Are there any good arguments against the EMH?
Over the years I’ve swatted away lots of arguments against the Efficient Markets Hypothesis. The question is not whether the EMH is “true,” how could it be? Almost no economic model is precisely true. The question is whether it is useful. I find the EMH useful, and anti-EMH models to be almost completely worthless. I’m still looking for the model that will tell me how to beat the stock market. Just when I was starting to warm up to Shiller’s model, he missed the huge bull market of 2009-11.
Lots of academics in finance departments think you test the EMH by looking for market anomalies. Yet even if the EMH were completely true, there should be millions of anomalies out there—roughly one million for each 20 million data correlations that one examines. So that can’t be right.
A better idea is to see if there is evidence that others have found anomalies that are “real,” i.e. not just coincidence. The way to test that is to see if other people have discovered actual market inefficiencies, i.e. if excess returns are serially correlated. Yet studies show that if there are persistent excess returns to better than average mutual funds, the gains are so small, and so hard to spot, that this fact is virtually useless to the average investor.
The last stand of the anti-EMH crowd (in my comment sections) was to point to hedge funds. They argued that only dopes invest in mutual funds, and that all the smart money goes into hedge funds, which are in turn managed by the smart stock pickers. And they argued that hedge funds consistently earned above average rates of return, at least in aggregate, and over an extended period of time. Now it looks like even that isn’t true:
There is no doubt that hedge-fund managers have been good at making money for themselves. Many of America’s recently minted billionaires grew rich from hedge clippings. But as a new book* by Simon Lack, who spent many years studying hedge funds at JPMorgan, points out, it is hard to think of any clients that have become rich by investing in hedge funds (whereas Warren Buffett has made millionaires of many of his original investors). Indeed, since 1998, the effective return to hedge-fund clients has only been 2.1% a year, half the return they could have achieved by investing in boring old Treasury bills.
I’ve done much better than that investing on my own. And then there’s this:
How can that be, when traditional performance measures for the industry show average returns of 7% or so? The problem is a familiar one in fund management and is the equivalent of the “winner’s curse” that occurs with auctions (the successful bidder is doomed to overpay). Take a whole bunch of fund managers and give them an equal amount of money to invest. The managers that perform best initially will tend to attract more investors, and so will gradually become bigger than the moderate or poor performers (who will eventually go out of business).
But the manager will not perform well indefinitely. By the time a bad year occurs, the manager will be running a much larger fund. In cash terms, the loss on the expanded fund may easily outweigh the gains made when the fund was smaller. The return of the average investor will be lower than the average return of the fund.
And this:
What is true for individual funds also turns out to be true for the industry as a whole. Between 1998 and 2003 the average hedge fund earned positive returns every year, ranging from 5% in 2002 to 27% in 1999. Back then, however, the industry was quite small: overall assets only passed $200 billion in 2000.
That strong performance attracted the attention of pension funds, charities and university endowments at a time when their portfolios had been clobbered by the bursting of the dotcom bubble. They duly piled into “alternative assets” like hedge funds and private equity. By early 2008 the hedge-fund industry had around $2 trillion under management.
But that year turned out to be the annus horribilis for the hedge-fund sector. The average performance was a loss of 23%. In cash terms the loss for that single year was more than double the industry’s total assets under management in 2000, when it was still doing well. Mr Lack reckons that the industry may have lost enough money in 2008 to cancel out all the profits it made in the previous ten years.
One of my least favorite maxims is; “the market can stay irrational longer than you can stay solvent.” I consider that to be a cop out for losers. If the market is actually irrational, you set up a long term investment strategy to take advantage of that inefficiency. You don’t gamble everything on one role of the dice. There should be “anti-EMH” mutual funds that invest on the assumption of market inefficiency, and these should tend to earn above normal rates of return on long term investments.
The past five years should have been an absolute gold mine for the anti-EMH types that supposedly dominates the hedge fund industry. Just think about it. Shiller says stocks are way too volatile, and the US stock market has been incredibly volatile since 2007. No need to worry about the market staying irrational for too long, the long run adjustments occurred quite rapidly. Then we had the mother of all housing bubbles in 2006, another great opportunity for people to rake in profits from market inefficiency. The year 2008 should have seen extraordinary profits to the hedge fund industry, with all that “irrationality” being corrected. Instead they lost more than they’d made over the previous decade.
One guy did beat the market by betting the housing bubble would collapse, but I recently read that he’s been doing poorly ever since. And we all know about the unfortunate stock market call by Mr. Roubini in 2009.
The anti-EMH crowd needs to face facts. Even the smart money can’t beat the market, except by luck.
What is there not to like about the EMH? It’s an aesthetically beautiful theory. It’s survived every test thrown at it. And yet almost everyone thinks it’s wrong. All the really cool, smart, contrarian bloggers ought to love the EMH. I don’t get it. Maybe intellectuals are put off by a theory that implies that the rabble are (collectively) smarter than they are. But that’s not the right way to look at things. The rabble all oppose the EMH. Go to a bar and start talking to the first drunk you meet. I guarantee he will have an opinion on where markets are going. I guarantee he won’t say “predictions are impossible because all publicly available information is already incorporated into asset prices.” Who are you with; him or me?
PS. Many pundits claim the housing bubble shows the EMH is false. This post shows why they are wrong.
PPS. Many people claim that wild speculative bubble always lead to crashes. This academic study shows that’s not true:
Abstract. The collapse of an investment mania usually reminds people that the phrase “This time is different” is dangerous. Recollections of this mantra then typically either state outright or at least imply that “It is never different.” However, there is at least one counterexample to this cautious view, a giant and wildly speculative investment episode that was profitable for investors. The British railway mania of the 1830s involved real capital investment comparable, as a fraction of GDP, to about $2 trillion for the U.S. today. It faced withering skepticism and criticism, much of it very reasonable, as its supposedly rosy prospects were based on extrapolation from the brief experience of just a couple of successful early railways. Yet by the mid-1840s, it was seen as a great investment success.
The example of the railway mania of the 1830s serves as a useful antidote to claims that bubbles are easy to detect or that all large and quick jumps in asset valuations are irrational.
Of course there were many countries that had housing “bubbles” in 2000-06, where prices never collapsed.
Tags:
31. January 2012 at 05:57
So, markets are efficient and impossible to beat. However, another kind of market keeps paying, year after year, ridiculous amounts of money to people who fail to provide returns to clients’ assets. Where is the efficiency there?
31. January 2012 at 06:13
The EMH defined as a no-arbitrage argument a la Jensen is hard to refute, it’s very strong on simple “so why isn’t anyone else picking up the $100 bill” grounds.
The EMH defined as a convergence theorem toward some alleged fundamental value (e.g., net present value of future returns), or the EMH defined as a welfare theorem (optimal allocation of the capital stock by whatever social welfare function of your choice) is much harder to prove.
It is, I think, generally acknowledged that:
(1) even the weak-form EMH of Fama is strictly false in a statistically significant way; there have always been nontrivial transaction costs and trader commissions that prevent the observed price record from adhering to the EMH. Thus one is restricted to the intuitive form of there being no profitable arbtirage,
(2) but the intuitive argument is very, very strong: even where statistically significant divergences exist, over worrying areas like assorted systematic anomalies and apparently implausible risk preferences by traders, the divergences are very small in magnitude from a perfectly-no-arbitrage EMH,
(3) noise traders are necessary for the no-arbitrage EMH to be true at all, so the welfare effects can easily be ambiguous,
(4) conversely, a whole industry is making a lot of money out of these noise traders. One could claim that everyone has zero ex ante ability to predict whether they will be a noise trader or a knowledgeable trader, but this goes well beyond the EMH as a market hypothesis and into the EMH as an individual hypothesis,
(5) it is phenomenally hard to prove that prices converge to any plausibly fundamental value, due to the sheer strength of the no-arbitrage argument itself. Local no-arbitrage stability is not the same as global stability, but proving global stability would entail the possibility of fundamental analysis and nobody can do that to an interesting degree. The EMH is irrelevant as a macroeconomic hypothesis the moment someone says “locally stable sunspot equilibria”.
31. January 2012 at 06:14
Okay, (4) and (5) are not actually generally acknowledged 😛 I do claim (1) – (3) are, though, by the likes of Fama himself.
31. January 2012 at 06:25
EMH iff P=NP (http://arxiv.org/abs/1002.2284)
Since P != NP, markets are probably not (weak-form) efficient.
31. January 2012 at 06:25
Since P is probably not equal to NP, markets are probably not (weak-form) efficient.
On the other hand, one might take this for evidence that P=NP.
31. January 2012 at 06:26
What do you call consistently putting out a better product than the rest of the market?
Certainly that is an investment.
And certainly you are beating the market.
And certainly it is not luck.
31. January 2012 at 06:33
How do we explain the results of systematic bubbles in an experimentally-controlled environment?:
http://www.theatlantic.com/magazine/archive/2008/12
/pop-psychology/7135/
“Here, finally, is a security with security””no doubt about its true value, no hidden risks, no crazy ups and downs, no bubbles and panics. The trading price should stick close to the expected value.
…
But that’s not what happens. Again and again, in experiment after experiment, the trading price runs up way above fundamental value.
…
‘You get bubbles and crashes.’ Ninety percent of the time.
“
31. January 2012 at 06:39
Scott you wrote: “Almost no economic model is precisely true. The question is whether it is useful. I find the EMH useful”
Useful for what, you never say, which I think is a cop out?
31. January 2012 at 06:51
I think I agree with everything that david says, but I think he left out the argument about prices reflecting the expectations of investors in liquid markets (which is sort of the definition of EMH, the others are more like implications of EMH). This definition leads to the no-arbitrage argument, but I have never seen how it leads to the second two. There are plenty of reasons to expect that investors expectations are not consistent with the fundamental values of the securities. The welfare argument is similar. The allocation can only be socially optimal to the extent that the allocation reflects the expectations, preferences, and time horizons of the individual investors in society.
31. January 2012 at 07:13
A major problem with the EMH is the lack of a testable definition.
Fama has said that in order to test the EMH you need an asset pricing model. An accurate asst pricing model obviously can’t exist (else why would prices ever diverge from it).
31. January 2012 at 07:18
@David, regarding no arbitrage, The Limits of Arbitrage,
Andrei Shleifer & Robert W. Vishny, http://pages.stern.nyu.edu/~cedmond/phd/Shleifer%20Vishny%20JF%201997.pdf
Fama seems to have moved from “the price is right” to “it’s hard to beat the market.” As you note, the former is rather difficult to prove (I’d say impossible, even in principle), while the later is essentially trivial – see Lake Wobegon.
31. January 2012 at 07:20
Why do you assume that models that actually produce financial gain will ever be public?
31. January 2012 at 07:21
@John Hall, “the argument about prices reflecting the expectations of investors in liquid markets.” Just what does “reflect” mean? How might this argument be tested?
31. January 2012 at 07:22
Also, the EMH is a very poorly defined term/ theory. It’s like saying ‘holding monetary policy constant’.
31. January 2012 at 07:28
Neal: that result doesn’t really show what it claims to show. Their notion of efficiency is computationally intractable, yes, but that just means that it’s intractable for all of us and we still can’t come up with a strategy that beats the market.
31. January 2012 at 07:29
Yes, in his recent papers Fama has moved to “it’s hard to profitably beat the market”. To be fair, this is still an achievement compared to what was conventional investment wisdom in 1970, and it is a far cry from “Keynesian beauty contest” views of capital markets. We can readily acknowledge that prices seem to be near local optima of risk-adjusted net present values, even if we haven no idea where other optima exist or whether the optima themselves move in some non-fundamental-related way (sunspots!).
@Ritwik – even Fama acknowledges that there exist people with insider knowledge who earn statistically significant supernormal returns, from tests of the strong-form EMH. The magnitudes are not very large, however.
31. January 2012 at 07:33
The criticism of EMH is that markets create bubbles as our hindsight becomes 20/20. IOW, they don’t know what they don’t know. So inevitably in the short run there is a correction, as in housing bubbles and dotcom fiascos. How does one write an equation to take advantage of this, especially when information is NOT perfectly accessible and the implied financial model may be as complex as the economy itself.
In the long run, the market, like all of humanity, is on a search for value. That search is emergent, plagued by failure, and in many cases influenced by cultural meme, error and human avarice. Some of those motives may be honorable and even rational, but it does not mean they are correct or even sane.
But the ultimate problem with EMH is that it is irrelevant, especially because it misses causes of shifts and their realignment. Network theory may bear more fruit, since it is an economy’s nodal relationships and coordination that often produces emergence and new paradigms, and so their changing relationships ultimately present most of our increasing knowledge and value.
In this regard, Hausmann and Hidalgo are doing cool stuff.
31. January 2012 at 07:51
David,
The magnitudes of people who can get into Harvard (or the U of Chicago, for that mater)are also not large. This does not mean that selection to elite universities is best described by a random walk. Nor does it imply that every Harvard admit has to be a legacy kid.
The Fama/Samuelson implication of ‘abnormal success = inside info’ is equivalent to saying that you cant get into Harvard unless you’re legacy. The argument of ‘how many people can successfully beat the market’ is like saying ‘how many kids can successfully get into Harvard?’.
For the EMH to be ‘untrue’, even in the practical sense, investment success doe snot have to be widespread. It only has to be replicable. No amount of measuring the average return of mutual funds or hedge funds across the years or decades even hits at the heart of the problem.
All that we know is, if you’re not very very smart/ don’t work for a large institution, don’t punt. Which is good advice. But not sexy enough to be called EMH.
31. January 2012 at 07:53
“We can readily acknowledge that prices seem to be near local optima of risk-adjusted net present values”
We can?
To know this, we’d have to know future cash flows and an appropriate discount rate (likely a risk-free rate plus an adjustment for risk). Using market rates seems circular. Where else would you get this information?
31. January 2012 at 07:56
I’m going to try and get you all to come at this another way one more time…
What do you call Michael Jordan?
Is your problem with the analysis that Michael Jordan didn’t actually compete against the market, he just competed with other basketball players?
Because technically everyone is a basketball player.
31. January 2012 at 07:57
Bubbles and crashes say absolutely nothing about the truth of the EMH, in any of its forms. Markets incorporate all publicly available information, but that is not the same as omniscience. Reality is Bayesian, thus we (and markets) constantly adjust our estimates of specific outcomes as new information becomes available. You can interpret the wild volatility since 2007, or even since August 2011, as a pendulum swing from the Bayesian observation that there are two medium term (~2 years) outcomes emerging for developed economies: a return to trend growth rates (if not the prior path); another echo crash brought on by too tight money, this time in the Eurozone, delaying recovery for another few years.
31. January 2012 at 07:59
foosion – studies of long-term returns, presumably. I admit it gets more dubious here.
31. January 2012 at 08:01
Morgan Warstler, I can only suggest acquiring at least a passing knowledge of what the argument is about before trying to contribute to it.
31. January 2012 at 08:03
david,
“I can only suggest acquiring at least a passing knowledge of what the argument is about before trying to contribute to it.”
Why would he break with four decades of tradition re: the EMH? 😉
31. January 2012 at 08:14
On the “it gets more dubious” note, I think Fama’s term for it since the 70s has been to call it the “dual hypothesis problem” and to try to separate the EMH into a “asset price formation” problem and a separate “plausibly rational price movement” problem. It is acknowledged that the latter is a lot easier to argue than the former!
31. January 2012 at 08:15
What Cthorm said.
“All information available to the markets” does not mean “complete information about the future”, and includes opinion, biases etc. Fama said such himself as early as 1970 in a paper. Fama also said the EMH does not require any specific assumption about the distribution of returns. All it takes for the EMH to be at least weakly true is that no one actor consistently predicts the future correctly. In its core the EMH is a corollary to the Hayekian view of distributed information in the economy, formalized as a capital market theorem.
The empirical evidence for the EMH to be true lies in the fact that no investor actually owns all assets, as yet, something that would be bound to happen in the long run were (s)he to outperform others consistently.
31. January 2012 at 08:28
David, that’s why I continue to assert that the EMH is not well defined.
mbk, there are many traders who do very well over the course of their careers. The key step for EMH is to figure out a way to separate skill from luck. Alas, we don’t have enough data to determine if success is skill or luck (e.g, get a million people to toss coins and someone will get 20 heads in a row). I believe Fama has a paper that acknowledges this.
Of course no investor owns all assets. No matter how skilled you are, it makes no sense to keep going once you have sufficient wealth.
31. January 2012 at 08:29
… nope. Grossman-Stiglitz for distributed information without EMH. Any standard perfect-information market model for EMH without distributed information.
Also wrong, sadly. You can earn supernormal profits without going on to own the universe, by simple token of consuming your abnormal returns instead of reinvesting them.
31. January 2012 at 08:34
Here’s a presentation by John Cochrane that you might find interesting:
http://faculty.chicagobooth.edu/john.cochrane/research/ppts/hedge_talk.ppt
Particularly the part about hedge fund returns after removing survivorship bias.
31. January 2012 at 08:39
You can’t beat the EMH because its a “postmodern” theory, like “deconstruction” or something of that nature. EMH is always proven because, in the last instance, you are always just getting to the right price, just getting that last little bit of information. And if something was mispriced, well,the information wasn’t known yet. EMH really isn’t that deep.
Oh, and by the way, it’s not true. Read: “The superinvestors of graham and doddville”- just google it, it’s a talk given by Buffett years ago. And don’t tell me that Buffet and company are just proving EMH because they make the market efficient. Well, yes, they do, and if that is the case then the EMH isn’t telling us anything. Guys like Buffett, Klarman, etc. beat the market consistently.
31. January 2012 at 08:40
@foosion – I think Fama is prepared to defend particular asset pricing models, if you poke through his more recent papers.
31. January 2012 at 08:41
Whoops, wrong Fama 1996 paper. This one: http://www.jstor.org/pss/2329302
31. January 2012 at 08:57
The EMH is a non-falsifiable way of looking at the economy. It is not a predictive theory. The notion that the EMH has never been falsified, as if that supposedly firmly establishes its legitimacy, is flawed.
Nothing on the empirical side could ever falsify the EMH. Even if a pro-EMH advocate was shown example after example after example, the alternative explanations and interpretations of which do contradict the EMH, the pro-EMH advocate can always explain the empirical examples using EMH and simply denying the competing explanations and interpretations, without directly considering those explanations and interpretations.
The EMH, if it is going to be established as a valid theory, must stand up not to empirical observations, but to competing theoretical explanations and interpretations of those empirical observations, which of course means EMH has to stand up to economic logic.
What is the logic of the EMH? The logic is not actually about “the market”, but rather an individual’s performance over their lifetime (it doesn’t actually speak of precise time periods, and I’ll show why that is below). The theory argues that because prices are informationally “efficient”, an individual cannot achieve returns that consistently beat average returns (risk adjusted), given the available information at the time of the original investments.
What logic can possibly refute this? Well, suppose I showed you an example of an innovator who over the course of, say, 20 years, consistently beat out the competition, and amassed a huge fortune far greater than the average investor. This is the empirical fact. What is a possible interpretation of this that doesn’t refute economic logic?
An anti-EMH explanation is that the investor was, during those X years, relatively superior at earning money, which of course means he was also, during those X years, relatively superior at producing goods for others, because of his entrepreneurial ability and knowledge.
A pro-EMH explanation is that the investor was, during those X years, relatively superior at earning money, which of course means he was also, during those X years, relatively superior at producing goods for others, because of having a X year string of higher than average luck.
Now, this is the real reason why anti-EMH and pro-EMH advocates argue against each other: The anti-EMH crowd holds the philosophy that individuals earn higher returns primarily because of having a superior ability and knowledge than the average, and not just luck. The pro-EMH crowd holds the philosophy that individuals earn higher returns primarily because of having higher luck than the average, and not just ability and knowledge.
They each tend to “prove” their case by pointing to empirical examples the explanations of which seem to contradict the other side’s explanations.
Sumner is pro-EMH, so when he is presented with an empirical example, the explanations of which contradict pro-EMH, he utilizes the vagueness of the lack of a definite time period inherent in the pro-EMH theory, and say “Yes, Mr. Smith did earn exceptional returns….but only for a time, and it matters that then his returns fell later on, as his relative ability and knowledge were illusory. Thus, it’s not his relative ability and knowledge that explains his return in those X years, it’s his luck. I win.”
I am anti-EMH, so if Sumner presents me with that same empirical example, the explanations of which contradict anti-EMH, I too can utilize the vagueness of the lack of a definite time period inherent in the anti-EMH theory, and say “Yes, Mr. Smith did earn exceptional returns….for that time, and it doesn’t matter if his returns fell later on, as his relative ability and knowledge might have changed. Thus, it’s not luck that explains his return in those X years, it’s his relative ability and knowledge. I win.”
The way to settle this debate once and for all is to make explicit what it means for someone to earn “CONSISTENTLY higher returns than average.” Does that mean 1 month? 1 year? 2 years? Of course if you wait long enough, any truly exceptional investor will eventually get old, lose his former abilities just like a sports athlete, and eventually his returns will fall (if he continues to manage and have the same responsibilities and duties as before). So the pro-EMH crowd can ALWAYS wait and point to an individual’s eventual returns after a string of good [performance; or luck], and say “Yes, he did well, but now he’s not doing so well.
I think the reason why EMH has won over many adherents is that it gives the rabble a justification for explaining why their returns are relatively poor and why other individual returns are so high. Sumner says he hates the quip: “the market can stay irrational longer than you can stay solvent”, on the basis that he considers that to be a cop out for losers. The irony is that there is a strong case to be made that the EMH is itself a “cop out for losers.” A loser can always explain their relatively poor performance, and other individual’s relatively superior performance, on the basis of “luck” and not ability or knowledge. It gives the rabble an open door to invest in the stock market, by essentially telling them “Don’t worry, you’ll make on average what everyone else makes. If anyone makes a killing, it’s because of luck, and so you might be the next to win at the wheel!”
All pro-EMH is at the end of the day is a philosophy that holds individual humans to gain primarily because of luck, not just their ability and knowledge. All anti-EMH is at the end of the day is a philosophy that holds individual humans to gain primarily because of ability and knowledge, not just luck.
No empirical data devoid of interpretation can ever contradict either theory. This is a philosophical debate on the nature of individual human life, that’s all. You either accept that individuals earn higher returns than other individuals primarily because of having superior ability and knowledge, or you accept that individuals earn higher returns than other individuals primarily because of having superior luck.
Whichever philosophy you choose, you will always be able to explain any and all empirical data. The problem for Sumner is that he wants to believe EMH is an empirical predictive theory, so that he can say it’s right because it’s never been empirically falsified. In reality the EMH is an a priori philosophy about human life.
31. January 2012 at 09:12
@David, I don’t see how there ever could be a useful asset pricing model for EMH purposes. If there were a model that could reliably tell us whether prices were correct, why would anyone ever trade at another price?
The link you gave is gated. If memory serves, that paper is one of the main Fama-French three factor model papers, showing that if a one-factor model breaks down on X and Y, adding X and Y to create a three-factor model has more explanatory power than the old model.
31. January 2012 at 09:19
Major Freedom: except for the fact that, you know, Scott has specifically and repeatedly mentioned the experiment that could falsify the EMH.
31. January 2012 at 09:27
Size, Value, and Momentum.
31. January 2012 at 09:40
“Major Freedom: except for the fact that, you know, Scott has specifically and repeatedly mentioned the experiment that could falsify the EMH.”
That’s the problem! That’s the very point I am making. He incorrectly believes that the EMH can be falsified via an empirical experiment that is devoid of explanation other than EMH.
