Tuesday, December 14, 2004

Frank Knight and Ellsberg's Paradox

This is not about other posts on the Internet, but rather the reeling of my tired brain while I was stuck in traffic today....and boy, I hope this post - more than any other - elicits comments...
(But feel free to scroll down to the next post if this starts to put you to sleep)....

Last summer I finally read Daniel Ellsberg's classic, Risk, Ambiguity, and Decision (just republished in hardcover), which includes the famous Ellsberg Paradox and the "balls-in-the-urns" experiments. (Simplified explanation for those unfamiliar: these were experiments documenting that people are more averse to uncertainty than they are to regular risks of known proportions, usually to an irrational degree). I was confused, honestly, that I could not find a reference to Frank Knight's earlier classic, Risk, Uncertainty, and Profit anywhere in Ellsberg's book. This seemed strange because Knight is generally credited with being the first to draw a conceptual distinction between risk and uncertainty (a distinction that is still controversial). Another book I read at the same time, Judgment Under Uncertainty (Kahneman, Slovic & Tversky) seemed to make the same omission, unless the reference happened while I was sleep-reading (very possible).

So, sitting in traffic today, I was thinking about Knightian uncertainty and the "Unconscionability Doctrine" in Contracts (a project I've been working on since last summer that has so far gone in circles). The rule is disfavored by creditors, of course, because it provides an opportunity for losses from defaulters - and most of the arguments against the rule seem to run along the lines that it disproportionately burdens the very businessmen and financiers who offer the most transactional opportunities to the poor (besides criticisms of the rule's overall inelegance). There is thus a chilling effect on opportunities for the poor. Valid concerns, of course. But there is something else going on with this rule: it is not just about tipping the scales slightly to one side or the other (like most legal rules), but is actually sort of an "anti-rule," and undefined, squishy, equitable loophole that allows a court to "break" the usual bright-line rules in Contracts. It constitutes the introduction of uncertainty into the transactions, for both parties, besides the supposed costs of insuring against default, etc. (which are simple, actuarial risks under Knight's paradigm). According to Knight, this introduction of uncertainty should create the possibility of "true profits" (for both sides?) where the possibility would not otherwise exist - perhaps encouraging transactions. Yet Ellsberg (I think) would indicate that the introduction of uncertainty brings in a chilling effect on transactions, besides any chilling effect from the rules being pro-consumer. This is where I end up every time I start working on this project - with this "unconscionability paradox" or contradiction, a circle I never break out of. Sigh.

I did, however, think of something else that helped me connect Knight's ideas with the Ellsberg Paradox (somebody be kind enough to email me if everyone else already knew this): Knight, after spending his whole book building up his definition of entrepreneurs, risks, uncertainty, and profit (which I think most of us would call "windfall" as a more descriptive term), says in the last chapter that he suspects uncertainty works in the negative in the aggregate. In other words, the aggregate amount of windfall profits realized by successful entrepreneurs in society is outweighed by the aggregate losses of the unsuccessful entrepreneurs. He concludes that entrepreneurialism must constitute an overall net loss for society as a whole, then admits that this is really confusing because it goes against the foundational assumptions of capitalism - and then quickly says farewell and ends his book (and then the Great Depression happened).

If Knight is right (I like how that rhymes), this would certainly add a touch of legitimacy and intuitive rationality to Ellsberg's Paradox, wouldn't it? Aversion to uncertainty - in fact, the greater aversion to uncertainty than to quantifiable (read insurable) risk - makes a lot of sense if it is a losing proposition for the species overall. Well, maybe this will strike others as a good case for listening to talk radio in traffic instead of letting my mind wander. But for me, Ellsberg's Paradox just moved from the category of "bounded rationality" into classic Posnerian rationality, and the wedding between Knight's ideas and Ellsberg's was complete. Next step is to talk myself out of the unconscionability thing...