both start with the– some of the stories about book being worse now are credible. If intangibles matter a ton more than they used to, price to forecasted earnings might pick that up really well because that’s in your ability to make earnings. Book might have more trouble picking that up. Now, again, their paper is deeper than mine. Mine is brute force, they go through explanations where they think that’s not as extreme as they think. But they and I absolutely agree. I could live without any price to book. That’s not a heroic statement, that is simply because price divided by anything, fairly correlated substitutes or druthers, if you have 15 to 20 value measures is to give price to book, a non-plurality weight but in line with many of the other factors, it is an aspect of value.
A small thing that’s actually a big thing at times, and I think you know this from our work. And this is debatable, someone else can take the other side, but we don’t believe in using value to take an industry bet or a sector bet. We wrote a paper on that. I told you I saw a lot of sentences like that in 1994. And pretty much since then, I’ve not been using value to take an industry bet. That helps with book. I would agree comparing the book of a bank to a tech company can get a little odd. So, a lot of the criticisms people have get a little weaker when you don’t– everyone thinks about value as tech versus textiles, tech versus financials. You can never take out that bet perfectly because you don’t have a perfect industry map. Some companies cross industries. But you could take out a lot of it. You could compare banks, to banks. And some of the criticisms, and they discuss this go away there, but I absolutely agree with them. Book should not have any special standing in a tongue-in-cheek way, because I’m more than a fan of Fama and French, but I do blame Fama and French and Frank Russell for this.
Fama and French made it their central measure. They’ve written plenty of papers saying, just like the rest of us, a lot of the other measures are really pretty darn similar, but we’re going to stick. They do like it probably a little marginally better and they have a good reason for it, it’s [crosstalk] lower turnover.
Tobias: Just that it’s just more stable.
Cliff: Yeah, exactly. It’s lower turnover than some other value strategies. On net, we think we trade for fairly cheaply, we’d still prefer the mix. So, they might have a difference of opinion. But as always with those guys, it’s a well-founded difference of opinion. Between them and Frank Russell, making it a very large part of their value indices. A fair amount of the world is locked into thinking of price to book as value or at least as quant academic systematic value. And this is not just AQR, I think you’d be hard-pressed to find many practitioners who give it even particularly special standing.
And let me add my favorite irony. We have a page in a presentation, but we only look at five. It gets a little crazy if you look at everything, but we look at five major valuation indicators. I think we look at price to book, price to sales, which you actually compare it to what’s called enterprise value, not sales– I’m not going to be getting into that sales before interest, so you want to compare it to the whole capital structure. Free cash flow, trailing earnings, and forecasted earnings. And just for fun, we have our own periodic chart of the order. They’re either up or down, and then the order of which did better.
Obviously, value has been a successful long term and a very poor short-term strategy. And you see that where the boxes are clustered, they’re usually on the same side of zero, except in years where values return was middling. And then the small differences can make one negative, one positive. But if you look over this whole period, price to book is the worst of the five. It is the best of the five in the last three years.
Tobias: [chuckles]
Cliff: Which I just find– not that I’m enjoying the last two years at all, but I find that to be a really entertaining irony to it. And I actually think there’s some economics behind it. Bear with me, this gets a little convoluted, but I’ll get there. Let’s say we assume the price to book is a reasonable but not the best value indicator, the other ones are better. And by the way, when I say it’s been the best of the last three years, that means the least worse, the least bad. Price divided by anything, broadly speaking, you can always find something– the price divided by anything, it’s not been a good thing to be doing for a while. But price to book has been the best of a bad lot.
Imagine price to book is worse, what does worse mean? A book contains noise in it that’s not relevant for valuation. It doesn’t necessarily make it backwards. Sometimes, people think if book value is a noisy measure of true value, then it has to not work at all. No, you add noise to a process, you make it worse, but you don’t change the sign. It suddenly doesn’t become a suicide strategy. But if it is a worse version of value, what are you going to see? You’re going to see it over the long term, not do as well, which is absolutely what we see, it’s the fifth of five. But if you have a period where value of the concept, not a particular measure, if people truly are freaking out and selling anything that’s not a darling and pouring all their money into the darlings, the better you measure that, the worse you’re going to do.
