Asymmetrical Information

Jane, I think your formulation: "In short, prices may not match the intrinsic value of the stock. But the odds are against you systematically outperforming the market in estimating what that intrinsic value is" is exactly the right way to put it. But this is a long way from strong-form EMT or arguably even semi-strong, which is why I think it's fair that the EMT is not correct.

To echo you, when I say the market is (relatively) efficient, what I mean is just that over time, the market price of a company is most likely to be the best forecast possible of its intrinsic value. Now, the best possible forecast could be way off, but it's going to be very hard to consistently come up with a better one. This is an attenuated definition of efficiency, but I think it's the correct one. Wisdom of crowds, and all that.

what? you mean that investing on wall street is basically like gambling? you're kidding me?

what's the name for the theory that describes the tendency of all academic study in economics to drift gently towards common sense? no doubt soon there'll be a backlash as economists feel a bit of job pressure (or even existential pressure), and they'll start generating theories again - like, say, the 'perfect information paradigm' that claims that yes, there is such a thing as perfect information after all.

sorry, i sent my post before i'd finished it, but the facetious punchline no longer seems appropriate. so, on a related matter, what will happen when everyone figures out what's actually going on and the whole world switches to index funds?

Sometimes it rains on Derby Day, sometimes the government attacks you with an axe, sometimes a plane flies into a building - all interuptions to the expectations. And sometime you find 100 original Ford shares in a shoebox.

Wow, a post from James Surowiecki! I swoon.

I am unfortunately not as familiar with the efficient markets literature as I should be. My experience with the theory comes entirely from going to business school at Chicago, where all the professors came down pretty heavily on the efficiency side. But none of them seem to be advocating the insane "markets always produce the metaphysically true price" position that I see detractors making fun of, not even Kevin Murphy or Eugene Fama. What they seemed to be arguing was that there was no way to systematically get to a better price. That's why they tend to be so anti-regulation; all my professors recognized that the market failed, but in cases where the failure was of human judgement, they expected potential regulators judgement to fail in the same way.

But if you tell me that there are people who are making what I'd call the "superstrong" EMT argument, I will be happy to join you in excoriating them for their silliness.

Jane hits the nail pretty squarely here. To show that EMH is wrong you need to describe a strategy that consistently outperforms the market on a risk-adjusted basis. Just pointing out that some prices get silly every now and then doesn't do it.

The value investors have a case of sorts, but its strength is unclear to me.

Meanwhile, all who knew the dotcom bubble was absurd will please report the profits they made by shorting internet stocks.

First, index funds are indeed the only sane investment for 99% of investors.

Second, Warren's last name is spelled Buffett.

Third, in Mr. Buffett's defense is his own version of the coin-flipping story (a shorter summary is here).

Everyone with any interest in the stock market should read the full version of Buffett's article. He begins by setting up exactly Megan's point about bell curves and standard distributions, and then goes on to make a counter-argument. To whet your appetites, here's the set-up passage:

"I would like you to imagine a national coin-flipping contest. Let's assume we get 225 million Americans up tomorrow morning and we ask them all to wager a dollar. They go out in the morning at sunrise, and they all call the flip of a coin. If they call correctly, they win a dollar from those who called wrong. Each day the losers drop out, and on the subsequent day the stakes build as all previous winnings are put on the line. After ten flips on ten mornings, there will be approximately 220,000 people in the United States who have correctly called ten flips in a row. They will each have won a little over $1,000.

Now this group will probably start getting a little puffed up about this, human nature being what it is. They may try to be modest, but at cocktail parties they will occasionally admit to attractive members of the opposite sex what their technique is, and what marvelous insights they bring to the field of flipping.

Assuming that the winners are getting the appropriate rewards from the losers, in another ten days we will have 215 people who have successfully called their coin flips 20 times in a row and who, by this exercise, each have turned one dollar into a little over $1 million. $225 million would have been lost, $225 million would have been won.

By then, this group will really lose their heads. They will probably write books on "How I Turned a Dollar into a Million in Twenty Days Working Thirty Seconds a Morning." Worse yet, they'll probably start jetting around the country attending seminors on efficient coin-flipping and tackling skeptical professors with, "If it can't be done, why are there 215 of us?"

It isn't the EMT that is the problem, it is uncertainty in the Capital-Asset Pricing Model (CAPM). The more of an asset's future value that is dependent on price appreciation, particularly multiple appreciation (which quite possibly cannot be forecast) the more likely the valuation is subject to these self-reinforcing belief-driven bubbles. Mordecai Kurz explains, for those with an appetite. This is a less academic summary.

I have mentioned Kurz' theories here and here.

"Meanwhile, all who knew the dotcom bubble was absurd will please report the profits they made by shorting internet stocks."

Which isn't what the EMT claims. In fact, it argues against short-term investing strategies. Here's an amusing book by an economist who knew the dotcom bubble was absurd, lost his daughter's college fund trying to prove it, and then wrote about it:

http://www.amazon.com/exec/obidos/tg/detail/-/0814406491/ref=cm_cr_dp_2_1/002-4052385-9894416?v=glance&s=books&vi=customer-reviews&me=ATVPDKIKX0DER

Mindles,

If the efficient markets theory (EMT) relies on a capital asset pricing model that has quite a bit of uncertainty in it (due in part to the "rational" beliefs mentioned by Kurz in your posts), then the EMT leaves room for analysis to produce market-beating returns, doesn't it?

The general idea expressed in one of the posts you link to--that fundemantal economic volatility is down while total market volatility is up--suggests that there's money to be made if one can "correctly" price a security that is mispriced by the market (per the RBE-influenced CAPM); asusming one has the patience and werewithal to outlast some period of being wrong until the market price again reflects the "fundamental" price.

Am I missing something, or is that part of the point you are making?

