 Printed and electronic copies are for personal use. Any unauthorized distribution by fax, email or any other means is prohibited and is in violation of copyright. If you are interested in redistribution, reprints or a subscription, please contact us at subscriptions@indexuniverse.com or 212.579.5833.
|
||||||
| The Fundamental Debate |
| Thursday, 01 January 2009 00:00 |
 Robert Arnott VS.  Paul Kaplan Recently, a debate was staged at the University Club in Chicago, Ill., between Robert Arnott—chairman and founder of Research Affiliates and one of the leading developers of noncap-weighted indexes—and Paul Kaplan, vice president of quantitative research at Morningstar, to get to the bottom of some of the biggest issues surrounding fundamentally weighted indexes, including:
Larry Siegel, director of research, Ford Foundation (Moderator): Is fundamentally weighted indexing simply a value strategy or tilt, rather than a new way of thinking about capital market equilibrium? Tom Philips of Malbec Partners has pointed out that fundamentally weighted indexing is mathematically equivalent to a market-cap index, where that index has been reweighted to give a heavier weight to low price-to-earnings, price-to-book, price-to-sales, price-to-employees and so forth. How do you respond, Rob? Robert Arnott, chairman and founder, Research Affiliates (Arnott): It would be very difficult to disagree with a mathematical truism. The Fundamental Index idea is a very simple idea. It involves weighting companies in a portfolio in direct proportion to the economic footprint of the company, measured by fundamental metrics of company size such as profits, sales, book value, dividends, or in our case, a blend of those. If you used, for example, a profits-weighted index, then by definition, weighting companies according to their profits is mathematically equivalent to weighting companies by market cap divided by their relative PE ratio. So the mathematical equivalence is undeniable. I've said in the past that the Fundamental Index idea is a value strategy of a very specific sort, and it's one that suggests a change in our frame of reference, from a cap-weighted-centric view of the world, to a company-centric view of the world—one that weights companies according to how big and prosperous they are today, not based on how big they might be in the future. Moderator: Paul, I'd like you to respond to Rob's comment about my first question. Paul Kaplan, vice president of quantitative research, Morningstar (Kaplan): I think Rob and I are in complete agreement on this. This is a mathematical truism. It's a value-tilted portfolio. And what's interesting about it is—unlike traditional value indexes—it includes every stock in the market portfolio. It tilts towards the ones that have higher yields. Moderator: Would you agree that fundamentally weighted indexing is in some sense a better approach to value indexing than previously constructed value index funds? Kaplan: Different indexes are designed for different purposes. So, for example, at Morningstar, we've created value, core and growth indexes which divide the market into three segments. The first purpose of these indexes is to really monitor the market … to show how value did versus growth over the past month, quarter, year or five years. Another purpose of that type of index is diversification. If you own an active manager in growth, but he doesn't have any active ideas for value, you could use a pure-play value index. The Fundamental Index approach gives you an overall value tilt in a diversified portfolio. It is an intriguing idea because it does include all of the stocks. The shortcoming that I see with it are that the weights are based upon variables like book value, earnings, dividends and so on, and there's no accounting for how much risk is in each stock. And that seems to run somewhat contrary to the usual kind of portfolio construction principles that we adhere to—that two stocks might have, say, the same expected future earnings growth, but we would want to put less weight on the one that's riskier. Moderator: When I've tried to explain fundamentally weighted indexing to people who simply haven't heard of it, I've said, "There are two ways that you can earn a better return than the cap-weighted market. One is that you can look at information that other people don't look at, and that's what most security analysts try to do. Of course, there are so many security analysts that it's very hard to beat them. The other way—which seems to be a little easier—is to not look at information that everybody else is looking at. That information, in the case of fundamentally weighted indexing, is the price." Does this statement characterize your strategy fairly, Rob? And second, is there information in the price that investors should pay attention to that the strategy does not? Arnott: I think the short answer to your question is yes. You've characterized the essence of the Fundamental Index approach, which is to strip price out of the weighting formulation. The reason to do that is that the price contains an error. We don't know what the error is. But we do know that every stock that's trading above its eventual fair value is weighted