دانلود رایگان مقاله انگلیسی همبستگی خودکار و خوشه ای در مدل سازی ورشکستگی شرکت های تولیدی - اشپرینگر 2018

Abstract

The interest in the prediction of frms’ bankruptcy is increasing in the recent recession period 2008–2012, when, in Italy, the number of distressed manufacturing frms 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 diferent ways, to take into account the spatial efects that can highly afect bankruptcy probability. We propose to apply the spatial Autologistic model and the Logit Regression Tree, with the aim to fnd evidence of spatial dependence and spatial heterogeneity in bankruptcy probability, of the manufacturing frms of Prato and Florence (Italy). Our application shows that spatial contagion efects are an important issue when modelling bankruptcy probability. Moreover, the application of the regression tree analysis shows the presence of three diferent clusters, with heterogeneous behaviours.

Conclusions

In the present paper we applied the bankruptcy logit model on the manufacturing frms of the Local Market Areas of Prato and Florence (Italy), to identify the probability of the frms to survive or to fail and exit from the market in the period 2008–2012. The potential contagion efects on interconnected frms, generated by chain reactions and liquidity tensions, will produce a positive spatial efect on neighbour frms. Our concern is that the presence of this spatial dependence and spatial heterogeneity is a very important characteristic of spatial data, that cannot be neglected, when studying the propagation efects of bankruptcy. Therefore, we extended the classic Logit model in two ways, for taking into account these two different spatial efects: the spatial Autologistic model and the Logistic regression tree. Our results highlight that the spatial model outperformed the classical one and the heterogeneity in the relation defning bankruptcy probability suggests the presence of three diferent clusters. Our spatial model shows that the probability to be a solvent frm is three times higher when similar solvent frms are in the neighbourhood. All our estimated models confrm the presence of a positive efect of age and size on the probability to survive and, instead, that the economic activity is not relevant to the defnition of the homogenous groups. Moreover, the spatial distribution of the frms among the three identifed clusters doesn’t follow the boundaries of the LMAs. These fndings suggest that the negative efect of the recession and its propagation on the small and medium enterprises can be defned by the spatial propagation and the frm’s value chains. Therefore, the clusters framework seems to be explained more by the frms’ performance, than by the result of the social cohesion within the local community of people (Ortega-Colomer and Molina-Morales 2016). These results are in line with Porter assumptions and should also confrm the assertion of Ramazzotti (2010) that in period of stagnation or recession, the conditions for the persistence of industrial districts may disappear, leading to diferent types of local organisation, such as BC. Future research directions will extend our analysis by merging together both methodologies proposed in the present paper, and applying a spatial autologistic model in the classifcation tree procedure. Furthermore, the analysis should be extended by including more variables in the estimated model, with the aim to explore how innovation and globalization can afect survival probability of the frms.