Simulating retaliation in payment systems: Can banks control their exposure to a failing participant?

This paper assesses the impact of an operational failure at one of the biggest participants in the Dutch interbank payment system, varying the time at which the disruption takes place. Liquidity levels equal historical levels. The impact of such a disruption is quantified in terms of the additional liquidity needed in order to settle all payments than can settle given the banks' intraday reserves and collateral facilities. Assuming the disruption lasts for the remainder of the day, banks are faced with costs, as they need to borrow this additional liquidity overnight from the market or from the central bank. As could be expected, the second-round effect (the number of unsettled payments among healthy banks) of an operational disruption is highest when it occurs early during the day and lasts for the remainder of the day. So are the additional overnight liquidity needed and the costs of overnight credit. Furthermore, the paper introduces different possible reaction patterns from the stricken bank's counterparties. These counterparties can react according to two basic rules: they stop sending payments to the stricken bank either after some pre-determined time or after their exposure to the stricken bank reaches a certain level. From a cost perspective, reacting is more effective when determined by the individual exposure of the stricken banks' counterparties. However, even an immediate reaction does not prevent banks from running losses following the failure of a major participant. This leads to a reflection about the bilateral relations between the stricken bank, considered to be a node in a partial star network, and the other banks. How much each payment system participant can control its exposure to the stricken bank depends on the degree of reciprocity in the value of bilateral payments.