This means that market prices are predictable but only after the passage of 10 years of time or so.
I can 'predict' most things that happened in 1999 with relative accuracy.
In a market in which prices are set by emotion in the short term but by economic realities in the long run, investors should be changing their stock allocations in response to big price changes, not practicing Buy-and-Hold strategies.
This is not true at all. The issue is something like "What is maximum economic risk adjusted growth?"
EMH true believers think that the extra .2% per year we get is worth bank panics and depressions every 20 years.
Most non-autistic people think differently. They correctly note that in a world with lots of bank panics and depressions, they have good odds of losing everything because of somebody elses bad decisions. They correctly note that maximizing gains is not optimizing risk.
Even if the EMH was true - even if it was - using it as the exclusive guide in how to run the economy is stupid. We don't want to maximize our wealth exclusively, we as a citizenry, like relative stability much more.
How much of your net worth would you give up to guarantee you would not lose the rest? Pure EMH people say "I would never take this option", but most people would say something like 10%
We have laws and restrictions on every other area of complex human behavior. Would you really claim that we should abolish all stop lights and traffic laws?
Barkley, page 35 of Cochrane's text book: "[In deriving the general price equation] we have not said anything about payoff or return distributions. In particular, we have not assumed that returns are normally distributed or that utility is quadratic."
Its when you write down a theory of the stochastic discount factor --- when you give a specific theory of how the market prices risk --- that you *may* have to make assumptions about the distributions of returns. The CAPM does have these assumptions, but this is not the only risk model.
Brock and Lakonishok a long time ago showed that technical analysts can beat the market. Robin is right that it is unlikely that any regulatory body will be able to replicate this. However, there may well be good reasons for certain kinds of increased regulations, such as to either outlaw certain particularly dicey derivatives, or at least to force the markets to be more transparent and public.
TGGP,
Fama's reply is disingenous. I have no idea what Fama has told his students, but one cannot find anything about fat tails in Cochrane's book. If Fama spoke of fat tails, his students managed to completely forget it. His defense of "portfolio theory" as such is not of standard CAPM, but only of the most general sort of rule, such as a tradeoff between risk and return. Wow. It is also the case that Fama disagreed with Mandelbrot on some crucial points, such as that second moments vanish asymptotically, a key part of his argument for fat tails, although they can still arise if fourth moments do so, even if second moments do not.
I'm not claiming any relation to reality, but couldn't a regulatory model be shown to A: improve the accuracy of EMH valuations (Likely for a growth cost) and B: Prevent systematic valuation errors without claiming better valuation in the first place?
What value does denying EMH have for people who want more regulation?
People often confuse EMH with "the market is rational;" I certainly did prior to writing my Master's dissertation on the subject. The reality is that the market does not appear to be rational. The reality is also that humans are, in general, not systematically capable of out-predicting the market. Arguing that the market is Efficient in a strong sense, and then concluding that, say, regulation could not improve things, because the market is already Efficient, does not follow from the much weaker fact that people cannot routinely predict the market.
In other words, the real problem with EMH is the claim that people cannot systematically outperform the market because the market is so damn smart. People also couldn't outperform the market if prices were determined by pulling numbers out of a hat. The Crazy Market Hypothesis, or the Largely Random Market Hypothesis, don't square quite as well with good ol' neoclassical econ, so they get rather short shrift.
If there is such a thing as overvaluation, the market cannot possibly be efficient. An efficient market is one that sets prices properly. An overvalued market is by definition one that is not properly valued. So all that we need to do to show that the market is not efficient is to show that overvaluation is a meaningful concept.
Yale Professor Robert Shiller did this in 1981. He showed that through the entire history of the market stocks have generated better long-term returns starting from fair or low prices than they have starting from high prices. Valuations have always affected long-term returns. The market has never been efficient. At least not in the way that the conventional theory posits.
The confusion stems from the fact that the market must be efficient in the long run. The entire purpose of a market is to set prices properly. So there is a good deal of efficiency after 10 years and a high amount of efficiency after 15 years. Market prices are set by emotion in the short term and by economic realities in the long term.
This means that market prices are predictable but only after the passage of 10 years of time or so. It also means that the conventional investing advice of today (rooted in the old understanding that efficiency takes place immediately rather than only after a long passage of time) is the opposite of works. In a market in which prices are set by emotion in the short term but by economic realities in the long run, investors should be changing their stock allocations in response to big price changes, not practicing Buy-and-Hold strategies.
