Don’t forget about those black swans

Matt Yglesias is back from his honeymoon, and better than ever.  But I’m not completely sold on this argument:

Here’s a fun Intrade price anomaly that showed up this morning. The markets indicate that there’s more than a 3 percent chance that neither Barack Obama nor Mitt Romney will win the presidential election. That’s clearly way too high.

Maybe, but is it possible that we underestimate the chance of something unusual happening?

1.  One of the two major candidates is assassinated, and the replacement is elected (as in Mexico’s 1994 election.)

2.  Ditto, except one pulls out due to health problems, or scandal.

3.  A third party candidate comes out of nowhere to get elected.

To be sure, I view all of these as being very unlikely.  But 3% is low odds.  It’s basically saying once in ever 130 years you’d expect something really weird to happen in US presidential politics during an election year.   That’s a long time!  Given all the weird things that have happened, how unlikely is it?  Some might counter that none of the three scenarios I’ve outlined have occurred in the US during an election year (my history is weak so I’m not certain.)  But mind-bogglingly unusual things have happened on occasion.  On November 10, 1972, what kind of odds would Intrade have given on neither Nixon nor Agnew being President on January 1 1975?

And no, I don’t plan on putting my money where my mouth is—I’ve already placed my wager on the same candidate I bet on in 2008.



21 Responses to “Don’t forget about those black swans”

  1. Gravatar of Morgan Warstler Morgan Warstler
    3. May 2012 at 17:06

    Just make sure you spell my name correctly in your dedication…

  2. Gravatar of Benjamin Cole Benjamin Cole
    3. May 2012 at 17:19

    My sentiments exactly.

    We live in an era in which if Romney is caught, or even sort-of caught, with his pants down with a cute girl (or guy), he will have to withdraw. Remember Edwards?

    I am still puzzled by the way Santorum, young and good-lookingish, suddenly folded, with home state Pennsylvania on the horizon. Did that make sense? No girl problems?

  3. Gravatar of Becky Hargrove Becky Hargrove
    3. May 2012 at 17:50

    Crossing my fingers for Obama.

  4. Gravatar of ssumner ssumner
    3. May 2012 at 18:19

    Morgan, ????

    Ben, Romney having an affair? Now that really would be a million to one against.

    Becky, Did you make the same “investment” I did?

  5. Gravatar of Steve Steve
    3. May 2012 at 18:25

    Intrade also has Ron Paul at ~2.5% chance of becoming president and Hillary Clinton at ~0.5% chance. So the mystery is no more.

  6. Gravatar of wrigglypiggly wrigglypiggly
    3. May 2012 at 18:40

    i’d be much more willing to buy matt’s argument. i remember on the night of the iowa caucuses seeing the combined chances of romney, santorum, and paul winning jump to a cumulative 115%

  7. Gravatar of ssumner ssumner
    3. May 2012 at 18:45

    Steve and wiggly-piggly, I certainly understand the market is not perfectly liquid, and hence is a bit inefficient. I was interested in another issue; whether black swan events are as rare as we assume—I think they are surprisingly common.

  8. Gravatar of Steve Steve
    3. May 2012 at 18:48

    Intrade has dumbed down the interface. The used to show highlight the bid-ask, but now they should last trade converted to a percentage.

    My first reaction to this piece was to go make a quick risk free 3%. So that proves Yglesias wrong. And it left me disappointed that no such opportunity exists.

    Actually there is a opportunity to short every candidate and make about 0.2%. Not worth it. That’s because a number of candidates have a minimum bid, including Johnson, Gingrich, and Santorum. Interestingly, Biden has no bid. Even if Obama dies in office, Democrats are presumed to rush in Hillary Clinton to replace Biden for the election.

  9. Gravatar of Steve Steve
    3. May 2012 at 18:54


    I didn’t see your reply when I posted this.

    Ron Paul is the black swan. I’d short sell him and take 2.5% but that would be like shorting a credit default swap. If the political armageddon trade becomes popular, I’d lose 40x.

    Hillary reflects about a 1% chance of Obama dying or resigning in the next 6 months (assuming a conditional 50% of her winning if nominated). That seems about right.

  10. Gravatar of Becky Hargrove Becky Hargrove
    3. May 2012 at 18:55

    I do think we are in a black swan timeframe but I keep hoping things will stay as calm as possible politically. No bets here, same vote as before, (my Ron Paul vote was before that) and no need to give Obama a hard time as I did that already, in three letters before he was elected.

  11. Gravatar of John Thacker John Thacker
    3. May 2012 at 20:41

    A recent example of #2 occurred in the US Senate in 2002, when a scandal came out about Robert Toricelli and he withdrew from the race after the ballot deadline (once the polls showed that he was not going to win). However, the New Jersey Supreme Court decided that it was in the interest of the state to have a competitive race, and allowed the New Jersey Democratic Party to substitute Frank Lautenberg’s name on the ballot, and Lautenberg won.

