Spatial autocorrelation and clusters in modelling corporate bankruptcy of manufacturing firms

The interest in the prediction of firms’ bankruptcy is increasing in the recent recession period 2008–2012, when, in Italy, the number of distressed manufacturing firms increased sharply. The most popular model applied by bankruptcy researchers is the logit model (logistic regression model). In the present paper we extend this classical model in two different ways, to take into account the spatial effects that can highly affect bankruptcy probability. We propose to apply the spatial Autologistic model and the Logit Regression Tree, with the aim to find evidence of spatial dependence and spatial heterogeneity in bankruptcy probability, of the manufacturing firms of Prato and Florence (Italy). Our application shows that spatial contagion effects are an important issue when modelling bankruptcy probability. Moreover, the application of the regression tree analysis shows the presence of three different clusters, with heterogeneous behaviours.