Analyzing the determinants of corporate_household delinquency rates using deep learning

The Korean financial industry has recently faced the issue of rising loan delinquency rates amid rapid increases in loan volumes. This study analyzes the main factors affecting loan delinquency rates through regression analysis and a deep learning LSTM model. Using data from the Bank of Korea, we identified significant variables through regression analysis and enhanced prediction accuracy with the LSTM model by learning the patterns in time series data.<br/> The results show that higher interest rates increase delinquency rates, while an increase in loan balances decreases delinquency rates. This implies that financial institutions can more accurately predict loan delinquency rates and manage risks effectively. The LSTM model&quots application confirmed higher predictive power than traditional statistical approaches. Consumer price index, consumer composite sentiment index, bank loan balances, and household loan interest rates were key factors for household delinquency rates, while producer price index, bank loan balances, and corporate loan interest rates were significant for corporate delinquency rates. These findings are valuable for formulating financial policies and managing risks.