Defender of Efficient Markets Hypothesis Gives Away the Game
For a while now, I have been a lone voice crying in the wilderness, claiming that the Efficient Markets Hypothesis (EMH) had no empirical content. It was not a falsifiable theory, but rather a way of looking out at the world and organizing it. Note that this isn’t a criticism–it’s what Misesians think about pure economic theory in general! However, the reason I stress the point is that a lot of EMH proponents think they are being really “scientific” in the same way that quantum physicists are. Here’s what I wrote a while ago in response to hilarious defenses some economists gave for the EMH in the wake of the financial crisis:
The efficient-markets hypothesis comes in various forms. There is, indeed, a large empirical literature, in which Fama and others conducted falsifiable tests. However, as I hope I’ve demonstrated with the quotations above, in practice the efficient-markets hypothesis is actually a tautology, or a way of viewing the world.
There’s nothing wrong with using a priori mental frameworks to parse economic reality; indeed that is one of the defining characteristics of Misesian praxeology . However, as the quotes show, many of the EMH apologists think they’re independently confirming the EMH, when, in fact, their goggles simply force all evidence into conformity with their presuppositions.
So it was refreshing to see (HT2 Krugman) Stephen Williamson–in his attempted takedown of Zombie Economics‘ John Quiggin–say the following:
The Efficient Markets Hypothesis: For Quiggin this is “the idea that prices generated by financial markets represent the best possible estimate of the value of any investment.” Here, Quiggin is badly confused, but maybe the finance practitioners are not helping him out much. Market efficiency is simply an assumption of rationality. As such it has no implications. If it has no implications, it can’t be wrong.
Exactly, Prof. Williamson. I’m glad you agree with me.
Incidentally, this young lad Noah Smith is pretty clever. Too bad he seems to think Chancellor Palpatine is the best leader the Galactic Senate has ever had.
I don’t get the Palpatine thing, but I may just be slow this morning.
DK, I mean Noah has a lot of potential, but he’s been seduced by the Dark Side (Krugman and DeLong).
Oh haha.
Yes – the free association between Krugman and the Sith apparently doesn’t leap to my mind as readily as it does yours!
Personally I don’t consider it tautological. Of course, it’s ridiculous, but it’s not unfalsifiable. I don’t think its simply an assumption of rationality, without implications. It seems to be a financial theory based on a number of the same flawed assumptions of neo-classical thought that leaves itself open for falsification because it makes claims about price behavior(i.e. under certain situations the market WILL NOT do X; e.x the shares of large, profitable businesses should not trade to a penny and than back up in a matter of seconds, as we saw in the flash crash)
We can measure and analyze price behavior. It’s pretty easy to show – almost on a daily basis – where market prices DONT represent the best estimate of value.
It is not falsifiable because any test rests on two hypotheses: (a) the markets are X efficient; (b) the model is correct. Tests can fail because (a) is false or (b) is false and it is not knowable which is the case.
Most of us financial economists agree that it is not binary, anyway. We see degrees of efficiency, based on the abundance of information.
It is not falsifiable because any test rests on two hypotheses: (a) the markets are X efficient; (b) the model is correct. Tests can fail because (a) is false or (b) is false and it is not knowable which is the case.
You don’t think we can know whether a model is correct?
Given that every test is implicitly a joint hypothesis, one cannot know which of the two has failed. Thus, we can only fail to reject the null if both (a) and (b) are true. I have yet to see a study that has done this and is universal to all time periods and locations.
Given that every test is implicitly a joint hypothesis, one cannot know which of the two has failed.
Suppose we want to prove that a guy is the father of a certain baby (let’s say that if he is shown to be the father then he is on the hook for child support). So we do a paternity test, and it comes back positive. The guy says, “ah, but given that every test is implicitly a joint hypothesis, one cannot know whether I’m the father or whether the test is inaccurate. Therefore you have proven nothing.”
Do you see the flaw in the guy’s reasoning?
Except that in your example, it is not a joint hypothesis. DNA testing has been proven accurate independently of paternity testing. This is not the case for the EMH or the models. Neither EMH nor any of the models have been independently shown to be an accurate depiction of reality. This isn’t to say it can’t be done; just that it hasn’t and it’s not clear how it could be done.
