Algorithm for solving the reverse stress testing problem of the bank's loan portfolio on basis of system dynamic models of borrowers

D.S. Kurennoy


The key mechanisms to determine the effectiveness of bank's activities is the credit risk management. One of the main procedures recommended to banks for risk assessment is reverse stress testing. Reverse stress testing is the construction of realistic scenarios that lead to a given level of financial loss, or the formation of realistic scenarios that maximize bank losses. Knowledge of such scenarios allows banks to mitigate the consequences of their implementation.

The current credit risk assessment models are not suitable for solving the problems of reverse stress testing. Widely known models do not take into account the structure of specific companies, do not allow investigating the development of crisis scenarios in time and assuming a large sample of data on similar enterprises. To avoid of the noted disadvantages of the traditional approaches allows the use system-dynamic models of borrowers. System dynamics provides the possibility of reproducing the enterprise’s structure under study in the form of continuously interacting elements and external factors. The links between the elements are described by functional dependencies and differential equations that determine the dynamics of the company and the degree of its stability in relation to various macroeconomic scenarios. The article is devoted to approximate dynamic programming algorithm, which solves the reverse stress testing problem for the bank's loan portfolio on the basis of system-dynamic models of borrowers. The algorithm uses classical quasi-Newton optimization methods. The results of the method work are compared with the results of genetic algorithms. The statements proved in the article justify correctness of the algorithm and give an idea of its applicability to nonsmooth optimization problems arising in the context of reverse stress testing and system dynamics. The main tool for implementing the described algorithm is Matlab.

Full Text:

PDF (Russian)


Bollerslev, T. Generalized autoregressive conditional heteroskedasticity // Journal of Econometrics. - 1986. - N 31. - P. 309-328.

Engle, R. F. Autoregressive conditional heteroskedasticity with estimates of the UK inflation // Econometrica. - 1982. - N 50. - P. 987-1008.

Flood M. D., Korenko G. G. Systematic scenario selection: stress testing and the nature of uncertainty // Office of financial research. 2013.

Forrester J. W. Industrial Dynamics / MIT Press. 1961.

Glasserman P., Kang C., Kang W. Stress scenario selection by empirical likelihood // Office of financial research. 2012.

Guidelines on stress testing / Committee of European banking supervisors. 2010.

Gurný P., Gurný M. Comparison of credit scoring models on probability of default estimation for us banks // Prague economic papers. 2013.

MATLAB. The Language of Technical Computing. Using MATLAB. MathWorks Inc., 1999.


Powell W.B. Approximate dynamic programming: Solving the curses of dimensionality / Wiley. 2007.

Principles for sound stress testing practices and supervision / Basel committee on banking supervision. 2009.

Schwarz, G. Estimating the dimension of a model // Annals of Statistics. – 1978. – N 2. – P. 461-464.

Schwefel, H. P. Numerical Optimization of Computer Models / Wiley. 1981.

Sterman J. D. Business Dynamics: Systems Thinking and Modeling for a Complex World. Boston: McGraw-Hill Companies. 2000.

Bellman R. Dynamic programming. – M.: Foreign Languages Publishing House, 1960.

Boks Dzh., Dzhenkins T. : Time Series Analysis, Forecasting and Control, M.: Mir Publishers, 1974.

Izmailov A. F. Solodov M. V. Numerical Methods of Optimization. – M.: Fizmatlit, 2008.

Kurennoy D. S., Golembiovskiy D. Yu. Estimating the probability of oil company default based on system dynamics model // Issues of Risk Analysis. 2018. №02. P. 86-92.

Kurennoy D. S., Golembiovskiy D. Yu. Construction of business default scenarios based on system dynamics models // Journal of Applied Informatics. 2017. №05. P. 11-28.

Kurennoy D. S., Golembiovskiy D. Yu. System dynamics credit risk model of an oil company // Issues of Risk Analysis. 2017. №01. P. 6-22.

Tot'myanina K. M. Review of models of default probability // Financial risk management journal. – 2011. - №01 (25). – P. 12-24.

Ukazanie Banka Rossii ot 07.12.2015 N 3883-U. URL:

Ukazanie Banka Rossii ot 15.04.2015 N 3624-U. URL:

Сisar I. F. MATLAB Simulink. Computer modeling of economy. – M.: SOLON-PRESS, 2008. – 256 pp.


  • There are currently no refbacks.

Abava  Кибербезопасность MoNeTec 2024

ISSN: 2307-8162