Bookblogging: The rise of the efficient markets hypothesis

by John Quiggin on July 22, 2009

A bit more from my book-in-progress. I’m currently toying with the title Zombie Economics: Seven Economic Ideas that Aren’t Dead but Should Be. As always, I’m keen to get suggestions on this, and on improvements to the text. I’m particularly happy to have putative errors pointed out. If I agree with you about the error that saves me from putting it in print. If not, it will be a point I need to anticipate and respond to.

The rise of the EMH began relatively modestly with the argument that the prices of assets such as stocks cannot be predicted from their past movements in they way claimed by “chartists” and “technical analysts”. In the popular terminology, prices follow a ‘random walk’. This idea had been put forward as early as 1900 in a neglected paper by a French statistician, Louis Bachelier, but it was not rediscovered until the 1950s.

The simple idea behind the random walk hypothesis was that, since everyone in the market could see the history of prices, any predictable pattern would soon be exploited and the very process of trying to exploit it would eliminate the pattern. The random walk hypothesis went against the powerful human tendency to find patterns in data, whether they exist or not. But it stood up well to initial statistical testing, and has done so ever since.

None of the patterns typically analysed by students of stock market charts, such as trends, reversals and support levels, appear to be of any use in predicting stock price movements. There remains some dispute about whether subtler features of the behavior of stock prices are consistent with the possibility of a profitable trading strategy based solely on observation of past prices.

A handful of anomalies such as the ‘weekend effect’ (prices tending to fall on Fridays and rise on Mondays) have been observed. However, the effects are usually too small to permit traders to gain significant profits after trading costs are taken into account. And most disappeared not long after they were discovered.

Two explanations of the disappearance of anomalies, both consistent with EMH might be considered. First, it may be that once the anomaly was publicised traders sought to exploit it (for example by selling on Thursday and buying at the Friday close). But such a pattern would produce a price increase on Fridays, and a decline on Mondays, wiping out the anomaly it sought to exploit.

The second, simpler, explanation is that the anomalies were just the product of human pattern-finding. The famous curse of the zero years provides an illustration. Beginning with the election of William Harrison in 1840, who died of pneumonia in 1841, all Presidents elected in zero years, up to and incluidng JFK died in office. It was after the Kennedy assassination that the curse was apparently discovered. But the next potential victim, Ronald Reagan, elected in 1980, survived for two terms and lived to the age of 93. George W. Bush also survived two terms, and the curse is now forgotten.

At a more sophisticated level, Andrew Lo, Director of MIT’s Laboratory for Financial Engineering has argued that because of investor irrationality, asset prices display some momentum over time. But this claim remains controversial, as does the performance of algorithmic trading strategies designed to exploit such patterns.

Among economists, the random walk hypothesis, now referred to as the ‘weak form’ of the EMH, is fairly generally accepted, and even the sceptics agree that any violations of weak-form EMH are subtle and hard to exploit. In a striking instance of the inefficiency of financial markets, however, investment banks continue to employ “technical analysts” using charting methods, decades after such methods have been shown not to work. The human desire to believe that there must be a way to beat the odds is reflected in the prevalence on the Internet of “systems” guaranteed to make you a winner betting on the horses or at the roulette table.

The success of the random walk hypothesis showed that the existence of predictable price patterns in markets with rational and well-informed traders was logically self-contradictory. It wasn’t long before economic theorists realised that the same point applied to other kinds of information, such as information about the likely future earnings of companies. If this information is publicly available, then traders should take it into account, just as they do with the past history of the stock price. So, the stock price will be the best available estimate of the future value of the stock, taking account of all

The key steps in the discovery of the strong EMH were taken independently by Paul Samuelson, the leading Keynesian economic theorist of the postwar era, and Eugene Fama, who soon became a leading figure in the free-market Chicago school, and is widely regarded as the father of modern finance. As we will see, they took the idea in rather different directions.

There was one more subtle distinction to make before the EMH assumed its modern form. The arguments so far concerned publicly available information, but what about information that was only available to some people, such as company insiders, or customers? Some theorists argued that such information would inevitably be reflected in market trades. Others stuck with the traditional focus on publicly available information.

Fama proposed a distinction between the weak form (EMH), which excluded profitable trading based on price history, the semi-strong form, which extended the claim to cover publicly available information, and the strong form, which claimed that the stock price incorporates all information held by traders, whether it is public or private.1

Although the EMH made a big difference to the way economists viewed financial markets, it had much less impact on financial markets themselves. An entertaining and economically literate description of the stock market scene in the 1960s, The Money Game by ‘Adam Smith’ (a pseudonym for George Goodman) describes ‘a random walk professor choking on his icecream at the thought that there are people called “technicians” who claim to forecast the stock market”, but makes it clear that the vast majority of Wall Streeters believed the technicians more than the economists.

The economic theory that really changed thinking in financial markets was the model of pricing options1 developed in a 1973 paper by Fischer Black and Myron Scholes and subsequently formalised by Robert Merton. The model was named Black-Scholes, but Merton got his share of the glory when he shared the 1997 Nobel Memorial Prize in Economics with Scholes (Black had died two years earlier).

The Black Scholes model showed that, under plausible assumptions, it was possible to duplicate the payoff from an option2 by a combination of trades in the original stock and in high-grade bonds. Hence, the ‘right’ option price could be calculated by looking at the interest rate on bonds and the variability of the stock price. If the market price differed from the Black-Scholes price, traders could make money with little or no risk by combining trades in the two markets.

