Foreword
The Efficient Market Hypothesis (EMH) is often cited, or at least alluded to, as the reason why everyone should just buy a basket of all stocks in the market, and then hold them passively. (1)
However, while I believe that the general advice is reasonable (2), the premise is, I believe, flawed.
As usual, a reminder that I am not a financial professional by training — I am a software engineer by training, and by trade. The following is based on my personal understanding, which is gained through self-study and working in finance for a few years.
If you find anything that you feel is incorrect, please feel free to leave a comment, and discuss your thoughts.
EMH, eh?
The EMH is generally attributed to Eugene Fama, in his seminal work “Efficient Capital Markets: A Review of Theory and Empirical Work”, though as the name suggests, many of the core ideas of the EMH did not come from Fama, but from others before him.
The gist of EMH is best summarized by a quote from the very first paragraph of that paper:
A market in which prices always “fully reflect” available information is called “efficient”.
Fama, E. (1970). Efficient Capital Markets: A Review of Theory and Empirical Work. The Journal of Finance, 25(2), 383-417. doi:10.2307/2325486
Note the key words: “always”, “fully reflect”, “available information”.
In other words, the EMH proposes that any public information is instantaneously (i.e.: always + fully reflected) incorporated into the prices of any securities affected by that information.
Models, models everywhere and not a single forecast to trade on
There are 2 main reasons why I believe the EMH is wrong — one is a technical reason, and the other is based on empirical observations.
Technically speaking…
The EMH is not really a hypothesis, so much as it is a model. It is a model of how financial instruments are supposed to behave, and the idea is that using that model, you can then make reasonable deductions about financial assets (or more accurately, their prices).
By definition, a model is a simulacrum of the original — models abstract away certain details of the original, to achieve a simplified representation.
Therefore, models are, by definition, wrong — when you remove certain aspects of the original in order to achieve the model, you are, in effect, creating something that is not a perfect reflection of the original, and thus it will never predict every single nuance of the original.
However, this doesn’t mean all models are useless! Within the assumptions on the parameters used to create the model, the model could very well be very predictive. For example, a simple model of the Sun is that it rises in the East and sets in the West. This is a model of how the Sun operates, but with the implicit assumption that you are observing the Sun on Earth, in a spot a little bit removed from the absolute North and South poles. If, say, you are observing the Sun from Mars, then this may not hold true any more. So, while this model is useful, because everyone I know is on Earth and none are on Mars, it is actually wrong — it implies the Sun revolves around the Earth in a prescribed path, instead of the other way around.
Ergo, all models are wrong, but some models are selectively useful.
EMH? This. Is. Empirical!
Going back to definition of EMH, note that it explicitly states that publicly available information are instantaneously reflected in the prices of security. Well, how often do you hear market moving information about stocks? Maybe once a day? Once an hour? Every few minutes?
But how often do stock prices move? If you have access to tick level information on stock prices, you’ll notice that they literally move every few microseconds. Microseconds. Are there really “publicly available price moving news” every few microseconds? If not, then why are the stock prices moving if they supposedly “always ‘fully reflect’ available information”? (3)
At a more high level, there exists easily observed price discrepancies in the stock markets. Take, for example, the stock symbols GOOG and GOOGL. Both are stocks of Alphabet Inc., the parent company of Google. GOOG represent class C shares which have exactly the same financial/economic interests as GOOGL, the class A shares. However, GOOGL, the class A shares, have voting rights on top of the financial/economic interests, while GOOG, the class C shares, only have the financial/economic interests.
Given that GOOGL = GOOG + “voting rights”, and voting is optional — you can choose to vote or you can choose to abstain, which means voting rights have a value strictly above 0, we should arrive at the conclusions that GOOGL should always trade at least as high as GOOG, and possibly a little bit higher. Right?
Well, would you look at that…
There are some who claim that prior to Q3 2021, because Alphabet Inc. does buybacks primarily via GOOG, therefore GOOG tends to trade at a higher price compared to GOOGL. I have no idea if that’s accurate, but on the face of it, it seems accurate enough — in Q3 Alphabet Inc. announced that they’ll also buyback GOOGL and the gap closed significantly.
Before the EMH crowd screams “Eureka!”… think about it. A stock buyback is essentially the company taking $N of cash and exchanging it for $N of its own stock. It is a financially and economically neutral move, i.e.: stock buybacks, according to the EMH, should not impact the company’s stock price at all.