  12. Gravatar of John Thacker John Thacker
    3. May 2012 at 20:41

    Oops, Robert Torricelli.

  13. Gravatar of Peter N Peter N
    3. May 2012 at 21:20

    There’s good reason to believe extreme events aren’t that rare. Many sorts of events aren’t normally distributed.

    This stuff has been pretty thoroughly analyzed, and I have nothing to add to the extensive literature, which is much clearer than I’d be likely to be.

    Try Wikipedia (their coverage of math topics is extensive) under heavy-tailed distributions, and links for there.

    Also this NASA article

  14. Gravatar of FillerCrowley FillerCrowley
    3. May 2012 at 23:19

    “And no, I don’t plan on putting my money where my mouth is””I’ve already placed my wager on the same candidate I bet on in 2008.”

    I really don’t think Nader’s gonna win this time around, Scott.

  15. Gravatar of backyardfoundry backyardfoundry
    4. May 2012 at 01:43

    Off topic, but regarding an old post of Sumner’s on Australia:

    Any Australia updates in light of some of the shifts highlighted in the following link? And I’m not making any claim’s about Shedlock’s chops; I merely want an easy way to point to resources.

  16. Gravatar of dwb dwb
    4. May 2012 at 03:21

    six sigma events happen all the time in the financial market.

    thats part of the reason the Fed has not yet done its job: it has not bought any insurance. Shocks tend to be assymetrically growth reducing. who knows when the next 9-earthquake, Tsunami or whatever will hit.

  17. Gravatar of Browsing Catharsis – 05.04.12 « Increasing Marginal Utility Browsing Catharsis – 05.04.12 « Increasing Marginal Utility
    4. May 2012 at 04:01

    […] Scott Sumner says to beware of black swans in elections. I made this argument years ago and thought of formalizing it, but a colleague who knows far more than me about econometrics told me that fat tails don’t matter for binary choice models. I don’t agree with that, but I’m also not someone well-equipped to argue to the contrary. […]

  18. Gravatar of Greg Chaudoin Greg Chaudoin
    4. May 2012 at 05:33

    Don’t forget that there is a time-value-of-money issue. Even if there is a 100% chance that either Romney or Obama will win, we have to account for forgone lost interest. In this situation you can place at the Intrade odds on both Romney and Obama and get get a sure GBP 1 payoff. If the odds sum up to 97%, then you get a 3% return on investment over the next over the next six months. Annualized, this is a 6% return. Pretty nice return compared to money market rates.

    But if you factor in black swans (scandals, death) the odds of both winning is lower than 100% and the expected return in reduced.

    Still, buying both shares seems like a good bet. And if you lose–say hello to President Ron Paul, or Joe Biden–then paying off the loss will be the least of your problems.

  19. Gravatar of Jason Jason
    4. May 2012 at 13:49

    There is a natural tendency for humans to wildly overestimate small probabilities (q.v. Kahneman). The odds of at least one airline fatality in the US in a year would probably fetch a couple percent on InTrade. Planes would have to be falling out of the sky left and right to have that work out from an efficient market viewpoint.

    The truth is that Intrade does not differentiate between “Obama is elected” and “Obama is elected given it is Obama vs Romney”, and so some people are purchasing contracts with an implicit “black swan” probability while others are not. As there is no way to short the black swans, their quasi-contracts will be inherently overpriced (e.g. the 3%) due to both the inability to short them and human nature that overestimates low probabilities.

  20. Gravatar of ssumner ssumner
    4. May 2012 at 14:40

    Steve, Ha! Clinton has a much higher chance than Paul–there’s an arbitrage opportunity.

    From your next comment it looks like we agree.

    John, Yes, and there have been other cases–I think a Minnesota Senator died in a plane crash while running for re-election (but I seem to recall his replacement didn’t win.)

    PeterN, Yup.

    Fillercrowley, Is he even running?

    backyardfoundry, I don’t see where any of that has any bearing on my Australian post. They are doing about as I expected. The 2005 property bubble still hasn’t burst, and probably never will.

    dwb, Good point.

    Greg, Good point.

    Jason, I don’t follow your airline example. Crashes aren’t that uncommon.

  21. Gravatar of Major_Freedom Major_Freedom
    7. May 2012 at 20:06


    six sigma events happen all the time in the financial market.

    They wouldn’t be six sigma events if they ALWAYS happened. They’d be one or two sigma events.

    thats part of the reason the Fed has not yet done its job: it has not bought any insurance. Shocks tend to be assymetrically growth reducing. who knows when the next 9-earthquake, Tsunami or whatever will hit.

    People like you are intellectually responsible for this. Inflation of the money supply leads to a reduced demand for money, and a reduced demand for money is a reduction in the very insurance you say is needed.

    In a free market money standard, money holding is rightfully the best insurance there is. In a fiat standard, money no longer serves as a good insurance, and thus insurance itself is no longer effective.

    When money is backed by nothing, insurance rests with the trust worthiness of the state money printer. That’s no substitute for the rigors of market discipline.

Leave a Reply