A sandbox technique

How Large Are Default Spillovers in the U.S. Financial System?

An Upper Bound on Network Default Spillovers Our goal is to estimate expected default spillovers for the entire U.S. financial network, across all institutions and asset classes, with the fewest assumptions possible. To do so, we use a new way to deal with the problem of data deficiency proposed by Glasserman and Young. They show that for a very broad class of models, even though the actual expected spillover losses cannot be computed without knowledge of all bilateral exposures, an upper limit to these spillovers can be estimated using only node-specific information (that is, without a precise breakdown of the nodes' counterparties or the magnitudes of obligations to them). In particular, the upper limit is based on each node's probability of default, its total outside assets, and its ratio of inside liabilities to total liabilities. Thus, at the cost of estimating an upper bound on spillovers instead of their actual values, the data requirements are greatly reduced. 
The upper bound shows how much larger expected losses in the connected network are, when compared to expected losses in an analogous disconnected network with its interconnections removed (see part 1 of this series for details). Estimating the upper bound in this way enables us to quantify the effects of the network structure, without needing additional assumptions about the initial shocks to each node’s outside assets. 

This is really asking what the upper bound is on pit boss operations across a liquidity net.    It can be observed during adiabatic change, a node at a time should stress the network sustainability.  The probability that a node exceeds its basket size, one way or the other. Some nodes should be more susceptible if they tend to remain skewed.

The network is a collection in which should be the generator, embedded to a mostly sustainable channel.  A days work in the sandbox.