Arnold Kling  

Morning Commentary

Flubs... Who Likes Democracy?...

Richard Posner writes,

The efficient markets theory, when understood not as teaching merely that markets are hard to beat even for experts and therefore passive management of a diversified portfolio of assets is likely to outperform a strategy of picking underpriced stocks or other securities to buy and overpriced ones to sell, but as demonstrating that asset prices are always an adequate gauge of value--that there are not asset "bubbles"--blinded most economists to the housing bubble of the early 2000s and the stock market bubble that expanded with it.

The distinction that he attempts to draw (many economists make the same attempt) is an interesting one. The claim is that prices can be "wrong" (quite far from fundamentals) in an efficient market. That may sound fine, but in order to believe it, you have to believe either:

a) when a price is wrong, it has about an equal chance of moving toward right or moving toward more wrong

b) when a price is wrong, it has a higher chance of moving toward right than of moving toward more wrong.

If you believe (b), also known as mean reversion, then you don't quite believe that markets are efficient. I believe (b). There are times when I think that the market is wrong and I am willing to bet against it. I have a vague sense that I have done better than a purely passive investment strategy, but the fact that I do not seem to want to do a careful audit should give pause.

Read Posner's entire essay, which gives an unusually balanced treatment of the issue of what the financial crisis says and does not say about the state of the economics profession.

A reader asked me to comment on Demyanyk's Ten Myths about Subprime Mortgages. I agree with the substance but not the tone of the essay. If your point is that the crisis was caused by a combination of factors, then going through a subset of factors one by one and calling them "myths" is confusing.

Comments and Sharing

COMMENTS (13 to date)
Lance writes:

Would the EMH even apply to housing?

The housing market is not a liquid market, and I have never heard anyone defend the appreciation in home prices as "rational" by pointing to the EMH.

david writes:


If you've ever heard someone saying "if higher house prices weren't rational, someone would have bet against them by now", you've heard an implicit version of the EMH.

More generally, it's not hard to find examples of the EMH being applied to housing.

mattmc writes:

The EMH still works here at setting the price at each moment. The problem in bubbles is that people's predictions of the future value of an asset are wrong, causing them to be willing to bear a higher price now.

Shayne Cook writes:

The excerpt from Posner's essay cited above is not only wrong, it's wrong-headed - (I've not yet read the entire essay, so I may be taking his assertion[s] out of context):

1.) I don't believe that "most economists" were "blinded to" the early 2000s housing bubble - only the most vociferous and politically oriented economists were. And even they were merely ignorant of the possible (probable) large scale detrimental effects of the "housing bubble". This follows Arnold's thesis of the "suits" and "geeks" - the "geeks" knew and warned, the "suits" blissfully ignored.

2.) The validity of EMH can only be evaluated within the legal and regulatory framework in which the market exists. Given the governmental and regulatory bias towards (unsustainable) personal home ownership that existed during that period, there was little hope that housing prices could move anywhere but "more wrong".

As an aside, the U.S. is currently immersed in a "health care bubble". I'm making that statement in order to provide an historical reference to the future "Posners" who will claim five years from now that "most economists" were "blind" to it. And given the government's continued bias toward further (unsustainable) "free" health care for everyone, health care prices will continue to move in the "more wrong" direction.

Tom Dougherty writes:

While I agree that there is something wrong with mainstream business cycle theory, Posner answer to what is wrong is completely wrongheaded. Posner criticism is the same as Brad Delong’s and Paul Krugman’s of mainstream business cycle theory: if only old-time Keynesian economics were the dominate theory in academia the recession would have been predicted and prevented.

Old-time Keynesian theory is not equipped to explain or predict this or any other recession. More importantly, government policy is still Keynesian in practice, yet despite this, Keynesian policy was unable to prevent this recession. Anyone who thinks, like Posner, that if we only had some old-time Keynesian economics around the recession would have been prevented, deserves a whipping.

Mike Rulle writes:

Your belief in mean reversion is interesting (well, not belief, but leaning toward it). I do not think EMH says their cannot be bubbles, only that bubbles cannot be predicted before hand. They, rightly I believe, own the "null". To disagree, one has to set up an experiment (even an historical one) with operational definitions which would be difficult to create.

Lance asks if it applies to housing. California and NJ housing prices rose at similar rates between 2000-2007, but California had a true bust and NJ "just" a recession.

