Bankruptcy prediction - A litterature review

Abstract

Developing a failure prediction model for small businesses requires recognizing some conceptual issues that analysts may face. Indeed, these problems mainly come from the absence of an empirically validated theoretical model that represents the process of progression of SMEs towards failure. They also arise from different perspectives that can be linked to the analysis. This reflects the diversity of expectations of the partners of SMEs, from a simple ambition to measure the risk of failure, a vision of "predicting failure", to  a predilection to identify the financial factors announcing managerial problems that must be corrected to avoid failure, the “prevention of failure” vision. Thus, according to Malecot (1997), according to this perspective "if we manage to foresee the risk of failure of a company sufficiently early, it will be possible to remedy it in a certain number of cases, notwithstanding the diversity and the multiplicity of causes at the origin of these failures”. Thus, we concluded, first of all, that in order to be in a failure prevention logic, the construction of the model must be achieved based on data relating to companies failing economically, not legally. Then, to assure the extrapolation of the developed models, the distribution of the samples by group used to develop these models must be representative of reality. Thus, the option of stratification with a balanced representation of observations in each group should be eliminated.