“Being There” revisited

Five years ago I did a post entitled Being There.  I compared Warren Buffett to the character played by Peter Sellers in the famous film.  I pointed out that people tend to be superstitious.  They don’t accept unusual coincidences.  Thus if someone outperforms the market for 20 years in a row, the general view is that it can’t be luck—after all the odds are a million to one against.  People forget that just as someone must win the lottery, in any group of a million investors it is a logical necessity that there has to be one who is luckier that all the others. Here’s a test I proposed back in 2009:

I suppose this should be testable.  If the EMH is correct then the top ten richest Americans should not see out-sized returns, once they have reached that pinnacle of success.  I have no idea whether the data exists to do this test, but is would be a good way of resolving the issue of whether Buffett just got lucky.  When similar tests are done with successful mutual fund managers, it turns out to be merely dumb luck.

At the time, many commenters claimed that hedge funds had greatly outperformed the market in recent years, disproving the EMH. It’s well known that hedge funds have since done relatively poorly.  But how about Mr. Buffet?  Here’s the NYT:

A new statistical analysis of Mr. Buffett’s long-term record at Berkshire Hathaway has just been done, and it’s come up with some fascinating insights about his abilities, past and present, and about the chances that the rest of us have for beating the market. Using a series of statistical measures, the study suggests that Mr. Buffett has indeed been blessed with an impressively big dose of alpha over a very long career.

But it also reveals something that isn’t impressive at all: For four of the last five years, Mr. Buffett has been doing worse than the typical, no-frills Standard & Poor’s 500-stock index fund “” so much worse that it’s unlikely to be a matter of a string of bad luck. Mr. Buffett has begun to behave like an investor with no alpha at all.

Why am I not surprised?  And don’t say, “it’s harder to do well when you are big.” It’s true that it’s harder to do extremely well when you are big, but it’s not hard to outperform the market when you are big, if you truly have “alpha.”  To see why, assume Buffett is only able to find $5 billion in good investments each year, but has $50 billion to manage.  Then put the $5 billion into the good investments, and index fund the other $45 billion.  If he truly had alpha he’d still be outperforming the market.

“It shows how amazingly difficult it is to keep beating the market, even for a master like Warren Buffett,” Mr. Mehta said in an interview. “And it suggests that just about everybody else should just use index funds and not even think about trying to beat the market.”

I certainly agree with the second point, but I disagree with the first point.  It should read: “It shows how difficult it is to keep winning the lottery, even for someone who has already won Megabucks.”

Sometimes people claim my anti-EMH pro-EMH arguments have no testable implications. That’s wrong.  I’ve been doing this for 5 years and again and again I’m being proved right and my critics are wrong:

1.  Back when Bitcoin was $30 I did a post skeptical of “bubble” claims.  The odds were probably at least 10 to one against me being right (as the potential upside was far more than downside, and hence far less likely) but I was right.  Even after the recent drop I’ll be glad to buy any Bitcoins you’d like to sell me at the “bubble” price of $30.

2.  I was skeptical of hedge funds.

3.  I was skeptical of the Oracle of Omaha.

4.  I was skeptical of Robert Shiller’s ability to give useful market timing advice.  He did not recommend buying stocks in March 2009.

5.  I was skeptical of the claim that the 2006 house price boom was a bubble.  We now know that Canada, Australia, New Zealand and Britain did not crash, after similar price run-ups.  This suggests the US crash was not pre-ordained, rather just “one of those things.”

Since I started blogging in early 2009, events have strongly supported my pro-EMH claims.  Anti-EMH models are completely useless.

PS.  Ok, I was wrong about one thing.  Last March I half jokingly said “stock prices have reached what looks like a permanently high plateau,” echoing Irving Fisher’s infamous 1929 prediction.  Yes, I was wrong—stocks have gone much higher over the past 13 months.

PPS.  Noah Smith has the best piece on high frequency trading that I have read so far.  (Tyler Cowen has also had some good stuff.)  Noah says we know almost nothing about whether it is good or bad, and I know far less that Noah.  Which suggests I have negative knowledge.

HT:  Clark Johnson.




78 Responses to ““Being There” revisited”

  1. Gravatar of Major_Freedom Major_Freedom
    6. April 2014 at 14:01

    To reject EMH doesn’t obligate an acceptance of the opposite of what EMH claims. One can reject both. There is a third way.

    If every investor sought to invest in the same portfolio that follows from EMH, such as a maximally diversified index fund, let’s call it the “apathetic portfolio”, then they wouldn’t be able to purchase it, because every other investor would be seeking to purchase that same portfolio, and thus not be in a position to sell it.

    Similarly, if every investor sought to invest in a portfolio that follows from anti-EMH, such as the empirically confirmed theory that stocks with low P/Es consistently outperform all other stocks, and hence “consistently beat the market”, then no investor would be able to purchase such a portfolio, because no other investor would be in a position of selling what they too want to purchase.

    Not all rejections of EMH consist of pretending that an investor can consistently beat the market, no exceptions.

    For myself, I reject EMH because I reject its characterization of the capital markets. This doesn’t mean I accept the polar opposite extreme however.

    The theory that individual investors can utilize their superior knowledge and skill relative to the rest of the population of investors, to beat the market as long as those other investors remain relatively inferior in knowledge and skill, then EMH is a false description of what is going on as long as that is true. The market is “inefficient” here, qua EMH, because an investor is gaining more than the market average due to relatively superior knowledge and skill, which is to say not luck.

    The capital markets “work” only if investors behave in such a way that EMH holds they should avoid. Investors trying to beat the market is the market itself. For this reason, adhering to a theory that rejects EMH is far more useful than adhering to EMH.

    EMH really is non-falsifiable, for if Warren Buffet were still beating the market today, then the pro-EMH will just say “He has been lucky so far, just wait…”.

    And then we wait and wait. If they die, then they were lucky the whole time. If they survive and eventually lose relative to the market, then the pro-EMH crowd concludes AHA! See? EMH is correct!

  2. Gravatar of A A
    6. April 2014 at 14:08

    This kind of evidence closely resembles reasoning from a price change. Ignoring Warren Buffett, the alpha in this story is just a plug in to match realized and expected returns. As a measure of “skill”, it is ambiguous unless skill equals alpha as a semantic rule. The market reacts to behavior, which can be motivated by thoughtful reasoning, or by dumb luck, at varying points, and in different combinations. So to rely upon the appearance, or lack thereof, of alpha as a measure of historical luck requires an assumption about the consistency of whatever motivated the alpha in the first place .

    There also seems to be an unexplained temporal weighting in this argument. Recent underperformance is taken as being more meaningful than underpeformance at earlier periods. Why would this stretch represent a better forecast of future performance than previous stretches? If your test is validated when reality matches your prediction that underperformance will occur, then you can be validated at various soft spots in a given manager’s career. But then, if the manager subsequently improves performance, the test resets with the claim that future underperformance will again validate the theory.

    Also, even if you accept the assumptions of alpha as a measure of luck, and the temporal weighting of recent underperformance, why is the stock price of Berkshire Hathaway a proxy for management performance? Stock prices represent forecasts of performance, so differentials between benchmarks over a given period don’t necessarily represent actual underlying results. The market can view benchmark SP as being extremely volatile in 2009, and company BH as being more stable.

  3. Gravatar of Major_Freedom Major_Freedom
    6. April 2014 at 14:24

    And if anyone is interested, there is an irony in this blogpost.

    Sumner is claiming to have been consistently proven empirically correct regarding the history of capital markets, in the face of other individuals who have been proved wrong by implication.

    He is claiming to have been able to consistently beat the average level of knowledge (which consists of those who agree with him and those who disagree), which is to say he is claiming to have a positive knowledge “alpha”. If he has superior knowledge, then he should be able to beat the market that is a function of both superior and inferior knowledge.

    One cannot empirically confirm a theory that would contradict the very nature of why empirical predictions are not possible in the economy as per the theory!

    Sumner doesn’t seem to understand EMH. If we consider for the sake of argument that EMH allowed for empirical predictions, then EMH could not even be a description of what the capital markets are. EMH would have to remain hypothetical, meaning it must admit the possibility that non-EMH theories of the capital markets are true. If it did not do so, if it did not admit any such possibility, then it would not be an empirical theory at all. It would be non-falsifiable.

    And that is exactly what Sumner is doing. He is refusing to even entertain the possibility that non-EMH theories are correct. He can observe Warren Buffet beating the market for 20 years, or Joe Shmo beating the market for 1 day, and his EMH theory tells him to identify this investing history as luck, no exception. This is not empiricism. This is a priorism masquerading as empiricism to gain credibility in the positivist crowd.

