Predicting stress in the banking sector: Malaysian evidence

This paper attempts to identify predictors of bank stress in Malaysia by computing a bank stress index and analyzing the index against bank specific variables and macroeconomic variables. This study utilizes the cointegrating VAR model to identify the existence of cointegrating relationships among both bank specific variables and macroeconomic variables towards bank stress. ARDL is also used to add robustness to the cointegration analysis. A forecast of the variables’ dynamic relationship is quantified by applying both orthogonalized and generalized variance decomposition (VDC). The impact of a shock in macroeconomic variables is also portrayed via an impulse response function and a systemic shock is simulated based on persistence profile. It is found that based on the index, bank stress levels in Malaysia have increased over time within the recent 15 years period. In the short term, non-performing loans to total loans (NPL) and lending spread are found to be the main predictors of bank stress. However, in the longer term, NPL and bank capital and reserve variation are the main predictors of bank stress. Bank stress has been found to have a strong cointegrating relationship with all the macroeconomic variables used in this study, and when the macroeconomic variables were shocked individually, it was found that there is an increased volatility in bank stress. Based on persistence profile, external shocks are found to have a greater and longer effect on the economy compared to internal shocks such as a disruption in the financial market or bank failure