He wrote above:
“A better idea is to see if there is evidence that others have found anomalies that are “real,” i.e. not just coincidence. The way to test that is to see if other people have discovered actual market inefficiencies, i.e. if excess returns are serially correlated. Yet studies show that if there are persistent excess returns to better than average mutual funds, the gains are so small, and so hard to spot, that this fact is virtually useless to the average investor.”
Who cares about mutual funds? What about individual investors, the subject matter of the EMH?
What does he mean by “persistent”? He can always say high returns were not persistent by just expanding the time period. It’s easy to explain it away when you just extend one’s time horizon beyond whatever time period the investor did achieve superior returns. Oh, that investor achieved higher than average returns? For 1 year? That isn’t “persistent.” Oh, that other investor achieved higher than average returns for 3 years? That isn’t “persistent.”
The vagueness of the time period is needed as the final escape clause to deny all anti-EMH arguments.
There are many examples in which individual investors have achieved superior returns, but the pro-EMH can always interpret those examples as proof of EMH, by just calling attention towards the time periods the individual did not achieve those superior returns, or explaining it away by luck.
31. January 2012 at 09:44
Scott, how do you explain Parrondo’s Paradox if anti-EMH is not “useful”? This sure seems like arbitrage to me…
http://www.youtube.com/watch?v=cajg7HTCiBM
31. January 2012 at 09:55
Scott my question is the opposite: is there any good argument for the EMH? Arguing about hedge fund returns doesn’t do it.
I see that at at best you argue based on various market returns, not from a macroeconic argument. Yet when you claim that,
“I find the EMH useful, and anti-EMH models to be almost completely worthless. I’m still looking for the model that will tell me how to beat the stock market.”
I find your argument misleading because as far as I know you are not a major market participant. How exactly has the EMH model been useful for you? I’m guessing that when you say that you mean useful as an economist. There are two things-the EMH as it is useful-or not-for market particpants and economists that use it for their models.
By the way, arguing that the S&P has a 10% return and so that makes it efficient doesn’t prove EMH much less that there are no good arguments against it. Just because the S&P is 10% doesn’t mean that eveyone or even most people in it will make a 10% return.
Hedge funds are actually closer to a meritocracy but the results vary a lot. In 2008 it was a good year for hedge funds and trading in particular. The last few years have been better for buy and hold investing.
Over the long haul it is tough to get rich from playing buy and hold. At best it might be a high yield savings account but as there is more risk you have to factor that in to in assessing it’s value.
Trading startegies-what hedge funds engage in is swining for the fences. Those who do well can make huge returns. Those who miss lose verything.
As someone who has played the market both buy and hold and more aggresively as a trader I can tell you that the EMH is largely irrelveant to market particpants. ON the trading floor you wont find many agreeeing with you though the mutual funds will love you for carrying their water.
Still when you adjust for inflation the S&P still isn’t where it was at the top of the 90s bull market; the Nasdaq is much lower than it’s 2000 high.
31. January 2012 at 10:00
The EMH can never be disproved by empirical data. This is Eugene Fama’s joint hypothesis problem. Any so called test of EMH is a test both of a model of expected returns and market efficiency. If the test fails you don’t know which is wrong.
That said, one major assumption about your post is that investors in mutual funds are stupid. You should really look at Berk and Green “Mutual Fund Flows and Performance in Rational Markets”. The point is that the gains from trade from manager skill in stock picking obviously go to the managers and not the buyers of mutual funds in a competitive market. Therefore, Bayseian consumers will flood into successful mutual funds, and the gains from successful stock picking go to the managers through extra fees. There isn’t persistence in returns because of something called closet indexing, which is that there are diminishing returns to stock picking so bigger funds have to look more like the index.
I would expect a similar mechanism to be at work for hedge funds. The gains from trade go to the hedge fund managers, not the institutional investors. The fact that hedge funds don’t outperform, is not a proof of efficiency.
Anyway, I think the EMH is wrong because I think the models of expected returns needed to justify housing, gold and stock prices during certain periods are implausible.
31. January 2012 at 10:09
You wouldn’t, hence the (3) I mentioned above. For even the weak EMH to be meaningful, there have to be noise traders. Otherwise, trivially, every trader would realize something’s up when offered a trade and there would be in fact no trade at equilibrium, not even trade at the model-implied market-clearing price (except for liquidation for immediate consumption). I can’t remember who first made the argument rigorously, but LeRoy comes to mind.
31. January 2012 at 10:11
david, I understand it, I want to see an actual line drawn by those pro-EMH (I’m assuming thats you) between provable skill and luck.
As an example: in the laws surrounding sweepstakes / skill game / contest rules it is impossible to nail down exactly, nearly to the point it almost gets epistemological, and you routinely run into grey areas based on avg. consumer performance.
Explain where the line gets drawn. Shut me up.
31. January 2012 at 10:21
@Jason – or you could pick a model of expected returns which is exceptionally simple (stock split tests, etc. – the only requirement is a plausible level of consistency). Or a model which is complicated but does seek to explain most behavior (ICAPM multifactor asset pricing etc., or whatever newfangled model Fama is toting these days). Fama himself designed his tests to try to evade the dual hypothesis problem.
The point is that it is in principle possible to falsify the EMH by making a careful choice of asset pricing model, one which is held to be true either by virtue of its falseness being implausible or its support from other fields (behavioral econ, portfolio theory, etc.) and thus forcing the market efficiency part to take on the explanatory load.
31. January 2012 at 10:21
> Lots of academics in finance departments think you test the EMH by looking for market anomalies. Yet even if the EMH were completely true, there should be millions of anomalies out there””roughly one million for each 20 million data correlations that one examines. So that can’t be right.
To check my understanding: this is a reference to statistical significance, p=0.05 (1/20 = 5/100), right?
> The example of the railway mania of the 1830s serves as a useful antidote to claims that bubbles are easy to detect or that all large and quick jumps in asset valuations are irrational.
Odd. I remember reading a fair bit about that, and the takeaway was the first railway mania was only a modest success which wiped out a fair number of investors, and the second British railway mania was an unmitigated disaster.
31. January 2012 at 10:23
Morgan has this right. In every other endeavor, we have models that include entrepreneurship, skill, information asymmetry, etc., so that market actors constantly look for and find excess profits. Sometimes these profits are fleeting and sometimes they last a lifetime. But, when we turn to finance, we have this EMH virus.
I have a buddy that has a knack for selling furniture to schools. I have another buddy that has a knack for valuing small cap stocks. They are both doing work that anybody could do. They don’t have a secret code. They both work in relatively efficient markets. And I suspect that they will both do well as long as they apply their skills to their endeavors. Should I tell my salesman buddy that EMH says his commissions are just the result of statistical noise?
I and practically everyone else I know would make terrible school furniture salesmen. Does this prove the EMH in school sales? Should this cause my buddy to lose confidence?
31. January 2012 at 10:29
Scott, the comment section seems to indicate that the two of us are the only people in the world that feel happy to use EMH. I think the problem basically is that people have a had time understand that the market is the best “aggregator” of information and knowledge known to man. It is hard to believe that the collective wisdom of the crowd should generate efficiency despite the fact that we are all surrounded by idiots (some of us are even idiots ourselves), but that is nonetheless the case.
Your sceptic commentators should have a look at this little clip from BBC: http://www.youtube.com/watch?v=iOucwX7Z1HU
31. January 2012 at 10:33
When you say ‘swatted away’ you mean ‘hand-waved away’ by arguing that it’s impossible to know what market prices should be so crashes don’t count (your first post on the EMH says this of the 1987 crash).
EMH proponents continue to move the goalposts and I’m never sure what they’re defending. If you’re saying ‘markets are decent at processing information’ then yeah, I can get on board with that. But it has no implications.
Basically, what I’m asking is for you to state some explicit, falsifiable implications of the EMH.
31. January 2012 at 10:37
@Morgan Warstler
I’m going to state the EMH qua Fama: securities markets price assets in a way that reflects all and only the information contained in some given information set: past prices, or past prices and all public information, or past prices and all public and private information.
That’s it. You tell me where your notions of luck and skill enter.
31. January 2012 at 10:39
@Lars Christensen – Grossman & Stiglitz? Profitably aggregating decentralized information conflicts directly with allocatively efficient results.
31. January 2012 at 10:49
Colby, the Fama-French three-factor model is market, size & value. Only recently have they taken momentum seriously, which would be a fourth factor.
31. January 2012 at 10:51
Lars,
Your argument has two separate parts, and one does not follow the other.
‘The market’ > ‘The individual’ does not imply that ‘The market’ = ‘correct’.
Further, neither of these has anything to do with ‘anti-EMH is not useful’, which is Scott’s viewpoint.
EMH is useful to the extent that humility about one’s general lack of knowledge and insignificance is useful. That’s about it.
Fama’s joint hypothesis postulation is as close to formal finance gets to striking at what I see is the heart and soul of EMH. I have a somewhat different take, that of replicability (and hence non-randomness).
For example, consider the proposition that ‘Warren Buffet making super-normal returns year after year disproves EMH’ usually countered by ‘Warren Buffett is a six-sigma event, there had to be one in the history of investing and that is Buffett’, or ‘Buffett has insider non-public information and is hence successful’.
This itself suffers from the problem of joint determination and poor specification. Consider the fact that at the prime of his investing days, Buffett used to read a 1000 annual reports in a year. Let’s replace Buffett’s investing strategy by the simple heuristic of ‘read a 1000 annual reports in a year’. The question of whether you will make similar returns as Buffett is now two-fold
1. Does reading a 1000 reports in a year allow you to beat the market consistently.
2. Can you read a 1000 reports in a year consistently.
Measuring your returns (long term, short term, whatever) is calculating AND(1, 2). A negative answer to this means that either 1 or 2 could be false. But to my mind, EMH is disproved as long as the answer to 1 is true. So measuring returns does not tackle the fundamental problem.
31. January 2012 at 10:58
@David, “For even the weak EMH to be meaningful, there have to be noise traders.” If there was an accurate pricing model that was used to test EMH, it would be widely publicized. It’s hard to see how anyone would thereafter be a noise trader.
“securities markets price assets in a way that reflects all and only the information contained in some given information set”
Care to clarify “reflects”? It seems a rather vague term. Does it imply some degree of accuracy or does it merely mean thinking about something?
31. January 2012 at 11:03
david,
‘all the information contained’ is a very poorly defined term. All the information contained in past prices and public information means different things for people of different skill. This can be a source of super-normal returns. This is the line to argue against the ‘is anti-EMH useful’ claims which assume the conventional meaning of ‘past prices and public information’.
Or, if one were to really take the ‘all the information contained’ in its most literal sense, then the second part of Fama’s formulation leads to Samuelson’s result of stock prices following a random walk and hence the log-normal distribution, which is demonstratably false.
31. January 2012 at 11:08
“bookies price bets in a way that reflects all and only the information contained in some given information set: past prices, or past prices and all public information, or past prices and all public and private information.”
would you agree this is also true?
31. January 2012 at 11:10
I’ll even give you my next question as we head down the garden path:
Will you stipulate that US Congressmen have been outperforming the market YOY?
31. January 2012 at 11:15
The EMH is a worldview not a hypothesis. Can you name one event that would disprove it?
31. January 2012 at 11:19
read page 1
http://www.berkshirehathaway.com/letters/2010ltr.pdf
summary: overall gain 1964-2010:
Berkshire Hathaway per share book value- 490,409%
S&P 500: 6,262%
31. January 2012 at 11:26
😛 There’s no accounting for individual stupidity… irrational people can exist. They just earn suboptimal returns.
If everyone was perfectly rational under limited information, we would have the no-trade equilibrium. We observe trade under limited information, so at least one individual is not perfectly rational.
It implies the market model of choice. This part of the hypothesis is left deliberately unspecified and is instead combined selectively with separate accounts of how asset prices are formed.
31. January 2012 at 11:39
Lots of commenters are claiming that the EMH is “empirically unfalsifiable,” since whenever anti-EMH types point to Bob who got 25% annual return for 30 years straight, the EMH proponents stick our fingers in our ears and yell “luck!!!” and there’s no way to tell the difference.
Well, there is a way: is Bob *any* more likely to earn above market returns in his 31st year?
No.
Imagine if every time students took the SAT, they got a score completely uncorrelated with their score on previous tests. Pick your favorite explanation:
a) the SAT is measuring the current level of something important called skill, but skill one day (/month/year/etc.) is unrelated to skill the previous day (/month/year/etc.).
b) the SAT is measuring luck.
31. January 2012 at 11:42
I would have thought the entire history of humanity and the immeasurable amount of bubbles thereof would be sufficient to disprove the EMH but apparently they aren’t.
31. January 2012 at 11:47
A few observations.
1) It’s not so much about analyzing existing information. It’s about predicting the future.
2) If you listen in on an analyst call for any listed company, you will discover that most analysts (supposedly the experts) don’t know shit about the companies they are covering.
3) There are definitely people out there who are good stock (asset) pickers. It’s harder to pick a good picker than it is to pick a good asset.
4) The market has a huge short term bias.
5) Asset managers are better salesmen then they are asset managers.
Just some observations…I have know idea what this says about EMH.
31. January 2012 at 12:10
Eric Falkestein has an interesting take on EMH were he says that you can beat the market by buying the safest equities because people are willing to pay for more volatility. It is like buying lottery tickets, people like the possibility of a big payoff, so much so they are willing to pay something for it. He explains that T-bills and AAA only under perform because they are needed for transactions and to fulfill some laws.
His web page:
http://www.efalken.com/
So people want stock swings that can net them huge returns and so under value low beta stocks like KO and PEP and over value high beta stocks.
I think that is not really a challenge to EMH though same as gambling is not people are paying for the volatility.
31. January 2012 at 12:24
Scott, you are torching strawmen.
You simply don’t comprehend or choose not to acknowledge the core of what is wrong with EMH analysis.
The whole history of EMH is the history of statistical relations between different streams of numbers — numbers chosen on the basis of particular theories of what number stream should cause or correlate with some other number stream.
Rules out BY ASSUMPTION is most everything that Hayek identifies as the core causal explanatory element in economics — say, everything Steve Jobs brought to his businesses at Apple, NEXT, PIXAR, etc. which allowed jobs to use his local judgment, adaptive perception, skilled understanding, constantly updated future relative price relation insight, etc. to turn a few thousand dollars into a multi-billion dollar fortune.
I.e. the news that of Steve Jobs returning to Apple doesn’t count as “information” in numbers and their functional relations only world of the EMH and it’s “scientific” history.
And the deep heterogeneous production process and relative time consumping structure across all relative prices and technologies involved in Hayekian macro causal explanation also completely stands outside of the simple minded number stream correlations at the heart of EMH “models”.
So it is a great and fraudulent canard to suggest that appeal to EMH decides anything when it comes to the core issues of what provides good “micro” and “macro” explanations of patterns in the coordination of market plans.
31. January 2012 at 12:25
Lars you are right many don’t agree with EMH. What I agree with though is your post about me today.
Appreciate it. Lars Christensen writes a post about me http://diaryofarepublicanhater.blogspot.com/2012/01/lars-chistensen-writes-post-about-me.html
31. January 2012 at 12:38
@John S, “Well, there is a way: is Bob *any* more likely to earn above market returns in his 31st year?”
The only way to know is to look at Bob’s returns for his 31st year. However, the vast majority of traders have a track record much shorter than 31 years. In other words, we don’t have enough data to answer your question on a systematic basis.
Requiring above market performance every year is hardly a way to separate luck from the alternative. Few would deny that baseball players (or pick your favorite sport) have skill, yet not everyone gets a hit every time at bat or performs above average every year. One bad year doesn’t prove much of anything.
We don’t have enough data to generate a statistically significant answer.
31. January 2012 at 12:40
I don’t know who you are trying to persuade or how you think you are persuading them, with these arguments, Scott.
The house price / house price to rent ratio nearly doubled from where it was in 1999 and has since gone down back to about where it was in 1999 — when the ratio was back on a gently rising long term trend.
And I can’t believe you’ve actually informed yourself on this issue, e.g. the namable quants who made a killing on the massive financial distortions generated by the housing / mortgage market.
Straight up question for you Scott.
Have you read Michael Lewis’s _The Big Short_ or any of the other books documenting that in fact these guys did see what was happening before it happened, and they made huge fortunes acting what they were seeing?
If you really want to engage this issue in a helpful way, engage the details found in Lewis’s book, or one of the other books on the topic.
Enough shadow boxing of phantoms — and enough torching of strawmen.
31. January 2012 at 12:47
david 31. January 2012 at 06:13
Thanks for actually adding something to the conversation …
31. January 2012 at 12:47
David,
You are positing a level of irrationality that is rather extreme. There may be people that dumb, but not many. How many people go into stores and offer to pay over the marked price?
People can rationally disagree about the correct prices and therefore trade. Think standard rational micro supply and demand curves. Securities markets should not be different in this respect than any other market.
“This part of the hypothesis is left deliberately unspecified”
That’s a good summary of one of the major problem with EMH – it’s left deliberately unspecified 🙂
31. January 2012 at 12:53
In equilibrium hedge fund capital should earn the same (risk adjusted) return as stocks and bonds. That’s true regardless of market efficiency. In the extreme efficient market case, there is almost no capital allocated to hedge funds. In the extreme inefficient market case, 99% of capital is allocated to hedge funds (in other words, markets are pure casino and serve no investment purpose).
31. January 2012 at 12:54
Drunkeynesian, The EMH doesn’t claim that output is produced efficiently.
David, Good points.
Neal, The issue is not whether the EMH is “true,” it’s not, it’s whether anti-EMH theories are useful.
They aren’t.
Morgan, I call that “Apple.”
Frank, Studies ahve shown that classroom experiments don’t always generalize to the real world. I am interested in what goes on in the real world, and whether anti-EMH models are useful. It seems they aren’t.
OGT, You said;
“Useful for what, you never say, which I think is a cop out?”
I find these sorts of comments to be really annoying. All of market monetarism is based on the EMH. The EMH told me monetary policy was disastrously off course in late 2008.
foosion, I think “it’s hard to beat the market” is the right way to test it. There’s no tautology there (it has been tested by Fama) as anti-EMH types like Shiller claim you can beat the market.
Ritwik, I don’t assume that. That’s exactly my point, they won’t be public. Google my “alchemy” post.
Jim, I find the EMH very useful.
Cthorm, Good point.
david, You said;
“On the “it gets more dubious” note, I think Fama’s term for it since the 70s has been to call it the “dual hypothesis problem” and to try to separate the EMH into a “asset price formation” problem and a separate “plausibly rational price movement” problem. It is acknowledged that the latter is a lot easier to argue than the former!”
Whenever I read finance types trying to get philosophical I think that they might do well to read some actual philosophy–especially Richard Rorty.
Jay, What’s the punchline?
Greg, If the EMH were true you’d expect successful investors like Buffett to occur on occasion, I did a post on that once (I think called “Being There.”)
Major Freedom, It is falsifiable. The proper test is whether some mutual funds consistently outperform others. That’s the right test, and it passes.
Rafe, I’ve never heard of the Parrando’s paradox? Is it an anomaly?
Mike Sax, I find it useful as an academic and as an investor and as a voter. As an investor I use index funds, and earn higher returns than my colleagues who don’t believe the EMH. As an academic I use it to infer the effects of monetary policy from market reactions. As a voter I vote against politicans that tell me they support regulatory regimes that assume the EMH is false.
Jason, The version of the Anti-EMH you present is not useful to me. I have no evidence that mutual fund success is based on anything more than luck, thus no reason not to buy index funds (which outperform managed funds.)
more to come . . .
31. January 2012 at 13:00
Scott, your reflections on this topic remind my a lot of my small children when tey are BSing me patent absurdities and thinking they are getting away with it …
I don’t know if you intend your reflections to be so perversely infuriating, but I would put money on the fact that you well know what you are doing.
31. January 2012 at 13:00
Scott, your reflections on this topic remind my a lot of my small children when they are BSing me patent absurdities and thinking they are getting away with it …
I don’t know if you intend your reflections to be so perversely infuriating, but I would put money on the fact that you well know what you are doing.
31. January 2012 at 13:03
Scott, this is NOT an argument, it’s an non sequitur
“One guy did beat the market by betting the housing bubble would collapse, but I recently read that he’s been doing poorly ever since.”
31. January 2012 at 13:08
Have you ever read about the evidence on dual listed companies? It doesn’t necessarily conflict with the certain definitions of the EMH, but it is pretty compelling evidence that something funny is going on.
You have 2 companies in different countries with dividends that are legally bound to be the same forever, and yet you have different prices, sometimes significantly (e.g. 30% for royal dutch/shell in the 80’s) and sometimes for a long time. These differences don’t seem to be able to be possibly accounted for by taxation, and the differences are correlated with indices of the markets they are on.
The NPV of these stocks must be the same or almost the same, and yet the prices differ substantially. What’s up with that?
31. January 2012 at 13:09
gwern, To your first question–yes.
Regarding the mania, that’s why academic studies are useful. It was once believed that the tulip mania was also completely irrational.
kebko, I don’t doubt that some people are better stock pickers than others, that has no bearing on the usefulness of the EMH to me.
Lars, Thanks for the link–there are a few others who agree with us.
Unlearningecon. The implication of the EMH being wrong is that well managed mutual funds should consistently outperform less well managed funds. But they don’t.
John, See my previous answer.
Morgan, I agree that insiders might have an advantage (haven’t studied Congress.)
Greg, Google my post “Being There.”
John S. Excellent comment.
dtoh, I agree.
Floccina, That’s possible.
Greg Ransom, I agree that a few got lucky in 2008. We know it was luck, because they’ve been “unlucky” ever since.
Max, That may be true–I have an open mind on the issue. It’s not clear why people seem to gamble more than they should. Perhaps the market’s only inefficiency is that most people wrongly think it is inefficient.
31. January 2012 at 13:12
Be honest, Scott, you are claiming that anti-EMH considerations are not true, that they are not explanatory.
This not “useful” stuff is just BS to skirt the matter at hand.
31. January 2012 at 13:20
Scott,
EMH is useful (at least for me personally). As Lars said, markets are the best aggregators of information.
But I am pretty sure that NYSE is more efficient today than it was 30 years ago. The problem with EMH is that it doesn’t let us ask the question “Is the market more efficient today than it was some time ago?”.
The next step is to ask: “Which market is more efficient: TIPS market or NYSE?”. EMH prohibits such questions, even though they are very important. For example, in June 2008 TIPS market and NYSE gave you completely different answers to the question : “What is the expected NGDP for the next two years?”.
31. January 2012 at 13:23
Rob, I did read that. Can you remind me why investors didn’t just buy the stock on the cheaper market?
Greg, I’m always honest on this blog.
31. January 2012 at 13:23
Though I have chimed in on the “anti-EMH” side, I would add that a good stock picker must believe in efficient markets. They just need to be loagy enough to temporarily get predictably off track here and there. I earn excess returns by buying when they are loagy and selling after they correct. These trades are never made in a context of complete certainty, so the gains from skill will never be bid completely away, and the notion that they are a statistical anomaly won’t easily be put aside.
I would agree that most of Scott’s examples from the recent comment are useful positions stemming from EMH belief. But, I don’t think we would take it so far as to say, for instance, that today’s bond prices provide a level of certainty about near future inflation rates.
There are traders making excess profits, and those traders make the market efficient. Because this is a feedback system, the existence of these traders is both proof for and against EMH.
31. January 2012 at 13:24
123, I do agree that it is a matter of degree. If we had an NGDP futures market we wouldn’t have to worry about inefficiencies in the TIPS market.
