Tuesday, March 1, 2011

More on Keynesian Uncertainty

- "Lord Keynes", at Social Democracy for the 21st Century, blogs about the Economist article I mentioned the other day here. He connects Keynesian uncertainty to Post-Keynesians specifically, which I found a little odd - I asked him why in the comment section. This may be a case of heterodoxy in search of distinction (similar to the wonderful Bryan Caplan characterization of the Austrian reaction to being informed that all economists are subjectivists - "Oh - well then we're radical subjectivists").

- Jonathan blogs about the same article here. This is less about the uncertainty arguments and more about questions of market efficiency and bubbles (which was the subject of the original Economist article, after all).

- Donald Rumsfeld is on a book tour that I think is relevant to this question. His book is called "Known and Unknown", based on his 2002 formulation of "known knowns, known unknowns, and unknown unknowns". This captures the essence of Keynesian uncertainty in a nice, short-hand way. Keynesian uncertainty boils down to a couple important propositions - first, that probability is a branch of logic. And second, that we can think about probability in a categorical way (which roughly corresponds to Rumsfeld's categories) depending on our ability to quantify probabilities for certain events.


  1. Why call it Keynesian uncertainty exactly? Sounds largely like a rehash of Montaigne, Sterne, and the other moderate skeptics of the Renaissance-Enlightenment.

  2. Because I'm discussing Keynes's point.

    Why call it "Keynesianism" at all? Why not call it "neo-Malthusianism". What do you want me to say, Gary?

    Keynes is particularly concerned with the relationship between logic and probability. I haven't read Montaigne or Sterne, but I somehow doubt they plumb these depths specifically. I checked, and Keynes does not cite Montaigne or Sterne. He does cite Peirce, and that is one influence that people have identified as anticipating Keynes on this - probably more relevant than the ones you mention.

    A better question would have been "why not call it Knightian". Knight and Keynes published their books on uncertainty in the same year. I think Keynes has a much deeper grounding in the literature on probability (which is no slight on Knight - that was not what Knight was aiming at), and in subsequent work Keynes developed his understanding of uncertainty much more broadly than Knight. Knight in a lot of ways describes a special case of Keynes's views on the way we react to uncertainty.

  3. I know what you mean, Daniel.

    When I first read LK's pieces on post-Keynesian uncertainty, I wondered, "Do post-Keynesians think they invented this concept?"

  4. "I checked, and Keynes does not cite Montaigne or Sterne."

    Then he wasn't terribly bright it seems. He probably never read Rabelais either, the philistine.

  5. You are perfectly correct that the concept of uncertainty in the Keynesian/Knightian sense is held by other economists.

    I have added an addendum:


    "I asked him why in the comment section."

    Your comment did not show up in my blog comment system. Apologies. Maybe you can try again.

  6. Arnold Kling has a much more interesting write-up: http://econlog.econlib.org/archives/2009/03/keynes_probabil.html

  7. Ya - I noticed that it didn't come up. My fault I'm sure - I did it before my first cup of coffee. No matter. Good post!

  8. Gary - Kling's post is good, but I think it's odd how he contrast frequentism with subjectivism.

    The point isn't that frequentism is wrong and subjectivism is right. The point is more that some things are more amenable to frequentist probability and some things are less amenable. Those three groups that he lays out in the beginning were the three that Keynes laid out (if it wasn't clear), but to my knowledge Keynes never rejected the first two or pitted the second against the third. He just noted that the existence of the third poses serious problems.


All anonymous comments will be deleted. Consistent pseudonyms are fine.