Modeling the Probability of Default of the Investment Projects

Abstract

Alexey Vladimirovich Morgunov - National Research University "Higher School of Economics".

E-mail: morgunov1986@mail.ru

The main objectives of the research: analysis of different methods of estimating the probability of default of investment projects; the practical application of the method of multiple choice for Russian investment projects (previously the author used the method of binary choice, the results of the practical application of which is published in 2015 in the journal of the New Economic Association" №2(26)) and comparison between methods of binary and multiple choice in accordance with the results of the validation using wellknown validation characteristics and criterias. In this article the basic approaches used in modeling the assessment of default probability for investment projects in a commercial bankes in the framework of the approach based on internal ratings (IRB Approach) were discussed. The results of the research connected with building models to estimate the probability of default for project finance transactions using multiple-choice method with high accuracy ratios (discriminatory ability) and stability (t-student test performed with confidence probability more then 85%) and obtaining a quantitative comparison of the quality of the developed model of multiple choice with the quality of the previously developed model of binary choice. For the most precise of the compared models (binary choice model) recalibration for 2015 year, taking into account the combined macroeconomic indicator developed by the author in 2015 year, was made. According to the results of the recalibration the decision making algorithm on project participation for credit institutions using K-S statistic was corrected. Obtained algorithms and models may be used by credit institutions to develope rating models based on internal ratings (IRB Approach). The feasibility of using the obtained results may be checked by banks according to results of the validation of the models on their own credit portfolios of investment projects. By results of research the conclusions about reasonability of application of the methods of binary and multiple choice to assess the creditworthiness of investment projects (the accuracy of the obtained models is high) and about the strong influence of  IRR on the creditworthiness of the projects were made.