Field NoteJune 30, 2026 (Tue) · WPSC Congress Day 2

Shrinking Cities, Growing Cities — A Living-Area Optimization Model for Balancing Efficiency and Equity, Presented at WPSC 2026 Helsinki

On Day 2 of the World Planning Schools Congress in Espoo, Finland, an oral presentation shared a linear-programming-based planning support module for deciding where to consolidate and where to add amenities when shrinking and growing districts coexist within a single city.

On June 30, 2026, Day 2 of the 6th World Planning Schools Congress (WPSC 2026) was held at Aalto University's Otaniemi campus in Espoo, Finland. In the Track 1 (Accessibility & Mobility) session, I gave an oral presentation on research refined over the past several years together with the Seoul Institute, Hanyang University, and Gachon University. The title was "Balancing Efficiency and Equity in Walkable Living Areas — Optimised Amenity Placement under Demographic Change." It is about a planning support module that uses linear programming (LP) to answer, when shrinking and growing cities coexist within a single city, the question of where to cut amenities and where to add them.

Start of the presentation — the title slide, framing the challenge of balancing efficiency and equity in walkable living areas amid demographic change.

Why This Problem, Why Now — An Asymmetric Crisis

Korea is undergoing the fastest demographic transition in the OECD. Total population turned to decline in 2020, and by 2035 the share of residents aged 65 and over is projected to exceed 30%. The problem is that this decline is not uniform within cities. Old town-center living areas are aging and shrinking, while new districts on the urban fringe are growing rapidly. Planners face two opposite questions at the same time: "Where should facilities be consolidated to save budget?" and "Where should facilities be added to make walkable access more equitable?"

Introduction — Korea's rapid demographic shift and the legal institutionalization of living-area planning.

The introduction laid out Korea's rapid demographic change. With the 2024 revision of the National Land Planning Act, living-area planning became a statutory plan, bringing this question formally into the planning system.

Why Busan Was Chosen as the Empirical Testbed

Busan contains the full spectrum of demographic trajectories within a single city. The 770 km² metropolitan area, home to 3.3 million people and divided into 6 mid-tier living areas, was broken down into 19,469 250-meter grid cells, and 3,979 facilities across 16 amenity types were analyzed. Gangseo and Gijang are growing; Dongnae and Haeundae are stable; Gangdong (Buk, Sasang, Saha) and the old town center are shrinking. Because growth and shrinkage coexist within the same city, Busan was an ideal stage for testing opposite prescriptions with a single module.

At the podium — explaining the growth, stability, and shrinkage trajectories of Busan's 6 mid-tier living areas.

A Four-Layer Planning Support Module

The core idea is that four layers share a single pipeline.

Layer 1 — Multimodal Walking and Transit Network: Building travel-time matrices between 250-meter grid cells using OSRM (walking), R5R+GTFS (transit), and the T-Map API (observed trips)
Layer 2 — 2SFCA Accessibility Diagnosis: Weighting facility capacity (supply) and demand by distance to compute accessibility indicators that capture even facilities that are "close but overcrowded"
Layer 3 — Living-Area Delineation via Louvain Community Detection: Considering both the travel network and the accessibility network together, deriving stable living-area boundaries through 100 iterations
Layer 4 — Linear Programming Optimization: Two opposing models

Explaining the methodology — from accessibility diagnosis (2SFCA) through living-area delineation (Louvain) to optimization (LP), all layers run on a single reproducible open-source pipeline.

Two LP Models — Efficiency and Equity

Model A (Efficiency) vs. Model B (Equity)

Category	Model A — Facility Consolidation (Efficiency)	Model B — Accessibility Maximization (Equity)
Target areas	Shrinking areas	Growing areas
Objective	Reduce operating costs by closing underused facilities	Improve accessibility by adding new facilities in underserved grid cells
Equity mechanism	Constraint minimizing travel-cost increase	Weight w = 1/(current accessibility) — weighting toward underserved areas

The two models are selected automatically based on each area's demographic trajectory tag. Same pipeline, opposite demographic phases: consolidate where the population is shrinking, expand where it is growing — within the same year, the same plan.

Results — Viewed Citywide

Gangdong (shrinking): 3 facilities consolidated, operating costs down −24.1%, with weighted travel cost rising only +6.8%
Gangseo/Gijang (growing): 6 new facilities added (a lifelong learning center, an integrated care center, a library, a senior center, a childcare center, and a "Deul-rak-nal-rak" children's facility), accessibility for the bottom 10% of grid cells up +56%
Busan citywide: net facility change of −1, operating costs −5.6%, bottom-10% accessibility +27%, Gini coefficient −0.076

The key finding is that efficiency (cost savings) and equity (improved accessibility for the bottom tier) improved together rather than trading off. Model B showed a Pareto improvement, raising the floor without cutting into higher-performing areas.

View of the Track 1 session — a series of presentations on accessibility and mobility, followed by lively Q&A with the audience.

Validation on the International Stage

In the discussion following the talk, attention centered on the value of the audit trail of the LP-based optimization — the documented rationale for which facilities are opened or closed — for public deliberation and community participation processes. In fact, in the Gangdong pilot, 2 of the 3 closures proposed by the LP model matched the outcomes of expert review under the 2025 living-area plan; the one discrepancy was arguably meaningful in itself, revealing a "documented rationale" in place of what had been an "implicit preference."

Presentations by fellow researchers in the same session — sharing perspectives with international comparative studies addressing accessibility and equity from angles such as the "three urban fabrics" and typologies of walkable streets.

Seoul, and the Next Steps

This module can be transplanted to any Korean city preparing a living-area plan. Next up is Seoul — a pilot along the Eunpyeong–Seodaemun axis in 2027. Follow-up work includes alternative equity weights based on income and elderly population share, multi-period LP for 10-year investment planning, and stochastic LP robust to population forecast error. The full codebase is planned for release under the MIT license.

Closing the presentation — when shrinking and growing cities exist within the same city, planning must be able to answer, under one consistent principle, what to reduce and what to add.

What this experience reaffirmed is that efficiency and equity are not an either-or choice but something to be designed together. Data and optimization are merely tools that make that choice transparent — the question of what kind of city to leave behind still belongs to planning itself.

Presenter: Joonyoung Choi (Seoul Institute) · Co-authors: Sunyong Eom (Hanyang University), Jiho Yeo (Gachon University), Yuri Lee (Seoul Institute) / WPSC 2026 Helsinki · Track 1 (Accessibility & Mobility) · June 30, 2026

Originally published as a field note on the Naver blog (novacite) and rewritten here in English. Photos by the author on-site unless noted. © Dr. Junyoung Choi · ← Back to the WPSC 2026 Article Series