The second thing is that if the null hypothesis of efficiency cannot be rejected, there is no problem. It’s only when we reject the null that there’s a problem.
I’ll put it to you this way, if we run a test of the EMH and it comes back positive, we can then be as certain about the validity of the EMH as we are certain about the accuracy of the test. No test is going to give us Cartesian certainty, but it can give us some, and some tests more than others.
The problem is that proponents of the EMH and praxeology certain act as if their theories have empirical content. Bob said a couple of months ago, for example, that if Bernanke was able to dispose of the excess reserves banks were sitting on without causing either an economic collapse or severe inflation, then he would have to rethink his adherence to the Austrian theory (there were a bunch of other caveats to this, but they were all empirical). But why? If the Austrian theory is purely a matter of deduction, then surely it can’t be refuted by something as petty as events.
The truth is that both the EMH and praxeology do have empirical content. If they didn’t then they wouldn’t be worth bothering with. The idea that they do not seems mainly to be used as a way of dodging criticism when the empirical predictions of the theory don’t turn up.
> If the Austrian theory is purely a matter of deduction, then surely it can’t be refuted by something as petty as events.
I don’t find it strange that events confirm or disconfirm a theory or school of thought. As to whether something “has empirical content”, I’m not sure what that means.
I think both EMH and praxeology purport to help us understand the real world, and if they do a bad job of it, then we are correct in reconsidering our acceptance of them.
Help out a lay person here. How would you prove EMH to be false?
How would you prove EMH to be false?
Which version of the EMH?
It’s not falsifiable. See my comment to Jon O. above.
FYI, I don’t think I can accept a Jon O. type explanation…
“We can measure and analyze price behavior. It’s pretty easy to show – almost on a daily basis – where market prices DONT represent the best estimate of value.”
If this were true, there are a lot of people that would like to speak w/ Mr. O.. A skill like that would be worth billions (or trillions?). The question is: Can you identify these issues real time? Or, do you need 20/20 hindsight to tell you these prices were wrong? Looking back and pointing out errors doesn’t disprove EMH.
Hmmm… now that I think about it, even if Jon O. did have this skill — the skill to out rationalize the market — all it would prove is that he had better tools. Once Jon O. comes out and informs the world of his tools and trade, those tools will cease to work so well for him. His added rationality would be included in market pricing. Isn’t that what EMH is all about?
I don’t mind being corrected if I have screwed something up here.
Well tell me if you need hindsight for this one. Below is a bar chart of Exelon Corp. In 2008 they made about $2.8 Billion, a utility company with relatively consistent profits . At they time of the flash crash they were valued at about $30 Billion. During the flash crash, in a matter of a minutes, the stock traded down to 1 cent per share. That would value the company around $10 million. Then in a matter of minutes the market shot back up to a more reasonable valuation back in the 10s of billions. (The chart doesn’t show the low of .01 because the exchanges arbitrarily busted a number of legitimate trades[my system shows a low of .41,] but the trades did occur and apply to a comprehensive theory like EMH, either way the point is the same)
http://chart.finance.yahoo.com/z?s=EXC&t=2y&q=c&l=on&z=l&a=v&p=s&lang=en-US®ion=US
As for your last points there are people who are billionaires because of their ability to spot underpriced or overpriced assets. Ask Warren Buffet or Steve Cohen or Jim Simmons.
Oh, and why would Jon. O come out and inorm the world about his awesome abilities if its profitable? This argument sounds like something I’ve heard from professors who taught BS financial theory without knowing the first thing about how financial markets actually work. You can create/believe all the cute theories you want but until you realize that people aren’t always rational, profit-seeking, homogneous automotans; markets don’t always move towards equilibrium; and prices are dependent over time, you will be lost.
On an anecdotal note: I was trading S&P futures during the flash crash. When the market broke all liquidity disappeared and I can assure you that almost no one was making a rational decision as to the proper value of future S&P earnings. Sell orders were coming in out of pure panic, technical selling, and forced liquidations. IT HAD NOTHING TO DO WITH MARKET VALUATIONS !