Value Works And Random Doesn’t
And that sounds very self-serving. I am actually saying that we’ve done worse than price to book alone, because we’re just that much darn better than it, which is a tough sell. But it’s also why I think we’ve done better. Our valuation indicators have been better than price to book over the long term. Whether it’s self-serving or not, I think it’s pretty clean. Imagine price to book was a terrible measure of value, not backwards. Again, you didn’t automatically lose money, but it was a random number. It really didn’t pick up the value effect at all. What would it look like? It would be the worst value measure long term because value works, and random doesn’t. Throwing darts doesn’t outperform. It would be the best in a horrendous value period because it’s not capturing value. That is too extreme.
I think price to book does capture a fair amount of value and I would not– and Fama-French is trade-off against turnover, they could even be right. But I do think it is somewhat ironic that price to book being worse is actually a better thing when you’re in a– everyone hates the real fundamental concept of value, it’s better to measure it poorly. And all the haters would have preferred to be in price to book for the last couple years.
Tobias: I like it. There are several great points in that– [crosstalk]
Cliff: Tough times, Toby. You have to find your joys wherever you can.
Tobias: It’s tough, I agree. As a pure value guy, it’s a very rough run, but I draw some inspiration from reading the AQR papers telling me why it’s not that bad. So, I really appreciate you guys publishing as frequently as you do.
Cliff: No, we say it is that bad, it’s just going to get better.
Tobias: That will work for me too. Hope is all I need. You talk about intangibles– and in this paper, there’s a talk about intangibles, and correcting for intangibles should lead to better returns. Dealing with share repurchases, there’s no greater outperformance for companies that do high share– or they’re not worse at being assessed using these value measures.
The Reason That Value Hasn’t Worked Recently Is Not Because It’s Too Well Known
The two that I really wanted to turn to, just very quickly. One is that the value strategy is too naive and too well known, too easy to implement and that is the reason why it hasn’t worked. And the rejoinder that you make in the paper is that the spread is very wide. And the one that I find particularly interesting is that, the accrual anomaly which worked for a while and became very popular and maybe got too much capital invested towards it, and sort of famously went away. And that’s often the thing that folks talk about that you can oversaturate these strategies with too much capital. But the accrual anomaly has come back.
Cliff: It has, and we never abandoned it. We never gave the accrual anomaly tremendous weight, but we didn’t lower it when it was unpopular, and we stuck with a modest weight to it. Single factors, irrespective of all our stories and our explanations, they don’t have super high Sharpe ratios. And Sharpe is not the be-all, end-all measure. Again, the world is fat tail, I use it as a same way everyone else does, is a way to compare. If you have a particular fat tail strategy say in option one, you never want to use Sharpe. But to get to an overall process balancing many factors, we think you can get to a north of 0.5 Sharpe uncorrelated to markets. That’s awesome, if you can do it. That’s a Sharpe above stock market returns uncorrelated. It’s like having a second stock market. It still goes through horrible periods and that’s for the whole process.
So individual factors can’t have– they have to have lower Sharpes after trading costs, after everything else, after implementation. So, something like accruals, if it’s a 0.2 Sharpe, don’t hold me to the numbers, I’m making it up. 0.2 Sharpe, square root of 10 years is roughly 3, it takes about a 0.7 negative STD event to have that be flat or negative for decade if it’s as good as you think it is, and you’d want at least a small amount of this in what you do. Which is why we keep an open mind, we always try to investigate, are we wrong? Has something changed? But we try very hard not to be the knee-jerk people to say, “Well, it hasn’t worked for a while. It must be done.”
