Modeling Gentrification with Theoretical Physics of Complex Systems: Epidemic Spreading

Abstract

Since Glass researched gentrification for the first time in history, most of the conventional research on gentrification have focused on the social aspect of the phenomenon conducted with qualitative methodology. Despite of some strengths of qualitative methodologies, research with quantitative methodologies would suggest more solid basis to the research on gentrification with urban data acquired from real space.

Quantitative approaches to analyze and simulate gentrification have appeared after 2000. They considered numerous possible geographical, economical, and social factors and variables that might be related with gentrification. They calculated and simulated gentrification numerically based on a "decision making algorithm" and the change of social, economic, and physical characteristics of a neighborhood with time.

However, few research that analyzed gentrification analytically using research methodologies from theoretical physics could be found. With more concise equations and variables, the quantitative analysis using physical methodology would help our understanding on mechanisms and dynamics of complicated gentrification phenomenon conceptually.

Components such as buildings, roads, transportations systems, and people in a city compose complex "urban network systems" and interact with each other, generating diverse urban activities. Urban form and structure can be considered as spatial network systems, "graphs whose nodes represent the dynamical units, and whose links stand for the interaction".

The gentrification phenomenon, a result of complex combination of numerous urban activities, could also be analyzed and described using complex systems by viewing neighborhoods as spatial network systems.

This research focuses on the physical aspect of gentrification. Physical transformation can be measured with the demolition and construction of building because in gentrified neighborhoods many buildings are repaired and replaced rapidly.

This research is based on the hypothesis that neighboring buildings interact with each other, i.e. when a building in a neighborhood is replaced with a new one, buildings next to that would be pressured to be changed: the development of a building is commonly followed by developments of contiguous buildings, which is caused by attracting visitors and/or new residents and increase of property value.

An assumption of this research is to define the gentrification phenomenon as the "spreading of reconstructions" by applying the epdemic spreading model. The purpose of this research is to build a model describing the "spreading of constructions" in a bounded gentrifying area.

According to the interpretation of gentrification as a process involving the spread of an infection, the old buildings correspond to S (susceptible state) and the new buildings relevant to I (infection state). Based on this relationship between the gentrification and the epidemic model, when the gentrification is progressed, there would be no step corresponding to the recovering process. As there is no process that new buildings transform back to the old buildings. Among the various epidemic models, I selected the SI model which does not consider recovering process to calculate the gentrification.

The gentrification models consist of three factors. In addition to the epidemic spreading mechanism, I introduced two other factors influencing reconstructions, deterioration and economic boom, for modeling gentrification. According to the combination of three factors, I defined the following four models to modeling gentrification: The Basic Model, Age Model, Economic Boom Model, and Integrated Model.

This model was applied to three residential communities in Seoul which is gentrified with commercialization, containing 400 to 500 buildings within walking distance, Bong Cheon, Hong Dae, and Garosu Gil.

According to results of modeling, reconstruction of buildings has spread most rapidly in Bong Cheon where land use was changed in 1998. On the other hand, Garosu Gil, where transformations have occurred by remodeling the existing buildings, was redeveloped in a rate less slowly than the rate of reconstructions induced by deterioration. However, reconstructions in Garosu Gil were most brisk from 2001 to 2004, economically booming periods.

By using two model testing indicators, RMSD and MAPE, it was verified that considering two factors, the Age factor and Economic Boom factor makes the modeling more articulate to describe the change close to actual data. Particularly in the Integrated Model, values of MAPE were less than 8% in all of the three sites (Bong Cheon: 7.7%, Hong Dae: 4.8%, Garosu Gil: 2.5%) and values of RMSD were not more than 1% (Bong Cheon: 1.0%, Hong Dae: 0.6%, Garosu Gil 0.7%).

With the Integrated Model, Buildings in Bong Cheon where Parameter G is the largest are predicted to be most rapidly reconstructed, 70% of buildings being predicted to be reconstructed by 2040. In Garosu Gil with Smallest G, 40% of buildings are predicted to be reconstructed by 2040. 50% of buildings in Hong Dae are expected to be reconstructed for the same period.

A similarity of the trend of parameter G was found in 3 sites. It implies the correlation between parameter G and economic condition of real estate market. The relationship was confirmed by comparing the trend of G and the Housing Price Index in Seoul.

This research confirmed that land use change can accelerate the reconstruction of buildings. The rapid reconstructions can lead to the increase of housing price and rent, followed by the displacement of residents in the sites. Public interventions to prevent gentrification are required to the areas whose land use was changed, especially whose limit of FAR has been increased. For the site where reconstruction of building is activated like Bong Cheon, more careful interventions by local government are required to facilitate the supply of affordable housings.