31. January 2012 at 13:26
Scott,
We might have to worry. If the TIPS market gave me a different expected NGDP reading than the NGDP futures market, I might trust TIPS more if, for example, TIPS market was more liquid, etc.
31. January 2012 at 13:27
ssumner:
“Major Freedom, It is falsifiable. The proper test is whether some mutual funds consistently outperform others. That’s the right test, and it passes.”
No, that is not the right test, because A. it ignores individuals who can consistently earn high returns (e.g. Warren Buffet, Bill Gates, Steve Jobs, etc), and B. you are again keeping the word “consistently” undefined and purposefully vague, such that you can always explain away high returns as consistent with EMH by denying market “inefficiencies” during the time they made high returns and saying subsequent times after swamps those high returns out.
Suppose I am an individual investor, and for 50 years, say from 25 to 75, I earn more than the market average. I amass a fortune. Then in my old age I begin to lose my touch, my abilities wane, my knowledge isn’t so superior, and over the next 2 years I remain stubborn, I refuse to retire, and I end up losing everything I earned.
Would this be an example that confirms EMH, or would it be an example that violates EMH?
If you say it confirms EMH because of the losses, then presumably, if I did retire at 75, that would be a violation of EMH, right?
If you say it falsifies EMH because of the 50 years of amassing a fortune, then how do you explain Warren Buffet, Bill Gates, Steve Jobs, and all other INDIVIDUAL innovators who consistently earned higher than the market average?
I can’t see any justification for EMH. The real world, and logic, refutes it.
31. January 2012 at 13:29
Scott, here are some excellent take-downs of Friedman’s instrumentalist/”pragmatic” account of economic “science”. I’d recommend taking a look. Alexander Rosenberg does a good job, but there are many.
Why is a spark plug in an gas engine useful? Because is causes a fire to ignite.
But consider, the spark plug does not cause the fire to ignite because it is useful …
Similarly an explanation of how nuclear fission works is useful for a number of reason, e.g. the explanation allows us to build power plants.
But consider, the fact that nuclear fission is useful does not provide us with an explanation of how or why nuclear fission occurs …
31. January 2012 at 13:29
For what it’s worth, I think it’s clear that Buffett implicitly buys into EMH. Just look at how often he says people should just invest in low-cost index funds rather than actively managed funds.
31. January 2012 at 13:32
Scott,
I read the post “Being there.” If i read it correctly, EMH suggests Buffet would be the “lucky one.”
Buffett tries to counteract this type of argument in his article “THE SUPERINVESTORS OF GRAHAM AND DODDVILLE.”
The article targets the EMH. He states that he is not the “lucky one” but that there are many investors of his strain and with similar training (from Ben Graham)that have historically acheived excess returns. He names them and their investing records in the article. The article is a fun read with some elementary statistics in the beginning on coin flips. I beleive he does a pretty good job. Have you read it? What do you think?
Coincidentally, I’ve always thought that Fama’s LVH factor captures a lot of what the value investors practice. I haven’t yet reconciled what this means for the EMH.
31. January 2012 at 13:32
Instumentalism / pragmatism in economics is a giant punt to avoid confronting the 4 ton elephant in the conference room of the tenured economists — incompetence and failure when it comes to providing and articulating a sound and coherent causal explanatory science.
31. January 2012 at 13:54
Sorry, if people repeatedly invested in something that gave worse returns than treasuries, markets are not efficient.
31. January 2012 at 14:38
Since Greg Ransom has seen fit to laud me, let me spell out why EMH and Austrian capital theory conflict so badly: the most aggressive forms of the EMH (tied to asset pricing models) imply that the entire capital market does behave, in the aggregate, as if it were a single representative security with a market risk and market return. Asserting that you can always do as well as the market by buying the market portfolio is identical to asserting the existence of homogenous K.
What need of time-structure of production? Worse, how do prices act as a coordinating, revelatory mechanism if people just keep buying the market portfolio, and information is generated by irrational noise traders instead of calculating producers?
In short, if Fama is right then the New Keynesians should celebrate – the EMH itself is invoked in the usual three-equation New Keynesian setup – and Austrian theory collapses in the darkest ignominy. Whoops. Of course one can modify EMH, ACT, or both to salvage the wreck but the conflicts are pretty fundamental.
31. January 2012 at 14:44
Scott,
Parrondo’s Paradox is fairly general and widely applicable. Here’s a good survey paper:
http://rafefurst.files.wordpress.com/2008/02/icand07_derek_v4.pdf
In there the author talks about an application to markets called “volatility pumping”. The video link I sent you was a press conference by a Colorado professor who proved you could combine two negative EV instruments into a winning combination.
31. January 2012 at 14:49
1. Fama and all other EMH theorists admit that EMH necessarily implies the existence of profitable asset-picking or arbitrage. Otherwise there would exist no mechanism for keeping prices efficient.
At equilibrium, the marginal value of the investor’s labor has to be equal to the marginal value they could reap elsewhere. But there’s nothing that says that the returns have to be uniformly distributed. If investors spend a billion man-hours investing, they should generate, let’s say, 50 billion dollars in returns. It matter not at all whether 49 billion of those dollars go to the best 10, 100 or 10000 people — unless, that is, such a skewed distribution provokes exit from the market (in which case the marginal price of labor will increase or higher prices will induce new entry; this is, in a sense, the story of hedge funds, the transition of investors who invest as retail investors to investors who work professionally in an industry).
In other words, the ability of some people to consistently beat the market does not defeat EMH — even if those people obtain their returns from skill alone and not from luck.
2. That said, I believe only in weak EMH; I do not think semi-strong is well founded, in part because of the prevalence of asset bubbles. It might be hard to distinguish an asset bubble from a price increase at 20,000 feet, but on the ground it’s easy. Here’s one way. You know how Fabrice Tourre bragged about selling garbage to widows and orphans? Nobody says that about selling Apple stock.
Also, the failure of risk management at many financial institutions implies a failure of EMH. Again, if your feet are on the ground, you’d quickly see that EMH would have been impossible. Financial reporting was far too opaque for investors to adequately assess their risk; indeed, the opacity produced uncertainty (not just risk), and in an amount underestimated by most pundits. That may be why the collapse took many by surprise; that fact doesn’t mean that the market was efficient.
3. The truth of EMH may or may not matter all that much in economics, but it matters for regulation and even more in law. A lot of corporate law is based on assumptions of efficient markets (and diversification per CAPM) — and also, a narrow but extremely important part of corporate law that depends in part on its rejectin (the legality of the poison pill). Securities law even more so.
And of course, as Sumner argues, it has policy implications for financial regulation. Sumner has given short shrift to the policy options available to regulators if EMH is not true (e.g. one can require structural solutions that kick in when certain meta-conditions are met, without requiring regulators to foresee bubbles). And so it is quite important whether EMH is or is not true. Can we expect regulators to demand better disclosure than market participants? The answer to me is clearly yes.
31. January 2012 at 14:57
Let me introduce modus tollens to David ….
If P, then Q.
Not Q.
Therefore, not P.
31. January 2012 at 15:26
Steve Jobs and Bill Gates have no relevance to the EMH. EMH is a theory about capital markets investing. Steve Jobs and Bill Gates did not make their fortunes by investing in capital markets. They invested in non-tradeable (or non-traded, in any event) assets like intellectual property.
The EMH does not state that there cannot be any winners in industry. To the contrary, it presupposes the existence of companies like Apple or Microsoft. If there were no Apples, then that would simply manifest itself as a reduction in risk (and a decline in the return of the market portfolio, since we would be chopping off the tops of the best performers), and the the theory survives intact.
31. January 2012 at 15:35
I see no valid reason to think that the NYSE today is more efficient than it was 15 years ago (going back 30 years takes us back to fractional not decimal prices, and I’m not sure what effect that would have on market efficiency).
Liquidity is often thought to increase efficiency but that’s not generally true. I’ll grant for the sake of argument that stock markets are more efficient than, say, real estate markets (although my subjective experience is that NYC real estate is pretty darn efficient), but, in my opinion, the idea that an increase in volume from 10B to 100B (or the decrease in time to sell from 1 second to 1 ms) adds efficiency is unfounded.
31. January 2012 at 15:45
This question always bothered me too!
Until I read Howard Marks:
“I find it hard to accept volatility as a comprehensive, sufficient or highly useful measure of risk.”
He does an excellent job of dressing down this assumption here:
http://www.oaktreecapital.com/MemoTree/Risk%2001_19_06.pdf
31. January 2012 at 15:51
Greg Ransom,
I think the Fama factors are relevant to EMH in the following way. I think EMH becomes a tautology if you expand it to include non-financial returns. The Fama factors basically correlate to affiliation and reputational risks and costs. There are affiliation costs to investing in low P/E, small, low market to book stocks. And the momentum effect is capturing the transition of stocks from low to high status or vice versa. Investments can be Veblen goods.
This is why you don’t see long term outperformance among mutual funds, because these reputational and affiliation effects are important in that context, and so they are investing based on more than financial returns. Look at the characters in The Big Short. When they were right, their customers were screaming & kicking trying to get out. To get long term outperformance, you need to be a private investor or someone like Warren Buffett who has so much reputational capital that he can ignore affiliation costs. Even he saw his following fall away when he determined to stay away from the tech bubble.
Even the best investors make many mistakes, so if they are acting as a professional agent and their customers have the right to leave, a good investor who is looking for large market errors is bound to experience a lot of capital flight. The worst thing that can happen to a professional is to be an outlier who is also wrong.
31. January 2012 at 16:16
Try Rob Arnott’s RAFI on http://www.researchaffiliates.com/rafi/index.htm for a well argued theoretical basis and empirical evidence to support the case against EMH. $87bn is currently managed using RAFI methodology.
Research Affililiates argue the 2-4% excess returns from fundamental indexing compared to cap weighting arise from pricing inefficiencies. A cap weighted index will be overweight overvalued stocks and underweight undervalued ones. Weighting by other measures of size, e.g. sales, profit, NAV, break the link with price thereby avoiding pricing inefficiencies. This makes it possible to beat the (cap weighted) market. Therefore EMH is at best flawed.
‘Lecturing birds on flying’ is a more aggressive attack on quantitative methods in finance generally, EMH, Black Scholes and academic finance in particular, a la Taleb. Great fun.
There is plenty of work around looking at modelling markets in new ways.
E.G. Nature December 2011. “Networks: control of the super-corporations.” 737 companies, mainly financial, dominate (80% control) a global shareholding network of 43,000 multinationals, the links forming a butterfly shape like the www. Barclays is number 1 with 4%, and the system is very prone to contagion in crises. There’s another piece I cant find right now about the similarities in behaviour of phase shifts in e.g. earthquakes, asthma attacks and market crashes. EMH as such is not relevant now.
However it, along with CAPM etc., has been useful in justifying and enabling the management of vast amounts of money with economies of scale and sufficiently high charges to finance asset gathering commission payments. Not entirely bad if it encourages saving.
31. January 2012 at 16:23
Greg Ransom,
It’s not persuasive to criticize a theory by saying “this is different from what Hayek said was important.” Those of us who do not venerate Hayek (or any other economist) are left wondering why Hayek’s ipse dixit matters.
You’ve used a great many buzzwords and thrown around a number of demeaning characterizations. What you haven’t done, however, is provide any argument that actually addresses the EMH or Scott’s comments about it. I’ve read the Big Short. It doesn’t help your case very much. You can’t disprove EMH by showing that some trades were very profitable. Nearly all financial trades prove profitable for one side or the other. That the winners “knew what they were doing” tells you only that it’s the winners who write history.
It’s a lot more interesting to look at the losers or the early winners, and what they knew. When people are buying into securities because of an expectation of non-fundamental price appreciation, when they are chasing yield to keep up with the Joneses, when they are making economic decisions under moral hazard conditions (the important moral hazard problem not coming from the government but rather the option-like bonus structures that private firms gave to their employees) . . . that tells you a lot more about EMH.
31. January 2012 at 16:58
Scott’s already admitted he’s wrong.
“Morgan, I agree that insiders might have an advantage (haven’t studied Congress.)”
That’s all you need to defeat EMH.
The insider advantage is the whole point of Libertarian economics, at any one moment finding / having specialized knowledge that you can use to take advantage of someone else’s (including the whole market) factual ignorance – whether in a trade of business strategy.
One of you was ACTUALLY correct, and whether it is from a closed Congressional hearing, or your unique personal history as a plumber with a brother who runs boat cleaning company – what you use to invest or invent is inside information.
The only other question is whether over time, certain people tend to HAVE and ACT ON more insider information…. and the upside to that info.
Bill, this is neat, but non-responsive:
“But there’s nothing that says that the returns have to be uniformly distributed. If investors spend a billion man-hours investing, they should generate, let’s say, 50 billion dollars in returns. It matter not at all whether 49 billion of those dollars go to the best 10, 100 or 10000 people “” unless, that is, such a skewed distribution provokes exit from the market (in which case the marginal price of labor will increase or higher prices will induce new entry; this is, in a sense, the story of hedge funds, the transition of investors who invest as retail investors to investors who work professionally in an industry).”
No once cares about the aggregate, the point here is that people with the most access to inside information (intelligence) and the most willingness to act on it – will tend to recover MORE of the returns than those without it.
31. January 2012 at 17:01
[…] Source […]
31. January 2012 at 17:08
BW:
“Steve Jobs and Bill Gates have no relevance to the EMH. EMH is a theory about capital markets investing. Steve Jobs and Bill Gates did not make their fortunes by investing in capital markets.”
*Sigh*, both Jobs and Gates owned HUGE quantities of stock in their companies. They were both investors!
Yes, they did make their fortunes by owning stock that appreciated in value, on account of their own efforts, their own abilities and knowledge.
You see, this is the flaw in EMH. It doesn’t even consider individual investors taking advantage of market “inefficiencies” in making their fortunes.
Market inefficiency is not a bad thing. It is inherent in a division of labor economy. Along with division of labor, there is specialized knowledge, and specialized knowledge means there are always opportunities for specialists to “exploit” information asymmetries. I don’t know why small government, market oriented economists are so shaken by market inefficiencies. It’s not like it ipso facto implies central planning is warranted.
31. January 2012 at 17:39
Why do EMH posts always garner so many comments? It’s the same argument over and over…
BTW, you write:
“I’m still looking for the model that will tell me how to beat the stock market.”
Realize that this cuts both ways. If the EMH is right, then every model you can imagine is inferior. If the EMH is wrong, then a model could be right but there’s no reason the market would validate it (except over the very long term). So, 2 years of data is… more proof the EMH is wrong. 🙂
What is the point of this argument again?
31. January 2012 at 18:40
It is almost impossible to prove or disprove the EMH IMO. First, the markets where most people “test” the EMH are those where there is a lot of information (e.g. stock, bond markets) – so almost by definition one will not find much inefficiency there (clearly there is a research bias towards markets where researchers can get more info). Second, outcomes are probabilistic, so to test the EMH often requires some counterfactual analysis. For example: M&A. Suppose company A is taking over company B. Very often, there be a spread between the takeover price for B and the market price. Is that inefficiency? No: the outcome is binary (either company B will be taken over with regulatory approval, or not, as in the recent case of AT&T & T-Mobile). And, if the takeover does not happen, sometimes company B will be worth less than prior to the announcement (because a strategic option is foreclosed). The rational market price is a probability-weighted average of the two binary outcomes, but there is no way to know if the market price is “correct” unless one runs an experiment 1000 times (which you can’t do). The best one can do is backsolve (assuming rationality/EMH) and determine the implicit probability of takeover, and decide if it seems reasonable. Same with the interest rate on soverign italian debt. Is 5.9% reasonable? Sure, a bank can borrow at 1% and buy at 5.9%, but there is still some probability of catastrophe….
Now, there are some exceptions: There are certain markets where there are structural impediments (often regulatory). If one builds a power plant in a deregulated market like New England for example, one has 30 years worth of electricity to sell (and 30 years worth of fuel like gas or coal to buy) but (for various regulatory reasons) no one will buy electricity more than about 36 months at a time. To get a 30-year firm fuel supply you really have to own a mine or some gas wells and a pipeline, a producer will not commit that far out. A bank will only lend against “locked-in” cashflows unless they have access to other collateral… So because of the structural assymmetry between buyers and sellers the “market price” in many markets is not the “EMH price”. (for example, the bid-ask on a power plant is about 30%!) I have seen some analysis that claims that this is a violation of the EMH, but in my opinion this is just poor market design (and you can’t really take advantage of this either because there is no good risk-free way to make a trade, and electricity is not very storable).
that said, i think the onus is on someone to disprove the market price is unreasonable. There is usually a very good reason for a “pricing anomaly” and one has not dug deep enough into the risks. If its too good to be true it usually is. In a simple-minded sense, it is true that I can name many markets that are incomplete and so one cannot hedge away many types of risks. But: i do not necessarily think that violates the EMH per se (does it?) – usually the market price reflects those issues and one cannot capture any “excess” return without lots of risk.
31. January 2012 at 18:53
“One of my least favorite maxims is; “the market can stay irrational longer than you can stay solvent.” I consider that to be a cop out for losers.”
This comment insults most reasonable people’s intelligence. Towards the end of 2009 I was shorting the market only to be stopped out after painful losses. Who would have thought that the US government would happily be spending 1.6 trillion more than it brought in to keep the economy afloat. Who exactly knew what the FED was doing, minus the major players in the market, to keep everything from falling apart.
31. January 2012 at 19:13
Where did you get the 1 out 20 anomalies? Is that based on a 5% significance level?
31. January 2012 at 20:06
I figured I would chime in for what I imagine is the silent majority that supports EMH as useful: you go Scott!
When talking with anti-EMH folks, I always find it useful to avoid a binary discussion of whether EMH is true or false. Instead, it seems more productive discuss what types of information they don’t think are incorporated into prices, how long it takes new information to get incorporated, and how long arbitrage opportunities exist before everyone else catches on–all for a given market.
BTW, I find it amusing that Morgan think the existence of insider gains disprove EMH when one of the primary distinguishing characteristics of the different strengths of EMH is whether insider information is incorporated in prices.
31. January 2012 at 20:19
Kevin, one more time, please pay attention, the idea that “the market” including inside information is the best possible mind of god snapshot of information at any one moment, or that this process clears itself going forward into the end of days is NOT in dispute.
MY point is that EMH allows for certain individuals to “at any one moment” to the be windfall recipient.
And that certain individuals put themselves in this position more than other individuals goinf forward into the end of days, and what they do to accomplish that is not LUCK.
Take “intelligence is genetic” and admit that smarter = wins more often, AS FACT.
And from there, EMH is an uninteresting statement about the wisdom of crowds, it never turns the corner and says “batting over .500 till you’re dead isn’t a skill.”
I don’t have to outrun the bear, I just have to outrun you.
31. January 2012 at 20:21
Scott you said:
“Useful for what, you never say, which I think is a cop out?”
“I find these sorts of comments to be really annoying. All of market monetarism is based on the EMH. The EMH told me monetary policy was disastrously off course in late 2008.”
I guess that makes us even. I suspected, of course, that you were implying something like that, so the implied hypothesis is Hedge Fund managers aren’t very good at their jobs; therefore the S & P 500 is a good indicator of monetary policy (if one accepts your definition).
Perhaps it is, but nothing in your post supports that. If you want to make that argument the burden of proof is on you to show that relationship.
31. January 2012 at 20:42
This is not that simple. Steve Jobs invested huge relative sums of his personal wealth and made huge bets.
“Steve Jobs and Bill Gates have no relevance to the EMH. EMH is a theory about capital markets investing.”
And Scott does not use the EMH for a restricted purpose, he uses it to make immense claims about the causal process and causal structure of the market.
31. January 2012 at 20:43
“When the price of a stock can be influenced by a ‘herd’ on Wall Street with prices set at the margin by the most emotional person, or the greediest person, or the most depressed person, it is hard to argue that the market always prices rationally. In fact, market prices are frequently nonsensical.” –Buffett
Here is the lie: It’s wasted time trying to outperform the market.
Each of the three analysts (Schloss, Knapp, Ruane) that worked at Graham-Newman with Buffett in the mid-1950’s had a Buffett-like record, and they all used the same investing framework investing in different securities. How could that be efficient?
Economists can’t deal with inefficient markets, because they are boxed into a world of assumptions (ration trumps emotion and bias), models (most things are too complex to model), and equilibrium (never actually get there). To an economist with a hammer, everything looks like a nail.
31. January 2012 at 20:45
BW, that’s not what I’m saying. I’m merely identifying a causal domain of reality that no one actually denies by pointing to the scientific giant who called everyone’s attention to that causal domain — sort of like mentioning Newton to remind everyone of the causal role of gravity …
BW wrote,
“It’s not persuasive to criticize a theory by saying “this is different from what Hayek said was important.” Those of us who do not venerate Hayek (or any other economist) are left wondering why Hayek’s ipse dixit matters.”
31. January 2012 at 21:03
“The EMH defined as a no-arbitrage argument a la Jensen is hard to refute, it’s very strong on simple “so why isn’t anyone else picking up the $100 bill” grounds.”
I worked at a trading firm that, year after year, made great profits doing arbitrage. There are continuous arbitrage opportunities, and the reason you weren’t seeing them is we were scooping them up before you noticed them. They would exist for times between a couple of seconds and a minute or two.
31. January 2012 at 21:07
david,
“… nope. Grossman-Stiglitz for distributed information without EMH. Any standard perfect-information market model for EMH without distributed information. ”
Other models exist, of course. That’s one of the problems with economics, other models always exist that explain the same datum with similar predictive power, or absence thereof.
“You can earn supernormal profits without going on to own the universe, by simple token of consuming your abnormal returns instead of reinvesting them.”
Point taken, but in the context of capital markets some kind of compounding is a reasonable assumption.
Foosion,
as another poster noted, the existence of Buffett doesn’t preclude the EMH as long as 1- he’s balanced out by other investors that are equally unskilled as he is skilled and 2- he doesn’t always exploit the same price “anomaly”.
And Buffett or any number of examples show that some people accumulate wealth far beyond personal consumption needs.
31. January 2012 at 21:24
What the debate is NOT:
It’s not a debate over whether a a horse-blindered focus on a data stream of quarterly report numbers from the past can allow one a statistical procedure to beat other investors.
As far as I can tell, what “EMH” party-liners are claiming is that to any challenge to the rhetoric of the “EMH”ers is a commitment to the the opposite assertion, i.e. a commitment to the claim that folks can come up with a statistical procedure (applied to a number series generated by a stack of quarterly reports) for “outpicking” the market . But that is not at all what “anti-EMHers” are claiming — in large measure because what the “EMH” ideologues what to claim for their faith goes complete beyond the horse-blindered data and statistics from which the ideology came, i.e. the history of the EMH is just the history of applying statistics to simple streams of standard “investor” numbers to make future number stream projections, and drawing inferences from that. But the claims for EMH go completely outside that very limited and blinkered base of data.
What the debate IS:
The debate is about all of the immense and ungrounded implications for the causal structure and process of the market which Scott draws from verbal definitions of “EMH”, implications and definitions conceived completely OUTSIDE of and conceptually beyond the context of the horse-blindered number sets and statistical findings that in fact historically and practically ground Fama’s conception of the EMH.
31. January 2012 at 21:30
Gates and Jobs weren’t investing in their companies as capital markets investors. Their equity stakes came from 1) having founded the company and 2) being compensated in equity by the board of directors. The EMH does not say that people cannot get rich from entrepreneurship.
31. January 2012 at 21:44
BW we’ve covered this… you are being non responsive.
31. January 2012 at 21:49
The history of the EMH isn’t much more than this.
Statisticians looks at relations between some very simple past numer streams generated by the stock market and corporate quarterly reports.
