To model agent relationships in agent-based models, it is often necessary to incorporate a social network whose topology is commonly assumed to be "small-world". This is potentially problematic, as the classification is broad and covers a widerange of network statistics. Furthermore, real networks are often dynamic, in that edges and nodes can appear or disappear, and spatial, in that connections are influenced by an agent's position within a particular social space. These properties are difficult to achieve in current network formation tools. We have therefore developed a novel social network formation model, that creates and dynamically adjusts smallworld networks using local spatial interactions, whilst maintaining tunable global network statistics from across the broad space of possible small-world networks. It is therefore a useful tool for multi-agent simulations and diffusion processes, particularly those in which agents and edges die or are constrained in their movement within some social space. We also show, using a simple epidemiological diffusion model, that a range of networks can all satisfy the small-world criterion, but behave quite differently. This demonstrates that it is problematic to generalise results across the whole space of small-world networks.

In this paper we analyse the long-term costs and benefits of bailout strategies in models of networked banking systems. Unlike much of the current literature on financial contagion that focuses on systemic risk at one point in time, we consider adaptive banks that adjust risk taking in response to internal system dynamics and regulatory intervention, allowing us to analyse the potentially crucial moral hazard aspect associated with frequent bailouts. We demonstrate that whereas bailout generally serves as an effective tool to limit the size of bankruptcy cascades in the short term, inappropriate intervention strategies can encourage risk-taking and thus be inefficient and detrimental to long term system stability. We analyse points of long-term optimal bailout and discuss their dependence on the structure of the banking network. In the second part of the paper, we demonstrate that bailout efficiency can be improved by taking into account information about the topology of and risk allocation on the banking network, and demonstrate that finely tuned intervention strategies aimed at bailing out banks in configurations with some degree of anti-correlated risk have superior performance. These results demonstrate that a suitable intervention policy may be a useful tool for driving the banking system towards a more robust structure.

We create an agent-based banking model that allows the simulation of leverage cycles and financial contagion. Banks within our model adapt their investment strategies in an evolutionary manner according to the success of their competitors, creating an endogenous interbank loan network and a dynamic asset market as they try to maximise profit by adjusting their leverage. The system exhibits periods of slow risk growth and fast insolvency cascades, allowing us to assess both the size and frequency of those cascades over a long time-frame. We demonstrate that banks endogenise systemic risk into their leverage behaviour when the asset market is subject to either low or high levels of volatility, but are less successful for medium volatility when the number of bank insolvencies is maximised. We also show that in a low volatility environment, banks are more susceptible to systemic contagion. While the majority of insolvencies occur through the asset side of the balance sheet due to fire sales causing a rapid depreciation in the asset price, failures through the liability side of the balance sheet tend to be correlated, acting as an amplification mechanism to create far more serious cascades. By creating realistic cycles of growth and collapse, the model provides a suitable framework for performing further policy tests.

Participants in interbank payment systems manage a stream of payment requests of varying priority to minimise their total costs. However, individually optimal strategies may conflict with system-wide optimality and can lead to inefficient equilibria, where banks cannot meet obligations in a timely manner. We construct a model of a collateralised payment system and demonstrate that socially optimal states exist in which banks should delay a proportion of non-priority payments in an internal queue, but banks' strategising behaviour leads to liquidity hoarding and increased systemic cost. We discuss how this behaviour can be reduced using measures available to a regulator.

We create an agent-based banking model that allows the simulation of leverage cycles and financial contagion. Banks within our model adapt their investment strategies in an evolutionary manner according to the success of their competitors, creating an endogenous interbank loan network and a dynamic asset market as they try to maximise profit by adjusting their leverage. The system exhibits periods of slow risk growth and fast insolvency cascades, allowing us to assess both the size and frequency of those cascades over a long time-frame.We demonstrate that banks endogenise systemic risk into their leverage behaviour when the asset market is subject to either low or high levels of volatility, but are less successful for medium volatility when the number of bank insolvencies is maximised. We also show that in a low volatility environment, banks are more susceptible to systemic contagion. While the majority of insolvencies occur through the asset side of the balance sheet due to fire sales causing a rapid depreciation in the asset price, failures through the liability side of the balance sheet tend to be correlated, acting as an amplification mechanism to create far more serious cascades.By creating realistic cycles of growth and collapse, the model provides a suitable framework for performing further policy tests.