Collins Conference Room
Meeting
Speaker: 
Alan Kirman (CAMS, Ecole des Hautes Etudes en Sciences Sociales, Paris)

This event is closed.

Abstract:   Heterogeneity is the reality of our world. We have an economic system that is composed of individuals who differ in their characteristics and in their choices, and who change themselves and their environment through their interactions. The fact that agents differ from one another is the basic driving force behind all economic activity. If agents were identical there would be little interest in their interacting with each other. When agents are too similar in their beliefs or characteristics we are rapidly led to “no trade” theorems in total contradiction with the amount of trade that actually prevails It is clear that one must take full account of the interaction between agents.

Heterogeneity almost inevitably leads the economy to be a complex adaptive system whose properties are emergent. In its most apparent form, emergence can lead to congestion or stampedes, even though on an individual basis no one is taking actions intended toward that end. This complexity leads to computational irreducibility, where analytical methods fail the task. The economy is not susceptible to being modeled as a system in which individuals behave according to some axioms where their behavior generates an equilibrium which can be shown to exist and whose characteristics can be solved.

Furthermore, having many heterogeneous agents can lead the system to be overidentified, so even if the tools of standard methods could be applied, the result will not be a closed system. As a result, heterogeneous agent models move from deduction to simulation. Integrating heterogeneity into the economist’s worldview not only changes the approach to analyzing models, it also changes the objective of the models. If heterogeneity matters, a different set of heterogeneous agents will lead to a different result. And this in turn means that rather than seeking a theoretical result, we are in the world of pragmatism, of engineering, of case studies. 

Banks are often thought of as single decision-making units but our main example examines the feedbacks from one part of a large financial institution to another. Again, the different functions of the various agents are associated with different rules and the sort of cascading events that, for example, characterized the Bear Sterns collapse can be clearly seen in this sort of framework. Notice that, as we make more realistic assumptions about the actors in the market or institution, the model becomes more specific and cannot be simply applied to any financial institution. To repeat, we move from theory to pragmatism.

 (Joint work with Richard BookstaberOffice of the Chief Investment Officer, University of California, Oakland, CA, United States)

Purpose: 
Research Collaboration
SFI Host: 
Sam Bowles