    The only way Sumner can avoid this contradiction, would be if he accepts the possibility that he might be wrong about EMH being an accurate description of the capital markets. That it is possible for an individual investor to consistently beat the market with something other than luck. That rejection of EMH might be a more useful, more accurate to the reality of markets approach.

    To address the numbered statements that allegedly show Sumner is being empirical, it’s rather easy to see he’s not being serious:

    1. If the price of Bitcoins did collapse from that $30 price point, and stayed below $30 since that time until today, then Sumner would have been claiming “Bubbles are implied in EMH. EMH is not falsified.”

    2. If Sumner’s “skepticism” of hedge funds turned out the other way, then he would have said that this is implied by EMH

    3. If Buffet was still beating the market, then Sumner would have claimed that Buffet was still being lucky, and just you wait for me to be right!

    4. If Shiller’s predictions were accurate, then Sumner would have claimed that he got lucky, that EMH is still true, and that sometimes being right is what happens in EMH.

    5. If the housing boom in the US persisted until today, as it did in Canada, then Sumner would have said that this is exactly what to expect from EMH. That all the bubble talk in the US was wrong. blah blah blah.

  4. Gravatar of Vaidas Urba Vaidas Urba
    6. April 2014 at 15:25

    “I was skeptical of Robert Shiller’s ability to give useful market timing advice. He did not recommend buying stocks in March 2009.”

    But his model did recommend holding more stocks in March 2009 than say in March 2008.

  5. Gravatar of Frances Coppola Frances Coppola
    6. April 2014 at 15:45

    Good grief.

    The UK’s housing market DID fall in 2008-9. But it did not fall anything like as much as property markets in US, Spain or Ireland. And the reason for this is that unlike the US, Spain and Ireland, the UK did not have excessive residential property construction in the run-up to the financial crisis. The UK’s housing market is supply constrained. It was never going to fall much. I thought everyone knew that.

    Nor is it going to fall any time soon, despite the fact that it is probably overvalued by at least 30%. The UK’s residential property market is propped up not only by ridiculously tight planning regulations that seriously restrict supply, but (since 2011) by a combination of fiscal and monetary policy (FLS and H2B), and now also by inflows of foreign capital.

    However, commercial property prices in the UK DID crash in 2008-9, and it is this that did the real damage to UK bank balance sheets. Commercial RE prices have fallen by 60% so far and are still falling.

    Really, Scott, if you are going to talk about the UK, do your homework.

  6. Gravatar of Scott Steele Scott Steele
    6. April 2014 at 15:58

    Hey Scott – I like the post; EMH is a good theory and is supported by many empirical studies. However, I do not think the bit about hedge funds supports your EMH argument.

    1. “Hedge fund” could mean a lot of different things with all their different strategies. Therefore, I would reject the notion that they should be judged by their performance relative to the “market” however you decide to define that (S&P 500?).

    2. It is much easier to define the market/benchmark for a mutual fund (especially equity mutual funds) and therefore tests looking for alpha are easy. And I agree, for a mutual fund manager to consistently beat the market and generate serially correlated alpha, it is tough because most of the market is priced efficiently.

    3. Many hedge funds strategies involve looking for assymetric returns. They invest in things with limited downside and huge upside. And as we have established with EMH being generally correct, these are rare and may only come along once every 5 years+. And because many hedge funds are not limited to a specific area of the market (like a mutual fund manager would be) they can look for these opportunities and invest in them. If this is true, then I would expect many years of ok, but probably relatively low returns followed by a year of an incredibly large return. Because in that year, one of their bets paid off huge. So even if you could define their “market” the hedge funds would not have serially correlated alpha. The returns may even look lucky, until you examine a track record of 20 to 30 years (maybe even more). But due to the assymetric nature of the bets and the fact that most market mispricing occurs in years of panic, a typical hedge fund will do ok for a few years and then have excellent returns, in a bear market. This negative correlation with the market makes them an important part of your portfolio.

    P.S. I tried to stay away from talking about any of this in terms of “risk” because I think the topic is poorly defined and understood. And yet, we throw around the term frequently like everyone knows exactly what it is….

  7. Gravatar of TravisV TravisV
    6. April 2014 at 16:00

    Frances Coppola,

    Quickly, just a bit of context on Sumner. In the post above, he said “the US crash was not pre-ordained.”

    Remember: a key part of Sumner’s argument is that monetary policy was far too tight in 2008 and 2009 in both the US and the UK. If NGDP had kept growing 5% in 2008, 2009, 2010 and 2011, then overall average commercial real estate prices might not even have fallen.

    As for the average U.S. home price maybe it would have fallen 0%. Or maybe 20%. But instead, due to tight money, it fell far far far more than that.

  8. Gravatar of ssumner ssumner
    6. April 2014 at 16:05

    A, You misunderstood the post, there is no reasoning from a price change involved. It’s out of sample tests, that’s why we care much more about what happened recently than what happened earlier. The earlier data is “selected” via data mining. Ask yourself why anyone talks about Buffett at all. Is he intrinsically interesting? No, he is examined because he was successful–data mining.

    I’m not saying that the data tells us anything about Buffett’s alpha. I’m saying that the data that others claim shows Buffett had alpha, shows no such thing.

    Vaidas, I’m not interested in his model, I’m interesting in his advice. That’s why he’s famous–advice. Anyone can fit a model to past data, but I want to know if that model gives good advice going forward. His did not.

    Frances, Yes, I’ve done my homework and know all of that–but it has no bearing on anything I said my post.

  9. Gravatar of A A
    6. April 2014 at 16:41

    But this post is not referencing an out of sample test. It was inspired by a data point that agreed with an existing bias. Isn’t Buffett mentioned because of the NYT and statistician posts?
    You could have made a similar claim prior to any other underperforming period, and made the same inference about the value of recent data. The value of out of sample information is that it continues out of sample, not that it is recent.

  10. Gravatar of A A
    6. April 2014 at 16:45

    The response to Vaibas is also misleading. If you are interested in a given pundit’s advice, then why are you measuring the value of such advice by the stock price performance of the stock he manages. As someone who frequently cautions against reasoning from a price change, are you not skeptical about the wisdom of measuring “performance” by relative price movements of a company stock and a benchmark index?

  11. Gravatar of ssumner ssumner
    6. April 2014 at 18:04

    Scott, I think you misunderstood the point. I’m happy to agree with the claim that it’s really hard to judge the performance of hedge funds, and that one needs to be careful. I was responding to others who claimed hedge funds had done well. By their metrics it is no longer true.

    A, I don’t understand your first comment. This post was a follow-up to my 2009 post. That’s the sense it which it is “out of sample.” People said “Look Buffett did great, doesn’t that disprove EMH” I said (in 2009) no it was luck. Then I said “if I’m right he won’t do great going forward.” And he didn’t. Not sure why that’s so unreasonable.

    I also don’t understand your second point. My discussion with Vaidas had nothing to do with a stock that anyone was managing, it was about Robert Shiller.

    Also see my reply to Scott (above), as you may have missed the point of the post. This is in response to critics of the EMH. They set the rules, I’m simply playing their game. If you want to argue that their game is silly and simplistic and should be ignored, I’m all on board.

  12. Gravatar of Lorenzo from Oz Lorenzo from Oz
    6. April 2014 at 18:19

    Frances: their is no single US housing market, there are lots of different housing markets. And some of them are every bit as supply-constrained as the UK market for much the same reasons–regulatory land rationing.

  13. Gravatar of Lorenzo from Oz Lorenzo from Oz
    6. April 2014 at 18:24

    Frances: That should be “there is”. Also, the housing bubbles were much stronger in some US markets than others. While residential “over-building” was a problem in only a few.

    In Ireland, for example, there were massive supply constraints in more desirable areas (driving up housing prices) and over-building in less desirable ones responding to those price signals (leading to empty estates after the crash and migrant workers left).

    In Australia, we have the most expensive housing in the Anglosphere (outside Hong Kong) because government regulation can create a land “shortage” even in Australia. The old Soviet joke that if the Soviet Union took over the Sahara within 5 years there would be a shortage of sand is not a joke in the Australian housing market.

    Of course, we adopted the British system of land rationing (sorry, building-by-approval).

  14. Gravatar of Daniel J Daniel J
    6. April 2014 at 19:41

    Hey Scott, Michael Pettis has an interesting new post. I was wondering to your thoughts on it. It reminds me of your modeling using logically connected sentences and building up from there, he does that very well in this post.


    Even though you disagree with the whole inequality debate because of your focus on consumption and not income, he makes some points that are hard to ignore. I would take issue with him seeing the US as dependent on trade as other countries, yet the weird things happening to the lower and lower interest rate kept kicking the back of my head until I couldn’t ignore his point on savings rates.

