Markets don’t overreact to disasters

I recently came across the following in a local paper called The Improper Bostonian:

When the British Petroleum rig exploded in the Gulf of Mexico, the stock plummeted from about $60 to $27.  it was an emotionally charged issue, and rumors of bankruptcy for BP echoed widely.  But once the tragedy was off the front pages, the stock rallied to $46.  The sell-off had been exaggerated by emotions and media frenzy.  After the tragic earthquake in Japan, big market sell-offs occurred again.  Two weeks later Japanese stocks bounced back, and a few commentators even talked about how the resulting potential for change created bullish markets.

This is false.  Markets don’t react emotionally to disasters.  The Japanese market did not overreact, it processed information very rationally.  In contrast to John Spooner’s assertion, the initial reaction in Japan was actually an under-reaction, and prices fell further as more information came in about the severity of the nuclear problem.  Then prices rose somewhat when the worst case was avoided.  Here are some Nikkei closing levels:

March 10  10,410

March 11  10,254   (Tsunami occurred at 2:46 pm)

March 14,  9,620

March 15   8,605

March 16   9,093

March 17  8,963

March 18  9,207

March 21  9,608

After that, prices tended to level off, roughly 10% below pre-tsunami levels.

Here’s the mistake Spooner made.  Take any negative news shock.  Then follow the stock market from the date of the shock to the present.  In 99% of cases the current price will be higher than the lowest post-shock price.  That’s simply the nature of any random walk series.  Only in the rare case where the current price IS the lowest post-disaster price, will current prices not be higher than the post-disaster lows.

At first glance it looks like the “overreaction” created a great buying opportunity.  But that’s an cognitive illusion, as no one at the time knew which price was the post crash low. If you bought Japanese stocks on the Monday after the tsunami, you paid too much–the tsunami had not yet been fully priced into stocks, as the extent of damage was still not fully realized.  Almost by definition, the low point (whether in Japan or the BP case) will be an overreaction to the disaster, as it will incorporate the most pessimistic damage estimates of the entire post-disaster period.  But in real time we don’t know when that overreaction occurred, each price point is fully rational, given what investors know at the time.

Just one more example of how cognitive illusions cause many people to mistakenly reject the EMH.

BTW, I am not suggesting that all opponents of the EMH are irrational, just that some anti-EMH arguments are based on cognitive illusions.  There are obviously some more respectable anti-EMH arguments, although in the end I don’t believe those arguments have practical applications.



10 Responses to “Markets don’t overreact to disasters”

  1. Gravatar of Joe Joe
    14. July 2011 at 10:23

    Has any economist ever done an epic centuries long analysis of the stock market’s reaction to news events. He could go year by year and show how the market exactly responded to to each major news event.

    It would be very interesting. And VERY difficult.

  2. Gravatar of marcus nunes marcus nunes
    14. July 2011 at 10:28

    In dealing with humans, I don´t think you can say there´s no “emtional reaction”. There surely is. The point is that the “emotional” part of the reaction is quickly “priced out”.
    Natural disasters are physical phenomena the effects of which can “quickly” be evaluated and priced.
    Dealing with the implication of “words” is much harder (not only because they can “move around” much more quickly (Bernanke-Fisher, for example).
    Just see the drawn out impact Kennedy´s “Steel industry speech” had on the stock market, and compare that with the 1 day impact of his assassination (in Brazil´s institutional set-up, for example, the effects would be exactly reversed!)

  3. Gravatar of marcus nunes marcus nunes
    14. July 2011 at 10:32

    It would be painstaking, but quite possible. I haven´t done a “century long” evaluation but have examined several instances both in the US and Brazil, mostly relating to “words spoken by authorities”.

  4. Gravatar of John John
    14. July 2011 at 11:36

    You admit that EMH isn’t testable or disprovable, right? I don’t think there’s anything wrong with an untestable hypothesis, any ceteris paribus assumptions fall into that category, but if your methodology is based on the scientific method, the EMH view is problematic.

  5. Gravatar of John John
    14. July 2011 at 11:52

    Question: are thinly traded markets less efficient? It seems that the number of people involved in speculating would make a difference in EMH.

  6. Gravatar of Scott Sumner Scott Sumner
    14. July 2011 at 14:27

    Joe, Yes, that would be interesting.

    Marcus, Yes, people react emotionally. But I was trying to show that what looks like overreaction isn’t really overreaction at all–it’s completely rational.

    John. You said;

    “You admit that EMH isn’t testable or disprovable, right?”

    No I don’t. There is only one rational way to test the EMH, and it’s already been done. Test for policy or investment implications. Try to see whether others have found flaws in the EMH. If they have, then excess returns should be serially correlated. And they aren’t, or are to such a small amount that it has no investment implications. This suggests the EMH is true, or false but close enough to be a useful approximation of reality. I’m fine with either one.

