Schumpeterian Innovation in Modelling Decisions, Games, and Economic Behaviour

Von Neumann’s standard paradigm of a game in extensive form and Kolmogorov’s standard model of a stochastic process both rely on constructing a fixed state space large enough to include all possible future eventualities. This allows a typical single-person decision problem to be represented as a decision tree. Yet not all eventualities can be foreseen. Also, any practical decision model must limit the state space rather severely. In this way the standard paradigm excludes not only Schumpeter’s ideas regarding entrepreneurship, innovation and development, but also Shackle’s “unexpected events”. This paper proposes an alternative approach using “decision jungles” with an evolving state space. ∗Many thanks to Kenneth Arrow, Ken Binmore, Johannes Binswanger, Stan Metcalfe, Christian Seidl, Joanne Yoong, and a referee for their helpful suggestions, while absolving them of all responsibility for my errors or omissions. The paper has evolved from work done in applying for a Marie Curie Chair to be held at the University of Warwick from April 2007, under Contract No. MEXC-CT-2006-041121 with the European Community.