Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
The problem of recovering the entries of a large matrix of expenditure, trade, or income flows from limited-incomplete multisectoral economic data is considered. Making use of some consistency and adding up restrictions, the problem is cast as a pure inverse problem and specified within a nonlinear optimization framework. Estimates of the unknown entries are provided along with an overall measure of uncertainty for the complete matrix and a measure of uncertainty for the individual elements. Artificial and real data are used to illustrate how the procedures may be applied and interpreted and to gauge performance under entropy and squared error measures. Copyright 1994 by MIT Press.