(Disclaimer: I'm just a lawyer with a passing interest in these matters, and what follows may betray my fundamental ignorance.) Doesn't the thesis that the market will rationally set prices presuppose that the buyers (or at least the buyers representing most of the money) are themselves rational? If the only buyers were Vanguard, Buffett, etc., then you would expect rational pricing. But if you inject lots of amateur buyers with enough money to affect prices, won't you see increasing degrees of irrationality in prices, at least until those people go banrkrupty or withdraw from the market after getting burned? And isn't this what happened in the 90s? i.e., Lots of amateurs with lots of money (at least in the aggregate) enetering into the market, buying individual stocks on rumor and speculation in an attempt to outperform the benchmarks?

Jim- hat to say 'yes and no'

I do believe there is money to be made - but not easily and not necessarily swiftly.

You will notice that the regimes characterizing large changes in valuation multiples tend to be very long - longer than the 1-3-5-and even-10 year horizons in which managers are evaluated.

I actually disagree with the implicit comment above that few people considered the market overvalued in the bubble. Many, many did. The trouble was, nobody knew when it might correct. There's a saying in the Investment Management biz - "Late[early] is the same as wrong".

Most of these observers simply capitulated, others gave up - most famously Julian Robertson. There's someone who was 'right' and couldn't make money.

I find myself recommending Fooled By Randomness again.

" The weak version says only that you can't make excess returns--basically, profits above the economic costs of finding the investment opportunities--based on historical financial data. I think it's safe to say that pretty much every economist thinks this is true..."

There are many players in high-frequency trading, which use statistical analysis of historical financial data. The ones I know about make money.

If we assume the "many coins flipping" theory of investing success, and if we assume that this theory is era-independent (I infer the latter from the logic of the argument), this suggests the following:

If we look at a statistically significant sample of "successful investors" from some time in the past, their investing performance from that time in the past to the present should show a distribution that is very similar to that of any other group of serious investors.

This seems to be a falsifiable hypothesis; does the available evidence suggest that it is true? (Note that I don't know whether it is true or false.)

Mike, markets can rationally set prices even when many of the buyers in the market are irrational, as long as they're not all irrational in the same direction. (A bubble is precisely what happens when people are irrational in the same direction.) That's actually what's most remarkable about markets, that in them individual irrationality can produce collective rationality. And it isn't just markets, either. This is true of a whole host of collective decisionmaking mechanisms, ranging from the parimutuel pool at the racetrack, to aggregating people's guesses as to how many jellybeans are in a jar. (Apologies for the self-promotion, but this book I just wrote called "The Wisdom of Crowds" deals in large part with this exact problem.)

Jane, I think in practice you’re right: there are few people, even at Chicago, who would make the superstrong EMT argument. But the theory, which still informs a lot of economic work and a lot of work in corporate finance, does posit that the market price at any one moment is not just likely to be the best possible forecast, but that it’s actually “optimal.” It’s in that sense that I think it’s fair to say EMT is wrong. But the more modest definition of efficiency that I think you and I agree on (and that I think in practice most economists would adopt) seems to me to be roughly correct.

What’s especially interesting is that the assumptions that underlie the EMT and the CAPM – perfect information, rational expectations, homogeneous investors – are all clearly wrong. And yet the outcomes that markets reach, most of the time, are not actually that different from those that the EMT would predict: market prices are generally as good a forecast as you can get based on the information that’s out there, it’s very hard to make money betting against the market, etc. EMT is, strictly speaking, not right, but it gets close to the truth about market outcomes, and to that extent it’s still valuable.

Patrick,

As usual, your point is unclear to me. One hears a lot these days about how obvious the bubble was and how only fools got taken in, but these same talkers don't seem to have cashed in. I think there's a lot of hindsight in these comments.

"I been in some big towns,
I heard me some big talk"

Mike Ryan,

There have always been "noise traders" in the market. As long as the irrationality they introduce is not consistently bearish or bullish they will not bias prices, but will increase volatiity.

Mindles,

I understand your point, and agree, somewhat. I don't know how Robertson tried to capitalize on the bubble. Presumably he shorted a lot of stocks that obstinately refused to go down on his schedule. So was it a bubble when he shorted? Or only later, when prices were even higher? These things get tricky.

It's a little bit fun to 'combine disciplines' here, but the EMT, especially in its weak form, is an example of 'socially constructed truth'.

Can you make money in the stock market with just price data and a small amount of fundamental data, and do so consistently enough to prove that it isn't luck, or you're extremely lucky if you do. Yes, you can. There is one thing to remember about the EMT in the form that it is taught in school, not that it isn't salveable with modifications, it's entirely the creation of finance professors.

What do you call a finance professor who discovers an algorithm that delivers consistent above average returns (i.e. non-lucky)? An ex finance professor who is currently working for an investment bank, who also doesn't publish his results. The remaining finance professors are by default the ones who haven't discovered any such algorithm, and though this is not charitable to say, but is in fact the case, aren't going to due to lack of talent. With the usual dollop of academic modesty, they then assume that if they haven't done it or cannot do it, it is impossible, because finance professors of course know all there is to know about securities markets. Therefore, the EMT.

Whoops, EMT in above post should be EMH.

Im not very knowledgeable about investing. In fact, when Brando was first reported dead and 20 million dollars in the hole and about to have his home seized, I thought to myself, what great timing. Now I realize Brando still had money, but Brandos whole 'I could have been a contender' attitude with money and then dying with his creditors at the door would have been a great way to go out. I suppose it would have been even better if he had made his creditors an offer they couldnt refuse.

Ive always considered the stock market very similar to sports gambling. In sports gambling, the number you can 'buy' a team at (point spread) isnt based on reality, its based on the publics perception of reality. I think the stock market is very similar. For a while, if you had access to buy at IPO, you were almost guarenteed to make money if you bought and sold fairly quickly. Imo this was based on the publics perception of the IPO, much more than the companies actual business. In the late 90s there were also all kind of companies trading at insanely high prices that hadnt turned a profit over there entire existance. Imo, this doesnt refect the real world value of those companies, but the publics perception of their likely future performance. Final point to this unusual analogy is that in both the stock market and sports wagering the only people that win consistantly are those investors that have access to more information than the general public.