I think getting hung up on what behavior or decisions are “rational” or not is a mistake. Perhaps an avenue of exploration for the Chicago moreso than the Austrian school.
What is clear is that the participants in the market at the time of the flash crash determined Exelon to be nearly worthless. Surely these participants were algorithms trading on momentum in some way. Perhaps this is evidence of the folly of momentum trading.
But it is easy to see the flash crash as a temporary temporal dislocation, whereby for a small window, there were way more sellers than buyers. The buyers reasserted their presence shortly afterward and the market carried on.
My feeling on efficient markets is that distributed signals via market participants in a profit / loss system will tend to beat any single prognosticator. The market reflects the sentiment of its participants. The implication is not that no one can outperform the market — rather that it is impossible to choose the person who will outperform the market.
1) What new information came about in that ~10 minute period that led people to believe a $30 billion company was worth $10 million and then $20 billion+ again?
2) It doesn’t matter if they were algos or not. EMH is a theory about financial markets. You can’t exclude certain parts of the market that invalidate the theory.
3)EXC is just an extreme example to prove a point. Every day markets make movements that can not be explained by even the weak form of EMH.
4)No, the implication of EMH is that you can’t beat the market.
1) Market participants seemed to be trading on momentum / psychology / rigid stop-loss strategies. The new information may have been a 10% price drop, perhaps on fundamental news.
2) I’m not excluding algos. They are agents acting in pursuit of a goal.
3) I don’t find those market movements disconcerting. I’ll be the first to admit that I don’t understand all the implications of strong or weak EMH. I am not a proponent of formal EMH. I simply believe the market a) reflects real world sentiments and b) allocates capital more efficiently than any central planner
4) I don’t think so. Otherwise it would be simple to disprove by counterexample.
Er, clarification regarding #4. EMH may imply that *I* cannot beat the market. But I don’t think it implies *no one* can beat the market.
Likewise, I am 99.99% sure your license plate is not “ABC 123”. This is a much different proposition than saying *no one’s* license plate is “ABC 123”.
I understand what you’re saying and agree with most of it.
4)You’d think so but apparently a “good” theory is immune to reality.
I’d just like to point out that actually in a stock market situation, the price seen is only a reflection of the price of those shares that are trading — not a true reflection of the value of the overall company, although many people presume this to be the case.
Think about it this way, I buy a bunch of shares for $10 per share with the intention of holding onto them for 10 years and hope to sell for $20 per share after 10 years. It really does not matter to me if someone else crashes the price down to $5 per share for a day or two in the meantime… my subjective valuation of the shares remains that they will be worth $20 in 10 years time. I am unwilling to sell them for less than that, and on the day that the price comes crashing down I simply don’t sell.
Shares only trade at the margin where bids and offers overlap, but behind that margin sits a whole bunch of subjective valuations that don’t overlap… and this is perfectly normal.
Scott, in your reflections here, you are falling into the trap that the big guns at Chicago fall into. At first you claim that nobody can beat the market, relying on EMH-type arguments. But then you admit that even if somebody did, it’s exactly what the EMH predicts.
See how no matter what happens, the EMH called it?
This was actually one part of the that review that annoyed me immensely. Enough so that I blogged about it. http://increasingmu.wordpress.com/2011/08/05/other-assumptions-of-efficient-market-hypothesis/
I didn’t look at it from the standpoint of non-falsifiability and the ERE. But why base an entire field (modern portfolio theory) on a framing mechanism with no empirical content? I have no problem with with using EMH or the ERE as a way of think about the world, but many economists “drink the koolaid” and actually believe things operate this way. On certain margins, Quiggin is right.
More to the point, I think you need other assumptions to get to EMH, just as you need other assumptions to get ERE, so Williamson is oversimplifying regardless.
Yes Bob, and clearly your predictions and those of other skeptics have been far superior to those of the market, hence your tremendous wealth and early retirement. lmao
You’re a market fundamentalist who apparently thinks the market is often hugely wrong for many years at a time in critical ways, instead of learning to listen to the market and seeing where it wants to go. You’re a living internal contradiction and a clown of the earth.