It took some time for financial traders to come to grips with the Black-Scholes model, and, while they did, sophisticated “quants” or “rocket scientists” who understood the model made big profits at the expense of old-fashioned traders working on rules of thumb and “seat of the pants” judgement. Eventually, the quants came to dominate the market and prices came more and more into line with the Black Scholes rule. The quants went on to design more and more exotic derivatives on which to practise their skills, and the role of finance theory was established, seemingly on a firm foundation of success.

There was something of a paradox here. The Black-Scholes pricing rule shows how an option price ought to be determined in an efficient market. But traders can only make a profit using Black-Scholes and similar rules to price derivatives if the market price deviates from the ‘correct’ price, that is, if the efficient markets hypothesis is not satisfied. This paradox was given a rigorous formulation in a famous 1980 article by Sanford Grossman and Joseph Stiglitz, one of the contributions that later earned Stiglitz the Nobel Prize in economics

Economists have wrestled with the Grossman-Stiglitz paradox for a long time without working out a completely satisfactory solution. The most common view was one that seemed to preserve the efficient markets hypothesis while justifying the huge returns reaped by financial market professionals. This is the idea that the market is just close enough to perfect efficiency that the returns available from exploiting any inefficiency are equal to the cost of the skill and effort that goes into discovering it.

1. Since the weak form of the EMH is relatively uncontroversial and mostly unimportant, I will use the term EMH to refer to the strong and semi-strong versions from now on. Where the distinction between the two is important, I will try to make it clear which one I mean.

2. An option is one of the simplest kinds of financial derivatives, that is, assets derived from other assets. An option gives you the right to buy (or sell) a given stock at a given price and on a given date

{ 55 comments }

1

Donald A. Coffin 07.22.09 at 9:29 pm

Let me begin by saying that I do not do the economics of finance (I’m a labor/urban economist).

However, I have thought that one way of looking at the (strong) EMH is to ask about sudden changed in asset prices.

The EMH, in its strong version, essentially argues that prices (especially of financial assets, but there’s no reason why one would restrict its applicability to such assets) incorporate all relevant information and are, in principle, correct.

I would suggest the following mental experiment. Let’s leave aside long secular movements in the prices of things and look instead at abrupt changes in prices. The EMH says that something happened to the information environment that makes the initial prices and the final prices both correct. So, for example, consider these changes:

Between October 22 and October 29, 1929, the Dow Jones Industrial Average fell by 36%.
Between October 16 and October 20, 1987, the DJIA fell by 26% and the S&P 500 fell by 24%.
Between January 19 and January 25, 1998, the Indonesian Rupiah fell by 41% relative to the US dollar.

That’s three large and abrupt changes in the price of financial assets.

Now, I don’t think it’s reasonable to expect the EMH to predict these abrupt changes. But it would be nice if it could explain them. But what do we want an explanation to do? Three things, I think. First, to identify the change that occurred in the information environment. Second, to explain why that change led to the observed outcome. And third, to validate that explanation by identifying other, similar changes in the information environment leading to similar changes in asset values. And, by “similar changes in the information environment,” I mean a fairly close similarity. Note that my emphasis is on explaining changes in asset prices.; I think that emphasis is consistent with what we should expect a theory (like the EMH) to do.

In talking with my colleagues in finance, I get the impression that no one has proposed explanations of events like these that would fulfill what I take to be a minimum standard by which to evaluate the EMH. I could of course, be extremely wrong about all this…

2

Ceri B. 07.22.09 at 9:46 pm

#1. I really like this take on the title. It’s much more engaging to me, and does not leave me feeling I must have misunderstood or am missing the joke.

#2. A conceptual nitpick of the purest nit sort: The human desire to believe that there must be a way to beat the odds is reflected in the prevalence on the Internet of “systems” guaranteed to make you a winner betting on the horses or at the roulette table. Strike “on the Internet” from that sentence. These things are, to the best of my knowledge, no more common online than off, after all.

#3. A request for data: How much do banks and others spend on technical analysts doing the refuted stuff? Dollars, people, any sort of brief citation in there would be welcome.

#4. The stylist in me thinks the flow is bad here: The model was named Black-Scholes, but Merton got his share of the glory when he shared the 1997 Nobel Memorial Prize in Economics with Scholes (Black had died two years earlier). Something like “Black died in 1995; in 1997, Scholes shared the prize with Merton” might do the job.

#5. I like Donald Coffin’s comment@1 about what we’d like an EMH to do.

3

Substance McGravitas 07.22.09 at 10:08 pm

After “taking account of all” there’s some text missing.

4

Phillip Hallam-Baker 07.23.09 at 12:06 am

The EMH is one of those claims that may have some utility in certain situations as a modeling assumption but has subsequently been beatified as axiomatic and then canonized as immutable truth.

It is not just that some people believe in the EMH, it is the way they react to people who challenge this assumption. To doubt the EMH is to expose yourself as a complete fool whose uninformed ideas are unfit for consideration on any topic. Thus it is not necessary to answer those who question the EMH as they clearly have no standing.

A similar effect is seen in the treatment of rational choice – again a perfectly good simplifying assumption that I apply myself from time to time in attempts to predict the targets selected by cyber-criminals. The population is question is clearly not rational, most of their profits go straight up their nose. But rat choice works acceptably well.

