APPLICATION OF SPECIFIC METHODS IN ASSESSMENT AND DEVELOPMENT OF RATING SYSTEMS OF FINANCIAL INSTITUTIONS

Abstract

The rating system is an important factor in risk management and financial performance assessment. Until the beginning of the eighties of the last century, credit risk assessments of clients were carried out in a traditional way, which was mainly reduced to intuition and subjective assessment of internal rating by the management of financial institutions, while counting on their ability to make quality decisions based on knowledge and expertise. After that period, the global standardization of the rating system was carried out, which enabled an easier assessment and comparison of financial institutions. Activities related to evaluations and development of rating systems in financial institutions today must necessarily be an integral part of their ongoing operations and risk management culture, and it is also important that financial institutions are able to respond to the specific minimum requirements of the internal rating system, risk management process and capabilities assessments of their necessary components. For some risk exposure classes, the Basel Committee proposes a basic methodology by which financial institutions take their own risk assessment as an input, while assessments of additional risk factors are carried out through the application of standardized rules. At the same time as the basic methodology, advanced methodologies have been established that allow the use of one's own internal assessments of risk components. Wide use of such assessments is an important part of the dynamic and risk-sensitive IRB approach (Internal Rating Based) in such a way as to recognize and differentiate those financial institutions that are able to conduct a sufficiently valid and quantified risk assessment. Along with the standardization of these activities and the development of information systems, quantitative models are largely included in this type of analysis in order to improve the objectivity of predicting the probability of default (PD - Probability of Default) and expected losses. These models include financial indicators, macroeconomic conditions and historical data on loans and borrowers. Regression analysis, discriminant analysis, panel models, hazard models and neural networks are the most common and sophisticated techniques that can be used to assess the probability of debtor default. It can be concluded that, through the results obtained through the mentioned methods, the management of financial institutions can look more complexly and objectively at the real pictures of potential debtors, which in the end enables a better assessment of their default, and therefore the mentioned results of the analysis have a favorable effect on the development of the rating of financial institutions. It is recommended that when approaching this type of analysis, several techniques and methods mentioned above are simultaneously used and that the causes of the differences in the results are thoroughly analyzed, which in any case reduces the probability of the risk of non-payment.