CEF 1997: The Dynamics of Regional Interaction, Growth and Agglomeration - a simulation approach based on cellular automata

The Dynamics of Regional Interaction, Growth and Agglomeration - a simulation approach based on cellular automata

Max Keilbach - Technical University Berlin, FR 6-5

If we look at the spatial structure of an economy we observe a very heterogenous structure. On the one hand, we can see regions with high spatial density of production factors (i.e. agglomerated areas) and others with a low spatial density of these factors (i.e. rural areas). On the other hand we observe different regional growth dynamics, i.e. regions with a dynamic evolution of economic activities and other regions where growth is rather moderate.

The model to be presented starts from the hypotheses that these two phenomena are closely related to each other i.e. the regional growth dynamics cannot be seperated from the dynamics of agglomeration. Factors of production that are only locally available contribute to the local growth dynamics but have also an influence on agglomeration.

On the formal level, the model will integrate two lines of economic research, new growth theory and economic geography. It will be an endogenous growth model that includes local infrastructure and/or localized knowledge spillovers. Thus, it will emphasize the role that local interactions of factors play in fostering regional growth and agglomeration. The market structure will be a monopolistic competitive one, where factors will migrate according to profit opportunities in other regions. As opposed to capital that will migrate globally, labor will migrate only locally, due to regional preferences.

The model will be briefly presented in an analytical form for a two-region economy. Then, I will analyze a multi-region economy using a simulation approach. The simulation will be based on a cellular automaton (CA), where the automaton represents the economy and each cell represents a region. This approach should be ideal to represent regional interaction due to local spillover capacities of certain factors since it explicitly models local interactions. Based on the simulation results, the value added of this approach as compared to the analytical approach will be discussed. I expect the paper to contribute on four levels:

The model should demonstrate that if we consider growth and agglomeration as two phenomena of the same origin it is not reasonable anymore to discuss economic growth in terms of convergence-divergence. Convergence might rather be considered as a statistical artefact due to the location of borders.
Comparing both, analytical and simulation approach, I should be able to emphasize the merits of the simulation approach. The structure of agglomeration with its different densities of production factors and multiple centres cannot be analyzed in a two-region economy. Therefore a multi-region economy has to be considered that cannot be handled analytically.
The link of endogenous growth models to a complexity approach (a simulation of production structure within a framework of locally interacting agents) that will be done in the paper is — to my knowledge — still rather unexplored.
The spatial structure to come out of the simulation can be supposed to be a Christaller-Lösch one. A discussion of the equivalence and/or relation of these two approaches would then be indicated (if time schedule allows for).

Scheduled for Session 7.5 Learning Algorithms

[ Agenda | Sessions | Program ]

	The model should demonstrate that if we consider growth and agglomeration as two phenomena of the same origin it is not reasonable anymore to discuss economic growth in terms of convergence-divergence. Convergence might rather be considered as a statistical artefact due to the location of borders.
	Comparing both, analytical and simulation approach, I should be able to emphasize the merits of the simulation approach. The structure of agglomeration with its different densities of production factors and multiple centres cannot be analyzed in a two-region economy. Therefore a multi-region economy has to be considered that cannot be handled analytically.
	The link of endogenous growth models to a complexity approach (a simulation of production structure within a framework of locally interacting agents) that will be done in the paper is — to my knowledge — still rather unexplored.
	The spatial structure to come out of the simulation can be supposed to be a Christaller-Lösch one. A discussion of the equivalence and/or relation of these two approaches would then be indicated (if time schedule allows for).