But some economists seem to claim that rat choice is an immutable law of sociology.

Another example is Pareto optimal. Again a simplification that helps get a result out. If something is Pareto optimal then everyone should be willing to consider a change (but not necessarily on what the change should be). But some claim that Pareto optimal is the same as optimal. So higher taxes can never be ‘optimal’.

And somehow each of these ideological mis-interpretations curiously result in justifications for whack-job conservative policies. Who would have thought?

5

Tim Silverman 07.23.09 at 12:55 am

Just a clarification …. People don’t really use Black-Scholes to price options (at least not where I come from). They use Black-Scholes (or, more usually, other, more sophisticated and realistic models of the price process) to express prices as volatilities, which have the advantage over prices of being more stable and easier to understand and model; also, if you’ve got a good model of the price process of the underlying, you can separate the level of the underlying price from its volatility (whatever that means in the context of the model) and treat them as independent. If you’re an options trader, you typically hedge away the part of your risk due to the underlying (i.e. the delta) and trade the volatility. At that point it’s like trading any other risky variable. If the market is illiquid (or semi-liquid) you can make money by knowing more than other market participants, e.g. by being the only one who has the faintest idea what they’re doing. If the market is liquid, you make money by market-making, or, if you’ve got cheap, fast access to enough markets, by arb trading. If the market is very very liquid, you eke out a subsistence on the crumbs of tiny spreads on a huge volume. With the number of assets out there, there is actually quite a lot of illiquidity and some weird prices. Of course illiquidity is quite unhelpful if you suddenly need to flatten your position for some reason…

You can also, I think, effectively add premiums for providing frills—funny structuring on payoffs, timing, etc—that other people don’t offer.

(I’m just a casual observer—maybe an actual expert could comment on where the money comes from. Different markets are very different.)

6

Neel Krishnaswami 07.23.09 at 1:12 am

#5. I like Donald Coffin’s comment@1 about what we’d like an EMH to do.

I think you’re asking for too much — the weak EMH (the version with empirical support) can’t do that. There are several different accounts of why bubbles are possible, all of which are compatible with weak EMH, and no really compelling way to distinguish between them.

Off the top of my head, there’s the DeLong-Shleifer-Summers-Waldmann account of noise traders, and how they can persist in a market and distort prices. Then, there’s the rational bubble story, which starts with the Grossman-Stiglitz account for the imperfections that can start bubbles, and then observes that rational traders can inflate bubbles as long as they believe they can find a greater fool. Third, bubbles in the financial markets can be created by asset bubbles in other, less-efficient/liquid markets, such as real estate bubbles, inflating the stock market. (Here, the problem is the lack of arbitrage opportunities due to the incompleteness of markets).

We can’t even test these alternatives, because of the McCloskeyan point that it’s possible for all of these predicted effects to happen sometimes, and the actual problem is identifying their relative size and frequencies. This means that you need complicated models which can accommodate all of these possibilities, which cash out as models with lots of parameters. In turn, such models are hard to decisively test: you can end up getting stuck doing parameter-fitting because you need a lot more data than you’ve actually got to statistically reject the theory.

7

Ali 07.23.09 at 1:23 am

John,
This bit seems a bit ill-put:

The success of the random walk hypothesis showed that the existence of predictable price patterns in markets with rational and well-informed traders was logically self-contradictory.

You’ve just spend a few paragraphs talking about the argument over and the empirical verification of the weak EMH. But then you say this shows it’s “logically self-contradictory”? How can evidence show that an idea is logically false? And if it’s logically self-contradictory, why would you bother recapping the evidence against it?

You need to rephrase this. I think what you want to say is that there are “arbitrage” arguments proving that certain things are logically self-contradictory (or more aptly, self-eliminating), and that the evidence for the random-walk hypothesis shows that these type of arguments can apply to the real markets.

8

andthenyoufall 07.23.09 at 1:41 am

There is some poetic justice to the fact that the mindless “zombie” meme just won’t die.

But if you insist on an undead theme, maybe “Cursed Creators: Seven Dead Ideas that Dominate (or Terrorize, Menace) Economics.”

9

mcd 07.23.09 at 3:18 am

Let me weigh in late on the title.

Why be nice? Rightwingers aren’t nice. Grab people by the throat. Call it

The Seven Big Lies of Economics

Or Economics: The Seven Big Lies if you prefer.

10

The Raven 07.23.09 at 4:11 am

But, see Mandelbrot on prices.

11

Evil Bender 07.23.09 at 5:10 am

I can’t add insight into the economics questions, but I must say I love this variation on the title.

12

Kenny Easwaran 07.23.09 at 5:13 am

I wanted to point out the missing text at the end of the 10th paragraph that Substance McGravitas mentioned in #3.

I was also wondering, like Ali at #7, how evidence could show that something was logically contradictory. Did the evidence lead them to notice something that they were then able to prove as a theorem? Or was the logical impossibility noticed first and then the evidence convinced people who were skeptical of the axioms involved? Or did the evidence motivate a later theoretical development that allowed people to prove the proposed theorem?

Finally, you don’t really say what the Grossman-Stiglitz paradox is. Is it the idea that in an efficient market, no one can make money? I’ve heard reference to a result that seems to show that in an efficient market with no information asymmetries, no one can buy or sell anything, because the potential buyer and seller both have all the relevant information, and therefore value the good at the same price, so neither sees herself as profiting off the other. Is that what this is?