They didn’t see much opportunity using those statistical patterns to make money in the stock market using those numbers and patterns.
Then better mathematicians came in and found tiny and temporary patterns that did allow for profitable arbitrage — and the Quants were born.
And the folks who said that the statistics didn’t show any opportunity for “beating the market” pointed to what was happening and said, “Look, any opportunity for beating the market using statistics will be used, so any opportunity for using statistics to beat the market will find itself eliminated.”
But note well, in all of this, we have never left the Flatland world of statistical relations between a very small horse-blinded set of number streams.
“EMH” ideologues are hardly more than residence of Flatland who illegitimately and cluelessly insist that the truths about Flatland apply to the fully-dimensional real world.
31. January 2012 at 21:58
Jobs sold all of his Apple stock in 1985 for a 1/4 billion $$$ and then bought one other company and financed the creation of another.
Spending a 1/4 billion $$ on a manufacturing plants, offices and employees is a capital investment, and is a form of competing in capital markets, with an aim for selling future stock in capital markets.
Again, the silly game of at one moment restricting the EMH to a small data set of stock and quarterly report streams and then in the next moment applying the EMH to each and every corner of the whole of the market process is a mugs game, and needs to be exposed for what it is.
BW wrote:
“Gates and Jobs weren’t investing in their companies as capital markets investors. Their equity stakes came from 1) having founded the company and 2) being compensated in equity by the board of directors. The EMH does not say that people cannot get rich from entrepreneurship.”
31. January 2012 at 22:17
Anyone watch a 3 cups and balls trick preformed by a magician?
That’s what this EMH stuff is — it’s a misdirection game.
And the big switch — made with classic magician’s misdirection — is made moving without justification from an ostensively defined concept (e.g. pointing to a set of statistical facts about relations between a small stream of financial and investment numbers) to fantastic claims about the causal mechanism of the market.
And the magician’s misdirection — the magic trick — is made by illicitly pretending to say just one thing while saying two insuperably incompatible things using two different and non-univocal senses of the term “information”, e.g. “information” as the statistical relations that can be pulled out a horse-blindered and limited data stream of numbers, and “information” as every alternative, open-ended, growing (and often rivalrous) understanding of the the world than anyone can possibly have from any and every possible perspective.
31. January 2012 at 22:37
Ransom we won on any flow of argumentation the opponents are stuck. They have only one dumb response: luck
The corner cannot be turned – arguing the aggregate doesn’t nullify that someone has to win and that doesn’t happen randomly.
You are exactly right the reality is escalation- the old get weak and die the young are smarter and outwit and conquer that fact is what proves EMH wrong because evolution Darwinism explains the phenomena scott is promoting
31. January 2012 at 23:14
[…] circularity posing as profundity from Scott Sumner on the Efficient Markets Hypothesis (EMH): The past five years should have been an absolute gold […]
1. February 2012 at 00:15
To specifically answer the point on Paulson, his poor returns may falsify the stance that it’s easy for hedge funds to beat the market. But they certainly don’t prove the other way, that market prices are always fundamentally correct based on all publicly available information. What set of circumstances were housing prices in Las Vegas or the pets.com stock price reflected their fundamental valuation? The same goes for CDO’s which only needed about 10% of underlying subprime mortgages to default.
To be honest, this is the infuriating thing. It’s clear that the proposition “market prices give fundamental evaluations reflecting all publicly available information” has clearly been falsified. There simply isn’t a single set of circumstances where these asset prices would have realized that long-term NPV. Unlike market prices, NPV can actually be actuarially calculated and when no acturarial calculation exists to justify a certain asset valuation, it’s safe to say that the asset price was not reflecting the fundamental value.
To get around how the central proposition of the EMH has been falsified, I see EMH proponents going in all sorts of other directions and strawmen arguments, such as how tough it is to beat markets. True, a falsification in and of itself may not have much use, but the falsification should give some pause as to whether completely unfettered free markets always do the best job of asset allocation. That’s really a completely different and much more complex discussion. The falsification is the easy part.
1. February 2012 at 04:50
I thought this was one of your worst posts. You say that EMH has 5% anomalies but then you say that criticism is wrong because it is not 100% true. Really! You are beginning to sound like other economic bloggers that what you believe is correct and others must be wrong because you are right.
1. February 2012 at 05:01
Re: dual listed comanies
Scott, as for the reasons why anyone would buy the more expensive ones, I really don’t know. It could be that (for example) Dutch investors just don’t want to invest in UK companies and so if suddenly more Dutch people invest in stocks, Royal Dutch temporarily moves up. It all sounds pretty ad hoc though.
Dual listed companies do seem to be the most persuasive piece of evidence against EMH that I’ve seen, and I’m not sure why they aren’t brought up more. If you had two funds, one that only bought more expensive pairings and one that only bought less expensive ones, and they both held on to them forever, so that their benefit was the actual dividends, we can guarantee which one of them would perform better.
As for arbitrage, Wikipedia tells us basically the dread phrase that “the markets can remain rational longer than you can remain solvent”:
“An important characteristic of DLC arbitrage is that the underlying shares are not convertible into each other. Hence, risky arbitrage positions must be kept open until prices converge. Since there is no identifiable date at which DLC prices will converge, arbitrageurs with limited horizons who are unable to close the price gap on their own face considerable uncertainty. De Jong, Rosenthal, and van Dijk (2008) [4] simulate arbitrage strategies in 12 DLCs over the period 1980-2002. They show that in some cases, arbitrageurs would have to wait for almost nine years before prices have converged and the position can be closed. In the short run, the mispricing might deepen. In these situations, arbitrageurs receive margin calls, after which they would most likely be forced to liquidate part of the position at a highly unfavorable moment and suffer a loss. As a result, arbitrage strategies in DLCs are very risky, which is likely to impede arbitrage.”
And as an example, also from wikipedia:
“Lowenstein (2000) [6] describes arbitrage positions of Long-Term Capital Management (LTCM) in Royal Dutch/Shell. LTCM established an arbitrage position in this DLC in the summer of 1997, when Royal Dutch traded at an 8 to 10 percent premium. In total $2.3 billion was invested, half of which long in Shell and the other half short in Royal Dutch (Lowenstein, p. 99). In the autumn of 1998 large defaults on Russian debt created significant losses for the hedge fund and LTCM had to unwind several positions. Lowenstein reports that the premium of Royal Dutch had increased to about 22 percent and LTCM had to close the position and incur a loss. According to Lowenstein (p. 234), LTCM lost $286 million in equity pairs trading and more than half of this loss is accounted for by the Royal Dutch/Shell trade.”
1. February 2012 at 05:37
[…] Sumner defending Efficient Market Hypothesis.In response: […]
1. February 2012 at 07:24
Then it’s not a corollary! It does not follow; in fact it is neither sufficient nor necessary! Aaaaah!
FWIW the Grossman-Stiglitz intuition moves in the opposite direction of Hayek; the more valuable the information that the price mechanism reveals, the worse the price mechanism is at allocation and we should really resort to a central Walrasian auctioneer instead. And then Oskar Lange rises from his grave to point and laugh.
1. February 2012 at 08:51
If EMH was simply the statement that it’s very hard to outperform the market over a long enough period of time, I don’t think many people would have a problem with it. It’s rather the assertion about market prices instantaneously reflecting all relevant information that stretches one’s credulity. One doesn’t need to believe in the second statement to believe in the first.
1. February 2012 at 09:11
Let me recommend this post by Gene on categorically different kinds of understanding of particulars of place and time.
http://gene-callahan.blogspot.com/2012/01/why-no-model-can-beat-market.html
Scott is you typical 1950s scientistic social scientist’s false view of how humans are situated in the world and how they know the world — the sort of crude scientism / poaitivism that is _always_ joined at the hip with “pragmatism” and relativism.
1. February 2012 at 09:35
@Jimo: How often will we reject a null hypothesis (P=0.05) if the null hypothesis is true?
@Greg: Beethoven’s symphonies show serial correlation. That is, after hearing one of Beethoven’s symphonies, we expect that another Beethoven symphony will be better than average. The same is true of free throw percentages, marathon times, SAT scores, IQ scores, etc. This is because these contests measure something besides luck. If skill or judgment create opportunities for higher average return, EMH critics *must explain* why skilled investors with good judgment on Wednesday predictably turn into average-skilled investors with average judgment by Thursday.
1. February 2012 at 09:40
@Morgan. I find it amusing that your response to my observation about your position on insiders is to succinctly restate the EMH itself.
So your problem with EMH is that it only applies to the 99% of people who don’t have any reasonable priors that they have access to very special information?
If all we’re arguing about is the 1%, I’m happy to say “Peace, brother”.
1. February 2012 at 09:44
John, I wouldn’t articulate the point the way Gene does, but the distinction stands.
1. February 2012 at 09:47
This isn’t true:
“EMH critics *must explain* why skilled investors with good judgment on Wednesday predictably turn into average-skilled investors with average judgment by Thursday.”
Koch can see the opportunity of the McDonald’s brothers opportunity, and never see another opportunity again.
1. February 2012 at 10:42
@Greg:
EMH: “Our opinions about the market being over- or under-valued are wrong.”
You: “No, sometimes our opinions are right–but only in extremely rare, literally once-in-a-lifetime cases.”
Seems to me that we agree: EMH isn’t “true,” but it delivers accurate predictions 99.9% of the time.
1. February 2012 at 10:52
John,
The reason why skilled investors’ returns may vary is that the absolute skill of an investor is not the real variable. The variable is the skill of investors relative to all other investors.
If the market as a whole becomes less rational, due to liquidity constraints, asymmetrical information and other market imperfections, then an investor of the same absolute skill will see higher returns.
It’s how markets can and will, as a whole, change their collective level of rationality is the great use of anti-EMH arguments. The semi-strong form of EMH says that unfettered free markets will always be the most efficient at allocating capital. Seeing the heights of the housing and tech bubbles should give that view some pause.
1. February 2012 at 10:58
Of course, that’s not what I’ve said. Nice try. This sort of game — the game of willfully violating the principle of charity — can go on endlessly.
John S.:
“EMH: “Our opinions about the market being over- or under-valued are wrong.”
You: “No, sometimes our opinions are right-but only in extremely rare, literally once-in-a-lifetime cases.””
1. February 2012 at 10:59
As a layman (i.e., I am not an economist) who used to “believe” in EMH, a few observations from this bruised and battered and often-humbled investor:
1) If your goal is to achieve index-like returns, then invest passively in that index.
2) Timeframes matter–anyone can look like a genius or a fool, depending on what timeframe is picked. The critical question is what timeframe is appropriate for you? John Paulson’s Advantage Fund, for instance, through 2011 has underperformed the S&P 500 six out of the last ten years, but over that same ten year period, that fund outperformed the S&P by 66%. Is Paulson a genius or a fool? Depends on when you gave him money.
3) Some people are really good investors, they are not difficult to find, and they usually don’t give a hoot about indices and relative performance and arbitrary time frames–where is it written that you must have positive performance in a calendar year?
4) It is wise to emulate the good investors, and not necessarily blindly invest in their flagship fund. For instance, John Paulson very publicly invested all of his own money in his gold fund, which has since greatly out-performed his flagship fund. (I risk sounding like a shill for Paulson–I have never met the man nor placed money with him. I just think Scott and others have uncharacteristically taken a cheap shot to make a point.)
Thus, my experience has taught me that EMH is a worldview and not a workable investment model as it has no predictive ability in the capital markets–you are just simply along for the ride. Graham and Doddsville, though, has been a much more rewarding investment model.
1. February 2012 at 11:04
Look, the issue isn’t investors & the stock market.
Scott is making claims about the structure of production across time and the relation of constantly adapting and learning individuals working from particular perspectives in the overall market order on the basis of local knowledge and ever evolving perceptual skills, etc. This goes completely beyond the stream of numbers looked at in the “Random Walk” literatue, etc., which serves as the class of entities providing the ostensive definition / cognitive content of Fama’s EMH.
1. February 2012 at 11:10
All it took for the original “Quants” to make millions was just a bit of background understanding of financial markets and a tiny expansion of the class of data streams beyond that found in the original”Random Walk” literature, an expansion of data sets suggested not by math alone, but in the first instance by their background understanding.
1. February 2012 at 11:18
Scott, are there any good arguments for extending EMH beyond the data streams of the “Random Walk” literature which ostensively define and gave cognitive content to Fama’s EMH?
Other, than how “useful” it is to misuse the EMH to confuse people about rival causal explanations of macroeconomic patterns that lie completely beyond Fama’s EMH data set and story.
1. February 2012 at 12:26
123, But the TIPS market doesn’t give you an expected NGDP reading.
Major Freedom, You said;
“Suppose I am an individual investor, and for 50 years, say from 25 to 75, I earn more than the market average. I amass a fortune. Then in my old age I begin to lose my touch, my abilities wane, my knowledge isn’t so superior, and over the next 2 years I remain stubborn, I refuse to retire, and I end up losing everything I earned.
Would this be an example that confirms EMH, or would it be an example that violates EMH?”
It tells us nothing about the validity of the EMH, because you’d expect people like that regardless of whether it’s true or not. Not a lot of people, but some. Which is what we observe.
Greg Ransom, I think the EMH is useful, and I think we have an explanation of why it works.
MP, Good point.
Greg, I read something about that a long time ago, but don’t recall the specifics.
Arthur, No, that’s not true. The EMH doesn’t prevent people from making repeated errors.
Rafe, I don’t have time to read many of the links, what’s a quick summary of the conclusions?
BW, I focus mostly on the weak form of market efficiency. BTW, many government regulations are inconsistent with market efficiency.
I agree about Gates and Jobs, and I agree about the market not getting significantly more efficient.
Giles, Anomalies don’t really tell us anything about the EMH. There’s no proof they will continue to hold in the future.
Morgan, No, the weak and semi-strong versions do allow for insider trading to work.
statsguy, You said;
“Why do EMH posts always garner so many comments? It’s the same argument over and over…”
If you look closely I’m not being repetitive. Each time I swat down one additional argument. When I used to point to mutual funds my commneters would throw hedge funds back in my face. Now I’m dealing with that.
dwb, I agree.
GetReal1, You said;
“This comment insults most reasonable people’s intelligence. Towards the end of 2009 I was shorting the market only to be stopped out after painful losses. Who would have thought that the US government would happily be spending 1.6 trillion more than it brought in to keep the economy afloat. Who exactly knew what the FED was doing, minus the major players in the market, to keep everything from falling apart.”
I would have thought that. That’s why I was long, and made lots of money. I think people who are wrong tend to blame it on market irrationality, not their own flawed model.
Jim, Yes, 5% is the standard used in anomaly studies that I’ve seen. These studies are literally meaningless.
Thanks Kevin.
OGT. Notice how I said nothing about stocks, and you simply assumed that’s what I meant? There are other markets. Unfortunately we are short one badly needed market–NGDP futures.
Crush, If they all used the same investing framework then I’m not surprised their returns were correlated.
more to come later today . . .
1. February 2012 at 12:47
Scott,
The Colorado professor showed that two assets, each with negative expected value over time can be combined in a portfolio with rebalancing to produce a positive EV.
Discovered in 1996…
“Parrondo’s paradox is the well-known counterintuitive situation where individually losing strategies or deleterious effects can combine to win.
…
Over the past ten years, a number of authors have pointed to the generality of Parrondian behavior, and many examples ranging from physics to population genetics have been reported.
…
In its most general form, Parrondo’s paradox can occur where there is a nonlinear interaction of random behavior with an asymmetry, and can be mathematically understood in terms of a convex linear combination.
…
Many effects, where randomness plays a constructive role, such as stochastic resonance, volatility pumping, the Brazil nut paradox etc., can all be viewed as being in the class of Parrondian phenomena.”
1. February 2012 at 12:53
Matt, SAT scores are standardized–scores are dependent on your intelligence relative to all other test-takers. Yet despite this we still see serial correlation in SAT scores.
Random variation in returns will reduce serial correlation–you need to show why they’ll eliminate it entirely.
Greg,
First, I understand that the issue goes way beyond investors and the stock market. But that does not make evidence from the stock market irrelevant. Debates about evolution vs. creationism are about far more than the fossil record. Your theory has implications. If the evidence doesn’t match, you need a convincing explanation or you need to change your theory.
Second, I am sorry if I misunderstood your point. I asked you to explain why returns are not serially correlated if skill affects returns. You responded by saying that Ray Kroc only got one chance to buy McDonald’s. I interpreted this example as a claim that we very rarely get opportunities to use skill to identify opportunities. Otherwise I just don’t see how it has any relevance to the lack of serial correlation in the data. But if I’m just not getting the point, I’m honestly curious.
If making the right call one time is not statistically related to making the right call the next time–for whatever reason!–I simply don’t see how I can have any confidence in my opinion that the market is under- or over-valued. Why should I be the exception to the rule? If insanity is doing the same thing repeatedly while expecting different results, what is it when I do the same thing as a few hundred million other people while expecting better results?
1. February 2012 at 13:16
ssumner:
I said: “Suppose I am an individual investor, and for 50 years, say from 25 to 75, I earn more than the market average. I amass a fortune. Then in my old age I begin to lose my touch, my abilities wane, my knowledge isn’t so superior, and over the next 2 years I remain stubborn, I refuse to retire, and I end up losing everything I earned.
Would this be an example that confirms EMH, or would it be an example that violates EMH?”
You said: “It tells us nothing about the validity of the EMH, because you’d expect people like that regardless of whether it’s true or not. Not a lot of people, but some. Which is what we observe.”
But you just said before that EMH would be falsified if there are investors who earn persistently higher than average returns.
I just gave you an example of an investor who earned persistently higher than average returns, for 50 years, and you’re saying it says nothing about EMH?
I’m speechless.
1. February 2012 at 14:13
John S., it can by a unique historical situation, or it could be a series of opportunities, and it could include failures, i.e. what I’m talking about isn’t stuck in this box:
“sometimes our opinions are right-but only in extremely rare, literally once-in-a-lifetime cases.””
1. February 2012 at 14:15
John S., a counter-example counts as a counter-example even if it doesn’t typify all possible counter-examples.
A simple logical point.
1. February 2012 at 15:14
Let’s assemble a few more reminders so that we get as clear as we can about what is going on in this discussion of EMH.
It cannot be emphasized enough that the EMH got its cognitive content by ostensive definition — by pointing to a very small and horse-blindered stream of numbers and some very simple statistical reflection on those numbers (e.g. price to earnings ratios and stock prices across time).
When Fama talks of the EMH, and “the ‘information’ available to an investor” in the first instance he has in mind is this bare bucket of numbers and relation between numbers, with the barest possible additional cognitive content added to the mix (e.g. constricted concepts of ‘profits’ and ‘earnings’ and ‘prices’). That’s where the whole literature comes from.
So, when someone says that the EMH tells us that:
“An investor cannot consistently achieve returns in excess of average market returns given the information available at the time the investment is made.”
we need to make sure we aren’t confused. “Information” just means a horse-blindered and very select stream of common financial numbers related across time. It doesn’t mean what the ordinary person means. That is, the work the word “information” is doing here is very different work that it does in ordinary language.
The sentence more properly should read:
“An investor cannot consistently achieve returns in excess of average market returns given a thin stream of numbers found in stock prices and earning reports or similar simple financial numbers available at the time the investment is made.”
That is exactly the bucket of “information” and statistical findings that the conclusion derived from those “data” which provides the EMH its cognitive content.
You can’t rely on the use statistics and price/earnings ratios and stock price data streams (and conceived of from the point of view of a vacuum) to beat other investors.
The challenge is list the premises and explain the justification for those premises that take you from that type of case to completely different cases which don’t satisfy those conditions.
What we don’t want is conceptual confusion and ambiguity in the attempts to do this — we don’t want people failing to perceive when they have left the simple Flatland of a tiny set of horse-blindered number streams, and entered into the wider world where folks have alternative, unique, and constantly updated understandings of the opportunities in the world.
My perception is what we get is a repeated mistaking of the features and conceptual categories of Flatland for the open-ended and ever changing features, conceptual categories and opportunities of the real world. Profit opportunities and how our understanding allows us to achieve them has a different aspect in these two different world.
1. February 2012 at 15:15
“Matt, SAT scores are standardized-scores are dependent on your intelligence relative to all other test-takers. Yet despite this we still see serial correlation in SAT scores.
Random variation in returns will reduce serial correlation-you need to show why they’ll eliminate it entirely.”
The intelligence of other test takers is also pretty much constant. So if someone has a high SAT score, they will still likely test in the same percentile years down the road.
That’s not the case with the market, for various reasons. People with money ultimately pull the trigger on whether to buy any sort of security, but the people with money are not necessarily the same people who can do complex financial analysis. Nor are they the same people who can pick people who can do complex financial analysis. The huge principal-agent and information asymmetry problems means that money-holders have to drastically simplify how they score managers and how they approach money management in general.
There’s also the pesky problem that there is, in fact, a lot of serial correlation when one looks at longer time-horizons. Buffett, for example, has done worse than the S&P in a number of years, but has beat the S&P 500 in every five year period since the 60’s. The time-horizon issue is exactly what you would expect when most investment managers are evaluated yearly, even though investors usually have longer time horizons.
The time horizon issue is also exactly the reason for dual-listed companies and closed-end mutual funds to have issues in their price which clearly do not correspond to the underlying fundamental value. Dutch Shell should nominally pay dividends to both listings equally, but in reality no investor has the time horizon to realize the arbitrage gains from buying one and shorting the other. LTCM arbitraged the difference between 29.5-year rates and 30 year rates, and we saw how that worked out for them.
1. February 2012 at 15:25
Matt, You said;
“To get around how the central proposition of the EMH has been falsified, I see EMH proponents going in all sorts of other directions and strawmen arguments, such as how tough it is to beat markets. True, a falsification in and of itself may not have much use, but the falsification should give some pause as to whether completely unfettered free markets always do the best job of asset allocation.”
The EMH tells us nothing about whether unfettered free markets are desirable.
jimo, You said;
“I thought this was one of your worst posts. You say that EMH has 5% anomalies but then you say that criticism is wrong because it is not 100% true. Really! You are beginning to sound like other economic bloggers that what you believe is correct and others must be wrong because you are right.”
It might be my worst, but you misunderstood my argument. I’m saying that published anomaly studies that claim to refute the EMH do nothing of the sort, because the EMH predicts that lots of anomalies will occur. Which part of that do you disagree with?
Rob, I’d need to know why Dutch investors don’t buy UK Shell stock before hazarding an opinion. Has anyone asked them?
LTCM made the mistake of being too leveraged. If you do that there is a chance you’ll fail. But suppose you create an “Anti-EMH mutual fund,” that does nothing more than invest in mispriced assets like the Shell oil example, and lots of other unrelated mispriced assets. Shouldn’t those mutual funds out-perform the market over time? Especially if they owned many different assets? Perhaps the inefficiencies are to small to overcome the expense ratio.
o. nate, I don’t think that second statement is precisely true, but I think it’s a good approximation of reality. But again, I think it’s wrong to consider this from the “is the EMH true?” angle. The important question is “are anti-EMH models useful?”
AlanMB, I can’t see how any of those life experiences would make you skeptical of the EMH. We know Paulson had some great calls–that’s consistent with the EMH.
Greg, Not all markets are equally efficient. It depends on factors like liquidity. Housing is less efficient than stocks, as it’s hard to sell houses short.
1. February 2012 at 15:30
Rafe, Why does the combined value become positive?
Major Freedom, What I meant was that it would be falsified if, on average, investors who did well in the past were more likely to well in the future. I may have worded my response poorly.
