Today we welcome Rob Arnott to the show! Rob is the founder of Research Affiliates and is a prolific writer who has published hundreds of articles for many different journals. We know firsthand, the power of Rob’s work, and how it can alter the way you think about investing, and this depth of knowledge, coupled with his ability to make complex topics understandable makes him a dream guest for us! Rob is the co-author of The Fundamental Index, and we get some insight into this subject along with many other groundbreaking areas he has worked on.
We cannot stress enough the rarity of Rob’s gift for getting difficult ideas across in a deliberate and approachable way, and this is apparent through this illuminating conversation. For us, it was quite surreal to speak to someone so influential, and listeners can expect to come away with a greater understanding of ‘smart-beta’, intangible assets, forecasting, and some insight into the interesting areas of earnings dilution and ‘the big market delusion’, before Rob shares some very surprising information on factor momentum at the end of our chat. So for this and a whole lot more, in a truly stand-out episode, be sure to listen in!
Key Points From This Episode:
Rob, there are tremendous this amounts of assets in cap weighted index funds today. And we talked on this podcast a lot about how they’re pretty good investments for a lot of people. Can you talk about some of the drawbacks of cap weighted indexing?
Absolutely. Well, firstly cap weighting is a perfectly good answer, you own the market. For of its good features and bad features, the turnover’s nice and low, the fees are negligible, and so you capture the return of the market, which is great. But you also chase every fad and bubble that comes along. You also underweight every stock that is trading at apocalypse lows, and you also miss out on rebalancing alpha.
We all think it makes sense to rebalance with our asset allocation of stocks with sword, trim your stocks, top up a little bit on the bonds and yet doing it within the equity portfolio doesn’t happen with indexing. Cap weighting’s biggest Achilles’ heel is that any stock that’s overpriced, by which I mean a stock that is priced above what the future truly has in store for it, something you can’t know, and that is destined to under perform is above its fair value weight in the portfolio.
And every stock that’s below its fair value weight is destined to out perform and vice versa. So the implication is you overweight the overvalued, underweight the undervalued. This criticism was leveled at index construction way back in 1957, when S&P launched the S&P 500. It was easily refuted by the index providers by saying, “Yeah, but that’s absolutely true, but you can’t tell me which is which.” And that is accurate, but if you overweight the overvalued and underweight the undervalued, most of your portfolio will underperform, too little will out perform. So you have a performance drag.
Now, if you break the link with price, so that you’re weighting the portfolio using equal weighting or fundamental index or DARTs, anything that breaks the link with price, an overvalued stock might be overweight or might be underweight, an undervalued stock might be underweight, the errors cancel. Jack Trainer wrote a paper in 2005 showing that if the market is not perfectly efficient, and if there’s a mean reverting error, that is the market sets a price that’s too high and that mean reverts or too low.
And that mean reverts as the market’s constantly seeking out fair value that your alpha relative to the market will be equal to the variance of the mean reverting error in price. And that’s important because if the markets let’s say 90% efficient, if you have 40% issue-specific variability in price, that’s 16% variance. If 90% of that is efficient is correct movements based on new news. That means you have 160 basis point drag from waiting proportional to price from cap waiting.
And that drag disappears if you break the link with price. And this is actually showed up in the live experience of FAFI strategies measured relative to cap weighted value indexes. Why value indexes, because RAFI has a value tilt roughly pari passu with the value indexes.
So RAFI being Research Affiliates Fundamental Indexing, which you’ve built a business on. Can you compare that to tilting towards the traditional Fama–French factors like size, value and profitability?
Sure, absolutely. Cliff Asness has suggested that smart beta is factor tilting and vice versa. I would respectfully disagree. The origin of the term smart beta was Towers Watson coined the expression in 2007 based on RAFI. RAFI was the inspiration for the term smart beta. And what they thought was smart about it was that it created a mechanism for country trading against the market’s most extreme bets for capturing a rebalancing alpha.
And they correctly noted that anything that breaks the link with price, be it minimum variance, equal weighting, DARTs should add the same alpha. And so they said smart beta is anything that breaks the link with price. Okay. Factors, does cap weighted quality break the link with price? Of course not. Does equal weighted quality break the link with price? Even there no, because quality stocks are going to be priced at a premium.
Does momentum break the link with price? Heavens no. So I would argue that the original definition of smart beta is not the same as factor investing. It’s strategies that break the link with price. Now, fundamental index itself has a stark value tilt. Why? You’ve got these stocks trading at lofty multiples. RAFI says, “Thanks for those gains. Yes, the outlook for the company is wonderful, it’s a great company, no clouds on the horizon. Sky’s the limit,” but that’s in the price, the price is already reflecting all of that.
So it’s forward-looking return expectation is a market return unless it exceeds those lofty growth expectations. So why don’t I reweight it down to its economic footprint? There are value stocks trading really cheap and RAFI will say, “Thanks for the discount. Now I get it, the company’s got headwinds, it’s got some major problems. The leadership is in apt.” It’s already in the share price. And because it’s already in the share price, the only way it’ll underperform is if its business does worse than the market thinks.
So why don’t I just go ahead and top it up to its economic footprint? So RAFI will bring down growth, bring up value, creating a start value tilt. It just so happens that that value tilt almost matches the average value tilt of the cap weighted value portfolios. Why? Because I mean, we’re including all the growth stocks. It’s because the cap weighted value portfolios throw out all the growth stocks, but then they cap weight the value of stocks. So they’re going to overweight the overvalued value stocks, underweight the undervalued value stocks.
And RAFI will simply say, “The deeper the value is, the bigger are overweight.” So we have a value tilt roughly pari passu with the value factor. Now, what we’ve found in the last 17 years since we went live is that RAFI has managed to beat the value indexes by roughly one and a half percent, sometimes 2% depending on the geography. Above 2% in emerging markets, one and a half to 2% in U.S. and international, above 2% in small companies.