13

Katherine 07.23.09 at 8:17 am

How about Zombie Economics: Seven Economic Ideas that Just Won’t Die ?

Either way, I think the zombie imagery is good – evocative of long dead ideas shambling around trying to eat people’s brains.

14

Katherine 07.23.09 at 8:29 am

Fama proposed a distinction between the weak form (EMH), which excluded profitable trading based on price history, the semi-strong form, which extended the claim to cover publicly available information, and the strong form, which claimed that the stock price incorporates all information held by traders, whether it is public or private.1

Hmm, although I feel that you’ve explained pretty well the weak form (the random walk hypothesis right? Although I cannot for the life of me work out why that is an appropriate description – perhaps I’ve missed something), I’m not so sure that the semi-strong form has been adequately described to disinguish it.

Also, I’m not sure why the “weekend effect” can be called an anomaly, when it seems to be a perfect example – i.e a predictable pattern that would be/was exploited, thereby eliminating the pattern.

15

Katherine 07.23.09 at 8:33 am

Also, if you are going to explain what an “option” means, you might also want to explain what a “high-grade bond” means., if you are explaining things at that level. I personally am much more familiar with the meaning of the former than the latter, rather than the other way around.

16

Katherine 07.23.09 at 8:42 am

Me again:

At a more sophisticated level, Andrew Lo, Director of MIT’s Laboratory for Financial Engineering has argued that because of investor irrationality, asset prices display some momentum over time. But this claim remains controversial, as does the performance of algorithmic trading strategies designed to exploit such patterns.

I find this a bit confusing. When you refer to asset prices displaying “momentum over time”, do you mean bubbles? If not, what do you mean? And if you do, is that controversial? Or perhaps you mean that asset prices gaining momentum means that it takes them a bit of time to (re)adjust to new information, and there are those who claim they can exploit that by getting in there before that readjustment?

Apologies if I’m being a dunce.

17

James Wimberley 07.23.09 at 9:02 am

Isn’t the Grossman-Stiglitz paradox a general feature of capitalism? All capitalist investors strive to make real profits, not a mere living at the competitive market equilibrium; their behaviour requires that markets be inefficient. If we ever had efficient competitive markets, capitalism would die. We could still have stock exchanges, run by computer algorithms and maintained by 9-to-4 bureaucrats earning $100,000.

Not really a suggestion for the title, but you could try to work in a joke about “Alfred Nobel, Baron Samedi of Chicago”.

18

Tracy W 07.23.09 at 9:24 am

A good summary of the history of Fama’s EMH, well at least the bits I know about independently. (I am aware of the possibility that praise from me will worry you deeply. :) )

Out of curiousity, can you name those theorists who believed that private information would inevitably be reflected in trades?

19

marcel 07.23.09 at 10:40 am

Following up on mcd (@ 9), and in an Al Franken vein:

The 7 Big Fat Lies of Economics, and the Lying Economists who Tell Them.

20

dsquared 07.23.09 at 11:52 am

1. Like the new title

2. I still believe that the evidence on chartism is much less cut and dried than that (in particular, head-and-shoulders patterns have passed some reasonably rigorous tests) but I doubt it’s worth adding anything more on a digression.

3. Not sure if it would be a digression, but shurely the real sting of Samuelson’s paper on iterated expectations and random fluctuations is that the value of the stock has to anticipate not only future values of the stock, but also the actual NPV of the dividends? This is important because a) it motivates Shiller’s Paradox which you might want to write about and b) ties the EMH in Fama’s sense to the economically important sense in which the stock market is (or isn’t) providing real information about actual transactions.

21

arthur 07.23.09 at 12:22 pm

Since Tracy W doesn’t want to argue, I will have to step into the breach.

From what I know, Grossman-Stiglitz arises due to the conflict between 1) having an efficient market and 2) the fact that acquiring info to trade on inefficiencies are costly. This is quite different from the inconsistency of profiting from market efficiencies while trading with a model assuming market efficiency (which I suspect is what you are writing). The conclusion in the final paragraph is okay though.

22

Johannes 07.23.09 at 12:48 pm

Hi!

I think your description of the efficient market hypothesis is way too simple.
There are so many more information-bits about it, that make it (not) work.

You might want to read this book:
“Inefficient Markets: An Introduction to Behavioral Finance”

It’s quite academic, but any one page leaves you with more insight than all those other shitty books about beating the markets etc.

23

Odm 07.23.09 at 12:57 pm

@17: According to my economics teacher, unlike in finance, the expected return of the owner is considered a cost in economics. Under perfect competition the producer will be earning just enough profit to convince the owner that he wouldn’t be better off owning a bakery.

24

Odm 07.23.09 at 12:58 pm

(My silly example assumes the owner is not already a baker.)

25

David Speyer 07.23.09 at 1:26 pm

Complete layman here, but one who thinks he is pretty smart. :)

There is something I’m not getting in all of these discussions of the EMH. As you explain it, the various forms of the EMH say that we cannot predict stock prices, or at least cannot do so without investing significant resources in learning nonpublic information.

Why do people say this means prices are correct? I can’t predict whether it will rain in Kansas tomorrow, but that doesn’t mean Kansas is getting the right amount of rain in any sense I can think of.

