A Macroeconomic Stress Testing Framework for Credit Risk in Banking

Abstract

The past decade has witnessed high volatility in the global macroeconomic conditions and has shown how the banking system is exposed to a variety of risks which can assume systemic dimensions and can affect the financial stability of a country adversely. Given the importance of financial stability, organizations like International Monetary Fund (IMF), World Bank, Bank for International Settlements(BIS) and the Central banks across the globe are striving to evolve a robust Macro Prudential and Systemic risk assessment framework. Macro stress testing is an integral element of this exercise. In the Indian context also, Macro stress testing has evolved over the last few years; however, given the advancements in this area across the globe, the research is still in a very nascent stage. Against this backdrop, our study is an attempt to contribute to the ongoing research efforts in this area and provide a reference point for reassessing and reviewing the existing macro stress testing practices in the Indian context. We propose to modify the existing macro stress testing model for credit risk as developed by Reserve Bank of India in terms of endogenous variable selection, calibration of stress testing scenarios and modification of the macro stress testing model. In the dissertation, Top-down approach of stress testing has been adopted for the Indian banks using quarterly data pertaining to the time period 1996Q2 to 2016Q4. The macroeconomic variables employed for the study are GDP, CPI, Exchange rate, Oil, Market Capitalisation of NSE, Short-term interest rate and Long term interest rate. For the empirical analysis, Vector Error Correction Model (VECM) technique has been employed to investigate the dynamic impact of changes in the macroeconomic variables on the Default Ratio which has been taken as a Credit Risk indicator. Wald Test, Granger Causality and Toda Yamamoto test have been employed to investigate the short term relationship between the variables...