So the value add has been very much in line with what Jack Trainer would have predicted in his 2005 article examining why fundamental index and other valuation in different strategies should add value. We also have a quality tilt, quality stocks traded a premium multiple. So we’re down weighting the most of the time. Crummy stocks are usually trading at a discount, so we’re reweighting them up. So we do have a quality tilt against quality.
We have a size tilt in favor of small cap, but it’s not a very powerful tilt. Think of it this way, the cap weighted market has a small company tilt, the RAFI index relative to the cap weighted market has a small cap tilt. What do I mean by that? There are probably 50 companies with bigger sales than Tesla that are smaller in market cap than Tesla. They might even be 100. And so cap weighting by pushing more money into Tesla has a small company tilt, RAFI by pushing Tesla down out of the top 50 has a small cap tilt because it’s pushing up smaller cap names.
So interestingly, the size tilt is a very special size tilt. It is an exact mirror image of the market’s small company tilt
Talk a little bit more about fundamental indexing? You talked about economic footprint, you talked about breaking the link with price. When you build the fundamental index, what are the weights?
Well, basically generation one fundamental index. The FTSE RAFI indices launched back in 2005 are based on four metrics. How big is the company based on its sales? Let’s say ExxonMobil is 3% of all publicly traded companies sales, just as a hypothetical example. Cashflow let’s say ExxonMobil is 2% of all in the market today. Book value, let’s say they have a book value that’s 3% of the total. Dividends, let’s say their dividends are 2% of all dividends paid out.
You could argue endlessly, is it 3% of the economy or 2% of the economy? Or you could just average the four and say it’s two and a half percent. That’s what I’m going to hold. Now the beauty of this is not getting the weight right. The two and a half percent isn’t that important. It’s having a steady anchor to concentrate against the market’s shifting expectations. So if ExxonMobil arises and its fundamental economic footprint doesn’t, you’re going to say, “Thanks for the gain.” I’m turning you back.
If it tumbles and the fundamentals don’t, you get to say, “Thanks for the discount. I’m going to top it back up.” And so you’re concentrating against the market’s most extreme bets. What’s interesting is that RAFI, the portfolio, you look at the top 10 or top 20 holdings, it looks like a market portfolio with a distinct value tilt. You look at the trades, the stocks that it’s buying are going to be the ones that have fallen the most in the last year relative to their fundamental economic footprint.
Those are the doggiest of the dogs. Those are the companies where you look at the buy list and you think, why on earth would I buy these? You look at the sale list and they’re the most beloved extravagant high flyers. And you’d say, “Why do I want to trim these?” And so the flavors of fundamental index, the generation one winds up with a stark value tilt, a slight small cap tilt an anti quality tilt and it profits from essentially none of those. It profits from rebalancing.
Yeah. Interesting. Very interesting. Okay. Now we have a fundamental indexing is one of the many, many quantitative strategies that have come into the marketplace in the last… I’ve been in this industry for the last eight years. And even since then, it’s been a pretty crazy proliferation. How much weight when an investor is sitting down and looking at a back test of a strategy, how much weight do you think they should put on the back test when they’re deciding, yes, I’m going to invest in this thing.
Very little. I have never seen more money move from one strategy to another based on back tests. I could create a back test with extravagant performance if I set my mind to creating a back test with extravagant performance. That’s not going to do anything other than allow you to earn massive profits if you can back up the clock to the beginning of the simulation. Can’t do that and past is not prologue.
So if you look at the back tests and put your money in the strategies with the best back test, you might be putting your money in good strategies. You might more likely be putting money into heavily data mined strategies where the strategy was to maximize the back test. And worse than that, the historical performance may be based on those styles of companies coming more and more into favor, pushing the price up.
If the relative valuation soar, your historical performance will look great, and if there’s any meaner version, your future performance will be dreadful. So one of the most interesting tests we did, we did a series of papers called Alison factor land back in 2016 and ’17. And one of the most interesting things we did was to pose the question. Let’s say, you want to do active factor investing. Let’s say you take eight factors.
And we chose all of the usual suspects and a couple of extras thrown in. Let’s say you choose to use just the three factors with the best trailing ten-year performance. Now, firstly benchmark, how did an equal weighting of the eight factors work? It worked beautifully. It gave you about a 3% alpha over the course of 50 years. But of course it did because we chose them based on factors that have historically worked. That’s data mining.
Now, you take the strategies with the best trailing ten-year performance and use just the three with the best performance over the last 10 years and keep reassessing once a year. Now, if you do that, your alpha drops in half. Let’s say you do the opposite, you pick the three factors with the worst performance. Nobody’s going to do that. Your alpha doubles, 6%.
So what you’re doing is selecting on strategies that have become cheaper. And because the eight were themselves selected based on 50-year past performance, they all work. So you’re anchoring on ones that, with the blessings of hindsight you know worked, you don’t know that they will work in the future, but they all worked. And then you’re secondarily selecting on the ones that you know work, emphasizing the ones that are out of favor, and so your alpha doubles.
So if you use back tests to choose your strategies, you are self-selecting to be in the ones that have outperformed their own expectations and are poised to disappoint.
Fascinating to think about. It’s like you would maybe not choose value right now because it doesn’t look so good.
That’s exactly right.
Let me as my question. Ben has eight years in, I have almost 30 years in this industry, so I’ve been through a number of ugly periods for the value of factor included in the most recent one. So perhaps you can talk to how normal extreme factor drawdowns are in investing, particularly in factor investing.
It’s really interesting, people have been attacking value because of the drawdown and basically saying, “It doesn’t work anymore.” In point of fact, we had a paper in the first quarter edition of The Financial Analyst Journal entitled Reports of Value’s Death Have Been Greatly Exaggerated. What we did is we looked at the narratives for why value has failed and will remain useless.
