Knightian Uncertainty Workshop
24 September 2007 at 11:17 am Peter G. Klein 3 comments
| Peter Klein |
The authors of this blog find Knightian uncertainty a useful concept for understanding both entrepreneurship and the economic theory of the firm (e.g., here and here). So we were pleased to learn about a workshop on Knightian uncertainty organized earlier this month at Columbia University by and Daniel Beunza David Stark. Nassim Taleb was there, as were Douglass North, Anna Grandori, Bruce Kogut, Adam Branderburger, and others well known to readers of this blog. Buenza summarizes the discussion and offers some commentary at the Socializing Finance blog.
One problem with the treatment of Knightian uncertainty in the management literature (not necessarily in the presentations above) is that the concept itself is not always defined precisely and consistently. Terms like Knightian uncertainty, radical uncertainty, case probability, complexity, ambiguity, and plain old uncertainty are often used interchangeably. Sometimes what is meant is ignorance of the relevant probability distributions. Sometimes it is ignorance of one’s own ignorance. Sometimes it means simply the lack of common (Bayesian) priors. To move forward, more clarity is surely needed.
Entry filed under: - Klein -, Entrepreneurship, Management Theory, Theory of the Firm.
1.
srp | 24 September 2007 at 8:54 pm
Having just completed a revision on a paper where disagreement between agents is central (although I don’t get into the Knightian thing), I agree that more clarity about different sources of uncertainty would be a good thing. It seems to me that while disagreement often is the result of some sort of ambiguity about states or incompleteness of supports of distributions, it’s also consistent with Bayesianism (as Ryall and Van Den Steen have modeled explicitly).
I think one modeling difficulty is often self-imposed: the desire to endogenize initial conditions of belief. The results of disagreement are often going to be the same regardless of the source of the disagreement. A self-consistent story that rationalizes the disagreement and gives a beginning-of-time-to-the-end-of-model-date (is there a German word for that?) explanation is more than we need most of the time. If a physicist is studying what happens when a particle of type A collides with a particle of type B, it’s really not necessary to have a theory of why they’re colliding in the first place, where they came from, etc. Disagreement is kind of like that.
2.
Peter Klein | 25 September 2007 at 2:23 pm
Steve, I couldn’t agree more. I have argued recently that the concept of “opportunities” in entrepreneurship research is like the concept of disagreement as you characterize it above. Perhaps we don’t need to spend so much time trying to model or conceptualize the psychological foundations of opportunity formation or recognition. Instead, we can simply posit that entrepreneurs act on the basis of perceived opportunities and go from there. It isn’t necessary to specify what opportunities _are_, in other words; we can focus on what opportunities _do_. (Jay Barney doesn’t agree, as those of you attending this year’s AoM session on opportunity discovery can attest.)
Similarly, we could perhaps diffuse some of the controversy over bounded rationality by focusing instead on contractual incompleteness. Perhaps all we need for a theory of the firm is the idea that contracts are incomplete. Whether they are incomplete because of cognitive limitations, or because some variables are unobservable (or unverifiable), may be beside the point.
3.
srp | 25 September 2007 at 7:57 pm
Up to a point, I agree with you. But if you want to give normative advice about how to search for opportunities, or understand what factors lead people to find more opportunities in some circumstances than in others. it’s going to be harder to avoid the underlying constraints on recognition. With the kind of disagreements I’ve been worried about, in contrast, the causes are orthogonal to the consequences and (less so) to the cures. I’m not sure that’s true for understanding differential opportunity recognition.
But you make me wonder–could we try a radical redefinition and say that entrepreneurial opportunities are potential investment/actons about which actors disagree? Then the equilibrium world would be one with no disagreement, kind of a hypothetical end point that is only observed in restricted circumstances. Two problems with this: First, I think differential awareness is not the quite the same thing as disagreement, and I think that awareness is important in explaining entrepreneurial opportunity. Second, as I noted in my first comment, there are models of disagreement using differing Bayesian priors that model it in an equilbrium way–everybody is aware of the disagreements, agrees on the nature of the disagreement, fails to be surprised by any event, etc.
I like the idea of entrepreneurial = disagreement + differential awareness though, at least better than I like uncertainty-bearing. Focusing on heterogeneity makes the foundations more social and less psychological, along the lines you advocate. Requires some reflection.