Replication data for: Spatial Enclaves or Regional Catalysts? A Spatial Econometric Typology of Special Economic Zones in Indonesia

1. Research Hypothesis We hypothesise that Indonesia's Special Economic Zones (SEZs) primarily function as isolated "spatial enclaves" or generate negative backwash effects, rather than acting as equitable "regional catalysts", due to top-down governance and weak local supply chain linkages. 2. Data Description & Collection This cross-sectional spatial dataset covers 514 Indonesian districts and 17 active SEZs in 2023. Official sources include BKPM (district investment), the National SEZ Council (SEZ data), and BPS (regional controls like GRDP and education). Spatial weights (Queen contiguity and KNN=7) were generated using BIG administrative boundaries. 3. Notable Findings Spatial Durbin Model (SDM) estimations and LISA clustering reveal a highly unequal investment distribution. The data categorises SEZs into distinct typologies: most fail to generate positive spillovers (operating as enclaves), some drain nearby capital (backwash), and rare exceptions like KEK Sorong act as true regional catalysts. 4. Interpretation Economic policy does not operate in an empty space. Without mandatory backward linkages, capital injected into SEZs remains trapped within zone boundaries or flows back to the economic core (Java), leaving peripheral host regions with minimal long-term prosperity. 5. How to Use & Replicate Designed for full reproducibility. Raw data includes Excel files and zipped shapefiles. Users simply run the master_R_script.R script in R Studio. This single script automates NA filtering, multicollinearity (VIF) diagnostics, LISA calculations, SDM estimations, and robustness checks. Log-transformations for zero-investment districts are handled automatically to output a final results workbook.1. 研究假设 我们提出如下研究假设：鉴于自上而下的治理模式与薄弱的本地供应链关联，印度尼西亚经济特区（Special Economic Zones, SEZs）主要以孤立的“空间飞地”模式运作，或产生负面的回流效应，而非成为公平的“区域增长引擎”。 2. 数据描述与采集 本横断面空间数据集覆盖2023年印度尼西亚的514个行政区与17个活跃经济特区。官方数据来源包括印度尼西亚投资协调委员会（Badan Koordinasi Penanaman Modal, BKPM，负责行政区投资数据）、国家经济特区委员会（National SEZ Council，负责经济特区数据）以及印度尼西亚中央统计局（Badan Pusat Statistik, BPS，负责地区国内生产总值（Gross Regional Domestic Product, GRDP）与教育等区域控制变量数据）。空间权重矩阵（皇后邻接权重（Queen contiguity）与7近邻权重（KNN=7））通过印度尼西亚国家地理空间信息局（Badan Informasi Geospasial, BIG）的行政边界数据生成。 3. 主要研究发现 空间杜宾模型（Spatial Durbin Model, SDM）估计与局部空间自相关（Local Indicators of Spatial Association, LISA）聚类分析结果显示，投资分布存在高度不均等的情况。本数据集将经济特区划分为不同类型：多数特区未能产生正向空间溢出效应（以飞地模式运作），部分特区会吸纳周边地区的资本（产生回流效应），仅有极少数例外如索龙特定经济区（KEK Sorong）可真正成为区域增长引擎。 4. 研究阐释 经济政策并非在真空环境中推行。若未强制要求本地后向关联，注入经济特区的资本将被困于特区边界之内，或回流至经济核心区域（爪哇岛），致使作为特区承接地的外围地区难以获得长期的经济繁荣。 5. 使用与复现方法 本数据集旨在实现完全可复现性。原始数据包含Excel文件与压缩格式的形状文件（shapefiles）。用户仅需在R Studio中运行master_R_script.R主脚本即可。该单一脚本可自动完成缺失值过滤、多重共线性（方差膨胀因子（Variance Inflation Factor, VIF））诊断、局部空间自相关计算、空间杜宾模型估计以及稳健性检验。针对零投资行政区的对数变换操作将自动完成，并输出最终结果工作簿。