Multi-agent simulation of residential property value dynamics
Multi-agent based simulation (MAS), a new approach to modeling human societies moves away from the limitations of traditional large scale modeling. Individuals respond to their environment based on their own preferences and objectives, and the responses of others in that environment, with a larger pattern that emerges from these interactions. MAS agents ability to respond to and interact with local environment avoids the commonly cited problem of spatial autocorrelation in empirical studies involving spatial attributes such as property values. ^ But to what extent can MAS effectively model the property value dynamics associated with individual's attitudinal change towards residential dissonance and tax incentives? ^ This research uses MAS to examine the impact of changes in individual's attitude towards residential dissonance on property values and determine the contribution that a tax incentive program such as Keystone Opportunity Zone has on surrounding property values. ^ The effectiveness of MAS in modeling property value dynamics associated with changes in individual's attitudes towards residential dissonance and tax incentives is examined relative to the set of empirical observations (BRT data) and not against other methodology such as traditional statistics. The findings showed that MAS could be used to model property dynamics associated with changes in individuals' attitude towards residential dissonance and tax incentives. Firstly MAS could be used to test speculations about certain anticipated outcomes of a phenomenon. Secondly, MAS could be applied to simulate alternative scenarios by simply changing the parameter values. In this dissertation, the MAS results showed that the lower the tolerance (associated with a higher mobility in the movement behaviors of agents) the higher will be the mean property values. This pattern is similar to what is commonly seen in a buoyant property market. The MAS results showed that in the absence of the KOZ, the property values were lower with an estimated loss, in terms of MAS simulated mean property values, of about 0.77% per year. ^
Urban and Regional Planning
Yang Liang Chua,
"Multi-agent simulation of residential property value dynamics"
(January 1, 2005).
Dissertations available from ProQuest.