1. February 2012 at 15:37
“An investor cannot consistently achieve returns in excess of average market returns given a thin stream of numbers found in stock prices and earning reports or similar simple financial numbers available at the time the investment is made.”
This is a very good way of stating the extent of the EMH as it exists today. Maybe a better way is to also add “when one has to be fully invested in US securities and has a time horizon of one year.”
If you have these constraints, as virtually every mutual fund and most hedge funds have, then very few investors will beat a dart board consistently. If one actually reads what Warren Buffett, Seth Klarman, and others say, they would be the first to admit that stock-picking over short time horizons is not worth the transaction costs.
But these studies and constructs do not admit an option other than picking an index fund or a managed, fully invested mutual fund. There’s the option of investing in alternative assets, where even EMH purists would say investors with very long time horizons can make outsize returns due to their huge liquidity risks and other issues. “Can” doesn’t mean that every alternative investment will do well. Plenty fail. Just that there are more opportunities for finding fundamental undervaluations.
But there’s also the option to walk away from the stock market and invest in bonds or even hold cash. In 1999, Buffett wrote an article saying that the stock market could never justify its current levels, under any plausible combination of interest rates and future corporate profit growth. Then he similarly wrote an op-ed to buy stock in late-2008, for similar reasons of future cash flows. This time the bubble was in bonds and not stocks and investors were undervaluing future cash flows from stocks.
In both cases, Warren Buffett was long before the actual peak and trough of the market. Also, somebody taking his advice would have done poor relative to other market actors in a short time frame. In 1999, stocks did better than bonds for another year. In late 2008, stocks didn’t bottom out until 2009 and in the meantime treasuries would have done better. But over five years from when Buffett said to move into and out of stocks due to fundamental cash flows, the investor following his advice would have done better than the investor trying to maximize one-year, relative returns.
1. February 2012 at 15:55
“The EMH tells us nothing about whether unfettered free markets are desirable.”
I hate to sound cross, but then what does the EMH say, exactly? What is the precise, falsifiable proposition of the EMH that has been empirically proven? As far as I can tell, it only goes as far as saying beating the market over a relatively short time horizon is incredibly difficult since the market doesn’t leave 100 dollar bills lying on the ground. But if you extend the time horizon, or if you say that the market’s prices always reflect all publicly available information, then it has been falsified.
The last assertion, that prices reflect all publicly available information, is particularly dangerous. If it were true, then any sort of government regulation or action would be nothing but wasteful overhead. But if it’s not true, then the proper level of government involvement in capital allocation is somewhere between circa-1900 libertarian markets and communism. That brings about a lot of cognitive dissonance in libertarians, including myself, but it’s also true.
I also think the anti-EMH view informs personal investing, as I say above, but that’s far less direct of a point. The main thing there is that investors should at least think through the fundamental aspects of investing and not just plow their money into index funds under EMH dogma.
1. February 2012 at 19:15
ssumner:
“Major Freedom, What I meant was that it would be falsified if, on average, investors who did well in the past were more likely to well in the future. I may have worded my response poorly.”
I don’t blame you for any wording. It’s the theory that is poor. It can’t help but be explained in a way that doesn’t make sense.
This latest explanation still suffers from the same vagueness in time periods. How long does someone have to do well, before we say that yes, they did well in the future after doing well in the past? If we look at Steve Jobs’ stock performance 2005-2011, then if we were to ask the above question in 2006, then from 2006 to 2011 we would be observing someone doing well after they did well in the past. Why do other investors have to be able to do what Jobs did, before EMH is falsified? If market inefficiencies exist, it will take only one person to prove they exist. It doesn’t require a democratic process, it doesn’t require groups or populations of “investors”. It just takes one investor to show it.
As for time periods, what time periods are we talking about? It makes no sense to leave the time periods undefined, and reserve the escape clause of not defining the time periods so that we can always point to when investors did poorly after doing well. Why do we have to stretch out the time period considered in such an ad hoc fashion, to whatever arbitrary time period is needed to include their average or poor performance? If market inefficiencies exist, then it only takes one investor making one investment decision to show it. The fact that you may not accept it on the basis of one investor, and one investment decision, is really only a smuggling in of some silly democratic voting process to establish truths.
If I say the dark side of the moon has craters, and you don’t believe me, then all it will take to prove it to you is one astronaut taking one picture. It doesn’t need a whole bunch of astronauts taking a whole bunch of pictures.
What does “investors” mean? Are you saying the entire population of investors? If so, why do all investors have to have the ability and knowledge to exploit market “inefficiencies” in order for the EMH to be falsified? Wouldn’t it be far more reasonable to think that if the market does have “inefficiencies”, then only the most able and the most knowledgeable investors will be able to exploit them for periods of time, whereas the less able and less knowledgeable will not, that is, until they become as knowledgeable as the previously most knowledgeable investors, but by then there will already be a new group of highly able and knowledgeable investors exploiting inefficiencies?
Define the time periods, and explain why there has to be more than one investor to show EMH is falsified, and maybe EMH wouldn’t sound so hokey.
1. February 2012 at 20:00
Major Freedom: If my theory is that a coin-flip has a 50% probability of yielding heads, and I have 100,000 people flip a coin fifteen times to test my theory, you don’t prove me wrong by pointing to one guy who got fifteen heads in a row. Far from it. With large numbers of trials, we fully expect improbable events to occur–we just don’t expect them to happen in any *particular* case. EMH is 100% consistent with “some people beating the market year after year,” just like my hypothesis that the lottery is a losing game is totally consistent with there always being a winner.
1. February 2012 at 20:40
david,
I used neither ‘sufficient’ nor ‘necessary’. To me distributed information and the EMH are ideas that point in the same direction, rely on similar fundamental ideas on how the world works.
Here’s Peirce via Wikipedia on corollaries: “in corollarial deduction ‘it is only necessary to imagine any case in which the premisses are true in order to perceive immediately that the conclusion holds in that case,’ whereas theorematic deduction ‘is deduction in which it is necessary to experiment in the imagination upon the image of the premiss in order from the result of such experiment to make corollarial deductions to the truth of the conclusion.'”
But I hope we won’t now start a semantic discussion on what the “true” definition of a corollary may be.
I can’t really comment on Grossman-Stiglitz, I’m not familiar enough with it. But it looks to me like a conflation of micro and macro effects. Any trade is effectuated because the trading partners both see an advantage to it. This difference in viewpoints does not mean the market as a whole can’t be efficient. Precisely it is the contribution of the different opinions that will result in the aggregate price. Think thermodynamics. The fact that thermal gradients exist does not mean the second law of thermodynamics is wrong. The second law just says that temperature gradients will tend to disappear over time and that an equilibrium temperature exists.
1. February 2012 at 21:44
If all we’re arguing about is the 1%, I’m happy to say “Peace, brother”.
Kevin,
Who cares about the bottom 80%>
Those who spend time in the top 20% (bigger than 20%), all exploit personal experience to beat everyone else.
You CANNOT shrink my arugment by arguing 1%.
It doesn’t work.
The math of the bear proves you wrong.
I don’t have to outrun the bear, I just have to outrun you.
Having out-sized personal wins in the W / L column happens BECAUSE OF IQ (birth), it happens because of tenacity, and it also happens because of inherited wealth (birth).
But your shitty “peace brother” is NON-RESPONSIVE – my argument is that what happens in the aggregate is meaningless because any one man based on his own personal choices can increase the W and decrease the L.
For you to be responsive, you have to argue that the aggregate = fatalism.
And you are NOWHERE close to turning that corner, in fact you are getting your ass kicked.
The 80 / 20 rule says 20% get 80% – it is the same rule that get 1% a good chunk of 80%, the and there is nothing you can bring to the table to turn that into a meaningful takeaway from EMH.
Swing, and a miss. brother.
1. February 2012 at 21:53
Kevin, I think I can make the argument you’ll agree with:
VC face the EMH in the mirror daily, entrepreneurs not so much.
And why? Because VC all admit what matters most is the management. Execution is about talent and skill (which tend to skew young). Betting on equity investments is more of a pooled / average game.
If this wasn’t true, you’d be closer to convincing. To be convincing, you’d have to argue management / super star athletes don’t matter.
1. February 2012 at 23:26
What does the EMH say other than that on average returns will be average.
2. February 2012 at 07:06
“But the TIPS market doesn’t give you an expected NGDP reading.”
The TIPS market, the stock market and all other markets give you an expected NGDP reading indirectly. The problem is that this reading might differ from NGDP futures, and this difference is not stable.
2. February 2012 at 09:52
Scott, what do you thing the Quants have been doing for generations now, making baskets? No, they’ve been making fortunes.
Scott wrote,
“suppose you create an “Anti-EMH mutual fund,” that does nothing more than invest in mispriced assets like the Shell oil example, and lots of other unrelated mispriced assets. Shouldn’t those mutual funds out-perform the market over time? Especially if they owned many different assets?”
2. February 2012 at 09:55
This “efficiency” notion is brain dead and is inspired by bits of math, not the ever changing woukd with endless open-ended opportunities which require skilled and ipevery increasing perceptual judgment to imagine and capture.
Scott writes,
“Greg, Not all markets are equally efficient. It depends on factors like liquidity. Housing is less efficient than stocks, as it’s hard to sell houses short.”
2. February 2012 at 09:59
Let me put that differently. The issue isn’t efficiency as constructed into bits of math.
The issue is situationally unique opportunities which require constantly updated and adaptive skilled perception to perceive and capture in unique historical and geographical moments.
I can’t believe you don’t get this ….
Scott writes,
“Greg, Not all markets are equally efficient. It depends on factors like liquidity. Housing is less efficient than stocks, as it’s hard to sell houses short.”
2. February 2012 at 10:04
What has Apple been doing and those who perceived what Apple was doing?
Why didn’t anyone else perceive that and do it instead? Why did anyone bankroll Steve Jobs rather than dozens of others doing “the same thing”.
Wealth production and market coordination into an unknown and changing world requires more than efficiency, it requires situationally unique skilled perception, constant adapting vision into unique historical and geographical and technological adaptive spaces, etc.
Math and math constructs ostensively defined as “efficient” seems to have closed off the ability of many to conceptualize whole domains of causation and profit within the real world.
2. February 2012 at 10:12
Scott,
You ask, “Rafe, Why does the combined value become positive?”
There are a number of answers to this question and at the risk of sounding flippant or irreverent, I believe the answer is the same as “Why did Markowitz get the Nobel Prize in Economics?” I believe that Modern Portfolio Theory is an application of Parrondo’s mathematics. It’s related to the Kelly Criterion. It’s related to Brownian Ratchets. The links I included have more detail, here’s one: that has a simulation of “volatility pumping” which is an instance of Parrondo’s Paradox:
http://www.financialwebring.org/gummy-stuff/volatility-pumping.htm
If you are familiar with the two-envelope problem, once of the theorists suggests that the paradox can be explained by broken symmetry (i.e. before you choose it’s a informationally symmetric, but after you chose the symmetry is broken):
http://www.physorg.com/news169811689.html
And presumably the process of breaking symmetry creates latent value (i.e. information) that wasn’t there before. If you chose to act on that information you can reap that value for yourself in the form of profit.
2. February 2012 at 11:08
@JohnS “That is, after hearing one of Beethoven’s symphonies, we expect that another Beethoven symphony will be better than average.”
Joseph Heller wrote one good book. Nothing else worth reading, IMHO.
That does not mean his one good book was luck.
2. February 2012 at 11:15
And anyway, JohnS, you were abusing my analogy: it was not between Beethoven and investors, it was between two different things that cannot be done by a model.
2. February 2012 at 12:00
Greg Ransom,
All you are doing is attacking a theory for what it does not cover. The EMH is a theory about prices of securities in secondary capital markets. That is all.
You wrote:
“Spending a 1/4 billion $$ on a manufacturing plants, offices and employees is a capital investment, and is a form of competing in capital markets, with an aim for selling future stock in capital markets.”
It is a capital investment. It is not the same thing as buying and selling securities. It *uses* the capital markets to acquire funding, but taking out loans to finance business activities is just not at all the same as buying and selling stocks for profit. This shouldn’t be a hard thing to understand.
If you think the EMH isn’t interesting because it doesn’t address entrepreneurship, fine. It doesn’t interest you. That’s quite different from saying 1) that it is false or 2) that it is useless.
My own view is that weak-form efficiency is clearly true (you generally can’t make money by historical prices, subject to the caveat that the equilibrium condition assumes some level of profit-making in this sphere), and that semi-strong efficiency (incorporating not only historical prices but “public information”) is not true.
It seems as if Sumner and I agree on these points, which is fine with me. I hate to always disagree with the blog author.
2. February 2012 at 12:08
John S:
“Major Freedom: If my theory is that a coin-flip has a 50% probability of yielding heads, and I have 100,000 people flip a coin fifteen times to test my theory, you don’t prove me wrong by pointing to one guy who got fifteen heads in a row.”
Terrible analogy. You are assuming the notion that investors are lucky when they win, when you use the analogy of a coin toss.
Humans aren’t coins. Humans don’t act randomly. They learn. They act. They make choices.
“EMH is 100% consistent with “some people beating the market year after year,” just like my hypothesis that the lottery is a losing game is totally consistent with there always being a winner.”
EMH isn’t an empirical theory. You can NEVER be wrong if you assume it is true and then explain away everything that could ever happen.
Sumner said that EMH would be falsified by the existence of mutual funds persistently beating the market, without defining “persistently”, then he said it would be falsified by the existence of investors doing well in the future after doing well in the past, without defining “in the future after doing well in the past.”
Now you’re saying EMH can never be falsified because one or more investors beating the average investors “year after year” can always be interpreted as someone having a prolonged lucky streak.
In other words, you are, necessarily, viewing humans in an “EMH” way. You view human investors as nothing but roulette players, such that any prolonged win streak must be chalked up to prolonged good luck.
Like I said, EMH is a cop out for losers.
2. February 2012 at 12:46
Matt, There are all sorts of reasons that free markets might not work well. There is monopoly power, externalities, etc. These issues have nothing to do with the EMH, which says that asset prices rapidly and efficient incorporate new information.
Major Freedom, You don’t look at individual traders, as those aren’t picked randomly. Suppose you looked at the all time luckiest roulette better ever. What would his record tell you about the ability to beat the odds at roulette?
You do a systematic study. Do investors that outperform one year tend to outperform the following year? You look at everyone, not just Buffett.
dtoh, It says lots more. It says that if you look at the investors who did well in the past, they will do average in the future. That wouldn’t be true of sports. The players that did well in the past tend to do well in the future. It’s a radical theory.
123, Maybe a very crude estimate, but why not go with NGDP futures, which would provide a very superior estimate?
Greg Ransom, You are confusing unrelated issues. The EMH has nothing to say about Steve Jobs and Apple Computer. It about how asset prices incorporate new information. Not whether smart guys can invent neat products.
2. February 2012 at 13:32
Scott, give me. break. Jobs was no inventor, and there is all the difference in the wolrd between an inventor and an entrepreneur/investor/visionary.. Ask Woz ….
These are increasingly dumb “responses”. The original Quant figured out ways to beat Vegas. and no statistical data steam can prove or falsify any hypohesis of who is using tools to do so and who is not, any pattern can happen by luck — but we know from other means what is happening.
Your whole game is to pretend we only have numbers and tests wih numbers. As i say. Nothing here but platefulls of dumb.
Scott wrote:
“Greg Ransom, You are confusing unrelated issues. The EMH has nothing to say about Steve Jobs and Apple Computer. It about how asset prices incorporate new information. Not whether smart guys can invent neat products.”
2. February 2012 at 13:42
Scott is the insufferable philosopher in the class room who demands of the students that they prove to hims that there are chairs in the room — and that they only use statistics and a dowloadable stream of computer numbers to do so.
It’s a category mistake and a burden of proof mistake and a mistake about the capacities of humans and a mistake about how we use the word “know” and a mistake about how we identify chairs, etc.
Those taking the EMh beyond the data sets used in the Random Walk literature are making similar gross errors .. and it is their burden to prove thy aren’t.
2. February 2012 at 13:45
Scott, do you know anything about the history of the Quants and the Random Walk / EMH literature, if so, what exactly?
I fankly can’t gage if I’m engaging someone who in ffact does know anything of this history.
2. February 2012 at 14:20
Scott, define “assets” and define “information” …
Again, I see a 3 cup and ball shell game being played, and not a thing more.
Scott writes,
‘”It about how asset prices incorporate new information. Not whether smart guys can invent neat products.”
2. February 2012 at 14:27
This “efficiency” notion is brain dead and is inspired by bits of math, not the ever changing woukd with endless open-ended opportunities which require skilled and ipevery increasing perceptual judgment to imagine and capture.
well, for me the EMH notion is driven by:
1. lots of traders I have personally seen who have been blown up trying to “arb” the market. By blown up I mean lost lots of money. Quants and non-quants alike.
2. review of lots of investment bank presentations. if you do not know who the sucker in the room is, its you.
“risk free” means exactly that: It means there is no way to lose money. If you are beating the market by playing outside the “normal” rules as the quants did at vegas (card counting), thats not a violation of EMH. thats a market structure flaw and you are capturing gains from improving market design. There are 5000+ hedge funds employing brilliant people to beat the market, yet most do not. I think the vast amount of evidence is that its excruciatingly hard to “beat the market.”
2. February 2012 at 14:30
Scott,
You say “It says lots more. It says that if you look at the investors who did well in the past, they will do average in the future. That wouldn’t be true of sports. The players that did well in the past tend to do well in the future. It’s a radical theory.”
Well….no! How do you think Larry Bird would perform these days?
And your mutual fund example doesn’t prove the EMH because a) there is a difference between the fund performance and the performance of individuals who actually manage the funds (or the parts of the fund), b) as the fund gets larger it becomes more difficult to take advantage of an arbitrage without moving the market adversely, c) as the fund gets larger, it will trend toward the average (e.g. if a fund was so successful they managed everyone’s assets, their return would exactly be the average), and d) the preferences of asset managers change as they become wealthier (a successful asset manager is more likely to want to play golf than to read 10Ks than an asset manager who was still relatively poor and trying to establish a reputation.)
Bottom line, you are probably earning a better return with your assets in an index funds than you would handing them over to an active manager, but if you managed them yourself, you would probably be doing better than the index fund…. at least until you became so rich you decided it wasn’t worth the effort.
2. February 2012 at 15:32
ssumner:
“Major Freedom, You don’t look at individual traders, as those aren’t picked randomly.”
Woah….investors who have the ability and knowledge to exploit market inefficiencies aren’t “random” either! They are specific people with specific abilities and specific knowledge that other specific investors lack. There is nothing random to it! Why must investors be picked randomly? Picking random investors will necessarily be a picking of average investors, and so you will never see anything other than average returns. But then you’d just be begging the question.
“Suppose you looked at the all time luckiest roulette better ever. What would his record tell you about the ability to beat the odds at roulette?”
I am hopeful that others can see this, but it is VERY clear to me that EMH is definitely a philosophy used to interpret investor and entrepreneur behavior. You clearly view investors as nothing but gamblers at a casino, who can only ever beat the average investors by hitting a lucky streak, the way roulette players hit winning streaks.
What if we view investors as able and knowledgeable in exploiting opportunities that other investors are too dumb witted to notice or take the time care about? Oh but that would a priori violate EMH wouldn’t it?
OK, if you want to view investors not as able and knowledgeable individuals who can exploit market inefficiencies that the average investors can’t or won’t exploit, but rather as nothing but gamblers who can only persistently win through luck, then I will direct your attention to the casino owners. Casino winnings are biased towards the house and biased against the customer side gamblers. They exploit inefficiencies derived from relative gambler ignorance and relative casino intelligence. Casinos tend to “persistently” win over time.
And since you didn’t define “persistently”, I won’t define “persistently” either, so there.
Casino side gamblers therefore tend to “persistently” beat the average population of customer side gamblers. “House wins”!
Sometimes, however, like the crackerjack team of students from MIT, customer side gamblers can “persistently” beat the House. That team developed a way to “persistently” beat the House, by using their relatively superior ability and knowledge to exploit the market inefficiencies derived from relative House ignorance and relative customer gambler intelligence.
Of course, able and knowledgeable investors..I mean gamblers…don’t have to remain relatively more able and knowledgeable than the average forever, since people’s ability and knowledge can change, and eventually the MIT team was caught. The casino then reasserted itself by learning about the team’s methods, and devising counter-measures. Then the casinos had once again become relatively more able and knowledgeable than the average customer side gamblers, and they once again exploited market inefficiencies to “persistently” win. That is, until the next time there are more able and knowledgeable invest..gamblers, who are able to exploit the NEW set of market inefficiencies that now exist between all parties involved.
And on and on it goes, as each set of inefficiencies is exploited over time, making way to new knowledge and new innovations, absorbing the old inefficiencies as new knowledge, which then changes the structure of the economy, leading to new inefficiency opportunities to exploit.
I interpret what happens between casino owner gamblers and customer side gamblers as a game of wits, not a pure gamble of luck. Most of the time the casino wins, and I argue that is because casino owners have state laws on their side, the resulting market inefficiencies of which they exploit to “persistently” beat the average customer side gambler. Without those state laws, you’d probably see more for more exploitations of inefficiencies from customer side gamblers, and far fewer from casino side gamblers. Or maybe we wouldn’t see hardly any gambling casinos at all, because highly able and knowledgeable customer side gamblers might be able to bankrupt every casino that opens. Who knows. The state probably won’t eliminate pro-casino, anti-gambler laws, because casinos are a huge cash cow for taxation.
2. February 2012 at 15:35
ssumner:
“You do a systematic study. Do investors that outperform one year tend to outperform the following year? You look at everyone, not just Buffett.”
How in the world can “investors”, meaning all investors, outperform the market, meaning themselves? You’re making no sense.
Sorry, if market inefficiencies exist, they will be exploited by individuals, not everyone. That’s just silly. Inefficiencies are endogenous. The population of investors can’t be at the same high ability and knowledge to outcompete themselves.
I don’t know about anyone else here, but I think the defenses of EMH are just downright muddleheaded.
2. February 2012 at 15:59
“Bottom line, you are probably earning a better return with your assets in an index funds than you would handing them over to an active manager, but if you managed them yourself, you would probably be doing better than the index fund…. at least until you became so rich you decided it wasn’t worth the effort.”
The key word is probably. PIMCO has a number of funds that consistently outperform their benchmark index. It’s a very simple model: derivative exposure to a target index (e.g. SP500, but it doesn’t matter which), with actively managed fixed-income collateral. PIMCO Stocks Plus is a good example. Unless your an employee or someone with access to the institutional class the fees probably cut into that, but that’s not an issue for me.
2. February 2012 at 16:32
you can’t judge the EMH by whether a manager can outperform an index because managers have some leeway to choose – one has to decide if the idex is appropriate. Bill Gross has a slightly above average track record (with some doozy errors) but also has the advantage of being able to go on CNBC and talk his book up when he is selling (no, i’m sure he neeeever does that). Dennis Gartman measures his alpha-funds performance against the 3 year canadian bond rate even though he’s actively managing a portfolio that is short a buch of stuff and long a bunch of other stuff that all adds up to significant risk, more risk than the 3 month canadian bond yield!
2. February 2012 at 16:54
Major Freedom, “returns are distributed according to luck” is the prediction that EMH makes. You can’t refute EMH by providing evidence that is fully consistent with that prediction. If you provide evidence that “skill causes higher returns,” you will provide strong evidence that EMH is wrong. But the mere existence of high returns provides zero evidence about the reason for those returns, and provides zero evidence that returns are distributed non-randomly. Period.
