Richard Rorty and the efficient markets debate

I use the efficient markets hypothesis in my research and in my blog.  Once I started looking at the world through the EMH lens, I found it much easier to understand the relationship between policy and the financial markets—particularly in my research on the Depression.  Here I’d like to do three things; indicate why I believe markets are more efficient than they seem, acknowledge that there are events that look like market inefficiency, and then argue that those perceived inefficiencies, even if real, don’t have the policy implications that many people assume they have.

Last Sunday I discussed cognitive illusions, aspects of economic theory that are highly counter-intuitive.  I regard the EMH as one such economic theory—strange, but (almost) true.  Let’s start with all of the studies showing market inefficiency.  Many of these studies show that there are market patterns, or anomalies, that seem inconsistent with the EMH.  But how do we know these aren’t just coincidence?  One answer is that we can use statistical tests, such as the example of the correlation between equity returns on Wall Street and rainy days, which I discussed a few days ago.  But are those statistical tests reliable?  They tell us that certain events would be extremely unlikely to occur by chance, but what does that mean?

If I go to Vegas and observe the numbers 17-34-23-1-23-5-31-7-15-25 show up on a roulette wheel, I could argue that the odds were more than one in a quadrillion against that exact combination showing up on that particular wheel at that time.  In response, anti-EMH types would argue that anomaly studies don’t just find patterns, they find interesting patterns.  But there are lots of interesting patterns, far more than you might think at first glance.  I’m always noticing interesting patterns in random strings of digits.  Indeed the “uninteresting” string of numbers that I just listed, are actually very interesting—can you see how?

I am not an expert on EMH research, so let me just say that as long as asset prices seem to follow a path even close to the random walk predicted by theory, I am not going to be very impressed by academic studies finding “statistically significant” anomalies.

But of course there is a much more powerful argument for the anti-EMH position; the large asset price movements we occasionally see that don’t seem to have any fundamental explanation.  Some of the most famous are the 1929, 1987, and 2000 stock market bubbles, as well as the 2006 housing bubble.  Of course there are many more, but I think you will agree that these are some of the primary examples that EMH opponents point to.  I’d like to consider these events from several perspectives, beginning with the question of fundamentals.

1.  The 1929 bubble is the easiest to explain with fundamentals.  Studies have shown that stocks were not grossly overpriced in 1929, and the collapse has a very good “fundamental” explanation—the Great Depression.

2.  I know of no explanation for the 1987 bubble.  The collapse (comparable to 1929) occurred when the overall economy was doing fine.  Some argue for “computer trading,” as if computers have free will.  I am not saying they are wrong, but then the correct term would be “really stupid computer programmers.”  If correct, this is a good argument against the EMH position.  But even here, one must be careful not to push things too far.  At the time EMH opponents probably assumed more than just a inexplicable price change—they probably assumed that the bubble’s peak represented irrational exuberance.  Were stock prices too high before the 1987 crash?  We had no way of knowing then, and we still don’t really know.  There is enormous uncertainty about what the stock market should be trading at, based on “fundamentals.”

I find it interesting that anti-EMH hypotheses often seem to be in conflict with each other.  Some talk about “irrational exuberance” at various market peaks.  Others point to the extraordinarily high average rate of return on long term equity investments in the U.S., which greatly exceeds the return on bonds (even adjusting for the higher risk on stocks.)  Of course this latter anomaly implies that stock prices in the U.S. were far too low throughout much of the 20th century.  So which is it?  Is the stock market often way too high?  Or way too low?  And why do we hear so little discussion of “negative bubbles?”  Is there even a word for the concept?

The 2000 bubble seems about half way between 1929 and 1987, there were some fundamentals involved, and traders certainly were looking at firms where there was great uncertainty (unlike say GM or Ford in the 1950s and 1960s.)  But even so, in retrospect the valuations look far to high at the peak.  Once again, however, a cautionary note for the anti-EMH crowd.  Didn’t Robert Schiller mention the famous “irrational exuberance” phrase to Greenspan in 1996, before the stock bubble occurred?  If so, this shows how hard it is to offer useful investment advice, even if you sense the market is overvalued.  Those that exited in 1996 would have missed both the peak and subsequent crash.  I also recall reading that Galbraith predicted a crash in January 1987, far too early to help stock investors.

It is widely assumed that the 2006 housing bubble was irrational, and perhaps in part it was.  But again, let’s not get too overconfident.  In fact irrational overconfidence—the same psychological trait that may generate bubbles, also generates an excessive level of confidence that we can spot bubbles.  We all tend to remember when we make a correct prediction, or even have a correct hunch.  But how often do we remember our failures?  And if we forget the failures, do we grossly overestimate our batting average?

How many remember frequent predictions of bubbles, that turned out not to be bubbles?  I have already argued that even today we don’t know for sure that stocks were overvalued in 1929, 1987, or 1996.  There are respectable models that show they were fairly priced.  (The 2000 NASDAQ is different.)  How many people recall all the predictions of coastal housing bubbles during the long divergence after 1980, when an enormous gap gradually developed between housing prices in Middle America, and housing prices in LA/SF/NYC/Boston?  Guess what, those bubble predictions were wrong.  The housing price gap never really closed.  Of course in any rapidly rising markets there are some pullbacks, and when these pullbacks occurred, the EMH opponents crowed “I told you so” about the ridiculous prices in coastal markets.  But each new cycle the gap got wider, and in the long run the pessimists were wrong, even this crash hasn’t significantly narrowed the gap.

The housing bubble in 2004-2006 was partly driven by rapid immigration from Latin America (as was the bubble in Spain itself!), and also by a perception (which turned out false) that coastal zoning constraints were spreading into interior markets.  Many Hispanic immigrants were snapping up older ranch houses, allowing native born Americans to move on to bigger McMansions.  The immigration crackdown in 2007 dramatically slowed this immigration (as did the worsening economy.)  Population growth estimates going several years forward fell sharply, hurting housing speculators.  Ground zero of the sub-prime bust is in working class areas of the Southwest and Florida.  Any guess as to who bought homes in those areas?  In addition, after 2006 nominal GDP growth slowed gradually, and then very sharply, to a rate far below the level any rational investor could have anticipated in 2006.  Even today, few people seem to realize the impact that going from plus 6.5% to negative 6.5% nominal growth has on housing prices.  This didn’t trigger the collapse, but it dramatically deepened it.

I am certainly not arguing that fundamentals can fully explain the sub-prime crisis.  There may have been some irrationality, especially in interior southwestern markets where land was still fairly plentiful.  I’m sure that lots of explanations that have been offered (such as the Black Swan problem) have some merit.  Consider this, however, if investors are foolish to ignore the risk of Black Swan events, why should we trust prob. values in anomaly studies?

More importantly, even if the anti-EMH position seems correct in a few cases, I will argue that it does not have the policy implications that many assume.  So let’s ask ourselves this question; what are the policy implications of the anti-EMH position?  And is there any evidence in the famous anomalies to support these policy implications?

It seems to me that there are two basic policy implications of the anti-EMH position, investors can do better than indexed stock funds, and regulators should prevent market bubbles that are likely to be disruptive to the broader economy.  And I see zero evidence to support either of these policies, even if one accepts the view that many recent market anomalies cannot be explained by fundamentals.  Let’s start with the easy case—managed stock funds.

EMH opponents argue that asset price movements are somewhat predictable, due to various market anomalies.  We have a highly competitive investment industry made up of some of the smartest people from the best universities who are working day and night to outsmart markets.  And many of them do.  The problem is that they are not successful enough to push me away from indexed funds.  A few managed funds will do better than indexed funds, but (ex ante) we don’t know which ones.  So I don’t think there are any investment opportunities (open to the general public, i.e. me), that can reliably beat indexed funds.  And don’t tell me that mutual fund X or hedge fund Y did such and such.  Ex post there are lots of investments that have done very well, but that information is of no value to investors unless excess returns are serially correlated, and I don’t believe they are.

The big public policy issue, of course, is not whether indexed funds are a good investment, but rather whether the sub-prime fiasco shows the need for tighter regulation.  I am really surprised by how many people seem to simply assume that the recent crisis shows the need for better regulation.  You already know some of my objections to this view—the pro-regulation position is usually based on the assumption that the housing crisis somehow caused the current recession.  This is despite the fact that virtually every cutting edge macro text says that monetary policy determines NGDP growth.  And yet the moment a crisis hits we get a sort of mass amnesia, and reputable economists are suddenly assuming that the rapidly falling NGDP is not a failure of central banks, but rather of commercial banks.  As if it is suddenly the job of commercial bankers to manage monetary policy!

I know that nobody will buy this argument, so I should just give it up.  But even if I am wrong, even if the commercial bankers are 100% to blame for the current recession, the sub-prime fiasco does not support the anti-EMH argument for tighter regulation.  I don’t doubt that one can find some arguments for regulation (moral hazard, too big to fail, etc.), but what I do deny is that any of these arguments are related to market inefficiency, to the anti-EMH position.

One can look at the sub-prime fiasco from a theoretical perspective, or a empirical perspective, but what one cannot do is compare an ideal regulatory scheme to actual banking practices.  No one doubts that we would be better off if we could go back in time and install a regulation banning sub-prime mortgages in 2004.  But if we had that ability, the bankers would have also known what was coming, and would never had made the loans in the first place.

[I hope no one gives me the silly moralistic argument that the villains got off Scott-free, while innocent investors were left holding the bag.  The villains are exactly the people who have lost $100s of billions of dollars, even after the bailouts.  I know it is never that way in Hollywood movies, but real life is different.  This never would have happened (even without regulation) if bankers could have seen into the future.]

So the anti-EMH argument for regulation must be based on the following; bankers are irrational and make lots of foolish loans.  Regulators are rational and can see that these loans are too risky, and can protect bankers from hurting themselves.  At a theoretical level this doesn’t even pass the laugh test.  But what happened in practice?  What position did the “regulators” take in this crisis?  First we need to define “regulators,” who are much more than just the low-paid Federal bureaucrats that oversee the banking industry.  Regulators are the watchmen, those who watch the watchmen, and those who watch those who watch the watchmen.  In other words:

1.  The President
2.  Congress
3.  The Fed
4.  The media
5.  Most academics
6.  Nouriel Roubini

Guess how many of these institutions warned us about the sub-prime crisis.  Now guess how many were encouraging banks to behave even more recklessly than they did.  Unless we plan on making Roubini dictator of the world, there is zero evidence from the sub-prime crisis that simply giving regulators more power would have helped.  And how do we know that even Roubini wasn’t just lucky, and might miss the next fiasco?

This is a good time to trot out my favorite philosopher, Richard Rorty.  In a recent book he quoted an old pragmatist maxim; “that which has no practical implications, has no philosophical implications.”  I would re-word that slightly for the current discussion:

That which as no practical implications; has no implications for economic theory.

Thus Rorty suggested that it was pointless to argue about whether something is an objective fact or a justified belief, as we have no access to an extra-human perspective, and thus can never resolve the debate.  To take another example, imagine two people looking at Mt. Monadnock.  One says “That’s a very small mountain.”  The other says “No, that’s just a big hill.”  Pointless debate, isn’t it?  Now imagine two economists look at the tech bubble.  One says “boy, those rational investors made a big mistake,” whereas the other says “no, the investors were irrational.”  Another pointless debate.

The only aspect of the EMH/anti-EMH debate that is of any interest is the policy implications.  Are there any practical implications to the anti-EMH position?  I have no trouble looking at some of the bubbles we considered here and saying: “Boy, those investors sure seemed irrational.”  But it doesn’t help, because unless the anti-EMH people can find some interesting policy implications (investment advice for me, or public policy advice that will sway my vote), I would just as soon toss the anti-EMH theory away.  I have found the EMH to be very useful in all sorts of ways mentioned in previous posts.  The anti-EMH position?  Not so much.

(To forestall some objections based on a misunderstanding of my argument, let me reiterate that I am not saying the sub-prime crisis does not point to the need for tighter regulation, I am saying that any regulation that comes out of this crisis would have to be justified on grounds other than market inefficiency, unless you can convince me that future regulators will be able to predict markets better than future financiers.)



98 Responses to “Richard Rorty and the efficient markets debate”

  1. Gravatar of Bill Woolsey Bill Woolsey
    21. March 2009 at 07:10

    I don’t agree with the notion that there is no need to think about speculative bubbles unless there is some kind of policy relevance. Or, rather, what sort of education is appropriate for market participants is always “policy” relevant.

    We cannot all free ride on “the market.”

    I also think it is important that monetary instituions be robust in the face of speculative bubbles. I don’t know how they could stop them, but they shouldn’t break down when the bubble pops.

    The same is true of banks. Independent of monetary policy, having all the banks close when a bubble pops is a bad thing. To me, this suggests a need for rapid bankruptcy for banks. Or making sure that if many banks fail, the sound ones can expand.

    So why is bubble theory important:

    1. Improved education of market participants.
    2. Monetary instituions that are roubust in the face of bubbles.
    3. Banking institutions that are roubust in the face of bubbles.

    The notion that bubble theory is only useful if we can identify bubbles and make money off of them or else, can have regulators somehow prevent them, does not exhaust the possibilities.

  2. Gravatar of Patrick R. Sullivan Patrick R. Sullivan
    21. March 2009 at 10:07

    Even worse, the regulators created the subprime problem. They insisted the efficient home lending market was racially discriminatory and ‘reformed’ it. As I’ve linked to before (but, you’ve been getting new readers, here it is again):

    ‘This report concludes that, in an attempt to increase home ownership, particularly by minorities and the less affluent, virtually every branch of the government undertook an attack on underwriting standards starting in the early 1990s. Regulators, academic specialists, GSEs, and housing activists universally praised the decline in mortgage-underwriting standards as an “innovation” in mortgage lending. This weakening of underwriting standards succeeded in increasing home ownership and also the price of housing, helping to lead to a housing price bubble. The price bubble, along with relaxed lending standards, allowed speculators to purchase homes without putting their own money at risk.

    ‘The recent rise in foreclosures is not related empirically to the distinction between subprime and prime loans since both sustained the same percentage increase of foreclosures and at the same time. Nor is it consistent with the “nasty subprime lender” hypothesis currently considered to be the cause of the mortgage meltdown. Instead, the important factor is the distinction between adjustable-rate and fixed-rate mortgages. This evidence is consistent with speculators turning and running when housing prices stopped rising.’

  3. Gravatar of Milton Recht Milton Recht
    21. March 2009 at 10:22

    Regions with the greatest price appreciation during the “housing bubble” were the areas with the most home development. If the prices in the housing market were irrational, developers would have tried to make abnormally high profits. In addition, there would have been a rush to sell land to developers and share in some (all) of the abnormal profits.

    Material plus labor costs increases would not account for most of the appreciation of new home prices. Substitutes exist for many house components and material suppliers could increase production.

    However, land cost increases might have accounted for a large part of the appreciation. Landowners would have shared in the excess profits available to developers by raising the price of future home plots. Additionally, the marginal cost of land increases as development occurs. The land in an area that is the cheapest to build upon is used first. As development continues, the land remaining for development is the more costly, requires greater preparation and has more government regulatory approval uncertainty.

    I have nor seen anything about homebuilders making abnormal profits for their risks. I have also not seen anything about the developers’ land and plot preparation costs to show that there were abnormal profits to developers.

