On uncertainty

Richard and Tim’s latest spat concerns the role of uncertainty, as distinct from risk. Richard says:

Keynes pointed [that] the number of circumstances where we can make the predictions neoliberal economists think possible are remarkably limited. He said the future is not probabilistic as they suggest in most cases: it is actually uncertain.

To this, Tim replies - correctly - that Hayek was fully aware of the importance of uncertainty.

But Richard has a point, which is hidden by his silly conflation of neo-liberal and neo-classical economics: Hayek was a neoliberal economist but not a mainstream neoclassical one.

If we define neo-classical economics as stuff we learn at university (I’m using the Royal We) then it does, I suspect, emphasize probability over Knightian uncertainty. When I did my masters (25 years ago!) we learnt about expected utility theory, regret and even prospect theory. But uncertainty barely featured. Yes, the Ellsberg paradox got a mention. But like Allais’ paradox, it was something we looked squarely in the face and moved on from.

Cynics might say this is because orthodox economists were wedded to a scientistic faith that economics should be mathematical. Expected utility and its variants lend themselves to a maths that is suitable for university exams, but uncertainty doesn‘t. Yes, Larry Epstein, among others, has done important mathematical work on ambiguity, but this is hardly good exam material.

Herein, though, lies a nasty fact. Academic economics’ elevation of probability over uncertainty reinforced our human tendency to prefer risk to uncertainty. And - wishful thinking being so powerful - this preference gave rise to a belief that future outcomes really could be quantified and managed. But the banking crisis showed that this belief was an expensive mistake. Yes, Hayek could have told us this. But banks weren’t listening to Hayek.

Another thing: Although Hayek saw uncertainty as a reason to reject central planning, the man who first emphasized the distinction between risk and uncertainty - Frank Knight - saw uncertainty as a reason for the existence of hierarchical, centrally-planned organizations, namely firms:

When uncertainty is present and the task of deciding what to do and how to do it takes the ascendancy over that of execution, the internal organization of the productive groups is no longer a matter of indifference or a mechanical detail. Centralization of this deciding and controlling function is imperative.

And another thing. I really can't be arsed to define neoclassical and neoliberal properly. My blog, my (rough) definitions.

Comments

[UnlearningEcon]

Hayek was not aware of the importance of uncertainty - he used it in a completely different way to Keynes:

http://socialdemocracy21stcentury.blogspot.com/2011/10/michael-emmett-brady-on-hayeks-concept.html

[ShodanAlexM]

The other peculiarity with this sort of debate is why Hayek is being invoked at all.

Hayek's economics has absolutely nothing to do with modern macroeconomics. His understanding of the economy bears no relation to any of those that currently prevail or have any leverage over policy.

He is some respects not a million miles from Keynes on uncertainty, but that is why Keynes' original thesis had to be emptied of all content by Hicks and the neoclassical synthesis. Uncertainty was removed from the story in order to tame Keynes into the IS/LM framework. In the process they completely lost the point. And they lost the aspects of Keynes that brought him closer to Hayek.

Uncertainty deeply threatening to contemporary economics at an ontological, epistemological and methodological level, which is why it persists with risk. If you embrace the importance of uncertainty and the idea that the future is currently underdetermined - there are genuine choices today that will shape which future comes to pass - then the whole edifice of formalism falls to pieces.

[Jorjun]

Software simulation is a good way to model the mechanical possibilities I would say. This sort of technology was unavailable cheaply enough, until recently. I would be surprised if there are not economic simulations running in a lab somewhere that are reasonably representative of the current economic situation. Then I would like to see what happens to a representative model when the minimum wage is abolished. I suspect more economic activity will take place, and wages will go back to being a sliding scale, instead of the massive employer leg-up that it has become (by creating a surplus pool and reducing bid prices). Hell we can imagine what would happen if there was a minimum price for pork : an animal welfare catastrophe.

[Will Davies]

One problem with uncertainty, as opposed to risk, is that it produces a very uncomfortable form of politics. It doesn't mesh with our public, bureaucratic and media desire for accountability. To acknowledge uncertainty, a politician, regulator or expert has to stand up and say "I know that this policy may produce a bad outcome, but I cannot tell you how likely that is". That person would be treated as incompetent. Uncertainty leaves political judgement looking very naked: how one responds to it depends on values and instinct, which are harder to defend publicly than statistics (partly because even laypersons understand them, whereas few of us understand statistics).

This all contributes to the power of a category that neither Knight nor Hayek discussed, namely ignorance. Bankers, Murdochs and politicians now prefer to say "I had firm knowledge of these risks, but knew nothing whatsoever of x", as this seems a safer strategy than saying "I recognised that x was possible, but did not know how probable".

[Luis Enrique]

I don't know anything about this topic .. In most settings, in the presence of uncertainty, is it reasonable to say people make a guess that amounts to supposing a probability distribution? If so, how would economics modelling need to change from just supposing people face probability distributions? I suppose guesses might be more unstable in light of new info,

[Ray Sawhill]

Frank Knight is 'way underappreciated. I almost never see his name come up, whether in the econ press or on econblogs. Any idea why?

[Uggs on Sale]

When uncertainty is present and the task of deciding what to do and how to do it takes the ascendancy over that of execution, the internal organization of the productive groups is no longer a matter of indifference or a mechanical detail. Centralization of this deciding and controlling function is imperative.