  15. Gravatar of Jim Glass Jim Glass
    6. April 2014 at 19:45


    Buffett’s Alpha

    Andrea Frazzini, David Kabiller, and Lasse Heje Pedersen

    November 21, 2013


    Berkshire Hathaway has realized a Sharpe ratio of 0.76, higher than any other stock or mutual fund with a history of more than 30 years, and Berkshire has a significant alpha to traditional risk factors.

    However, we find that the alpha becomes insignificant when controlling for exposures to Betting-Against-Beta and Quality-Minus-Junk factors.

    Further, we estimate that Buffett’s leverage is about 1.6-to-1 on average. Buffett’s returns appear to be neither luck nor magic, but, rather, reward for the use of leverage combined with a focus on cheap, safe, quality stocks.

    Decomposing Berkshires’ portfolio into ownership in publicly traded stocks versus wholly-owned private companies, we find that the former performs the best, suggesting that Buffett’s returns are more due to stock selection than to his effect on management. These results have broad implications for market efficiency and the implementability of academic factors.

    Buffett has himself always endorsed the EMH as a very close approximation of reality (as the Times story notes).

    Buffett also notably publicly wagered a few years back that in coming years hedge funds would under-perform the S&P 500 for that very reason, because their managers can’t systematically produce returns that cover their fees, which is just how things have gone.

    If you read Buffett’s biography you find that even if his own alpha-after-leverage isn’t as miraculous as is so often claimed, he is still an extraordinary investor in many ways. E.g. when he was running his first investment fund in his early days, at the top of the 60s boom market with him producing ‘bonus’ returns on top of that, causing investors to beat on his doors trying to get into it, he closed it down because he felt couldn’t find any investments reasonably enough priced to buy. You just won’t find many other investment managers who pass on big guaranteed ‘fee profits’ like that. (For 9 out of the 10 of the rest, that’s exactly the moment of opportunity they’ve been waiting for to take the money and run.)

  16. Gravatar of Steve Steve
    6. April 2014 at 20:07

    “To see why, assume Buffett is only able to find $5 billion in good investments each year, but has $50 billion to manage. Then put the $5 billion into the good investments, and index fund the other $45 billion.”

    This isn’t a valid counterargument because Buffett would pay, perhaps, $15 billion in tax in order to implement this strategy. If your null hypothesis underperforms Buffett by $15 billion after tax, then it’s a bad null hypothesis. Or Buffett is a genius for discovering that it pays to underperform sometimes in order to cut Uncle Sam out of the loop.

    Diminishing returns to scale is a universal challenge of active money management. The people who can beat the market don’t want your money and don’t show up in the academic studies. People who can’t beat the market do, and do.

  17. Gravatar of Vivian Darkbloom Vivian Darkbloom
    6. April 2014 at 20:09

    “Further, we estimate that Buffett’s leverage is about 1.6-to-1 on average. Buffett’s returns appear to be neither luck nor magic, but, rather, reward for the use of leverage combined with a focus on cheap, safe, quality stocks.”

    The first part of that sentence makes a very good observation even though the “focus on cheap, safe, quality stocks” appears to me as the essence of good “stock picking”. It is not sufficient to compare the performance of Berkshire stock as a proxy for Buffett’s stock picking ability. Yet, it appears that most if not all studies of Buffett’s performance as a stock picker do just that by comparing the performance of that stock with, say, the S&P 500 (and that Berkshire leverage is significantly enhanced by the tax sheltering inherent in making those investments via an insurance conglomerate). The value of Berkshire’s stock would also reflect any dividends distributed by the companies it invests in so it is also necessary to use total returns of an index rather than merely the index itself.

    Does the above-referenced study eliminate the effect of Berkshire’s leverage and tax advantages so that these effects can be eliminated from the overall performance? Why is it valid to compare the performance of Berkshire stock with the performance of “any other stock or mutual fund” as the above study (based on the abstract) appears to do when the latter are generally precluded from using leverage?

    Has anyone ever attempted to simply track Berkshire’s publicly traded stock purchases and sales so these effects can be eliminated? I doubt such a study is possible because Berkshire would not be required to disclose precise purchase and sale dates of may of its investments.

  18. Gravatar of Vivian Darkbloom Vivian Darkbloom
    6. April 2014 at 20:16

    “Nor is it going to fall any time soon, despite the fact that it is probably overvalued by at least 30%.”

    Francis, the two clauses of this sentence need to be reconciled in some manner.

  19. Gravatar of Vivian Darkbloom Vivian Darkbloom
    6. April 2014 at 20:16

    Sorry, I meant Frances.

  20. Gravatar of Steve Steve
    6. April 2014 at 20:53

    Noah’s post on HFT isn’t good because he doesn’t address (and doesn’t seem aware of) the issues with market structure.

    For example, the vast majority of retail order flow is sold to “internalizers”, i.e., people who trade against the retail orders for their own book. In theory this helps the retail customers, because it allows them to get prices in between the bid-ask (price improvement). Retail brokers (including Vanguard and Fidelity) collect payment for selling their customer order flow. But why would people pay for order flow and then offer price improvement? The theory is that retail is dumb money, so the “internalizer” can make money by offering a tighter bid-ask. But this also means that there’s an adverse selection issue as public limit orders get adversely selected. And the price improvement for retail is miniscule relative to the skimming. On the other hand, discount brokers may be able to offer competitive low commissions precisely because they are allowed to skim from customer orders.

    Here’s an interesting quote: “For example, you can look on Ameritrade’s 606 report for Q2 2010, and see that 83% of market orders are sold to Citadel for about .0015/share on average.” http://blog.themistrading.com/the-internalizers-and-the-flash-crash-let%E2%80%99s-talk-real-villains/

    Then there are issues with exchanges selling price feeds, co-location, etc. Each of these issues needs to be studied separately. The term HFT is a catch-all that encompasses gee-whiz fast trading as well as various regulatory, architectural and order-routing arbitrages.

  21. Gravatar of Benjamin Cole Benjamin Cole
    6. April 2014 at 21:16

    Excellent blogging.

    One quibble even though it is within the context of agreeing.

    Warren Buffet’s Berkshire Hathaway was a long record of acquiring private companies, the Diary Queens, Benjamin Moore paints etc.

    Sometimes an artificial boost can be had through an increase in multiples. Buffett can buy a Diary Queen at 9 times earnings, but the stock market values Berkshire Hathaway at 12.

    And buying whole private companies is not stock market investment. Obviously, there is a very imperfect market for the sale of very large private operating companies. The owners of which might be enamored of Warren Buffet and his promises not to fire everybody.

    Despite all that, Warren B, still can’t outperform the market….

  22. Gravatar of Kevin Erdmann Kevin Erdmann
    6. April 2014 at 22:12

    So, let me get this straight….your evidence that EMH holds and that nobody can reliably outsmart the consensus is that over the past 5 years you’ve…..reliably outsmarted the consensus,…and you outsmarted the consensus by repeatedly taking the odds that EMH would hold.

    Excuse me while my head explodes. 😉

    At the risk of petty self-promotion, I prattled on about EMH recently:

  23. Gravatar of Vaidas Urba Vaidas Urba
    7. April 2014 at 03:52

    “I’m not interested in his model, I’m interesting in his advice. That’s why he’s famous-advice. Anyone can fit a model to past data, but I want to know if that model gives good advice going forward. His did not.”

    He got his Nobel for his model, not for his personal advice. We should ignore his public pronouncements and we should check his model instead to see if it gives good advice going forward. In March 2009 I was not aware what Shiller was saying, but I knew that his model implied that you should hold more stocks than you had one year ago. And just before the market lows, Grantham (who uses Shiller Cape-like model) sent out his famous ‘Rigor Mortis’ letter:

  24. Gravatar of J Mann J Mann
    7. April 2014 at 05:00

    Major Freedom and Kevin,

    IMHO, the question for the EMH is whether Scott can consistently make money trading on his insight that the EMH applies.

    The EMH doesn’t predict that no one can ever be right about anything, it predicts that you can’t consistently beat a normal market using public information.

    If you come up with a way to profit if an average of mutual funds fail to outperform index funds, let me know – I will gladly invest.

  25. Gravatar of ssumner ssumner
    7. April 2014 at 05:08

    Daniel, I don’t agree with his claim that inequality and savings rates impact AD. In my view monetary policy determines AD.

    Jim Glass and Vivian. Good comments.

    Steve, I don’t understand your point about taxes, surely indexed funds are not taxed at a higher rate than managed funds?