in the cap-weighted portfolio above that eventual fair value weight, and every company that's below its fair value is below its fair value weight. If you can just randomize those errors, that does add value. But to your second part of your question, is there anything in the price that investors should consider that is not considered in the Fundamental Index approach? I would say your question answers itself. If there's information that's in the price, how do you profit from it? It's in the price! Moderator: Paul Kaplan asks, "Doesn't fundamentally weighted indexing assert that all stocks should have the same PE, where E is a proxy for earnings or dividends—whatever you're dividing price by—while market-cap-weighted indexing is an argument that the different PEs that are put on the various stocks by the market are correct." What evidence do you have, Rob, that all stocks having the same PE is a better approximation of the economic values of the companies than all market PEs being correct? Arnott: I have no evidence that all stocks being at the same PE ratio is a better representation of fair value for companies than the market price … indeed, the evidence runs to the contrary. I've done some new work on a notion that we refer to as "clairvoyant value," in which you take historical prices and you ask, "What were the subsequent cash flows from that investment actually worth?" I found something interesting—unsurprising, but startling in its magnitude—that companies trading at premium multiples have subsequent growth relative to companies trading at discounted multiples, which is highly correlated with those multiples. That excess clairvoyant value is correlated with the ex-ante pricing with a 50 to 60 percent correlation; it's awesome. So the market is doing a brilliant job at gauging which companies deserve premium valuation multiples. What we also found is that a fair spread between growth and value stocks is about 40 percent for the one-sigma outliers, while the actual average spread is about 80 percent; so the market pays up for growth about two-to-one relative to what it should pay up. Moderator: So you're saying the market can forecast which companies are going to grow but then assigns too high a multiple for those companies? Arnott: Exactly. The way the Fundamental Index approach adds value is categorically not to have a better representation of fair value. Categorically, it is to have a sensible and intuitive anchor for rebalancing. Most of us in the investment world believe that rebalancing is an important part of our tool kit. Why do we have rebalancing stop at the borders of the stock market? Why do we want to have 31 percent in technology at the peak of the bubble in 2000, rather than having a rational anchoring mechanism for rebalancing and contratrading against the excessive bets, in either direction, that the market will make from time to time? Moderator: Paul, do you think that there's an aspect of fundamentally weighted indexing that hasn't been discussed that causes it to have some flaw? I ask this because if the market can distinguish between companies that are going to have future high growth rates, and companies that are not, then there is information in the price, which ought not to be completely ignored. Kaplan: Well, I think that suggests that the market price should not be ignored if the market actually does a good job of identifying companies that have future growth. Arnott: I think a lot of the controversy about the Fundamental Index idea is that so many people have trouble shifting their frame of reference from owning stocks, and thinking of the market-cap-weighted way of owning companies, and thinking of the company scale as the anchor. Ultimately if a company is at a premium multiple, one of two things must happen. Either the [corporate fundamentals] must eventually rise to show that that premium valuation multiple is fair. Or the price must fall to correct that error. And so, if we grant that there is that mean reversion tendency, then ultimately the one scenario in which the Fundamental Index idea would be expected to under-perform is a scenario in which the market is paying too little premium for the growth that it anticipates. That is to say that the market discerns growth and value but has the opposite of hubris—inadequate confidence in its ability to discern opportunity—and pays too little a premium for growth relative to value. In that scenario, the Fundamental Index approach would under-perform. Kaplan: I think what you're describing is the principle of value investing: that the market prices are not equal to fair values, but over time the differences are eliminated. This is the principle of value investing that goes back at least to Graham and Dodd. So you're enunciating principles with which we've all been very familiar in this industry for many decades. And so I view what you're doing in fundamentally weighted indexing as simply an automated passive approach to … Arnott: I'm shocked, shocked that you used the word passive! But go on. Kaplan: … passive in the sense that you can do it with a computer. You don't need an analyst to execute the strategy. Arnott: I have never claimed to invent fundamental investing. I