26

SamChevre 07.23.09 at 1:28 pm

I like the proposed title much better.

And I think it’s Donald Coffin’s question that is key. I’ll say “It’s the wrong question,” but I expect that your answer is different. Saying “is on average the best available model” does not mean “will explain everything.” Betting against the EMH is like drawing to an inside straight; sometimes you win, but on average you lose.

27

Kevin Donoghue 07.23.09 at 1:54 pm

David Speyer,

Quite a few economists have made the same point. James Tobin was one of the most eloquent. Recently Krugman directed us to Summers on ketchup economics.

I think it’s a point which John Quiggin ought to take up. The EMH wouldn’t sound quite the same if it was called the Unpredictable Price Hypothesis or something like that. There’s a touch of Unspeak in using the term “efficient” to mean “behaves like a well-oiled roulette-wheel.”

28

Phillip Hallam-Baker 07.23.09 at 1:55 pm

There is another model that should be considered as at least plausible: that a financial market is rigged by various colluding interests who make their money from the people who don’t know about the conspiracy.

Of course in the penny stocks world we know that the markets are frequently rigged, if a penny stock moves without news the cause is more likely to be pump and dump than anything else. But the history of the mortgage derivatives markets in recent years shows that prices were set by the short term interests of the human participants in the market and not the putative buyers and sellers.

I think that this is a major shortcoming in this discussion, the trader and the employer are treated as a single entity with identical interests, they are not. Nor are the interests of the broker or financial adviser the same as the client.

In practice, as we all know, the trader is going to maximize their personal trading profits over the quarter. And that has really only a casual relationship to the real value of the underlying assets.

When I was part of an S&P500 company it was hard enough to work out what the results were for a given quarter according to GAAP. How well we had really done that quarter was likely rather different. And forecasts of what the results would be in the next quarter, or the next year or the year after that pretty soon dissolved into question marks.

In short, the people running a public company with all the information on the company results do not have a precise understanding of the value of their company. The idea that the markets could arrive at a better view is pure fantasy.

Another flaw in the EHM is the idea that markets even give prices. If you look at yahoo finance you will see a market cap for Microsoft arrived at by multiplying the price of the last trade by the number of shares outstanding. But the real market value of Microsoft is the amount of money it would take someone to buy control of the entire company and the market cap figure is only a lower bound on the real value.

And even if a company is acquired, the acquisition price only tells you the value that it had before the acquisition.

29

stefan 07.23.09 at 1:59 pm

The Black Scholes model showed that, under plausible assumptions, it was possible to duplicate the payoff from an option.

I’m not so sure about the ‘plausible assumptions’ here. Since when is costless continuous time trading plausible? This isn’t a minor approximation since the expected trading volume in the model is infinite. Addressing approximate option replication with trading costs is a hard problem, as is how to trade deviations from BS pricing. This is not to argue that BS pricing is useless, simply that the assumptions aren’t plausible.

30

P O'Neill 07.23.09 at 2:33 pm

I believe that there’s a sentence in one of the relatively early Fama articles on EMH along the lines of “if one judges the fertility of a field by the quantity of manure dumped on it, the EMH is indeed an extremely fertile field”. Might be ripe for an update.

31

Delicious Pundit 07.23.09 at 3:39 pm

In The Long Run We Are Undead: Seven Economic Fallacies That Won’t Be Killed.

32

James Surowiecki 07.23.09 at 3:48 pm

I think one of the key papers to debunk in this chapter would be Robert Shiller and Jeeman Jung’s “One Simple Test of Samuelson’s Dictum for the Stock Market”: http://ideas.repec.org/p/ysm/somwrk/ysm315.html, where they argue that there’s evidence that since the 1920s, the prices of individual stocks have been reasonably efficient, not in the simple sense that they’ve been unpredictable, but in the sense that they reflect the future free cash flow of the underlying companies. They see this as supporting Samuelson’s argument about the stock market, which is that it’s macro-inefficient, but micro-efficient.

33

Odm 07.23.09 at 3:59 pm

@29: Clever title, but a little pessimistic, no? I’d like to think that we’ll eventually put these fallacies to rest…

34

Billikin 07.23.09 at 4:02 pm

It is really not too difficult to explain why the weak EMH holds and yet hedge funds can make enormous profits. During the 1980s in one or two hours a day I used to find many market inefficiencies to exploit. Almost all of them were too small to make money at because of commissions. Others, while theoretically profitable, were too ephemeral or entailed the risk of legging into a hedge, where you might lose money on one trade of the hedge before completing the other. Still, I could find a profitable trade every week or two.

Hedge funds use computers to find market inefficiencies in seconds or fractions of a second and to execute trades before they evaporate. They also pay low commissions. In addition, they can take large positions and use leverage.

35

Daniel 07.23.09 at 4:18 pm

You have the weekend effect backwards. Prices tend to rise on Fridays and fall on Mondays.

36

Delicious Pundit 07.23.09 at 4:22 pm

Zombies With Charts: How Economic Fallacies Eat People’s Brains

37

James Wimberley 07.23.09 at 5:19 pm

Odm in #22, 23: True, but you illustrate my point. A world in which capitalists only just think it worthwhile to show up at the office isn’t one in which you have bubbles and crashes. It also isn’t one in which they take risks on new products and technologies. You don’t have to be a Randian to think that it’s animal spirits – the hope of making a pile – that drives real entrepreneurs and traders. Their collective interest is to keep markets opaque and uncompetitive, and they are pretty good at ensuring it: see Adam Smith.

