Analyzing the Profitability Performance of SMEs Using a Neural Model

Special models of artificial neural networks (ANNs) have proven their worth in various and sometime unexpected domains. In this paper, our focus was to develop an ANN application in order to analyze the financial performance of the SMEs in Romania. For historical reasons, this sector seems to be still weakly developed in that country, both quantitative (being situated on one of the last places in the EU's entrepreneurial intensity) and qualitative, having a weak economic performance with a modest contribution to GDP. Literature shows the importance of this sector for the economies of different countries, and diverse scientific methods used for its description and analysis. One of our research purposes was the identification of those factors that condition the profitability of companies, thus providing useful directions and possible strategies for developing the SME sector. The selected information source was represented by the annual balance sheets, from about 8000 of medium-sized companies in Romania. As a means of verifying the obtained results, econometric methods were used, such as regression analysis, which could identify and validate the models that emphasize the dynamics with different influence factors. The conclusions obtained could prove their utility in both the investigation of the combining quantitative methods (ANN and regression), and in the SME sector management plan.