Abstract:
This thesis is the outcome of theory development research into an identified gap
in knowledge about systemic risk of the global financial system. It takes a
systems-theoretic approach, incorporating a simulation-constructivist orientation
towards the meaning of theory and theory development, within a realist
constructivism epistemology for knowledge generation about complex social
phenomena. The specific purpose of which is to describe systemic risk of failure,
and explain how it occurs in the global financial system, in order to diagnose and
understand circumstances in which it arises, and offer insights into how that risk
may be mitigated.
An outline theory is developed, introducing a new operational definition of
systemic risk of failure in which notions from evolutionary economics, finance
and complexity science are combined with a general interpretation of entropy, to
explain how catastrophic phenomena arise in that system. When a conceptual
model incorporating the Icelandic financial system failure over the years 2003 –
2008 is constructed from this theory, and the results of simulation experiments
using a verified computational representation of the model are validated with
empirical data from that event, and corroborated by theoretical triangulation, a
null-hypothesis about the theory is refuted. Furthermore, results show that
interplay between a lack of diversity in system participation strategies and shared
exposure to potential losses may be a key operational mechanism of catastrophic
tensions arising in the supply and demand of financial services. These findings
suggest new policy guidance for pre-emptive intervention calls for improved
operational transparency from system participants, and prompt access to data
about their operational behaviour, in order to prevent positive feedback inducing a
failure of the system to operate within required parameters.
The theory is then revised to reflect new insights exposed by simulation, and
finally submitted as a new theory capable of unifying existing knowledge in this
problem domain.