There are good arguments against EMH. But “some people have beaten the market repeatedly” is not one of them.
2. February 2012 at 17:07
Major Freedom:
“How in the world can “investors”, meaning all investors, outperform the market, meaning themselves? You’re making no sense.”
Read Scott’s post again. He’s saying you look at the subset of investors that outperformed last year and you see if they tend to outperform again next year.
If you’re going to call your opponents muddleheaded, I suggest reading more carefully.
2. February 2012 at 17:58
@dwb
I’m not citing his performance as evidence against EMH. I think the weak-form EMH is true. I’m just pointing out that there are reasonable alternatives to index funds that can consistently beat that index. Maybe I’m a cynic, but I don’t think there is much on CNBC that ISN’T talking their book, and I have that channel on all day throughout the workweek.
2. February 2012 at 18:09
Maybe I’m a cynic, but I don’t think there is much on CNBC that ISN’T talking their book
we agree on that. entertainment, and not even good entertainment.
2. February 2012 at 19:39
Scott, you’d do well to state clearly the many things everyone hear is actually arguing, so you are clear you understand the points.
Everyone else, Scott has already punted off entrepreneurs, he’s punted on insider info (which is a huge gaping hole), and he’s probably willing to admit the first quants kicked the shit out of everyone (until the game got raised).
It isn’t interesting unless he say Michael Jordan was actually just lucky.
—
I do think the admittance that entrepreneurs make capital investment based on skill JUSTIFIES my argument hat they should be the favored class of investors.
Since theirs isn’t luck, Scott would agree we should want them to keep the most of their money for re-investment.
2. February 2012 at 20:22
Scott, it’s about whatever you want to say it is about at the moment and the game changes every single time you “weight in”.
You’re playing Calvinball.
I don’t see anything else being put on the table.
And at the core of your game of Calvinball is the game of making category mistakes about an every changing variety of completely non-univocal varieties of “information” and “assets”.
Scott wrote,
“It about how asset prices incorporate new information.”
2. February 2012 at 22:01
Greg Ransom is doing little to dispel common stereotypes of Hayek-heads. Cf. Dunning-Kruger effect.
Calling people and their responses dumb does nothing to change reality. In reality, there are three forms of the EMH hypothesis: weak, semi-strong, and strong. Strong EMH posits that all private information is incorporated into asset prices. Since nobody believes that to be the case, it is pointless to point to inside information as a “gaping hole” in the theory.
Likewise, almost nobody believes that weak-form EMH is false. Let me share something with you: quant traders don’t make money by predicting future price movements from past price movements, except maybe a few at the margin. All that “technical analysis” mumbo-jumbo is made-up foolishness designed to sell books and DVDs and investor seminars.
So the entire argument boils down to semi-strong efficiency. Once again, it is worth noting that *even if EMH is true* there will be *some* investors who outperform the market, even by *skill.* That’s because markets can’t incorporate information unless people are given an incentive to do so. That’s one reason why it does not contradict EMH to point to Warren Buffett. Then, of course, there is the separate question of luck, which is relevant no matter how much you protest to the contrary. A rudimentary understanding of basic statistics should tell you that.
As for the idea that EMH is “non-falsiable” or considered to be “a priori” true — you’re confusing the scope of a theory with its predictive power. Newtonian mechanics, for instance, is a terrific theory of physical motion within its scope, which is that of objects that aren’t very tiny moving at speeds that aren’t super fast. You can make a lot of highly accurate predictions about the world with F=MA. So too with weak-form EMH.
I won’t even get into the irony of a Hayekian complaining about lack of empirical evidence.
2. February 2012 at 22:20
BW,
Glad to have you fully aboard the Morgan train!
“If you think the EMH isn’t interesting because it doesn’t address entrepreneurship, fine. It doesn’t interest you. That’s quite different from saying 1) that it is false or 2) that it is useless.”
Since we agree entrepreneurship is a skill based activity, you will ADMIT we should logically favor their gains in tax policy, since they alone (according to your own definition) violate EMH.
I love when you lose points on the flow.
(tingles)
3. February 2012 at 00:10
[…] financial columnists, etc.) do not outperform the market. Here are some explanations I’ve seen (the following are paraphrased commenters’ […]
3. February 2012 at 00:26
Dtoh: “Well….no! How do you think Larry Bird would perform these days?”
I’ll bet you that the NBA players with 3-point FG percentages in the top 10% of the NBA in 2011-2012 will have above-NBA-average percentages in 2012-2013. I’m absolutely serious and would bet you up to, say, $2000. Let me know if you’re interested.
3. February 2012 at 07:20
John S:
“Major Freedom, “returns are distributed according to luck” is the prediction that EMH makes. You can’t refute EMH by providing evidence that is fully consistent with that prediction. If you provide evidence that “skill causes higher returns,” you will provide strong evidence that EMH is wrong. But the mere existence of high returns provides zero evidence about the reason for those returns, and provides zero evidence that returns are distributed non-randomly. Period.”
John S, “returns are distributed according to ability and knowledge” is the prediction that anti-EMH makes. You can’t refute anti-EMH by providing evidence that is fully consistent with that prediction. If you provide evidence that “luck causes higher returns,” you will provide strong evidence that anti-EMH is wrong. But the mere existence of high returns provides zero reason for those returns, and provides zero evidence that returns are distributed randomly. Period.
“There are good arguments against EMH. But “some people have beaten the market repeatedly” is not one of them.”
Yes, it is. If market inefficiencies exist, then individuals should be able to exploit them. It’s not the case that all investors have to be able to exploit them, it’s not the case that mutual funds have to be able to exploit them, it’s not even the case that those who did will in some undefined past have to do well for some undefined future.
Again, for the millionth time, EMH is not an empirically derived theory. It is a way of interpreting why investors earn higher than average returns.
“How in the world can “investors”, meaning all investors, outperform the market, meaning themselves? You’re making no sense.”
“Read Scott’s post again. He’s saying you look at the subset of investors that outperformed last year and you see if they tend to outperform again next year.”
He never said anything about “last year” and “this year”. He never said anything about time periods at all, and for good reason. Any time he gives would of course be arbitrary, thus blowing the whole EMH theory up.
Why does it have to be last year and this year? Why can’t it be the past 6 months for those individual investors who had relatively superior ability and knowledge for the past 6 months such that they can exploit market inefficiencies for the past 6 months, and then we observe some of those individuals making higher than average returns for 6 months, some of them 1 year, some of them 2 years, some of them 3 years, etc, thereafter? Anti-EMH is fully consistent with explaining why this occurred. Those individual investors having relatively higher ability and knowledge for various periods of time, were able to exploit market inefficiencies for those varying times while they had relatively higher ability and knowledge.
If you’re going to say anti-EMH is wrong, then you have to at least show instances that can’t be explained by anti-EMH. All Sumner has done is referred to empirical examples that can all be fully explained by anti-EMH.
The term “persistent” is left undefined because every EMH advocate realizes he has to retain for himself an escape clause to explain away all instances of what one could call “persistent” high returns. By not defining the time, the EMH advocate can just go beyond the “persistent” time his opponent refers to, and then ad hoc defines “persistent” to include MORE than that time, to when the investor conveniently earns average or below average returns.
Look, you can view investing as nothing but gambling, and investors as nothing but gamblers, but you cannot claim anti-EMH is refuted by empirical data when every single instance of empirical data is fully explainable by and consistent with anti-EMH theory.
The silly tests that Sumner is proposing are not tests that would actually prove anti-EMH false. They are artificial, and they don’t even consider the anti-EMH position. The anti-EMH position consists of individual investors having relatively superior ability and knowledge to exploit market inefficiencies FOR ANY period of time. From arbitrage desks that exploit inefficiencies and thus collapse those inefficiencies over a timespan of a minute, all the way up to individual investors like Buffet who exploit inefficiencies and thus collapse inefficiencies over a timespan of years, no empirical data has so far ever refuted anti-EMH. Just like no empirical data has ever refuted pro-EMH.
How can that be? It’s because both EMH and anti-EMH are ways of INTERPRETING historical data.
For Sumner to say “EMH would be refuted by observing ALL investors who performed well in the past perform well in the future” is like an anti-EMH advocate saying “anti-EMH would be refuted by observing ALL investors who performed well in the past perform well in the future.”
And wouldn’t you know it? If I also keep the time period undefined, then I can point to every single investor who did well in some arbitrary past also do well in some arbitrary future, by just picking the mid-point so that they perform well on both sides of that point. You think this is arbitrary? Well then so is selecting a time period of success to be “last year”, and “this year.”
Why a year? The market process is continuous. Inefficiencies aren’t set in stone such that they last no less than one year and one year only, so that we are compelled to look at investor’s yearly performances. Ability and knowledge are constantly changing, sometimes slower, sometimes faster. They change from person to person, and even with the same person. Market inefficiencies are a product of people’s relative ability and knowledge. It is simply absurd to insist that the time period of exploiting inefficiencies be “one year.”
The catastrophic flaw with EMH is with it’s rigid, non-human, mechanistic conception of the market, along with a complete ignorance of the entrepreneurial process. Entrepreneurs are the individuals who exploit market inefficiencies, not some random population of investors connected to “higher than average returns last year.”
Sumner is looking at numerical data and insisting that there should be some fixed pattern, when what he should be looking at are the individuals who produce that numerical data, which will enable him to interpret entrepreneurs as relatively able and skilled at exploiting market inefficiencies, and correcting them, before others can do it and/or before others choose to do it.
You guys are looking at data like it has a life of its own, trying to find fixed patterns based on arbitrary criteria, all of which is abstracted away from individual entrepreneurs. ANY model about the market economy that abstracts away from individual entrepreneurs is no longer talking about the market economy. They are just measuring the dimensions of trees as they rustle in the wind.
3. February 2012 at 08:36
Major_Freedom makes an important point in the comments to Bob Murphy’s comment on this post:
“I think those who adhere to EMH are people whose minds tend to collapse to rigid stability concepts of final equilibrium and whatnot. That’s a big problem because the real world market takes place over time, and people’s ability and knowledge changes too. Individuals have used their ability and knowledge to beat the market for periods of time. Sure, they can’t keep doing that once their ability and knowledge becomes generalized, but by that time, there will already be new individuals with new abilities and new knowledge.”
3. February 2012 at 08:45
Major_Freedom writes:
“EMH is not an empirically derived theory.”
There are a couple of ways to read this sentence, but however your read it, we should acknowledge that in the first instance the “insight” of EMH is nothing but a way to characterized statistical facts about simple streams of numbers, e.g. stock prices, etc.
What is _NOT_ empirically derived — what come straight out of the back pants pocket — is the move from this data stream to everything in the entire world lying outside of this blind stream of numbers pouring out of a bucket.
3. February 2012 at 08:52
Scott: “Maybe a very crude estimate, but why not go with NGDP futures, which would provide a very superior estimate?”
EHM says you cannot make money by arbitraging NGDP estimates from NGDP futures and TIPS markets. But EMH does not say which estimate is superior for the purposes of public policy.
3. February 2012 at 09:08
Greg, Jobs didn’t get rich by picking stocks, he got rich by managing Apple. The EMH has nothing to say about those skills.
If you think it does, then you don’t understand the EMH.
dtoh, You said;
“You say “It says lots more. It says that if you look at the investors who did well in the past, they will do average in the future. That wouldn’t be true of sports. The players that did well in the past tend to do well in the future. It’s a radical theory.”
Well….no! How do you think Larry Bird would perform these days?”
I don’t know if this is a joke, but I’ll take it seriously. At the beginning of the season you can predict that Lebron James and Kobe Bryant will score more than the average player. Those predictions will usually be right. You can’t predict at the beginning of the year which mutual funds will earn higher returns. That’s the key difference.
I doubt I could pick stocks better than an index fund. I wouldn’t even know how to start.
Major Freedom. The point is you need to figure out whether such skilled people exist, or whether the very rich were just lucky.
You said in reply to me;
“You do a systematic study. Do investors that outperform one year tend to outperform the following year? You look at everyone, not just Buffett.”
“How in the world can “investors”, meaning all investors, outperform the market, meaning themselves? You’re making no sense.”
I never said everyone could outperform the market.
Morgan, Yup, Jordan wasn’t just lucky.
123, Perhaps there’s no ironclad proof, but surely you’d expect oil price futures to better predict future oil prices than orange juice futures could predict future oil prices. So I don’t see where you are going with this argument.
3. February 2012 at 09:20
Scott,
To be clear, you do agree with my claim:
The capital gains of entrepreneurs ought to / should be be treated as different and unique since LUCK isn’t the operable factor.
This suggests raising taxes on the capital gains of investors so we can lower them on entrepreneurs, so that entrepreneurs can make more investments IN THEMSELVES.
Because they have a higher chance of success than normal investors, and because they actually build new products and services…
My preferred model is better than the current one OR yours which treats them evenly.
Have you changed your mind yet?
3. February 2012 at 09:21
Scott, you are using the EMH to infer much more than the “Random Walk” statistical attributes of stock prices across time.
And it looks to me like you don’t know Jobs’s biography.
Scott,
“Greg, Jobs didn’t get rich by picking stocks, he got rich by managing Apple. The EMH has nothing to say about those skills.
If you think it does, then you don’t understand the EMH.”
3. February 2012 at 09:28
No, the anti-EMH position is about the competitive perception and discovery of adaptive spaces that never previously existed — and it’s about the scientific illiteracy of the picture of “the market” which sees it as a perfectly efficient GE construct, with pockets of “market failure” inefficiency to exploit.
It’s a conflict of visions between a scientifically illiterate understanding of the market process derived from mistaken reifications of bits of math versus a rival understanding of the market process as a discovery procedure of the sort we find in the real world.
Major_Freedom wrote,
“The anti-EMH position consists of individual investors having relatively superior ability and knowledge to exploit market inefficiencies FOR ANY period of time. From arbitrage desks that exploit inefficiencies and thus collapse those inefficiencies over a timespan of a minute, all the way up to individual investors like Buffet who exploit inefficiencies and thus collapse inefficiencies over a timespan of years, no empirical data has so far ever refuted anti-EMH. Just like no empirical data has ever refuted pro-EMH.”
3. February 2012 at 09:31
Scott, here’s a heads up. Someone who is “always honest” doesn’t play games with non-responsive statements which don’t engage the issue at hand, and which effectively dissemble.
3. February 2012 at 09:36
Scott: “Perhaps there’s no ironclad proof, but surely you’d expect oil price futures to better predict future oil prices than orange juice futures could predict future oil prices. So I don’t see where you are going with this argument.”
Your oil-orange juice analogy is misleading in several ways:
1. Oil futures markets are more much more liquid than orange juice futures markets, while NGDP futures markets would be much less liquid than S&P 500, TIPS and regular treasury markets
2. Orange juice market is not involved in the transmission of oil price shocks, while S&P 500, TIPS and regular treasury markets are very important for the transmission of NGDP shocks
3. Orange juice market doesn’t really process oil-related information, while S&P 500, TIPS and regular treasury markets are one of the most important places where NGDP-relation information is processed
3. February 2012 at 09:39
Scott, Jobs sold 1/4 billion of one asset, and invest that money in two other assets.
No amont of BS can contribute anything to an honest discussion which doesn’t engage such facts.
But bring it on …
3. February 2012 at 09:53
BW, I never mentioned insider trading, and your remarks on Hayek and the empirical makes it plain that you’ve never read Hayek.
And BW, is you are acknowledging the fact of entrepreneurial discovery, judgment and perception into previously unknown and non-existing adaptive spaces, then you’ve effectively rejected the Flatland mental picture of “information” which lies behind that falsehoods about the real world unwittingly articulated by EMH fundamentalists.
3. February 2012 at 09:57
I suppose the Dunning-Kruger effect would also cover people who are begging the question and so little comprehend the matter at hand that they fail to be aware that they’ve done so.
The whole use of the word “information” here BEGS THE QUESTION and fails to engage the issue at hand.
Do I need to explain again what that is?
BW writes,
“Calling people and their responses dumb does nothing to change reality. In reality, there are three forms of the EMH hypothesis: weak, semi-strong, and strong. Strong EMH posits that all private information is incorporated into asset prices. Since nobody believes that to be the case, it is pointless to point to inside information as a “gaping hole” in the theory.”
3. February 2012 at 11:17
Major Freedom (and Greg Ransom), “skill” that only gives excess returns in random periods doesn’t seem much different than “luck.”
Imagine I have a huge number of young students take a test each day. But each time, I switch the subject to something fairly esoteric. Keynesian macroeconomics, Babylonian history, human anatomy, etc. Now here are two observations I make about my students’ performance:
1. There is a normal distribution of scores.
2. If I take only students who scored above the 90th percentile on the first test, and observe their performance on the second, third, fourth, etc. tests, they score (on average) in the 50th percentile on each one. The same is true of the performance of the 90th+ percentile scorers on each test.
So, what’s going on here? Are the students who do well “skillful” or “lucky”?
Well, from one perspective, skill might sometimes matter a lot. Students don’t always guess randomly, they know specific esoteric subjects. Maybe Timmy’s dad is Paul Krugman. Maybe Bobby’s mom is a doctor. Timmy scored well on the first test, and Bobby scored well on the third test, and they used skill to do so.
From another perspective, luck was the only factor. On any given day, I can give a test on any one of a thousand subjects. Since the students have no idea what the test will be about before they take it, none of them can go to school in the morning expecting to score any better than average.
So–it’s possible that investors get excess return for “skill,” but that skillful investors still just get average returns–because the circumstances that allow them to beat the market one day aren’t present the next. Sometimes they discover an “adaptive space” that never previously existed. Sometimes they don’t. But note the implication: none of us can expect to beat the market, because we don’t know *in the present* whether our skills are helping us. We’ll only know in the future, after we see our returns. If you try to beat the market, either you’re betting that your skill will turn out (unpredictably) to be relevant and profitable, or you’re betting that you’ll be lucky.
3. February 2012 at 11:18
BW,
Read Roy Cordato’s _Efficiency and Externalities in an Open-Ended Universe_, Esteban Thomas’s _Prices and Knowledge_, and Don Lavoie’s _National Economic Planning_ (esp. the last chapter), and then email me and we can begin an intelligent conversation on “information”, efficiency and the various uses of these notions in the domain of economic science.
You might also brush up on case of Kuhn & Popper against modern Baconian empiricism — which helps highlight the difference between theory laden perception and rival understanding vs the conception of the world as a stream of “given” data or “given” information coming out of a jar.
This will give you a leg up on getting on the other side of the Dunning-Kruger effect.
3. February 2012 at 11:43
At some point this discussion has become little different than discussing whether or not STOP signs are red or green with a man who turns out to be a color blind man who doesn’t know he is color bling — at some point you realize the man really has no idea that there are two different colors, red and green.
The task then becomes the tricky one of finding a way to articulate, demonstrate, and explain to him what you are talking about when you say that STOP signs are red and not green.
3. February 2012 at 11:50
John S., you need to think harder about this … the opportunity to most easily and profitably exploit the adaptive space of fast food and drive-thru food came and went — it didn’t exist forever. And not everyone had the same perceptual skill and judgment of future opportunity spaces to recognize these profit opportunities.
The examples could take an infinite space of different forms, and would include small “adaptive” perceptual and judgmental differences.
And note well — Scott is NOT restricting the “lesson” to be learned for macroeconomic theorizing to what the statistical facts of a stream of stock price numbers can “tell us”.
John S. writes,
“Major Freedom (and Greg Ransom), “skill” that only gives excess returns in random periods doesn’t seem much different than “luck.””
3. February 2012 at 14:08
Greg, you accurately summarized the premise of my comment:
The opportunity to most easily and profitably exploit my knowledge of (Subject X) comes during the first test, and doesn’t last forever. And not everyone has the same knowledge to recognize/take advantage of that opportunity. There are near-infinite subjects, and small differences in cognition could be adaptive for different subjects (literature vs. economics vs. biology).
3. February 2012 at 15:29
John S,
“The opportunity to most easily and profitably exploit my knowledge of (Subject X) comes during the first test, and doesn’t last forever. And not everyone has the same knowledge to recognize/take advantage of that opportunity. There are near-infinite subjects, and small differences in cognition could be adaptive for different subjects (literature vs. economics vs. biology).”
And the guys who do it once, are MORE LIKELY to do it again compared to those who have never done it.
Note: we are specifically talking about the guy who ACTUALLY BUILT the drive through, not the 1000 others who thought about it.
This is the corner you can’t turn, the skill function is the actual building of the thing, raising capital, figuring out business processes… doing all the real entrepreneurial shit.
—–
We see it again and again in this thread, you are MAKING MY ARGUMENT that we should favor the gains made by entrepreneurs above all other kinds of investor gains.
What they do is skill AND it is best kind of gain for this country.
—-
The next question: is the guy who had inside information more likely to have inside information again?
Yes, because he made gains increased his wealth and got closer to decisions makers.
—-
The next question: Is the guy with the fastest pipe, biggest computer, etc. more likely to make gains than the guy without them?
Yes. He’ll beat those people who are trying to make the same side of his trade.
—-
There are advantages other than skill, knowledge, that was also fleeting, but would predict greater likelyhood of future success.
I do like the idea of HUBRIS and OLD AGE (being outperformed by the young) being the great equalizers that weaken gains int he end, but that isn’t your argument.
You are trying to turn this into rolling dice, and that’s wrong.
3. February 2012 at 16:07
Wow, this thread really blew up since my last post.
My interest in EMH is what it says about the relationships of fundamental prices vs. market prices. Forgetting the fact that investors buy and sell behind the scenes, the relationships between fundamentals and market value often get very, very, very out of whack.
By out of whack, I don’t mean we had an unexpected nominal shock in late-2008 that forced the Dow to go down to 6000. I mean how the fundamentals never made sense for many market prices, even under the rosiest possible macro and micro scenarios.
Despite, people voluntarily bought at prices way too high for fundamentals and voluntarily sold at prices way too low for fundamentals. There are two things that might explain this (past the usual liquidity risk premium and such):
1. Simple market failures, whatever those are. Principal-agent problems, lack of information, outright fraud (i.e. lemon problem), etc.
2. Investors have extremely high discount rates which may mean they’re acting rationally by trend-following even though over a few years the strategy is terrible.
The variance in market vs. fundamental prices would be purely academic if the valuations didn’t have such a huge effect on capital allocation. When investors misallocate capital, then their standards of living are adversely affected.
Therefore, it’s in the government’s interest to maximize RGDP by regulating markets if more regulation would reduce capital misallocation by more than the regulatory compliance costs.
And why wouldn’t the market itself demand more disclosure like the government requires? I believe that financial disclosure and fraud prevention cannot really be done privately because of the large free rider problems. It can be done by short sellers, for example, but short sellers only make money if the stock is already overvalued. The company itself might increase their valuation from more disclosure, but then somebody outside the company has to check their disclosure. That’s the job of accounting auditors, but they are paid by the company themselves. They have their reputation on the line, but there’s also a good bit of money to be made in the short-term.
Anyway, in short I think we’re mostly looking at the EMH from the wrong perspective. It’s not all about how investors could possibly be consistent in beating the market. The EMH is all about capital allocation. You know, a hypothesis of the efficiency of markets. The fact that government could make markets more efficient is a blow to free market ideology, but it also could be true based on gross mispricings in the past.
3. February 2012 at 16:49
“magine I have a huge number of young students take a test each day. But each time, I switch the subject to something fairly esoteric. Keynesian macroeconomics, Babylonian history, human anatomy, etc. Now here are two observations I make about my students’ performance:
1. There is a normal distribution of scores.
2. If I take only students who scored above the 90th percentile on the first test, and observe their performance on the second, third, fourth, etc. tests, they score (on average) in the 50th percentile on each one. The same is true of the performance of the 90th+ percentile scorers on each test.