And we find that each of these arguments is somewhere between flawed and deeply flawed. The argument that is most powerful is, well, value is usually defined by price to book value and it’s a terrible measure. Yes, it’s a terrible measure. It ignores all intangible assets of a company. If I buy a desk for 1,000 bucks for a staffer, that goes straight onto the balance sheet as an asset. If I invest a million dollars in R&D it doesn’t, it was an expense.
Now, what if like the desk, you assume I didn’t spend a million without expecting to get it back, so let’s add it to the balance sheet, add R&D to the balance sheet, and then just like the desk, amortize it out. Because if it hasn’t paid for itself within 10 years, it was a stupid investment or a stupid purchase of a desk. Now, if you view things that way, you find that price to intangibles adjusted book is about twice as powerful in its cumulative efficacy as ordinary book value.
You wind up with Fama-French value, you wind up four times as wealthy after 60 years as with Fama-French growth. Adjust for intangibles, you wind up eight times as wealthy.
But in point of fact, value has been attacked because of this drawdown. Every factor has some heroic drawdowns, value is getting the attention because the drawdown has been now and has been protracted. If you use Fama-French value, get reached to peak in relative performance in first half of 2007, and it’s been downhill ever since. So you wind up with a 13 and a half year drawdown.
If you use price earnings ratios, your all time peak of relative performance was 2014, so it’s a seven-year drawdown. If you use price to sales, it was 2017, so you’ve got a four and a half year drawdown. If you use RAFI weight to cap weight as a composite measure of value tacitly, that’s the same as looking at relative price to cashflow, relative price to earnings, price to book value, price to dividends and price to sales averaging those four relative valuation metrics.
That’s what comes when you take fundamental weight to cap weight, that measure of value peaked in early 2018 and had a three and a half year drawdown. So bottom line is, no matter how you cut it, values had a dreadful run the last four years. If you made the mistake of anchoring on book value, it’s been a 13 and a half year drawdown. But it’s not the biggest drawdown any factor has seen. Low beta has had drawdowns just as big, momentum has had drawdowns that are even bigger.
And take it all the way back to the Great Depression, momentum had an 80% drawdown.
Oh my goodness. Quality has yet to have a severe drawdown, but quality was first published in 2013. And by the way, hasn’t worked since 2013. So it was it published as a result of a data mining exercise or was it published because it’s a great idea? Jury’s out on that. So it is normal for factors to have massive drawdowns. This is one of the appeals of multi-factor and it’s one of the reasons that multi-factor has become popular, although it is ferociously over sold and over-hyped by advocates of multi-factor.
The narrative is these factors all work. They’ve all been vetted by academia and proven to be effective. No, they’ve been proven to be effective in back tests. These factors all work at different times. Yes. True. Therefore, combining means you’re going to smooth your ride. You’ll have less jarring drawdowns, less extravagant wins. Yes, but the drawdowns are not diminished as much as you might hope because the drawdowns are often correlated.
The cliche on Wall Street is that the only thing going up in a market crash is correlation. So that’s a worrisome component. And the final part of the sales pitch is, these all work historically over long periods of time, they work at different times. So if you blend them, you have near certainty of winning if you’re patient over a three to five-year span. No, you don’t have near certainty, you have odds modestly in your favor. So correctly positioned, multifactor is a very useful concept that’s egregiously overhyped.
That’s an important takeaway, because I think because I think that factor diversification idea is definitely heavily sold, especially for dealing with bad outcomes like the ones that you were just talking about.
Right. It’s a legitimate argument, but let’s not soft pedal the downside risk, which can be large when correlations spike in a factor crash. And let’s not over-hype the average expected return or the likelihood of success because there you’re just extrapolating the past and the past was cherry-picked. You didn’t use any factors that didn’t work historically. It doesn’t guarantee that they will work in the future.
I want to come back to intangibles. Dimensional had a paper out I think late last year, maybe it’s early this year, I can’t remember. But they showed that if you are using profitability and value together, that intangibles don’t matter quite as much. You’re saying that intangibles are super important for value, so what’s the answer? How important are intangibles?
Well, intangibles are very important for price to book value measures of value. If you’re using price to book to define value, you’re including intangibles if there’s been an acquisition. Good will is intangibles. You’re not including it if it was organically grown internally in the business. And so you got this weird mishmash of including some intangibles and not others. If you look at the Russell 3000 today and at add in R&D as an intangible amortize out over 10 years, what you find is that intangible book value is as large as tangible.
So the adjusted book value is about twice as large and good will from acquisitions captures about half of that. So you’ve got a big gap and you’re misspecifying things badly. DFAS paper is actually absolutely spot on. If you use profitability as co-metric alongside of value, it mitigates the damage done by excluding intangibles. But why wouldn’t you want to fix the book value one so that the combination of value and profitability is at least modestly better?
We’ve done that within RAFI. We’ve modified the RAFI strategies that we run and the ROE strategy that we run for PIMCO, all include intangibles. Now with RAFI with four components, sales profits, book value, dividends, and buybacks, only one of the four is affected. And so the impact is modest in back tests with that caveat, the effectiveness of RAFI improves 10 to 15 basis points. But why not do that? Why would you just leave that on the table if you know that there’s a way to improve what you’re doing?
So I would agree with DFA that profitability is the best compliment to price to book on unadjusted for intangibles. But I would disagree that that’s a reason not to include intangibles.
They’d have a better product and a better process if they did.
Let’s jump back to value stocks just for a very quick question. Is it different this time for value stocks?