Yes, I think we agree, which isn't surprising, because I loved your book--everyone reading this should go read it, post haste.

You make a point that I've been wrestling with, because I just finished reviewing John Kay's new book, in which he trashes Milton Friedman's argument that rational models work even if rational individuals don't. But he doesn't really engage with the argument; just points out that it sounds silly. It does. And yet it works.

I think the question about Warren Buffett (sorry about the typo) is whether or not he himself is a case of selection bias. If we posit that there are a largish number of people out there investing along basically the same lines, some of them will form statistical clusters just by chance. However, the papers will write up Warren Buffett, who has outperformed the market; they will not write up Joe Smith and his disciples, who are now selling insurance in Peoria because their investment business went bust. As I've said in person, I think it's certainly plausible that Warren Buffett has, in fact, discovered some alpha. But to be really sure about that, you'd want to take two random groups of individuals and set one of them to investing by "value" metrics, while setting another to investing by some other strategy. This, of course, will not happen.

HH, the question is not whether you make money; people investing in the stockmarket will, on average, make money as long as the market is on a broad upward trend. The question is whether they make more money than they could have by jamming it in an S&P500 index fund and leaving it. On average, active managers of any stripe slightly underperform the S&P (due to trading costs), and chartists apparently significantly underperform.

jmct, that's an interesting point, but it seems to me that the community of finance professors is small enough that they generally notice when one of their own leaves and begins minting money as a proprietary trader. Also, as LTCM illustrated, finance professors who've found a secret way to mint money sometimes haven't.

"HF, the question is not whether you make money; people investing in the stockmarket will, on average, make money as long as the market is on a broad upward trend. The question is whether they make more money than they could have by jamming it in an S&P500 index fund and leaving it. On average, active managers of any stripe slightly underperform the S&P (due to trading costs), and chartists apparently significantly underperform."

High-frequency traders make money on average regardless of the direction of the market.

Of course I am aware that active managers underperform the S&P. However, HF traders are currently not of the same stripe. You have no data on their profits, since they are limited to investment banks and the larger hedge funds, who don't disclose their gains.

Eventually the knowledge will diffuse, as it always does, and the opportunities will close down. This will probably happen when more and more participants enter the field and cut the gains, forcing more risk to be taken for corresponding less profit. In hindsight researchers will mistakenly suppose that it was always thus, and that the profits currently generated were actually hiding more risk than traders thought, rather than questioning (as they should) their pre-suppositions about the EMT.

If I may add a slightly off side remark: While I believe in the weak form of EMT, I am also concerned with what constitutes "The Market" It's certainly not the S&P 500. It may not even be the Wilshire 5000. Does it weight all world stock markets equally? What about bonds and REITS, etc? How should my probably income factor into that market basket.

So if I really want to buy an index of THE market, what should that be?

Should have read "How should my probable income stream factor into that market basket?"

As James points out, the strong EMH is implicit in a lot of policy arguments. For example, the standard case for unrestricted international capital flows is that markets will produce the best possible estimate of the value of assets denominated in particular currencies. If EMH holds, this is self-evident. If not, it's an empirical question whether various forms of regulation and control might improve welfare, for example by reducing fluctuations in exchange rates.

On making money out of the dotcom bubble, the option of going short was only available to those with almost infinitely deep pockets (deeper than those of George Soros) given that no-one knew when the bubble would end. But as I point out in the original post, anyone could profit by selling their stocks in say, 1998, and waiting for sanity to return. That's what I did.

Good post, Jane. I came away with the same generaly sentiment from Chicago, if only a slightly greater belief in semi-strong than strong EMT.

Couple thoughts, though, and I haven't read the whole comments section or the link to the Buffett article, but I was under the impression that some of what made value investors successful was essentially inside information (or at least less than completely public info). Now some of that may have changed since the full disclosure rule went into effect, but at one time, companies routinely gave previews to analysts covering their stocks. Also, in the less than public info department, I can imagine that Buffett gets to tour plants and operations of potential investments at a degree that you and I might not. Surely this access allows him and those like him an opportunity to find greater value that the general investing public would not. I think it is this imperfect information problem is the biggest factor undermining EMT and making the theory semi-strong at best.

But if Chicago is right, even if the asset price isn't the intrinsic price, there is nonetheless no systematic way to reach a better price. That is not the same as saying that there is no better price.

I sold out my pitiful holdings in 1998 as well, except for my 401K. As it turns out, I would have done better to hold on and sell them when I went to school. Early is the same as wrong . . .

"To echo you, when I say the market is (relatively) efficient, what I mean is just that over time, the market price of a company is most likely to be the best forecast possible of its intrinsic value."

Quiggin got to this point first, but...from about late 1997 to mid 2000 it was very easy to make a better forecast of dot-com companies value than the market was doing. The problem was that with the way short selling works one would have needed very deep resources to trade on one's superior knowledge and actually make a large amount of money on it. People with those kind of resources and access were often better off trying to place their bets on public gullibility and buy IPOs.

At the macro level, the history of the 20th century U.S. stock market shows a lot of this kind of pattern, not a little. Just looking at the overall index level, over the 20th century the market has consistently overshot in its real growth predictions, then consistently overadjusted downward, then overshot again. The market has alternated long periods of stagnation (29-mid 1940s, 1967-83, suspect 00-??) with long periods of frentic growth (82-00, 47-67). One can ascribe this to new info and say that the predictions were always perfect, only the info changes, but that is somewhat tautological.

Nobody sensible would argue that the market is not an important, valuable, and in many ways effective institution for capital allocation. But the larger argument is how good it really is in an absolute sense and how much supplementing it needs. On this theme, one must also add that the market does not and is not designed to value a company's true overall contribution to social value (net of externalities); government action must force externalities to be internalized before this is true.

