Shrinking Large-Scale Population-Projection Models by Aggregation and Decomposition

During the past two decades social scientists have come to model dynamic socioeconomic systems of growing size and complexity. Despite a heavy reliance on ever more sophisticated high-speed digital computers, however, computer capacity for handling such systems has not kept pace with the growing demands for more detailed information. Consequently, it is becoming ever more important to identify those aspects of a system which permit one to deal with parts of it independently from the rest or to treat relationships among particular subsystems as though they were independent of the relationships within those subsystems. These questions are, respectively, those of decomposition and aggregation, and their application toward ‘shrinking’ large-scale population projection models is the focus of this paper.