Well, certainly it is because we’ve never had a drawdown this long or this deep, even the tech bubble, people think of the tech bubble as the decade of the ’90s. Now, it was really ’97 to 2000, it was three years. And so value underperformed growth by 4,000 basis points using the Fama-French methodology, using price to book. It got it back in the space of 14 months after the low, it got the whole thing back. So the snapbacks can be profound, but it was a horrible drawdown, a little under three years.
And this one, no matter how you cut it is at least three and a half. And by some measures 13 years, and it’s deeper, the value underperformed growth by 40% in the tech bubble and by 58% using price to book from 2007 to 2020. But here’s what’s interesting, the value effect has worked just fine the last dozen years. Now, that sounds like a radical preposterous statement. Here’s what I mean, value has gotten cheaper relative to growth by a larger magnitude than its performance drawdown.
Let’s say you have a stock that fell 58% like value relative to growth. Let’s say the stock’s price to book value fell 68%. That means the book value is risen. Are you going to look at that and say, “Get me out of here, I can’t stand the pain”? Many investors would say that. But if they looked at the drop in price to book value, they might say, “I can’t believe how cheap this is. It’s the cheapest ever. It’s cheaper than it was during the tech bubble. Let me at least rebalance in. And if I can possibly persuade my clients to do so, let me over rebalance and overweight value at this stage.”
Now the problem with overweighting value when it is cheap is that you will never get the turn right. So you will look like an idiot and you will feel like an idiot until the turn happens, and that might be a long wait. If when you’re wrong and it continues to underperform, you rebalance again and top it up again. What you do is you assure that you have peak exposure at the turn. Now, that is enormously important. It goes totally against human nature, but that’s the best way to add value with mean reversion strategies, with concentrating against the market’s biggest bets, which are sometimes crazy bets.
So you’re an incredible communicator on this, Rob. From a retail investor perspective, how important is it to have a story that gets back to what is really going on with a company like the distress story or these reasonable arguments to make as opposed to academic factor description? Is that story important to stick through times of under-performance?
Yes, if it’s a story that encourages you to buy low and sell high. No, if it’s a story that encourages you to chase fads, and so stories are dangerous. Tesla’s sales in 2021 are probably going to be up 20, 30% over 2020. That’s a great story. Does it justify a price to sales ratio that’s roughly 100 times that of the big three? Wow. So be careful of stories. They’ll lead you into bubbles.
One of the things that I’m proud of is that we came up with a workable definition for the term bubble back in 2018 that can be used in real time. Everybody talks about bubbles and band is around the term without any definition. And you look back and say, “Yep, that was a Japan bubble in 1990. Yeah, that was a tech bubble in 2000. Yep, that was an oil bubble in 1980. Yup, this is a problem, EV bubble, but I don’t know.” Well, we do know if you have a rigorous definition.
Let’s define a bubble as something where if you use a valuation model like discounted cashflow, you would have to make implausible, highly unlikely growth assumptions to justify the current price. And as a quality check on your analysis there, the marginal buyer cares nothing for valuation models. Okay? Apple, Microsoft, are these bubbles. No. You can look at those stocks and you can use a discounted cashflow model and make aggressive assumptions, but not implausible.
And yes, there are marginal buyers who use valuation models. Tesla, AMC, GameStop, Netflix, how many buyers of these stocks are using valuation models? None. How extravagant to your growth assumptions have to be to justify the price? Extravagant to the point of implausibility. Tesla, if you assume 50% annualized growth for the next 10 years, 50, then 50, then 50, then 50 compounding. You get to 55 times today.
Amazon grew at 26% a year. That’s enough to give it 11 times the size that it was 10 years ago. Now that is extravagant growth. That is phenomenal. But you’re assuming at 50% growth that Tesla will have five times the ten-year growth of Amazon. Let’s assume that at the end of the 10 years, it has the highest profit margin of any major automaker in any of the last 10 years. Well, the peak profits in any year in the last 10 years was a little over 10% after tax.
If you do that and you use BWA bond yields to discount back to today with no risk premium for equities, you wind up with a valuation of $430 a share. Now it’s 600. So that definition in a bubble can help you identify bubbles in real time. This doesn’t mean your short sell bubbles. Bubbles can go longer and further than any skeptic could possibly imagine. Zimbabwe stock market in the summer of 2008 as they were entering into the late stages of their hyper inflation, the first six weeks of the summer, the currency fell tenfold.
People in Zimbabwe plowed their money into the stock market to get away from Zim dollar-dominated anything. And the stock market rose 500 fold. Not 500%, 500 fold-
Wow.
… in dollar terms, 50 fold. So you could have at the start of the summer said, “This is going to zero, I don’t want to own these stocks. In fact, I’m going to put a 2% short position on.” Six weeks later, you would have been wiped out. Now, by the end of the summer, the currency had fallen another 100 fold and the Zimbabwe stock market essentially went to zero and stopped trading. So you would have been right, but bankrupt.
So be very careful about bubbles. They can go a lot further than you expect, but you don’t have to own them. Just avoid them. Anti-bubble stocks, trading at discounts so deep that you’d have to make implausibly bad assumptions to not earn a risk premium. Those you can own. The downside risk is not 50 fold, the downside risk is 100%, so a modest, long position is totally fine.
You had a paper with Brad Cornell who was a guest on this podcast. I think he was actually our most recent guest at the time that we’re doing this recording. You guys had a paper on the big market delusion. Can you describe what that is and how it plays into bubble formation?
Sure. I’m indebted to Brad for coining the expression, big market delusion. But anytime you have a massively disruptive technology or a massively disruptive reshaping creation of a new enterprise or a new industry or disruption of an existing industry, what you often have and by the way, disruption is normal. There’s always something disruptive. That’s the nature of the evolution of a healthy capitalist economy, but disruptors get disrupted.
