Neural Networks and their application in the fields of corporate finance

Eric Severin (1, 2, 3)

1 CES - Centre d'économie de la Sorbonne
2 SAMOS - Statistique Appliquée et MOdélisation Stochastique
3 CIS - Laboratory of Computer and Information Science

This article deals with the usefulness of neuronal networks in the area of corporate finance. Firstly, we highlight the initial applications of neural networks. One can distinguish two main types: layer networks and self organizing maps. As Altman al. (1994) underlined, the use of layer networks has improved the reclassifying rate in models of bankruptcy forecasting. These first applications improved bankruptcy forecasting by showing a relationship between capital structure and corporate performance. The results highlighted in our second part, show the pertinence of the use of the algorithm of Kohonen applied to qualitative variables (KACM). More particularly, in line with Altman (1968, 1984), one can suggest the coexistence of negative and positive effects of financial structure on performance. This result allows us to question scoring models and to conclude as to a non-linear relationship. In a larger framework, the methodology of Kohonen has allowed a better perception of the factors able to explain the leasing financing (Cottrell et al., 1996). The objective is here to explain the factors of the choice between leasing and banking loans. By using different variables, we highlight the characteristics of firms which most often use leasing. The corporate financing policy could be explained by: the cost of the financing, advantages of leasing or by the minimization of agency costs in leasing, we highlight a relationship between resorting to leasing and credit rationing.

Mots clés: neural netwoks; SOM; corporate finance

Domaines: Economies et finances

Eric Severin. Neural Networks and their application in the fields of corporate finance. 2008. ⟨hal-00325117⟩