19 thoughts on “Granger causality

  1. Brooks

    The mid-Jan to mid-Feb data are interesting in that hangover searches spiked both same day as mixed drink searches spiked, and the day after. My half-serious theory/hypothesis would be that during this period people planning to make/order a lot of mixed drinks (or choose between mixed drinks and other drinks) that evening still had memories of that particularly bad New Year’s Day hangover fresh in their minds, so they also researched hangovers prior to their evening drinking to have a better sense of risk/reward, but of course once they were out and had consumed a couple of drinks, rationality went out the window and they ended up with a hangover the next day, searching for something to mitigate the pain from both the hangover and from kicking themselves.

  2. Ironman

    Makes sense, especially with the blue spikes mostly either leading or coinciding with the red ones. I also like how the blue spikes most often occur on Saturdays!

  3. Steven Kopits

    Cleary, economists have a lot of spare time.
    Interesting tool, though, tricky to use correctly.

  4. Charles N. Steele

    Hypotheses for Brooks:
    (1) Info on how to deal with the IRS and guns are substitutes.
    (2) After trying to read incomprehensible tax code, people become angry and increasingly interested in guns (hence the lag)
    (3) After reading up on IRS, people realize they have tax rebates coming.
    (4) (most likley) The gun searches spike on weekends; this is recreational searching. The IRS searches spike on Mondays and slowly decline — this is work-related.

  5. Charles N. Steele

    To Brooks: “irs” & “cooking” is almost identical to “irs-guns.”
    To Anonymous: “mixed drinks-morning after” … beautiful. You’re turning this into an art form.

  6. Brooks

    Charles,
    My failure to think of weekdays vs. weekends reminds me of something George Carlin said. (Paraphrasing) A year is a natural unit of time. A month is a natural unit of time. A week is NOT a natural unit of time…which is why the pigeons in the Financial District can’t figure out why no one is there to toss them bread on weekends.

  7. GNP

    Interesting. Try “fuel prices” and you will notice two distinct peaks in 2005 and 2008. If you compare “fuel prices” to “fuel economy”, you will note a lone spike in “fuel economy” in 2009.

    “Terrorism” exhibits an interesting downward trend.

  8. Spooky

    It actually looks like if you plot ANYTHING with ANYTHING else, you more or less get this pattern, i.e. there is some sort of seasonality within the day. We’d probably need to strip out the cycle.

Comments are closed.