    John, I’d guess so, but studies of prediction markets show you don’t need all that many traders to get reasonable efficiency.

  7. Gravatar of Matt Waters Matt Waters
    14. July 2011 at 15:25

    I admit I have issues with the EMH, but a good piece of evidence for the EMH is Buffett’s editorial to buy stocks in October 2008, if I recall correctly. Stocks are indeed higher today, but it was six months or so before we actually hit the bottom.

    But like I said, I have issues with the EMH. Here is Buffett’s letter where he compares Book Value of Berkshire with the value of the S&P with dividends reinvested:

    Particularly important is how Buffett has done every five-year period. Relative performance to S&P once year can be misleading, because a 50% gain is not equivalent to a 50% loss. If a stock gains 50%, it only takes a 33% loss to go back down to its previous level. But if a stock loses 50%, it takes a 100% gain the next year to go back to its previous level. And very rarely does an investor just put money in for a year. Five years, at least, is the typical time frame for most savings.

    So, in comparing 42 different five year periods in the report, Berkshire’s book value beat the S&P’s five year results in all 42. The probability of that happening by pure luck is 1 in 2^42, or about 1 in 600 times the world population. For this to result just through the binomial distribution, everybody in the world would need to manage about 600 different stock funds, manage all those 600 funds for 42 years, and then 1 of those stock funds’ performance would equal Buffett’s.

    And if you take a 42-year periods, Berkshire’s book value has gone up 490,000% while the S&P has gone up 6,000%. Most of this gain may be because the markets were less efficient before the 80’s, but that still doesn’t explain why all five-year periods after 1980 have done better than the S&P. 2^20 is still 1 in 1,000,000. That may be more in line with the EMH coin-flipping philosophy, but there are funds other than Berkshire which have also beat the S&P given five-year periods.

    I will admit that this analysis isn’t perfect because the five-year events are obviously not independent and the binomial distribution assumes independence. My probability skills are not sophisticated enough to adjust for the lack of independence, but hopefully the 2^42 number gives some pause.

    To me, the most straightforward explanation then is that Buffett does have a method to routinely beat the market once you take into account five-year periods rather than one-year relative performance. I believe there are behavioral finance reasons for this, since so many investors tend to buy high and sell low due to fear and greed and not because of underlying fundamentals. This investor psychology does create mispricings, even though they are acting contrary to their own long-term self-interest.

    And EMH would be fine as a tool to just say that it’s very difficult to predict the direction of the market or beat the market. However, I think the EMH causes us too often to take the market’s prices as gospel, when clearly the market can and will create some gross mispricings. The market caused AAA tranches of CDO’s to show a <1% chance of default when now more than 50% are worthless. That just cannot be explained just by new or private information. That is a terribly gross mispricing under any conceivable set of circumstances. Same with tech stocks and junk bonds in the 1980's.

    Part of that mispricing was due to the EMH mentality, where investors did not have to dig into actual future cash flows. After all, if the market had low yields and the ratings agencies gave them AAA, what was the need to really dig into the future cash flows? That's my fear with the EMH, that people will rely too much on the market's price without actually making sure they get more cash than they put in. The Efficient Market Hypothesis also says that markets are always, well, efficient and implicitly argues for no regulation despite government backstop of banks. I know that you're not for bank bailouts, but as things stand with moral hazard, we clearly need strong regulation and the EMH mentality helps stand against that.

  8. Gravatar of FXKLM FXKLM
    14. July 2011 at 17:00

    Even if investors “over-react” to disasters in the sense that asset prices decline by more than the loss in expected value caused by the disaster, that could still be consistent with EMH. In the middle of a disaster where the full effects are still unknown, volatility will spike and asset prices will decline due to the increased risk premium as well.

  9. Gravatar of Lorenzo from Oz Lorenzo from Oz
    14. July 2011 at 17:56

    If people could systematically identify superior investment opportunities, would that not then feed into prices? So, the sensible claim is not that such cannot exist (I am thinking of studies which indicated Congressional share portfolios did better than the share market overall), but that they cannot persist. (Congress acts in ways which affects share prices, even without oversight “tips”, so they have a permanent information advantage; but it keeps being a “new” information advantage.)

  10. Gravatar of Scott Sumner Scott Sumner
    15. July 2011 at 09:03

    Matt, That’s wrong, as there are only 8 or 9 independent periods. The case of Warren Buffett actually supports the EMH. It predicts that for every million investors, one should get once-in-a-million type success. And that’s what we observe. Google my post “Being there themoneyillusion”

    FXKLM, Good point. It also depends how this excess volatility is correlated with other world markets.

    That’s one reason I’d like to eliminate insider trading laws, so we could do what Congress can.

Leave a Reply