MODELLING BANK DISTRESS EVENTS: THE EFFECT OF NETWORK CONNECTEDNESS

• Main Author: Aini, Mutiara
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/36847
• Institution: Institut Teknologi Bandung

Predicting bank distress is important since it is strongly related to financial stability. This study introduces the early-warning model that incorporates banking network connection to predict bank distress events for Indonesian banks. We employ variance decomposition to estimate the weighted connection among banks and use it as the network variable, both directional networks and net connections, in the model. Logit pooled regression is used to see the relation between distress events probability with network, bank-level, and macroeconomic variables. The usefulness of the model is calculated by taking into account the policymaker’s preferences of signaling a false alarm or missing a stress event. The aim of this study is to show that by incorporating networks the performance of the model is increased. The results show that asset, liquidity, and management quality are significant to the distress probabilities and by incorporating network variables, the performance of the model does increase. There is an important note that increasing the connection, directionally, is good as it is related to a lower distress probability. But the connection to others must be larger than from others so that the relation of the network on distress probability is still negative.