Companies Predisposition to Horizontal Agreements Modeled by Neural Networks

The paper advances a method of predicting, based on their market share and turnover, which companies in a given market would be subject to hard-core agreements between competitors, which break the provisions of the competition law. We track the correlations between the market share, turnover and anticompetitive behaviors, and we construct a neural network model to discriminate between companies not entering anticompetitive agreements, and companies, which are vulnerable to this kind of anticompetitive practices. The conclusions can be extended to various sectors of activity and to various company sizes.