Now, the accrual anomaly is harder than value in that– I don’t know any one stop– let’s look at it to see if it’s been arbitraged away. But you got it exactly right and I referred to it before. We worry about value being arbitraged away. But that’s not what we’re seeing. We’re seeing a world where no one on earth seems to give a darn about value. I am fond of saying if value has been arbitraged away, someone forgot to tell the prices. I also occasionally paraphrase Yogi Berra, who famously said about a restaurant, “That restaurant is so crowded, no one goes there anymore.” People are saying value is too crowded, but nobody goes there anymore. It is a worry.
The nice news is you probably get a consolation prize if value gets arbitraged away. The last few years are glorious, because what is arbitraged away? Someone buying where you’re overweight and selling where you’re underweight, narrowing that spread, and that accrues to your benefit. The bad news is at the end of that period, you have to be paying attention and have some way to say, “Well, this isn’t going to work no more.”
I have a running joke. And again, I did this two and a half years ago when I was really talking more about what if this ever gets arbitraged away? Not today as why is it ever going to work again? But I used to look at the youngest AQR person in the room. And whoever that was, I’d be like, “Well, imagine this gets arbitraged away over the next 10 years.” We do better than our normal kind of act s strong because we have this wind at our backs.
The world mistakenly thinks I’m better than I am because credit and blame and I’ve seen both– me and the more general AQR, they think we’re better than we are. I don’t know if I’m retiring in 10 years, but I’m making that up. I’ll be 63, it’s a little early. But I retire, turn it over to the woman to my left, to the youngest person in the room and say, “The good news is you’re in charge. The bad news is, I used up all the return forever.” Again, we do other things that wouldn’t all be bad news, but it would be at least mildly bad news for her going forward as the new head of the firm to have no value spread. And people would mistakenly say, “Boy, we used to do well when Cliff was here.” It has nothing to do with me.
Value Investing Because It Is Too Well Known
That could happen one day, but people saying that value is too well known to work, I think of all the explanations for why value might have problems in today’s world that is by far the weakest, because I think we have a very clear indicator of what that would look like. And we’re seeing dramatically– not just dramatically, hundredth percentile dramatically, new hundredth percentile dramatically the opposite. I don’t get how you have a world where so many people do value, they’ve arbitraged it away, so the prices are radically more different than normal without a fundamental justification we can find. So, yeah, I’m pretty dismissive of that one.
***
Do Fundamentals Still Matter?
Tobias: That’s a good segue into– one of the questions that they ask is, do fundamentals still matter? They have a very interesting approach to assessing that– create this look– it’s completely cheating, which they say in the paper. The point is that you take the forward earnings estimates for a year in advance. So, for 2020, we get the 2021 estimates of 2022, and we’re able to trade on that basis. And then you can look at– of course, that generates an incredible Sharpe ratio. It’s a great strategy.
Cliff: It’s really good.
Tobias: If I could get those forward estimates, I’d be very happy. But you’re not able to do that, of course, in the real world, but it shows that there are periods of time where even with those forward estimates, you still can’t generate good returns and the two that really stand out, ‘98, ‘99 and 2019 and 2020. So, do fundamentals still matter?
Cliff: Not so much currently. When I say I think ‘99, 2000 was a bubble and I do think at least cross-sectionally among stocks, I’m willing to use that word now. I am making a statement that again– I don’t want to go irrational versus not perfectly rational, do fundamentals not matter? I think they matter considerably less than they normally do or than they should. I won’t say that the expensive companies aren’t better companies. I include this in my latest blog, one of the final tests I do is looking at some of the quality measures the literature looks at. The most famous one, I think, being Robert Novy-Marx’s gross profitability to assets.
There’s no magic to that. I think, again, just like value, there are other ways to measure it. But I chose something very obvious that cheap companies on average have lower margins. World is not perfectly rational, but it’s not totally insane. You’re normally giving up—so, if you’re pure value investor– and a Graham and Dodd style value investor can look at more things than just price, so they can account for this. But if you’re only trying to buy low price to fundamental assets, you are going to be buying worse companies almost always.
