Banking Products and Retail Funding: The Effect on Non-Maturing Deposits

Abstract Non-maturing deposits are a growing and critical funding source for banks, yet their uncertain cash flows require granular methods to assess depositor behavior and manage liquidity and interest rate risks. Existing approaches rely on aggregate time series, treating all depositors as homogeneous, a key limitation this work addresses. Using a proprietary dataset of 845,000 retail customers and over 6.5 million annual observations from a large Italian bank (2013–2022), we employ fixed-effects panel regressions, quantile regressions (10th, 50th, 90th percentiles), and a Difference-in-Differences framework with staggered adoption. Correlational results show banking products increase deposit volumes by 26–94%, while credit products reduce them by 30–59%. Causal DiD estimates are systematically larger, consistent with negative selection attenuation in fixed-effects models. Of six products examined, only three generate statistically significant treatment effects: salary accreditation (+ 114%), payment cards (+ 34%), and credit lines (− 54%). Personal loans, utility debits, and financial advisory services fail the parallel trends requirement, suggesting their correlational associations reflect customer selection rather than genuine product effects. These findings establish product composition as a first-order determinant of NMD stability, with direct implications for Asset-Liability Management behavioral models and Basel III liquidity framework runoff rate calibration.