Liquidity management of derivatives linked to interest rates

Liquidity management for portfolios with interest-rate-linked derivatives typically relies on scenario-based projections of expected cash flows. We show that plain expectations under either the real-world measure P or a risk-neutral measure Q are fragile: they depend on the choice of numéraire and may yield inconsistencies when cash-flow timing is stochastic or products are multi-callable. We apply a measure-consistent alternative proposed by Fries (2025): compute expected cash flows as discounting sensitivities with respect to zero-coupon bonds. This viewpoint reconciles cash flow forecasting with replication, and it naturally handles stochastic payment times. In the context of life insurance ALM, we adopt a separation of market and actuarial drivers and discuss when actuarial averaging is a sound approximation for liquidity planning. Finally, building on the replicating-portfolio literature, we compare cash-flow matching versus terminal-value matching objectives and verify theoretically predicted bounds between them (Natolski and Werner 2018) by Monte Carlo experiments in a LIBOR Market Model setting using bonds, caplets, and swaptions. The results support discounting sensitivities as a robust basis for liquidity management while clarifying when classical expected cash flow projections can mislead.