And often, this happened with railroads, it happened with telephone, it happened with radio and TV, it happened with autos. It happened with computers. You get into this state where there are a handful of companies that are seen as the dominant disruptors. They’re all priced as if they’re all going to win, but they compete against one another, so they won’t all win. Worse than that, disruptors can get disrupted. Palm was worth more based on the PalmPilot than General Motors.
Wow.
And three years later, the Blackberry just blew them out of the water.
Blackberry was priced extravagantly and five years later, the iPhone blew them out of the water. So disruptors get disrupted, people forget that. In our paper on big market delusion, we focus the spotlight on electric vehicles. The electric vehicle makers in 2020 had sales that were approximately 2% of the auto industry. That 2% was priced at 45% of the industry’s market cap. The rest of the industry, the conventional auto makers produced 98% of the revenues and commanded 55% of the market cap.
Net-net, the EV makers were priced at roughly 100 times as much per dollar of revenue as the conventional makers. Now where this gets really interesting is are these guys going to be disrupted? Yeah, maybe by the conventional automakers. Something that practically nobody has noticed is that more electric vehicles are produced by these conventional automakers than by the EV specialists.
The EVs specialists have less than 50% market share in EV. So is that a bubble? Yes, for sure. Is it a big market delusion? Yes, for sure. You had Nicola was priced at multi-billion dollar market cap before it had $1 of sales. Nevermind profits, it’s not clear it will ever have profits, but before it had a single dollar of sales, that means its price to sales ratio was infinity. Okay. Well, that’s a little weird.
And also it’s interesting to note that in terms of price to sales ratio, Tesla, after rising 600% last year, who was the second cheapest of the EV makers on price to sales why because –
That’s unbelievable.
… second cheapest. Price to sales ratio was I think 30 to one.
Because they have sales.
Yeah, because they had sales.
You got an older paper. I think you wrote it around the time of the last technology bubble on earnings dilution. When I read this paper, it completely blew my mind, so thank you for the paper. But can you describe what earnings dilution is and how it can relate to a bubble formation?
Sure. This was a paper written with Bill Bernstein back in 2003 and we entitled it The Two Percent Dilution. What does that mean? In a capitalist system, companies make profits. If the economy grows 10 fold over the course of two or three decades, aggregate earnings will have probably grown tenfold. So does that mean that you’ve got that tenfold growth in earnings in the stocks that you buy in an index fund at the start of that period? No.
Growth of the economy consists of new enterprise creation plus growth of existing enterprises. If there’s any new enterprise creation, then by definition, earnings growth of existing enterprises must be slower than macro economic growth. What did we find? We found that earnings growth during the 20th century, we used the Dimson Triumph of the Optimists database to test over a full century, the 20th century. And we found that GDP grew roughly twice as fast as the earnings of indexes of stocks.
Now, that was 2% hence the 2% dilution. So what did we find? We found that, think about it this way. 2% dilution from new enterprise creation means that the stocks that you buy today, if you don’t trade them, if you don’t move into the disruptor, if you don’t trade them after 35 years, you’re going to own half of the stock market, the other half will be newcomers. Well, look at the market cap today. What percentage of the Russell 3000 is companies that didn’t exist 30 years ago?
Well, over 50%. So you would have enjoyed economic growth, a little less than economic growth, about 2% a year, plus the dividend yield, plus inflation, a very, very nice rate of return and a rise in valuation models to boot. Wonderful rate of return, but less than most people would expect. I still hear senior, highly respected academics talk about how earnings obviously with the macro economy.
Collectively they do, per share, they don’t. And this is a very important nuance. What it means is that your dollar can’t be invested in existing companies and new companies at same time. Sure, you can carve out some of your existing investment and put money in venture capital. But now you’ve got to pockets of money, one capturing new enterprise creation, one capturing existing enterprise growth. You can’t capture the sum of the two. You can’t do that, not without using leverage. So therein lies just a fundamental misunderstanding of how the stock market works.
Wow. It really is a head spinner. Just to keep this simple, do you think investing in revolutionary or these disruptive industries is a good idea?
Often no. And I say often no for the very simple reason that you get big market delusion. If you were a buyer of railroad stocks in the early days of railroads, the extreme early days, the 1830s when they weren’t practicable, they were just experiments, yeah, you made some good money. If you waited until the 1840s when people started building tracks between cities and trains, no, you are overpaying because those were the social media companies of the day. Those were the extravagantly expensive companies of the day.
RCA in 1929 was priced at something like 100 times earnings and never after 1929 achieved those price levels, let alone valuation multiple levels again. So you bought at the all time high of the company. So you have to be very careful, the disruptors are very disruptive, but here’s how they disrupt. They don’t disrupt by picking the pocket of the marketplace and putting it in their pockets and into their shareholders pockets.
They disrupt the market by creating something inexpensive that benefits society more than it benefits their own shareholders. And it only benefits those shareholders if the benefits to the shareholders exceed the markets, probably lofty expectations. But there are periods of time when growth is fantastic. ’97 to 2000, 2014 to 2020, you didn’t want to be out of those stocks. And if you were underweight them, we were, it was a brutally painful experience. It doesn’t mean that you were investing in an unwise fashion, it just means that you weren’t chasing the popular and beloved companies.
And one of the things we find is that top dogs, the companies that are at the top in market cap over long periods of time underperform by pretty astounding margins.
That’s an important point that you touched on there. So a thread that we’ve been talking about through this disruption discussion is that the disruptors get to the top and there’s a bit of a winner take all situation in many cases. You mentioned railroads, but we’ve mentioned Tesla. Those companies, once they’ve gotten to the top, don’t tend to produce good returns, this is what you’re saying.
Right. They produce great growth. Cisco fell, I think 88% in the aftermath of the tech bubble. During the period of time the price fell 88%, the company sales rose 40%.
Wow.