Bob, if you were being interviewed no how wrong your economic predictions have been, you’d sound somewhat like this guy: http://video.cnbc.com/gallery/?video=3000037435
What a joke.
David, remind me of your explanation for why gold and silver have shot through the roof in the last two years. Do you subscribe to Krugman’s Glenn Beck theory?
I couldn’t help myself, David S., and I followed your link. As usual, you have contributed nothing to the conversation. I have been saying for a long time–most recently in April–that the fundamentals did not justify the stock market’s heights, and that it could collapse. It drops 4% yesterday. To remind me of how dumb I am, you compare me to a guy saying the stock market drop makes no sense, in light of the economy’s solid fundamentals.
I’m tempted to say your link is irrelevant, but that would be too nice. Your link is the exact opposite. If you posted Peter Schiff hemming and hawing about price inflation, that would have made sense.
Bob, I can simply refer you to the World Gold Council and emerging market demand versus world supply. That explains most of the run up in gold prices.
http://www.gold.org/investment/statistics/
But, of course, I don’t have to explain anything. I’m not the one continuously drumming on about coming high inflation, even having been wrong for years. What nerve you have to claim you were right about this market downturn when your model is wrong. lmao
And what you and Rhen have in common there is that it’s like you’re speaking from alternate realities. Each of you is in his own universe, in which your crude, anti-market models of the world are somehow justified by what you see around you. You are both obviously clowns and the market makes fools of both of you daily.
David S. wrote:
Bob, I can simply refer you to the World Gold Council and emerging market demand versus world supply. That explains most of the run up in gold prices.
Oh, you explain the increase in price by saying demand has increased more than supply. My model certainly can’t handle that type of explanation…
Davis S. wrote:
You are both obviously clowns and the market makes fools of both of you daily.
Except Thursday.
The argument that EMT is unfalsifiable is not merely an argument against EMT, it is an argument against all real-world economic observations. Consider for example the statement, “Obama’s stimulus package was a failure.” you could point to unemployment and say the statement is proven, but it’s just as easy for someone to claim that unemployment would have been a whole lot worse without the stimulus. Once again, it falls back to circular reasoning and largely comes down to your point of view.
To some extent you can get around this with computer simulation. The problem with real-world economics is that you have only one sample point, one world to observe, and you can’t go back and reset the situation for comparative purposes. On a simulator, you can run as many worlds as you like, and you can go back and reset them as many times as you like. Thus you can make accurate comparative statements about whether the simlation reached some sort of optimal endpoint.
Admittedly, the nature of economic calculation is that we have no method to accurately simulate human thought in detail, but in terms of an overall system applying economic calculation as a simplistic “maximum profit” motivation can achieve quite plausible outcomes at the aggregrate level.
Needless to say, I’m not the first guy to go down this track, and I might recommend a book called “At Home in the Universe” by Stuart Kauffman (1995). Just a warning that this book is written from the persepective of a strong belief in Evolution, so some of it might jar with religious readers. Moreover, Kauffman doesn’t waste time arguing the pros and cons of evolutionary theory, that’s not the intention of the book. Readers just have to accept that it is written from an evolutionary perspective.
One of the things Kauffman does is to construct a simple formalised simulation methodology to explore what he calls “rugged fitness landscapes”, in such a way that the difficulty of the optimisation problem can be tuned from very easy problems, to very complex and difficult problems. What Kauffman’s simultion discovered is that a Stalinist central planning approach is good at solving easy optimisation problems in big systems… better than allowing each individual to attempt a local optimization. However, a network of individuals works better at solving high complexity difficult optimization problems. In addition, Kauffman explores the concept of various sized firms that operate somewhere between the Stalinist central planning and complete individualism, and it turns out that for a given fitness landscape there tends to be a best fit firm size that handles that landscape.
I’ve not really done real justice to the book in just a few paragraphs, and it’s one of those hard work books to get through, but the point is that it does apply a scientific approach in a way that is repeatable and falsifiable.