There is such a thing as market efficiency. But the old understanding is more wrong than right. The idea that investors don't need to change their stock allocations in response to big changes in stock prices has caused the biggest loss of middle-class wealth in the history of the United States. So we do need to get that old understanding corrected. And policymakers should be encouraging investors to lower their stock allocations when prices get to insanely dangerous levels (rather than to buy stocks regardless of price, a "strategy" that benefits only The Stock Selling Industry).
The distribution of returns has nothing to do with whether the market is efficient or not.
Case in point let's say I am selling a security that has a random payoff of $100 with 1/10 chance and $0 with 9/10 chance. Let's also say this security is not correlated with any other asset price, so that all marginal buyers are risk neutral.
As long as the security is prices at $10 it is "efficient" but the returns to it certainly are not normally distributed.
Bureaucrats may not know better than the markets what the price should be, but they can know whether a pricing system is stable or unstable; that is, whether or not the pricing system suffers from serious positive feedback.
When a pricing system has sufficient negative feedback, the EMH is a very useful hypothesis. The market's best guess is reflected in the price. With instability, the pricing system can give spurious results, results sometimes worse than a central planner's.
Case in point: look at the silver market back when the Hunt brothers were trying to corner the silver market. The market was doing a terrible job of pricing silver. The commodities markets suffer terribly from positive feedback. Contract holders are able to double up or double down. A price rise finances the bull's ability to make the price rise higher. Conversely, a price fall finances the bears. Too much leverage makes for positive feedback, turning speculation into a poker game.
You can find similar action with thinly traded penny stocks. Short traders can be squeezed by manipulators. When manipulators buy up enough shares, the shorts get margin calls forcing them to buy even though they are bears. The markets reflect the inverse of their opinions.
Nope. Proving that some particular method doesn’t work doesn’t require demonstrating a better method. For instance, a matchmaking service might claim that it can pair ideal partners based on their phrenological indices. Strict reliance on this method has produced some successes, and also some catastrophic failures. Does one have to have a reliable system for picking successes to be able to say that a system which produces frequent gross failures is defective?I'm not convinced that an after-the-fact demonstration of gross failures is sufficient to say that the process was not efficient. The problem is that the failures might have been truly impossible to forsee with the information available at the time.
An example of a market that would radically undervalue and overvalue (in retrospect!) securities, yet is nonetheless efficient: Consider securities which have a payout set at $0 or $1, based on the toss of a fair coin. Until the coin toss, these will all be valued at $0.5. After the coin toss, a decision to purchase a losing security at say $0.49 will look like a gross failure - but there was, in fact, no better valuation at that time.
The only way to demonstrate that an estimate truly is inefficient is to display a better one, computed at the same time as the inefficient one.
As you will see later in my posts pointing to Mandelbrodt/Taleb, having a range of predictable doesn't exclude the random. How else does one explain a bubble or a crash.
Markets exist for "dead call options" which are headed out of the money and into an exercise value of zero. Their prices usually tend to "bounce around" above zero until finally the exercise date is reached at which point they are not exercised.
In an efficient market all such options would go to zero and stay there...
(Hint: equities are call options on the value of the underlying company, at least for a corporation with limited liability. )
I'm not sure how regulators push the goal of rationality.But rational investment is the goal of the (villainous, if you believe the media) algorithmic traders.
Civilians hear the term "efficient" and think it means something it doesn't.
"Efficient" is a loaded adjective for what the EMH really means, which is that if you had a way to reliably predict price movements, you'd keep it to yourself! Or if you didn't keep it to yourself, others would use and over-use the prediction until it didn't work any more.
So how about we call it the "Unpredictable Markets Hypothesis"? That would better reflect the meaning of the concept.
Unpredictability does not mean that the price is actually of any use to anyone. It's perhaps the best prediction of the future price, but perhaps not a prediction of anything else.
This means that market prices are predictable but only after the passage of 10 years of time or so.
I can 'predict' most things that happened in 1999 with relative accuracy.
In a market in which prices are set by emotion in the short term but by economic realities in the long run, investors should be changing their stock allocations in response to big price changes, not practicing Buy-and-Hold strategies.