So they were growing handily, they were growing beautifully, but they’d been priced at 180 times earnings. And so they fell to something like 30 or 40 times earnings and the earnings were now materially higher, but the drop in valuation multiple swamped the growth of the business. And that happens again and again. Do I think Tesla is going to have great sales growth over the next five years? Yes, I do. In fact, I think an EV friendly administration almost guarantees that because we’re going to see EV subsidies again most likely.
Do I think that’ll turn into appreciation of the share price? No, I don’t. Do I believe that strongly enough to put a big short on Tesla? No, I don’t. Because again, even though I think it’s a bubble stock, could it go up five fold from here? Of course, it could.
Right. All right. We’ve touched on value, we’ve touched on growth or high priced stocks. How do you go about forecasting the expected returns of a factor portfolio?
We actually have an interactive website that I would urge your viewers to take a look at. If you go to researchaffiliates.com and look in the upper right corner of the page, you’ll see Smart Beta Interactive, which is an interactive website that looks at factors like classic Fama-French value or a blended aggregate measure value, long value, short growth, and then ask the question, how expensive is this today relative to the past?
Looks at past relationships and finds, oh, when this factor is trading really cheap, the subsequent alpha is both much more reliable and much larger. And out of that comes up with a forecast of forward-looking returns. So if you look at that website right now, it’s saying, “Quality? Maybe not.” It’s trading at a premium to historic norms, expected alpha is negative. Low volatility, low beta, it’s trading at a large premium to historic norms. Maybe not. It’s projected alpha is negative.
Momentum two months ago was saying, “Chase the FANG stocks,” and was trading at all time record valuations, stay away. Now that those FANG stocks have faltered quite a bit, momentum is no longer saying, “Chase the FANG stocks.” It’s saying, “Chase these basket of companies that’s a mishmash and it’s priced in line with historic norms. So maybe now you can turn momentum back on.” Value was last August trading cheaper than any time in history, including the peak of the tech bubble.
Wow.
The spread in valuation between growth and value was normally five to one on a price to book value basis. In 2007, it was four to one, meaning the value was trading rich. Had we had those measures back then, we might’ve said, “Maybe soft pedal value is not likely to pay off very big.” Peak of the tech bubble, it was 10 to one, in the COVID aftermath, it got to 12 and a half to one, 25% richer than at the peak of the tech bubble.
So with that kind of extravagant discount, basically the narrative was we’re going to see sweeping bankruptcies because of the COVID lockdowns and the companies affected by that are certainly not going to be the FANGs, it’ll be the value stocks. They’re going to see sweeping bankruptcies, so they’re not nearly as attractively priced as you think they are, because stocks always look cheap on their way to zero.
All right, then you get the hints of a vaccine coming, hints of a slowdown in COVID cases and all of a sudden value starts to make a comeback for the very simple reason that people say, “Oh, maybe these bankruptcies aren’t going to happen.” This stock was priced at a deep discount because it’s allows the company and an even deeper discount because it might go bust. Well, if it might not go bust, it’s only worth this lesser discount. It snaps back.
And so we see that again and again that value is most reliable, not in bear markets, but in the first two years aftermath of the bear market, the early stages of the renewed bull market. And we’re seeing that again right now, it’s powerful.
So do you think it’s possible to add value by timing exposure to these factors?
The short answer is yes. The longer more nuanced answer is, as always be careful about making too big a bet. When we wrote, How Can “Smart Beta” Go Horribly Wrong? Cliff Asness and others attacked us and basically set up a straw man and knocked it down. Said, “These guys are saying you should time factors and put your money in whatever is cheapest.” And if you do that, you’re taking on a lot of residual risk tracking or relative to the market and it’s not worth it.
Well, it’s absolutely correct, you make bets that are scaled. If a factor is trading a little cheap or a little rich relative to its historic norms, make a negligible bet. If it’s trading at all time record cheapness, we have the confluence of value being cheap all over the world, especially in emerging markets, and emerging markets being roughly half the Shiller PE ratio relative to the U.S. So half off for EM, biggest discounts ever for EM value.
I have a little over half of my liquid net worth in EM value by way of a fundamental index-based emerging market strategy.
Wow.
That means that I have half of my net worth and other stuff, so I don’t put all eggs in one basket, no matter how attractive the basket.
It sounds extreme when you say half is in EM, but you rightly point out that the other half is another stuff.
Well, it’s certainly not as extreme as Elon Musk having 95% of his net worth in one stock.
That is true. Not even comparable in terms of how extreme those bets are.
Right.
I want to come to momentum to finish our conversation here. You’ve done some excellent papers on momentum, including factor momentum, which is just fascinating. Momentum looks unbelievable on the data, live funds. You have a paper looking at mutual funds, live funds have not looked so great. Why do you think that is?
There’s several reasons for that. Momentum is the most powerful factor in historical back tests, was first published in ’93. From 1999 to date, standard momentum trailing year excluding the latest month, standard momentum has had a negative return since ’99. That’s 22 years since it last worked. Or more accurately, it had a momentum crash in 2002, when the market turned. It had another momentum crash in ’09. It had an intervening bull market where momentum worked beautifully from ’03 to ’08, but the crash in ’09 wiped out 15 years of alpha and it struggled back and then had some other faltering, including a sort of a mini crash in the last couple of months.
So the crashes have outweighed the bull markets and net-net, momentum hasn’t worked in 22 years. So that’s a starter. Momentum investors will say, “We don’t use standard momentum.” Okay. How have their strategies worked since they created their variant of momentum? Typically, not very well. There are exceptions though. The second problem with momentum is that the turnover is humongous. So you’re trading costs, unless you can trade dirt cheap, buying stocks that are on a rocket ship without paying a trading cost and pushing the price up higher with your own purchase. Unless you can do that, trading costs will eat your alpha alive.