    Your last point is something I sometimes hear. But I must respond to existing claims of market inefficiency. And that’s what I’ve done. After I shoot them all down people claim there are other secret geniuses that no one knows about.

    Or maybe they are just other lucky people.

    Steve, Your second point seems consistent with Noah’s post–there are lots of complex issues that need to be studied. He mentioned some of them, but didn’t claim to mention them all.

    Ben, Good point.

    Kevin, Yup. An out of sample test that went very, very well from my point of view.

    Vaidas, I believe he got his Nobel for both. There are lots of good models out there, that perform well using historical data. His fame came from his public statements. People like Shiller get judged by their statements. If he has created a good model, but is unable to make good predictions from HIS OWN model, how likely is it that the model would be useful to me?

    The model doesn’t tell you exactly when to buy or sell–that’s a judgment call. I find it interesting that the creator of the model is not able to use it effectively. That suggests to me that the model is useless. Alternatively, how has the model done “out of sample,” since I started blogging in 2009? Even if the model was “correct” in March 2009, it’s been mostly wrong.

    Suppose someone invents a new car, but is unable to drive it. Would that make you want to try out the new car?

  26. Gravatar of Michael Byrnes Michael Byrnes
    7. April 2014 at 07:28

    Jim Glass wrote:

    “Buffett also notably publicly wagered a few years back that in coming years hedge funds would under-perform the S&P 500 for that very reason, because their managers can’t systematically produce returns that cover their fees, which is just how things have gone.”

    Buffett is winning his bet, BTW.



    The S&P 500 underperformed the hedge funds he bet against in the first year (which started on 1/1/2008) but after 6 years (of 10) he was ahead by about 30%.

  27. Gravatar of Steve Steve
    7. April 2014 at 07:29

    “I don’t understand your point about taxes, surely indexed funds are not taxed at a higher rate than managed funds?”

    Suppose you have two investment options:

    Option A: Outperforms market by 5% annually for 5 years, but has random unforecastable variation vs market thereafter.

    Option B: Index fund.

    Obviously, you choose option A and get the initial 25% outperformance kick. But then you are stuck with option A forever, unless you want to pay big taxes to switch to B. If the expected return is the same after five years, why switch?

  28. Gravatar of Steve Steve
    7. April 2014 at 07:29

    “I don’t understand your point about taxes, surely indexed funds are not taxed at a higher rate than managed funds?”

    Suppose you have two investment options:

    Option A: Outperforms market by 5% annually for 5 years, but has random unforecastable variation vs market thereafter.

    Option B: Index fund.

    Obviously, you choose option A and get the initial 25% outperformance kick. But then you are stuck with option A forever, unless you want to pay big taxes to switch to B. If the expected return is the same after five years, why switch?

  29. Gravatar of Steve Steve
    7. April 2014 at 08:14

    “Your second point seems consistent with Noah’s post-there are lots of complex issues that need to be studied. ”

    The problem with Noah’s post is that he says there’s lots of stuff to study, but he hasn’t studied that stuff. It would be better if he Hadahpinion.

    The thing about “HFT”–really a catch-all for today’s byzantine market structure plus all the paid data feeds and order flow–is that no one would design it from scratch the way it is today.

    The good thing about HFT is that markets are relatively robust to flawed structure. People figure out how to adapt, and manage around the available order types and toll takers. People have strategies like using market orders, small order sizes, and dark pools.

    The problem with HFT is mostly the piss-off factor. People don’t like spending so many resources studying market structure and investing in trading systems.

    Imagine if there were a $20 toll on the Mass Pike at Allston. And then everyone ran out and bought GPS systems and real-time traffic monitoring systems to figure out if Soldiers Field Road, or Beacon St, or Route 9, or back roads are better options on any given day. That’s HFT in a nutshell.

  30. Gravatar of Doug M Doug M
    7. April 2014 at 08:14

    Buffet was at the top of the forbes 400 list in 1990. If you are going to say, “Americans should not see out-sized returns, once they have reached that pinnacle of success”, you should look back 20 years in Buffets case.

    But, there is a lot more to Buffets success than stock picking. Much of his out-performance is “structural.” Buffets primary holding is a group of insurance companies that have cheaper funding than most other investors. Cheap money is the secret to his success.

  31. Gravatar of TravisV TravisV
    7. April 2014 at 08:21

    First Yglesias video for Vox:


    Good part: He implicitly associates higher interest rates with easier money.

    Bad part: Yglesias says “trying to reduce debt might actually make the debt situation worse.”

    He should just say “easier money makes the debt burden smaller / tight money makes the debt burden larger.”

  32. Gravatar of Mark A. Sadowski Mark A. Sadowski
    7. April 2014 at 08:51

    Off Topic.

    James Hamilton discusses the usual research on QE’s effect on the term structure.


    “A number of other studies have looked at the variation over time in the maturity structure of U.S. Treasury debt and have found statistically significant correlations between the maturity structure and relative yields; see for example Kuttner (2006), Gagnon, et. al. (2011), Greenwood and Vayanos (2013), Doh (2010), and Hamilton and Wu (2012) (ungated version here).”

    I respond in comments by discussing my results, in which I replicate Gagnon et al’s estimates during the time in which there has actually been QE, and find that QE significantly *raises* the term premium.

    Hamilton also mentions the recent paper by Hayashi and Koeda, of which I have a much more favorable opinion.

  33. Gravatar of Mark A. Sadowski Mark A. Sadowski
    7. April 2014 at 09:12

    Off Topic,

    Gavyn Davies seems to think the latest IMF World Economic Outlook understates the role of loose monetary policy in lowering real long term interest rates.


    “…The IMF says that the main reason for the drop in real rates in the 1980s and 1990s is obvious: the easing in monetary policy that occurred after the 1979-82 Volcker tightening…Note also that quantitative easing by the central banks is hardly mentioned at all among the forces that have held real rates down, which is interesting in view of the amount of attention it has received from investors since 2008. It seems odd that the IMF has assigned almost no importance whatever to QE, but it certainly cannot be the main factor, since two-thirds of the fall in real yields occurred before QE even started…”

    The relevant sections of the IMF WEO can be found on page 7 and 11-13:


    In my opinion it’s typical boilerplate intended to be written on an accessible level. Evidently Davies seems disappointed that they didn’t dumb it down to the “low interest rates equals loose monetary policy” level.

  34. Gravatar of Mark A. Sadowski Mark A. Sadowski
    7. April 2014 at 10:04

    Off Topic.

    It’s pieces like these which make me wonder, when it comes to monetary policy, which is worse, the American or the British press?


    “…Since mid-2008, the Federal Reserve has expanded its balance sheet fourfold. The Bank of England has practised QE on a similarly larcenous scale. In truth, the eurozone has been doing plenty of QE too, the ECB’s balance sheet having surged from around €1,700bn in 2009 to over €3,000bn three years later, before partially falling back. But so sensitive is the German public about money-printing and its eventual inflationary impact, that euro-QE has been cleverly masked. Various schemes, hidden behind technical names such as ESM and Target-2, have expanded the eurozone’s money supply in a way the currency markets understand, even if ordinary voters don’t.

    So far, the Anglo-Saxon world’s overt QE has out-printed the eurozone’s sotto voce variant, so the currency advantage has been ours. Now, official determination to get the euro back over $1.40 is so strong that even Berlin is signalling it might back explicit QE…”

    The size of the ECB balance sheet is not very informative. A more relevant measure is the size of the monetary base. Whereas the Fed’s monetary base has increased 349% since August 2008, the ECB’s has only increased by 29%.

    The reason why the size of the ECB’s balance sheet is not very informative because the ECB has always had a lot of liabilities that are not part of the monetary base.

    On average from February 1999 through August 2008 only about 63% of its liabilities were part of the monetary base. This is in contrast to the Federal reserve where the corresponding figure is 93%. The most recent figures for each institution are 54% for the ECB and 91% for the Fed. In short whereas the size of the balance sheet is a fair approximation of the size of monetary base in the case of the Fed, it is nearly twice the size of the monetary base in the case of the ECB.

    In the Fed’s case the additional liabilities are: 1) reverse repurchase agreements with foreign official and international accounts, 2) non-reserve deposits 3) liabilities associated with Maiden Lane, etc.

    In the ECB’s case the additional libilities are: 1) revaluation accounts, 2) non-reserve deposits, 3) current accounts and deposits with the Eurosystem held by central banks, other banks, international/supranational institutions and other foreign depositors, 4) current accounts and liabilities under repo transactions which are usually investment transactions using foreign currency or gold, 5) paid-up capital, legal reserves, other reserves and retained earnings, 6) counterparts of special drawing rights allocated by the IMF, etc.

