Clustering of territorial areas : a multi-criteria districting problem

Endogenous resources, economic profile and socio-economic issues are the criteria that define the level of development and identifying features of a territorial unit. The territorial units that organize the country, in political and administrative terms - parishes and counties -, have a hierarchical structure, which initially reflected the organization of productive activities as well as the traditional State organization. The success of development policies addressed to territorial agglomerates depends on the homogeneity of their territorial units. In this context, the clustering of territorial areas can be stated as a multi-criteria districting problem. Thus, this paper aims to propose a framework for obtaining homogenous territorial clusters based on a Pareto frontier that includes multiple criteria related to territories' endogenous resources, economic profile and socio-cultural features. This framework is developed in two phases. First, the criteria correlated with development at the territorial unit level are determined through statistical and econometric methods. Then, a multi-criteria approach is developed to allocate each territorial unit to a territorial agglomerate, according to the Pareto frontier established. The framework is applied to a set of parishes and counties of the Central Alentejo region in southern Portugal. Results are presented and discussed in the scope of a regional development strategy. The results of multiple linear regression analysis show us the most important variables in explaining the differences in development in the area considered. We conclude, as expected, that the more elderly the population or the higher the school drop-out rate, the lower the area's development. On the other hand, the greater the active population or the rate of employment in tertiary social activities, the greater is the development. In the 2nd part of the analysis, we started from the current situation in terms of administrative organization of parishes. The results of the Max-p-model show that tests to increase the homogeneity between parishes, using the variables of population size and area, it is possible to reduce the disparity between parishes, reducing the number of units. The simulations show that the number of parishes may be lower if the variable of analysis is population size. This result takes into account the wide disparity of the population in current parishes, as well as the small number of inhabitants in most places.