Knightian Risk and Uncertainty
A company is thinking of building a hotel. If nothing goes wrong, the profit will be 100. The known problem is that after it buys the land, the company will not get zoning permission for a big enough parking lot, and the profit will be -100. After much thought, the company concludes that not getting permission has probability X=.3. But the company knows that something else as yet unidentified might lead to a profit of -100. The company estimates the probability of this to be Y=.1.
But the company is much less sure of Y than of X. It knows that after building this hotel, its estimate of the probability of a zoning problem will still be X. It’s estimate of the probability of other problems, though, will change to be either Y+Z or Y-Z, with equal probability.
Why should the company be more bothered about Y than about X? Why care about uncertainty over uncertainty? The Ellsberg Paradox shows that we do care, though really introspection is enough to see that.
Maybe it is that we do not wish to look foolish. Everybody might be agreed about X. Some people, though, have better advance estimates of Y.
Or, even if nobody is watching, if I build the hotel and find I had neglected some important problem, I will feel stupid.
Schumpeter, and the Madman quote from Tocqeville in my norms paper with Posner are perhaps relevant here. Some people will pick a low value for Y (even zero), while others will pick a high value– we are biased, but we do not know our biases.