I'm not arguing that what happened in 1997-2001 was rational; only that there may not be a better systematic way to reach a price. Why didn't Alan Greenspan pierce the bubble? He certainly knew it was there, after all. But he was limited by the same factors that non-governmental actors were. And when we see systematically bad information processing in markets, as I'm more than prepared to admit that we can, we see exactly the same systematically bad information processing problems in democratic government, since the voters are the same people who are buying the bad shares. The more democratic a government is, the less it is even possible to "correct" market failures. Where government diverges from the market in its information processing, as far as I can tell, it diverges in ways that are unsalutory: higher noise-to-signal ratio, only positive feedback loops instead of positive/negative mix, centralisation away from sources of information, lower value placed on information gathering by actors, and so forth. The technocratic dream of efficient regulators is considerably farther off than the libertarian dream of perfectly efficient markets. But that's another rant . . .

"And I'm acquainted with a value investor, a disciple of Warren Buffet's, who makes a convincing argument that the Uncle Warren School of Value Investing in fact does generate excess returns for those who follow it closely. This may well be true, but it's somewhat recursive: value investing (like efficient markets) only works as long as a lot of people don't think it works"

Wouldn't there be an argument that Buffett's type of value investing is more like a private equity play, though? Take damaged assets, take a more aggressive role in management than a typical passive investor, fix 'em up, and more on.

Liked your "weak", semi-strong" and "strong" analysis of the EMT, BTW. Think Quiggin was far to quick to proclam its death.

Two points of a statistical nature:

I think it's safe to say that pretty much every economist thinks this is true

Absolutely not. Every economist who has kept up with the literature over the last ten years knows that there are serious problems with weak-form efficiency when applied to high-frequency data. Over-reaction in short term price movements is now an extremely well-established phenomenon. Andrew Lo has a number of working papers on his website to this effect.

The very existence of Vanguard, I would argue, is proof of this fact. Vanguard does not have a "buy and hold" strategy; it reproduces the performance of the SP500. It manages to achieve this despite the fact that it pays commission and slippage every time the SP500 is rebalanced and is regularly obliged to sell stocks cum dividend. Given that it still manages to reproduce the performance of the SP500 to within an epsilon, it must be systematically generating a positive return from its trading activities.

2. The "coin flipping" argument about Soros and Buffet is not statistically valid. If it was the first time anyone had ever made it, then that might be fair enough, but people have in fact been trying to explain Soros and Buffet away in this fashion for more than ten years. This means that we are not looking at the simple mathematical truth that a twenty-year coin tossing competition will produce winners. We are looking at a ten year competition which produced winners, then a second ten year sample which produced the same winners. That would be laughably statistically improbable (I also note that the sample size mentioned in the post above is screwed up; far fewer than 225 million Americans have any material stock market investments).

And finally, I would note that turning a market inefficiency into profits is not at all a trivial idea. You can't get more simple than the idea that if Royal Dutch and Shell trade at a ratio other than 60:40, then there is an arbitrage, but I have sat next to a man whose job was to carry out that trade and you would be very surprised indeed at the logistical support he needed in order to do his job. Finance professors do move into the investment community, and they often do very well there.

I just wanted to flag the fact that jn's point about "what is the market" is very important, in my view. It seems to me to provide an explanation for why an uninformed index investor can't match the performance of informed investors. I've never had the time or technical skill to formalise this particular argument though.

John, I don't think you needed infinitely deep pockets, as betting on a drop doesn't mean going short. Buy puts instead. You could have gone to an invesment bank, and asked them to come up with a portfolio that would pay you off as much as possible for a $1000 bet in the event of a crash of x% magnitude over the years y-z, and they would have given a quote (with their customary huge markup). You'd need to buy the puts with strike prices determined by at various prices.

One very crude portfolio would be to simply buy puts every day (with strike price given by current market prices), and as soon as the crash hits you'd recoup more than all your losses.

Soros made a bet (I guess on the timing and magnitude of the crash), and he was wrong. Were he right, he'd be even richer now.

Markets behave like nonlinear systems, where small changes in initial conditions bring wildly different results. Efficient markets require a perfect understanding of market dynamics, and this has not been achieved.

Chicago professors don't argue that there are no arbitrage opportunities, D2; they argue that the economic cost of finding those opportunities is approximately their economic return, and hence that the market is at least semi-strongly efficient. Nor did I argue that finance professors don't go into the private sector and do well; only that if they were making returns that suggested that they'd found some surefire way to game the market, their former colleagues would, in fact, notice.

Because the investment community is ruled by the herd mentality, markets are illogical and inefficient in the short run.

The markets are logical and efficient only in the long run. (over 5 years). That is why most people should invest only for the long term.

One very crude portfolio would be to simply buy puts every day (with strike price given by current market prices), and as soon as the crash hits you'd recoup more than all your losses.

Sadly, this strategy would require very deep pockets ...

Chicago professors don't argue that there are no arbitrage opportunities, D2; they argue that the economic cost of finding those opportunities is approximately their economic return,

Perhaps they do, but this argument is clearly a structural one which is not particularly well related to the statistical concepts of weak, semistrong or strong efficiency. Those are all statements about the information content of prices.

and hence that the market is at least semi-strongly efficient

A claim which (subject to the caveat above over whether it is meaningful at all) has been thoroughly empirically falsified. The near-term overreaction, long term under-reaction effects are pretty well established and are not consistent with weak-form efficiency unless one assumes that the returns earned are compensation for a pretty weird kind of risk.

Nor did I argue that finance professors don't go into the private sector and do well; only that if they were making returns that suggested that they'd found some surefire way to game the market, their former colleagues would, in fact, notice.

This is a sociological statement about finance professors which is not necessarily true, and in my view most likely untrue.

