This repository includes the code used to perform stress testing based on economic scenarios using loan portfolio. The application was originally done on MATLAB and we extended it to the open-source Python programming language.
The Python programming language, version 3.7.13, and its statistical libraries were used to build all applied models. The Python script stress_test.py in the python folder contains the script necessary to reproduce the results.
All operations, from data preprocessing to model estimation, were carried out on the Google Colaboratory platform.
Bank stress testing is a framework for analyzing the impact of simulated extreme, but plausible, economic and financial conditions to ensure that banks have sufficient capital to maintain operations during such situations. Banking stress tests were introduced by central banks and authorities in charge of banking supervision in the late 1990s. From that time, the more frequent banking and financial crises, and in particular the Asian crisis of 1997, had highlighted the role of the deterioration of macroeconomic factors in the triggering of banking crises which were not sufficiently taken into account in other methods of banking regulation and supervision.
The applied generalized linear model takes into account, in addition to loan information, economic data in order to predict the probability of default. An adverse economic setting is then defined in order to assess and analyze the rates of default under such conditions.
The data folder contains the portfolio of outstanding, and historical, loans as well as economic and financial time series under normal and stressed conditions.
The generalized linear model in this application was built and estimated using the statsmodels library.