Complexity and Multilayer Networks in Economics and Finance
Begeleiding : Prof. Koen Schoors
(promotor, Faculteit Economie en Bedrijfskunde), Jan Ryckebusch (promotor),
Benjamin Vandermarliere, Andres Belaza, Milan van den Heuvel, Ken Bastiaensen
One of fhe most puzzling things about the financial crisis of
2008 is that it was caused by a relatively small group of financial
products. The subprime exposure, which was often believed to be rather
marginal, gave rise to widespread problems, and started a domino
effect which made the entire financial system to topple. In
physicist's jargon, this means that a local perturbation affected the
system as a whole. This was possible because over the years, the
financial system grew more and more complex. Thereby, all agents in the financial system are highly
interconnected very by means of unrecognized links. Policy
makers tried to stabilize the system by devising rules for each
individual player. In physics it is well known that "The whole is most
often more than the sum of its parts", and systemic risk is also at
play in the highly interconnected financial system.
The financial calamities of 2008 has boosted a lot research into the complexity of interbank networks in particular, and
in economics and finance in general. This field is highly
interdisciplinary drawing on big data manipulation, advanced statistical analyses, computational sciences, finance, game
theory, network theory and complexity science. We propose two master's thesis subjects in a timely and
very exciting field of interdisciplinary Physics.
Subject 1: Controllability in interbank networks
The application of network theory to interbank systems already
resulted in major advances in our understanding about their topology,
interaction dynamics, and growth. A major issue is the endeavour to
steer the evolution of the banking system in the direction of
optimized control. Thereby, one aims at defining the rules of the operation of the
interbank credit system such that it evolves in a certain direction.
The master's thesis will address the issue of structural
controllability of interbank credit networks under different
conditions. Using principles of network theory one can identify the
nodes which need to be controlled in order to achieve a status whereby
one can drive the network from an initial state to a selected final
state. Thereby, it is essential to possess an effective method or
algorithm to identify the driver nodes at each instant of time.
Controllability in network science is a relatively new theory. After
having mastered the concept of controllablity, the student will study
a version of the CRISIS
agent-based interbank model. She/he will analyse the time series of
the evolution of the interbank system under different scenarios. In
particular, a study will be made of the connection between
controllability and the underlying dynamics of the simulated
network. The concept of control very much hinges on the (non)existence
of a link between different nodes in the network. Traditionally,
interactions between nodes are aggregated into a single network
layer. In practice, interactions can adopt very different forms and
some interactions may be more effective than others to gain control of the nodes.
To this end, one has developed the concepts of multilayer networks ( A
recent paper on the topic of Multilayer Networks
). In the context
of the research for the master's thesis, it will be investigated
whether the theory of Multilayer Networks provides a superior way of
defining the controllability of the system.
Subject 2: Discrete Scale Invariance and Exchange-traded funds
An exchange-traded fund (ETF) is a fairly new financial product which enables investors to easily own and trade a share of large and well-diversified portfolios. In the total net assets distribution of these ETFs we have found indications of discrete scale invariance (DSI). The first goal of this thesis will be to analyze rigorously whether there actually is DSI or not, where the student will implement the Lomb diagram technique
. The second goal of the thesis, if the DSI hypothesis is confirmed, will be to understand and model the underlying mechanism.
[Terug naar overzicht