Debunking Economics, Part XIII: Alternatives to the EMH

The Efficient Markets Hypothesis is pretty much indefensible. It is based on ridiculous assumptions: all investors have access to money at the same interest rate, have the same information and interpret information in the same way. It also has counterfactual implications: according to the EMH, markets would stay in equilibrium and move only when new information became available (which they don’t); people would not consistently outperform the market (which they do); and in its strongest form, it actually implies that bubbles can’t exist. The only defence its proponents seem to be able to muster is that it can’t predict anything (and sometimes, that economists full stop can’t predict anything). I could go on, and have, as have others. But what’s more important is exploring the many available alternative theories of finance. This is the purpose of chapter 15 of Steve Keen’s Debunking Economics. Keen goes through and assesses the major alternatives to the EMH on by one.

First, Keen mentions the obvious choice: behavioural finance. But he doesn’t really explore all the different heuristics and biases that people experience in financial markets – that would and has taken entire books. Instead, he objects to the way that EMH proponents initially defined ‘rationality.’ Apart from basically meaning prophetic, it was based on a misreading of John von Neumann, the creator of Game Theory, who said that his definition of rational would only apply when games were repeated enough times. A game with an unlikely but large loss as one of the possible outcomes looks less appealing when you do it once than if you play it 1000 times, allowing the losses and gains to even out.

Hence, Keen touches on something that others have mentioned: the whole idea that behaviour is either ‘rational’ or ‘irrational’ is not a useful way to think about human behaviour. In fact, behavioural finance retains some unfortunate implications carried over from economics: that we need to reduce everything down to individuals making choices, and that if only people behaved how economists think they should, then financial markets would be efficient. Having said that, behavioural finance is promising and useful field, though so far it is still in its early stages with no clear forerunning theories.

There are, however, a few theories which have been fully developed, and look incredibly interesting. The additional bonus is that they are complementary to each other (and to behavioural finance).

Fractal Markets

A fractal is a pattern that looks the same no matter how much you zoom in or out (see above). So it’s no surprises that one of the implications of the fractal markets hypothesis is that markets display similar patterns of behaviour over a day, month, business cycle or what have you. The fractal markets hypothesis models price movements as a function of previous price movements, which explains the emergent fractal pattern, and also means that stock markets will exhibit a tendency for volatility to produce more volatility, something contrary to the EMH.

A skeptical reader might suggest that this implies future price movements are easy to predict, if only one had the relevant formulas. But a system as complex as this would be highly dependent on initial conditions: just a tiny error in the initial values would soon produce results that were wildly offbase. This is what happens with weather models, and is why weather predictions are more likely to be right the closer you are to the day. It is actually probable that calculating prices accurately ould be computationally impossible using a fractal model.

But this might beg another question: why is the stock market not more chaotic? This is explained by dropping one of the assumptions of the EMH: that investors trade with identical time horizons. Similarly to von Neumann’s observations, a trade that looks bad for a day trader due to large potential losses at any one time, could look good for a long term trader if it has net positive yields over a given period. Hence, introducing heterogeneity makes the model more realistic. A highly promising theory.

The Inefficient Markets Hypothesis (IMH)

Provocatively named by its originator Bob Haugen, who has written three books full of data contrary to the EMH. The IMH suggests that markets systemically overreact to price movements, and hence cause incredibly inefficient allocations of resources.

Haugen identified three sources of price movements: event-driven, error driven and price driven. The EMH assumes away the second two, but Haugen has calculated that the third one accounts for up to 95% of stock market volatility, because price movements create a self-perpetuating spiral as investors seek gains or cut losses. Haugen has concluded that the stock market in its current form is a serious drag on investment, and suggested reducing the length of the trading day or simply having one auction per day.

Econophysics Approaches

Physicists have recently turned their hand to economics, and, due their strong empirical bent and the relative lack of data in economics, have been drawn to finance, where streams of data are readily available. Keen comments that much of Econophysics would perhaps be better named ‘Finaphysics.’

There has been a plethora of suggested approaches from the physicists, mostly applying their various chaotic theories to economics: earthquake models, power laws, the Fokker-Planck model. Keen does not go into much detail here because, again, it would take an entire book. He briefly goes over Didier Sornette’s earthquake model, which has been used to make explicit predictions about the future of stock markets. Keen directs the reader to this website, which supposedly tracks its predictions, though I cannot find anything after a quick look.

So there are many alternatives to the EMH, and each involve making explicit predictions and drawing on data, rather than handwaving ‘the market is volatile we can’t do anything about it’ statements. Personally, I consider the fractal markets hypothesis the most promising framework, and it is also one that can easily incorporate elements from the other approaches. I look forward to future developments in all of thee theories.


, , , , , ,

  1. #1 by Aziz on October 22, 2012 - 1:41 pm

    21st Century Economics is likely to be defined by the use of the concept of disequilibrium.

    • #2 by Ben Brennan on October 23, 2012 - 12:37 am

      I get excited anytime I read a paper with a differential, only to see that they assume equilibrium and drop the term without bothering to see if it’s a stable equilibrium or not.

    • #3 by Unlearningecon on October 23, 2012 - 1:35 pm

      That’s certainly an optimistic perspective, but we’ll have to wait at least a generation in my opinion.

  2. #4 by john77 on October 22, 2012 - 1:56 pm

    The EMH is just that – a Hypothesis which does not represent the real world. It was a simplification in order to make it easier for American academics to do the sums – the real world is fantastically complex. Using EMH, the academics have developed theories, some of which have some use in the real world..
    You seem to be assuming that the EMH pretends to be the real world. It doesn’t. In the real world we have death and taxes and dealing costs and insider dealing (which would be pointless under the strong version of EMH) and investors with different time horizons and mutual funds with investors with different risk preferences and ….
    “Modern Portfolio Theory is neither modern, nor a portfolio, nor a theory” – several people independently coined that phrase (not me – I was born too late). Salesmen use the phrase MPT to make their sales pitch sound impressive but anyone even half-seriously studying investment theory knows that MPT and CAPM doesn’t apply to the real world and is, at best, a first approximation. There are some useful ideas, such as portfolio diversification, derived by the Chicago guys using EMH but most of them can be obtained from classical investment theory without needing the EMH.
    Behavioural finance is not a theory about investment – it is a study of investORS.
    Price movement can and does create further price movement as a consequence of the new fad of “momentum investing” (and the older fad of chartism aka “technical analysis”) but it is difficult to know how much a lot of the evidence is unreliable because one cannot know, without mind-reading ability, how much price movement is a result of the earlier price movement and how much is due to slightly delayed reaction to the news item that triggered the first price movement. Fractal theory is very pretty but it is just a pretty face

    • #5 by Unlearningecon on October 23, 2012 - 1:37 pm

      I don’t understand why so many defenders of economic orthodoxy have the ‘it’s just a theory’ attitude.

      Theories/hypotheses can be falsified. You can’t just cling to them no matter what. The EMH has been proved false again and again, as have its applications in finance.

      Behavioural finance tells us the behaviour of investors, which predicts what they will do in markets, and therefore the behaviour of markets. And characterising fractal theory as ‘just a pretty face’ is simply lazy. It fits the data far better than the EMH.

      • #6 by john77 on October 23, 2012 - 3:19 pm

        The Efficient Markets Hypothesis is NOT A THEORY. so it cannot be proved wrong, merely demonstrated to be unrealistic, which I have done on several occasions.
        What you *should* be saying is that MPT has been proved wrong many many times – I myself have proved (to a standard that could be, and was, exposed to scrutiny) it wrong more than once.
        You seem to equate Theory of Investment with the behaviour of stock prices in markets – that is only one part of it and, as has been repeatedly demonstrated, a minor part. The majority of investment returns come from interest and dividend payments and capital returns rather than proceeds of sales.
        Fractal theory, like the “random walk” applies *only* to day-to-day movements in stock prices. Just what does it tell you about dividends, returns on capital employed, gearing risk, unquoted “private equity” in public companies, private companies, property investments, risk premiums in both equity and debt markets, the benefits of financial gearing when return on capital exceeds the cost of borrowing, the tipping point when it becomes better to have a rights issue instead of increasing further, the benefits and costs of operational gearing in a cyclical growth industry, etc?
        “lazy” – one of my papers which as a side-effect noticed by the editor showed all three versions of EMH did not work (I had got bored with pinting that years before)disproved MPT was 70 pages long. But I look at returns on investment not just day-to-day share price movements.
        Over what term have you measured how good a fit fractal theory gives to stock price movements – the last seminar I attended that covered “momentum”.concluded that it worked well if you sold out after 4-5 months but after two years the strategy showed a loss *even before expenses* – now that is from a proponent, so is unlikely to be unfair to it: fractal theory would have momentum investing working over any timescale.
        Presumably you were born after the invention of “Modern,Portfolio Theory” and have had your hearing saturated with claims that MPT is *the answer* as you seem to be unaware that the likes of Warren Buffet, Anthony Bolton and Bob Yerbury use Classical Investment Theory which involves a damn sight more hard work than looking at pretty pictures of scattergraphs of stock prices. It still works over any period long enough for temporary fashion trends to end (which can be a long time – Tony Dye got sacked six months before the dot-com bubble burst).

      • #7 by Unlearningecon on October 23, 2012 - 3:41 pm

        The clue is in the name: hypothesis. It can be proven to be wrong, and hence abandoned!

        I will freely admit I know little to nothing about modern portfolio theory. You say the fractal markets hypothesis fits only specific periods of time; do you have any links to papers that demonstrate this?

      • #8 by john77 on October 24, 2012 - 9:44 am

        You don’t seem to get the point. The proponents of EMH say “let us suppose that the market is efficient and see what results we get”. We can state until we are blue in the face (and have done so) that in the real world we have death and taxes and inefficient markets but that doesn’t invalidate *all* the theories derived from EMH and it won’t be completely abandoned because a few of them actually have some use.
        All you need to know about MPT is that it was fashionable but is, and always was, wrong
        I did *not* say that “fractal markets hypothesis fits only specific periods of time”. What I *did* say was that it only applies to market prices movements which is a relatively small part of total return and is only small fraction of investment theory. I then said that momentum investing which appears to be the reason why someone invested the fractal markets hypothesis only works if you sell out after a short period.
        There is no link to the paper I listened to and the access to the pdf is behind a paywall but it does contain some references, one of which is If you read through that there is a graph on page which effectively refutes fractal.

      • #9 by Unlearningecon on October 24, 2012 - 10:23 am

        This goes back to Friedman. You can’t just have a black box of a theory with counterfactual prepositions about an economy and say ‘well it can still be useful’ – we need to know causal mechanics. And you haven’t even shown exactly how the EMH has proven useful – as far as I can tell, most of the time economists use their financial models they blow up companies (LTCM) or the economy (2008).

        And you seem to have a double standard here. We can ‘refute’ fractal but the EMH holds a revered position?

        What if I said ‘fractal is just a theory. Sometimes it’s useful; sometimes it isn’t. You can go on about reality all you like but fractal isn’t concerned with reality, it’s just a tool.’

        Well, that would be stupid. I’m happy to look into evidence against fractal but the EMH needs to be thrown out entirely, along with epicycles and the phlogiston theory of fire.

      • #10 by john77 on October 24, 2012 - 11:14 am

        You are asking someone who has been denouncing MPT since before you were born to justify the use of EMH??
        i) Diversification reduces risk
        Well, we knew that but apparently US academics wanted to prove it and EMH allowed them to do so
        Using EMH you can fund that 20 equally-weighted stocks give you 90% of the diversification benefits in terms of reducing stock-specific risk of complete diversification across the whole equity market. That may not be precisely true but is useful because it enables people to see whether a fund manager is deliberately taking a risk in concentrating his portfolio – conversely whether adding more stocks to a portfolio is a strategic decision or just opportunistic because they look cheap.
        ii) Higher risk is correlated with higher return and investors trade off one against the other.

        LTCM blew up due to a combination of hubris, a political misjudgement and the use of numbers from the normal distribution when dealing with a distribution which simply had to be non-normal because it is skewed. EMH does not state there is no sovereign risk – in fact EMH allows for shocks from unforeseen events such as Russia defaulting.
        The 2008 crash had a lot of causes, the largest of which was the debt bubble where hundreds of millions of people and thousands of companies accumulated unsustainable levels of debt. UK personal debt increased by £1 trillion in a decade. Lots of us said that it was going to end in tears: just because we aren’t professors of economics, that doesn’t mean that we don’t exist. There was even a new subsector of financial services created to deal with the flow of insolvencies turning into a flood. The actual timing of the 2008 crash caught me by surprise because I thought Bear Sterns deserved to go under more than Lehman so after the US Treasury arranged a bail-put for Bear Sterns I didn’t expect it kill Lehman, but I and others predicted a deflation when consumers stopped *increasing* their borrowings. .

      • #11 by Unlearningecon on October 24, 2012 - 4:00 pm

        But the EMH can’t ‘prove’ anything because it isn’t a model of financial markets. It s a model of an alternative universe – as you say, we already know that diversification reduces risk (that’s obvious), and there are plenty of correct models that imply that, too.

        The EMH may allow for shocks but one feature of fractals is that highly volatile events are likely to create more highly volatile events. The EMH has a hard time dealing with this.

      • #12 by john77 on October 25, 2012 - 9:48 am

        Not quite – *your misinterpretation* of the EMH belongs in an alternate universe. The EMH does not require that “all investors have access to money at the same interest rate, have the same information and interpret information in the same way” – it merely requires that market prices reflect all known information about the stocks. Parts of Modern Portfolio Theory require these things.
        EMH is an assumption, using which academics “proved” things. The assumption is unrealistic but so is your statement “highly volatile events are likely to create more highly volatile events”.- markets have alternated between periods of high and low volatility and high volatility has been triggered more often by news than by market movements – “Black Monday” which followed the first hurricane to hit England in three centuries was the most volatile trading period I remember and volatility progressively reduced again over the next.few days. Actually EMH, with which I have said I disagree, fits this better than fractal theory.

      • #13 by Unlearningecon on October 25, 2012 - 11:14 am

        This is plain wrong. Have you read Sharpe’s initial formulation? He makes all of those assumptions and appeals to a Friedman style argument to justify them.

        I said all this in my earlier post on the EMH, which I linked to in this one. You are weaving between incredibly weak definitions of the EMH and stronger ones in an attempt to deflect criticism. I don’t even know why you’re doing it considering you think the EMH is wrong.

      • #14 by john77 on October 25, 2012 - 12:43 pm

        I am plain right.
        Sharpe’s initial formulation is of the Capital Asset Pricing Model, not of the EMH and he uses EMH as just one of his assumptions in developing the CAPM and introduces those three (not phrased in quite those words) on the *ninth* page of his paper as additional assumptions (which he admits to being unrealistic) in order to derive a theory as to what would happen in an equilibrium situation. .
        You seem to be saying the CAPM is based in an alternate universe so EMH is in an alternate universe.
        If all investors did indeed have the same expectations and the same rate of interest etc you would not have any market at all, let alone an efficient one. Markets depend on the existence of buyers and sellers who must, almost by definition, have different preferences or different views in order for one to wish to sell when the other wishes to buy.

      • #15 by Unlearningecon on October 25, 2012 - 3:45 pm

        Hmm yeah perhaps I was wrong about that.

        But my overall point remains: the EMH has no worth. It uses normal probability distributions. It uses an odd definition of rational investment. It is at odds with empirical reality. There is no point in keeping a hypothesis that is so wrong, as it has been falsified.

