出版社：Faculdad de Ciencias Económicas, Universidad de Buenos Aires
摘要：We present a discussion of the concept of information in Economics.
We begin by presenting the prevalent model, closely associated with
the Maximization of Expected Utility hypothesis. While it has proved
to be useful in many ways, it also exhibits many shortcomings, which
we illustrate with some examples. On the other hand the evidence accumulated
by Experimental Economics shows that the actual behavior
of economic agents does not necessarily respond to the traditional characterization
that stems from von Neumann and Morgenstern’s (1943)
work.
We present, instead, the idea of information as represented by a
modular, updatable and linguistic structure. We draw from Computer
Science the concept of ‘database’ as an adequate structure with those
features. Logics provides a way of characterizing the selection of an
appropriate database up from real world data and underlying constraints.
These constraints do actually configure the information held
by an agent.
Anomalies in rational behavior as well as non-Bayesian information
updates can be modeled by means of this approach, although its full
development is still ahead.
