On Forming Linear Combinations of Accounting Data to Detect Constant Small Losses

Recently, there has been renewed attention focussed on the problem of using material accounting data from more than one accounting period to make inferences about the state of material control in a facility. This renewed activity was prompted in part by the so-called Rosenbaum report [1] that underscored the inability of usual MUF [2] calculated for a single accounting period to detect small losses that accumulate over time. This deficiency in the MUF statistic has been obvious to workers in the field of safeguards accountancy for many years and, for internal control purposes at least, facilities tend to put more emphasis on the accumulated MUF, rather than on any individual MUF, as a measure of the long term state of control. This emphasis on accumulated MUF does not mean that a single MUF is not a useful statistic; it is intended to detect a single large loss that occurs within the accounting period in question. Thus, both concepts are important, the one to detect continuing small losses and the other to detect single loss events.