So, what’s going on here? Are the students who do well “skillful” or “lucky”?”
Let me take a crack to extend this analogy to how the market really works.
First of all, the people who happen to have money are those that, in the end, pull the trigger on buying and selling. They may or may not be “investors” themselves, in the sense of someone who analyzes the markets and picks securities. But they do have to pick those investors. Since they have limited time, they have to use a limited set of information to decide on someone to invest their money.
The easy thing to say is that they should just pick an index fund. That’s pretty much what I do, in fact. But ultimately security prices have to be determined by those who are actively in the market. For anyone active in the market, they must either invest the money themselves or pick an investor.
The market is also a voting machine, not a weighing machine. So the collective action of the active traders ARE the market. For no easy arbitrage opportunities to exist, ready buyers need to equal ready sellers at a given price.
To get back to your analogy, how are these investors graded exactly? It should be that investors are graded on long-term capital accumulation. In this sense, they would not need to beat the S&P 500 in a single year, but have a portfolio that will generate more (discounted cash flows)/(capital invested) over 3, 5, 10 years.
However, since the ultimate holder of money cannot know exactly how much discounted cash flow a given investor will generate, they have to use tools other than diving down into the nitty-gritty of specific securities and evaluating fundamentals. Often it’s simple and easy enough to just pick the investor which happens to have the longest string of outperformance compared to the S&P 500.
The pressure of money-holders on investors should mean that investors buy low and sell high, right? Well, yes and no. For fundamental valuation, the actual cash flows may be for years into the future. A liquidation or a company selling itself, for example, may not happen for a long time. In the meantime, the market valuation will be determined by market sentiment and nothing else. Market sentiment will eventually change, but that may be too long for the money manager to keep the business.
Predicting market sentiment is not just for the money manager to keep their job, but they also often get bonuses for chance outperformace over a year’s time. From their perspective, they maximize their utility by betting on market sentiment, upward or downward. There’s just little money in long-term fundamental value itself unless their personal incentives are better aligned.
Predicting market sentiment is the test you’re talking about with very little serial correlation. That’s very, very, very difficult, with millions of economic actors determining a securities’ price at the end of the year. For long-term value investor, that shouldn’t matter. They just see the cash flows and the liquidation value. Buffett, for example, greatly underperformed the S&P during 1998-2000. But money-holders often don’t have the same long-term thinking and go with those who have flipped heads enough times on predicting market sentiment.
4. February 2012 at 00:32
[…] the market” over time except by being extremely lucky. The evidence supporting this idea is surprisingly strong, at least to […]
4. February 2012 at 06:10
Morgan, Nope, haven’t changed my mind.
Greg, You said;
“Someone who is “always honest” doesn’t play games with non-responsive statements which don’t engage the issue at hand, and which effectively dissemble.”
I recommend reading Tyler Cowen, especially the parts about not assuming the worst in your opponents. I can only understand about one out of three of your comments, so it’s no big surprise I don’t always respond the way you’d like.
I still say Jobs tells us nothing about the EMH. The EMH allow for highly successful entrepreneurs. It says that Apple stock should reflect all available information–and it does.
123, I disagree. Both oil and OJ markets reflect both general inflation and market-specific relative prices.
Inflation (TIPS) reflect both NGDP shocks and supply shocks. So the analogy works for me.
Matt If regulators were smarter than the market they’d be out in the market getting rich. They have no ability at all to determine when asset prices are too high or too low.
The real world is full of uncertainty, and this is why both markets and expert views bounce around. Last August stocks plunged as investors feared a US and Europe double dip. That turned out to be wrong. But experts (economists) would have been of no help as they were also running around like chickens with their heads cut off, warning about a double dip from the euro-crisis.
Shiller is another potential “expert” who thinks he has the stock market figured out. So why wasn’t he giving a strong buy signal in March 2009? Weren’t stocks undervalued?
You said;
“Buffett, for example, greatly underperformed the S&P during 1998-2000. But money-holders often don’t have the same long-term thinking and go with those who have flipped heads enough times on predicting market sentiment.”
There really isn’t much evidence that Buffett is a talented investor. Even if the EMH were completely true, you’d expect a few Buffett’s to pop up now and then. Perhaps he was lucky.
4. February 2012 at 06:26
“I recommend reading Tyler Cowen, especially the parts about not assuming the worst in your opponents.”
Ah yeah, you might also then explain WHY you don’t think entrepreneurs gains should be favored over the stock pickers.
One is skill, so they are more likely to put gains to their next venture.
Real answers make it easier to get your butt whipped.
4. February 2012 at 09:45
Morgan, you realize you’re the only person in this thread who’s been beating the “entrepreneurship” drum? The rest of us are debating the efficient market hypothesis. The fact that no one’s taken your invitation to debate a pet issue doesn’t mean we’re fleeing in terror from the force of your arguments. It’s just that… we’re already discussing something else, and it is only tangentially related to entrepreneurship.
Matt, thanks for your responses. I agree that prices often seem ridiculous. But they usually seem much more ridiculous in hindsight. Even if bubbles and capital misallocation are big problems, I’m worried that building a government agency with more incentive to identify and correct long-term capital misallocations than the rest of the market would be an incredibly difficult task. The historical record of government/academic experts doesn’t give me a lot of confidence that they wouldn’t be cheering on the “bubbles” along with everyone else.
You tell the most plausible story I’ve heard for why we should expect no serial correlation in a non-EMH world. But if capital misallocation is an obvious and large problem, isn’t there a *huge* amount of money on the table for investors willing to focus on the long term? It seems like LTCM shouldn’t have had any problems identifying price-value mismatches *far* more lucrative than the tiny difference between 30- and 29.75-year bonds.
4. February 2012 at 15:12
Scott, once again, please define “information” — the whole conversation turns on the difference between (a) statistically correlated streams of numbers, i.e. “given information” and (b) the difference ways people understand the opportunity spaces in the world.
It’s fairly equivalent to the difference between Baconian empiricism applied to stream of “given” kinds of items dropping out of a bucket, and rival theory-laden conceptual schemes for understanding the world, e.g. as you find in most of science (e.g. read Popper and Kuhn and Hansen and Toulmin, etc.).
Or you might read a book on this topic: Esteban Thomas _Prices and Knowledge_, Don Lavoie, _National Economic Planning_, or Roy Cordato, _Efficiency and Externalities in an Open-Ended Universe_, etc.
The whole notion of “reflect all available information” IS QUESTION BEGGING.
What part of the logical notion of _begging the question_ don’t you understand?
Scott writes,
“The EMH allow for highly successful entrepreneurs. It says that Apple stock should reflect all available information.”
4. February 2012 at 15:19
Scott, I’ve read it. Have you? Really?? Your remarks don’t reflect that.
Here’s part of what Cowen suggests.
Make the smallest, tiniest effort to re-state or re-articulate the point being made.
Let’s start with Bacon vs Popper & Kuhn.
Do you really not get and have you really never heard of the difference between the Popper/Kuhn notion that perception and explanation is theory-laden, and the crude “empiricist” notion that science proceeds by some accounting or statistical derivation or “inference” for a series of “given” bare experience kinds?
Just say YES or NO.
After that, we can take the next step to how this applies to the open-ended world of profit opportunities in the human world.
Scott writes,
“I recommend reading Tyler Cowen.”
4. February 2012 at 15:29
Scott, I would guess that Cowen would recommend that people not move the goal posts repeatedly in the middle of a conversation — you’ve done that in the EMH conversation moving from housing prices and the whole housing economy, tp the whole of the structure of the economy, and then back to a restriction to just stock prices and professional investors.
(Note that Cowen also likes to move goals posts and take shots at people in the process — and he’s a likely to move the goal posts on his own principles of charity as anything else.)
4. February 2012 at 15:31
How many times have I invited you to start at a common place so that we can do what Cowen recommends — attempt to flesh out a common area where all sides are able to express what the other is sayings?
It tells me a lot about your understanding of and concern for Cowen’s recommendations that you have no interest in this core part of what he says.
4. February 2012 at 16:25
John S.
My bugaboo is treating entrepreneur gains as favored over conventional investor gains.
It appears those arguing for EMH as described here would need to support my goal. Within it is a tacit admittance that certain kinds of investors outperform the market far more reliably.
Just insisting I’m not talking about people picking stocks without actually responding to my point doesn’t diminish my argument.
There might be an argument for treating gains made by a skilled “out perform” investor the same way as all the idiots trying to beat the monkey picking stocks – but you aren’t making it.
The fact is right now, we treat entrepreneurial SMB gains WORSE than we treat stock and real estate investors, and they are where the new job growth comes from, so letting them keep their skilled based gains makes tremendous sense.
4. February 2012 at 17:52
First, let me say that I take the middle-of-the-road attitude that Malkiel has on the EMH. This means I think both the efficient markets religion guys like you, as well as the completely anti EMH-guys are wrong. But you are more wrong then they are, because you should know better, while the other side consists mainly of technical analysts. And since this is one of the few academic subjects I know very well because I wrote my master thesis on it and it’s very very close related to my work, I will explain why both of you are wrong.
First, the prediction of the EMH is not that the markets will not make systematic errors (this is plain RE); if that were the case I’d accept your arguments above. The prediction of the EMH is that market prices incorporate all relevant information and THEREFORE markets won’t make any systematic errors.
Now, the very strong version of the EMH says that markets price everything so smoothly that nobody can’t benefit from any information, i.e. the efficient markets religion is simply and plainly wrong. In my view, yes, you can’t beat the market, but that’s all because this kind of thing is done by people better equipped than you. Yes, there are opportunities for non-random alpha, as well as anomalies that you can benefit from, it’s just that if you were clever enough to do all that, you wouldn’t be trading your pension but you’d be trading the capital of Goldman Sachs. But just because YOU can’t beat the market, doesn’t mean that no-one can. It’s just that you’ll never meet the people that do. And you’ll never meet them because hedge funds and more so mutual funds are so insanely overpriced that even though some of these guys beat the market, people are stupid enough to pay them insane fees so you’re better off with a passive index; if you are a rational investor you’ll stay away from the people that beat the market because they’ll simply charge you more than the value they give you. The same thing is true about the traders in my industry, even though some institutions have by now managed to learn how to pay market-beating traders fairly.
I have three serious kind-of-empirical arguments to support my case:
A. Momentum is correlated with return and this is not data mining. Your “well, in 20 million data points you’ll get a million anomalies” is, how shall i put it.. cute. But exactly the same anomaly in the US and Sweden and Norway and Denmark and so on is not an anomaly, but a systematic pattern. Fama himself accepts that much.
B. I, myself once found a pretty good market anomaly, i.e. you could predict part of the variation in large cap stocks by having a good estimate of their brand name value. I wrote about this here:
http://orionorbit.blogspot.com/2011/05/more-on-efficient-markets-hypothesis.html
Yes, i know the above is kind of trivial (which explains why i’d probably make a poor academic economist, i’m always interested in trivial stuff), but look, it’s another case where the Fama-French residual is not white noise as the EMH (even in it’s #3465 ex-post-resurrected version) predicts. My explanation for this anomaly is that whoever trades large cap stocks at the big banks got their pricing models wrong, underestimated the monopoly power of having a good brand name, or whatever. So for a small period of 6 years there was (and maybe still is) a way to beat the market. This is why the religious version of the EMH is wrong; there is always a trade that you’d expect some investment bank or mutual fund to have made, but they didn’t. Markets are only sort of efficient and pocket change lies here and there.
C. My experience on the trading floor tells me that some people are good traders and some aren’t. Now, this section is written in a weird way because I got a contract and wish not to get sued for posting company stuff on Scott’s blog, but I’ll give it a honest shot with as much as I can publicly disclose. So what my experience tells me, is that if you take the time series of traders’ records (Professional traders, not the poor souls that get ripped of from brokers that financial literature usually looks at) and look at their alpha over a long period (or actually ANY kind of time frame), the lags of the alpha are very very significant. People that made good risk weighted returns last month or last year are more likely to make good risk weighted returns next year. It’s a pity that I can’t share any data on this, but consider for instance Goldman Sachs (no, i don’t work for them, just using a bank people know), they still DO pay their traders a lot of money. In fact, if you believe in the EMR, you have to disbelieve in private banks, because if banks are so stupid to pay 20% of their profits to bonuses for employees that could be replaced by an undergrad investing in the risk free asset, you’re actually saying that if the US government nationalized GS and fired the traders they’d be running the bank more efficiently than the private sector. Either this, or the EMR is wrong and those traders are paid because they DO beat the market. You can’t have it both ways here Scott.
Oh, and before you respond that my sample is biased or something like that, yes, I know, it might be. But on the other hand, I remember that when I started as a trader back in late 07 early 08, I made lots of trades and the frequency of trading increased so that by early 09, I’d be making around one trade a day, usually something involving the Swedish crowns or Norwegian stocks. And because I was mostly trading against traders that had very little information compared to mine on how to price Norwegian dry shippers and whether Riksbank would go negative, i was making money consistently (let’s say, in each and every month, regardless of whether the Swedish crown had gone up or down, around 65% of my trades were profitable). So when at the end of the quarter, my boss had in front of him say 50 trades that I was in favor of and maybe 30-40 would be winning ones, the chances that I would have done this by flipping a coin are very small -i stayed 11 quarters and kept doing that. If the markets were efficient, he could have fired me and hired a robot to invest passively, but he didn’t. Because chances are that the bank would have made less money.
Oh and by the way let me add that I don’t write the above to play the blog’s evil capitalist, in fact I quit the trading floor a bit more than a year ago and I would strongly discourage everyone for working with that kind of crap, whatever the reward. I’m not trying to play the rich guy that blows the tobacco of my $500 cigar on your tenured face, the above is a sincere account of my experience.
And based on what i’ve seen, there just too many anomalies and too many people consistently making profitable trades for the EMR to be true. To me the more watered down version of the EMH that Malkiel defends in his book seems much more reasonable. And the noise trader model of the Shleifer gang a much more accurate description of what happens in the markets, even though unfortunately it doesn’t make any brave testable predictions.
Don’t get me wrong, I still think that markets are efficient enough not to be worth to bother trading, unless you have lots and lots of million. Beating the market can be done if you have the skills, time and analytically tools, but it’s just not as easy as people think and in all likelihood you’ll get burned. Which is why these days I keep most of my money in passive ETFs.
5. February 2012 at 03:05
Scott: “I disagree. Both oil and OJ markets reflect both general inflation and market-specific relative prices.
Inflation (TIPS) reflect both NGDP shocks and supply shocks. So the analogy works for me.”
It is not possible to extract the general inflation information from the orange juice futures and use it the Department of Energy work, as relative price movements swamp everything.
On the other hand, market monetarists extract the general inflation and supply shock information from TIPS, Treasuries and S&P 500 all the time. The NGDP futures arbitrageurs would do the same, and according to EMH, TIPS – NGDP futures arbitrage would be no longer be profitable. But since the required expected return on arbitrage activities is both time varying, and dependent on the NGDP shocks, a central bank that pegs NGDP futures would have a lower NGDP stability than a central bank that allows some variation of NGDP futures prices according to the information received in other markets.
5. February 2012 at 06:52
Morgan, I don’t think there should be any taxes on investment income, I just favor consumption taxes. If stock pickers consume their lucky gains, they should be taxed.
Greg, OK, my definition of “information” is empirical or theoretical evidence that leads people to revise their subjective estimates of the value of assets.
orionorbit:
You said;
“First, let me say that I take the middle-of-the-road attitude that Malkiel has on the EMH. This means I think both the efficient markets religion guys like you,”
I see, it’s a religion to me. Then how do you explain this comment at the very beginning of my post:
“The question is not whether the EMH is “true,” how could it be?”
Yup, I sure sound like I have a religious faith in the EMH.
I won’t respond to all your arguments as you are attacking a belief I don’t hold. Your comments apply to someone who thinks the EMH is literally true, and obviously I don’t think that. So they have no bearing on my argument.
BTW, I’m not impressed that you found an anomaly, there are millions of them out there. I could find lots of them if I wanted to. One has to do with which team wins the Super Bowl. The AER published one on rainy days in NYC.
One final point. You say Fama “accepts” your argument on anomalies. He also supports the EMH. That should tell you your arguments don’t refute the EMH.
123. I don’t agree that the central bank could predict variations in the risk premium of NGDP futures. But if they think they can, then by all means go ahead. As long as I’m allowed to buy NGDP futures and profit at their expense, I have no objection. What I oppose is situations like late 2008, where I knew NGDP was going to undershoot, and the Fed wouldn’t let me get rich on that knowledge. As long as that opportunity is out there, I’ll let the Fed do as much speculation as they want, as much manipulating of the monetary base as they want, and we’ll see who’s right.
5. February 2012 at 07:19
scott, NON-RESPONSIVE.
In a perfect world, I’d take consumption taxes too.
That is NOT the argument I am making, and you are not being intellectually honest.
The question is: Compared the current capital gains system, assuming EMH we ought to and should favor entrepreneurs gains OVER the stock pickers.
I’m talking about something in an either A or B, not some C counter plan.
Now please answer.
5. February 2012 at 13:46
Morgan, increasing capital gains taxes (all other things equal) would presumably:
1. Increase the supply of entrepreneurs, as wealthy people deciding whether to invest their money or start their own business are more likely to do the latter.
2. Make us feel good because we like industrious entrepreneurs and now they’re taxed less than the lucky investors.
3. Increase taxes on borrowing money (you can’t just tax one side of a transaction), making life more difficult for entrepreneurs who don’t have large amounts of capital available.
Ignoring 2, which I don’t care about, explain what evidence or theory makes you sure that effect 1 dominates 3 when it comes to the amount and quality of entrepreneurial economic activities.
5. February 2012 at 14:11
Scott:”I don’t agree that the central bank could predict variations in the risk premium of NGDP futures. But if they think they can, then by all means go ahead. As long as I’m allowed to buy NGDP futures and profit at their expense, I have no objection. What I oppose is situations like late 2008, where I knew NGDP was going to undershoot, and the Fed wouldn’t let me get rich on that knowledge. As long as that opportunity is out there, I’ll let the Fed do as much speculation as they want, as much manipulating of the monetary base as they want, and we’ll see who’s right.”
The Fed won’t be able to predict the variations in the risk premium of NGDP futures. But it should be able to measure the risk premium, and take it into account when adjusting NGDP peg.
Suppose a hedge fund that was long stocks and short NGDP futures goes bust. As a result, NGDP futures market starts showing inflationary pressures, with TIPS and S&P 500 indicating deflationary pressures. The Fed should do nothing in this case, as neither money demand nor money supply hasn’t changed (or a bit of monetary stimulus might be warranted if the collapse of a hedge fund increases money demand).
5. February 2012 at 14:35
John S:
“Major Freedom (and Greg Ransom), “skill” that only gives excess returns in random periods doesn’t seem much different than “luck.”
It’s isn’t random. It is exactly correlated with the periods of time in which individuals have relatively superior skill such that they can exploit market inefficiencies.
You say it’s not visually different compared to if individuals won and lost according to luck. You’re absolutely right. It isn’t visually different. That’s exactly why EMH is not an empirical theory, but a way in which one interprets what one sees.
“Imagine I have a huge number of young students take a test each day. But each time, I switch the subject to something fairly esoteric. Keynesian macroeconomics, Babylonian history, human anatomy, etc. Now here are two observations I make about my students’ performance:”
“1. There is a normal distribution of scores.”
“2. If I take only students who scored above the 90th percentile on the first test, and observe their performance on the second, third, fourth, etc. tests, they score (on average) in the 50th percentile on each one. The same is true of the performance of the 90th+ percentile scorers on each test.”
The market is not a controlled experiment, and is not composed of only one field of inquiry at a time. Bad analogy.
“So-it’s possible that investors get excess return for “skill,” but that skillful investors still just get average returns-because the circumstances that allow them to beat the market one day aren’t present the next.”
WHY aren’t they present? It’s because investors have already exploited the inefficiency, the market has since changed, and now new inefficiencies present themselves that are borne out of the market process itself.
5. February 2012 at 19:07
Major Freedom, you may call EMH as a “non-empirical theory,” but most economists would a particular model of human behavior in order to make certain predictions. In this it’s exactly like every other economic model out there. And what matters isn’t whether the model is realistic but whether its predictions are accurate. You’re making many attacks on the *model*, but none of your criticisms explain why the evidence is consistent with the *predictions*, the only part that actually matters.
So maybe you’re totally right about “skill” vs. “luck.” Let’s assume that you are–that any instance of above-market return is due to skill, not luck. The EMH, as a model, is thus wrong. Anti-EMH, as a model, is right.
So now that we’re done with the unimportant question of whose model is true, let’s get to the much, much more interesting question of whose model makes accurate predictions.
EMH predicts that the distribution of “wins” and “losses” will be indistinguishable from chance. Having a high IQ score won’t be associated with higher return. Having a good education won’t be associated with higher return. Having won big in the past won’t be associated with higher return. And so far, in the vast majority of cases, for the vast majority investors, that prediction holds up extremely well.
Anti-EMH predicts… well, whatever you want, really. Apple is undervalued (or overvalued). Stocks are overvalued (or undervalued). Gold is undervalued (or overvalued). The list goes on. And here, the same empirical evidence says unequivocally that our anti-EMH predictions will do no better than chance.
And that’s why EMH is useful, even if it isn’t true, and anti-EMH is useless, even if it is true.
5. February 2012 at 21:26
Just a couple of quick points:
John S. makes a good point about how, if prices were bad all the time, then there would be a huge amount of money to be made. In fact, I agree that for most well-functioning, liquid markets, market prices will tend to the fundamental prices for this reason. Real markets are a mix of a “Keynesian beauty contest” where everyone only cares about market sentiment and where everybody buys purely on long-term fundamentals.
That’s also why Buffett hasn’t really had an eye-popping year in beating the S&P 500 since the 70’s. His performance since then has really been a slow pace of compound, geometric growth by having longer time horizons than the rest of the market. If Buffett continually beat the S&P by large amounts year over year, then the market would have indeed figured what the hell he was doing and do that.
But since there is always a lot of money to be had when the market falls out of line with fundamentals, we have to ask what the hell is going on when the market does fall out of line with fundamentals. There are EMH-friendly reasons, such as transaction costs and discounts on alternative assets due to illiquidity. But tech stocks had low transaction costs and were highly liquid. AAA CDO’s meanwhile were illiquid, alternative assets, which should have meant a discount to fundamental value and not the large premium observed pre-2007. My best explanation is that, despite $100 bills laying on the ground for patient, liquid short-sellers in either case, the strength of the bubbles outlasted both the patience and liquidity of short-sellers end investors, as happened to Julian Robertson whose hedge fund blew up before his .com shorts could come to fruition.
To go back to Scott’s point on regulators, government intervention to “fix” this issue has a high bar to climb. Less capital has to be misallocated under regulation than under no regulation. Contrary to what Scott says, regulators do not have to outsmart the market, predict market movement, pop bubbles or anything like that. What regulators can do is step in to provide information where providing information to the market has a large free-rider problem.
For example, banks kept so much AAA CDO’s on their balance sheets (i.e. they never intended to make markets in them) because of the low cost of funding they had. The repo and money markets for banks should have suffered due to their high amount of information asymmetry and principal-agent problems. But since nobody could envision AIG and broker-dealers going under, many pension funds, insurance companies and corporations invested with them on the basis of brand name, their supposed capital according to regulators, etc. Short-sellers did eventually expose Bear and Lehman, but short-sellers can only make money after the shorted companies have already received too much capital. Otherwise, information digging into companies has a big free rider issue which government may be able to solve.