have never claimed to invent value investing. What I have, I think, done is to come up with a new method of constructing a portfolio. You can call it indexing. We do. You can call it an active strategy. Others do. But it is a portfolio that simply mirrors the look of the economy. And what we've shown, through vast empirical evidence, is that historically that captures a lot of return that is left on the table by cap weighting, with its chasing of bubbles, crashes, fads and shifting expectations. It's not the value tilt, on average over time, that creates the incremental return, it's the contratrading against the market's constantly shifting bets. Kaplan: I'm not sure I see a distinction. But, I guess, one thing I would point out is that the idea of using a measure of the economic footprint rather than market capitalization in constructing your index … I believe the credit should go back several decades to MSCI, for they created the GDP-weighted index. You all remember back in the late '80s when on a market-capitalization basis Japan had this huge weight which everyone thought was ridiculous. So the way it was corrected, or addressed, was to use GDP weighting. Arnott: I would agree that we weren't the first to use fundamentally derived weights; absolutely right. Moderator: GDP weights are a fairly close antecedent. I remember thinking that MSCI might lose a lot of business because it was so easy to beat the MSCI EAFE benchmark just by holding a below-benchmark weight in Japan during the '90s. But, in fact, during the '80s, the situation was the opposite: It was almost impossible to beat EAFE because Japan was the only country that was going up. So reweighting the index to a more fundamentally oriented set of weights did not necessarily produce a better index or a worse index, it was just a different index. It outperformed tremendously in the '90s, having underperformed in the 1980s. And we might expect something similar to happen to a fundamentally weighted index consisting only of U.S. stocks if you have a bubble and then a crash. Kaplan: Which is exactly what happens if you look at the back history of a fundamentally weighted index through the bubble of the late '90s into 2000. When the bubble was rising, the fundamentally weighted index did very poorly relative to the market. But once the market correction came, the fundamentally weighted index was in all of the right places for the ensuing period in which value beat growth. Arnott: One of the neat things about the Fundamental Index concept is that the magnitude of its bets varies with the magnitude of the bubble. For example, take the GDP weight—and I would absolutely agree that that has the same intellectual underpinnings of the Fundamental Index idea, and the same source of alpha. GDP weighting, when Japan started its bubble, was a neutral weight. But by the time the Japan market reached its peak, with GDP weighting, you had a maximum bet against Japan just when the bet was going to be most profitable. That's wonderful. It doesn't mean that you're going to win every quarter or every year. It does mean that the size of your bets will correspond to the magnitude of the markets' extreme bets. And the reciprocal happens on deep value. It's that rebalancing that's the principal profit mechanism, not the value tilt per se. Moderator: In 1954, when the United States finally surpassed the 1929 peak in the stock market, a group of wise men, including John Kenneth Galbraith, advised President Eisenhower to prepare for another crash and another Great Depression. What the market was telling us, however, was that the United States was emerging as the dominant economy in the world, and the 1954 price which exceeded the 1929 high on the Dow Jones was actually a tremendous bargain. We all wish we, or our grandfathers, had gotten in at that price. The Japanese bubble could have been the same phenomenon. It could have been actually a bargain. At the time, it was hard to figure out. In retrospect, it's easy to figure out because we know what happened later. How does fundamentally weighted indexing reflect the possibility that you might be wrong about when something is in a bubble or overpriced? Arnott: Suppose we could take the net present value of all future cash flows on every company and calculate it back to today and identify a current fair value based on the ex-post realized cash flows. Now, suppose I have those numbers and nobody else does. And suppose I come to all of you in the audience and say, "Here's the cap-weighted portfolio and here's the true-fair-value-weighted portfolio. I can't tell you if the true-fair-value-weighted portfolio is going to do better or worse today, tomorrow, this year. But I can tell you that the true-fair-value-weighted portfolio is correct, exactly the fair-value weightings. How many people in this room would choose the cap-weighted portfolio? Right; nobody would. Not a single person. If we have a cap-weighted portfolio, we know most of our money is in companies that are above fair value, whether they're high or medium or low valuation multiples. We know that they're going to revert toward fair value over time. And so we know that most of our money is in assets that