COMPUTATIONAL SOCIAL SCIENCE

An Agent-Based Model of the Housing Market

An Agent-Based Model of the Housing Market

A simple agent-based model of a housing market built in NetLogo for a term project in CSS 610. In the model only the interest rates are set exogenously and the experiments are run according to the actual historical interest rates from the US. The results of these experiments are consistent with the bubble burst from middle 2007.

The model represents a conceptual/theoretical housing market, where the prices of the houses and the mortgages are determined endogenously by the interacting agents (people and banks), while the shocks are induced only by controlling for the interest rates exogenously. The premise for developing this model was the current financial crisis and the housing market bubble of 2008-2009, particularly in the USA. The hypothesis tested here with respect to the housing bubble is whether the path-dependent tweaking of the interest rates has led to the current crisis. By building this simple model, we try to verify this hypothesis or at least to observe some emergent properties in the housing market. With this level 1 type of model (Axtell & Epstein, 1994), we can observe patterns of development out of the individual interactions of agents qualitatively and try to test them with the empirical data.

Figure 1. A typical initialized model view of the 'Housing Market' showing foreclosures (houses in pink)

The figures/snapshots below are plots of the behavior of the model throughout the entire scenario 1993-2009. The big drop in the banks’ balance sheets corresponds to the decrease in the house prices and the increase in the average mortgage rates (pointed at by the arrows). This is the peak of the bubble burst, which is followed by a period of readjustment.

Click here to read the paper or here to run the model.

Click here to see more models