I thought Shleifer, et.al., had shown that there was actually short-term underreaction to good news, and a gradual drift upward. I also don't think the "long-term underreaction effect" has been pretty well established. And the Jung and Shiller paper from 2002 shows, in Shiller's words, that "for U.S. stocks that have been continually traded since 1926, the
price-dividend ratio is a strong forecaster of the present value of future dividend changes. So, dividend-price ratios on individual stocks do serve as forecasts of long-term future changes in their future dividends, as efficient markets assert." That seems a fairly strong statement that prices do contain significant information about the future performance of companies.

Since I snoozed through all my college econ courses, my grasp of most discussions is superficial. So, thank you for these discussions of economics. Thankfully I learned enough to follow them, and hopefully correct the gaps in my knowledge.

One of the arguments in favor of some form of EMH has been that the very large majority of professional fund managers underperform a low-cost index fund. However, this statistic should (but seldom is) qualified in that it supports EMH in the case where the "investor" has to manage large sums of money (hundreds of millions up to a few billions of dollars) and may have to do so subject to fund rules that preclude many strategies (eg, if the rules require the manager to be 95% invested in equities at all times). Finding a persistant pattern in which one can invest a billion dollars without those investments affecting the pattern itself is difficult. For the individual investor working with a much smaller amount, and without the extraneous constraints, the situation might be quite different.

To be a bit more precise, the dotcom mania was evidence against the semi-strong version of EMH, since there was plenty of public information to say that prices were too high. The semistrong version is probably the most policy-relevant.

d2: "slippage every time the SP500 is rebalanced..."

What slippage? The rebalancing is based on a fixing price.

"And finally, I would note that turning a market inefficiency into profits is not at all a trivial idea. You can't get more simple than the idea that if Royal Dutch and Shell trade at a ratio other than 60:40, then there is an arbitrage, but I have sat next to a man whose job was to carry out that trade and you would be very surprised indeed at the logistical support he needed in order to do his job. Finance professors do move into the investment community, and they often do very well there."

And they often do poorly as well - finance professors tend to be looked down at in the profession as being too la-di-dah.

jq: " anyone could profit by selling their stocks in say, 1998, and waiting for sanity to return. That's what I did."

Probably sold in panic in the midst of the LTCM crisis...

Whoops, better take back that comment on a fixing price...! Bloody stupid Americas.

"If it was the first time anyone had ever made it, then that might be fair enough, but people have in fact been trying to explain Soros and Buffet away in this fashion for more than ten years."

I have no emotional (ahem) investment in the issue, but isn't it possible that some of the first 10 years was luck, and some of the second 10 years was self-fulfilling prophecy as Buffet and Soros watchers followed their lead?

Perhaps I'm not understanding the weak version of the EMH well enough, but it seems to me philosophically possible that there could be a system of analysis which gives made results from time to time, but which is nevertheless the best possible system for getting results on average. In other words it seems possible that even though a system (let us call it the fairly free market) produces sometime bad results, that it might be better than all other systems (government controls A-Z for instance), across any medium-long term.

Mindless H. Dreck writes:

"I actually disagree with the implicit comment above that few people considered the market overvalued in the bubble. Many, many did. The trouble was, nobody knew when it might correct."

Here's a funny data point from a january of 2000 slashdot thread, a few months before the bubble burst:

"Oh yeah -- dont sell. Thats another mistake. I know people who sell off thier options to pay for cars and crap and regret it later -- we are in a major tech boom its a little early to take the money off the table IMO. ... untill then ride the hype for all its worth."

some of the second 10 years was self-fulfilling prophecy as Buffet and Soros watchers followed their lead

I don't think so, but even if true, self-fulfilling prophecies are prophecies and thus not consistent with any form of market efficiency.

James S: as far as I am aware, underreaction is to earnings news and is ... reasonably well-established. "Overreaction", or more formally, the profitability of contrarian strategies, is quite likely due to lead and lag relationships between stock prices and very well established.

Dsquared, all of your points seem to amount to arguing that prices vary from intrinsic value over the short or medium terms, which as I've said several times, was a point hammered into me by the very advocates of efficiency you are decrying. What they said was that you couldn't make economic profits trading on the variance, not that there was no variance.

John, you and I rather prove that point. On average, we probably underperformed hotter heads who stayed in until the crash, because we missed a hell of a lot of runup that most people probably sold into while rebalancing. Or take someone I know who went short every year from 1997 to 2000. He made a killing in year four, but it's a good thing the crash happened then, because trading on that public information you cite almost bankrupted him.

On the topic of policy implications of this, I'm curious: in what way do you expect the government to systematically determine a better price than the market? What are its informational or decision making advantages over a distributed system? It seems to me that the mob mentality can as easily--if not more easily--prevail in a committee of twelve with no personal capital at stake.

Daniel, the underreaction is to earnings news, but at least in the Barberis, Shleifer & Vishny model, the underreaction is short-term, and in the long term there's overreaction, where you suggested there was "near term overreaction, long-term underreaction." I'm not sure how all these things can be simultaneously true. In any case, in Fama's 1998 paper -- and, to be sure, he may have an ax to grind -- he found that short-term overreaction to information was as common as short-term underreaction, so that there was no systematic market error.

Also, I don't understand what the "lead and lag relationships between stock prices" means. What are the stock prices that the relationships are between?

Dsquared,

I am curious as to the nature of the problems involved in carrying out the Shell trade.

Does not sports wagering hold an advantage for the short seller, in that there is a certain date of resolving the public's irrationality, thus negating the need for very deep pockets? There are people with published predictions on NFL games who have experienced annual success rates on their "best bets" between 60 and 70 percent for several years running. I don't know of anyone who does well when attempting to predict the outcome against the point spread on every game.

The fact that these public "best bet" predictions haven't simply been copied by the wagering public, thus bringing down the success rate closer to 50%, may indicate that the sports wagerer is, on average, more likely to be irrational in a single direction than than the equity investor, which increases the opportunity for the short seller, with the tremendous added advantage that the public's irrationality will be resolved by a certain date.

