Spatial Flood Risk Mapping in West Sulawesi Using a Regression Model with Moran’s Index and Its Implications for Risk-Based Disaster Insurance

Abstract

This study mapped the risk of flood disasters in six districts on West Sulawesi for the period 2015–2024 using a regression model combined with Moran's Index. The results of flood risk zone mapping are used to develop spatial risk-based flood insurance. The flood risk index (Y) was analyzed against the influence of five independent variables. However, in multiple linear regression tests, there are only two significant variables, namely the variables X2 and X3. Meanwhile, Moran's I test in six counties showed no statistically significant spatial autocorrelation for any variable (all p > 0.05), reflecting a very low test power with only six spatial units. Therefore, spatial dependency was tested through a spatial panel model on the observation of 60 data (10 years for six districts). The results of the Lagrange Multiplier test detected significant spatial-lag dependence, in addition to that the Spatial Durbin Model also provided the best stable match, including significant spillovers from neighboring population densities. Based on the normalized flood risk index, it is known that Majene is the area most at risk of flooding, as well as Mamasa which is the least at risk of flooding, with Pasangkayu, Mamuju, Central Mamuju, and Polewali Mandar at moderate risk. This spatial result rests on only six units and is interpreted as exploration. As a practical implication, the resulting hotspot-coldspot zoning is used to outline risk-based premium structures where high-risk districts attract higher risk burdens, while premium affordability is assessed relative to regional economic capacity (GDP), providing an actuarial bridge between spatial risk maps and disaster insurance designs.