    Unless developers were showing extraordinarily high rates of returns on their investments, I would go with efficient markets as the most likely and not a housing bubble. I would guess that demand for undeveloped land with high marginal development costs and risks was responsible for most of the price appreciation called a housing bubble.

  4. Gravatar of Phil P Phil P
    21. March 2009 at 13:59

    Scott, as you know I’m not an economist, so I’m not going to argue the technicalities of the EMH hypothesis, which I don’t understand anyway. I only poke my nose into these discussions when I feel there’s a common sense point I can make that may have merit. Because I share your Great Depression obsession, my eyes perked up at your statement that stocks in 1929 were not grossly overpriced. I think they were in terms of PE ratios. (Of course absent the GD stock prices wouldn’t have fallen by 80% but that’s rather beside the point). Robert Shiller’s home page has a spreadsheet of historical stock price and earnings data which I’ve looked at. I notice that stock market booms always are preceded by several years of above normal real corporate earnings growth. Not a big surprise really. The mistake people make is in projecting these indefinitely into the future. Now I think in analyzing bubbles you emphasize too much ex ante rationality. I don’t think Irving Fisher was irrational in believing stocks were fairly valued in 1929. I think he lacked the data and historical perspective we have.

    If I may venture a more general observation, having read your blog for a while now, it strikes me that since you are a monetary policy specialist, you tend to view everything from a policy perspective. The main point of your post, I think, is that 1) bubbles are not really predictable ex ante and 2) therefore regulators should not try to prevent them. I actually agree with your second point, as a practical matter. But I’m more interested in history than policy, and in whether we can determine whether bubbles exist by some objective criteria, even if only ex post. In the U.S. over the last century and a half, real GDP per capita growth has averaged a little over 2% a year, real corporate earnings growth a little less (not surprisingly). Maybe someday there will be scientific advances of a magnitude to produce huge long-term increases in productivity that will raise growth rates over historic norms, otherwise we would have to expect that periods of abnormally high corporate earnings growth will produce reversion to the mean. Similarly, as to housing, the last decade was marked by price to median family income ratios way out of line with historical norms. People can only afford what they can afford. If ever there was a bubble that was both predictable and predicted, this was it. (I don’t know enough about the regional price anomalies you point to but it’s not necessarily inconsistent with the broader point).

    One final brief point about regulation. I would like to see the simplest and most market friendly form of financial regulation. To me that means adequate capital ratios and limits on leverage. I don’t know if that’s really controversial. Anyway I don’t see a difference in principal between preventing excessive monetary growth and preventing excessive leverage and credit growth.

  5. Gravatar of Bill Stepp Bill Stepp
    21. March 2009 at 16:27

    Stock market crashes don’t happen as often as would be predicted by EMT (with prices following a random walk) as they do in the real world (see Mandelbrot and Taleb). So why assume that stock prices changes are random and that their distribution is normal when the empirical evidence points the other way?

  6. Gravatar of Fred Fred
    21. March 2009 at 20:01

    To a moron, everything looks like a random walk. The moron notices that some people succeed in business while others don’t. He concludes it’s all a matter of luck, because he’s too stupid to that some of these businesses are well run and others are poorly run and that is what causes success/failure. In other words, the stupider a person is, the more efficient (random) the world appears. Understanding reveals the inefficiency.

  7. Gravatar of VG VG
    22. March 2009 at 01:32

    The problem here is that you are only arguing that markets are no less efficient than regulators or random numbers. But this is meaningless. Randomness is not efficiency. Markets are not supposed to be “as good as anything else”. They are supposed to be better. If they aren’t then there is no reason to prefer them to dirigisme or lotteries.

  8. Gravatar of Bill Woolsey Bill Woolsey
    22. March 2009 at 03:34

    I think it is De Long who has been emphasizing how in the less than a year, some huge fraction of the world’s wealth was destroyed. I believe that the EMH suggests that he is correct. Is it true?

  9. Gravatar of Matt Matt
    22. March 2009 at 03:42

    I have a question:

    Is it possible that inflation will never set in despite the large sums of money and projected budget deficits?

    I submit this question also as a position I take based upon the current fiat money system we operate under.

    If the fed pumps money into banks or other entities in need of an injection of capital, they continue to produce money from nothing.

    In this sense, if the money never enters the market then inflation will not have a chance to set in – or am I extremely misguided?

    Thought would be appreciated!



  10. Gravatar of mike mike
    22. March 2009 at 03:59

    I like a lot of what you have to say.

    But moral hazard is the fundamental market failure, in the pure technical economic sense, not any ‘moral’ sense (it’s a tragic name for an economic concept). The market failure is probably only partly irrational expectations. The idea is that incentives become perverted toward traders taking on too much risk when one earns all the upside of a gamble and none of the downside.

    Robert Shiller does a nice job of explaining who these two ideas work together in his earlier work.

  11. Gravatar of ssumner ssumner
    22. March 2009 at 05:19

    Bill, Here I define “policy” broadly as any practical implication, public or private. I agree that monetary institutions need to do a better job of dealing with market instability, whether is is caused by irrationality, or just mistakes by rational people.

    Patrick, Yes, I like that paper. I basically agree with you, with one caveat. Like with the Great Depression, I don’t like mono-causal explanations. Despite all the bad regulation I think there is no getting away from the fact that foolish decisions were made by some private actors.

    Milton, Yes, land prices are the key. We are not far apart, as I also don’t like the anti-EMH position. My hunch is that one big mistake speculators made in interior markets, is overestimating how dramatically zoning constraints were spreading into their area.

    Phil, I don’t agree that the historical average P/E is the right benchmark. That averages good times and bad. But in mid-1929 they had no idea the 40 year statist nightmare was on the way. Policy was probably as close to optimal as we’ll ever get. (Zero inflation, low unemployment, trade and budget surpluses, low and falling tax rates, etc.,etc,)
    Also, even anti-EMH models say stocks are usually GROSSLY undervalued. So maybe 1929 was the right level, and the valleys in the index are way too low.
    On your second point, I don’t think you quite got one of the distinctions I was making. If you look back at past bubbles, you have to decide whether it was irrationality or rational mistakes. That’s really hard to do, indeed I argued it would be impossible, without bringing in policy implications. This is a subtle point. But unless there are policy implications, there is really no useful lesson to be learned from studying history. And I say a sterile debate over whether something “looked irrational” is useless without at least some potential practical implications.
    I think your views on regulations sound reasonable. I would just add that we already had such regulations and they didn’t work. That’s not being sarcastic, I do think you are right that we should improve these regs, just showing you how hard it is to do.

    Bill, I don’t think the EMH assumes normality. Maybe originally, but surely not in the more sophisticated versions today. There is no logical connection between EMs and normality, its just a simplifying assumption.

    Fred, It is interesting that managed mutual funds on Wall Street are almost entirely run by people you call “morons.” I don’t deny that there may be a few non-morons in the world (such as Warren Buffet), I do deny that there are enough to make the anti-EMH position have practical value for morons like me.

    VG, The randomness is not my argument, it is my response to silly anti-EMH papers that claim EMH can be refuted by non-random patterns established through “statistical significance” tests. I agree that there is more than randomness implied, and I address that in my comments on bubbles and fundamentals.

    Bill#2, Yes, Delong is right. The random walk of stock indices imply ther eis no “bounceback, the losses are permanent. That’s so counterintuitive that even i can’t wrap my mind around the idea. Here’s my pathetic attempt to understand how a monetary mistake could do so much damage:

    1. Severe recession (Phillips Curve)
    2. Greatly worsens banking crisis
    3. Much higher future taxes from ineffective fiscal stimulus
    4. Bad statist policies that don’t address the root causes (GM bailout, caps on pay, bad regulations, etc.)

    Matt, Yes, it is possible (see Japan). It all depends on whether the Fed wakes up and gets serious about nominal targeting. Right now, we are much more likely to undershoot the 2% target.

    Mike, I agree about moral hazard. The Shiller point about combining the two problems is interesting, but at the end of the day it doesn’t change my pragmatic argument about the anti-EMH position lacking useful policy implications. The regulatory failure I described occurred in a world that should have been well aware of the moral hazard problem. If they do better with regulations in the future based on what we have learned about the interaction, great. But I think they will only do better by focusing on moral hazard, not trying to outguess bankers on what loans are likely to be profitable.

  12. Gravatar of Randy Randy
    22. March 2009 at 05:20

    As someone who spent their life in the investment management industry investing other people’s money, I have one comment on this blog. While it seems one needs recourse to the EMH to explain the superiority of index funds to the “average” manager this is not actually true. Index funds outperform on average because of their substantially lower costs with respect to fees and other frictional costs including trading. Why does this work? Because the average is the average is the average. So, on average the market return is the return of the average investor. But, the average investor has much higher costs in implementing their active strategy and so on average active portfolios must in fact underperform. Furthermore there is no reliable way for a superior investor to identify him or herself to a potential client so for the normal person buying mutual funds, their purchase amounts to a random draw from the pool of investment managers. The public policy issue here however, is that if everyone indexed, the strategy would not work. In the absence of active management there is no price discovery mechanism occurring and as price divergence from fair value occurs somebody will make an above-average profit (in this world of indexers only presumably the brokers, market makers or specialists) leading others to try to earn above average profits as well. Thus, it seems in equilibrium we will have both active and passive investors. But again, relevant to your post, is that the performance of index funds is not evidence per se of the efficacy of the EMH.

  13. Gravatar of Terry Jaillet Terry Jaillet
    22. March 2009 at 05:31

    Your implicit assumption that bankers and bank investors are both synonymous and have the same incentives is false. The fact that bankers incentives encouraged them to write more mortgages, regardless of the impact on the risk profile of their firm, is one of the main reasons for the housing bubble.

    What is interesting to me is the idea that people can all act rationally, in their own self-interest, you the market as a whole can end up in an irrational place because of misaligned incentives.

  14. Gravatar of odograph odograph
    22. March 2009 at 05:59

    I think markets are unpredictable, impossible to “beat” in the traditional meaning of that word. That doesn’t imply to me that the EMH must be true. That would be like saying that any other fact endorses any other single theory.

    We see ships’ masts drop as the disappear over the horizon, which makes sense if you consider that the world is draped on the back of a giant turtle.

    FWIW, I’m more in tune with behavioralist and chaotic reasons for the markets being unpredictable. For the most part I find common-cause people who use EMH as their central organizing principle … but not in one fairly important area: Now and then the market sends us clear signals. Now and then it isn’t tea-leave reading. Now and then we can no the future … say for instance when you hear that everybody in a housing market is selling interest-only loans to no-doc buyers.

    That might be a clear enough signal to break through, and make you doubt that real estate prices are reflecting real risks.

  15. Gravatar of odograph odograph
    22. March 2009 at 06:30

    (I probably overstated “[know] the future.” I think we can see when warning flags change from yellow to red. And maybe we can see when the flags are taken in again.)

  16. Gravatar of Mike Mike
    22. March 2009 at 06:37

    Great blog!

    How do you feel about the various theoretical “Noise Trader Risk” -> “Limits to Arbitrage” line of argument against the EMH? EMH requires flawlessly executing shorts against noise traders, which is very difficult (and involves taking uncompensated risks) in practice.

    As for the statistics: It is not just equities, there is also compelling statistical evidence that “Underreaction, Overreaction, and Increasing Misreaction to Information” occurs in the options market (see the research by Poteshman,2002, among many others). Without getting to into it on a blog comment, it is harder to argue that it is just a slightly tweaked valuation problem when it comes to options.

    As for the subprime mortgage mess, regulators got it wrong precisely because they believed that the EMH would cause the incentives of ‘shadow banks’ to line up correctly, even though there was a conflict of interest since they were initiating the risk but not holding all of it.

  17. Gravatar of silvermine silvermine
    22. March 2009 at 07:21

    This is my perspective from where I live, Silicon Valley.

    The 2000 bubble was all about Y2K. Every company in the country bought new computers, services, software and programmers. People thought for some insane reason that the level of purchasing would be permanently sustained. But of course not, they were doing a one time upgrade for the Y2K issue. Once they upgraded, no one needed new hardware or software for a year or two, because they already just upgraded everything! And demand fell off rapidly.

    Companies started to go down, and as they did, there was a new thing called eBay they used to sell off their assets, meaning that people who needed a new server/router/whatever didn’t need to buy a new one, still! They could go pick up cheap servers and routers on eBay.

    And there was a lot of exuberance over people thinking the unsustainable Y2K buying spree would continue. It was based on fact, but the problem was they made assumptions that were nonsensical given what as really going on. Then there was panic.

    Once people lost money in the dot com crash, they put the money into real estate, because “real estate never loses value”. I *heard* these people. The people who were paper millionaires and lost it all, took what they had and bought property. Any property. At any price — because they were sure that the value of the house would soar and make them money before the ARM/balloon payment was due. They were sure they would re-finance or sell by then, moving to a new place. (Or they were buying second, third, fourth houses. Whatever.)

    It was the same people, I worked with them. I told then real estate didn’t always go up. They said silicon valley was a special case. They didn’t believe me.

    Other people panicked that prices were going up so fast, that if they didn’t buy now, they would be priced out of the market forever. They would miss their chance. The people who needed their property value to go up just encouraged that thinking, of course.

    I heard the justifications people made for 0 down crazy ARMs.

  18. Gravatar of Gabriel Durazo Gabriel Durazo
    22. March 2009 at 08:04

    >Indeed the “uninteresting” string of numbers that I just listed, are actually very interesting””can you see how?

    Sigh, this is exactly the sort of challenge I love. Alas, I’ve spent a half hour or so puzzling over it and now have got to move on to other things. Is there some sort of interesting property you have in mind, or are you just trying to get us to find one?

  19. Gravatar of ssumner ssumner
    22. March 2009 at 08:29

    I will get to all the new commenters later today, but I notice that someone said they spent a lot of time on the 10 random numbers, so I felt guilty and wanted to make sure people weren’t wasting their time. I don’t find these numbers interesting–my argument is that this is the sort of pattern anti-EMH types find interesting. Nine of ten are odd, that means the hypothesis that the wheel is biased towards odd numbers is confirmed at 99% confidence (if I did my calculations right, which is unlikely since its been a while since I took probability–I assumed 10 divided by 2 to the 10th, is that right?)

    I just found out about the Mankiw link, so I was unaware there were so many new visitors.

  20. Gravatar of Don the libertarian Democrat Don the libertarian Democrat
    22. March 2009 at 08:43

    I want to throw some philosophy into the discussion, and, to do so, I’m going to rough the edges up a bit to keep from writing a treatise. Although I agree with Rorty about theories/positions being more or less useful/practical, I approach the world from a different position.

    If your read Rorty, you will notice that he spends a lot of time talking about Dewey, Quine, Sellars, Ryle, Davidson, and Dennett. In my mind, all these philosophers are behaviorists. As Rorty says:

    Rorty provides this view with a label: “Explaining rationality and epistemic authority by reference to what society lets us say, rather than the latter by the former, is the essence of what I shall call ‘epistemological behaviorism,’ an attitude common to Dewey and Wittgenstein.” (PMN 174)

    To me, “Word And Object” was the best place to see this expounded. Now, in economics, there seems to be a problem differentiating causal explanation from correlative explanation. I think that’s because, to the behaviorist, there’s no difference. There are simply correlations of behavior. Consequently, when tax cuts were passed last year, and didn’t work out as expected, behaviorists are stuck. The results are the results. That doesn’t mean that they don’t offer other explanations, but they offer non-behaviorist ones and fudge the issue a lot. This explains their fascination with biology, especially Darwin, and avoidance of math and physics. The latter two are a barrage of anomalies for the behaviorist. I would argue that confirmation bias is also a major problem for behaviorists.

