Deposit beta / Pass-through rate
Definition
The proportion of a market rate change that a bank passes on to depositors on managed rate products. A deposit beta of 0.5 means a 100bp rise in the policy rate results in a 50bp increase in the deposit rate paid to customers.
What This Actually Means
If the Bank of England raises rates by 100bps and your savings rate only goes up by 40bps, your deposit beta is 0.4. The remaining 60bps widens your net interest margin — for now. The beta is never fixed: it reflects a combination of management pricing decisions, competitive dynamics, customer rate sensitivity, and the stage of the rate cycle.
Deposit betas tend to be lower at the start of a hiking cycle (customers are slow to notice or switch) and rise over time as competition intensifies and rate-aware customers begin to migrate. This path-dependency means a single static beta assumption can be wrong in different directions at different points in the cycle.
Where It Matters
The deposit beta is simultaneously an input to NII models, EVE repricing schedules, structural hedge sizing, and FTP frameworks. Because it feeds multiple systems that are often owned by different teams, the same beta may be applied inconsistently across models — or different betas used for different purposes without explicit reconciliation.
Beta vs. beta risk: the beta itself is the assumed pass-through rate used in models. Beta risk (see above) is the risk that this assumption turns out to be wrong. Both need to be owned, documented, and stress-tested — but they often aren't, because the beta is treated as a model input rather than a risk assumption in its own right.
Relationship to structural hedge: the deposit beta implicitly determines how much of the deposit base is treated as stable and long-duration for hedging purposes. A low beta implies deposits are sticky and insensitive to rates — supporting a larger, longer hedge. If the beta rises unexpectedly, the hedge is oversized and the bank is over-hedged relative to its actual liability profile. This is one of the most direct channels through which a modelling assumption becomes a P&L outcome.
Pricing strategy as the source: the deposit beta should in principle be derived from the business's pricing strategy — a forward-looking view of how management intends to price deposits across a range of rate environments. In practice, pricing teams typically think one move at a time: how much to pass through this hike, or this cut. Translating that into a coherent beta across a broad range of rate scenarios — up 300bps, down 100bps, a steepener, a flattener — is genuinely difficult and often not done rigorously. The result is a beta that reflects past decisions more than a considered forward pricing strategy.
Convexity in the beta: deposit betas are not constant across rate levels. They tend to be low in low rate environments — there is little margin to give away and customers have nowhere better to go — and expand as rates rise and competitive pressure builds. This convexity means a single beta applied across all scenarios will be wrong in a predictable direction: understating pass-through in high-rate scenarios and potentially overstating it at the lower bound. Models that use a flat beta across all rate scenarios are embedding a known structural error.
The historical data problem: historical betas are a tempting model input because they are observable and auditable. But they are a poor guide to future betas precisely because market conditions, competitive dynamics, and product pricing strategies change materially at different rate levels. The 2010-2021 low-rate period produced very low observed betas that told you almost nothing about behaviour in a 4-5% rate environment. A forward-looking view of the organisation's pricing strategy — informed by but not mechanically derived from history — is the more appropriate input. That said, backtesting the beta assumption against actual pricing outcomes remains essential: it is the primary mechanism for detecting model drift and holding the pricing strategy assumption to account.