With equity markets all over the planet betting heavily that the long awaited US recovery is now just over the horizon, there is only one small detail left to tidy up: is this for real? Of course the strongest argument in support of the recovery-on-the-way view is that it has failed to materialise for so long now that surely it's going to happen. But isn't there a simple probability fallacy about tossing coins that is applicable here? No, no.. you don't understand, you see there is also the business cycle, and from this we know that what goes down must come up, recovery is just a matter of time, and surely now sufficient time has elaspsed, ergo..... Well, no. Business cycle theory does enable you to predict a recovery will come one day, but it does not tell you how far down, or for how long you will go first. Nor does it tell you how pronounced the recovery will be when it does come. For all of this we need more information, a bit more theory, and some bloody good guess work.
So what does that leave us with. Well the recovery is coming because the markets are already pricing it in. But what if the markets are already pricing it in because the market participants are expecting it to come. Doesn't this then become a bit circular. I think that today, for reasons which will become clear a bit later, is going to be be Robert Shiller day here at Bonobo.
Ever since Brad's admirable sally into the world of the Socratic Dialogue I haven't been able to get away from the problem of herd behaviour , and the associated phenomenon of information cascades.
And this is where Shiller comes in, since he has an admirable paper entitled: Conversation, Information and Herd Behaviour.
Experimental evidence shows that an important reason why people tend to imitate others, to exhibit "herd behaviour," is that they assume that others have information that justifies their actions. The informational cascade models of Banerjee (1992) and Bickchandani et al (1992) are significant developments in showing some general equilibrium and welfare effects of such rational imitative behaviour. But these models as specified may be of limited applicability since they assert that differences across groups in herd behaviour can be attributed to the random decisions of the first movers. Differences across groups might be explained more often across groups in terms of different modes of interpersonal information transmission. Patterns of human conversation imply great selectivity to the kinds of information transmitted within groups.
The problem which interests Shiller in this paper is not whether or not herding behaviour exists, he obviously takes it as read that it does, but how the decisions of the 'first movers' spread to the rest of the herd. And here he gets rather interesting because he attempts to use a communicational model to explain differences in receptivity across groups. We might consider that the phenomenon of the IT early adopter, and the differing ways this unfolds across societies, could well be a case in point. What determines how many 'first movers' - those who, by definition, do not trust the judgement of others to form their own - exist in any given society? (Here's one for you Maynard on the 'malleability' of cultures). Trust, and the forms of communication and social networks associated with it within different societies, would seem to be a crucial concept here. (Come to think of it, and entirely on the fly, this should have some implications for how we interpret our 'confidence' indices). And what determines the velocity of propagation, from first mover to herd? Well these are really questions rather than answers, but sometimes the questions themselves point us towards the answers. Shiller's conclusion is as follows:
Groups may differ in complicated ways in terms of what may be called informational cascade facilitators. For example, some groups may observe others in relation to certain kinds of information more often than other groups, or have different theories than other groups about the relevance of other group members' decisions to their own decision problem. Conversation patterns may also vary across groups in terms of habits of revealing sources of information. A seemingly inessential group tendency to reveal or not reveal wher one heard an improbable story may make an important difference in the kind of informational cascades that develop in conversation. If one knows that a more reliable person believed the story enough to tell it, then one would more logically have more tendency to believe it and tell it . Much research could be done documenting differences across groups in such dimensions.
The key expression in the final part is 'a more reliable person'. This is what I mean by trust and how it operates. So, to sum up, and return to where we started, since I am one of those people who congenitally has difficulty trusting the judgement of others to form my own, I am a recovery doubter. My reasons, a bit of information (see next post) a bit of theory (see website) and some bloody guesswork!