    These liabilities have almost nothing to do with monetary policy and the ECB has a lot more kinds and quantities of such liabilities in part due to the fact that the eurozone is an international currency area.

    The ECB monetary base is currently €1,162.8 billion, which is down 34.5% from its peak level in July 2012:


    Reserve balances (current accounts and deposit facility) increased from €215.1 billion in August 2008 to €230.6 in February 2014, or by €15.5 billion.

    So if the ECB has done any “soft voiced” QE, it’s a tiny little whisper compared to the ad alta voce of the Fed and the BOE.

    In percentage terms, the ECB’s monetary base has increased less in the 67 months since August 2008, than it did in the preceding 32 months (December 2005 to August 2008). If the ECB has been doing lots of QE since the Great Recession, what on earth was it doing before?

  35. Gravatar of Edward Edward
    7. April 2014 at 10:31

    Krugman has a new column about oligarchy and inflation.
    Your thoughts, Scott?

  36. Gravatar of brendan brendan
    7. April 2014 at 10:34

    It’s hard to convey to non-investors just how deleterious scale (of assets under management) is on returns.

    Markets are extremely efficient. Most investors have no good ideas, ever. Great investors have a maximum of 4-5 good ideas at any one time. And market efficiency varies inversely with the scale/liquidity of the asset in question.

    Genuine alpha is rare because it requires: a great investor + small scale + patience + an informed investing base.

    Most alpha claims will be false because, a) real alpha is rare, and b) performance measurement is incredibly noisy.

    So Scott is almost always right about the EMH. But Buffett generated real alpha, and so do many others.

    *Among historically successful investors there is high correlation in their approaches. For example, they disdain market timing, they’re willing to patiently hold cash and wait for opps: “fat pitches”. High similarity among successful investors is evidence against the EMH; the kind of evidence that statistical tests do not account for.

  37. Gravatar of Mark A. Sadowski Mark A. Sadowski
    7. April 2014 at 10:38

    Scott has a post at Econlog that addresses Krugman’s related blog post (“Oligarchy and Monetary Policy”):


  38. Gravatar of Scott Sumner Scott Sumner
    7. April 2014 at 11:03

    Steve, Noah’s opinion (which is correct in my view) is that no one knows whether your “in a nutshell” analogy captures HST, or whether one of many other perspectives best captures its effects. It’s possible that many of the things (both good and bad) that people say about HST are true, but people have no idea how these things all net out.

    I understand your point about taxes in the more recent comment, but don’t see how that point relates to anything I said in my post.

    Doug, Yes, I’ve done a post on his cheap funding. Of course the test of alpha should look at many billionaires, to be more statistically signficant. But you are right that he’s done well until recently, well after first becoming very rich.

    Mark, That IMF quote is just appalling.

    Thanks for the data on the ECB monetary base. I always have have trouble interpreting eurozone data.

    Edward, Check over at Econlog.

    Brendan, I mostly agree. I would just add that I am agnostic on whether Buffett had alpha (ex ante) or just luck. Ex post he certainly had lots of alpha.

  39. Gravatar of Brian Donohue Brian Donohue
    7. April 2014 at 11:53


    I am sympathetic to the EMH but you sound like an amateur analyzing Buffett. You really should read his shareholder letters- I think you would learn some good stuff.

    He has repeatedly said over the years that the amount of capital he is forced to allocate necessarily limits his potential for return, and that if he could start over with a small amount of capital, he could again make outsized returns more easily. Feel free to disbelieve him, but don’t think you’ve demonstrated anything here.

    As far as the past five years, Buffett is again very clear that he expects to underperform the S&P 500 during bull markets. Considering the past five years has been a pretty big bull market… how does Buffett stack up over the past six years?

  40. Gravatar of TravisV TravisV
    7. April 2014 at 12:45

    David Glasner reviews Jeffrey Hummel’s paper back in 2011:


  41. Gravatar of TravisV TravisV
    7. April 2014 at 12:47

    Noah Smith:


    “Why didn’t the Fed fully reveal FRB/US model before now? It always seemed to me that it was basically because of embarrassment. Academic macroeconomists haven’t used or studied this type of model in decades (having abandoned everything else in favor of DSGE). In 2010, Chris Sims appeared to call models like FRB/US “something close to a spreadsheet”. Since most Fed employees are drawn from the same pool of people as academic macro (and interact with academic macroeconomists quite frequently), the fact that they use something like FRB/US must have been a bit embarrassing.

    So if my guess is right, the Fed’s publication of FRB/US indicates that the embarrassment is mostly gone. That is kind of interesting.”

  42. Gravatar of ed ed
    7. April 2014 at 13:30

    I suppose you might be right about Buffet, but I wouldn’t be so sure. I believe Berkshire has long had negative beta, so you’d expect under-performance in the type of up markets we’ve had for the last five years. I expect when you get a down year you’ll probably see Berkshire outperform again.

    (I see commenter Brian Donohue made essentially the same point.)

  43. Gravatar of Jim Glass Jim Glass
    7. April 2014 at 13:48

    Speaking of markets and HFT, Michael Lewis’s book claiming “the stock market is rigged” via HFT is reviewed at Slate by Felix Salmon.

    (Not to give away spoilers, but it starts out a bit less than laudatory and gets less so from there.)

  44. Gravatar of TravisV TravisV
    7. April 2014 at 19:26

    Benjamin Cole,

    Mr. Doom and Gloom! Check out Prof. Sumner’s forecasts for various countries back in July 2013!



    I’m glad to see Sumner retain Friedmanite optimism!

  45. Gravatar of Vaidas Urba Vaidas Urba
    8. April 2014 at 00:13

    “The model doesn’t tell you exactly when to buy or sell-that’s a judgment call. I find it interesting that the creator of the model is not able to use it effectively. That suggests to me that the model is useless.”

    There was a WSJ interview where Shiller said :’Q: What’s your personal track record: Are you more up than down? A: I have never done a personal analysis. I have to do that. But I believe that I’ve done well in timing the market, although not perfectly.” Many people have criticized Shiller for not measuring his own performance. Let me defend him. Performance measurement is irrelevant, and may be harmful psychologically, and some say performance measurement is one source of market inefficiency. However, it is extremely important to update your model when you get new evidence. I checked his website, he is constantly updating his model with new data. If you focus on specific dates and on measuring performance you lose statistical validity, and you are attacking a straw man – Shiller does not say that stock market is that predictable.

    “Alternatively, how has the model done “out of sample,” since I started blogging in 2009? Even if the model was “correct” in March 2009, it’s been mostly wrong.”
    We have observed some extreme movements, movements that are more extreme than changes in fundamentals. On one hand, extreme movements support Shiller’s theory of market inefficiency vs fundamentals. On the other hand they can reduce long term market predictability. What is the net effect? I don’t know, you actually need to do some calculations. It is also interesting to consider international experience, as US stockmarket experience is a bit of outlier.

    There is a very interesting interview with John Campbell who worked both with Shiller and Cochrane.
    Whole interview is very interesting, however the key point is that Shiller and Fama don’t really disagree on the degree of market forecastability. They disagree on sources of that forecastability. Are the reasons behavioral, or are markets forecastable because there are efficient changes in risk premia.

  46. Gravatar of ssumner ssumner
    8. April 2014 at 05:10

    Brain, You said;

    “I am sympathetic to the EMH but you sound like an amateur analyzing Buffett. You really should read his shareholder letters- I think you would learn some good stuff.

    He has repeatedly said over the years that the amount of capital he is forced to allocate necessarily limits his potential for return, and that if he could start over with a small amount of capital, he could again make outsized returns more easily.”

    If you are simply going to call me ignorant, but not tell me what is wrong with my argument, then there is really nothing for me to respond to. In contrast, I anticipated your argument, responded to it even before you made it, and you seem unable to counter.

    And yes, I am an “amateur” in analyzing Buffett, how could it be otherwise?

    Ed, So what’s his cumulative return since 2007?

    Vaidas, I know lots of people who tell me they’ve done well timing the market. I don’t believe any of them.

    The reason I put a lot of weight on Shiller’s recommendations is that I’m a pragmatist. I know that anyone can construct a model that forecasts well looking backwards, even if the underlying data is a random walk. But if the true model is a random walk, they’ll do poorly in forecasting out of sample. In that case I’d expect them to continually tinker with their model, revising it as they go along, to “fix” the problems. Here’s what you said:

    “However, it is extremely important to update your model when you get new evidence. I checked his website, he is constantly updating his model with new data.”