That depends on your time horizon, doesn't it?
This is not true at all. The issue is something like "What is maximum economic risk adjusted growth?"
EMH true believers think that the extra .2% per year we get is worth bank panics and depressions every 20 years.
Most non-autistic people think differently. They correctly note that in a world with lots of bank panics and depressions, they have good odds of losing everything because of somebody elses bad decisions. They correctly note that maximizing gains is not optimizing risk.
Even if the EMH was true - even if it was - using it as the exclusive guide in how to run the economy is stupid. We don't want to maximize our wealth exclusively, we as a citizenry, like relative stability much more.
How much of your net worth would you give up to guarantee you would not lose the rest? Pure EMH people say "I would never take this option", but most people would say something like 10%
We have laws and restrictions on every other area of complex human behavior. Would you really claim that we should abolish all stop lights and traffic laws?
You cannot claim you've never seen the Robert Schiller study about long term P/E ratios and future returns.
You've seen this study right? It is a great asset allocation tool if you have long term time horizons
Then, how about all of the Haugen studies?
I mean, this isn't even a contest. There are dozens of "anomalies" in stock markets - and this is the market that should be most "efficient"
Then there how about the famous noise trader study by Delong?
Some people call these anomalies, but I like to call them what they are - proof the EMH is BS and has always been BS.
You can lead a horse to water, but you can't make him drink.
Nicely stated, brazil84.
Barkley, page 35 of Cochrane's text book: "[In deriving the general price equation] we have not said anything about payoff or return distributions. In particular, we have not assumed that returns are normally distributed or that utility is quadratic."
Its when you write down a theory of the stochastic discount factor --- when you give a specific theory of how the market prices risk --- that you *may* have to make assumptions about the distributions of returns. The CAPM does have these assumptions, but this is not the only risk model.
Don Geddis,
Brock and Lakonishok a long time ago showed that technical analysts can beat the market. Robin is right that it is unlikely that any regulatory body will be able to replicate this. However, there may well be good reasons for certain kinds of increased regulations, such as to either outlaw certain particularly dicey derivatives, or at least to force the markets to be more transparent and public.
TGGP,
Fama's reply is disingenous. I have no idea what Fama has told his students, but one cannot find anything about fat tails in Cochrane's book. If Fama spoke of fat tails, his students managed to completely forget it. His defense of "portfolio theory" as such is not of standard CAPM, but only of the most general sort of rule, such as a tradeoff between risk and return. Wow. It is also the case that Fama disagreed with Mandelbrot on some crucial points, such as that second moments vanish asymptotically, a key part of his argument for fat tails, although they can still arise if fourth moments do so, even if second moments do not.
I'm not claiming any relation to reality, but couldn't a regulatory model be shown to A: improve the accuracy of EMH valuations (Likely for a growth cost) and B: Prevent systematic valuation errors without claiming better valuation in the first place?
What value does denying EMH have for people who want more regulation?
People often confuse EMH with "the market is rational;" I certainly did prior to writing my Master's dissertation on the subject. The reality is that the market does not appear to be rational. The reality is also that humans are, in general, not systematically capable of out-predicting the market. Arguing that the market is Efficient in a strong sense, and then concluding that, say, regulation could not improve things, because the market is already Efficient, does not follow from the much weaker fact that people cannot routinely predict the market.
In other words, the real problem with EMH is the claim that people cannot systematically outperform the market because the market is so damn smart. People also couldn't outperform the market if prices were determined by pulling numbers out of a hat. The Crazy Market Hypothesis, or the Largely Random Market Hypothesis, don't square quite as well with good ol' neoclassical econ, so they get rather short shrift.
If there is such a thing as overvaluation, the market cannot possibly be efficient. An efficient market is one that sets prices properly. An overvalued market is by definition one that is not properly valued. So all that we need to do to show that the market is not efficient is to show that overvaluation is a meaningful concept.
Yale Professor Robert Shiller did this in 1981. He showed that through the entire history of the market stocks have generated better long-term returns starting from fair or low prices than they have starting from high prices. Valuations have always affected long-term returns. The market has never been efficient. At least not in the way that the conventional theory posits.
The confusion stems from the fact that the market must be efficient in the long run. The entire purpose of a market is to set prices properly. So there is a good deal of efficiency after 10 years and a high amount of efficiency after 15 years. Market prices are set by emotion in the short term and by economic realities in the long term.
