Aggregation and decomposition for multi-divisional linear programs

This paper proposes the use of linear programming aggregation theory to study decentralization in linear programming models. Traditionally mostly standard decomposition approaches are used when decentralization is discussed. The authors consider a two-level organization and interpret the aggregation/disaggregation scheme in terms of information that has to be transferred between the central planning department and the divisions in a divisionally organized enterprise. Relations between this procedure and the classical decomposition approaches are discussed. The ideas are illustrated by means of a simple example for which also an economic interpretation of the procedure is developed.