The very brief period at the peak of the tech bubble, where you had maybe value investor nirvana, the spreads were not quite as wide as today but you were actually able to buy companies without giving up any quality. For the few on the call who remember that as adults, there were a lot of crazy companies, by no means do I say the expected companies today aren’t great companies that deserve to be expensive. I say they’ve gone too far, which is very different. In the tech bubble, there were a fair amount of companies that were gigantically expensive, that weren’t even the better companies.
But the simple result today, if we don’t have that– though we do have even wider spreads, but we don’t have the opposite either. Another thing that might give me pause as a value investor is cheap versus expensive, well, you generally give up some quality. If you normally gave up 10% on gross profitability of a portfolio of cheap against a portfolio of expensive, the most you ever gave up was 18%.
The tech bubble, you were giving up zero percent. Yay, you’re almost there. If today you were giving up 40% and it looked like an unprecedented drop and then it stayed there, you would really have to step back and say, maybe at the end of the day, we still believe in value. Maybe we think the world believes this is prominent and we don’t, but now you have to make tougher calls. Now you have to say this is something dramatically different than we’ve seen in the past, just not seeing it.
Again, so many of my stories, I know they start out sounding like they’re going to be interesting and the conclusion is always, and we don’t see that. But that is the point. We’re trying to investigate with an open mind every possible story. If we do find it, we will report it. But we repeatedly keep finding, that’s not it. I won’t tell you buying cheap stocks are suddenly the nirvana of the cheapest ever, and they’re actually just better companies, they are worse companies. There’s a price for everything. The price for the companies is way lower and way higher on the expensive ones than history and that is not because of any larger fundamental difference that we can discover.
And to be humble doesn’t mean it doesn’t exist and we’ve just failed to find it. We live in a world not of arbitrages but bets we think are smart, the same world you live in. We think when you’ve investigated everything that you can think of that anyone else can think of, that anything you’ve heard and not found the answer, and you’re facing wider spreads than ever in history, Occam’s razor says that’s probably a pretty good bet. Occam’s razors never approve. It says the simplest explanation is usually best.
So, I do try to stay humble, that’s why we keep looking. Is there something else out there? If so, I’ll admit, I’ll be disappointed. It would be hard for me to have to go tell the world, “Uh, we found it, this is why we–” [crosstalk] But I would do it, I would absolutely do it because anything else wouldn’t be ethical.
***
Correlation Between Value Investing And Interest Rates
Tobias: Last question, and it’s in relation to the value in REITs paper. This is a prevailing narrative that one of the reasons that value is not working is because rates have been very low, unnaturally low, perhaps crushed to sort of virtually zero and somehow that has created the conditions for which value doesn’t work. Your colleagues, Maloney and Moskowitz, have– and I find the duration argument theoretically pretty compelling that value stocks tend to be shorter duration and growth stocks are longer duration, so when the interest rates drop, it impacts the growth stocks more and that has created the scenario that we see now. They have gone through that in any implementation that you can possibly think of and they couldn’t find anything that sort of indicated that was the case.
Cliff: Yeah. This is in a series of unsatisfying conclusions where all I end up with is saying, “No, that’s not it.” We certainly and still find it to be a fairly intuitive story, particularly when long-term real rates are lower. Any kind of distant cash flows should be worth more. And I don’t like coining growth versus value, I prefer cheap versus expensive. But there is something to the idea of the expensive are normally stocks with longer projected tails, future growth.
My colleagues point out in the paper, it’s often not that simple. Sometimes, when you have a blow to future growth from low rates, it’s not clear whose growth is affected by more, so the actual numbers can change. But by and large, the real thing is just empirics. They step back and they look at value returns in correlations to interest rates, either alone or adjusting for many other things in the marginal correlation to interest rates, and it varies by specification, but they don’t find much. I keep planning, I haven’t got there. Like you’ve seen before, I often take my colleagues’ great detailed paper and do a brute-force simple blog inspired by it, where I both try to introduce the paper and do it in maybe a little bit more of an accessible way.
