Thursday, February 7, 2013

Uncertainty is Not Pessimism

An article in Bloomberg today by Caroline Baum is called "How a Nation Got Snookered by a Phony Narrative." I wrote about the prevalence and power of narrative in economics a few weeks ago in the context of the Federal Reserve. Baum's interest is in a narrative promulgated by politicians, business leaders, the media, and some economists that says that "uncertainty" has been causing businesses to postpone investment and hiring.

I was glad to read Caroline Baum's remark that uncertainty and pessimism are not the same.
No one speaks of uncertainty during good times. There was lots of uncertainty in March 2000, when the Nasdaq Composite Index breached 5,000 as investors bought shares of Internet companies with no revenue, no profits and, in at least one case, no known business. No uncertainty back then; just a case of irrational exuberance. 
In good times, the word uncertainty rarely appears in policy discussions. In bad times, it’s the default setting. Why not call it by what it really is, which is pessimism? When businesses say they aren’t going to invest because of uncertainty, what they mean is, they don’t think their investments will produce a substantial profit.
Baum's commentary is very relevant to part of the research I am working on now. When I presented a preliminary version of my research at a lunch seminar at UC Berkeley last semester, I began with this toy "model" which is actually simple enough to just describe in words. Here is the set-up:

Suppose Alice and Bob are business owners and they are facing the same opportunity to invest in a project. Next year, the economy may be good or bad. If good, the project will be highly profitable, and if bad, not so profitable. Alice and Bob have different beliefs about what will happen next year. 

Now, suppose Alice is 99% certain the economy will be bad next year. Bob thinks there is a 50% chance the economy will be good and 50% chance it will be bad. Who is more uncertain? Bob, of course. He has no inkling about which outcome is more likely, a situation known in probability theory as "uninformed priors." Alice is much more certain about what she believes will happen. Who is more likely to invest in the project? Bob. His expected payoff is higher because he puts more weight on the good outcome. So in this case the more uncertain businessperson invests and the more certain businessperson does not. Bob is uncertain but relatively optimistic, Alice is relatively certain but pessimistic.

I used this example to show that uncertainty is not pessimism. A popular measure of uncertainty, used in a number of interesting new economics papers, is the Economic Policy Uncertainty Index, which consists of three components:
One component quantifies newspaper coverage of policy-related economic uncertainty. A second component reflects the number of federal tax code provisions set to expire in future years. The third component uses disagreement among economic forecasters as a proxy for uncertainty.
If Baum is correct, the first component of the index may confound uncertainty with pessimism. If everyone were like Bob, the economy would be at maximal uncertainty. If everyone were like Alice, there would be very little uncertainty, but also very little investment, and the newspapers would likely refer to the economy as being "uncertain." Baum uses March 2000 as an example, saying that there was high uncertainty when the Nasdaq breached 5000 but nobody talked about it. Indeed, the newspaper coverage component of the Economic Policy Uncertainty Index actually fell in March 2000, as did the overall index.

The Alice and Bob example can demonstrate that uncertainty is also distinct from disagreement, the third component of the Economic Policy Uncertainty Index. Suppose Alice is 99% certain the economy will be bad next year and Bob is 99% certain the economy will be good. Then they are both very certain, but disagreement is very high. Conversely, suppose Bob and Alice both believe there is a 50% chance the economy will be good and 50% chance it will be bad. Then they are both very uncertain, but disagreement is low.

The Alice and Bob example takes no stance on the "true" probability that the economy will be good or bad or on why Alice and Bob believe what they believe. In my research, I am trying to address both of these issues. I am also trying to construct a theoretically-sound measure of uncertainty that is not confounded with pessimism or disagreement. I hope that this research goes well and that I can post some results in the future. I have another presentation at Berkeley on Friday, March 15 in Evans Hall. Contact me if you are in the area and would like to attend.


  1. Even if uncertainty and pessimism aren't the same thing, it seems like they could be inextricably linked.

    Like what if Bob and Alice's choices are interdependent? Maybe they're in a situation where they can trade, but in order to do so each has to pay a cost to make the good he/she is going to sell, and they commit to this cost before production takes place. If both make their respective good, they will trade, but if one of them bails on the plan, they won't. (I'm imagining a 2-period model, where in T=0 Bob and Alice commit to making vs. not making, and in T=1 their choices are played out [they pay the cost, make the good, and complete the potential trade], which I think boils down to a 2x2 normal form with two equilibria, both make and neither makes.)

    Now suppose Alice is risk-averse and gets an exogenous, positive shock to her uncertainty. Maybe her job is magically zapped by Zeus, and she doesn't know what her income will be next month. As a result she doesn't know if she'll be able to afford the production cost. This makes her less likely to stick to the good equilibrium, which makes Bob more pessimistic, which then makes Alice more pessimistic, and they end up in the bad equilibrium.

    Does that make any sense? Do you think it's a reasonable way to characterize the intuition behind what we hear in the public discourse about uncertainty being this terrible thing?

  2. Daniel, yes, that makes a lot of sense. That is part of what I meant by saying "The Alice and Bob example takes no stance on the 'true' probability that the economy will be good or bad or on why Alice and Bob believe what they believe." There is bidirectional feedback between uncertainty and economic outcomes, and you described one likely channel. Let's meet up in Berkeley the week of the 18th!

  3. Cool. Definitely. This is great stuff, looking forward to your lunch next month to see how it's evolved since the fall.

  4. This comment has been removed by the author.

  5. have you looked at Angeletos and La'O?

  6. Good post. I also agree that we need to dehomogenize uncertainty and pessimism.

    "Now, suppose Alice is 99% certain the economy will be bad next year. Bob thinks there is a 50% chance the economy will be good and 50% chance it will be bad. Who is more uncertain?"

    Can we really say that Bob is more uncertain? He knows the probabilities and so can perfectly hedge in the market to protect himself given either future state of the world. Same with Alice.

    Someone who is uncertain can't put any probability distribution whatsoever on the various future states of the world, and therefore can't hedge. They are forced to bear all the unpleasantness of uncertainty.

    Mind you, there are plenty of definitions of uncertainty. Mine might be different than yours.


Comments appreciated!