The mortgage crisis and evictions in Barcelona: identifying the determinants of the spatial clustering of foreclosures

ABSTRACT

The article identifies the determinants of the uneven spatial distribution of housing accumulated by banks via foreclosures in the city of Barcelona. Working with a new data source, we geolocalised foreclosed housing and analysed its tendency to spatially cluster. Using the bivariate version of the Local Moran Indicator, we confirmed the spatial correlation between the concentration of foreclosed housing and indicators of the socio-economic vulnerability of the neighbourhoods containing it. We also applied an OLS model to identify and weight the determinants of this clustering at the neighbourhood level. Our findings revealed that the growth of unemployment, the concentration of (non-EU) immigrant population and a greater presence of residents with low levels of studies were the key variables that explain the uneven geography of foreclosures in Barcelona. The results obtained also allowed us to characterize the spatial distribution of the housing accumulated by the banks during the mortgage crisis. As a result of the massive wave of evictions, banks have emerged as large-scale property owners in Spain and key agents for present and future housing policies.

KEYWORDS:

• Foreclosures
• evictions
• mortgage crisis
• spatial clustering
• deprived neighbourhoods
• Barcelona

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Aaron Gutiérrez http://orcid.org/0000-0003-0557-6319

Antoni Domènech http://orcid.org/0000-0002-1881-6679

Notes

1 ‘¡Queremos un país de propietarios y no de proletarios!’ (We want a country of owners not proletarians!) This was a famous statement made by José Luis de Arrese, Secretary for Housing of the Franco Regime from 1957 to 1960.

2 http://ec.europa.eu/eurostat/statistics-explained/index.php/Housing_statistics (latest access: 10/09/17).

3 Since this date, official data aggregated by provinces has been provided by the National Institute of Statistics (Instituto Nacional de Estadística, INE).

4 http://www.bde.es/f/webbde/GAP/Secciones/SalaPrensa/NotasInformativas/Briefing_notes/es/notabe120614.pdf.

5 On April 22, 2016, the interim council of ministers formulated an appeal of unconstitutionality against Law 14/2015, which had been passed by the Parliament of Catalonia, concerning the tax on empty housing stock. The interim Spanish Government also included in its appeal the implementation of Article 161.1 of the Constitution, which provided for the automatic issuing of a restraining order affecting any regional law that was subject to an appeal by the central administration. This effectively implied the suspension of the Law immediately after the first payment of fees by banks that accumulated empty housing stock without destining this to social uses.

6 http://www.bcn.cat/estadistica/angles/dades/barris/index.htm.

7 In the case of GeoDa©, this refers to a pseudo p-value. Statistical significance in this case is based on a permutation test. This is obtained by comparing the Moran's I indicator relating to a distribution reference of random spatial conditions, in this case, it was obtained from 9999 permutations.

8 Through an exploratory analysis of the data, we were able to check the linearity of the model, assuming the normality and linear independence of the independent variables. We also checked the multicollinearity, using a variance inflation factor (VIF). The results showed that multicollinearity was not a problem in our estimations as the highest VIF value was considerably below the limit of 10 (Chatterjee & Hadi, 2012; Esteban & Altuzarra, 2016). In fact, in order to prevent collinearity between variables (which could have potentially occurred between the percentage without studies and the percentage with tertiary level studies; and between the percentage of housing with surface areas of less than 60m2 and with surface areas of greater than 90m²), we introduced two synthetic indicators into the model: one for education and another for the area of the housing surface (see Table 2).

Additional information

Funding

This research was supported by the “Agència de Gestió d'Ajuts Universitaris i de Recerca” Agency for Management of University and Research Grants of the Government of Catalonia [grant number 2017SGR22].