There is nothing easier than seeing what one wants to see--after the fact. I know, because I do it all the time. Mean Reversion may exist, but first lets give it an operational defintion (in the context of how I assume you mean it, relatively long macro-time frames). Then see if we can test it.

Good luck!

Dorian Taylor writes:

I am a neophyte when it comes to economics, however I have recently become interested due to my exposure to information systems. I find the efficient-market hypothesis somewhat confusing in that it seems to exhibit wildly different behaviour at different scales (not unlike the disparity between classical and quantum mechanics).

I understand the most rudimentary form of a transaction to constitute the following: Two principals, two objects* to trade, and two partially-overlapping** sets of information about the objects, one of each for each principal.

(* could be an object, tangible or not, or a service, or a mix, aggregate, or whatever.)

(** The infosets would have to at least partially overlap because it is assumed that the principals know something about what they are trading.)

It also makes sense to consider a trade as a discrete phenomenon, as is the act of acquiring specific information. Each occur over an arbitrary, but logically atomic period of time, and transit from one state to the next instantaneously. I understand the reference trade scenario of the EMH has both principals arriving at a site to swap objects which they mutually agree, empirically and with complete confidence, have identical value.

Where the EMH begins to confuse me is first in distinguishing between goods and information. If two perfectly-informed and rational principals arrive at a site to carry out a trade, where did they get the information from? Information requires a physical carrier medium and therefore must somehow occupy space (even if just light). Information also begets information (e.g. you know that I know that you know, etc.). Given that, it is very difficult for me to imagine both principals to share the information required to evaluate an even trade without being one and the same.

It follows then that each trade must therefore exhibit an arbitrary disparity of value. Not necessarily zero-sum (i.e. both principals could be better off than before), but in a given trade, one principal must always make off at least slightly better than the other. What's more, either principal may or may not be aware of the disparity. I understand that with enough entropy there could be an equilibrium across the whole system, although in examining any part of it, we can observe monumental gaps in wealth and information.

I have been using the following model to understand economic activity:

Consider a single principal's effort to achieve an arbitrary state. In some cases it is an individual effort, in others it is acquired through trade. For each state, the principal must evaluate with which approach the economy lies. It must then locate another principal to carry out the transaction, as well as acquire the requisite information. This represents a recursive computation of arbitrary complexity, where every bit of information accumulated up to and including the transaction itself alters the state of the system. It is therefore impossible to gauge a priori completely effectively that the trade is even, and as such a heuristic must be applied. To add even more complexity, the subparts of any given transaction are not necessarily synchronous, and as such its state cannot be meaningfully evaluated until it exits.

In short, if we get the shaft on one transaction, we can recoup our losses (and then some) with another. However, the wealth of any individual principal (gained from trade) will amount to the sum of the value of the transactions in which it participates.

Pockets of disparity will therefore shift about the system and very likely collect in certain places, as any principal could acquire a superior means of evaluating transactions as a whole. And this isn't even accounting for the caprice of individual people, which I will leave out, because I have written enough already.


Felix writes:

Maybe someone can clear up my confusion about EMT.

If EMT implies that the market has the current best guess of the "right" price of a company's stock at some indefinite point in the future, then that price is really no help as we don't know when that future time is, exactly. It could be right now. It could be 10 years from now. It could be several times in the future. But how, exactly, is one to know when the time(s) are?

Anyway, is there even a "right" price at all? If so, is the "right" price as chaotic or distributed in value as stock prices are in time? Is a "market" a search function that tries to find that "right" price? If there is no meaningful "right" price, then what's going on here?

Mr. Econotarian writes:

I also find the concept of a "right price" a bit silly.

EMH stats that prices already reflect all known information, and thus no one has "special knowledge" that others do not have that is not priced into the price.

EMH does not state that prices reflect information that is unknown or unknowable, and future predictions of complex systems (such as markets) are typically less and less knowable as you try to predict farther and farther out.

Shayne Cook writes:

To Felix:

"If EMT implies that the market has the current best guess of the "right" price of a company's stock at some indefinite point in the future ..."