D2, I think a more established wisdom is that small-cap stocks have traditionally outperformed large-cap ones, and various rationale have been given (primarily less liquidity).

Will, I've heard of the oddly mispriced odds (eg, Basketball Stanford v Yale at even odds :)), but I'm sure the casino's would love people to try to arbitrage them. The transaction cost in most sports bets are fairly high. If you really think that there's an arbitrage there, so collect the data, and then give it a go. You may be right, in which case you'll make a fair bit of cash.

James, I'm working on that very set of issues right now. Possible dissertation topic. I'll be back to you in a few years on this. The model that I have in mind involves hidden state variables and something called a Kalman Filter.

In English, nobody can exactly pin down the fundamentals (e.g. productivity growth, monetary policy rules, whatever) at a point in time. The optimal thing to do in such a case is to underreact to new information since that information is potentially very noisy. I could certainly imagine other well-established phenomena like conditional heteroskedasticity popping out from this.

Maybe this is also a decent propagation mechanism for business cycles. I just don't know yet.

Actually, Jason, one of the biggest problems in sports gambling is that success will get you banned from the market, which is why very good wagerers tend to go into the tout business; if they simply sauntered up to the window, day after day, and placed bets that won 60 to 70 percent of the time, the sports book would probably no longer take their bets.

A few comments on the sports wagering. Most of the 'tout' numbers and TV shows are scams. For example, I know of several pay for pick phone numbers that split the country in half, giving one half of the country the favorite and the other half the dog in their 'pick of the week'. The benefit of this strategy is that at least half the people will win and likely be back to pay for a winner again the next week. An intresting additional point is that this service promised the next weeks 'pick of the week' for free if they were wrong the previous week.

I only bet football and boxing. Most years I end the season about even for football, but I did well last year. I prefer to support my local bookie rather than using the net wagering sites, but going into Dec on an 8 week win streak last year my man shut me off. Im pretty new to SC and dont know many people so I switched to a Net gambling site. I continued to do very well, but the Net service I used only paid off for free once a month. As my account went up, it became harder to get the same thrill from my typical bet, so I started bumping up my wager. Big mistake, I went cold, got smacked and the season wasn't nearly as profitable as it could have been. With a neighborhood bookie its pay on Tues and collect on Wed every week. By having my whole gambling nut in an account that I didnt empty or pay off every week I became irrational in pursuing the thrill more than the intelligent wager. For me, boxing is were I do best, but its a different type of bet that revolves around odds rather than a point spread. In a way, a bookie is like a stockbroker, he wants as much action as possible so his juice(commission) is large. The bookie wants an exactly even amount of money wagered on each side, so all he does is pay off the winners with the losers money and collects his 10% in juice.

This lengthy post isnt an attempt to hijack the thread. Its very evident to me that I have nowhere near the investment knowledge some of you have. In fact this is the most challenging topic and commentary Ive encountered on the web. Kudos.

I'm surprized by this, with online betting, a number of different casino's (just in Vegas), and many different countries to choose from, it really seems to me that it would take a fair amount of money (you could be betting $10-20K per game if you're confident) before you'd get noticed. Even then, you can switch to proxies. I'm sure you'd find plenty of people prepared to bet on your behalf for a modest cut.

I seem to remember back in New Zealand, the government used to run the TAB (sports betting), and so the employees there wouldn't really care if you kept winning.

Chris, "the optimal thing to do is underreact" sounds like a contradiction to me. If it's optimal, it's not underreacting. Also, isn't the Kalman Filter only really applicable to linear systems (with normal everything)? Good luck with the dissertation

Oh, just FYI, I remember over a decade ago, before international betting and the like were popular, it used to be common that you could arbitrage bookies in different countries. One time in particular I remember when the English rugby team was touring NZ, and the favourites in NZ were (of course) the All Blacks, while they were the underdogs in England. Needless to say, a group of us math students with family in England arranged for simultaneous bets imediately upon noticing this :)

"High-frequency traders make money on average regardless of the direction of the market....You have no data on their profits, since they are limited to investment banks and the larger hedge funds, who don't disclose their gains. "

This seems to me to be a rather significant non sequitur. How do you know if high frequency traders as a group consistently make money if the public can't?

I admit I'm unfamiliar with the strategies pursued by high frequency hedge funds. It seems to involve a fair amount of statistical modelling, which suggests [to me at least] a germ of technical-oriented prediction...am I incorrect?

Many of the funds use varieties of "stat arb," or statistical arbitrage, I think. This is heavy on modelling.

It is also commonly accepted that day traders as a whole lose money (quite a lot of money), largely through transaction costs.

Has anyone mentioned this article on the tulip bubble? (Via Craig Newmark).

Jane, the S&P 500 is at the same level as in 1998. Since dividends have been minimal, those who sold in 1998 and went to fixed-interest securities have done better than those who held on through the crash. In hindsight the optimal time to re-enter was end-2002, but things were more or less back to normal by the end of 2001, so shifting back to a standard mix of bonds and shares made sense any time from then on.

As you say, governments aren't immune from the same kinds of errors. It's an empirical matter which will do better or worse in particular circumstances.

Jane: You can make profits trading on the variances. Andrew Lo and his mates have repeated proved this, statistically and practically (Lo is a major adviser to Vanguard on a number of issues around this subject).

James: We're talking about two different phenomena. The Shleifer & Vishny work looks at reaction to news (usually earnings news). The Lo et al work only uses pure price data.

The "short term over-reaction/long term underreaction" literature began with the empirical regularity that if you ranked stocks by their performance over a short period (say a week), then over the next period, the "losers" portfolio would outperform the "winners" to a statistically significant degree. However if you carried out the same study using longer time periods you would get the opposite effect; "winners" outperformed "losers" if you looked at performance year to year.