    Is this a fair assessment? Probably not. I’m trying to be succinct, and I have a different view. But as I told Nick Rowe, behaviorism does seem to be the philosophy underlying much of modern economics. Anyway, since I approach economics through philosophy, but one centered on intentionality, I would be interested in how economists view their theories philosophically, if they do.

  21. Gravatar of Randy Randy
    22. March 2009 at 09:53

    Having reread the blog and some additional comments I would like to comment on this again from the practical perspective of the EMH as a practitioner, i.e., someone who invested other people’s money as a career. First, it is important to note that there are “forms” of the EMH normally identified as the weak form, the semi-strong form and the strong form. Nearly all academic debate and documented “anomalies” are taking on the strong form (basically, that current prices reflect all available information and there are no strategies that can consistently make a profit by betting against or current market values.) The weak form, that there exist no mechanical trading rules that can be used to make an excess profit is almost universally considered to hold, even by most active investors. So, it is important to realize that for most investment professionals they assume efficient markets as a relatively realistic first approximation of reality. Having said that, it is true that there are examples of divergence from easily calculable fair value. The example I most often use is the mis-pricing of the original S&P 500 futures contract back in the early 80’s. When first brought to market this contract consistently traded below fair value (quite easily calculated using cost-of-carry methods) and allowed investors to sell out of S&P 500 index funds and roll the contracts as a substitute and make significant excess profit. Basically, there was widespread arbitraging of the mis-pricing of the contract, and most of the large index fund providers offered variants of this product to their investors. This worked for a while but ultimately the arbitraging activity forced fair value convergence and the strategy no longer earned an excess return, though it took much longer than any economics or finance paper I have ever read would anticipate. I conclude from this that new instruments are often mis-priced due to institutional frictions and biases until the market can correctly value them. And remember this is a derivative, its correct value is actually not ambiguous in relation to the physical asset. The underlying asset may have high volatility in its market value but the derivative should and normally will track the underlying with perfect accuracy in the bounds of transaction cost frictions. So this would be in conflict with the strong form of the EMH. Why, because in calculus when you take a derivative and set it equal to zero to solve an optimization problem it is instantly solved in mathematics, but that same action in the human enterprise that is the market takes time to find the pathway to equilibrium pricing under certain hard to define in advance conditions but ones that can be seen when they occur as in the example above. I hope that this example is understood to be meaningfully different than say the Fama-French anomaly with respect to cap size and valuation characteristics, which is a statistical argument that has periods of market history where it sometime “works” and where it sometimes doesn’t. Which gets me to my final comment. I do not think it appropriate to deduce from the EMH that markets always have the “correct” value of an asset in some metaphysical sense of the term. It just means that there is no easily developed strategy that can be used to profit from the possibility that there has been a divergence from “correct” value. In other words, there exists no surefire way to know that we have diverged from correct value. Those are two pretty different things, can I develop a known-to-be profitable trading strategy or do I know that all assets are correctly priced in some Platonic truth sense at all times. One of the clear issues here of course is that most economic and finance models are single period and we live in a multi-period world. Maybe houses in Phoenix were correctly priced in 2006 for those time and conditions and are correctly priced today in 2009 for this time and condition. Who said prices are stationary. All the assets I have spent my life investing in are absolutely volatile assets with distributions of return and non-stationary covariance matrices to boot. It is beyond me how you could have even done a rudimentary look at the residential housing market and thought it always went up. Ever been to California or Texas in the 80’s? Anyway, to summarize an overly long and wordy post. The EMH in its strong form almost certainly is not true, there exist too many true arbitrage opportunities, particularly with newly created instruments, that persist for too long for that to be a reasonable first approximation. The weak form almost certainly is a reasonable first approximation but that does not mean that the assets are priced correctly, whatever that might mean. I tend to be in agreement with DeLong and some others on the fact that we have massive shifts in collective risk aversion parameters that have a huge impact on asset prices both positive (bubbles?) and negative (busts?)
    But really, who knows. But on the policy front I would note that we have known for a long time that the basic business of banking is volatile and risky, that bankers tend to over leverage in good times and hoard in bad, and that because of that we do (or used to) sensible things like limit their leverage and not ignore obvious agency problems such as originating mortgages but not having to bear the consequences of bad credit decisions. Just some reflections from an in the trenches practitioner, (now basically retired and free to blog.)

  22. Gravatar of STLADAMS STLADAMS
    22. March 2009 at 10:22

    The EMH is appealing to me as a scientist. I don’t think that an apparent anomaly that is not immediately explainable does any damage since there is always some lack of detail in our understanding. As I look at past events, and the current one, the problems seems to be tied to the assumption of our knowledge of the facts. Many of the issues are due to a lack of correct information or assumptions about the facts. As our decision making gets more complex, such as the quant models many firms use, the errors in our inputs become more critical. The AAA ratings on the loan packages that many banks bought led them to the wrong conclusions.
    The misrepresentation of financial details and other attempts to obfuscate the truth can lead many rational investors to make the wrong decision.

  23. Gravatar of Philo Philo
    22. March 2009 at 10:44

    Much of your philosophizing seems to have little to do with Rorty. You write: “To take another example, imagine two people looking at Mt. Monadnock. One says ‘That’s a very small mountain’. The other says ‘No, that’s just a big hill’. Pointless debate, isn’t it?’ I see no grounds
    for denying that there is a fact of the matter. It is clear that Mt. Everest falls under the concept MOUNTAIN, not the concept HILL; contrariwise for Bunker Hill–hence the names. In some intermediate cases, such as Monadnock (but note the name!), we are aware of the vagueness in our thinking, and it may not be worth bothering trying to clear this up. But lots of things that could be done aren’t worth doing; we didn’t need Rorty to bring this to light.

    You began: “Thus Rorty suggested that it was pointless to argue about whether something is an objective fact or a justified belief, as we have no access to an extra-human perspective, and thus can never resolve the debate.” Now, we don’t need “access to an extra-human perspective”; all
    we need is an understanding of the concepts OBJECTIVE and JUSTIFIED, as well as of the fact and the belief in question. And (since Rorty raises the issue) do we utterly lack “access to an extra-human perspective”? I’m willing to grant that people who think God (a non-human) communicates with them, or that (non-human) extra-terrestrials do so, are deluded. But to some extent we can read the “body language” of lower animals; we have some idea about their perspectives on things. And, while regretting my thus limited access to these extra-human perspectives, I rejoice in my access to the perspectives of a great variety of human beings, many
    of whom are very different from me. Why isn’t Rorty more favorably impressed by that?

    You conclude: “Now imagine two economists look at the tech bubble. One says ‘boy, those rational investors made a big mistake’, whereas the other says ‘no, the investors were irrational’. Another pointless debate.” Not so; in many cases this debate would have legal implications–e.g., if an
    investor had some fiduciary responsibility. And why think our grasp of the concept RATIONAL is so weak that we can’t resolve the debate rather easily? Admittedly, if it’s too hard to get the answer, it may not be worth doing. But the answer to almost any question is worth *something* (positive); as Samuel Johnson said: “If it rained knowledge, I would hold out my hand.”

    Of course, one reads your blog for the illuminating discussions of monetary economics, not for warmed-over servings of bad philosophizing.

  24. Gravatar of Gabriel Durazo Gabriel Durazo
    22. March 2009 at 10:49

    > I don’t find these numbers interesting-my argument is that this is the sort of pattern anti-EMH types find interesting. Nine of ten are odd,

    Oh, I see, I misunderstood then. Thanks for clearing that up. I had thought maybe there was some sort of math pattern that they fit to or something, like if you had given the numbers 1,1,2,3,5,8, etc. (fibonacci).

    >(if I did my calculations right, which is unlikely since its been a while since I took probability-I assumed 10 divided by 2 to the 10th, is that right?)

    10/2^10 is correct for the question, “What is the probability that exactly 9 of the 10 next random numbers will be odd?” I’m not sure if that’s equivalent to “that means the hypothesis that the wheel is biased towards odd numbers is confirmed at 99% confidence”, though.

  25. Gravatar of ssumner ssumner
    22. March 2009 at 12:16

    Randy, Why do managed funds have to under-perform? Suppose the market was 40% managed, 20% index and 40% individual. Suppose individuals earned 6%, indexed earned 8% and managed earned 10% (minus say a 1% expense ratio for a net of 9%.) I don’t see why that isn’t possible, indeed isn’t that pretty much what managed funds claim? Otherwise, how could they look their investors straight in the face? Now you might say that in practice it would be hard for them to outperform the individuals by enough to cover the expense ratio, which is a sort of unfair disadvantage. But that’s exactly what I was driving at by my “practical implications” argument–they don’t do well enough to compensate investors who have to pay for the opportunity cost of their time. But this raises another argument against me–much of the research done by them becomes a public good, so perhaps society should subsidize Wall Street professionals. In practice I don’t think we need to do that, as there is another “market failure” insuring we’ll have plenty of people trying to outwit the market despite the truth of the EMH. That market failure is called “human overconfidence.” BTW, both your comments were good, despite my reply.

    Terry, Even I knew that banks were lending to anyone who could fog a mirror in 2006 (and I don’t following housing closely)–how come these innocent “bank investors” didn’t know? Many of the bank investors are very sharp Wall Street pros who lost a fortune. Or did they know, and approve of what the banks were doing? If they didn’t approve, why didn’t they sell their stocks in 2006?

    Odograph, Your view may be right, but has no obvious policy implication, until you added the yellow/red flag at the end. But if the case you cite was as obvious as you claim, why didn’t smart investors short bank stocks in 2007. A few did, but most did not. Maybe it wasn’t obvious.

    Mike, I don’t believe anomaly studies because they misuse statistical significance. A t-stat=2.0 means nothing, although many economists seems to think it indicates “significance.”
    Everyone hears these stories about banks that initiated bad loans and sold them to gullible investors. The only problem is that when banks saw how much money gullible investors were making they bought lots back for themselves. There’s a reason so many banks are doing poorly.

    Silvermine, Both factors may have contributed, as you indicate. The tough issue is in deciding whether to call it “irrationality” or just a “mistake” by rational people.

    Gabriel, Sorry I wasted your time. (see my earlier reply.)

    Don, I have an earlier post on Rorty where I argue that economists use an eclectic methodology. But correlation/prediction is very important.

    Randy#2, Lots of good ideas–you may be right about the strong form of the EMH. I am actually fairly agnostic about the whole EMH. It proved very useful in my Depression research, but I used a weak version almost nobody would strongly object to. What triggered this post is that I realized the anti-EMH had no implications for me as a investor or voter.

    STLADAMS, I agree. People tend to underestimate how hard it is to figure things out. I believe I have as good an understanding of the monetary policy failure as anyone. But I didn’t predict it because I thought the Fed would act differently (more aggressively) in a crisis.

    Philo, I am certainly a bad philosopher, but not quite as bad as you assume. I totally agree with you about Mt. Everest. I specifically picked a borderline case to create a debate that had no useful implications. I would not have used Everest. I was trying to draw an analogy to make it easier for readers to understand what Rorty meant by “no useful implications.” I don’t mean to suggest the issues were identical to those in Rorty’s example. I don’t find your animal example very helpful. Yes, it is a non-human perspective (except that it is still our interpretation) but the God example is what Rorty had in mind–someone who could resolve basic ontological questions that are beyond our understanding. And why assume that Rorty is not impressed by variety?
    Also, your fiduciary responsibility example doesn’t contradict anything I wrote. I didn’t argue that a debate over the EMH was useless IN PRINCIPLE, I argued that there is no evidence out there that it is useful. If regulators could see market errors better than markets, then yes there would be practical implications. But I don’t see evidence of that, and I also don’t see evidence for your fiduciary point. Indeed it is equivalent to the regulation debate. Because if those with fiduciary responsibilities knew where the markets were going, I assume regulators could as well. But there is no evidence (from the abysmal response of regulators, and the massive losses of banking insiders), that either knew what was going to happen.

  26. Gravatar of Steve Sailer Steve Sailer
    22. March 2009 at 14:10

    “The housing bubble in 2004-2006 was partly driven by rapid immigration from Latin America (as was the bubble in Spain itself!), and also by a perception (which turned out false) that coastal zoning constraints were spreading into interior markets. Many Hispanic immigrants were snapping up older ranch houses, allowing native born Americans to move on to bigger McMansions. The immigration crackdown in 2007 dramatically slowed this immigration (as did the worsening economy.) ”

    Actually, it was more that the people who were getting loans in 2005 to the middle of 2007 didn’t have the earning capacity to ever pay off the mortgages. The whole Housing Bubble was a fraud made possible by things like George W. Bush’s argument that his goal of 5.5 million more minority homeowners by 2010 meant that down payments and documentation of income shouldn’t be required — the same “reforms” that Angelo Mozilo of Countrywide was calling for. The percentage of first time buyers in California who put no money down went from 7% in 1999 to 41% in 2006.

    There was never a hope that the new population of California could pay off half million dollar mortgages. Look at California’s National Assessment of Educational Progress school test scores. California barely ranks above Mississippi these days in average student achievement because of decades of illegal immigration. How are people who struggled to get through high school supposed to pay off $500k zero down mortgages?

    And they couldn’t find a greater fool to flip them to because who wants to pay more than $500k to live in a barrio?

    The worse default rates are seen in Southern California are in exurbs where working class people, white, Hispanic, black, Filipino, etc., bought too much house in a desperate struggle to get their kids into a school district not overrun by the kids of the illegal alien construction workers building the houses. Talk about circular reasoning!

  27. Gravatar of Steve Sailer Steve Sailer
    22. March 2009 at 14:14

    If you are looking for policy implications, one is obvious: political correctness makes stupid bubbles like the one in housing more common. Government policies in the name of giving minorities their fair share of the American Dream to both encourage and stop discouraging loans to people unlikely to pay them back were stupid. Moreover, nobody could say: “Why are we investing in all these half million dollars mortgages in California? Isn’t California full of Mexicans? How can a bunch of Mexicans pay back $4,000 per month after the teaser rate runs out?” If you put that in an email to your colleagues, it would show up in discovery of an anti-discrimination lawsuit, you’d get fired, and the CEO would have to promise La Raza that the bank would make even more stupid loans to members of La Raza.

  28. Gravatar of ssumner ssumner
    22. March 2009 at 14:36

    Steve, The whole PC push to get more people into houses (which came from both the left and from Bush as you noted) can only explain a part of the bubble, and probably a small part. If I’m not mistaken banks could have gotten by with far fewer subprime loans, and still pleased the various regulators and special interest groups that they had to placate. (Indeed, some bought back mortgage bonds in 2006-07 as “good investments.”)
    I’m not saying that your argument is wrong, in retrospect it does seem foolish to think all those working class families would be able to pay off $500,000 mortgages. Of course the banks assumed rising prices, in which case the mortgages would not have been defaulted on even if they couldn’t meet the monthly payments. But that just pushes the question back one step farther. I tried to indicate in my post that even I find the housing bubble hard to reconcile with the EMH. I don’t see any factors (including yours and mine) that can explain the scale of the fiasco. So I am not trying to suggest I have “the answer”, just that there might have been some factors that people overlooked.
    BTW; Spain had the same fiasco, and I doubt there are PC regulations in Spain forcing loans to “Hispanics,” if that term even has meaning in Spain. So there is a big worldwide mystery here.

