Factors contributing to banking incidents among SME's in France

Author : Herve LEBIGRE, Robert TROMMSDORFF

Abstract :The post-covid increase in bankruptcies of small and medium-sized businesses in France underlines the continuing importance of studying insolvency. Recent literature once again shows the need for in-depth analyses to assess the financial health of companies more accurately in models with ratios. It is in this context that we propose a study of banking incidents (BI), which constitutes a first original feature. BI involve the banker's decision to reject an unauthorized overdraft on a company's debit account. In this paper, we study possible groupings of ratios and accounting quantities using the PCA method. These will then be used to explain this leading indicator (BI), which is an important signal for the company and its stakeholders, such as bankers and chartered accountants. Our second original feature is that we are working on a database comprising the trial balances of 1,805 very small, small and medium-sized companies for the 2022 financial year, all clients of French chartered accountants. This provides us with recent and more detailed information than the simplified accounts available to the public, which are generally used in research for statistical studies on the risk of failure of this type of company. We have chosen to use two approaches, one for all the companies in the database and the other solely for joint-stock companies. In this first paper, we start by reducing the dimensionality of the data to identify hidden structures. The results of the analyses may sometimes seem counter-intuitive. Economic performance is clearly at the top of the dimensions in both panels. This axis ranks ahead of cash flow as the explanatory factor in the “all companies” database. Moreover, in the panel of joint-stock companies, the cash flow dimension no longer appears. The results of the logistic regression show models with a strong ability to predict cases with no banking incident in samples with a banking incident rate of no more than 14.5% of the base studied.

Keywords :Bankruptcies, financial health, banking incidents, predictive modeling

Conference Name :International Conference on Finance, Bank & Economics (ICFBE-24)

Conference Place Paris, France

Conference Date 6th Dec 2024