And third problem is a much subtler. One momentum is a monthly strategy typically. And if you look at the monthly returns, they chain out to be very powerful. But let’s say you buy a momentum portfolio and shorten anti-momentum portfolio today. And then you just hold those portfolios for the next three years, so you don’t trade them. What happens? Your high momentum stocks beat your low momentum stocks handily. And then peter out after about six months, and then roll over, and then give all the alpha back and then some turning negative about a year and a half into it.
And that’s even using data from the periods of time when momentum was brilliant. So the implication of that is with momentum more than any other factor, if you don’t have your sell discipline nailed down carefully, you’re going to give all the alphabet. It rolls over. Value starts modestly and then accelerates. It starts modestly because it’s anti-momentum and it’s held back by momentum. And then again, speed, quality has a trajectory that’s long-lasting, size has a trajectory that’s long lasting.
Momentum doesn’t, rolls over and dies. And so momentum can be very useful. I think the most compelling way we use momentum that anyone can use is not to create trades where you’re going to have to justify your trading costs and probably won’t or coupon. But to block trades, if you’ve got a stock that’s on a rocket ship and your valuation model says, “Get rid of this.” What’s the harm in waiting three months and seeing if it’s losing momentum?
If it’s in free fall, every value trap that’s on its way to zero will look cheap tick by tick all the way down. So do you want to buy it and buy it again, and buy it again, and buy it again and then see it at zero? No. If it’s in free fall, what’s the harm in waiting three months and saying, “Is it still in free fall?” If it’s not in free fall, maybe I’ll start buying. If it’s recovering, maybe I’ll start buying more aggressively.
So you can combine momentum with other factors by using momentum to block trades. Now, what’s the trading cost associated with that? None. In fact, it’s better than none, you’re not doing some trades. So you have negative trading cost impact. To me, that’s a give me, that’s a natural way to use momentum that almost assuredly is not going to hurt you.
Probably the most powerful result we had in our paper, Can Momentum Investing Be Saved? It was a deliberately provocative title, because the natural assumption is, it doesn’t need saving. It works. Yes, it does need saving. One of the results that we found was if you take mutual fund returns and regress them on their momentum exposure, how big is their momentum factor beta? How much does the portfolio look like a momentum portfolio?
The correlation of factor loading, low anti-momentum, high momentum, and the return is negative. So you find that the higher momentum strategies had lower returns than the anti-momentum strategies. I think the reason for that is, one, momentum has not worked very well since 1999, two, momentum rolls over and dies, and if you’re running a momentum strategy, chances are you’re going to wind up holding the stock way too.
And three, the trading custody you’re alive. So net of those three problems, can momentum investing be saved? Yeah. But got to be really judicious about you do it.
So there is a chance that it could make sense as a standalone strategy
Used aggressively with the extreme outliers. What you find is that the deciles of return don’t have a linear pattern to return. They have a linear pattern which falls off a cliff in the bottom decile and spikes high in the top decile. Concentrate in those extreme deciles and make sure you trade fast. So once it loses momentum, ditch it. Don’t hang on for the, give it all back phase and use those signals across everything you do to slow down trading in areas where the model is wanting you to buy stocks that are in a free fall or sell stocks that are soaring.
You mentioned trading costs earlier, so I know a lot of listeners might be thinking, well, you can trade for free, can’t you? But I assume you’re not talking about commission costs?
Commissions are near zero or are zero.
So there’s other costs, so does the other costs also make you decide whether it should be a large portfolio or a small portfolio that goes with the momentum strategy?
Well, if you’re using momentum to proactively trade, to incur trading costs, it better not be a large portfolio or you’re going to wipe out your alpha. If you’re using it to block trades, there’s no limit. Now, you can trade on Robinhood and Ameritrade and Schwab and other platforms for zero commission. We’ve all heard the cliché, if a product is free, you are the product. People will pay to know what you’re trying to trade and they’ll pay because that knowledge is more valuable than the trade itself.
So the real cost of trading is the price that you transact at relative to the price that would have prevailed if you weren’t trading, if you weren’t there. That’s unmeasurable, but it can be estimated and it’s assuredly non zero. If you’re buying, you will always pay at least the price that would have prevailed in the absence of your wanting to buy, always. And if you’re selling the price will always be at most the price that would have existed in the absence of your wanting to sell.
So viewed from that perspective, trading costs always take a toll. The other part of trading costs is what if the trade runs away? I can guarantee you that trades that run away, that you don’t get made are the ones that you really wanted to have made. And the trades that happened easily and the trade went away, the price went away the wrong direction are the trades you really didn’t want to have done. So you’re assuredly going to get done the trades that are the least valuable to you. You will not get done some of the trades that are the most valuable to you.
That also is a difficult to quantify cost of trading. Net-net, the cost of trading has to go up the larger your trades, because the price that would have prevailed in the absence of a $10,000 trade is probably within a penny of the price that you paid. If you traded a billion dollars, the price that would have prevailed in the absence of your billion dollars of demand will be a lot lower than the price you pay.
So those prices go up with higher turnover, with reliance on less liquid stocks. Momentum strategies look fabulous on paper, but much of the alpha was earned in stocks in the bottom 2% by market liquidity. Ouch, you couldn’t trade those. So there goes that part of the alpha. So you have to be very careful. I’m a believer in multi-factor investing. I just think that it’s been appallingly oversold.
Rob, we’re just talking about individual stock momentum, which is what everybody knows about. And here’s about some more recent research, including another one of your papers that blew my mind was on factor momentum. From that research, can you tell us, does momentum primarily exist in individual stocks like we tend to think about, or is it in sectors or is it in factors?
I would have said it exists in all three until we did the research. And in fairness, the leader of this research, Juhani Linnainmaa senior professor at Dartmouth is the originator of a lot of this research. But what he found was that factors have momentum. Factors have momentum right from the get go. You don’t have to wait a month like you do with individual stocks. If a factor has been profitable last month, it’ll probably profitable this month. And it doesn’t matter which factor you look at.
