Analisis Spasial dan Variasi Lokal Merchant QRIS Menggunakan Adaptive Geographically Weighted Regression

Abstract

Indonesia’s digital payment ecosystem increasingly relies on the Quick Response Code Indonesian Standard (QRIS). However, its distribution still exhibits clear spatial disparities between western and eastern regions. This study aims to analyze local variations and the determinants influencing the number of QRIS merchants across Indonesian provinces in 2024. The analysis employs Adaptive Geographically Weighted Regression (AGWR) with an adaptive bisquare kernel to capture spatial heterogeneity that cannot be explained by global models such as Ordinary Least Squares (OLS). The independent variables used include Gross Regional Domestic Product per capita, average years of schooling, digital infrastructure, urbanization rate, population density, number of Micro, Small, and Medium Enterprises (MSMEs), and internet access. The results indicate that AGWR outperforms OLS, with the Coefficient of Determination (R²) increasing from 0,806 to 0,976 and the Adjusted R² from 0,751 to 0,905. Additionally, the Akaike Information Criterion (AIC) decreases from 1003,417 to 967,981, while the Sum of Squared Errors (SSE) drops significantly from 1,91×10¹³ to 2,32×10¹². The empirical findings reveal that the number of MSMEs is the most consistent determinant of QRIS adoption across regions. Socioeconomic factors exhibit strong influence in Java but show limited relevance in eastern provinces such as Papua and Maluku, suggesting the presence of structural constraints in these areas. This study recommends implementing location-specific financial inclusion strategies rather than uniform national policies.