38

James Wimberley 07.23.09 at 5:30 pm

Possible chapter headings:
Vampirical tests of EMH.
Stake-holders and drivers.
Teenage were-traders in London.

39

Nancy Kirsch 07.23.09 at 6:43 pm

I could think of more than seven.

40

dsquared 07.23.09 at 7:30 pm

in the sense that they reflect the future free cash flow of the underlying companies.

in a very weak sense indeed; what the paper manages to show is that you can reject the null, at the 5% but not the 1% significance level, that future dividends (which is not the same thing as free cash flow) are totally irrelevant to equity valuations. I think I will term this the “Homeopathic Form” of the efficient markets hypothesis.

41

James Surowiecki 07.23.09 at 8:03 pm

Right, Shiller and Jung look at dividends rather than FCF — I hadn’t read the paper in a while, and I must have been projecting, since that’s what they should have been looking at. In fact, this is a strange lacuna in the argument for the EMH — apparently, no economist has actually bothered to study if there’s any correlation over time between stock prices and future FCF, even though from a fundamentals point of view that’s what you’d expect to see if stock prices were, in any sense, “right” (setting aside, if possible, the problem of shifting discount rates). I realize John and Daniel think this would be the very definition of a fool’s errand, but there are plenty of fools in academia, and it seems surprising — even given the fact that it would be a monumental task in terms of aggregating the data — no one’s done it.

42

Paul 07.23.09 at 8:40 pm

Personally, and I have to be candid, some economists come across as zombies to the average man or woman. I cringe when I hear the word “expert” any more. Being an expert does not mean that one necessarily will make rational and feasible decisions.

43

sg 07.23.09 at 10:18 pm

John, do you think it’s okay that your title sounds a bit too close to Lies, Damned Lies and Health Care Zombies: Discredited Ideas that will not die

I know it’s not exactly the same or anything, but it seems like a broadly similar pattern.

And isn’t the colon a bit too Northern Hemisphere for an Australian academic?

44

dsquared 07.23.09 at 11:24 pm

#39: basically the problem is that the CRSP data files contain good, clean data on dividends but not on free cash flow and it would be a hoolie of a job to go and backfill the data on FCF – I can certainly testify that out of the half-dozen commercial services currently providing long runs of accounting data for companies, not one of them is any good at all. Also you’d have to form an opinion about the correct rates of depreciation, and depreciation is a subject which economists instinctively seem to steer clear of – if I ever do a PhD I think it will be on the subject of depreciation.

My own comments got eaten above I notice – one of them was sort of pre-empted by James:

1) I like the new title

2) I still think you’re not being fair to the chartists here but doubt it’s worthwhile including any more material on predictability as it’s a bit of a digression already.

3) surely the big sting in Samuelson’s original paper is that you can’t arbitrage by simply holding the security and collecting the dividends either? This a) motivates the Campbell/Shiller excess volatility paradox, which you probably want to discuss and b) ties the Fama version of the EMH to the more economically interesting theory about capital allocation – ie, semi-strong EMH also has to rule out “buy and hold forever” strategies which can make supernormal profits, so it is logically committed to all the strong stuff that Tracey W (for example) don’t necessarily want.

45

Tim Worstall 07.24.09 at 10:16 am

I’m obviously missing something very important here (not all that much of a surprise, eh?):

“There was something of a paradox here. The Black-Scholes pricing rule shows how an option price ought to be determined in an efficient market. But traders can only make a profit using Black-Scholes and similar rules to price derivatives if the market price deviates from the ‘correct’ price, that is, if the efficient markets hypothesis is not satisfied. This paradox was given a rigorous formulation in a famous 1980 article by Sanford Grossman and Joseph Stiglitz, one of the contributions that later earned Stiglitz the Nobel Prize in economics

Economists have wrestled with the Grossman-Stiglitz paradox for a long time without working out a completely satisfactory solution.”

Doesn’t the paradox go away if we simply insert a few “at our current state of knowledge” qualifiers in there? For example, people didn’t know that options and stocks and bonds were related in their pricing in the way that Black Scholes says they are (just to float an idea), this knowledge was then discovered by Black and Scholes (and Merton), this new knowledge was then exploited and with lags (perhaps measuring the lags by how long people were able to make money from it) became incorporated into those market prices?

We don’t think it all that surprising that there are chemical reactions we don’t know about as yet, then someone finds out about them and then they become more generally publicly known. Or sub-atomic thingies that then inform electronics. Or even fusion recipes, putting together ingredients from different parts of the world to create new dishes (most of which don’t work but some indeed do)*.

The EMH, as I understand from the above, is about “publicly available” information. As I say, I’m clearly missing something but how is it a disproof of the EMH that when someone finds, discovers, manufactures or codifies previously “not publicly available knowledge” then prices or behaviours change?

Isn’t that, in some form, a proof of it? That financial markets, over time, adjust to this new information?

Or is it the “over time” qualifier that causes the problem?

*Yes, I am quite seriously comparing the creation and dissemination of the recipe of, say, kiwifruit Pavlova with new information about patterns in financial markets. Something new is discovered, knowledge of it is rare but over time that knowledge spreads to every damn cookbook in the land/every investor.

