Liquidity demands in real-time gross settlement payment systems can be enormous. To reduce the liquidity requirement, central banks around the world have implemented liquidity savings mechanisms (LSMs). The most effective LSMs are those that economize on liquidity needs by matching offsetting payments that have been submitted to a central queue and settling these payments using only the liquidity needed to cover the net obligations. Maximizing the value of payments settled in a queue given available liquidity is computationally difficult. Existing centralized queuing systems do not always meet this objective. Even when they do, the resulting outcome does not necessarily maximize system welfare. This paper seeks to improve upon existing centralized netting queues by making two fundamental changes. First, instead of making decisions on how much liquidity to provide to the queue before netting arrangements are determined, banks receive take-it-or-leave-it offers that determine which of their payments will be settled as well as their share of the liquidity cost. Second, rather than attempting to maximize the value or volume of payments settled in the queue, I propose using information regarding the instantaneous benefits and costs of participants to define a welfare measure for any set of netted payments. The full benefits of these two changes are realized through an application of the Shapley value cost allocation method, which ensures welfare maximizing netting proposals are always accepted.