This paper has two main parts. The first deals with sourcing assumptions underlying multiregional models including GTAP and TERM (The Enormous Regional Model). By assuming that all users of a good in a particular region source that good from all regions in common proportions, the dimensions of a multi-regional CGE database are reduced many-fold. This greatly reduces the solution time in simulations. The paper details the derivation of formulae on which the common sourcing assumption is based and the necessary additional identities required to ensure that the assumption holds. While the IMPLAN team has provided very detailed small region input-output tables for the US economy for many years, multi-regional modelling of the US economy is less common. USAGE-R51 is in the TERM school of multi-regional, sub-national models. Rather than be discouraged by the absence of interstate trade data, USAGE-R51 follows the TERM traditional of estimating such trade matrices so as to satisfy state level excess demands and distribute excess supplies. Exceptionally detailed regional employment numbers for the United States are available from census data. The master database of USAGE-R51 includes 497 sectors, including a split of electricity generation into different fuel types. It also includes 51 regions. Following the GTAP approach, the USAGE-R51 database is tailored for each application through appropriate aggregation.