46

Alex 07.24.09 at 10:57 am

My fundamental issue with the EMH is that the publicly-available information going into the sausage machine (aka trader) is by definition historical at the time of use. However, investment decision are by definition forward-looking. Therefore, whatever decision process takes place involves forecasting the future. It’s not like a control loop where you get real-time inputs and outputs; it’s more like one of the anti-aircraft gun sights that the first cyberneticians worked on, which works by tracking the target, computing a forecast of where it will be after a given lag time, offseting the gun there, and then re-checking.

Alternatively, if you use slightly different definitions, a lot of the publicly-available information itself consists of forecasts. (The distinction is whether you assume the trader does their own analysis or just uses other people’s.)

Commercial forecasting varies hugely in the degree of rigour and sophistication involved; but a hell of a lot is based on models where you set up some limiting factors (say oil above $50 and below $150, and today’s margins and market share less a fudge factor) and then assume X per cent CAGR out to the forecast horizon. This gives you a forecast; if you’re serious about it you can do a sensitivity analysis based on wiggling various factors, but the biggy is always going to be the CAGR you picked. And as Dsquared says about the BoE Inflation Report, this comes down to arguing about it.

Now, if the entire analyst community’s choices of forecasting assumptions were independent and normally distributed….but that’s never gonna happen, not least because of the custom of constructing “consensus estimates” which inevitably tend to act as an anchor for the next iteration of analysis.

So, to a very large degree, the publicly available information itself is going to be strongly dependent on the forecasting process. And, fundamentally, what the hell makes the difference between an analyst consensus CAGR of 3% or 6%, which is enough to transform the NPV of a given stock and swing a huge amount of capital reallocation, other than Keynesian animal spirits?

(Looking back on the .com boom, of course, a special element in this was the fact that so many forecasters’ pay was closely coupled to the valuation of the stocks they analysed – remember Jack Grubman? – and it didn’t help that the inputs to the forecasting process for some sectors were strongly influenced by a number of large frauds. )

Going back to the gun predictor, a huge amount of its performance was defined by the preset forecast horizon; too short, and there would be no time to adjust the aim (or for the rounds to reach the target) and it would constantly fall behind, too long and the target might change course during the lag, and anyway the gunner would be tempted to adjust the aim back, thus getting into a destabilising feedback loop. It’s in the nature of such a system that has historical inputs and forward-looking outputs that the assumptions you use will dominate its forecasting skill relative to the baseline.

47

Kevin Donoghue 07.24.09 at 11:04 am

Tim Worstall: …I’m clearly missing something but how is it a disproof of the EMH that when someone finds, discovers, manufactures or codifies previously “not publicly available knowledge” then prices or behaviours change?

It’s no problem for the EMH that a new discovery in chemistry moves the market, but it is a problem if market participants need professors to tell them how to price financial instruments. The profs are telling them how to model the market, but the EMH says the market already knows the model. Admittedly there are several versions of the EMH but any version which doesn’t assume that the players understand the game is vacuous. At some point the weakest version of the EMH becomes indistinguishable from Keynes’s casino, where the price is determined by what average opinion believes the average opinion to be.

48

Tim Worstall 07.24.09 at 1:50 pm

“The profs are telling them how to model the market, but the EMH says the market already knows the model.”

Maybe that’s the part I’m not getting then. From the description at the top of the EMH it says “publicly available” knowledge. A new method of valuation is new knowledge, isn’t it? Something not publicly available before it was, umm, made public?

“The arguments so far concerned publicly available information,” said JQ.

This is something I just noticed this morning and yes, it’s an entirely silly story but offered as a ludicrous example of my logic (or illogic perhaps). Around where I live we always have problems with ticks on the animals: we get a lot of shepherds grazing intermittently on the common land around us. Last year most of the local flocks suffered from brucellosis and so this year there are hardly any around (local shepherds tend to breed their own flocks, not buy stock in) as most were slaughtered.

We are now buying less tick spray than we did last year. OK to take it to absurd limits: if you see an outbreak of brucellosis then short the stock of those who make tick spray.

This is new information, information that was not previously known to the participants in the financial markets (OK, lt’s just pretend it is shall we?). And I’m assuming that we all learn more about the world each day, new information like this is arriving all the time.

So, perhaps it’s that I don’t understand the EMH itself, but from what has been said it seems that the claim is that it processes all of the information currently publicly known. Prices reflect, efficiently (?) the current state of knowledge. But that state of knowledge is continually changing and I’m still not getting how markets adjusting their valuations in the light of new knowledge disproves the EMH?

49

Bunbury 07.24.09 at 2:26 pm

If Paul Samuelson said in 1998 that the stock market is micro-efficient but macro-inefficient, then the EMH in its meaningful form is truly undead since that must be the closest thing to a death certificate in economics.

One thing that you can take from Shiller and Jung and other papers is that even if the EMH were true, there would only ever be very limited empirical support for it. At the moment there are some really very weak results applying only to the most liquid markets and it is very hard to see how there could be much more. The EMH is really not a balance of probabilities kind of thing. Without empirical support any believer in a broadly applicable or strong form of the EMH is engaged in an act of faith.

The dividend example is particularly interesting because it is impossible for both dividend growth and returns to be unpredictable without a fixed dividend yield.

