The local or global firm condition to reach the survival: an Italian SMEs spatial modelling research.

Abstract The traditional strategic management literature tends to assume that: a) the firm localization is a key driver to reach the competitiveness/survival of firm trough exchanges of knowledge within of the district and b) relies on external knowledge relationships necessarily means these are confined to the limited area (region, district, and other). Therefore, on the basis of this condition, in the last decades, a lot of researchers focused their analysis of bankruptcy prediction on basic data such as financial ratios, economic and industrial data, and business cycle indexes neglecting spatial condition. Therefore, a large number of studies have been focusing on how to improve the accuracy of prediction models, mainly by taking into account the increasing availability of firm-level data (Brunello and Langella, 2016; Jacobson and von Schedvin, 2015; Jones et al., 2017; Laitinen and Lukason, 2014; Lukason et al., 2016). In this way, economic-financial ratios are combined with information gathered from balanced sheets, applying new nonlinear statistical techniques seeking to discriminate between a healthy firm and a failure firm. This condition shows a scientific and empirical limit represented of a lack of spatial evaluation useful to analyses if the interaction of nearby companies has an influence on their chances of survival and if the behaviour of the firms change over the space. The paper tries to fill the gap considering spatial distributed data to predict the outcome variable at each location as a function of variables on the focal point and on nearby locations as well (Chen, 2011; Ciampi, 2015). To reach this aim, the research will be based on the spatial approach through the use of geographical micro data will provide a better explanation of the link between the business' insolvency and localization of the firms. 2 The authors analyse the Italian manufacturing SMEs' default probability determinants during two different periods: 2008-2012 and 2013-2016, using financial indices and micro geographical data.statistical techniques, as linear regression, linear discriminant analysis and related models, to the introduction of spatial effects, accounting how these can affect bankruptcy probability. Design/methodology/approach – The study is based on 12.241 manufacturing firms located in Central Italy. We apply spatial Autologistic model and Logit Regression Tree (LRT), with the aim to find evidence of spatial dependence and spatial heterogeneity in bankruptcy probability of the firms during two different periods: 2008-2012 and 2013-2016. Findings – The main findings confirm the usefulness of the spatial domain in bankruptcy analysis, to better explain the financial crisis of companies, validating the transmission impacts of failure process within the analyzed regions. Originality/value – Previous studies have been only occasionally applied the geographical analysis to the forecast of the bankruptcy of companies using financial indices and micro geographical data. Moreover, the use of an algorithm for model-based recursive partitioning provides a simple and clear graphical representation to identify spatial heterogeneous clusters of firms with different dynamics, in order to catch the increasing complexity in firms’ failure process.