I have looked at my value spreads that we’ve been talking about. This is related to returns but not quite the same because remember, you can have eight years of bad returns and not see a big cheapening if it’s fundamentally, so it’s not the same as returns. And I find a little bit more than negative. We don’t force false agreement at AQR. We don’t have a lot of people who think quant is the worst thing in the world and you should buy one stock, you can’t have complete disagreement.
But we share with people when we have different people at the firm who have at least nuanced views. And my research shows statistically, this is dodgier. It’s what’s called levels on levels. It’s harder to be confident in the statistics. You only need an infinite amount of data to asymptote to accurate statistics. It might feel like we’ve been an infinite drawdown, but it is not quite there yet. But I do find at least as a point estimate that interest rates explain why value spreads, a little bit more than my colleagues work. And they don’t literally look at that. So, it’s not a contradiction.
But the spirit that interest rates might explain things– and this is one I want to write up, I find it to be a little bit stronger. But then when I do the exercise that you must do next. All right, what are the spreads look like through time if we adjust for interest rates? What would the spreads be over and above low interest rates or under and below high-interest rates would imply? You find it is still well past the prior historical maximum. So, these are the least satisfying explanations in finance. When someone says, “Yeah, there’s some truth to that, but it’s not nearly enough.” No one walks away happy.
The people who believed in it kind of still believe in it, they grumble, test it– [crosstalk] The people who didn’t believe in it or mad that you conceded that there’s something there, but I do give a little credence to the story, at least in the spreads version, I do find some support for the story. If you want to tell irrational bubble stories, some support that the world then takes way too far is a fairly consistent story. There’s some economics here, people are not totally crazy, but they’ve gone nine times past what low-interest rates would imply. You could also– again, it’s a bubbly inefficient market story, but it may not be literally the… part which is just, can be rational pricing of cash flows, but the speculative environment of such cheap money.
For our story to be true, you have to get to a point of why are people paying crazy—or in our view, crazy prices for the good companies and crazy low prices for the less good companies. So, there’s nothing be some irrationality somewhere.
One simple story, and this is very hard to prove, these are just conjectures, is eight years of irrational loss to value has led to a two-plus year blowoff, get me the hell out of this stuff, that has been an irrational loss to value. Another version is the growth stocks have always had a bit more of a whiff of speculation to them than value stocks. And if we have a world of free money, even if it’s not literally the interest rates that justify it, might that be part of the bubble type environment, absolutely. Not approved, then there may be other explanations.
But, yeah, we think low-interest rates of all of them is probably the one with the most hope before looking at it, because there are inherently reasonable aspects to the story. Again, my colleagues tell you why it’s not quite as simple as you think but we all get the intuition of that. And it’s not nearly enough. If low interest rates are justifying 8% of a wider value spread, the 92% that’s left over still has no story to it, still seems to be a rather extreme mispricing that’s going on today.
Tobias: Well, that’s all we have time for today, Cliff. Thanks so much for–
Cliff: Well, is it okay If afterwards I just keep going?
Tobias: Absolutely. I’ll turn off the recorder and you can tell me what you really think.
Cliff: I meant alone, sometimes– no, I’m kidding.
Tobias: [chuckles] Incredibly grateful for the time that you’ve spent going through these papers. I realize that that’s– getting into the weeds is– it’s rare to get the guy at the top of the firm to get right down into the weeds, so I very much appreciate that.
Cliff: I live in the weeds.
Tobias: Very much appreciate the time. Cliff Asness, AQR, thank you very much, sir.
Cliff: Well, these were great questions and you’ve read all the stuff, which I always appreciate. So, thank you. I enjoyed it.
For all the latest news and podcasts, join our free newsletter here.
Don’t forget to check out our FREE Large Cap 1000 – Stock Screener, here at The Acquirer’s Multiple:
