The “Old” View of Finance
2.4 Do empirical studies of the EMH shed any light on the actual speed of information transferal?
dispersal. This potentially stands to be useful in the ensuing chapters simply because our evolutionary model of biological-like information formation is (by definition) a non-instantaneous event. An empirical jury coming down in support of a Strong EMH type paradigm as typifying most financial mar- kets at most points in time would effectively strike a killer blow against our hypothesis of an army of economically rational analysts all producing new ideas and disseminating them into the marketplace in search of monopol- istic reward. Thankfully the empirical jury is by no means so unequivocal in its verdict. Rather, any feedback as to the validity of any of the various alternate forms of EMH in the “real world” seems to be at best characterized as being in a permanent state of quandary – not able to make up its mind if Strong, Semi-Strong or even Weak EMH reigns supreme. This is the best of all possible results for our hypothesis – the reason for which will become clear in the next section.
2.4 Do empirical studies of the EMH shed any light on
23 Accuracy of
forecasts in generating
“excess” returns
Information transmission over market “events”
Power of
technical analysis
Seasonal patterns in time series
Firm characteristics and excess return Lataneet al. (1970);
Black (1973); Holloway (1981); Copeland and Mayers (1982); Dimson and Marsh (1984);
Rozeff (1984); Stickel (1985); Eltonet al.
(1986); Kiem and Stambaugh (1986);
Campbell (1991);
Pesaran and
Timmermann (1995);
Womack (1996); Barber et al. (2001).
Reilly and Hatfield (1969); Kraus and Stoll (1972); Pettit (1972);
Grier and Albin (1973);
Watts (1973); Firth (1975); Ibbotson (1975);
Dodd and Ruback (1977); Joyet al. (1977);
Charest (1978); Watts (1978); Aharony and Swary (1980);
Rendlemanet al. (1982);
Fosteret al. (1984);
Pierce and Roley (1985);
Jain (1988).
Fama (1965); Fama and Blume (1966); Levy (1967); Jensen and Bennington (1970);
Pinches (1970); Praetz (1972); Fama and MacBeth (1973);
Cootner (1974); Fama and French (1988);
Brush (1986); Conrad and Kaul (1988); Poterba and Summers (1988);
Pruitt and White (1988);
Glosten (1989); Fama (1991); Campbellet al.
(1993); Ballet al. (1995);
Benik and Bossaerts (2001).
Granger (1975); Rozeff and Kinney (1976);
Branch (1977); Dyl (1977); French (1980);
Gibbons and Hess (1981); Brownet al.
(1983); Gultekin and Gultekin (1983);
Reinganum (1983a);
Keim (1983); Berges et al. (1984); Lakonishok and Smidt (1984); Keim and Stambaugh (1984);
Tinic and West (1984);
Kato and Shallheim (1985); Keim (1985);
Keim (1986); Chang and Pinegar (1986); Harris (1986); Ariel (1987);
Joneset al. (1987).
Dimson (1979);
Banz (1981);
Reinganum (1981);
Roll (1981); Basu (1983);
Brownet al. (1983);
Stoll and Whaley (1983);
Reinganum (1983b);
Shiller (1984);
Balverset al. (1990);
Chanet al. (1991);
Reinganum (1992).
Don’t be overcome by the extensive categorization and citation in the Table 2.1. The point readers should focus on simply stated is that all these tests have a common denominator – they search for associations between market information and market returns. Systematic excess return from private information (such as analyst forecasts) or under-utilized public information (such as technical rules, seasonal patterns or firm characteristics) implies a refutation of Strong EMH. Why? Primarily because under the Strong EMH conditions – as outlined in the previous sections – such information should be already reflected in an asset’s price. Thus asystematicexcess return to any one investor from having an “informational-edge” is implausible under the austere Strong EMH world. What these studies attempt to do is to categor- ize this “informational-edge” into a variety of forms. In terms of specifics, forms (1) and (2) of our categorized EMH testing methodologies in Table 2.1 – testing the accuracy of analyst forecasts in generating excess returns and measuring the speed of information transmission during market “events” – are obvious in their rationale for empirical EMH testing in the respect that they explicitly seek to clarify the existence and transmission of profit-making private information as it percolates into the public realm. That said, the latter three forms of EMH testing are much more subtle in their empirical approach since they attempt to identify systematic excess return from spe- cific trading rules rather than attempting to measure informational impact directly.
For instance, form (3) of the EMH testing methodology highlighted in Table 2.1 – estimating the profit generating capacity of technical analysis – is especially interesting as it attempts to test the hypothesis that price trends are an informational variable to market participants (an idea seized upon in some of the “new” views within finance described in Chapter 3 and indeed, technical analysis comprises one of the four major information categories for our information byte components).18 If price trends themselves were to exhibit predictive patterns in a fashion consistent with the precepts of tech- nical analysis, then this would imply a refutation of the principle of Strong EMH. Why? Discerning future prices from patterns exhibited in past prices under the guise of technical analysis is totally at odds with the “instantaneous adjustment” principles espoused under Strong EMH. Indeed, the tradition- alists are particularly harsh in their judgment of the efficacy of technical analysis. To quote Malkiel (1996):
Technical analysis is anathema to the academic world. We love to pick on it. Our bullying tactics are prompted by two considerations: (1) after pay- ing transaction costs, the method does not do better than a buy-and-hold
18The other major informational categories being – fundamental, economic and political information.
strategy for investors, and (2) it’s easy to pick on. And while it may seem a bit unfair to pick on such a sorry target, just remember: It’s your money we are trying to save. (Malkiel 1996, p. 139)
As stated earlier, Strong EMH argues that all information (both public and private) that is relevant to determining the future price of an asset is instant- aneously incorporated into an asset’s price upon its production – this includes latent price information. Subsequently there is no incentive for any analyst (including technical analysts) to systematically produce information in an attempt to position themselves for reward once this information is dispersed to the wider investment community as effectively there is no opportunity to do so. Under Strong EMH principles, the production and dispersal of informa- tion are inexorably intertwined and are instantaneous events. Indeed, as also mentioned earlier, even the skill-set used to produce information is effect- ively in the public domain the moment it has been developed. In one fell swoop this totally removes the incentive for skilled analysts to devote them- selves to a career of producing information – especially when it comes to enhancing memes and themes which are already in place in the hope of an even larger price impact and therefore subsequent reward. Under Strong EMH it is assumed investors already know the full consequences of a given information byte’s price impact-that we will demonstrate in Chapter 4 has the potential to be governed by Complex nonlinear principles when various information bytes come together to form memes. We find this assumption of perfect foresight a little “heroic” to say the least. Indeed, in Chapters 4 and 5 we will formalize the concept of “latency” inallinformation bytes that when unlocked generates the price movements that technical analysts feed upon for their particular style of analysis. In short, in contrast to the traditional- ist’s viewpoint, Evolutionary Finance principles are highly supportive of the pursuit of technical analysis as a mechanism for alpha-generation.
As for form (4) of the empirical EMH testing methodologies identified in Table 2.1 – determining if asset prices follow a reliable seasonal pattern – this particular methodology follows in a similar vein to the conceptual basis for the inclusion of technical analysis as a separate category for EMH empir- ical analysis. That is, estimating the degree of efficacy of seasonal patterns as reliable alpha-generators tests yet again the foundation for a degree of latency in information. Any statistically significant correlation between par- ticular trading “seasons” illustrates a link between past and present prices and the Strong EMH null hypothesis is effectively rejected. Why? Simply because investors could use such information to formulate a reliable trading rule. Rather, investors should already know of such phenomena (assuming it exists) and this should be reflected in prices – thus denying any return from trading “seasonals.” In the tests cited within this category in Table 2.1, vari- ous correlations were tested over alternate time horizons with different lags but the same principles applied – if any associative tendencies were identified
between trading season and prices the Strong EMH hypothesis was rejected.
The results were mixed.
Finally, form (5) of our empirical EMH testing methodologies – the link between firm characteristics and the presence of excess return – provides perhaps the most esoteric of all EMH empirical test-beds. Attempting to expli- citly link firm characteristics such as size to the presence of excess returns implies a failure of market efficiency. Why? If a market were truly efficient then such descriptive characteristics of firms and its relationship with price should already be known by all investors – thus denying any form of excess return.19 Rather, under Strong EMH principles characteristics such as firm size that do on occasion generate excess return can only do so under the somewhat random practice of “noise trading.”
So, these are the various critiques, what to make of this extensive – albeit conflicting – array of empirical evidence as to which form of EMH best typi- fies markets in the “real world”? It is often quoted that it is impossible to step into the same river twice. Apparently, the same principle applies to markets. We believe the academic stalemate in failing to conclusively come down in favor of a particular form of EMH can perhaps best be described by the time-dependent nature of market efficiency itself – and, by default, the biological-like determined speed of information production and dissem- ination. As we will argue in Chapters 4 and 5, the forces governing the biological-like molecular formation of information are difficult to predict and have the potential to generate complex nonlinearities in pricing behavior.
Without doubt, this has direct implications for any attempt at measurement of market efficiency simply because at times the market may take on a some- what random appearance in response to these nonlinearities, while at other times, the market follows a more predictable pattern. Indeed, we would favor a spectrum rather than binary – all or nothing – approach to EMH testing where the market evolves from fulfilling the conditions from one form of EMH to another as information, price and indeed the supporting infrastruc- ture of the market itself adjusts through time in response to the genetic-like forces governing emergent information. Obviously, to undertake such ana- lysis one would require a better understanding of the true microfoundations of information generation and transmission – a charter which our Evolu- tionary Finance approach aims to fulfill and which is indeed achieved in our extensively documented Game Theoretic microfoundations of analyst behavior detailed in Chapter 6. But more on this later, for now let us final- ize our critique of the “traditionalist” elements of information assessment within the existing finance literature by outlining the “old” view on the actual mechanics of information arrival.
19Note we are referring toexcessreturns here. Obviously smaller firms would command a larger risk premium given their higher potential to fail.