As presented the Black Scholes argument is not terribly convincing. It, or rather related techniques primarily allow the making of markets in financial derivatives and the profit comes from mark ups that are usually tiny in relation to other costs involved in the business. To some extent the mere fact that such vast volumes of derivatives can be traded are a vindication of market efficiency since all such formuale are based on the hypothesis of no arbitrages – which is a related but weaker condition than market efficiency. As stated the example is essentially saying that the existence of bid offer spreads is a contradiction of the EMH. It is but derivatives markets are not the best example.

There is a rhetorical difficulty with this topic and several others in the element of truth contained in the EMH. It is hard to find free lunches in financial markets and there isn’t really another universal source of valuations to demonstrate is fundamentally better. So some deft distinctions are needed to avoid getting bogged down by anyone getting the impression that the bits that are true are what is being denied. Richard Thaler’s no-free-lunch versus the-price-is right distinction is a good attempt.

50

Not Really 07.24.09 at 3:18 pm

> Commercial forecasting varies hugely in the degree of rigour and sophistication
> involved; but a hell of a lot is based on models where you set up some limiting factors
> (say oil above $50 and below $150, and today’s margins and market share less a fudge
> factor) and then assume X per cent CAGR out to the forecast horizon. This gives you a
> forecast; if you’re serious about it you can do a sensitivity analysis based on wiggling
> various factors, but the biggy is always going to be the CAGR you picked. And as
> Dsquared says about the BoE Inflation Report, this comes down to arguing about it.

Then with all that carefully-constructed input, you sit down in an executive sales strategy meeting where the key decision factors driving this quarter’s pricing, discount, and hiring decisions are (1) the VP of Sales’ need to make simultaneous balloon payment on his mansion, his yacht, and his daughter’s college tuition (2) the VP of Manufacturing’s hatred for the VP of Sales.

51

Kevin Donoghue 07.24.09 at 5:01 pm

Tim,

Here’s a way of explaining why I think your reasoning about EMH gets into trouble. Suppose you saw forex traders quoting a spot GBP/EUR rate which wasn’t the product of GBP/USD and USD/EUR, with the result that you could make money by a simple triangular arbitrage. You would presumably regard that as evidence of market inefficiency. You would hardly defend the EMH on the grounds that traders needed time to learn how to multiply spot prices. Yet you defend a transitory inefficiency arising from a failure to come to grips with the Black-Scholes model. So traders are expected to know about multiplication, but not about stochastic calculus. Well, where do you draw the line? Just what level of numeracy does your version of the EMH require of traders?

I’ve no argument against any version of the EMH which merely says that markets are unkind to traders who are behind the times. That’s not even as demanding as the no-free-lunch EMH. It says there’s no free lunch unless you find a new way to get one; and when the other guys catch on, you won’t get any more free lunches. That’s so weak that to my mind it doesn’t even qualify as an EMH. I can’t think of any economist, living or dead, who has disputed it.

52

Bunbury 07.24.09 at 6:07 pm

Tim, your position doesn’t explain why the EMH is true rather than one of the alternatives such as Andrew Lo’s Adaptive Market Hypothesis or Bossaerts Efficient Learning Hypothesis.

In other words the market achieves does make it difficult to find free lunches and does incorporate new information but it is quite hard to describe accurately how much that is. The EMH is very difficult to state precisely without it becoming either trivial (it’s not easy to predict financial markets) or implausibly strong in a refusing to believe that there could be a £5 note on the pavement sort of way. The stronger statements assumes a level of perfection that is demonstrably not achieved in practice, possibly not achievable even in principle and would not be empirically verifiable even if true.

In justifying his AMH, Lo explicitly draws parallels with evolution. To stretch that metaphor, all theories of evolution assume that species are well adapted for survival but the EMH, in a form worthy of the name, assumes that they are in fact perfectly adapted and that therefore there should be an important purpose for the appendix or the hind legs of whales. Other statements recognise that there is some slack in the system even if it is not easy to exploit and might reasonably called Quite Efficient Market Hypotheses.

53

Matthew Kuzma 07.24.09 at 6:15 pm

Much better title. I think you could go move vivid with the subtitle yet. Seven Economic Ideas that Refuse to Die is punchier but begs the question. There are a lot of ways to beat the metaphor to death, which you certainly want to avoid. But I still think the subtitle could be a bit tighter somehow.

54

Tim Worstall 07.25.09 at 11:23 am

“Suppose you saw forex traders quoting a spot GBP/EUR rate which wasn’t the product of GBP/USD and USD/EUR, with the result that you could make money by a simple triangular arbitrage. You would presumably regard that as evidence of market inefficiency.”

I interned decades ago in a Forex room so I’ve seen people doing that as it happens. Yes, an inefficiency, one that gets pretty ruthlessly exploited too and thus shrinks…..

“It says there’s no free lunch unless you find a new way to get one; and when the other guys catch on, you won’t get any more free lunches.”

That was pretty much what I was trying to say. OK, so I’m a believer in a very weak, possibly too weak to even merit the name, EMH. Good, thanks for clearing that up for me. Not all that dissimilar from Adam Smith’s points about capital moving between sectors in search of “excess” profits. Once the “excess” has been identified it gets competed away. Which is pretty much where I got my vague thoughts from in the first place.

55

Henri Vieuxtemps 07.27.09 at 10:23 am

The NewScientist calls it “Falling out of love with market myths“.

Comments on this entry are closed.