  29. Gravatar of Steve Sailer Steve Sailer
    22. March 2009 at 14:45

    My argument is more subtle and far-reaching than that. It’s not just that the government held a PC gun to financial institutions heads and made them make bad loans. Among other things, irresponsible financial institutions wore a PC shield against regulators. Angelo Mozilo of Countrywide positioned himself not as the sleazeball boiler room operator he was, but as the leader of the war against racist redlining that was preventing minorities from getting their fair share of the American Dream. Here are excerpts from Mozilo’s prestigious Harvard address of 2003:

    “As President Bush said last October: “Two thirds of all Americans own their homes, yet we have a problem here in America because fewer than half of the Hispanics and half of the African Americans own their home. That’s a homeownership gap. It’s a gap that we’ve got to work together to close for the good of our Country, for the sake of a more hopeful future. We’ve got to work to knock down the barriers…”

    “While the number of minority homeowners has advanced recently, climbing from 9.5 million in 1994 to 13.3 million in 2001 – an increase of 40 percent – the fact remains that it is still not at a level equal to that of white homeownership. And as President Bush pointed out, the homeownership rate for African Americans is 47 percent and for Hispanic Americans it is 48 percent, a stark contrast to the homeownership rate of 75 percent for white American households. That means there is currently a homeownership gap of over 25 points when comparing white households with African Americans and Hispanics. My friends, that gap is obviously far too wide. It has been far too wide for far too long. And when adding new factors into the equation – like an influx of new immigrants or continued reduction in the supply of affordable housing – it has the potential to become far worse. …

    “One of the more obvious resolutions to the Money Gap is the elimination of down payment requirements for low-income and minority borrowers. Current down payment requirements of 10 percent or less add absolutely no value to the quality of the loan. [Down payments dropped precipitously. How’d that work out for us?]

    “Equally important, we must reduce the documentation required to make any and all loans; we should be able to approve loans in minutes, rather than days, and close loans in days, rather than weeks.” [i.e., liar loans, NINJA loans, etc.]

  30. Gravatar of Steve Sailer Steve Sailer
    22. March 2009 at 14:49

    Pundits like to blame “greed,” but the greedy we shall always have with us. What was missing in the Sand State housing bubble was fear. Where was the fear that rapidly Hispanicizing populations wouldn’t be able to pay back these huge mortgages? Where was the fear that people would stop wanting to move their families to school districts with abysmal test scores?

    Well, you weren’t supposed to talk about things like that in public. Or in private, either — your emails could end up as evidence in an anti-discrimination suit. We’ve propagandized each other for decades about how wonderful diversity is, so nobody would talk about how the increasing diversity of California, Arizona, Nevada, and Florida made them bad bets.

  31. Gravatar of Steve Sailer Steve Sailer
    22. March 2009 at 15:02

    So, mortgage dollars going to Hispanics to buy houses increased by 691% from 1999 to 2006, according to the federal Home Mortgage Disclosure act database. It almost octupled!

    Everybody who was anybody in America — Bush, Clinton, Obama, Mozilo, Barney Frank, Washington Mutual, the Greenlining Institute — thought that was just peachy.

    Well, it didn’t work out so well. 87% of the Housing Bubble (in terms of home value appreciation from 2000 to 2007) took place in the four Sand States of CA, AZ, NV, and FL. And that’s where about 7/8ths of the defaults have been.

    It would have been better if we had just had a blatant quota system just for minorities, because what we really did was cut credit standards for everybody, which made the Bubble bigger and the Crash worse.

  32. Gravatar of Steve Sailer Steve Sailer
    22. March 2009 at 15:13

    The Efficient Market Hypothesis assumes that, with the exception of inside information, everybody in the market has more or less the same information. What’s missing from EMH is the whole question of how people interpret that information. Some people interpret it correctly, other people interpret it incorrectly.

    Everybody knew that due to immigration, the quantity of people in the four Sand States was going up. What the market chose to overlook is that the quality of new buyers in those four states was going down because the immigrants to those states, many of them illegal, were, on average, below the old residents’ average in terms of earning capacity, credit-worthiness, education, relatives’ net worth, children’s test scores, children’s law-abidingness and the like — all the things that contribute to the two crucial questions of whether they will be able to pay off their huge mortgages or be able to flip them to others who want to move into their neigborhoods.

    Why didn’t the market correctly interpret this information that was publicly available about Hispanics having higher default rates, lower test scores, and so forth and so on.

    Because those are “hatestats.” Only evil racist people know statistics about things like that.

  33. Gravatar of Bailout and Financial Crisis News » Sumner on Financial Regulation Bailout and Financial Crisis News » Sumner on Financial Regulation
    22. March 2009 at 15:50

    […] are common in the blogosphere, but if you have the patience, they are well worth the time. In a recent post defending the efficient-markets hypothesis (EMH), he includes this […]

  34. Gravatar of TGGP TGGP
    22. March 2009 at 17:09

    You might be interested in Charles Davi’s first blogpost, The Not So Efficient Market (Theorem) Hypothesis.

  35. Gravatar of StatsGuy StatsGuy
    22. March 2009 at 17:16

    First, some small nitty points:

    1) The roulette wheel is a poor example, and extremely misleading.

    Statistical tests are not so crude as to limit themselves to exact combinations. They do not ask “what is the probability that this _exact_ event occurred?” They ask “What is the probability that an event this drastic, OR EVEN MORE DRASTIC, occurred by chance?” If the likelihood is small, then the simple chance is a poor explanation, and we conclude some causal effect is likely.

    2) 1987 – the essence of the Black-Scholes argument for the 87 crash is that certain types of trading strategies were wired into a large number of computers that all operated identically. Broadly speaking, this is an examply of endogeneity introduced by anticipatory planning (more on this later with regards to EMH). In the narrow circumstance of 87, the simultaneity of the computer programs caused trades to be executed in line with a particular plan – when so many market participants react simultaneously, they overwhelm market liquidity. Markets that otherwise appear “deep” become very shallow, and highly volatile.

    If even 5% of all market participants _simultaneously_ hit the sell button on all shares, what would happen to the price? To receive all of these shares, you would need that many people with buy orders – and buy orders become scarce when the market is in freefall because often people don’t even know _why_ it’s in freefall, but everyone presumes there must be _some_ reason.

    In this circumstance, the _individual_ programs were not “dumb”. Nor were programmers. However, they did not anticipate that their own actions could affect the market. This is what I mean by endogeneity…

    3) Now onto EMH – The list of reasons for inefficient markets is long and well documented. However, in highly liquid, low transaction cost markets (e.g. the stock market), we can eliminate a few common ones – incomplete contracting (the contract is complete – buy/sell), transaction costs (7.95 a trade is pretty cheap), moral hazard (bad incentives), externalities (for example, pollution), adverse selection (if anything, the stock market should select good traders not bad by killing off bad traders).

    That still leaves a couple big ones – asymetric information is one, which presumably should be addressed by such institutions as the SEC. In practice, inside information is THE lifeblood of Wall Street. It’s the reason everyone is in NYC. It’s a massive source of profits for the investment banks (or at least, was). For example, knowing that a big state run retirement fund has started buying into a stock means the bank can buy up the stock first and cut themselves some margin. (State run firms don’t start and then stop – they buy significant stakes, so if you see early buying you have very rich information.)

    The crippling of the SEC under the last administration certainly contributed to the everything-goes attitude on the Street. This can contribute to wide distrust of wall street for fear that the market is a rigged game – which should (one imagines) lead to underpricing of assets.

    Normally this may not be an issue, but if you focus on RETAIL investors in the past six months, there is a widepsread belief that they were taken to the cleaners by the insiders. That the insiders said “hold hold” even as they sold. One can point to the losses by hedge funds (in particular, som eleveraged ones), but average hedge fund losses are MUCH SMALLER than average mutual fund losses.

  36. Gravatar of StatsGuy StatsGuy
    22. March 2009 at 17:23

    The problem with Endogeneity:

    OK, those were the nitty points (above). Now some serious points:

    I’m not going to engage the cognitive/behavioral literatue addressing EMH – the irrationality of human cognitive processing is massively well documented. I do not know how well that has been linked to actual market functioning (although common wisdom and the observation of panic selling and ridiculous volatility levels in the last 6 months suggests there is a clear link).

    I would rather focus on something that has received less attention – endogeneity. This is the notion that markets create their own value. For example, when the markets store a significant amount of global wealth, a sharp drop in markets (for whatever reason – ration, irrational, or just randomness) can translate into an immediate downward wealth shock. This reduces spending, and can precipitate a recession all by itself (some argue this happened in ’87), which thereby reduces the value of stocks.

    Presumably, there are some negative feedback pressures (negative feedback is good in systems dynamics because it’s stabilizing). This would cause a shock to eventually “settle”, without causing a self-reinforcing downward spiral.

    An alternative (cyclical/cognitive) argument focuses on “overshooting” and “momentum”.

    For example, once a recession sets in, it can be hard to reverse until assets become so cheap that cash-holders pour in, and the market “momentum” reverses. Some may argue the markets then signal the recovery, and indeed “create” the recovery. (e.g. the ’83-84 recovery has been held up as an example of this)

    Interestingly, this dynamic can be broken by awareness of the dynamic.

    Likewise, EMH can be destroyed by _excessive_ faith in EMH.

    For example, consider the rise of ETFs (exchange traded funds that are predicated on past data showing that most mutual funds underperform against a weighted market index) which have been upheld as free-riding. If ETFs grow large enough relative to overall liquidity, they can create market distortions. Consider the limit – everyone invests with an ETF. Who supplies the information to the market?

    The counter argument is that, at the margin, we will see some opportunity for economic profits and eventually directed investing will achieve an equillibrium with ETFs. Certainly, in the long run everything hits an equillibrium.

    Consider another example – rebalancing and hedging. From the 1960s to the 1980s, we observed inverse correlations between many asset classes, which caused hedgers to recommend diversification across assets classes (stocks, bonds, gold, real estate, etc.). However, from the 1990s onward we start to observe this inverse correlation vanish, and become very high positive correlation. Why? Because so many people followed classic diversification/hedging strategies that it destroyed the inverse correlation that predicated those strategies. This high positive correlation between asset classes means reinforces systemic instability. A loss in one region of the globe shifts into another, wealth globally decreases, and the entire system can shift into a negative-spiral equillibrium.

    The problem – broadly speaking – is endogeneity between markets and market expectations.

    One of the great fallacies of EMH (other than the cognitive arguments which are so adequately covered) is that markets actually _have_ a natural steady state that is… well, steady. However if the equillibrium point is itself a function of what people _think_ the equillibrium point is, and if market dynamics are massively impacted by investment strategies that are predicated on past market dynamics, then we have a serious serious problem with system instability.

    Well, friends, we have a serious problem.

  37. Gravatar of Steve Sailer Steve Sailer
    22. March 2009 at 18:43

    “In practice, inside information is THE lifeblood of Wall Street. It’s the reason everyone is in NYC. ”

    Right. That’s one reason it was so silly for Icelanders to think that they could play high stakes finance for a living: a quarter of a million people out in the middle of the ocean, with few relatives or friends in key positions in New York or London to feed them inside info, and no big government to bail them out when they took too much risk. The Icelanders figured that because they were as good as anybody at whipping together a spreadsheet, that they could play in the big leagues. They were wrong.

  38. Gravatar of Patrick C Patrick C
    22. March 2009 at 22:54

    It seems like EMH people like to talk about incentives as well. I thought the idea with regulation is not that regulators can see problems any better, but that the incentives of bankers/executives are to maximize personal profit, perhaps shortsightedly, while the incentives of regulators are ideally, aligned with the long term health of the market and the nation in general.

    If anything, it seems to me that regulators performed poorly because their incentives were gradually perverted over time. Regulators were too often offered high paying jobs in finance after their careers in government and this made their incentives nearly identical to those of individual business-people. In effect there were *NO* regulators, because our regulators were made to act with the short term interests of business.

    If we correct the regulator incentive structure, it seems likely that we’ll have less regulatory failure. I certainly think that we would be tossing out the baby with the bathwater if we were to treat the most recent failure of regulators as a failure of all regulation.

    An obvious market failure we see today comes from what Delong calls “tunneling”. If a firm is likely to fail, then the firm’s employees are incentivized to extract value from the firm at shareholder expense. This seems like something that is clear and predictable and something that we can work to prevent by effective regulation.

    Maintaining a sufficient number of small firms in a market is also a useful purpose of regulators. Microeconomic theory only predicts efficiency when there are a large enough firms in a market. Also, if there even exists an industry dominated by one super efficient firm, all people would suffer if that firm failed even for reasons outside of its control. Firms will move to fill that gap, but it will not be immediate and in the interim there will be a lot of lost productivity and suffering.(see AIG) Finally, is there a reasonable alternative to having regulators at minimum stop things like fraud? Maybe they don’t always work, but they still often work.

    It seems like the author’s argument follows a form somewhat like this:
    a) 4 police were recently killed in Oakland.
    b) These officers failed spectacularly to prevent crime.
    c) Thus police cannot prevent crime any better than ordinary citizens.
    d) We should therefore abolish the police force.

    Perhaps someone can explain how someone can honestly make a case against at least *some* regulation. To me, the question is only, “how much”?

  39. Gravatar of ssendam ssendam
    22. March 2009 at 23:40

    First of all, isn’t money the original bubble (i.e. something that is priced out of line with fundamentals)? Monetary models must resort to tricks like cash-in-advance or MIUF because without them, money would have zero value. Or to put it another way, why is the price of gold about $1000 per ounce, when gold’s fundamental value is almost nothing? Why are goldbugs buying coins and burying them in their backyard, when it contributes nothing to their instantaneous utility in any period?

    Second of all, you ask for practical implications of an anti-EMH position. Here’s one: if you believe in the EMH, then there’s no particular need to understand how the financial markets work, or what the banks are doing, or where their assets come from – all you have to do is look at the current price. Yet that is exactly the attitude that got us into trouble. You mention the Fed, the media, Roubini, etc – well which one actually knew something about subprime mortgages, CDOs, CDS, and the whole alphabet soup? The rest were completely clueless (as are most cutting-edge macroeconomists). Or do you believe (like Lucas said a few years ago) that the the details of the financial system don’t really matter?

  40. Gravatar of Jeff V Jeff V
    23. March 2009 at 02:42

    Good morning,

    How do you define the EMH and Anti-EMH positions? I ask because there does not seem to be a consistent definition of EMH; frequently, it is just a vague statement such as “all information is priced into the market.”

    The anti-EMH, in my view, basically says that psychology is important; people aren’t hyper-rational and motivated entirely by greed, contrary to what is taught in graduate economics.

    It is hard to follow the arguments without consistent definitions.