      • #16 by john77 on October 25, 2012 - 4:38 pm

        All of those points are valid criticisms of MPT and are only a few of those that I have been making spasmodically (because I can get tired) for more than 30 years.
        I have been (and still am) critical of EMH but it is MPT (of which CAPM is a core component) that deserves your ire. EMH ignores market inefficiencies and most stuff derived from EMH ignores death and taxes but however much I complain people will continue to use it as a first approximation, just as people still use Euclidean geometry instead of Riemannian geometry to build houses even though they know the earth is curved and parallel lines meet at the poles (they also meet at the antipodes but most people don’t appreciate that). EMH does not, ipso facto, use normal distributions: some MPT theory does because that makes the maths easier, but let’s face it – if stock prices were normally distributed they could not more than double from the mean (although could from a point on the lower part of the distribution) and the number that have more than doubled is incompatible with the normal distribution.
        The example I gave – that 20 stocks gives you >90% of the benefits of diversification *is* a useful result (which cannot be obtained without some assumptions): there are others if I wanted to hunt them down (but I don’t – each of us have spent far too long arguing whether you should say MPT instead of EMH in your first sentence).
        Earlier you asked why I could just refute fractal theory but allow EMH to continue – the answer is that I *have* refuted EMH *many* times with the cry “death and taxes” (it sounds better than “dealing costs”) but I cannot stop people using EMH, however much I wish that I could.

      • #17 by Metatone on October 26, 2012 - 7:48 am

        I think John77 is confusing scope here. The EMH says stock movements are only about news. Fractal suggests that overall volatility can be connected to previous volatility – it’s not a statement that you can predict individual market movements in a way that generates profits. Profiting from volatility requires making actual up/down bets.

  3. #18 by Blue Aurora on October 22, 2012 - 3:36 pm

    At last, you finally arrive at the econophysics project, Unlearningecon! Benoit Mandelbrot’s “fractal markets hypothesis” actually inspired the works of the econophysics project, I might add. If you read the Physica A database, you can find references to “multi-fractality” in the econophysics literature. So it seems that the econophysics project may have the hope of reforming economics and possibly overthrowing Subjective Expected Utility – something Mr. Steve Keen does not explicitly criticise!

    • #19 by Unlearningecon on October 23, 2012 - 1:39 pm

      He does use this chapter to channel von Neumann’s criticisms of immeasurable utility, but I left that out as there’s only so much one can do. Are you aware of this criticism?

      • #20 by Blue Aurora on October 23, 2012 - 3:54 pm

        There’s a difference between Subjective Expected Utility and von Neumann-Morgenstern Expected Utility. Nevertheless, both approaches are used by orthodox economists as basic approaches to decision-making. (S.E.U. more than VNM Expected Utility from what I can tell at least.) But to answer your question, I’m not very familiar with this criticism, no. Still, S.E.U. decision theory needs to be criticised more often!

  4. #21 by Oilfield Trash on October 23, 2012 - 2:54 am

    Yes, but while you can approximate irrational behavior with ODE in models you still have to have some accepted laws to build on. Imagine how reliable our understanding of physics would be if someone discovered the speed of light was not constant and we could not figure out why.

  5. #22 by Matías Vernengo on October 25, 2012 - 2:52 pm

    I highly recommend on this subject the paper by James Crotty here

  6. #23 by Eric L on October 26, 2012 - 4:55 am

    “The IMH suggests that markets systemically overreact to price movements, and hence cause incredibly inefficient allocations of resources.”

    Agree on part A, but stock markets have little to do with allocating resources so I’m not sure it matters. None of the money going through the stock market goes to the companies whose income is being traded (except in the case of IPO); the function of a stock market is to give those who invested in a company before it was publicly traded a financial incentive to have done so.

    • #24 by Metatone on October 26, 2012 - 7:45 am

      You are correct about the real function of stock markets, but the theories all base around the allocating resources idea.

  7. #25 by A Castaldo on October 28, 2012 - 10:36 pm

    You could also have mentioned the Adaptive Market Hypothesis of Andrew Lo, one of the more interesting alternatives that try to go beyond the EMH in my opinion.

    • #26 by Unlearningecon on October 28, 2012 - 11:49 pm

      I had not heard of that, but thanks for the HT.

  8. #27 by Scott Dauenhauer CFP, MSFP, AIF on October 31, 2012 - 9:31 pm

    Reblogged this on The Meridian Blog and commented:
    An interesting series at the Unlearning Economics blog – this post is in a long line of posts that “debunk” economics. The point being that economists have been getting things wrong more than right and its because they are not working with models that make sense. I won’t proclaim to know THE model that does make sense, but there are some posts here that are worth reading.

  1. 10 Sunday Reads | The Big Picture