We looked at over 50 factors and we found that momentum worked in essentially all of them and was extremely powerful in some of them. So factor momentum is important. Now, if you take factor momentum and neutralize it by sector, so that you’re taking the factor loadings within sectors and asking, okay, how much is left? The vast majority of the factor momentum remains intact. If you take sector momentum and remove the factor component, nothing’s left.
So sector momentum is apparently entirely a function of factor momentum. Individual stock momentum is apparently almost entirely a function of factor momentum.
There’s a little bit left, but most notably there’s the tendency for the last month to mean revert. So the bid-ask bounce or the one month reversal effect, no matter how you characterize it is very real and is independent of factor momentum. But the longer term stock momentum appears to be almost entirely a product of factor momentum. So this has been fun research. It’s hard to get published because it goes against what a lot of people believe.
But we have it in at one of the top finance journals and just got back referee reports and they have some wonderful suggestions on ways to make the article a little tighter and a little more robust. I’m cautiously optimistic that it’ll be accepted once we make those changes. But there are lots of people into the factor community who just don’t like the idea that factor momentum could actually subsume stock momentum and sector momentum.
That’s a mind blowing finding. What’s the story behind factor momentum? Like the behavioral story for individual stock momentum, the behavioral story is fairly well known, I think. Is it the same kind of thing for factor momentum?
Here’s how I think it works. Let’s back up to 1960-ish when efficient market hypothesis was first proposed. It was proposed in conjunction with the notion of an equity risk premium. Now, the risk premium is supposed to be a reward for risk bearing. Let’s assume that’s wrong. Let’s assume it’s a reward for willingness to accept fear. There are some investments that are scary. Those investments, human nature or behavioral finance would say, those ought to have a bigger risk premium.
There are some investments where the marginal investor is terrified of missing out, so the fear goes the opposite way. The fear is, I’m afraid of not owning this stock. So a fear premium, unlike a risk premium can go negative. A fear premium unlike a risk premium doesn’t have to be correlated with an objective measure of volatility or a beta. Cap Em formalized this and said, “The risks that ought to carry a risk premium is beta.” And I would agree with that.
I think it was a seminal finding, one of the most important in the history of finance. But the key words are ought to, the risk premium ought to be proportional to beta. In fact, is proportional to fear. Now, if you have a fear premium, do you expect a value effect? Of course. The value stocks are troubled companies. They’re crummy, they got lousy management, lousy product, I’m afraid of owning them.
The growth stocks, fear of missing out, wonderful products, wonderful management. These are the companies I really want to own. So if anything, the fear premium, disappears or even goes negative. And if it goes negative, your expected return is lower than T-Bills. Would you expect a size effect? Sure. Little companies are not as well followed, not as well understood. You want a little more return to buy these companies than big well-established well-known firms?
What about momentum? Well, momentum is an interesting example. The more a stock is soaring, the more fear of missing out comes along, pushing down the fear premium and pushing up the price. And then it rolls over at some stage when you feel I’ve got my position and are they delivering the goods? Are they delivering on the reason I bought? Oh, maybe they’re not really going as extravagantly well as I’d hoped. All of a sudden fear premium starts to creep back in and it rolls over.
So most of the anomalies of finance would have been predicted if finance had labeled it a fear premium instead of a risk premium. This is what I think is the driver of momentum. And it’s not surprising that momentum is most powerful in factor land, not sectors, not individual companies, because factor capture broad attributes. And it’s those attributes that dissipate fear or that engender fear.
Can that factor momentum be used to improve portfolios?
Well, for starters, you’d want to pay attention to sectors, not from the vantage point of momentum, but from the vantage point of controlling tracking error relative to the market. Because the risk that matters most is not volatility for sure. People can make 50% lose 50%. And if their neighbors have lost 60% and they lost 50, they’re feeling miserable, but less miserable. So relative performance to your peer group actually matters more than any other form of risk in terms of human nature.
And so if you’re looking at factor momentum, you can use it to shape your bets. For instance, our dynamic multi-factor strategy made a big swing in the first quarter towards value. When was value at its cheapest? Last summer, but it was in free fall. So wait three months. Okay, the momentum seems to be turning. Let’s nibble away at it. Three months later, momentum has decisively turned. Wow, look at this momentum. And boy, it’s still really cheap, so jump in with both feet.
And so by using a blend of momentum with relative valuation of your factors, you can actually do very nicely with dynamic factor allocation, but don’t make too big bets. Our multifactor, I don’t mimic multifactor. Five factors, normal allocation is 20% each. We’ll go as low as five, we won’t kick it out. We’ll go as high as 35, which means tacitly, the most concentration you could have would be betting on three out of five factors. And so we spread our bets out gently emphasizing the ones that are cheap.
One of the things I’m proud of is the great majority of our strategies. Once you adjust for the factor or style bets that we’re making, for example, looking for alpha net or Fama-French factors, the vast majority of our strategies have had cumulative alpha that is material net of all trading costs, net of all fees. And there’s very few organizations that can claim that.
Rob, you mentioned being excited about the factor momentum research. Is there anything else that maybe I haven’t read yet? That’s not out yet that you’re excited about?
Oh, I would just say stay tuned. We’re working on a lot of different things and several of them look pretty exciting.
I’m excited to see them when they come out.
And our last question, Rob as always, how do you define success in your life?
Success to me is enjoying what I do and helping my family and friends enjoy what they do. Now, from a business perspective, I won’t enjoy what I’m doing if I’m not helping our clients to do well, so the end customer has to win. In terms of family, I want my family to be happy and comfortable. I want my family to every last one of them be doing something they love and something that makes the world a better place.