The residential sector is a crucial segment in Italy, as it represents the most significant component of household wealth, and homeownership greatly prevails (around 73%) in comparison to other European countries. Moreover, the construction sector is an important contributor to domestic economy (about 10% of the GDP), but the last market cycle had a substantial impact on the construction business. The most significant downsizing concerned the constructions, as they have lost more than 10 percent of companies, almost 20 percent of employees and around 30 percent of the gross fixed capital formation. A weak economic prospect hit the property market, dramatically reduced the volume of transactions and bring the house prices down in most cities. This prolonged downturn of the market freeze the capital flow in most of part of the country, but different towns experienced different reactions to the global economic crisis.

This paper aims at analysing the particular condition of the Italian residential market of the last 20 years, clustering the Italian cities into groups according to their behaviour and their characterization. The novelty of the approach proposed here lies in using not only market data and socio-economic covariates for the analysed cities, but also trends and dynamics shown by the adjoining and surrounding territorial units, according to the principles of spatial data analysis. Indeed, following the so-called first law of geography (Tobler, 1970), “everything is related to everything else, but near things are more related than distant things”. Therefore, we expect that the real estate market fundamentals in a city are affected, to a certain extent, by what happens in the neighbouring cities.

The method has been implemented using the data of the last two market cycles (expansion: 1999 – 2006; recession: 2007 – 2016), employing a variety of demand-side and supply-side variables, such as the number of transactions, house stock, house prices, building permits, quality of life, index of sustainability. The analysis has been conducted considering data organised at NUTS 3 level (Nomenclature of Territorial Units for Statistics), namely, the Provinces. Custer analysis is then used to determine the composition of different submarkets and to characterise them, in order to understand their different conditions and to assess the weakness of the various submarkets during the expansion and the downturn of the market.