EventsThe 2nd International Online Conference on Risk and Financial Management
Published
This submission belongs to the session S2. Machine Learning for Next-Generation Risk Modeling of the event The 2nd International Online Conference on Risk and Financial Management
Published date
28 Oct, 2026
Academic Editor
author-avatarThanasis Stengos
Citation
Wil Martens, Shin-siong Pang, Margin Compression Without Fee Substitution: Evidence from Euro Area Negative Rate Policy, in Proceedings of The 2nd International Online Conference on Risk and Financial Management, 2 November–4 November 2026, MDPI: Basel, Switzerland
Share
Email
Facebook
Twitter
LinkedIn

Margin Compression Without Fee Substitution: Evidence from Euro Area Negative Rate Policy

Shin-siong Pang 2
1. College of Management, National Sun Yat-sen University, Kaohsiung City, Taiwan
2. Department of Business Management, National Sun Yat-sen University, Kaohsiung City, Taiwan
Abstract

The prolonged period of negative interest rates in the Euro Area raises a fundamental question about whether unconventional monetary policy reshapes bank business models beyond traditional lending channels. Using a difference-in-differences design comparing Euro Area banks with non-Euro Area European banks over 2010–2023, this paper examines whether ECB negative rate policy (NIRP) induced a structural shift in bank revenue composition. The central finding is that NIRP significantly compressed net interest margins among Euro Area banks relative to the control group (β = -0.22 pp, p = 0.018), consistent with the deposit franchise mechanism. By contrast, there is no significant aggregate shift toward fee-based income, and balance-sheet retraction, rather than income substitution, appears to be the dominant adjustment channel. A supplementary causal-forest analysis shows that baseline operational efficiency and liquidity positioning, rather than bank charter type, best predict which banks adjusted most under NIRP, complicating prior heterogeneity findings based on institutional categories. This result survives covariate-overlap trimming, though propensity diagnostics indicate the exercise is constrained by near-deterministic treatment assignment given institutional covariates, a limitation we argue is itself informative for applying causal machine learning to macro-institutional treatments in banking. The findings suggest NIRP reshaped banking primarily through margin compression and efficiency-conditioned adjustment rather than uniform fee substitution, with implications for how unconventional monetary policy is evaluated and how bank heterogeneity is measured.

Keywords
Negative interest rate
bank business model
net interest margin
fee income
difference-in-differences
Euro Area
Selective Forecasting of Realized Volatility Using Prediction Confidence
Detection of Market Stress Through Adaptive Liquidity-Scaling Signals