Modelling spatial distribution of spill-over effects resulting from urban development

Declining growth is – despite current immigration – a phenomenon occurring especially in peripheral regions in Germany lacking job offers and accessibility. Decline is characterised by decreasing number of inhabitants, a high debit per capita in terms of dues for infrastructure and services and in many cases by high debts of the municipality. In terms of a sustainable urban development and a long-term calculation municipalities try to retrofit the building stock instead of developing new building land. In order to keep demand and supply on the housing market in balance one can even suppose they reduce the building stock. However, many municipalities provide new building land to generate new buyers and attract young families. Those measures can trigger higher vacancy rates in the core (‘donut effect’). One reason is the neglect of spill-over-effects by municipalities when calculating urban measures. Empirical analyses of standard land values between 2011 (after census) and 2014 in ‘shrinking’ municipalities which developed new building land showed changes in the value of building stock. Yet, spill-over-effects have not been taken into account, but only costs for developing building plots. One reason is the problem of modelling the spatial distribution of these effects and the mapping to single real estates. Therefore, the resemblance of real estates is to be known as well as the spatial relation. This paper shows an approach for modelling the spatial distribution of spill-over-effects of building measures. Assuming that building measures affect ‘similar’ neighbourhoods in particular, the resemblance has to be determined first. The resemblance of housing estates representing different neighbourhoods was analysed applying cluster analyses. The data used originate from the real estate platform www.immobilienscout24.de and comprises real offers in the region of northern Eifel (Germany) between November 2015 and January 2016. Based on the clusters generated neighbourhoods were classified. In order to model the spatial distribution of spill-over-effects, this spatially related data were processed in a spatial lag model. The data were processed in R and linked to QGis. This enables municipalities to apply the analyses by themselves.