For decades, banks have viewed non-maturity deposits—the standard checking and savings accounts that form the bedrock of retail funding—as a stable, structural component of their balance sheets. These funds are generally treated as “sticky,” staying put even as market conditions shift. However, recent volatility in global markets has revealed a more precarious reality: deposit stability is not a fixed characteristic of a bank’s ledger, but a fluid outcome of market behavior.
A new behavioral modeling framework is challenging the traditional approach to non-maturity deposit risk under interest rate stress, suggesting that the probability of customer withdrawals is far more sensitive to real-time financial market shifts than previously modeled. By aligning behavioral estimates with the rigorous stress-testing scenarios mandated by the Basel Committee on Banking Supervision, the framework provides a more accurate lens for understanding how funding stability erodes when interest rates spike.
The core of the issue lies in Interest Rate Risk in the Banking Book (IRRBB), the risk that a bank’s earnings or capital will decline due to movements in interest rates. While regulators require banks to perform stress tests, many internal models suffer from a misalignment: they often hold macroeconomic variables constant while shocking interest rates. This creates a blind spot, as depositors do not react to rates in a vacuum; they react to the broader financial environment.
Closing the Gap in Regulatory Stress Testing
The proposed framework addresses this misalignment by utilizing a multivariate logistic regression model to estimate the probability of withdrawal events. Unlike traditional models that rely on lagging economic indicators, this approach focuses exclusively on “shockable” financial market variables. These are variables that move in direct synchronization with the supervisory interest rate shock scenarios prescribed by global regulators.
By relying on these market-consistent variables, banks can ensure that their behavioral assumptions about depositors are internally consistent with their regulatory stress tests. This prevents the dangerous assumption that deposits will remain stable while the cost of funding elsewhere is skyrocketing.
The shift in perspective is summarized in the following comparison of how banks have traditionally viewed these deposits versus the findings of the behavioral framework:
| Perspective | Traditional Structural View | Behavioral Modeling View |
|---|---|---|
| Deposit Nature | Stable, long-term funding | Market-dependent and volatile |
| Risk Driver | Customer loyalty/inertia | Short-term interest rate differentials |
| Stress Testing | Static macroeconomic assumptions | Dynamic, “shockable” market variables |
| Risk Category | Liquidity risk (separate) | Integrated Interest Rate and Liquidity risk |
The Mechanics of Deposit Flight
The empirical data suggests a clear hierarchy of what drives customers to move their money. Short-term interest rates emerge as the primary driver of withdrawal risk. When the gap between the interest a bank pays on a non-maturity deposit and the rate available in the wider market widens, the probability of a withdrawal event increases significantly.
However, the model also identifies “amplifiers.” While short-term rates trigger the initial urge to move funds, the slope of the yield curve—the difference between short-term and long-term interest rates—acts as a catalyst. A steepening yield curve can accelerate depositor responses, making the flight of capital more aggressive than short-term rates alone would suggest.
This indicates that bank funding stability is not merely about the absolute level of rates, but the shape of the entire interest rate environment. For risk managers, this means that monitoring the Federal Open Market Committee (FOMC) and other central bank signals regarding the yield curve is as critical as monitoring the overnight rate.
Regime Shifts and the “Covid Effect”
The framework further highlights that deposit behavior is regime-dependent, meaning the “rules” of how people move money change during crises. The analysis identified significant seasonal patterns in withdrawals, but more importantly, it noted a distinct shift during the Covid-19 pandemic.
During the pandemic, abnormal liquidity patterns emerged as government stimulus and central bank interventions altered the traditional relationship between interest rates and deposit stability. This suggests that behavioral models cannot be “set and forget”; they must be capable of adapting to new economic regimes where historical correlations may temporarily break down.
This finding underscores the argument that deposit stability is a market-dependent outcome. When the regime shifts—whether due to a global health crisis or a sudden pivot in monetary policy—the perceived stability of a bank’s funding base can evaporate quickly, transforming a structural asset into a liquidity liability.
Integrating Liquidity and Interest Rate Risk
the framework argues for an integrated view of risk. For too long, banks have treated interest rate risk (IRRBB) and liquidity risk as two different silos. One focuses on the value of the balance sheet, the other on the ability to meet cash obligations.
The reality is that they are two sides of the same coin. An interest rate shock creates a behavioral response (withdrawals), which in turn creates a liquidity crisis. By using a behavioral modeling framework that links these two, banks can more accurately assess their funding stability under extreme scenarios, allowing for more precise hedging and capital allocation.
Disclaimer: This article is provided for informational purposes only and does not constitute financial, investment, or regulatory advice.
The next critical checkpoint for these standards will be the ongoing reviews of the Basel III framework, as regulators continue to refine how banks must account for liquidity and interest rate risks in an era of higher-for-longer rates.
Do you believe banks are underestimating the volatility of retail deposits in the current rate environment? Share your thoughts in the comments or share this analysis with your network.
