# Model risk for non-maturing products

Most banks have significant amounts of non-maturing products on their balance sheets, such as current accounts and savings deposits. The contractual maturity is very small, so that the client could withdraw his or her money at short notice and banks can adjust interest rates at any time.

However, most banks assume a longer behavioral maturity determined by statistical models. Besides volume changes, the interest rate adjustment frequency is mostly considered and can be assumed to be relatively low. This approach is international and experience supports the procedure in general.

From a banking book perspective this approach means that a bank that finances medium-term assets using non-maturing products would have (nearly) eliminated any interest rate risk because the model ‘says’ that maturities match. From a liquidity perspective, everyone agrees that sufficient reserves should be held due to the risk of short-term withdrawals.

However, the interest rate risk perspective does not consider what would happen if the model assumptions are not valid. For example, if the modeled maturity and interest rate adjustments are actually shorter than computed, creating an interest rate gap between interest-bearing assets and liabilities. In this scenario, interest rate risk would practically no longer be eliminated in the banking book, resulting in a significant lower or even negative interest result.

This type of model risk needs to be assessed and, as well as holding sufficient liquidity buffers, it also needs to be covered by capital. Although bank regulations do not consider this requirement explicitly, it does not mean that it is not necessary. It would be time to at least understand the impact of such a potential risk model by trying to identify the impact on capital and costs.

One approach could be using scenario analysis and stress testing by applying different duration assumptions for non-maturing products in the context of measuring the business risk. Finally, it would become obvious how fragile current absolute (interest rate) risk figures are as banks usually assume cash flows of non-maturing products as quasi deterministic. There is no doubt that playing with the duration, particularly if long-term average maturities are assumed, the resulting Value at Risk would breach the risk limits.