Risk analysis and controllability of credit market

Abstract

Abstract. In this paper we present the behaviours of a simple credit market model built upon a static directed complex network topology. We found that the market structure plays a key role on the stability of such a model. Cyclic topologies accumulate more debt than acyclic ones, and when the total cash is less than the total debt, the market becomes unstable and exhibits chaotic behaviour. Exogenous creation of currency is used to control the market stability. Two regimes of control emerge depending on the credit market structure. First, power law control is needed for acyclic random topologies built upon Erdos-Rényi and Barabasi-Albert algorithms. Second, exponential control is needed to stabilize cyclic market built with the Watts-Strogatz algorithm.Résumé. Dans cet article, nous présentons les comportements d'un modèle simplifié du marché du crédit. La structure du marché est construite autour d'un réseau complexe statique. Nos résultats montrent que la structure du marché joue un rôle crucial dans la stabilité du modèle. Les topologies cycliques accumulent plus de dettes que les topologies acycliques et lorsque la dette accumulée excède le montant total des liquidités, le marché devient instable et converge vers un comportement chaotique. La création exogène de liquidité est utilisée pour contrôler la stabilité du marché. Deux types de contrôle qui dépendent de la structure du marché du créditémergent. Le premier type de contrôle est le contrôle suivant une loi de puissance. Ce contrôle est nécessaire pour stabiliser un marché basé sur les topologies acycliques du type réseaux de Erdos-Rényi et Barabasi-Albert. Le second type de contrôle est le contrôle suivant une loi exponentielle qui est nécessaire pour stabiliser un marché basé sur les topologies cycliques du type réseaux de Watts-Strogatz.

Altering control modes of complex networks based on edge removal

“…Therefore, we can also easily change a network from centralized modes to distributed modes. 4 and average degree <k>∈ [5,25]. For each <k>, we generate 100 networks.…”

“…Considerable research efforts have been devoted to analyzing control principle of various real-world networks. For example, network controllability has been used to analyzing the interbank loan network 4,5 , finding the underlying mechanism of human brain network [6][7][8] , identifying disease genes [9][10][11][12][13] and finding drug target [13][14][15] .…”

Impact of immigrants on a multi-agent economical system

“…The model we consider here has been proposed and developed in [ 22 – 26 ]. It is based on the interactions between agents sitting at the nodes of a given complex network (whether random, scale-free or from observational data).…”

“…Note that in [ 26 ], a credit market model has also been introduced to study the stability of a market in which banks lend money for a given time to other agents of the market. There, the behavior of the total debt is considered as well as the possibility to control the market away from bankruptcy.…”

“…This memory effect can of course be extended to more time steps, if needed. It is expected to dampen the “ping-pong” effect between agents, but not to remove possible chaotic regimes, as observed in [ 26 ].…”