Centralized or decentralized control of school resources? A network model

Abstract

The typical school district in the US consists of a central office overseeing primary, middle and high schools. The school district budget is allocated between the central administration and the constituent schools, who can spend these funds on personnel and non-personnel. We model this allocation problem as a network data envelopment analysis problem which solves for the technically efficient allocation of the budget within the district. The goal is to identify the allocation which yields the best aggregate performance for each school district in our sample. In our examination of 70 school districts in the Dallas, Texas area we find that test scores could be increased by approximately five normal curve equivalent (NCE) points by campuses reducing technical inefficiency and by an additional four NCE points by optimally reallocating the school district budget. Our illustrative model suggests that school districts could increase achievement test scores if more of their budgets were spent on campus personnel like teachers and less on non-personnel items like supplies, and if personnel resources were reallocated from the secondary to the elementary level. Furthermore, while the average school district in our sample allocates 21 % of their budget to the central office, our network model indicates that if resources were optimally allocated, the average school district would allocate only 16 % of their budget to the central office.