This means that market prices are predictable but only after the passage of 10 years of time or so. It also means that the conventional investing advice of today (rooted in the old understanding that efficiency takes place immediately rather than only after a long passage of time) is the opposite of works. In a market in which prices are set by emotion in the short term but by economic realities in the long run, investors should be changing their stock allocations in response to big price changes, not practicing Buy-and-Hold strategies.
There is such a thing as market efficiency. But the old understanding is more wrong than right. The idea that investors don't need to change their stock allocations in response to big changes in stock prices has caused the biggest loss of middle-class wealth in the history of the United States. So we do need to get that old understanding corrected. And policymakers should be encouraging investors to lower their stock allocations when prices get to insanely dangerous levels (rather than to buy stocks regardless of price, a "strategy" that benefits only The Stock Selling Industry).
Rob
The distribution of returns has nothing to do with whether the market is efficient or not.
Case in point let's say I am selling a security that has a random payoff of $100 with 1/10 chance and $0 with 9/10 chance. Let's also say this security is not correlated with any other asset price, so that all marginal buyers are risk neutral.
As long as the security is prices at $10 it is "efficient" but the returns to it certainly are not normally distributed.
Bureaucrats may not know better than the markets what the price should be, but they can know whether a pricing system is stable or unstable; that is, whether or not the pricing system suffers from serious positive feedback.
When a pricing system has sufficient negative feedback, the EMH is a very useful hypothesis. The market's best guess is reflected in the price. With instability, the pricing system can give spurious results, results sometimes worse than a central planner's.
Case in point: look at the silver market back when the Hunt brothers were trying to corner the silver market. The market was doing a terrible job of pricing silver. The commodities markets suffer terribly from positive feedback. Contract holders are able to double up or double down. A price rise finances the bull's ability to make the price rise higher. Conversely, a price fall finances the bears. Too much leverage makes for positive feedback, turning speculation into a poker game.
You can find similar action with thinly traded penny stocks. Short traders can be squeezed by manipulators. When manipulators buy up enough shares, the shorts get margin calls forcing them to buy even though they are bears. The markets reflect the inverse of their opinions.
Nope. Proving that some particular method doesn’t work doesn’t require demonstrating a better method. For instance, a matchmaking service might claim that it can pair ideal partners based on their phrenological indices. Strict reliance on this method has produced some successes, and also some catastrophic failures. Does one have to have a reliable system for picking successes to be able to say that a system which produces frequent gross failures is defective?I'm not convinced that an after-the-fact demonstration of gross failures is sufficient to say that the process was not efficient. The problem is that the failures might have been truly impossible to forsee with the information available at the time.
An example of a market that would radically undervalue and overvalue (in retrospect!) securities, yet is nonetheless efficient: Consider securities which have a payout set at $0 or $1, based on the toss of a fair coin. Until the coin toss, these will all be valued at $0.5. After the coin toss, a decision to purchase a losing security at say $0.49 will look like a gross failure - but there was, in fact, no better valuation at that time.
The only way to demonstrate that an estimate truly is inefficient is to display a better one, computed at the same time as the inefficient one.
As you will see later in my posts pointing to Mandelbrodt/Taleb, having a range of predictable doesn't exclude the random. How else does one explain a bubble or a crash.
Markets exist for "dead call options" which are headed out of the money and into an exercise value of zero. Their prices usually tend to "bounce around" above zero until finally the exercise date is reached at which point they are not exercised.
In an efficient market all such options would go to zero and stay there...
(Hint: equities are call options on the value of the underlying company, at least for a corporation with limited liability. )
I'm not sure how regulators push the goal of rationality.But rational investment is the goal of the (villainous, if you believe the media) algorithmic traders.
Civilians hear the term "efficient" and think it means something it doesn't.
"Efficient" is a loaded adjective for what the EMH really means, which is that if you had a way to reliably predict price movements, you'd keep it to yourself! Or if you didn't keep it to yourself, others would use and over-use the prediction until it didn't work any more.
So how about we call it the "Unpredictable Markets Hypothesis"? That would better reflect the meaning of the concept.
Unpredictability does not mean that the price is actually of any use to anyone. It's perhaps the best prediction of the future price, but perhaps not a prediction of anything else.