    That’s precisely the problem, he doesn’t have a correct model, but rather a model fitted to past random data. That’s why he continually updates it. My claim is that essential all market anomaly research is data mining, and hence is worthless. It’s only out of sample performance that matters. I do think that Shiller’s model is far better than most other data mining. It’s an impressive piece of work. But it still is not useful for forecasting future stock movements (in my view) and hence is useless as an anti-EMH argument. It may be useful for other purposes, but it’s not useful as a critique of the EMH. In that sense I agree with Fama’s interpretation.

    Many people misunderstand the EMH. They wrongly believe it says “you can not find anomalies in the data, you cannot find statistically significant patterns in past data.” That is NOT a prediction of the EMH. Thus 99% of anti-EMH papers that get published are testing the wrong model.

  47. Gravatar of brendan brendan
    8. April 2014 at 06:25

    Scott, you’re in that class with David Friedman, Eliezer Yudkowsky and Robin Hanson: people too smart/wise/rational to disagree with comfortably. I check myself when my opinion diverges from yours. (Tyler Cowen made this same point long ago: don’t disagree w/ Sumner.)

    But your version of the EMH is too extreme, a bias likely caused by the ubiquity of stupid anti-EMH arguments you rebuff.

    You don’t want to be confused with these academics still trying to rescue the CAPM, even though the sign of the BETA-Return relationship is empirically wrong!

    If I could recommend a slight tweaking to your EMH position, it’d be to give up the idea of micro-efficiency. Guys regularly uncover frauds from public info. Stock picker performance is cross sectionally persistent. Buffett really is legit, probability 99.99%.

    But successful market timers are mostly lucky.

    (John Hussman, a guy you linked to (from me) awhile back is a market timer, killed it 2000-2008, crashed and burned since. That is typical among them.)

  48. Gravatar of Jason Jason
    8. April 2014 at 06:38

    Wow, just wow. I’m going to have to disagree with you on this one Scott. Where to begin? Why are the last 4 years out of a 49 year winning streak significant at all?
    I have no idea why buffet doesn’t work with smaller pools of capital. Perhaps if he did, he’d still be outperforming. The article also mentioned that buffet buying entire businesses as we’ll, which is comparing an apples to… Cucumber (!) comparison when you’re measuring that against the s&p.
    Finally, it’s NOT just buffet. One of the things I like about him is that he’s quite modest for a billionaire. He conceded in an interview in the 80s that HIS career could be luck by itself, but not the careers of value investors as a whole. (I’m not talking about academic behaviorists like Schiller who ANALYZE the market based on p-e multiples, but self identified value investors like buffet and his mentor. Benjamin graham. They tend to statistically outperform.

    I have a theory about why people get so mad 🙂 when you critique Buffet. (Not that I am) it’s like a group of people looking at a Picasso in a museum oohing and aahing and one guy at the back snarking: “that’s not so impressive, if any of us spattered paint in a random pattern we’d have an equal chance at creating a masterpiece. It’s just luck.”
    Of course they would be pissed! 🙂

  49. Gravatar of Jason Jason
    8. April 2014 at 06:44

    Finally not all of us have the discipline, the Time horizon or the start up capital to become the next buffets, but if one of us could have some of those things, perhaps we might have a chance

  50. Gravatar of Mark A. Sadowski Mark A. Sadowski
    8. April 2014 at 07:02

    Off Topic.

    Interesting quote of Yves Mersch, who is an ECB Governor from Luxembourg:


    “…On Monday in London, Yves Mersch played down the risk of deflation and said that he didn’t see an asset purchase program, known as quantitative easing, as an imminent policy step for the ECB.

    “QE is, above all, a theoretical concept,” he said during a speech in London. “From theory to implementation is a long way.”

    The ECB sees “no imminent risk of deflation,” he added…”

    I agree with him on the deflation risk. But yoy HICP inflation is currently 0.5% and their goal is supposedly 2%. How badly do they have to miss their target before they admit there’s a problem?

  51. Gravatar of Mark A. Sadowski Mark A. Sadowski
    8. April 2014 at 07:09

    Off Topic.

    Here’s a depressing paper, if it’s an indication of current IMF thinking:


    Basically it asks the question, should monetary policy be saddled with additional policy goals that takes its attention even further away from nominal stability?

    The answer the paper gives is a bunch of equivications. But it’s sad that they’re even posing the question, given what the last six years have been like.

  52. Gravatar of Mark A. Sadowski Mark A. Sadowski
    8. April 2014 at 07:48

    Off Topic.

    Kunio Okina delivered an anti-Abenomics speech recently.


    “…Kunio Okina, a long-time BOJ official widely considered Japan’s most influential monetary economist in the 1990s and 2000s, delivered a speech Friday dissecting what he called “the Achilles’ heel of Abenomics,” a reference to the economic revival plan of Prime Minister Shinzo Abe. Abenomics relies heavily on turbocharging the restrained monetary policy long advocated by Mr. Okina and his disciples…”

    “…At one point in the speech, Mr. Okina compared current Japanese economic policy to the recent disastrous experience of Zimbabwe, which was plagued by rampant inflation and colossal government debt…”

    “…Here’s why. The main ammunition for the Kuroda bazooka is the large-scale purchase of Japanese Government Bonds. The BOJ injects money into the economy by buying JGBs from the banks that hold them. That pushes up the price “” and drives down the interest rate “” on the bonds. That makes it dirt cheap for the Japanese government to keep borrowing, which is convenient for Mr. Abe, since Japan’s government is the most indebted on earth, except, perhaps, for Zimbabwe.

    What happens if and when Mr. Kuroda succeeds in hitting his target of 2% inflation? At that point, he has a choice, Mr. Okina said in his speech. The BOJ could curb bond buying to steer interest rates higher. But that would raise the cost to the government of its outsized borrowing, “severely aggravating government debt problems.” It would also, he said, wreak havoc on the BOJ’s own balance sheet, since the price of the JGBs in its portfolio would fall.

    Or Mr. Kuroda can choose to keep interest rates low by continuing his big bond buys, and then “it will be impossible to halt inflation overshooting,” Mr. Okina said. In other words, Japan would have simply replaced a deflation problem with inflation…”

    It’s worth noting that Zimbabwe is *not* the most indebted country on the earth. Not even close. General government debt as a percent of GDP was 58.7% in 2013 according to the IMF. High debt is correlated to low inflation, and low debt is correlated to high inflation.

    The scenario that Okina describes is just not credible. No doubt you saw me presenting historical data in comments recently on this very scenario. In 80% of historical cases of advanced nations with high debt levels (above 90% of GDP) and inflation rates of 5% or more, real long term bond yields were negative, and usually significantly so.

    Here’s Okina’s slides:


    Slide #46 is interesting because it gives the parameters for Japan’s Cabinet Office Taylor Rule. Okina argues that this implies that the call rate is going to rise to 3.7% by FY 2017 when the Cabinet Office forecasts 10-year bond rates will yield 2.8%. He says it’s more realistic that long term bond yields will rise to 5%.

    Personally I will be floored if the bond rate rises from 0.6% to 2.8% in just three years. There’s no way it will ever rise to 5%. This is quite simply ridiculous.

    Note also on slide #48 there’s a highly alarming graph suggesting that central bank losses from QE could be as high as 4% of GDP in the US, 6% of GDP in the UK and 7% of GDP in Japan. Where do these predictions come from?

    The IMF of course. You’ll find them on page 22:


    The thing that alarms me about these predictions is not their size (I find them wholely implausible) but the fact that the IMF evidently thinks central banks should be run like a for profit corner savings and loan.

  53. Gravatar of Econ/Finance Fan Econ/Finance Fan
    8. April 2014 at 11:11

    Prof. Sumner,

    I am a huge admirer of your writings on monetary policy and I have learned a lot from them. However, your comments on Buffett’s performance are somewhat off-center. Regardless of what one thinks of EMH, we should at least subject the NYT article to more scrutiny.

    Here is a comparison of Berkshire (A stock) versus S&P 500 (I have used SPY as the ticker) for various start dates. All data from the website: http://www.buyupside.com/calculators/dividendreinvestmentbackportfoliocalculatorinput.php, and we assume that $1 is invested into BRK-A and SPY and we measure its final value as of end of day yesterday (with dividends reinvested for SPY).

