In this paper, we develop an agent-based model of the financial market. Agent-based modeling is a simulation-based technique that is gaining popularity in economics. In an agent-based model autonomously acting and interacting units (e.g. representing financial market participants) endogenously generate structures and system properties.

In our model, boundedly rational agents trade a financial asset. Their trading strategy thereby depends on their return forecast which is formed byeither considering fundamentals or technical analysis. Agents are endowed with a balance sheet composed of a risky asset and cash on the asset side as well as equity capital and debt on the liabilities side. The risky asset is traded among agents at an endogenously set price. We assume that agents actively manage their respective balance sheet in two regards. Firstly, they choose a portfolio which optimizes the ratio between risky assets and cash conditional on their current return forecast, and secondly they aim at a fixed ratio between debt and equity (leverage ratio). Agents are constrained in their ability to acquire and dispose of debt by the credit supply of a risk managing financier and credit frictions, which hinder agents to make immediate changes to their debt levels.

We simulate our model and show that it can reproduce several empirically observable facts and relationships. Although we initially endow all agents with identical balance sheets, the size distribution of agents quickly converges to a lognormal distribution, which is typically observed for investment banks. We furthermore observe a natural tendency for inequality to increase over time. When we impose low credit frictions on the model financial market, leverage becomes procyclical, which is also typical for investment banks.

In a next step, we vary central parameters of the model exogenously in order to identify their effect on financial stability. By varying the leverage target of agents, we find that an increased target goes along with increased price volatility, more bankruptcies and higher systemic risk. It is hardly surprising that increased indebtedness of its participants makes a financial system less stable. When, on the other hand, varying the degree of credit frictions agents are confronted with, we find that lower frictions, which can be interpreted as an increased fraction of short-term credit on balance sheets, provokes complex repercussions. Specifically, lower credit frictions decrease the number of bankruptcies that typically occur within a specified time frame, but at the same time increase the probability of extreme events. Severe liquidity crises that can lead to a collapse of the entire financial system arise more frequently. Low credit frictions thus lead to a more stable model financial system most of the time, while systemic risk clearly increases. However, the introduction of a lender of last resort and an entity that gradually unwinds bankrupt agents can help to stabilize the model financial system.

Fischer, Thomas and Jesper Riedler (2012), Prices, Debt and Market Structure in an Agent-Based Model of the Financial Market, ZEW Discussion Paper No. 12-045, Mannheim. Download


Fischer, Thomas
Riedler, Jesper


Agent-Based Model, Financial Markets, Leverage, Systemic Risk, Credit Frictions