This is slightly mis-stated, possible leading to some confusion on your part. In stock market fundamental analysis, the market attempts to establish (guess) the net present value of all future revenue/earnings streams from an asset to establish a current price for that asset, not its future price. That present value "guess" includes consideration of (at least) possible future stock price appreciation, interim dividend yields (if any), risks of impairment to future earnings, as well as all alternative investment options with alternative risk/future revenue stream probabilities. Company stock investing is an exercise in "right pricing" an asset, given all known information that could affect its future earnings potential. Note that this (very briefly) describes fundamental investing strategy.

Note that trading (as in "day-trading") is quite different. Trading strategies assume the "market" has mis-priced assets, short-term, and generally to a fairly small degree. Trading usually entails taking a leveraged "short" position in an asset, if the trader believes the market has currently mis-priced the asset too high, and a "long" position if the trader believes the market has mis-priced the asset low - usually via options. The option-based leveraging allows traders to multiply the returns on their "bet", compensating for the relatively assumed small market mis-pricing of the underlying asset.

Walt French writes:

A price is "wrong?" Only if it misstates a transaction that took place or an honest quote that was made. Everywhere and Always, if you want some cant.

As an investment advisor, I do not have the foolish luxury of thinking that a price "should" be X; rather I am charged with buying assets where I believe the price will rise, or shorting assets which I believe will fall. My kind has a strong interest in my price beliefs coming true (Exhibit A in another case, how wildly inefficient incentives are in motivating Pigovian solutions to economic problems).

Really, the good perfesser might seek out economists more experienced in Finance before he says what "should" be so, lest he be treated as a jet engine designer who comes across as foolish because he can't identify a motorcycle's blown head gasket while any farmer who maintains his own tractor would.

Here's a starter, and this is advice for some financial economists, too: even in the social sciences, an "hypothesis" states a yet-to-be-proved theorem. Theorems are based on postulates; NONE of the proposed non-trivial EMH postulates can be accepted for housing markets (and by obvious extension, investment in mortgages and their derivatives).

Ever teach Intro to Finance or Investments? Here's an essay question for the kids: if house prices are variable (check), "banks" borrow short and lend long (check), and loans sometimes default (check), said banks may find themselves owning a lot of real estate that cannot be sold for enough to cover its liabilities. Given incentives tied to capital structures, that bank can only claim the prices are "wrong" (with Treasury cheering them on) or submit to the FDIC. (1) Come up with contorted reasoning under which EMH has anything to do with markets for housing debt. (2) Ditto, for why bank stocks that derive their returns but limited risk from mechanisms above supplemented with social insurance, should be 1/4 of our corporate wealth. (3) Describe why studies about financial (sometimes, mis-) regulation, e.g., those of Diamond & Dybvig and Akerlof & Romer, are wrong, based on EMH. (Extra credit for #3) Why did the US even create the Fed, when, based on Randian precepts, it is unnecessary?

Felix writes:

...In stock market fundamental analysis, the market attempts to establish (guess) the net present value of all future revenue/earnings streams from an asset to establish a current price for that asset, not its future price...

Thanks, Shayne.

So if the "right" price is the present value of etc. etc., then that right price must be a distribution rather than an exact number. Distributed over a range of values, that is. And, it would be distributed to the extent that (unpredictable) randomness affects the future value. Which is unknowable, itself.


Shayne Cook writes:


You've got it - anything dependent on future activity is probabilistic (follows a distribution), rather than being "fixed point". And there are two other dynamics you should consider in this context:

1.) In efficient-market hypothesis (EMH), the availability of information that may affect future revenue/earnings streams from an asset is critical. Given that all information (and mis-information) is uniformly available, much of the current price variance of an asset is attributable to variance in individual market participants' variance in ability to assimilate that information. EMH has supposedly been discredited by the recent housing bubble, due to the market's supposed "failure" to assimilate available information and thereby identify and preclude bubble pricing. Hardly. Again, there was as much or more mis-information available to the markets as information. The three rating agencies' "AAA" grade on various "toxic" financial instruments is a glaring example. I rather think EMH was verified by recent events - once the markets discovered that mis-information was indeed mis-information, markets re-priced assets very efficiently, quickly and violently.

2.) There are risks and uncertainties, "known unknowns" and "unknown unknowns" respectively, that can affect the current "right" price distribution of an asset. Risk (the known unknown) can be arithmetically compensated for, uncertainty (unknown unknown, or "The Black Swan, see Nassim Taleb) cannot - by definition.

Comments for this entry have been closed
Return to top