    1/1/2002-now: BRK-A: $2.50, SPY: 2.07
    1/1/2003-now: BRK-A: $2.73, SPY: 2.68
    1/1/2004-now: BRK-A: $2.06, SPY: 2.00
    1/1/2005-now: BRK-A: $2.05, SPY: 1.88
    1/1/2006-now: BRK-A: $2.06, SPY: 1.71
    1/1/2007-now: BRK-A: $1.68, SPY: 1.49
    1/1/2008-now: BRK-A: $1.36, SPY: 1.53
    1/1/2009-now: BRK-A: $2.06, SPY: 2.48
    1/1/2010-now: BRK-A: $1.61, SPY: 1.87
    1/1/2011-now: BRK-A: $1.51, SPY: 1.53
    1/1/2012-now: BRK-A: $1.57, SPY: 1.47
    1/1/2013-now: BRK-A: $1.27, SPY: 1.26
    1/1/2014-now: BRK-A: $1.09, SPY: 1.04

    Sure, there are certain periods where the S&P has outperformed BRK-A, but the overall record is clearly in BRK-A’s favor (and even the period where S&P outperformed can be “rationalized” by the argument that the S&P index fell far more than BRK-A when the financial meltdown happened and therefore was relatively cheaper than Berkshire).

    Notwithstanding our difference of opinion on Berkshire, many thanks for your outstanding blogging.

  54. Gravatar of Patrick R. Sullivan Patrick R. Sullivan
    8. April 2014 at 11:48

    The French have words for it;


    ‘”The efforts that we are making to reduce our deficit, to implement structural reforms, improve the competitiveness of our businesses and lower labour costs should not be swept away by a too-high level of the euro,” he [the new Prime Minister] said.

    ‘Calling for an EU-wide growth initiative, Valls noted that the path to growth involved “large investments and policies to create jobs specifically targeting youth. Otherwise,” he warned, “all our efforts to reduce the deficit will be in vain,” he said.

    ‘Comparing the European Central Bank’s policies with its counterparts in the US and Japan, Valls noted that the ECB was less supportive of economic growth.’

  55. Gravatar of Kevin Erdmann Kevin Erdmann
    8. April 2014 at 15:59

    Econ/Finance Fan,

    I hadn’t looked that closely at BRK before. The kicker is that BRK’s beta is around .3 or .4.

    The Yahoo data going back to 1980 makes BRK look like a very high alpha stock in the 80s & 90s with normal beta, and it looks almost bond-like in the late 1990s & 2000s.

  56. Gravatar of Phil Phil
    8. April 2014 at 17:16

    You’ve made a fundamental error I think Scott. The error is that you haven’t adjusted for Buffett’s huge cash pile, or (relatedly) risk.

    This article has a go at both and comes to the opposite conclusion to your conclusion:


    Also, I don’t think you have understood the tax point mentioned above.

    Buffett owns some stocks with enormous deferred capital gains tax liabilities. If he sells those to put the cash into the S&P500, he will crystalise a giant tax bill. He may have calculated that it is better to defer the tax bill indefinitely and underperform the S&P500 over the next x years, rather than crystalise the tax bill and put $45 billion into the S&P500 and pick stocks with the $5 billion.

    Buffett has said he views the deferred tax liability as a non-callable interest free loan from the US government, with restrictions (that he not sell the stock).

  57. Gravatar of TallDave TallDave
    8. April 2014 at 17:42

    There is only one group of investors that consistently outperform the market: politicians, especially those in the U.S. Congress, who can legally trade on insider information and even pass laws to benefit their holdings.

    Buffet has started rentseeking on a massive scale, e.g. investing in oil shipping by rail while trying to kill Keystone XL.

  58. Gravatar of TravisV TravisV
    8. April 2014 at 18:35

    Charles Plosser’s latest speech: Madness!


    “I can be concerned about inflation, but then the question is, what do you do about it? We’re still adding accommodation and we’ve got record-low interest rates,” Mr. Plosser told reporters following a speech. “I don’t believe, for example, that buying even more assets than we are will help our inflation problem.”

    The Fed already has been making large asset purchases, he noted, “and inflation has done nothing but drift down. So it’s not obvious to me that continuing to buy would be an answer to concern about inflation.”


    Mr. Plosser said Tuesday he’s confident inflation will return to the Fed’s 2% target over time, in part because expectations remain well-anchored. “In the short run, I’m not terribly concerned,” he said.

  59. Gravatar of TravisV TravisV
    8. April 2014 at 18:40


    You’re one of the most insightful commenters here, so I was surprised and disappointed to read your paranoid views about Warren Buffett above. Buffett is one of my biggest heroes.

  60. Gravatar of Tony Hansen Tony Hansen
    8. April 2014 at 18:57

    I appreciate your passion for EMH, but if you factor for corporate taxes paid had Berkshire employed the strategy you discussed (10% into the best opportunity, the rest into an index), the taxes would mean they would still underperform the market.

    If you strip out the appropriate level of taxation on the capital growth & dividend components of the S&P results over the 5 years examined, Berkshire would have beaten the S&P in 4 of the last 5 years.

    If he can last long enough to see another 20 or 30% drop, I’m pretty sure the trouncing Berkshire will inevitably have in such a year will remind people that ‘Alpha’ can come in very large irregular chunks in the stockmarket…

  61. Gravatar of Peter Drake Peter Drake
    9. April 2014 at 00:34

    Doesn’t EMH include the possibility that new wisdom is generated as market circumstances change. Isn’t it possible that Buffet was able in the past to be among the first to recognize or generate new wisdom a significant number of times. Each such instance would provide an advantage over a period of time until the knowledge spread to other investors. But continuing to create new wisdom consistently over decades sounds like a hard thing to do. Your actions are closely monitored and the period of advantage would shrink. And as his fundamental outlook is replicated it might be very hard to have unique insights before everyone else.

    Regarding bubbles, I think they happen, in the sense that people base their judgments of value on a mix of factors, some of which are subject to mob-like effects. But financial markets have lots of sophisticated tools and investors who are aware of this and try to adjust.

    But housing isn’t the same. The residential home market is not significantly influenced by a class of professional investors. Not only are the mob-like effects not damped, but letting people walk away from underwater mortgages limits a home buyer’s risk when buying into a bubble. And sub-prime loans with cheap starter rates further encouraged short-term non-professional investing in housing. Calling it investing might be generous. It was more like gambling, or a ponzi scheme. There were people warning about it in 2004, but it kept working for years after that.

    I think part of the problem with housing is that there are no fundamentals. Who is aware of how much it would cost to replace their home if destroyed? The exact same house on the exact same size lot could sell for $240K, $480K, or $720K depending on where it is. You could say it depends on location but you could also say it depends on the expected value of the houses nearby. Once nearby houses go up, so does mine. Trying to parse the reasons why nearby houses went up is very hard. The only signals are price and time on market. And an influx of over-eager gamblers who don’t think they can lose will have a significant effect on both those signals.

    In contrast, in Canada you can’t easily walk away from an underwater mortgage. Buying homes with little or no money down is very hard. As a consequence, people were primarily buying homes for themselves, with mortgages they could afford. There have been minor corrections but without investors dumping houses its hard to see how the housing market _could_ crash without some other factor, like a steep rise in mortgage rates, to force people out of the homes they bought.

  62. Gravatar of Vaidas Urba Vaidas Urba
    9. April 2014 at 02:25

    “My claim is that essential all market anomaly research is data mining, and hence is worthless.”

    What about Fama’s value and size anomalies? Is that just data mining? Shiller’s CAPE and Fama’s value and size anomalies have similar levels of empirical support.

  63. Gravatar of James James
    9. April 2014 at 03:24

    These guys seem pretty good at beating the market:


    You know when a fund refuses to take your money, they have to be doing something right.

    I’d guess that most hedge funds, like most of anything else, are run by pretenders. Assuming so, average hedge fund performance would tell us nothing about the EMH. On the other hand, it only takes one consistent outlier to disprove the EMH.

  64. Gravatar of Scott Freelander Scott Freelander
    9. April 2014 at 03:52


    I think I understand that the purpose of this post is to point to the absurdity of some of the anti-EMH arguments you’ve read, but this is still beneath you. Buffett, aside from being an anecdote, is over 80 years old, and quite possibly over-the-hill.

    Yes, your argument is purposely silly, but it invites much misinterpretation.

    Of course, strong EMH views are self-contradictory for reasons Friedman pointed out decades ago.

  65. Gravatar of Scott Freelander Scott Freelander
    9. April 2014 at 03:58


    I should add that your pro-EMH posts are interesting and otherwise provocative, that particular perspective is not necessarily the best argument in favor of using immediate market reactions to guide policy, even distractions aside. I prefer to say, give the markets what they want, and right or wrong, there are self-fulfilling prophecies critical to cyclical stabilization.

  66. Gravatar of ed ed
    9. April 2014 at 08:49


    I found Buffet’s returns since 2007 that you asked for, but now I see that commenter “Econ/Finance Fan” has already done it. I agree that Buffet has significantly beaten the S&P 500 since the beginning of 2007, though not since the end of 2007.