are going to under-perform and too little is in assets that are going to outperform. The fascinating question is if you can construct a portfolio which has a clairvoyant value weight, plus or minus random noise. Mathematically it will have the clairvoyant value portfolio return, minus a tiny drag from the noise, but not minus a systematic drag for overweighting the overvalued parts and underweighting the undervalued. We went back over the last 50 years and looked at company size and took the correlation of the gap between clairvoyant value, the subsequent and realized value, and share price—i.e., the error in that initial price—and posed the question, "Is that correlated to the company size?" The short answer was no. The average correlation was zero, which means that, over the last 50 years, the fundamentally weighted portfolio would have had almost exactly the same return as that clairvoyant value portfolio that we all agreed would be the better choice. So really it's our randomizing of the errors, instead of systematically overweighting the overvalued and underweighting the undervalued with cap weighting, that's the source of the value added. Kaplan: With respect to price, what you're calling a randomized error is not, in fact, a randomized error; it is in fact the fair price multiple. Moderator: Please develop that idea, Paul. We haven't all read and carefully studied your paper. As I understand your paper, it's a very fully developed argument that what Arnott and [Jason] Hsu call a randomized error is actually correlated to the market so that it's not random at all. This point is the key to understanding why your position differs from Rob's; otherwise, it sounds like you're sort of both marching towards the same goal with a 0.8 correlation. You're not. Kaplan: The point is, how do these weights that are created from fundamental variables differ from those created from fair values? It's just a simple mathematical truism that if I take the ratio of one to the other, I get the fair market multiple, which is a component of how the market is going to be priced. And therefore, this so-called random error term is not a random error term at all. It is, in fact, reflected in the market capitalization. Arnott: That's exactly right. Kaplan: But Jason says in his paper that it's independent of market capitalization. So there's an internal contradiction that I'm seeing in the logic here. Arnott: There is not. You have price and you have fair value. The difference between the two cannot have zero correlation with both. In Jason's paper—and in our paper with Harry Markowitz—we make the assumption that the error is random, and uncorrelated with value. That means it must be negatively correlated with price. Now, classic efficient markets theory says the error is uncorrelated with price. And basically we reject that. We say that ultimately the two need to converge, and ultimately the anchor for value is the eventual future fundamental scale of the company. And because that's the anchor—and this is verified empirically on the clairvoyant value paper—we find that the error is utterly random relative to company scale, and it is empirically negatively correlated with price, which gets back to your point that value works; something that's been known for a long, long time. Empirically it has worked in countless markets around the world over long periods of time, albeit with dry spells, with periods that are disappointing. But you would expect that in a world in which price equals fair value plus or minus a truly random noise term. Moderator: Behavioral finance teaches that investors make systematic mistakes because they don't react rationally to information. Is it accurate to say that fundamentally weighted indexing relies on behavioral finance? Arnott: I would be hesitant to endorse the notion that it relies on behavioral finance. The two are joined at the hip, though. The presumption inherent in the Fundamental Index idea is that prices are wrong, and that eventually, the market identifies and corrects these errors. Therefore an anchor for rebalancing, even if it's the wrong anchor, winds up winning in the very long run. And so tacitly it does not rely upon a behavioral finance model for the world, but on the notion that prices are not exactly right at all times. That is a core principle of the Fundamental Index as a concept. Moderator: Paul, does behavioral finance make sense to you, and if so, is it enough for investors to be irrational or to have knowledge deficits or information processing deficits to get fundamentally weighted indexing to be a worthwhile strategy? Or would you do something else with the behavioral insight? Kaplan: Well, I think behavioral finance as it exists today is largely just a critique of classical finance, but it doesn't really present an alternative. The idea that the market does not price stocks correctly is the foundation of all active management in value strategies and in growth strategies. Since fundamentally weighted indexation is a value strategy, the idea is not something that's unique to fundamentally weighted indexing. Arnott: When people say "value investing," they just mean a preference for a low PB and shun the growth