Parsimonious models of financial insolvency in small companies

This study is an extension of current research on insolvency diagnosis. We intend to demonstrate that in small firms, the relevant information for the preventive diagnosis of insolvency can be synthesised in a model built upon a more reduced number of economic and financial ratios than the ones generally used in this kind of study. Our approach produces parsimonious models that can extract information from publicly available accounting-financial data. We demonstrate that using an extensive exploratory stage that will monitor the effects of correlation between financial variables, we will be able to build relatively stable models with a small set of variables. The results of the models built by resorting to discriminant analysis and to logistic regression present a similar accuracy to models previously developed. Our models present the advantage of including a small number of variables that can be interpreted in the light of current financial theory and therefore it reduces the number of financial data needed to make an insolvency diagnosis. This is particularly decisive when working in an environment of restricted information availability, which is very common in small companies.