NMDs Logistic Regression Model

This paper proposes a novel approach to modeling the volumes of Non-Maturity Deposits (NMDs), a key component of interest rate and liquidity risk in the banking book. Using a multivariate logistic regression model with autocorrelated data, we estimate the probability of cash-out (PoC) from the stable portion of NMDs. Uniquely, our model relies solely on variables included in the six supervisory interest rate shock scenarios prescribed by the Basel Committee on Banking Supervision, ensuring consistency with regulatory stress testing frameworks and avoiding misalignment with macroeconomic scenario assumptions. Empirical results confirm that financial market conditions—particularly short-term interest rates and the yield curve slope—are significant drivers of withdrawal behavior, underlining the role of the interest rate channel in monetary policy transmission. Seasonal patterns and the impact of the COVID-19 crisis are also found to influence depositors' behavior. By isolating the contribution of supervisory variables, our approach supports both risk management practices and regulatory compliance, offering a practical and forwardlooking tool for banks in assessing core deposit stability under stressed conditions.