stocks, like a classic Russell 1000 Value Index. Well, the Russell 1000 Value Index has a value tilt. It's about twice as much value tilt as the fundamental index. But it also cap-weights. So systematically it's going to overweight the overvalued value stocks. The Fundamental Index approach—because of its very specific type of value tilt—winds up with half as much value tilt as the value index, adding twice as much incremental return, which is really cool. Kaplan: Larry, I'd like to ask Rob a question ... What I see as a real controversy around fundamentally weighted indexing is the claims that are associated with it rather than the strategy itself. Jeremy Siegel published an article on the editorial page of The Wall Street Journal four days before they launched the product in which he made a claim that fundamentally weighted indexation represented a big revolution. What he said was that basically, all classical finance is all completely wrong. It needs to be replaced by this new "noisy market hypothesis." So the market-cap-weighted portfolio disappears. The default portfolio becomes the fundamentally weighted portfolio. Rob, do you have a very different view than Jeremy has on that? Is that his view, and not yours? Arnott: Our views are somewhat similar. But I'm certainly not going to reject the notion that there's been a certain amount of overselling of the Fundamental Index concept. If I've ever said anything factually incorrect, or materially exaggerating reality, I'd love to have somebody point it out to me. I don't think I ever have. And so the fact that some folks oversell and misrepresent, I have no comments about that. Yes, it's happened. Audience Member: It was mentioned that the market price does a good job of identifying the company that is either growing faster or slower but it just exaggerates that effect in the price. Would there be a way that you could introduce something reasonably simplistic into a Fundamental Index approach that would, in fact, have some nonprice-related measure of reasonably expected future growth? Arnott: Paul has, in a sense, done that with the collar strategy, which, I think, is a neat idea. We, at Research Affiliates, license the Fundamental Index approach through FTSE to anyone in the world who wants to license it. We don't run passive Fundamental Index money. By agreement with FTSE, we aren't going to compete with the licensees. We run enhanced Fundamental Index strategies and the enhancements are along the lines that you described. One of the enhancements that we use is quality of earnings; companies with poor quality of earnings that have a history of using aggressive accounting practices are smaller than they seem. And companies with conservative accounting standards are larger than they seem, and so we make that adjustment, and it's very profitable. So it's an illustrative example. Kaplan: Consider two companies with the same earnings: a fast-growing company and a slow-growing company. Let's say you used the DCF model to assign weights. Both companies have the same earnings, which you divide by the cost of capital minus a company-specific growth rate. You end up with different weights. That's the kind of risk and growth adjustment that is not done in fundamentally weighted indexing as Rob has defined it. But certainly you'd think it would be a sensible thing to do because, again, your basic lesson of portfolio theory is that if two stocks have the same earnings, but they differ in risk or prospective growth, you don't hold them at the same weight. Arnott: But you have to identify the high-growth versus the low-growth company in a fashion that is not in the price. Just because the high-growth company is at a high multiple and has high future growth, and the low-growth company has negative growth, that doesn't mean it's not in the price. In fact, your optimal portfolio might be sharply overweight the slow-growth company and underweight the fast-growth company. Kaplan: Yes, but you're not using the market price. You would not use the market price. You would use a model of intrinsic value. Moderator: If enough people fundamentally index, wouldn't a fundamentally weighted index converge to the cap-weighted-market-index return by mathematical certainty? Arnott: Absolutely right. I hope it happens, and not just for the selfish reasons that you might infer. I hope it happens in the sense that if the world shifts from cap weighting to fundamental weighting, then valuation dispersion will gradually narrow and the two portfolios will become the same. What that means is that the early adopters will enjoy enormous wealth accumulation. When the alpha goes away, it doesn't have to go negative. It can just go away. Moderator: Is there a downside to everybody adopting fundamentally weighted indexing, making all prices fair? Arnott: That won't happen.
Permalink | © Copyright 2008 Index Publications LLC. All rights reserved
More on this topic
(What's this?)
40 Value Stocks that Graham Would Buy
(Dividend Growth Investor, 11/28/08)
Blogroll
(Dividend Growth Investor, 1/1/09)
The Five Keys to Value Investing Profits
(Money Morning, 11/20/08)
Fairholme’s Maxims: A primer to value investing.
(Dah Hui Lau (David), 8/7/09)
|
