Introduction

Central banks in Europe, Japan, and Switzerland have experimented with negative policy rates to stimulate economies when conventional stimulus is exhausted. Negative rates have counterintuitive effects: they may reduce bank lending by compressing net interest margins, increase saving as households seek yield, and distort asset pricing by forcing investors into risky assets. A data-driven analysis using machine learning can quantify these effects across multiple dimensions, informing policy debates and investment strategies.

Mechanisms of Negative Rate Transmission

Bank Profitability and Lending

Negative rates compress spreads between lending and deposit rates. If deposit rates cannot go negative due to cash alternatives, banks face margin compression. Machine learning models using bank-level data measure this: banks with higher deposit ratios face larger margin compression and reduce lending. Quantifying this mechanism informs debate on negative rate effectiveness.

Savers' Behavior and Asset Demand

Negative rates reduce returns on safe deposits, pushing savers toward riskier assets. Does this drive equities higher or bond duration risk? ML models analyzing household portfolio data identify how negative rates shift asset allocation. Results suggest savers shift toward equities and credit, consistent with central bank intentions but with stability risks.

Causal Inference Methodology

Identify treatment (introduction of negative rates) and control groups (countries without negative rates) to isolate effects. Use synthetic control methods and difference-in-differences regression with machine learning to estimate causal impacts. Results: negative rates modestly stimulate credit growth but compress bank profitability, with mixed effects on overall economic growth.

Conclusion

Data-driven analysis of negative rate episodes reveals nuanced transmission mechanisms and unintended consequences, informing ongoing debates on monetary policy effectiveness at the lower bound.