Run-off balance sheet
Regulatory Definition
A balance sheet where existing non-trading book positions amortise and are not replaced by any new business.
EBA GL/2022/14
What This Actually Means
Assume your bank stops writing new business today. Every loan repays on schedule, every deposit rolls off at maturity, and nothing is replaced. The balance sheet shrinks to zero over time. This is the assumption used for EVE calculation under the SOT.
Where It Matters
The fundamental tension: EVE uses run-off as a deliberate theoretical construct — it measures the embedded risk in today's positions, not your forward business plan. But banks manage interest rate risk on a going concern basis: deposits are retained or replaced, the balance sheet evolves, and hedging strategy is built around earnings stability over time. Run-off assumes the opposite. Optimising risk management around a run-off metric can therefore generate behaviours that conflict with business strategy — hedges that make sense under run-off may be wrong for a going concern, and EVE-calibrated risk appetite limits may constrain activities that are entirely rational from an NII perspective. Many banks end up managing to two partially conflicting objectives simultaneously.
The deposit run-off problem: run-off still requires behavioural assumptions for NMDs, which have no contractual maturity. There is no reliable way of predicting how deposits would actually behave in a genuine wind-down — customer behaviour in a rundown is fundamentally different from steady-state, and historical data provides limited guidance. Any deposit run-off profile is therefore a modelling assumption with significant uncertainty attached.
The regulatory EVE bias: the SOT leaves little room for gap risk between deposit run-off assumptions and the structural hedge programme. If the two diverge materially, the resulting gap creates EVE sensitivity that can breach SOT thresholds. This creates a strong incentive — whether explicit or not — to align run-off assumptions to the existing hedge profile rather than deriving them independently. And the uncomfortable truth is that any independently-derived customer behaviour view would itself be unreliable — it is necessarily based on historical data that reflects past rate environments and competitive dynamics, not what will actually happen in the future scenario being modelled.
The conflation this produces is significant: a structural hedge profile is designed to stabilise NII on a going concern basis — it is an earnings management tool, not a prediction of deposit rundown. Using it as a proxy for run-off produces EVE metrics that are internally consistent with the hedge book but may not represent the true economic exposure in the deposit base. Supervisors are increasingly alert to this distinction, and the ability to demonstrate that run-off assumptions are derived independently of the hedge programme is becoming a meaningful point of scrutiny in SREP.