    – Jeff V

  41. Gravatar of ssumner ssumner
    23. March 2009 at 05:24

    Well, I said the EMH was very counter-intuitive, and you are all certainly going with the intuitive. But here’s my response:

    Steve, People like Mozillo don’t really explain what went wrong with the financial system. You are looking at the froth on top of the wave. When big investment banks managed by “the smartest guys on the planet” from Ivy league schools were pouring $100s of billions into mortgage bonds, it wasn’t to please the PC crowd. Again, some of what you describe happened, but it doesn’t even come close the addressing the EMH policy implications I discuss in my post.

    TGGP, Thanks for the link. I glanced at it, and will look more closely later.

    Statsguy, I strongly disagree on several points:

    1. The roulette wheel absolutely is a very relevant analogy. There are two kinds of anti-EMH studies. The hunt for anomalies (interesting patterns), and bubbles, or prices out of whack with fundamentals. This reflects the EMH positions that prices should reflect (estimated) fundamental, and should immediately incorporate all available information (hence no predictable patterns.) The wheel referred to the no patterns studies, such as the correlation between rainy days and stock prices, or the January effect, or a million other similar “studies.” These are very different from the “drastic events” issue you discuss, which is the bubbles I address later in my post. The studies I refer to are all basically worthless because they misuse classical statistical significance. The original idea of significance was that you developed a theory without reference to the data (say using logic, or pre-existing theory borrowed from a neighboring field.) Even if researchers did things that (honest) way, the results would be dubious because of publications bias—the tendency for journals to only publish “positive” results. But it is much, much worse. Reseachers don’t work this way. They look at the data and INFER hypotheses from the data. Then they test those hypotheses WITH THE VERY SAME DATA. Believe me, if you gave me a spread sheet of all the gambling machine outcomes in Vegas, I could find just as many “anomalies” as the finance professors find on Wall Street. Yes, you wouldn’t find Black Swan events, but those are consistent with EMH if the underlying fundamental shocks are non-normal.

    2. If computers were all wired to sell after sharp price declines, then that supports my really stupid computer programmer argument. Because such behavior is not only an example of technical analysis (which is basically astrology), but it is kind of a silly one. If you thought the stocks were a good buy at 2200, why would you want to sell a few hours later at 1800? And why would you program a computer to do this? Maybe they made this mistake–I’m not arguing that people are always rational. But if they made it only once, what are the policy implications?

    3. I agree that inside information is everywhere (indeed the Fed has inside info when it engages in QE.) Making insider trading illegal (except where contractual fiduciary relationships are violated) doesn’t make much sense to me–but I’m sure lots of you will disagree.

    4. If hedge funds are so much better than managed funds, wouldn’t managed fund operators who wanted to build a good reputation just invest their money in hedge funds? Maybe this is not allowed–but I thought I read that some managed funds had invested in Madoff (which isn’t exactly a hedge fund, but is similar.) Someone fill me in on the laws here–are hedge funds off limits to mutual funds? If so, then the law should be changed–and it’s not inefficient markets, it’s inefficient laws.

    Statsguy#2 You are totally wrong about 1987, it was not followed by a recession, it was followed by a huge economic boom. You may wish to revise your theory that the wealth effect of stock crashes causes recessions. You say there are some theories of the 1987 crash. They must not be very useful theories, because the October 19, 1987 event has only happened once in U.S. history. The Dow fell over 22% that day. How big was the next biggest drop in history? Only 13%. So if this is a good theory, where are the other super crashes (say more that 15% or 20%) that it should be able to predict?

    I don’t think recessions are hard to reverse, because they are not caused by falling wealth, (as the 1987 example clearly shows) They can be quickly reversed if enough money is pumped into the economy to create 5% expected nominal GDP growth.

    Regarding psychological studies showing irrationality, those don’t have anything to say about the policy implications of the anti-EMH position. Nobody argues that people are always Spock-like rational beings–the issue is whether their irrationalities allow us to spot non-fundamental asset prices. I say there is no evidence.

    Steve, Yes, Icelanders were wrong, just as all the sophisticated European investors who bought shares of the Icelandic banks were equally wrong. But I’m not sure what this says about EMH. That hypothesis certainly allows people to be wrong, it simply argues that they are not obviously wrong. If they were, why would all of these sophisticated investors have bought shares in their banks? (Or why wouldn’t others have shorted the shares.) And the crux of my argument is that regulators knew no better than markets–hence no policy implications. I focused on the U.S., but the equally big errors in Europe (which cannot be blamed on Bush’s (falsely assumed) anti-regulation ideology), shows that there is no easy regulation solution to market bubbles. Yes, we should have sound public policies–I totally agree. But even if we do, markets will make mistakes and regulators will continue to fail to spot those mistakes. That’s my key point.

    Patrick, Patrick, even if you are right, no policy implications flow from the subprime crash. Remember that “regulators” occur at all levels, and were not all bought off. Barney Frank (my Congressman) is a sort of regulator. He probably doesn’t plan to have a cushy job in the banking industry later. What about the media and academics? And if you argue I am naive, and that corruption runs all through the system, then that would make your “correct the regulatory incentive structure” hopelessly utopian. I think it would be much easier to correct the bank manager incentive structure. They are best placed to know which loans are likely to be good. If there is a problem in the market for corporate control–fix that problem. (Here I am referring to your “tunneling” argument.)

    It is not obvious that having many smaller bank firms would be better, although I am open to persuasion. In the early 1930s we had many thousands of banks fail. Canada had zero (even though they also experienceda Great Depression.) The Canadian system was dominated by 10 big nationwide banks. You could actually make a good argument that our decentralized banking system was one of the root causes of the Great Depression. Now I realize that you aren’t asking for that many little banks, and your argument may have merit, but this example shows how unpredictable events can be. We fix the last crisis, and then are surprised by a different next crisis. The very last thing I am worried about is a repeat of the subprime crisis–banks have learned a painful lesson. They’ll screw up again, but in a different and totally unexpected way.

    2. Who said there shouldn’t be “some” regulation. Not me. Reread the end of my post. Actually banking has always been highly regulated and the regulation got even tighter under Bush. Since 1934 the liabilities of the banking system have been (de facto) mostly loans from the government. We are not even close to a free market.

    The police example is not relevant. I am not making claims, I am responding to the claims of the anti-EMH crowd. It is they that drew the connections between police and crime (to take your analogy.) I am showing that the connections they drew have no policy implications. The specific example I used was the 2006 housing bust–it’s the one they point to. Yes, the anti-EMH policy might work in other cases. Show me one and I’ll shoot it down as well.

    ssendam, No trick is necessary. Economic theorists with their abstract models think there is some money mystery to explain, but there is not. Tokens with a fixed nominal price are very useful in transactions, and are also a good anonymous store of value.

    In your second point you said you were going to provide me with a policy implication, and then just provided another theory, the theory that people don’t spend enough time researching for themselves, but just rely on market data. What is the implication for policy? Should we subsidize the activity of searching for data? Should government workers do it? If you make that argument, I will point out that the vast majority of people are overconfident about there abilities (and hence don’t believe EMH–despite the fact that it is taught in schools) and thus we have no shortage of experts on Wall Street trying to outwit the markets. Yes, in theory what you mention might be a problem, in practice it is not.

    As far as who knew about subprime mortgages, even I knew, and I know almost nothing about the housing industry. So if I knew . . . The causes of our current crisis are 75% monetary policy, 20% subprime mortgages, and 5% (at most) the other alphabet soup you mention. So I wouldn’t blame Lucas for not knowing finance, I’d blame him for not understanding the need for a forward-looking monetary policy.

    Jeff V, The existence of irrationality is not the anti-EMH position, it is their explanation for their position. The actual anti-EMH position is that at times asset price OBVIOUSLY diverge from fundamentals in such a way as to make possible to forecast future price movements. EMH says asset prices reflect the optimal forecast of fundamentals, given all available info.

    To summarize. Not one commenter has provided me with a plausible and specific public policy implication for more regulation based on the anti-EMH take on the subprime crisis. The one example that intrigues me is the hedge funds. If it is illegal for managed funds to invest in them, that law should be repealed.
    Again, to be clear, I think there are many public policy implications that come out of this crisis, just not based on the anti-EMH position.

  42. Gravatar of odograph odograph
    23. March 2009 at 06:26

    Thanks for the response ssumner,

    Odograph, Your view may be right, but has no obvious policy implication, until you added the yellow/red flag at the end. But if the case you cite was as obvious as you claim, why didn’t smart investors short bank stocks in 2007. A few did, but most did not. Maybe it wasn’t obvious.

    Really “unpredictability for another reason” is my main point. When I get out to red/yellow flags I think we are very much in the domain of “decisionmaking in the face of uncertainty.”

    That is not the best domain in which to make decision, but sometimes we must. I mean, going back to the mortgage thing and associated recession … I decided to come back out of retirement in 2007, so that I’d be working through the coming downturn. Was I smart or lucky?

    I know that I’ll never know 😉

  43. Gravatar of odograph odograph
    23. March 2009 at 06:28

    BTW, re. policy – I think that is definitely in the domain of decisionmaking in the face of uncertainty.

    We can’t use theory, we have to use common sense.

    (We also have to make rules that work whether EMH is true or false.)

  44. Gravatar of Joe Joe
    23. March 2009 at 07:00

    Scott, as an earlier commenter pointed out, so long as you are referencing the weak form of EMH, you will get no argument from me. Anyone sticking to a strong form of the EMH should be mocked mercilessly (kidding, but the strong form of EMH is truly ridiculous).

    This brings me to my question though. You make the point that ex-ante it is nearly impossible to pick an active money manager to beat a passive index. You also make the point that other than Dr. Roubini, there weren’t a lot of voices of caution regarding the housing/credit bubble, especially from our regulatory apparatus; again showing the difficult of predicting the future. So I must ask, how then do you have any faith, ex-ante, in monetary policy? Why should we conclude that the men and women in charge of the Federal Reserve are able to know, ex-ante, what the optimal rate of short term interest should be in order to foster maximum employment and price stability? Why should we believe they could accomplish even one of those two mutually exclusive goals? Why shouldn’t the Federal Funds rate be left to its own devices? Why should the FOMC actively manipulate, because that’s what they do, the short term price of money? If you can’t trust money managers to beat the market, and you can’t trust regulators to identify bubbles in advance and successfully prevent them, why should you trust the Federal Reserve to “know” the “right” level at which to set the price of money?

  45. Gravatar of Markets are efficient « Nick Hudson’s Blog Markets are efficient « Nick Hudson’s Blog
    23. March 2009 at 08:36

    […] are efficient 23 03 2009 So defends Scott Sumner, an economics professor at Bentley University. I’ve added a link to his blog on […]

  46. Gravatar of Regulators, Risk, and Roubini » The Foundry Regulators, Risk, and Roubini » The Foundry
    23. March 2009 at 08:40

    […] Greg Mankiw, Bentley University professor Scott Sumner writes on efficient-markets hypothesis (EMH): So the anti-EMH argument for regulation must be based on the following: bankers are irrational and […]

  47. Gravatar of StatsGuy StatsGuy
    23. March 2009 at 08:54

    First, it is not my intent to be hostile, merely conversational. As I said, these are some nitty points.

    1) The roulette wheel:

    Thank you for clarifying your meaning. So you are essentially arguing that this is a selection bias problem in what gets published, not that the statistical tests are flawed? Sure, I buy this – although it’s important to note that the roulette wheel example of coming up with a _specific_ sequence is still improper.

    It think it’s easier to say the following: If you run a statistical test at 95% confidence, then 1 in 20 times you look at the data you will find meaningless anomolies that appear “significant”. And if you are a bad researcher (or one who rationally follows incentives in a perverse academic reward system) you will mine data to look for anomolies to defend a point of view that can get you published.

    Sure. It’s essentially a bayesian argument against the quality of the evidence, given that you have a very high prior belief in EMH. (It seems that prior beliefs are hard to shift without extremely compelling data…)

    2) Computer programmers were stupid… Sure. In a global sense, yes. That is exactly my point. The programs work (and indeed can be stabilizing), as long as you don’t factor in higher order effects (endogeneity). The programmers were not dumb. The economic theory, however, was incomplete – and dangerously so.

    3) When I write “This reduces spending, and can precipitate a recession all by itself (_some_ argue this happened in ’87)”, I should specify who the “some” are:

    Please note their argument – that the mini-recovery after 87 was engineered by the Fed and others during an election year, and was followed by a more prolonged recession.

    You reply: “it [87 crash] was followed by a huge economic boom”

    I’m sorry, looking at a graph like this:

    I have a hard time seeing the huge boom. It looks like GDP growth stayed steady a little while longer (till the end of the 88 election year) then took a nosedive. As the wikipedia article contends. Maybe that is long enough to argue the 89-91 drops didn’t result from the 87 crash (although the wiki article disagrees), but it’s hardly a boom.

    As to the risk of a wealth-shock inducing a recession, please note that the downward trend of 89-91 happened in spite of massive government efforts otherwise…

    I had a conversation with an investment banker who was involved in the October 87 crash a few years ago. On the day of the crash, everyone in the big investment banks was literally packing their bags. Then they got a phone call from the Fed, which told them they had an unlimited credit line (and that all the other big banks were getting the same credit line)… That meant a one way bet. They’re already insolvent, so if they bet with the Fed’s money the worst case is they are more insolvent. The best case is they make back their losses. And if everyone is making one-sided bets, the market stabilizes. (IMO, this is what Romer and Geithner are trying to engineer right now with the 85% loans.)

    I think I will stand by my contention that a large market shock creates risk of a recession, but I also agree that prompt and aggressive govt. action can mitigate (possibly eliminate?) this recession. And that such action is not always detrimental in the long term.

    4) Insider information – well, I guess I have a hard time believing inside information is a good thing given the obvious distributional consequences and the inobvious efficiency gains (indeed, it seems like depressed stock prices due to lack of trust in transparency of institutions is inefficient).

    5) Why don’t managed funds invest in hedge funds? The simple answer is that most can’t – managed funds usually have a prescribed set of investment activities they target. For example, big cap income, or small cap growth, or Euro-Pacific, etc. I suspect we were moving toward more integration between managed and hedge funds (e.g. managed funds behaving more like hedge funds, particularly with innovations like 130/30 funds and leveraged on-the-margin managed funds) when the bomb blew up. Retail investors were late to the party, and ended up holding the bag when it ended.

    6) Cognitive arguments – I believe I said that no one has proven the cognitive irrationality at the individual level translates into irrationality at the group level. I don’t know it can be truly proven (or disproven), even though people have strong beliefs pro/anti-EMH.

    7) I should have put the nits at the end, because my main charge is that EMH is _incomplete_ because it does not account for system endogeneity. In fact, most of the economics field is incomplete because it only addresses local system dynamics, and the pieces which try to encompass more of the system (like endogenous growth theory, or endogenous money supply) are widely dismissed as being mathematically intractable.

    As if mathematical tractability is a higher order priority than being empirically accurate.

    Honestly, the response of the Econ field to endogeneity reminds me of Einstein’s response to the Heisenberg Uncertainty Principle…

    “God doesn’t play dice with the universe”

    Except that he does.

  48. Gravatar of Randy Randy
    23. March 2009 at 09:19

    Scott, the example you provide is the only one where it might be true that active management can on average provide an advantage, or more accurately the only type of argument. Effectively, by saying individuals earn less than the market or professional fund managers you are saying that they will on average be less astute and hence underperform. This would clearly be a violation of the EMH and suggest that there is value in the incremental research of professional investors versus the research of individual investors. Having said that, there is to the best of my knowledge no information or studies that support that claim. In other words, individuals do not systematically underperform professional investment managers.