It used to be thought that this was because in the first experiment the "losers" had all experienced negative newsflow during the first period, and the market had "over-reacted" to this, while in the second experiment you were picking up something like Shleifer's under-reaction, plus persistency of earnings momentum.

Lo and Mackinlay's seminal 1990 paper, "When Are Contrarian Profits Due To Stock Market Overreaction?", however, suggested that there is no need to presume that the "losers effect" is due to overreaction. It's also possible (Lo would argue, likely) that the "losers" portfolio consists of a set of stocks which have not yet reacted to positive newsflow that has been incorporated into other similar stocks. In particular, large stocks might "lead" small stocks, which would account for numerous other anomalies which seemed to be caught up with the contrarian empirical phenomenon. Lo believes that contrarian profits effectively represent the collection of a payment for providing liquidity to the market, although (the last time I looked) he did not agree that the nature of this payment for service could sensibly be described as a risk premium.

A Kalman filter would indeed pick up this sort of lead-lag relationship, although you would need to be running it over a pretty large dataset, and you would also need (I continue to emphasise) a pretty strong logistic operation to actually execute your trades.

Bernard: there were a myriad things that needed to be done to run the RDS arbitrage, but most of them fell into the categories of "making sure that the price quoted on the screen is a price you can actually deal at" and "making sure that a large order in one of the two stocks does not come through and knock the prices even further out of whack".

Here's the paper, found it!

http://overreaction.behaviouralfinance.net/LoMa90.pdf

http://overreaction.behaviouralfinance.net/NaJe95.pdf

is a paper which disagrees with Lo and finds genuine over-reaction which cannot be explained by lead-lag relationships, btw.

Note, though, that all of these papers, like more or less every good paper in empirical finance of the last fifteen years which has addressed the subject at all, starts from the fact that contrarian investment strategies over short-term equity returns are, at least in principle, profitable. The claim that there are no players in the market able to realise these returns is a pure act of faith, and I reiterate that, as far as I can see, the existence of Vanguard proves that at least one big player is able to.

I have no doubt that it's in principle possible to make profits using contrarian strategies to exploit the kind of non-random trends Lo documents in "A Non-Random Walk Down Wall Street." But I have to confess I'm not sure what broader conclusions we can derive from this fact, other than that it refutes the EMH. It doesn't tell us anything about whether mechanisms other than markets can do a better job of allocating capital or predicting the future performance of companies -- since Lo isn't paying any attention to fundamentals at all -- and the potential superiority of these strategies seems to depend primarily on exploiting microstructural holes (like lack of liquidity) rather than better stock-picking (assuming that forecasting the future value of companies is genuinely possible, which I do).

Potential overeactions or underreactions to news and most other identifiable inefficiencies in the market don't look like eternal truths but phenomena that have yet to be well enough identified to make money that would remove them. If they do turn out to be fundamental it will surely be because they are not as irratinal as they seem.

When these phenomena are better understood the EMH, at least in its weak form, will be closer to being true, not less. In that sense the EMH is an attractor.

More problematic I would think are the inefficiencies caused by wealth weighting of knowledge. It is not obvious that wealth attracts financial knowledge of the sort that markets are supposed to be gathering accrues to wealth as accurately as its weight in the market implies.

I'm also not clear on the whole link between the actual mechanisms of feedback and its theoretical impact. For instance buying a stock tends to force its price upwards which is positive feedback. Given the existence of momentum trading outfits like AIM this is more than a theoretical issue.

I second James curiosity about what alternative systems might be more accurate than the market.

It doesn't tell us anything about whether mechanisms other than markets can do a better job of allocating capital

The "retained earnings plus bank loans" model is the most obvious alternative mechanism for allocating capital, and you would be hard pushed to suggest that it does a worse job since it is the main mechanism by which capital is allocated in the entire world, including the USA.

Jack, if the phenomena isn't identifiable using current technology then it can' t be considered bad that it isn't factored in. It seems in keeping with the "you can't make money without spending time investing in technology" hypothesis.

I agree with you, the more troubling assumption is that there is an implicit assumption that increased proclivity to invest accompanies increased certainty about the state of the market. The arguement for this is along the lines of, if you don't behave this way, you lose your money, so the biggest players will be those that do behave this way, and they'll be driving the market. It's a plausible arguement, but not a compelling one, to my mind.

The "retained earnings plus bank loans" model is the most obvious alternative mechanism for allocating capital, and you would be hard pushed to suggest that it does a worse job since it is the main mechanism by which capital is allocated in the entire world, including the USA.

I think this is a fine answer to the question as stated but is it really a substitute for the stock market? Germany and Japan got quite a long way with it but even they still needed a stock market and in both places the role of the market is getting bigger. I don't think the market did a better job in either place than it did where it played a bigger role. I'm not sure whether the famous tulip market is currently regarded as irrational but I'm guessing that it was financed by debt and retained earnings.

I think it is also necessary to consider how well the technique recovers from the mistakes it does make. Equities degrade more gracefully than banks.

The weak version says only that you can't make excess returns--basically, profits above the economic costs of finding the investment opportunities--based on historical financial data. I think it's safe to say that pretty much every economist thinks this is true, and that "chartists", who attempt to predict price movements based on past trends, are doomed to fail in the long run.

Well it has been demonstrated that you and pretty much every economist are completely wrong. For example, there was one study where the weather in New York was used to determine whether to buy or sell stocks, the theory being that gloomy weather would make traders overly pessimistic and sunny weather would do the opposite. The traders doing the study were able to make above market returns, even after transaction costs, by pursuing such a strategy, using only publicly available information (aka weather forecasts). That precise opportunity is likely gone now, but others await the imaginative investigator with access to nothing but public information.

In fact, it's why Chicago tends to preach index funds as the only sane investment for 99% of investors

The greater the number of people who believe such a thing to be true, the less true it will be. If everyone believes there are no arbitrage opportunities (and acts rationally on that belief), arbitrage opportunities will certainly exist. If everyone else bought only index funds, buying index funds would be a dumb investment.

