FORCASTING DE LA DEFAILLANCE FINANCIERE : ANALYSE DISCRIMINANTE

Corporate financial failure is a prominent issue in financial management. At first, initiated by RAMSTER (1931), then in a second wave with the work of BEAVER (1966), who introduced ratios to the prediction of financial failure by proposing a univariate analysis of the ratios, which is basically an analysis of the profiles of healthy and failing businesses. It was not until the end of the 1960s that the work of ALTMAN (1968) proposed failure prediction modeling using a multivariate analysis following the recommendations of BEAVER (1966). The ALTMAN model represents the first forecasting model that produced a discriminant function implementing several variables, which is a more suitable approach studying financial failure giving its complexity. Thus, a predicting financial distress model using multiple discriminant analysis is developed. The failure prediction literature then evolved along with the development of prediction methods, namely, artificial intelligence methods and data mining.The empirical analysis focused on the development of a predictive model by a commonly used statistical technique, namely discriminant analysis. We proceeded by estimating the variables with a strong discriminating power by linear regression method. The linear regression model is then weighted to obtain a model for classifying the observations into theoretical groups, which will eventually allow us to produce predictions with a forecast horizon of 2 years. The forecasting model was able to achieve a correctly classified sightings rate of 86%.The variables constituting the prediction model fall under the categories of ratios allowing the measurement of liquidity, profitability and indebtedness. The results acquired converge with the failure prediction literature that we have been able to consult. According to our model, financial failure comes down to problems of illiquidity, profitability as well as the level of indebtedness of the company