    In a sense, it would be a surprising market outcome if an investor were to consistently underperform and not modify their strategy, and there is no reason, at least for years now, why an individual could not switch from their losing strategy to an indexed strategy. So it is hard to conceive of a functioning market where participants have transparency on the market return and the ability “to learn”, that would result in one group of investors systematically underperforming another. This is partly what I meant when I said that at least the weak-form EMH can be used as a reasonable first approximation.

    What can occasionally occur is that you have institutional frictions that create clientele effects, where a group of investors with one characteristic can profitably trade with a group of investors that have different characteristics that can make it look like one group is outperforming another, but in fact they are both optimizing their unique expected utility functions. A good example of that was when dividends were taxed at a higher rate than capital gains. As you probably know from Modigliani-Miller there is no reason why an investor should favor receiving return in one form over the other, it violates arbitrage principles. But in a market with a significant percentage of participants that are either tax-exempt (pension funds, endowments, individuals in tax-deferred IRA’s, etc.) or taxable, it would turn out than on a pre-tax basis one should earn a higher return with dividend payers, ceteris paribus, than with non-dividend payers, under an unequal tax regime. Why? Because this equilibrates the return between the two asset types on an after tax basis, to the level of the cap-weighted marginal tax rate in the market. There is some evidence of this clientele effect having existed in the 70’s and 80’s until the tax regime was changed to dividend/cap gain neutral. (As an aside, I think this may explain some of the Fama-French data since dividend payers tend, on average, to have lower valuation characteristics.) It is also why I have always argued that differential tax rates of dividends and cap gains distorts economic behavior in an unfavorable way. You can tax them both to be lower or higher relative to labor income, but capital return taxes should be equal with respect to form of payment for minimum economic disruption.

    So, it is important to understand market constituency and determine the possibility of profitable clientele effect trades. Remember, most financial theory asserts all players have the same expected utility of wealth function adjusted for personal risk preferences. However, it is possible to postulate alternative utility functions that it would be rational to seek to maximize other than the traditional expected utility of wealth model. (After tax optimizing and immunizing liabilities are two pretty common examples.) While I personally believe that clientele effects are rare and fairly weak, they can be something useful to exploit when investing for an investor with specific characteristics.

    At any rate I stand by my assertion that the return of a properly constructed index (Russell 3000, Wilshire 5000) that covers the entire market is by definition the return of the average investor. I have concluded that why people pursue active management is a complex issue, but my belief is that if done properly and priced properly it creates an option to outperform that has some value whereas indexing is accepting average results (actually a touch below depending on the costs of the index fund you are in, for institutional investors indexing is nearly free, anywhere from 1 – 5 basis points, where as for individuals it costs between 10 and 20 basis points.) Interestingly, therefore, tax effects not withstanding, this means in some cases it may make more sense for individuals to take a chance on active management because the marginal cost of participating may be a bit lower for them. But the problem I noted earlier remains, there is no good way to identify the superior investors, or for them to self identify, because the volatility of active returns are high enough that historical results do not provide a statistically meaningful result given that the sample size is limited to the career span of the individual. On a broader level, the benefit of active management remains the price discovery process that it provides. I don’t see how prices could be “efficient” in an asset market without people trying to make an excess profit. I can’t come up with an equilibrium model where everybody is a passive investor, even though it makes the most sense for most people to be one.

  49. Gravatar of Regulators, Risk, and Roubini « Conservative Thoughts and Profundity Regulators, Risk, and Roubini « Conservative Thoughts and Profundity
    23. March 2009 at 09:27

    […] — nhiemstra @ 1:18 pm Via Greg Mankiw, Bentley University professor Scott Sumner writes on efficient-markets hypothesis (EMH): So the anti-EMH argument for regulation must be based on the following: bankers are irrational and […]

  50. Gravatar of Steve Sailer Steve Sailer
    23. March 2009 at 12:15

    One obvious problem with all the calls for regulation is that we had lots and lots of regulation of mortgage lending for “lower income and minority neighborhoods,” but the vast majority of it was pushing toward more lending in the name of diversity. The feds tended to overrule state regulators who were concerned about excessive lending by boiler rooms to people unlikely to pay back or even be able to understand what they were signing.

    It’s a little bit like an airliner crashes because the jet engine snapped off and everybody is standing around saying, “You know, airliners crash for a lot of reasons. Look at that one that crashed because of geese. No plane flies forever. And how do we know the plane crashed just because the engine fell off one minute before it hit the ground? And isn’t the real question why a plane can’t keep flying after the engine fell off?”

    And when you say, “Well, yeah, but in this case, we know the engine fell off, so why don’t we figure out why that happened and how to prevent that in the future?”

    And then everybody says, “What are you, some kind of enginist? We don’t think about engines in polite society. Only enginists think about that.”

    In other words, isn’t the most likely place our society is going to screw up, a priori, going to be exactly the place where free speech and public debate is most suppressed?

  51. Gravatar of odograph odograph
    23. March 2009 at 15:05

    One obvious problem with all the calls for regulation is that we had lots and lots of regulation of mortgage lending for “lower income and minority neighborhoods,” but the vast majority of it was pushing toward more lending in the name of diversity. The feds tended to overrule state regulators who were concerned about excessive lending by boiler rooms to people unlikely to pay back or even be able to understand what they were signing.

    I think regulatory capture was more complete than you credit. There were some diversity-lending programs which where harmless to the banks, just as the more safety-oriented regulation became harmless to them. They did what they wanted, at 30:1 leverage.

    Besides, this is one of those things again where we can ask, if the real estate bubble was fueled by bad US programs, why was it actually global? A few years ago, Shiller in Irrational Exuberance, said it was “Jet Set” cities around the globe. We’ve seen it break that way.

  52. Gravatar of Economic Blog, Everything Economic » links for 2009-03-22 Economic Blog, Everything Economic » links for 2009-03-22
    23. March 2009 at 16:07

    […] Richard Rorty and the efficient markets debate – The Money Illusion […]

  53. Gravatar of Steve Sailer Steve Sailer
    23. March 2009 at 22:28

    “if the real estate bubble was fueled by bad US programs, why was it actually global?”

    Well, how about the country most similar to America: Canada? America is dragging their economy down now, but, no, they haven’t had a foreclosure crisis, and for much the same reasons that large parts of America didn’t have one.

    “A few years ago, Shiller in Irrational Exuberance, said it was “Jet Set” cities around the globe.”

    Moreover, the mortgage defaults haven’t been worst in the Jet Settiest parts of the country, but in downscale places like California’s Central Valley and High Desert.

  54. Gravatar of Sören Höglund Sören Höglund
    24. March 2009 at 00:36

    Steve, it’s hard to not conclude that you’re an enginist, or at least someone with a vested interest in anti-engine sentiments, when you continue to claim engines are the problem even though the entire plane disintegrated.

    AIG was able to underwrite half a trillion of CDS:s without any of the assets needed to back that bet, but somehow the real problem in regulation is immigrants getting loans they can’t afford. That’s the important thing. Never mind that Wall Street was sucking up all the bad mortgages they could get so they could repackage the crap into a CDO and sell it as gold.

    To keep harping about minority loans in the face of that is just ridiculous.

  55. Gravatar of Jeff V Jeff V
    24. March 2009 at 03:43

    Thank you for the reply, Scott.

    I would agree that that is no Anti-EMH case for more regulation; it is reasonable to assume that the regulators are just as rational (or irrational) as the regulated.

    I think a bigger issue is the opposite case; using EMH to justify less regulation. Until the recent troubles, it had been argued that derivative markets did not need regulation, because the Market Efficiency of Rational Individuals meant that the market could regulate itself.

    I never quite understood this reasoning; why should some financial instruments (stocks and bond) be highly regulated, while others (various derivatives) are less regulated? Allowing some folks (Wall Street and Hedge Funds) to play by a different set of rules than other folks (Commoners like me) made it easier for the first group to game the system to their advantage.

    – Jeff V

  56. Gravatar of ssumner ssumner
    24. March 2009 at 04:18

    Odograph#1, I agree about uncertainty. When there is uncertainty it is very hard to prove or disprove the EMH.

    Joe, One reason that I didn’t talk about different versions of the EMH, is that I wasn’t defending it. In fact people seemed to overlook that I attacked it in places. Rather I attacked the anti-EMH position as having no useful implications. That’s very different. If I had to choose, it would probably be the weak EMH. In fact, i tend to like the “weak” version of just about all social science theories. It is a very imprecise science.

    I have no faith at all in the Fed predicting the business cycle. That is why I have published 7 papers in scholarly journals (maybe more than anyone else) advocating monetary regimes where the Fed doesn’t have to predict the cycle, but rather relies on market forecasts. But if they are going to insist on predicting, I am going to give them advice, as I am right now in this blog.

    Statsguy, I think we are very far apart:

    1. No, you do not understand my point. I am not just talking about publication bias. Even if journals published everything, you have the problem of people using data to search for hypotheses, and then testing the hypotheses with the same data. What gets sent to journals is already very biased. You seem to suggest that my “bayesain argument” is somehow controversial. No it is not. Everyone in economics and finance research knows that this is exactly what occurs. How do you think research takes place? It has nothing to do with my priors. I would say the same about pro-EMH research that shows some market follows a random walk. It is worthless. Robert Schiller would never believe such a study just because it had a T-stat of two, and Fama would never believe an anomaly study that had a T of 2. Both are correct to say “I just don’t believe that” when people throw studies in their face. They know the game. So I would be just as contemptuous if I was an anti-EMH guy.

    2. In point 2 you say programmers were stupid, but not dumb. Please define each term.

    3. OK, the boom didn’t last three more years, I checked and it lasted 2 years and 8 months. Unemployment was at 6% when the market crashed, and hit bottom at 5.2% when the business cycle peaked in June 1990. The slowing GDP growth is meaningless. Growth was strong–and its normal for growth to slow a bit toward the end of a long expansion. No serious economist would believe that there was any causal connection between these events. Sorry, but stock market ups and downs have little impact on consumption, there are all sorts of studies that show that. There are often correlations, because stocks sometimes predict recessions. When that occurs the two may move together. But when it doesn’t, as in 1987, then consumption is not affected. You must distinguish between correlation and causation.

    What massive government attempts to prop up the economy? Talking to an investment bankers about whether a recession would have happened is useless. As we found out recently, investment bankers are no more able to predict recessions that economists.

    4. The insider information issue is complex, but in general markets work more efficiently when asset prices incorporate all available info. If someone in a time machine could have gone back from 2008 to 2006, and sold short all sorts or real estate bonds, etc, we would be much better off. And that would have been trading on inside information. The only unfair distributional consequences are when someone violates their contractual fiduciary responsibility. And I said that should be illegal. Suppose someone studies geology for 6 years in college. And then becomes a prospector. He sees a land formation/rock layer that looks promising for oil. Is he morally obligated to tell the rancher about the oil? And thus let the rancher profit from all his years of study? But if not, that’s insider trading.

    5. When you say “prescribed set” I presume you do not refer to any law. If so, you have no argument, as funds can always change their prescribed rules to fit what clients want.

    6. Beliefs, yes. Evidence for the anti-EMH position? None.

    7. Your endogeneity assumption is interesting, now build a useful model that will help us predict markets. It’s not easy.

    Randy, Your example makes exactly the point I was making, there is no evidence that managers are significantly better at picking stocks than relatively uninformed investors, but if the anti-EMH position were true and useful, they should be better. Otherwise, what’s the point of the anti-EMH hypothesis? Certainly I am not saying anything new here. The reason so many finance-type vigorously oppose the EMH, is that it suggests they are not helping their clients by steering them to non-indexed funds.

    The EMH allows for tax considerations to impact relative rates of return (pre-tax.)

    Steve, I am afraid that I generally agree with the comments by Odograoh and Soren that follow. Part of your argument I accept. I think the PC theme in our society does inhibit honest speech about issues relating to race and ethnicity. If you make Thomas Sowell-type arguments in academia, you will be looked down on as a racist. And I also agree that a priori you would expect more societal screw-ups where free inquiry is inhibited. But even if accept all that, we don’t get anywhere near your case. I already conceded that some of what you are describing went on–banks pressured into lending more to low income groups and minorities. But I still think it was a small part of the overall subprime fiasco. Just because something a priori might be expected to lead to a certain problem, does not me that ex post any problem we have is 100% or even 40% caused by that tendency.

    BTW; The S&L crisis of the 1980s, which was totally unrelated to PC attitudes, was also concentrated in the southwest and Florida.

  57. Gravatar of StatsGuy StatsGuy
    24. March 2009 at 09:53

    1) Usage of data in the econ field – what you describe sounds very disheartening.

    2) Globally stupid, locally not so dumb.

    3) The Fed Response to the 1987 crash was immediate and very strong.

    It probably prevented an utter catastrophe. Even so, _some_ believe lingering effects contributed to the recession after the ’88 election.

    4) Asymmetric information causes inefficient markets. This is not currently under debate.

    5) Funds can’t easily change their charter – doing so requires informing all participants, etc.

    Under normal times, retail consumers do not shift money from fund to fund frequently – fund participation is fairly stable (transaction costs, etc.)

    However, the system of mutual funds was changing rapidly, with many new “innovations” – many involving using leverage in the same manner as hedge funds.

    6) Neither position has strong evidence – I agree.

    7) It’s virtually impossible; various computer models (World Bank, etc.) of the world economy which try to incorporate system dynamics not only do a lousy job of predicting growth; they don’t even agree with each other.

    The models that ignore endogeneity are easier to build, but simply wrong. Nonetheless, the econ profession prefers to build them anyway.

    It’s like the tale of the police officer who finds a drunk man looking for his keys under a lamp post.

    “Where did you lose your keys?” the office asks.

    “Over there,” answers the man pointing to the dark street.

    “Then why are you looking over here?” asks the officer.

    “Because I can’t see anything over there,” answers the drunk man.

  58. Gravatar of Randy Randy
    24. March 2009 at 10:44

    Scott, time to put this to bed. Basically, I agree. Weak form EMH is a good first approximation of reality. It should be the basis for public policy proposals. All of the other stuff is useful for investment professionals but there are few public policy implications. I do not think the strong form is as easily defended but I don’t think it has to be for the points you are trying to make. The purpose of posting on tax issues etc. was to note that things that can be explained in an EMH framework may look like divergences from EMH when really they are not, i.e., Fama-French anomalies may not be anomalies at all but rather rational clientele effects.

  59. Gravatar of ssumner ssumner
    24. March 2009 at 11:31

    Randy, I should put it to bed, but I can’t resist one more reply to Statsguy.

    3. There was no recession after the 1988 election. The economy boomed in 1989, with unemployment falling to the lowest levels in 9 years. And every single day the Fed prevents utter catastrophes. If the Fed simply kept interest rates constant, it would usually lead to depression or hyperinflation within 5 years. The point is that the Fed always needs to move rates around to steer the economy. Nothing special about late 1987.