    When arguing about whether Buffet has alpha, it is really important to realize that a large part of his out-performance over the last 20 years has come during down markets. This is consistent both with his measured market beta, and with Buffet’s stated investment philosophy.

    It’s also important to realize that Buffet doesn’t control his stock price, so the return to holding Berkshire over a short period will be an inaccurate representation how much value Buffet has created (or not created).

  67. Gravatar of Alfred Alfred
    9. April 2014 at 12:01

    “To see why, assume Buffett is only able to find $5 billion in good investments each year, but has $50 billion to manage. Then put the $5 billion into the good investments, and index fund the other $45 billion. If he truly had alpha he’d still be outperforming the market.”

    You have to consider that many portfolio managers get rated on their ‘tracking error’ (how correlated the PM’s results are to the market). A portfolio composed of $45B index fund $5B good stock pitches will have a tracking error of at least 90%. No PM wants on his record as potential investors will rightfully wonder why give their money to someone that will only track the market. So there is a strong disincentive to do this.

  68. Gravatar of Michael Sullivan Michael Sullivan
    9. April 2014 at 12:40

    Scott, I am skeptical of the idea that Buffet’s outperformance represents no evidence at all. He was a famous investor by the 1970s and was clearly identified as an outperformer by 1980. Since then his portfolio has grown about 94fold, while the s&P 500 has grown about 45fold. Annualized, that’s almost 14% versus 11.5% per year.

    Now, in searching for these results, I discovered a paper that suggests something interesting about Buffett’s investing. His return outperformance can be explained by his leverage ratio of 1.6-1. The hard to explain part is his volatility which is actually less than the market, despite this leverage, giving him a ridiculous sharpe ratio.

    I’m inclined to think that his strategy has produced some real alpha at some point. I don’t think you can dismiss his record after 1980 as data-mining, as he was clearly identified at that point as a brilliant investor.

  69. Gravatar of Scott Sumner Scott Sumner
    9. April 2014 at 13:33

    brendan, The best way to uncover fraud is to believe in the EMH. If the SEC had believed in the EMH they would immediately have understood that Madooff was a fraud, when the evidence was presented. Instead thay ignored that evidence.

    Stock picker returns are not serially correlated in the studies that I have seen.

    Regarding the odds of Buffett being lucky–note that the odds of winning megabucks is 100 million to one. Does that mean the winner was not lucky?

    Jason, You asked:

    “Wow, just wow. I’m going to have to disagree with you on this one Scott. Where to begin? Why are the last 4 years out of a 49 year winning streak significant at all?”

    For the simple reason that an out-of-sample test is about a million times more statistically significant than a test constructed via data mining.

    And no, I’m not at all surprised that other investors who bought the same sort of stocks that Buffett bought also did well.

    Mark, I agree that the focus should not be on deflation, but too low inflation (actually too low NGDP.)

    And when the Japanese worry about Zimbabwe, it makes one think of Hawtrey’s “Fire, fire in Noah’s flood.”

    Econ/Finance guy, Let’s say you are right that Buffett has done about as well since 2003, not worse. I still say that supports the EMH, as he used to be outperforming by a wide margin.

    Patrick. Good to see the Europeans are finally waking up to the role of the ECB.

    Phil, Maybe I’m wrong. I suppose my last defences would be to pit Shiller agaisnt Buffett. If Shiller is right, then why didn’t the smartest investor on the planet, and someone who doesn’t think the EMH prevents excess returns, follow Shiller’s advice?

    I was reacting to earlier claims about his long term performance vs. the S&P 500. In reacting to those claims I don’t think it was unreasonable of me to use the same metrics as his supporters had constantly used. After all, suppose Buffett had underperformed the S&P500 for 50 years, but had outperformed on a risk adjusted basis—would he still be famous? Obviously not. But I accept your point that his stock picking may still be outperforming, I haven’t studied that question as you have.

    Phil and Tony, I still don’t get the tax issue. That should be endogenous. Let’s say Buffett knew ex ante that he couldn’t manage more than $X amount efficiently, due to the tax penalty of moving funds around, and the difficulty of picking so many winners. Then he should gradually direct the new money getting invested into index funds, so that he never ends up in the tax bind. Or am I still missing the point?

    Peter, Good point. I favor the Canadian approach to housing (although even it is not perfect.)

    Vaidas, I should have been more specific. I meant “worthless as evidence against the EMH.” Fama doesn’t claim his anomalies show the EMH is wrong.

    Scott, Fine, but keep in mind that most anti-EMH stuff I read is complete nonsense, even when in respected publications like The Economist, or from respected economists. As bad as my pro-EMH posts are, they aren’t complete nonsense like many of the arguments on the other side. I really don’t understand why data mining and cognitive illusions are so hard for intelligent people to understnad, but based on what gets published today it apparently is.

    ed, Fair points, read my reply to phil.

  70. Gravatar of Scott Sumner Scott Sumner
    9. April 2014 at 13:38

    Michael, That’s a fair point. But again, I started blogging in early 2009. Since then everything has gone my way–hedge funds, Buffett, Shiller, foreign housing markets, Bitcoin,etc. It didn’t have to, but did. No reason for me to abandon my skepticism of the utility of anti-EMHism

    My only counterargument would be to ask if there were say 10 investment geniuses in 1980, and he was the best of the 10 in subsequent years—causing him to go from being famous to being very famous. I don’t follow investments enough to know if that counterargument is valid.

  71. Gravatar of Vaidas Urba Vaidas Urba
    10. April 2014 at 02:50

    “I should have been more specific. I meant “worthless as evidence against the EMH.” Fama doesn’t claim his anomalies show the EMH is wrong.”

    Yes, and Fama claims that Shiller CAPE anomaly represents efficient changes in risk premia. You need something else to decide if these anomalies are efficient or not. Shiller uses investor surveys for that. My own view is that both Shiller and Fama are 50% right, and the 2007 Greek bond bubble did most to persuade me.

  72. Gravatar of J Mann J Mann
    10. April 2014 at 04:43

    Thinking about Kevin’s first comment a little more, what does the EMH predict about belief in the EMH?

    My first cut is that the EMH predicts that markets that include a product where you could profit by trading on your opinion that the EMH is correct are priced in a way that reflects that belief.

    That’s super awkwardly stated, but for example, if there were a product that paid if mutual fund managers overall failed to beat the market average, the EMH would predict it would be priced as if the market believed in EMH.

    On the other hand, if Berkshire Hathaway or a particular mutual fund is priced in a way that can only be justified if Buffet and Munger or the fund managers are consistently able to beat the market, I’m not sure what the EMH has to say. On the one hand, EMH would suggest that the product is priced correctly. On the other, it would also suggest that the investment managers can’t consistently beat the market. Unless you grant that maybe the investment managers have a non-market advantage (inside information, government patronage, or superior ability to utilize assets once under their control), it seems like a puzzle.

  73. Gravatar of Morgan Warstler Morgan Warstler
    10. April 2014 at 05:52

    Mark Sadowski,

    “Basically it asks the question, should monetary policy be saddled with additional policy goals that takes its attention even further away from nominal stability?’

    That’s why I think it’s so important to lay out the conservative case for NGDPLT.

    Because it forces the Fiscal side to own economic performance, it forces the elected officials to all dance the exact same beat forever.

    It finally clarifies once and for all what approach to taxes and regulation works.

  74. Gravatar of ssumner ssumner
    11. April 2014 at 06:15

    J Mann, Do people know what stocks Buffett holds? Can they replicate his portfolio without buying Berkshire directly?

  75. Gravatar of TravisV TravisV
    11. April 2014 at 06:32

    Prof. Sumner,

    Yes, the SEC requires Berkshire to thoroughly disclose how much stock it owns of certain publicly-traded companies (partial ownership, Wells Fargo, for example) as well as other entities (100% ownership, BNSF, for example).

    You should look at their 10-K sometime!



  76. Gravatar of TravisV TravisV
    11. April 2014 at 06:35

    It’s possible for the average Joe to replicate the performance of Berkshire’s publicly-traded holdings (basket of Wells Fargo, Coca-Cola and IBM, say).

    However, he can’t really buy into the performance of entities that Berkshire totally controls (such as BNSF).

  77. Gravatar of ssumner ssumner
    12. April 2014 at 11:10

    Thanks Travis, that suggests the valuation of his company does not disprove the EMH.

  78. Gravatar of TravisV TravisV
    25. November 2015 at 08:36

    Oh yeah? Oh yeah?????

    “What is a Better Strategy – Invest in Berkshire, or Buffett’s Stock Picks?”


Leave a Reply