Hypothetically speaking…
This is where I’ll admit that I was being a little misleading. If you read Fama’s paper in full, you’ll realize that he didn’t actually say that EMH is correct. In fact, he fully admits the hypothesis is wrong:
We shall conclude that, with but a few exceptions, the efficient markets model stands up well.
Fama, E. (1970). Efficient Capital Markets: A Review of Theory and Empirical Work. The Journal of Finance, 25(2), 383-417. doi:10.2307/2325486
Note that he clearly stated there are “exceptions”, and that the “model” isn’t correct, but that it “stands up well”. More importantly, he doesn’t even call it a hypothesis, but a model.
Because a hypothesis is a proposition of what reality is, and as all budding scientists know, “no amount of experimentation can ever prove me right; a single experiment can prove me wrong”, i.e.: just a single counter example, or exception, can prove a hypothesis is wrong. And we have “a few exceptions” here.
Which is to say, it appears that Fama is fully aware that EMM(odel) is a model, with all that implies about a model. It is close enough to reality that it is a useful model in some cases, but it is wrong to assume that the model is always right.
Practically speaking…
In practice, the EMH is useful essentially when you are unable or unwilling (4) to delve deeper into the data. By abstracting away a lot of the complexities of modern financial system, the EMH provides a useful simplification of what happens in the markets, and allows us to ignore those parts of the markets which we don’t care to care about.
For example, when you are developing a trading algorithm for SPY, the number of things the perfect such algorithm will need to know about is basically limitless — interest rates, consensus interest rates predictions, possible Fed initiatives, major events happening around the world, etc. The list is, quite literally, endless.
To make a perfect trading algorithm for SPY is thus impossible. But that doesn’t mean that a profitable SPY trading algorithm cannot exist! The EMH suggests that for the most part, you can assume away most of the details, and focus only on those bits that you have an edge on. For example, maybe you really understand how interest rates and SPY interact. Well, then you can build a model and an algo off that model, which assumes everything else is priced in (5), and just trade based off your simplistic model. Maybe it works, maybe it doesn’t — the point is, the EMH does not predestine it to not work.
In other words — the EMH is useful if there are some things you simply don’t care to worry about right now. Maybe v2 of your model/algo will take those into account. But right now, you have money to make.
Passive investing
Coming back to “passive investing” (1) — if you are unable or unwilling (4) to delve deeper into the data/details, and you simply want a carefree, easy way of investing your money, passive investing is a reasonable answer (2). This is a corollary of “the EMH is useful if there are some things you simply don’t care to worry about right now” — in this case, you simply don’t care to worry about any of those things.
But understand that it is reasonable, only because you are willingly looking at the problem from 10’000 feet away, and thus missing a lot of the nuances and detail that others who are more attentive may see.
Footnotes
- I intentionally avoided using “passive investing” in the foreword, because that term is often overloaded — some people mean “buy and hold” (i.e.: don’t trade too much), some people mean “buy baskets of stocks reflecting the total market” (i.e.: don’t do active stock selection), and some people mean both. For the sake of this article, I’m going with “both”.
- It is “reasonable”, in that for most people, it is pretty good advice — most people are unlikely to do much better than simply passive investing (as defined in (1) above), though this is not always true in every case. Remember that financial planning isn’t about maximizing your returns, it is the reverse — it is about finding an acceptable level of return, then figuring out the least risky way of attaining that return. Therefore, in some cases, it may be reasonable to adjust your holdings. For example, if you work in tech and your company pays much of your salary in stock, it may make sense to hedge against a general tech stocks decline by overweighting non-tech stocks in your investing portfolio.
- There are some who claim that the stock prices themselves are “publicly available information”, and thus, the “current” price move is just a reflection of the “prior” price move, i.e.: the stock price is moving because the stock price moved and generated new information. This is mostly circular reasoning that falls apart upon even cursory examination — as noted, the information must be “fully reflected” in the price “always”, which implies the information must be priced in instantaneously. There is simply no “prior” or “current” in an instant.
- Unable here means, well, unable. It doesn’t necessarily mean “too stupid to”. Similarly, unwilling here means unwilling — it doesn’t necessarily mean “too lazy to”.
- By the powers vested in me by the EMH, I pronounced all those factors I don’t care about “priced in”.