5. February 2012 at 21:50
“And that’s why EMH is useful, even if it isn’t true, and anti-EMH is useless, even if it is true.”
Well, here’s the thing. A refutation of some scientific theorem, in and of itself, is almost never useful… past making sure not to rely on that scientific theorem.
To put in scientific terms, I wouldn’t say the EMH is “refuted” like a flat earth or the ether theory of light. It would more be like how relatively and quantum mechanics partly refuted Newtonian mechanics. Newtonian mechanics still work a lot of the time, but things could go horribly wrong if engineers for a satellite did not correct for where Newtonian mechanics was wrong.
For when the EMH could go horribly wrong, think of all the investors who dutifully plowed money into the stock market at its 1999-2000 peak. I have to be careful to say I am not talking about market timing. Like I said, predicting market sentiment is mostly based on luck. But if investors could have a more toned-down EMH view, which allows for bubbles and gross misprincings to still happen, then they may have steered clear of investing in the indexes in 1999-2000.
The same goes with the ultimate funders of the housing bubble, subprime CDO’s, who often used circular logic to defend holding them. They should be held because their price is high, and their price is high because everybody things they should be held. There was *a lot* of money to be made by shorting them, but for behavioral finance reasons (in particular the availability heuristic), very few people with money wanted to short them. In the “Big Short,” it explicitly shows how one investor said “Goldman Sachs is holding them and they must know what they’re doing.” They couldn’t comprehend a market where all these big players were so wrong. (GS later shorted the market before everyone else, but they were very long CDO’s at the time the comment was made).
In short, for both personal finance and economics, the fact that markets *can* have gross mispricings is the main, non-trivial point of the anti-EMH view. From there, it starts getting more complicated. Is the market today overvalued? How much regulatory disclosure should we require? Should we start money market funds and repo markets as pseudo-banks to stop runs, which can happen even if they’re fundamentally solvent? These questions may all answer in the way of “it’s not worth second-guessing the market’s value” or “it’s not worth the price of government intervention,” but asking the question is the first step.
6. February 2012 at 06:42
Morgan, I doubt the government could even distinguish between the two. Lots of entrepreneurs have huge stock gains.
123. No, I want to keep the peg the same, and have the Fed supply more money than the market wants, if they really think the market forecast of future NGDP (in the NGDP futures market) is higher than the optimal forecast. That keeps the Fed accountable. If they lose money, it shows your premise was wrong.
Matt, I agree with many of your points. Let me just say that if the EMH was as true as Newtonian physics, I’d consider that a great achievement. Indeed it would be the only model in economics to reach that level of accuracy.
I do believe the EMH is not literally true, but is more accurate that the QTM, PPP, Keynesianism, monetarism, perfect competition, and lots of other models.
6. February 2012 at 09:53
What is written on an astrology chart or the fact that a cat crossed the path is taken as leads people to revise there subjective estimates of the values of assets — as do growth numbers released by financial firms reporting short terms gains on portfolios that will generate long term losses.
We are back to EMH being a 3 cup and ball shell game.
Scott writes,
“Stocks should reflect all available .. empirical or theoretical evidence that leads people to revise their subjective estimates of the value of assets.”
6. February 2012 at 10:02
Geez, this is like saying bloodletting is justified because it is more accurate than poisoning someone with mercury.
Do you ever test these ideas out with simple paralell cases or with a audit of simple potential counter-examples?
Scott writes,
“I do believe the EMH is not literally true, but is more accurate that the QTM, PPP, Keynesianism, monetarism, perfect competition, and lots of other models.”
6. February 2012 at 13:38
MF Global revised their subjective estimates of the value of assets. asee on their perception and vision of the empirical and theoretical evidence — and EMH predicts that stock prices should reflect that, and because prices reflect that, no one can profit from thinking otherwise.”
Is that what you are saying, Scott?
I’m taking your direct words and I’m applying them to an actual empirical case.
Where did I go wrong?
Scott wrote,
“Stocks should reflect all available .. empirical or theoretical evidence that leads people to revise their subjective estimates of the value of assets.”
6. February 2012 at 13:39
“based on”
6. February 2012 at 14:45
Scott:”No, I want to keep the peg the same, and have the Fed supply more money than the market wants, if they really think the market forecast of future NGDP (in the NGDP futures market) is higher than the optimal forecast. That keeps the Fed accountable. If they lose money, it shows your premise was wrong.”
Your preferred policy pegs not only the NGDP futures, it also pegs the NGDP futures risk premium at zero, thus providing a subsidized insurance to market participants who are risk-averse to NGDP volatility. This subsidy is socially suboptimal, it also reduces Fed’s profits.
My preferred policy is to fix a NGDP path, and establish a corridor (for example +/- 0,5 percentage points); NGDP futures could fluctuate inside a corridor. This way Fed’s profits are preserved, and NGDP futures risk premium would be closer to socially optimal level.
6. February 2012 at 19:03
[…] will probably get mad at me for posting a link about the efficient market hypothesis, but screw him. I liked this […]
7. February 2012 at 07:30
Greg, You said;
“EMH predicts that stock prices should reflect that, and because prices reflect that, no one can profit from thinking otherwise.”
You must think Fama’s pretty stupid. People often profit from “thinking otherwise.” Do you think Fama doesn’t understand that fact.
123, I don’t think my plan pegs a risk premium, indeed I don’t even understand how such a concept is possible. The price is set by the Fed, the risk premium is determined by the market.
Any subsidy to traders is a second order problem compared to the macro goals of the policy–which need to come first.
If you make the corridor plus or minus 0.1%, then I guess I’d have no objection
7. February 2012 at 08:11
Scott,
first, my point about the religion was not entirely serious, if we are talking in strictly literal terms, i don’t literaly believe that about Fama either even though he’s wrong about this one.
now regarding your points on anomalies:
1. “My” anomaly is not comparable to the Superbowl, the main insight of the anomaly I found is not that if you invest in stocks with big brand names you make money, it’s that once you include brand name values in the fama-french model, the explanatory power increases, the significance of BM disappears and it is replaced by the brand name variable which has a different sign than the BM value; exactly what you would expect if the market was undervaluing brand names (If you have an alternative explanation i’d love to hear it, I of course accept that this could be a coincidence, but if it aint i can think of nothing else).
That said, it’s an anomaly that is trivial and you can indeed find some more like it, my point was that this is the reason you can actually beat the market, if you can spot this kind of thing when it’s present, as professional traders are paid to do. But anyways, i’ll leave it at that since this anomaly is trivial by itself, which is why I said I would make a really bad academic economist as I always seem to be interested in trivial stuff.
2. The “momentum” anomaly is another story. It’s not an anomaly, because it’s persistent in pretty much every market in the world (at least I’ve read papers confirming it for many more markets than the american one). Yes, if you found correlation between superbowl wins and market returns it would be spurious, but what if you found that in every European stock market the returns were correlated with who wins the local soccer championship?
And this is why the EMR guys like Fama are wrong, notice that he does not “acknowledge it” in a conventional way, he just joined a huge discussion about the findings of empirical finance, made some high profile publications where he examined the arguments of the other side (very convincingly by the way) except for this one, on which he said very little and nothing related to how this anomaly can be reconciled with his version of the EMH.
On the other hand the anomaly can be explained very well through the noise trader risk model, which also explains how you can believe that rational expectations in your model is a good thing but efficient markets isn’t (which is my main objection against the completely anti-EMH crowd), even though both boil down to the possibility of systematic errors by some profit/welfare maximizing entity. And it’s also the main reason I think you are wrong when you say (or imply) that there are no better alternatives to the EMH.
7. February 2012 at 10:19
Scott: “I don’t think my plan pegs a risk premium, indeed I don’t even understand how such a concept is possible. The price is set by the Fed, the risk premium is determined by the market.
Any subsidy to traders is a second order problem compared to the macro goals of the policy-which need to come first.
If you make the corridor plus or minus 0.1%, then I guess I’d have no objection”
If NGDP futures are pegged, and all market participants are risk neutral, then the market expects NGDP to hit the target. But since market participants are not risk neutral, there is a futures risk premium, and markets expect that NGDP will be above target (or below, depending on the sign of the risk premium). This is not a huge problem because the path targeting helps a lot. But a corridor peg is better, because in practice central bankers tend to see any subsidy to traders as a very big deal. The corridor could also help central banks to set the price of NGDP futures that corresponds to market expectations (not prices) being on target after making an allowance for the NGDP futures risk premium. BTW, there some papers from regional FEDs where researchers try to estimate the inflation expectations from TIPS data after making adjustments for TIPS risk premia.
7. February 2012 at 15:33
I’m trying to limn what your arguments and claims are, Scott, not
Part of the problem is the “the EMH” has no fixed meaning or content, as you use is.
I’d like to nail it now a bit for the purpose of engaging your case against Kling, Hayek and “asset bubbles”, but so far, it’s been like trying to nail jello to the wall.
Scott wrote,
“You must think Fama’s pretty stupid. People often profit from “thinking otherwise.” Do you think Fama doesn’t understand that fact.”
7. February 2012 at 16:48
Sorry, Ipad auto-edits are a disaster.
now = down
and make the end of that first sentence “not Fama”.
7. February 2012 at 16:51
Beyond the original referent for Fama’s insight — the early statistical work on stock prices, e.g. the Random Walk stuff and such, I don’t think Fama or anyone is has very clear necessary and sufficient conditions for distinguishing between Fama’s multiple levels of EMH — which is great for generating a literature and great for allowing people to cite “the EMH” when making expansive and open-ended claims, but not so great when you are doing science.
7. February 2012 at 16:57
Scott, which form of the EMH are you using?
“Fama proposed three types of efficiency: (i) strong-form; (ii) semi-strong form; and (iii) weak efficiency. They are explained in the context of what information are factored in price. In weak form efficiency the information set is just historical prices, which can be predicted from historical price trend; thus, it is impossible to profit from it. Semi-strong form requires that all public information is reflected in prices already, such as companies’ announcements or annual earnings figures. Finally, the strong-form concerns all information, including private information are incorporated in price; it states no monopolistic information can entail profits, in other words, insider trading cannot make a profit in the strong-form market efficiency world.”
8. February 2012 at 14:44
123, With level targeting the risk premium really isn’t a problem at all. Suppose it cause NGDP to consistently be 0.3% (on average) above target. In that case the growth rate of GDP will stil be 5%, except in the very first period wher eit will be 5.3%. And in the long run it’s the growth rate that matters.
Orionorbit, I think you are still missing my point. The way to attack the EMH is not to convince me that you’ve found a way to beat the market, but rather to convince me that others have found a way to beat the market. If what you say is true then the class of mutual funds and hedge funds that invest using these anomalies will outperform others. That means excess returns should be serially correlated.
Greg, I think the best case can be made for the weak version. I think semi-strong is also pretty accurate, and the strong form less so.
9. February 2012 at 10:39
Scott, yes, I was unclear about that in my first post. But actually most hedge funds do outperform the market BEFORE charges. Or to put it more clearly, there exists a non negligable number of funds (and traders) for which you can predict the next period’s alpha from the present period’s aplha to a reasonable degree, if you are measuring alpha BEFORE commissions. This is not consistent with the predictions of the EMH.
Now, if you argue something like “well yeah, but there are many hedge funds so some of them will outperform the market”, my response is true but trivial, because when you have even a small number of people that can beat the market, its natural that a few idiots will jump in to extract rents and their numbers will grow as long as there are enough idiots to invest money with them. This doesn’t mean that it’s impossible to beat the market.
And since you keep bringing up the same argument over and over, let me make a little over the top example that will hopefully kill this damn zombie once and for all. Let’s say we test 100m runners. Now, if we take any number of runners, some of them will run bellow 10 secs because of skills and training and most will run around 11 secs which is most consistent with the performance of the average athlete. Of course, the average runner will run some races around 10.50 and maybe if he runs the race of his life he can get bellow 10. We do know however that there exists a small number of runners that can be expected to run bellow 10 secs in many major races in their lifetime. The reason we know this is not due to chance is not the distribution of the entire sample of runners, but because the INDIVIDUAL scores show patterns of out performance that could not reasonably be attributed to chance.
Likewise for the EMH, it doesn’t matter that most people underperform. What matters is if there exist enough entities where tomorrow’s performance can be predicted from yesterday’s performance. If they exist, a central prediction of at least the semi-strong and the strong versions of the EMH is shown untrue. And to my knowledge there are such funds, but the more striking evidence of the existence of such skills is the fact that banks like Goldman Sachs pay most of their salaries to traders that buy stuff they think are going up. So here you either pick EMH + every major investment bank acting in an irrational and suicidal way for the past 100 years, or that the EMH is wrong.
Furthermore, this “individual outperformance” needs not be in the same asset class, or strategy, as you seem to believe (so that I can show you something like “there!! look!! funds who buy corn every full moon get rich!!!”. Real life traders and most hedge funds are not restricted in trading strategies, and even most mutual funds are restricted by asset class alone. And since funds and traders will move classes frequently, i can’t show you what you ask. So the only meaningful piece of information to determine whether the EMH holds is whether you can predict future alpha from the past alpha, so that you are incorporating the choice of fund class and trading strategy in your alpha.
Now if you want to say something like EMH holds under the assumption that no strategy switching is permited, yes, I can agree to that, but this is not at all consistent with Fama.
Finally, the above is more of an attempt to convince you that the noise trader risk model is the best approximation of the market that exists, not an attempt to refute the EMH. The EMH is refuted even by your own standards on the basis of the momentum effect (so we don’t even need to go to the serially correlated alphas of GS traders). I still haven’t heard an explanation of the presence of the momentum effect in so many markets that is consistent with the EMH…
10. February 2012 at 09:07
Orionorbit. I think you get the sports analogy wrong. If there are SOME traders who really have skill, then the statistical pattern should be like sports. Looking at ALL traders, should show a pattern where those successful one year tend to be successful the next. If you restrict your sample to some traders, you run the risk of data mining–in which case it might be luck. But if you look at all traders, and some really are skilled, then that should show up in study that looks at the conditional probability of an average trader outperforming given he did so in the previous year. You CANNOT do those studies with a restricted sample of successful traders, as you will be assuming the conclusion. I have no doubt that the annual performances of the world’s richest men are positively correlated. The same is true of the world’s most successful roulette players.
10. February 2012 at 13:14
Scott, keep in mind i am talking about alpha, NOT returns, that is serially correlated. That said, I think this is checkmate in two:
Selection bias is not an issue because we do have an estimate of the population mean, that is the historical return of the market portfolio. Or in the case of runners, the median running time for 100m. When we have the mean of the population and series of a finite sample, the results of the efficiency tests asymptotically converge to those for the case that our sample is the entire population, even if your sample is biased (every sample is). So the only way you can justify your argument is that i don’t have enough observations to establish that the results of our efficiency tests sufficiently converge. And yes, here’s where my case weakens because even though the samples that I am looking at ARE sufficient (we are talking a few dozen traders, weekly re-balancing over three years), nobody would ever share that kind of data. But I have seen them and GS DOES pay the bonuses to their traders and then you got guys like this:
http://www.hussmanfunds.com/pdf/hsgperf.pdf , so you can confidently “buy” my story if you assume the slightest shred of goodfaith on my part and/or you are not willing to concede that investment banks are irrational.
And now I think this is checkmate in one: Paul Krugman enters a casino. The casino players play only roulette and all of them have to keep betting in every round or they have to go. Paul Krugman watches 100 rounds of game selects the 10 players with the best records, he “picks the winners”. Paul Krugman gives each player a million and says “play for me”. Then they go and play 100 more rounds and again, the 10 players that Paul Krugman chose double his money. I.e., contrast to what randomness would predict, the winners Paul Krugman chose kept on winning, BIG. Let’s say that you estimate the probability this would happen randomly is 1 in a billion.
The next day, police arrests them all and Paul Krugman appears before of Judge Sumner (because it’s your blog, you get to be the judge and jury in this trial, also Paul Krugman does not get any civil liberties because he’s a damn hippy and you get to send him to jail based on your hunch). He says, “your honor, I swear we didn’t do anything tricky, our performance can be explained by chance and selection bias. I chose the winners after all!”. Do you send him to jail?
Question 2: If you had no idea about the rules of roulette so you could not calculate the population mean for the expected payoffs for playing a game of roulette (for all you know they could have been positive), would you have sent Paul Krugman to jail?
I am no grandmaster but think this is checkmate.
11. February 2012 at 07:02
orionorbit.
1. 3.8% over ten years is a lousy return, even I’ve done much better than that in the stock market. What am I supposed to be impressed about?
2. The Krugman story has no bearing on the real world, because there is little or no evidence of successful investors one year (on average) continuing to be successful the next. Fama’s done research on this. Yes, you can find individual cases where it happened, but no systematic evidence.
12. February 2012 at 15:40
1. I think I have said many times that i don’t care much about average performance, only aplha. 3.8% is a really lousy record if your beta is one but that guy’s isn’t. Also If you had followed the EMH prescription you would have actually lost money. Finally, your point about doing much better than that, is kind of supporting my view that out-performance in the stock market seems too correlated with knowledge of economics for markets to be efficient.
2. If I remember correctly Fama was using data from retail stock brokers in his research; no fair, these people are noise traders so his results are entirely consistent with the Shleifer viewpoint. If he does the same thing with Goldman Sachs traders and finds no evidence of serial correlation in risk weighted performance, I’ll concede he’s right on this one (still the EMH doesn’t explain momentum effects while Shleifer’s model does and is inconsistent with the impossibility of informational efficiency while Shleifer’s model isn’t).
13. February 2012 at 11:54
orionorbit, I’m in no position to know how that guy invested, just that 3.8%/year for 10 years doesn’t seem impressive to me. My own investments were not based on my knowledge of economics, I tend to buy index funds where possible, mostly international.
I would expect traders at Goldman Sachs to have done better than average, they are one of the most successful banks in all of world history.
Fama also looked at mutual fund returns, and this article shows that the high returns to hedge funds were a fluke, they’ve fallen back to earth.
6. March 2012 at 15:40
[…] Passive investing (e.g., indexing) is predicated upon the efficient markets hypothesis. To oversimplify, that hypothesis asserts that because asset prices reflect all relevant information and that investors act rationally on that information, it is impossible to “beat the market” over time except by being extremely lucky. The evidence supporting this idea is surprisingly strong. […]
3. May 2012 at 13:37
[…] the market” over time except by being extremely lucky. The evidence supporting this idea is surprisingly strong as a practical matter even though aspects of it are easy to […]
3. September 2012 at 16:26
[…] says that bubbles can’t happen, but weaker forms also exist, allowing its proponents to weave between the different types, effectively endorsing the strong form but falling back onto the weak […]
9. September 2012 at 04:50
There is a very simple argument–actual data. The EMH requires that price changes are continuous. There’s absolutely no reason for this.
Also, using a random walk to describe markets is complete and utter nonsense. Take the past 50 years of the Dow Jones data and find its Hurst exponent–it won’t be equal to .50, it will probably be around .6. Movements in the market are certainly correlated–which is actually what creates the “fat tails” that we see in financial data.
In reality, the market is filled with different investors of different time frames that respond in different ways to different pieces of information. What might be a must sell for a day trader might be a buying opportunity for a longer term investor. Different people respond differently to different sources of information because the market consists of people working on different time frames.
22. October 2012 at 04:51
[…] exist. The only defence its proponents seem to be able to muster is that it can’t predict anything (and sometimes, that economists full stop can’t predict anything). I could go on, […]
4. June 2013 at 23:20
Here are some thoughts I have on it his points:
“The managers that perform best initially will tend to attract more investors, and so will gradually become bigger than the moderate or poor performers (who will eventually go out of business).”
Size indeed is a hindrance to out-performance over a long time period. The best investors of today understand this constraint, and close their funds to new money as a result. This way they can have small enough capital to still participate in compelling investments. For instance, when Buffett was young he was able to compound money at above 30% for 15 years straight by being a stock picker and understanding where markets are inefficient. Today he has hundreds of billions of dollars to invest which makes his opportunity set much smaller (less liquidity), not allowing him to participate in the inefficient markets as much. That being said, he has still outperformed the market substantially nonetheless.
If you only read one thing this comment, read this: http://www.tilsonfunds.com/superinvestors.html
It’s an old transcript of a Buffett speech given at Columbia University (essentially the polar opposite of UChicago ideology).
Buffett always says: “In any sort of a contest – financial, mental, or physical – it’s an enormous advantage to have opponents who have been taught that it’s useless even to try.”
I agree with Sumner that the majority of hedge funds are not worth paying up 2/20 and the average person would probably be better off simply buying an index. BUT, for the minority of people who are willing to conduct thorough research, they can generate massive out-performance for extended periods by staying in markets that are inherently inefficient (like distressed debt where information is difficult to find and technically rigorous with bankruptcy speak, etc., or small capitalization stocks that are not profitable enough to have formal institutional research coverage). Small cap stocks are also often inefficient because the smart money is too big to participate. These shares are too illiquid as the smart money would have to take too illiquid of positions to make it worth their while.
He says: There should be “anti-EMH” mutual funds that invest on the assumption of market inefficiency, and these should tend to earn above normal rates of return on long term investments.
These funds do exist actually… They’re called value investors, and have outperformed the S&P for very long periods, see the above article (Buffett, Greenblatt, Mohnish Pabrai, Munger, Steve Romick, Seth Klarman, etc.). These investors all adhere to the same school of thought called value investing, which is a strategy that seeks to grow capital over long time periods and exploiting price inefficiencies that arise for whatever reason, be it emotional selling, systemic inefficiency, etc. They absolutely made a killing during the 2008-2009 crisis. Like he says, EMH is aesthetically beautiful and academics love it. However, it is not pragmatic because there are strategies that allow investors to exploit systemic inefficiencies – such as forced selling of securities by institutions.
Here’s another example of markets being inefficient. Sometimes a phenomenon happens when a stock trades for less than its liquidation value. I’m not sure how familiar you are with accounting, but a trivial example goes as follows:
Imagine company A has a stock price of 10 dollars. The company holds 12 dollars of cash, 5 dollars of inventory, and 5 dollars in buildings/equipment, for example. The company also has 8 dollars in debt. So in total the company has $22 in assets and 8 dollars in debt. So If you wanted to liquidate all the assets of the business, that should cost $14 because you could purchase the whole company, get $12 in cash, sell the inventory and equipment and buildings and get $22 in cash total. You would also have to pay off all the debt, meaning you would have 22-8 left = $14 per share. But for whatever reason, the market price of the stock is only $10. The market is essentially saying that the value of the business is not even worth the liquidating value of the assets. The company would be worth more if it didn’t even operate! If this happens, you could do asset arbitrage by buying all the shares and liquidating assets and pocketing $4 per share. That is value investing and this is one of the ways Buffett made his fortune – because often times the market puts abdurd valuations on stock prices that a resourceful investor can exploit. EMH says this situation should never occur, and in fact it happens all the time in small cap stocks. Would love to hear your thoughts.
18. April 2016 at 12:55
Can’t believe I missed this one.
Two links for anyone looking at this ancient history:
https://www.aqr.com/cliffs-perspective/hedge-funds-the-somewhat-tepid-defense
http://www.bloombergview.com/articles/2014-03-03/hedge-fund-genius-isn-t-really-sold-on-hedge-funds
18. February 2017 at 09:15
[…] http://www.themoneyillusion.com/?p=12801 […]
25. April 2017 at 02:34
[…] http://www.themoneyillusion.com/?p=12801#comments […]