What is it about?
Millions of Filipino students sit in classrooms far beyond capacity. One government fix is to pay tuition so students attend nearby private schools instead. We combined school location, enrollment, capacity, tuition, and road data from nearly 8,000 schools to model where students actually go and why. Distance matters about four times more than cost, and the real limit is not the size of the subsidy but the number of private school seats available. Raising subsidies alone will not fix overcrowding.
Featured Image
Photo by Wander Fleur on Unsplash
Why is it important?
Education subsidy programs are widely used to relieve overcrowded schools, and policy debate usually centers on whether the subsidy is large enough. Our results point elsewhere. Increasing the subsidy twentyfold raises enrollment by less than two percent, because more than five students compete for every available private school seat. Two findings matter most for policy: geographic proximity constrains school choice roughly four times more strongly than tuition cost, and over half of the untapped capacity lies in routes between schools that no student has yet taken. The method generalizes to any setting where individual choice must be reconciled with fixed physical capacity, including healthcare access and public transit.
Perspectives
Most of the effort behind this paper went into something that will not win any awards: reconciling school records that had been kept in separate systems, by separate offices, for years. It is tedious work, and it is invisible in the final model. But there was a moment when the joined dataset first rendered as a national flow map and we could see, for the first time, where students in three regions actually go. I would encourage anyone working with government data to take that infrastructure work seriously. The analysis is often the easy part.
Sebastian Ibañez
Department of Education
Read the Original
This page is a summary of: Student Flow Modeling for School Decongestion via Stochastic Gravity Estimation and Constrained Spatial Allocation, August 2026, ACM (Association for Computing Machinery),
DOI: 10.1145/3770855.3819058.
You can read the full text:
Resources
Contributors
The following have contributed to this page