Of course, you make the same point, "value investing (like efficient markets) only works as long as a lot of people don't think it works". The same is true for technical analysis.

It tends to work best in thinly traded areas of the market; you are vanishingly unlikely to make money (as Uncle Warren did), by buying Coke because you think it's undervalued.

Was Coke thinly traded when Buffet bought in? No.

Others:

One hears a lot these days about how obvious the bubble was and how only fools got taken in, but these same talkers don't seem to have cashed in. I think there's a lot of hindsight in these comments.

The time periods over which you could short a stock did not allow those who knew the bubble was just that to cash in - which was perhaps one of the things that allowed the bubble to occur. People knew the valuations were insane, but asking them to put an exact date on when the insanity would cease is a bit much.

The rational thing to do was to get out of the market completely, which many people did. Remember that Buffet at this point was regarded as a senile old man who didn't understand the "new economy".

Wouldn't there be an argument that Buffett's type of value investing is more like a private equity play, though? Take damaged assets, take a more aggressive role in management than a typical passive investor, fix 'em up, and more on.

Buffet is the most passive of investors. He does not involve himself with day to day management.

"Oh yeah -- dont sell. Thats another mistake. I know people who sell off thier options to pay for cars and crap and regret it later -- we are in a major tech boom its a little early to take the money off the table IMO. ... untill then ride the hype for all its worth."

The textbook example of the "greater fool" theory in action - and another argument for the inefficiency of markets.

Dsquared, all of your points seem to amount to arguing that prices vary from intrinsic value over the short or medium terms, which as I've said several times, was a point hammered into me by the very advocates of efficiency you are decrying. What they said was that you couldn't make economic profits trading on the variance, not that there was no variance.

And, as my first example shows, the very advocates of efficiency are provably wrong. To cling to a theory once it has been shown to be false is the definition of religion. What's most interesting is that your religion can not possibly be true if everyone believes it.

A quibble with your first point felixrayman.

For the definition of the weak EMH given above it is only required that in expectation or on average all such strategies won't make excess returns, not that it won't happen in particular instances.

Indeed for any given strategy, either it or its opposite will most likely make money. For example if they decided that traders would only pay attention when the weather was bad and would be distracted when it was good and taken the opposite strategy, they would have lost.

On its own it is just a coin toss that came up heads.

Also, the idea that Buffett is the most passive of investors is completely wrong. He was the key figure, for instance, in getting Douglas Daft fired as CEO of Coke, he was instrumental in the restructuring of Salomon after the bond-buying scandals in the late 1980s, to name just a couple of examples. It's also undoubtedly the case that at this point news that Buffett has invested in a stock makes other investors more interested in and even more likely to invest in that stock, so that the returns he enjoys are not purely the result of good stockpicking. Which isn't to denigrate his stockpicking -- I have little doubt that Buffett is one of the few investors who can consistently outperform the market. The problem is that there are almost no Warren Buffetts, and in any case it's impossible to identify them in advance.

It would be fascinating to have a video, made by hidden camera, of Buffett in the early days, when he was making calls to Midwestern businessmen, lawyers, and doctors, in search of capital. Was there anything discernable in his presentation that revealed what was to come, or was his pitch as banal as any other stock-picker?

Dsquared wrote "We are looking at a ten year competition which produced winners, then a second ten year sample which produced the same winners. That would be laughably statistically improbable..."

I don't have a dog in this fight, but is this true?

Lets say there are 16 million coin tossers/stock pickers -- the number in the post above -- i have no idea whether this number is accurate -- if the chance of beating the market in any given year is 1/2 then the chance of beating the market 10 years straight is at least (1/2)^10=1/1024; which means the expected number of winners is more than 16 thousand. out of these 16 thousand, we should expect more than 16 will outperform the market over the second 10-year period; hence, lynch and buffet.

Follow up on my previous post: yes, the probability of the same exact winners over two independent 10 year period is small; however, the probability that the set of winners over two independent 10 year periods have a non-trivial intersection is not small; the expected size of the intersection is bounded below by the expected number of traders who won each year over a 20 year period. so whatever way you look at it, soros and buffet should ultimately be explained by whether (# of traders)*(1/2)^20 is bigger than two or not.

But, empirically, the popular news media etc, etc did not pick out sixteen thousand people and call them geniuses in 1992/3. It picked Buffet, Soros and a very few others. I maintain that the question is very much more like the probability of a specific individual winning a ten year coin toss tournament, which would be very small indeed.

Dsquared wrote "But, empirically, the popular news media etc, etc did not pick out sixteen thousand people and call them geniuses in 1992/3.It picked Buffet, Soros and a very few others."

I see -- true, of course. Out of curiousity, who else besides Buffet and Soros?

To James Surowiecki:

I have no doubt that it's in principle possible to make profits using contrarian strategies to exploit the kind of non-random trends Lo documents in "A Non-Random Walk Down Wall Street."

Forgive me for not having read "A Non-Random Walk Down Wall Street." Perhaps this question is answered there, but, do you expect true long term profits or just normal rates of return on the investment made in finding and implementing the contrarian strategies (taking the "economics of superstars" observations into account)?

To me the EMH means that one cannot objectively expect (in as statistical a sense as is possible in this context) to make a greater than normal rate of return on investing in the stock market without having "inside information," interpreted very broadly to include things like special insights into the market, knowledge of new products that are not well known, special talent at identifying unusually competent on incompetent management teams, etc.

And, to generalize one step further, the expected return on attempts to gather this sort of "inside information" will, on average, yield a normal rate of return.

None of this rules out the existence of a Vanguard or a Warren Buffet.

Whether or not the EMH is right or wrong is entirely dependent upon the purpose to which it is put. It is "wrong" for activities that are trying to generate this sort of inside information (although only a fool would ignore it's wisdom). It is "right" for activities that aren't.

Comments are Closed.