    5. New funds can be set up, it happens all the time.

  60. Gravatar of Charles Trzcinka Charles Trzcinka
    24. March 2009 at 11:41

    I enjoyed this piece but I must say that my view is a bit different. I think the EMH is a very useful tool in any form and I always use it as a first approximation. Even if you examine all the episodes in history that are labeled a “bubble” you can find rational explanations. But it is also true that it is relatively easy to find exceptions to the EMH. In most cases there are transactions cost and “micro-market structure” explanations (e.g. most financial economists think that the 1987 crash was a failure of market structure). However, there are times that the market just seems out of any equilibrium and I have to admit that investors appear to be acting irrationally. Public policy needs to based on a rational market unless proven otherwise
    but certainly needs to take an empirical view. The commodity markets have long recognized the possibility of irrational trading and every contract have price limits and trading hour limits built in which cannot be justified by transactions cost. Public policy needs to take the same view.

  61. Gravatar of srp srp
    24. March 2009 at 14:12

    There is interesting experimental evidence that even when the fundamental is given by the experimenter, so that there is no possible question about fundamental values, bubbles still occur. However, those who make the most money in these experiments are not those who stick to the fundamental price but rather those who ride the bubble up and sell just before it pops.


  62. Gravatar of ssumner ssumner
    24. March 2009 at 15:38

    Charles, I generally agree with your pragmatic take on the EMH, but let me offer one cynical side comment. You mentioned how finance-types explain the October 19, 1987 crash. Of course those experts are looked to for explanations when something like that happens. When someone looks to you for an explanation, there is a lot of pressure not to say “I dunno,” especially if you are a high paid finance professor. There is really no economic theory that would predict the 1987 crash. Any theory is literally “ad hoc.” Unfortunately, there are no other stock crashes of comparable size, before or after. It is a unique event. (And I mean that literally, even though the term ‘unique’ is overused.) So what do my meandering comments imply? All theories of the 1987 crash are WORTHLESS, until and unless they can be verified with a different data set. BTW, I view the 1987 crash as the strongest argument that the EMH is not true, by far. Indeed I know of no other event that even comes close to being so seemingly unrelated to fundamentals

    srp, Thanks for the reply. I am not opposed to experimental economics, but for several reasons one has to be careful in interpreting the evidence. Some studies have shown that certain experimental results do not persist in the real world, when the “games” are played repeatedly. The article cites a similar pattern, although it also speculates on how you could still get real world bubbles. But nothing in the experiment proves that bubbles should be expected in the real world. In addition, one must also be skeptical of results, such as the experiment you cite, where there is a simple get-rich-quick strategy. We know in the real world it is really, really hard to beat the market. One thing I did like about the experiment you cited is that the action of the “speculators” was stabilizing. Earlier, many commenters questioned my view that speculation is stabilizing. But in the attached article the speculators were said to buy low and sell high, which is stabilizing.
    Of course in my essay on the EMH, I did not deny that bubbles occasionally occur, rather I argued the anti-EMH position had no practical (policy) implications.
    BTW: Although I greatly admire the writings of Virginia Postrel (are you a relative?) I think her comments on monetary policy are a bit of a stretch, for two reasons:

    1. Bubbles are not caused by expansionary monetary policy (monetary policy was much more expansionary in 1965-81, but bubbles have been more of an issue in low inflation periods like 1929 (maybe) 1987, 2000, 2006.)
    2. Low interest rates are not an indicator of an expansionary monetary policy–indeed often they are the result of deflationary monetary policies, as in late 2008.

  63. Gravatar of Jon Jon
    24. March 2009 at 19:07

    In addition, one must also be skeptical of results, such as the experiment you cite, where there is a simple get-rich-quick strategy. We know in the real world it is really, really hard to beat the market.

    When I was in school, I participated in some of Charlie Plott’s experiments. On a number of occasions, I pulled in real dollar payouts worth ~$100/hr. Can’t say I never knew whether there was something wrong in those particular experiments, but it was just as you say: there was a counterfactual get-rich-quick-strategy. I say counterfactual because I recall it being a very risky maneuver that kept paying off big.

    I do have a theory though: in the real market, there are lots of very smart players, but more importantly the smartest players comprise an outsized fraction of the volume. So its almost impossible to play against them and win. Whereas everyone else–the 401k, IRA crowd–is a plate of suckers. If you get in a game with just them, you can win big.

  64. Gravatar of Jon Jon
    24. March 2009 at 19:31

    “1. Bubbles are not caused by expansionary monetary policy (monetary policy was much more expansionary in 1965-81, but bubbles have been more of an issue in low inflation periods like 1929 (maybe) 1987, 2000, 2006.)
    2. Low interest rates are not an indicator of an expansionary monetary policy-indeed often they are the result of deflationary monetary policies, as in late 2008.”

    #2, I dead on agree; but #1? I’m skeptical. Monetary policy has been on afterburners since 1993. Sure, the base did grow faster at other times, but the changes to the reserve-ratio have made a big difference. Say what you will about the coupling of GDP+M3, M3 growth has been disturbingly rapid since the early 90s. The confluence of that with first a little bubble (tech boom) then a big bubble (housing boom) is just too much to be a coincidence.

    Maybe if the Fed had been paying attention to M3, we wouldn’t be in such trouble now:

  65. Gravatar of Jon Jon
    24. March 2009 at 20:18

    Everyone in economics and finance research knows that this is exactly what occurs.

    Yes, economics is actually one field where this problem is actually well understood! Thank-god. Much of the best (most accessible) exposition on this subject is in econometrics journals.

  66. Gravatar of Randy Randy
    24. March 2009 at 21:10

    Scott, I keep trying to leave this topic alone and then a comment is made I can’t let go. At least its an addiction without known adverse consequences. As someone who on Oct. 19, 1987 was trying to reposition a portfolio through a modest “program trade”, (ultimately pulled and I went to lunch to commemorate working on an historic day) I actually disagree that it is a day that potentially refutes the EMH. Some comments to defend that statement; many professionals had argued the market was badly overvalued for the previous six months and many TAA (tactical asset allocation) managers had already pulled a lot of money out of stocks and into bonds; while the day may have been a “black swan” the return for the year of 1987 is smack dab in the middle of the projected distribution of returns for equities; and which day refutes the EMH, Oct. 16th when the market was valued roughly 40% higher than the 19th or when it repriced on the 19th , or the 20th when it repriced again? As I tried to allude to in a previous post, the market is understood to be a very random asset, well approximated by a random variable with lognormal distribution, a positive first moment and with a quite high second moment to boot. (To get real geeky I also think it has fat tails, (kurtosis) and actually in the long run is skewed in the investor’s favor.) Do 4 standard deviation events occur all the time? No. Do they never occur? No again. Remember, when we say the market is efficient we say that there is no ex-ante way to construct a strategy that will outperform the market without access to special non-public information. It doesn’t mean that unexpected stuff doesn’t happen. Consider another example of what have been shown to be efficient markets, betting in sporting events. Since baseball is my personal favorite sport I will use that as my example. For anyone who is “knowledgeable” about baseball it is nearly inconceivable that the 1987 Minnesota Twins could win the World Series, but they did. Because the betting books in Las Vegas did not “predict” that outcome, i.e., forecast that as the highest probability outcome, does that mean that the betting books in Las Vegas are inefficient? Absolutely not, because it took special information or a belief in the occasional non-highest probability outcome coming true to make that bet. (Or like my good friend Steve, you are just an emotional Twins fan and can’t help yourself in believing it could happen.) It is definitively not proof that the market for sports betting in Las Vegas is inefficient. In fact, as more than one doctoral dissertation has demonstrated, the book is Las Vegas in the paradigm of a weak-form efficient market. No mechanical trading strategy (betting on the home team underdog and so on) has ever been demonstrated to work in Vegas. Weak form EMH, QED. Yet, the Twins won the World Series in 87. It is an error of category to say that because the most probable outcome in a probability distribution did not occur means the market is inefficient. That is what it means to manage the risks of probability distributions with finite variance. Things happen all the time that were not the most probable outcome but are well within the distribution. Oct. 1987 easily falls into that category. As I said in my original post, efficient does not mean “correct” in some metaphysical construct. It just means you had no way to know in advance how to make money from the event, on average. So, Oct. 87 has never given me a moment’s pause in my belief that the weak form EMH is accurate to a first approximation. And I don’t see any argument that makes sense why it should. As someone who spent their life investing the capital markets, I would say that without question the biggest mistake you can make is being sure you know something that the market doesn’t.

  67. Gravatar of Carl Futia Carl Futia
    25. March 2009 at 03:34


    Here is my test comment to the “Richard Rorty” post.


  68. Gravatar of Carl Futia Carl Futia
    25. March 2009 at 03:43

    Dear all:

    I tried to post this comment Sunday, but evidently and embedded link stopped it from getting published. But, at the risk of beating a dead horse, here it is:


    I think it is important to distinguish between two kinds of market “efficiency”.

    The first is efficiency in the sense of the Efficient Market Hypothesis (EMH). This asserts that markets don’t make too many economic mistakes – they generally price assets to reflect “rational expectations” of future discounted cash flows, dividends, etc.

    There is a second kind of efficiency which to me seems more plausible. This efficiency is a corollary of the No Free Lunch (NFL) principle. Markets are allowed to make big mistakes (bubbles and crashes) all the time, but NFL asserts that these mistakes must be unpredictable in both magnitude and timing using models estimated from historical data.

    Much is made by behavioral economists of statistical evidence which is inconsistent with EMH. But take a look at this 1998 paper by Euguen Fama; “Market efficiency, long-term returns, and behavioral finance”. It appeared in the Journal of Financial Economics, volume 49, pp. 283-306.

    In it Fama argues that the anomalies detected by the behavioral economists do not constitute a predictable component of market returns. This is precisely what one would expect from the NFL principle.

    Scott makes the crucial observation by noting that Shiller believed the U.S. stock market was irrationally exuberant in 1996 when he offered this phrase to Greenspan for his famous speech. The S&P 500 was at roughly the 650 level then, but would more than double over the subsequent three years. Even at 750 today the S&P total return since then has been about 3% (below historical norms but then so are current price-earnings ratios using Shillers measures). The point is that even the world’s greatest expert on bubbles couldn’t use his expertise to predict the one that made him famous!

    I think the data lend strong support to the assertion that, while markets often make mistakes – thus contradicting EMH, these mistakes are unpredictable using statistical models.

  69. Gravatar of Joe Calhoun Joe Calhoun
    25. March 2009 at 13:32

    It seems to me that to believe in EMH, one must believe that investors can act rationally without realizing it. What I mean is that the outcome, regardless of the motivations of investors, is rational. For instance, I think it was perfectly rational for investors to buy real assets during the years from 2002 to 2008 when the dollar was falling. I also have to acknowledge that most people buying commodities and real estate made no connection between the falling value of the dollar and the rise in the price of real assets.

    Having worked for many years as an investment advisor I can attest to the irrational behavior of individual investors, but in general it seems that the market as a whole is quite efficient. So in any market you might have some investors acting rationally while others are acting completely irrationally and yet the overall outcome is quite rational. And that is what EMH predicts does it not? That the aggregate actions of all the individuals acting rationally and irrationally will produce a result that is rational and efficient?

    As for policy implications, it seems to me that if the markets produce rational outcomes based on the price signals being generated, we should concentrate on the price signals. It seems to me that it is the Fed and the Treasury which need reform as they are the source of many of the price signals that produce rational but sub optimal outcomes. The housing and commodity bubbles (and why is that everyone talks about the housing bubble but no one talks about the commodity bubble?) to me were rational based on the price signals generated by Fed and Treasury policy. The Fed distorted the price of real estate (through interest rate targeting) while the Treasury distorted the price of the dollar. Investors acted rationally based on the price signals provided by the distortions of the monetary authorities. One can look back now and say that investors acted irrationally, but one can equally look back and say it was the Fed and the Treasury that acted irrationally. Frankly, I think all the other factors (such as CRA, etc.) were just noise in relation to the price signals produced by irrational monetary policy. From a policy perspective it seems useless to try and regulate the innate human response to the price signals provided by the market and therefore policy should concentrate on making sure that the price signals are not distorted by our monetary institutions.

    The question of course is how to do that and frankly I’m not sure there is an answer. During the last two centuries, we have had bubbles and busts when on a gold standard…and when we weren’t. We’ve had bubbles and busts when we’ve had much higher bank capital requirements than now…and when we’ve had lower capital requirements than now. We’ve had bubbles and busts with central banks….and without. So what’s the answer? As unsatisfying as it is, I think there may not be one…

  70. Gravatar of Alex Golubev Alex Golubev
    25. March 2009 at 14:47

    I’m sorry I missed this discussion in real time. There are a lot of great comments. However I think it helps to take a step back. IF EMH was not true, someone would be able to SET prices better than the market. While markets can and certainly do misallocate resources, it can hardly be argued that a group of traders or designated regulators can do a better job (buffets of the world are easily explained by the alrge sample size and dont’ forget survivorship bias. we’re all bound to become failed buffets). It is also obvious that there are such things as material information. A simple probability weighted bet on interpretations of that information (EPS, merger, etc…) would be quite valuable. So it’s not black and white and i see no utility in making it so. The only questions is of policy making. I see no problem with the market MISALLOCATING resources as long as there is no system that is more efficient at allocating them or at the very least allows the public to express it’s views on the allocation issues politically and financially. The specific of housing bubble (if anyone wishes to glorify the present) are in my opinion exagerated due to LEVERAGE. if everyone owned houses without borrowed money, this issue of misallocation wouldn’t be

  71. Gravatar of Alex Golubev Alex Golubev
    25. March 2009 at 14:50

    (oops PART II) a crisis. it is borrowed money that BK’s people. there is a reason why stock margin regulations were introduced by the SEC in the 30’s. it is only logical to apply that to housing. obviously requiring 50% downpayments would cripple the financial world at this point. noone said that history is path independent. my only point is that misallocation of free markets is a secondary culprit to leverage. good fight, good night 😉

  72. Gravatar of alou alou
    25. March 2009 at 18:47

    Dumbest post ever.

  73. Gravatar of ssumner ssumner
    26. March 2009 at 05:00

    Charles, I forgot to respond to your commodity market point. I’m not an expert, but I am a bit skeptical of trading limits. Do they actually do much good? What is the problem if a commodity price overshoots on one day, and then has to be nudged back the next. Does that hurt the economy?

    Jon#1 I’m not too sure about your comment about getting in the game with 401K people. Virtually all stocks are owned by both groups at the same time. So aren’t you always in both games at once? Consider my pathetic 401k index funds. What strategy beats that?

    Jon#2, You are very thoughtful, but like many people you may have a tendency to form preconceived ideas about monetary policy without thinking critically enough about the evidence. Greenspan’s monetary policy was more contractionary than any other Fed chairman in many decades. He inherited low inflation, and lowered it further. (I do agree that policy was a bit too expansionary in 2004-06, however.) Yes, interest rates were often low, but that was because he lowered inflation. There has always been a tendency to look at assets bubbles and reason backwards that monetary policy must have been expansionary. The 1927-29 boom was the only boom in the 20th century that was actually deflationary. Yet many consider the monetary policy to be wildly expansionary because of the stock boom.

    Jon#3, Thanks, it’s good to know I’m not the only one who has noticed this.

