Banks risk managing interest rates through assumptions and governance processes that move more slowly than markets. An ALM leader argues that structural hedging, behavioral deposit models, scenario analysis and decision-making authority must evolve to reflect faster customer behavior and increasingly volatile yield curves.
Banks need to give treasury teams the authority and tools to respond proactively to interest rate movements rather than waiting for governance processes to catch up with markets.
A group head of asset liability management at a digital bank warned that balance sheets cannot be managed dynamically if decision-making cycles move more slowly than the risks they are supposed to control.
Speaking at CeFPro’s Risk Evolve 2026 event, the ALM head challenged the assumption that banks are operating in a straightforward post-rate-hiking environment.
Instead, movements in the yield curve have exposed the fragility of strategies built around an expected return to normalization.
“That is not normalization. That is a repricing and uncertainty,” she said.
The consequences extend beyond the direction of rates. The ALM head argued that banks must consider how different parts of the balance sheet reprice at different speeds and how interest rate, credit spread, liquidity and behavioral risks interact.
Structural hedging is central to that challenge. It can provide earnings stability when rates fall, while creating negative carry as rates rise. But decisions about hedge coverage and duration should not be treated simply as technical treasury judgments.
“Is it a governance decision that’s masquerading as a technical one?” she asked.
Extending hedge duration to stabilize net interest income can simultaneously increase economic value sensitivity, meaning decisions taken within treasury can ultimately have capital consequences.
The ALM head argued that institutions reducing hedge coverage to avoid negative carry may find themselves structurally underhedged when reinvestment yields subsequently become most attractive.
The deeper problem is that nobody can reliably forecast rates. Market expectations have repeatedly been too early or too late even when the eventual direction was broadly correct.
“The practical consequence of any structural hedge calibrated to a single rate path is not a hedge, it’s a leverage expression of the forecaster’s view,” she said.
Customer behavior is making traditional assumptions equally vulnerable. Non-maturing deposit models remain crucial because their behavioral assumptions determine hedge notionals, which in turn influence hedge income.
Yet the 2022 to 2025 rate cycle produced evidence that conflicts with assumptions embedded in many older models.
Deposit beta behavior proved nonlinear, while digital banking allowed customers to move money considerably faster than models based on branch-era behavior anticipated.
“Customers change,” the ALM head said. “We can’t use the same assumptions we used before because of the internet, because of the new capabilities and new facilities and how the market changes very quickly.”
Her institution has developed a machine learning model intended to move beyond an excessive reliance on deposit beta assumptions and provide a better understanding of actual customer behavior.
For the ALM head, this changes the nature of balance sheet modeling itself.
“For me, ALM and non-maturing deposit turns from mathematical assumptions into psychology,” she said.
Stress testing must evolve alongside those behavioral changes. Standard parallel interest rate shocks remain necessary for regulatory purposes, but she argued they are insufficient for contemporary balance sheet management because they can miss sequencing risk.
Markets can move rapidly from expectations of rate cuts toward higher-for-longer pricing without an equivalent movement in overnight rates.
Different types of curve movement can also be driven by very different forces, including geopolitical shocks transmitted through energy prices and inflation expectations.