    Randy, I am actually going to have to argue the other way here. You’re view is very defensible, but I see one weakness. You argue than the market was “understood to be overvalued.” Isn’t that a definition of market inefficiency? My point is not that the post-crash level was too low, but that it was implausible that both the pre- and post-crash levels were rational in terms of fundamentals, because the fundamentals didn’t change radically in 24 hours. Also, what does the 4 standard deviation refer to? Wasn’t the crash a 26 or 28 standard deviation event? (down 22% where the average change is about .8%)

    I’m also not sure if the Twins case was comparable. I don’t know the exact odds, but for a 500 team to beat a 650 team in a 7 game series (or even two series) can’t be anywhere near a 26 standard deviation event.

    Despite all my reservations, you do make some good points here, and I’m not willing to rule out you being right and me being wrong about the 1987 crash. Just because I don’t understand it, doesn’t mean there isn’t some rational explanation.

    Carl, Yes, I agree. The question of how to distinguish between what you call mistakes and others call irrationality is a subtle one. That’s why I ended up with my pragmatic test; if there are no practical implications, then the irrationality view has no usefulness.

    Joe, I agree with much of what you say, I would just reiterate what I said to Carl, and add that the term “irrational” is vague. Individual investors may be rational, but not well informed, or irrational. The difference is subtle, and the key is how markets aggregate all these views, as you say.

    Alex, Good points. I’m not an expert on leverage. But given one market failure (moral hazard creating by government bailing out banks and depositors, I do understand that there is a case for regulations on leverage–indeed it actually may be stronger in housing than stocks, as the collapse of a housing bubble may damage the financial system more than the collapse of a stock bubble.

    Alou, Yours or mine?

  74. Gravatar of Jon Jon
    26. March 2009 at 07:24

    “Virtually all stocks are owned by both groups at the same time. So aren’t you always in both games at once? Consider my pathetic 401k index funds. What strategy beats that?”

    No, I mean in the experimental setting. I was not in a market with hot-shot x and y. I was participating in a market with other students all of whom start-out with equal shares of wealth.

    “Greenspan’s monetary policy was more contractionary than any other Fed chairman in many decades. He inherited low inflation, and lowered it further. (I do agree that policy was a bit too expansionary in 2004-06, however.) Yes, interest rates were often low, but that was because he lowered inflation.”

    Should I take my lumps and go home? I don’t particular think that we can look at the interest rates and know whether the policy was expansionary or not. So, I certainly am NOT a member of the ‘rates were low’ Greenspan was profligate camp. You skipped past the point that I actually made, which is that there are not meaningful reserve requirements anymore. You cited that the growth-rate in monetary base slowed down, but so? It isn’t ceteris paribus because a much smaller increase in the base now translates into a much larger increase in broad money.

    There are enough distortions in the exchange-value of the dollar that I do not think you can draw such strong conclusions from CPI.

    I’ve laid out a testable hypothesis: if you compare the ‘price-inflation’ of traded and non-traded consumables, there is an asymmetry with the latter experiencing much headier inflation.

  75. Gravatar of alou alou
    26. March 2009 at 13:58

    Both, ssumner, but I meant mine to be.

  76. Gravatar of Alex Golubev Alex Golubev
    26. March 2009 at 15:46

    alou is evidence against the weak for of EMH (beyotch). Sorry, ssumner, blogging is subject to offensive driving.

  77. Gravatar of ssumner ssumner
    26. March 2009 at 17:36

    Jon, I did misunderstand your 401k example.

    I don’t think I used the monetary base argument. I agree with you that the base may not be very informative, so if I said something like that, I shouldn’t have. The import price issue is valid. Perhaps I should have focused more on NGDP growth than inflation. Had I done so, then import prices would not have distorted my argument. Indeed my feeling that policy was a bit too expansionary in 2004-06, dovetails with you import price argument, as Chinese imports were holding down the CPI during that boom. But inflation got pretty high in the 1965-81 period, so high it overwhelms any terms of trade arguments. Pick any inflation measure you like and it was very bad. I know that 16 years isn’t that significant a sample, I guess I am also going with my priors–which are that fully anticipated inflation shouldn’t really cause bubbles–it ought to be priced into wage and loan contracts. So in the end I am not convinced by any explanation for why we have recently had so many bubbles, but at the same time I acknowledge I don’t have decisive counterarguments to you hypothesis.

    Alex, I’ll have a big celebration when we hit 1000 comments–alou is helping me get there quicker.

  78. Gravatar of Randy Randy
    26. March 2009 at 20:50

    Scott, don’t know if you will return to this post but my real point is that the EMH doesn’t preclude rapid and significant change in asset values. It just means that there is no reliable way to forecast those events. And because of that, as I mentioned earlier the mistake in the current discussion about the EMH in general, (there was a really wrong headed post at Mark Thoma’s today) is that because the market hit a price that was ultimately “wrong”, whatever that means, doesn’t mean the EMH is invalidated. There is no metaphysical truth in prices, just current prices. I am sorry I mentioned TAA because that was misinterpreted I am afraid. All active investment strategies act as if the market is inefficient, but in so doing their price discovery activity is what creates at least a weak form of efficiency. And finally, fundamentals may not change radically from day to day but collective risk aversion parameters certainly can and I would say that financial economists have done a good job of demonstrating that changes in collective risk tolerance explains a significant amount of financial asset volatility. And there is nothing about that fact that invalidates EMH either, in my opinion. Remember prices basically reflect the mean of a big distribution and I do not think Oct. 1987 was a 26 standard deviation event either. And all of 1987 wasn’t even close. Return for the year of a bit over 5% isn’t even a one standard deviation event.

  79. Gravatar of ssumner ssumner
    27. March 2009 at 17:06

    Randy, I agree with most of what you say. But I was referring to the October 19th crash only. So I do think it is implausible that the rational expectations view of the fundamental value changed 22% on a “slow news day” when in all the rest of history it has never changed more that 13% even on devastating news days. So I was trying to give the other side the benefit of the doubt. But I also agree with you, as in my work on the Depression I often found seemingly inexplicable price changes did have an explanation, if you understood how markets were reacting to policy. So I usually give the EMH the benefit of the doubt–it is more useful than the alternative.

    On standard deviation, I think you mean the month of October as a whole wasn’t that strange, but that specific day really was something like a 26 SD event, unless I am mistaken.

  80. Gravatar of Sumner on Financial Regulation – Economics – Sumner on Financial Regulation - Economics -
    28. March 2009 at 08:50

    […] are common in the blogosphere, but if you have the patience, they are well worth the time. In a recent post defending the efficient-markets hypothesis (EMH), he includes this insight:So the anti-EMH argument for regulation must be based on the […]

  81. Gravatar of AJ Witoslawski AJ Witoslawski
    28. March 2009 at 09:51

    All of the crises you mentioned were caused by credit overexpansion. I think that most economists would agree that money is non-neutral in the short term but neutral in the long term. Credit expansion happens when banks and/or the government via its central bank creates new money (real or electronic) and either loans it out or buys securities with it. Wouldn’t it follow that this newly created money would artificially raise the prices of some assets and artificially increase the profitability of some businesses?

    For example, increased credit would make construction much more profitable because increased credit means lower interest rates, which means more people can afford to buy houses and more businesses can build factories, rent office space, etc. Likewise, other businesses which rely on credit, such as the automotive industry, would be helped with lower interest rates (credit expansion).

    What causes the bust is when the newly created money spreads throughout the economy. When lower interest rates promote borrowing and spending on capital goods and durable consumer goods like those I mentioned, the workers that make those products would ultimately get a significant portion of that income. Those construction workers and car manufacturers would ultimately get higher wages and spend that money. This new money would push up prices in other sectors of the economy besides durable goods manufacutring. Prices would begin to increase, including the costs of production for those manufacturers, forcing them to reduce real wages and/or lay off workers. This, in turn, would mean less spending, which would hurt retailers and other businesses which do not produce durable goods sensitive to interest rates.

    My point here is that stock prices could seem “irrational,” but that’s only because they’re being misdirected by credit expansion. Increased profitability for businesses sensitive to interest rate changes could cause a rise in the price of their stocks, increasing investor confidence.

    As for evidence, the Spanish economist Jesus Huerta de Soto uses some in this book:

    You can also look at Federal Reserve data which shows falling investment preceding every single real recession:

  82. Gravatar of Lorenzo (from downunder) Lorenzo (from downunder)
    29. March 2009 at 00:19

    I hate it when macroeconomists “discuss” “the” housing bubble. There were housing bubbles across the Western world in jurisdictions where officials have discretionary power over land use. Where land use was able to respond directly to demand for houses, there were no housing bubbles. (Krugman’s “Zoned Zone” versus “Flatland” extended beyond the US.)

    It seems fairly elementary economics to me. Constrained supply increased the price response. That then created an effective “one-way-bet” in those housing markets. That added the demand for an inflation-beating asset to the demand for housing. Once people lost confidence in houses as inflation-beating-assets, that part of their value evaporated (i.e. the “one-way-bet” stopped being so).

    I am certainly not expert in the Efficient Market Hypothesis, but I do wonder if a market for houses-as-houses is the same as the market for houses-as-houses plus houses-as-inflation-beating-asset.

  83. Gravatar of record label business plan » Blog Archive » It’s time for a new strategy – make money the “stupid” way record label business plan » Blog Archive » It's time for a new strategy - make money the "stupid" way
    29. March 2009 at 00:37

    […] are common in the blogosphere, but if you have the patience, they are well worth the time. In a recent post defending the efficient-markets hypothesis (EMH), he includes this […]

  84. Gravatar of ssumner ssumner
    29. March 2009 at 09:42

    AJ, I don’t agree that all the bubbles I discuss were due to credit overexpansion. In fact of the 4 (1929, 1987, 2000, and 2006), only the last one was caused by credit overexpansion. Indeed the 1927-29 stock boom was accompanied by deflation. I also don’t agree with your hypothesis that credit expansion leads to asset price crashes. This theory has very strong implications for investment strategies, and is thus wholely inconsistent with the EMH. Buy your theory only fools would be invested in the stock market after a long credit-fuled boom. But that theory is easily refuted, as at any given point in the history stock ownership is dominated by very rich, very shrewd indiviuals and institutions. I strongly oppose any theory that implies stock movements should be predictable, which is why I think the key mistakes in monetary policy occurred in October 2008, not before.

    Lorenzo, I did point out in my essay that zoning (and expectations about future zoning) constraints played a key role in the housing bubble. I agree with your point about the importance of zoning, and this helps explain why no bubble occurred in texas. But Arizona and Nevada are still a bit of a mystery. It seems that bad forecasts were made in two areas; future zoning constraints, and future population growth.

  85. Gravatar of Ed Lopez Ed Lopez
    31. March 2009 at 05:29

    How do you respond to a reverse causation critique? I.e. The
    supply of older ranch houses is made greater in areas where there is abundant suburban/inland new construction, so these areas are where immigrants settle. Or do immigrants not respond to variation in housing prices in his EMH model?

  86. Gravatar of ssumner ssumner
    31. March 2009 at 05:41

    Ed, Yes, there could be some reverse causation. I was just showing that there could be some fundamentals involved, I was definitely not saying “here is my theory of the bubble.” The burden of proof is on the anti-EMH crowd to show that there is no conceivable fundamental explanation. I just tried to show that there were at least some POSSIBLE fundamentals involved. As I said, the housing bubble looks a little fishy, even to me.

  87. Gravatar of Är marknader effektiva? « Nonicoclolasos Är marknader effektiva? « Nonicoclolasos
    21. July 2009 at 21:44

    […] Här utvecklar han argumentationen (och läs den innan du kritiserar honom). EMH lever och frodas, som det verkar, även om forskare i behavioral finance ofta är av annan uppfattning – för en introduktion, se t.ex. Andrei Shleifers bok Inefficient Markets. […]

  88. Gravatar of Some links about the Efficient Market Hypothesis « A Little Knowledge Some links about the Efficient Market Hypothesis « A Little Knowledge
    11. August 2009 at 13:13

    […] Scott Sumner […]

  89. Gravatar of TheMoneyIllusion » Tight money causes bubbles TheMoneyIllusion » Tight money causes bubbles
    10. January 2010 at 08:39

    […] Many readers may find my views on bubbles confusing.  Last March I explained why I don’t think the debate over the EMH is very interesting.  The question isn’t […]

  90. Gravatar of Tight Money Causes Bubbles Tight Money Causes Bubbles
    10. January 2010 at 21:20

    […] Many readers may find my views on bubbles confusing.  Last March I explained why I don’t think the debate over the EMH is very interesting.  The question isn’t whether the […]

  91. Gravatar of Efficient Markets Hypothesis « William’s Continued Adventures Efficient Markets Hypothesis « William’s Continued Adventures
    11. January 2010 at 08:20

    […] Scott Sumner (pro-EMH) […]

  92. Gravatar of Juan Juan
    9. July 2011 at 10:47

    The efficient markets hypothesis forms the core of faith based economics that flies in the face of reality and critical thinking. How anyone could argue that because the President, regulators, congress and others did not foresee and act on the upcoming crisis is a valid arguement against regulation that would have prevented the crisis is idiocy. As Upton Sinclair said “It is difficult to get a man to understand something when his job depends on not understanding it.” No amount of faith in the purity of markets can overcome the overwelming proof that when incentive structures are skewed to encourage fraud and all those participating in the system (the crooks, regulators, media, preseident and congress) have a self interest in perpetuating the fraud, that fraud will occur. Those that predicted the crisis were not “lucky” they just didn’t have a self interest in perpetuating the fraud and were not blinded by the religion of an efficient markets hypothosis. A defense of a simplistic and now bankrupt economic theory reminds me of my kid who found out that Santa does not exist but still defends his existance because he woke up and some of the cookies were eaten.

  93. Gravatar of ssumner ssumner
    10. July 2011 at 07:11

    Juan, Actually, everyone on Wall Street has an interest in debunking the EMH, because if true many would be shown to be hacks.

    And I do favor regulation of banking, as long as we have deposit insurance.

    I plan 3 more pro-EMH posts in the next week–stay tuned.

  94. Gravatar of TheMoneyIllusion » Links to my views on money/macro TheMoneyIllusion » Links to my views on money/macro
    29. September 2011 at 09:01

    […] A post defending the EMH. […]

  95. Gravatar of My journey to the left wing, and back again « Of Flies and Fly-bottles My journey to the left wing, and back again « Of Flies and Fly-bottles
    3. November 2011 at 11:04

    […] Scott Sumner awakened me to the importance of monetary over fiscal policy (and, indeed, the arguable irrelevance of the latter) as the appropriate mechanism for macroeconomic management. And argued that the falsity of the efficient markets hypothesis doesn’t have the policy consequences I thought it does. […]

  96. Gravatar of More ‘Free Market’ Double Standards « Unlearning Economics More ‘Free Market’ Double Standards « Unlearning Economics
    14. December 2011 at 08:11

    […] Reality trumps theory. But I love the EMH so much…and RET…and […]

  97. Gravatar of ‘That’s OK in Practice, But Does it Work in Theory? « Unlearning Economics ‘That’s OK in Practice, But Does it Work in Theory? « Unlearning Economics
    10. January 2012 at 04:40

    […] Scott Sumner, defending the EMH: I know of no explanation for the 1987 bubble.  The collapse (comparable to 1929) occurred when […]

  98. Gravatar of TheMoneyIllusion » Are there any good arguments against the EMH? TheMoneyIllusion » Are there any good arguments against the EMH?
    31. January 2012 at 05:45

    […] Many pundits claim the housing bubble shows the EMH is false.  This post shows why they are […]

Leave a Reply