A dynamic-time dependent spatial autocorrelation detection for East Java's Covid-19 regional percent of cases, March 2020-March 2021 (Indonesia)

Abstract

Covid-19 regional percent of cases is one of the regional variables that dynamically interact across space and time. It exhibits a time trend, and at one point in time, it may form clusters of regions with similar values. Since Covid-19 is an infectious disease, the regional percent of cases also exhibits spatial dependence across regions. The time trend indicates the possible time lag of the spatial dependence, and the spatial dependence analysed at one point in time may be undetected. This situation was observed in the 38 regions of East Java. It gives an incorrect impression of the nature of spatial dependence, leading to an improper policy formulation. To capture the spatial interaction more accurately, this study accommodates the time-dependent dynamic nature of the variable into the formulation of the Moran's I index for a set of spatial panel data. A simulation study is conducted to confirm the accuracy of the proposed index, especially when the degree of contemporaneous spatial autocorrelation is high. The proposed index also succeeds in detecting the time-lagged spatial autocorrelation of East Java's Covid-19 regional percent of cases. It provides a better understanding and policy recommendations regarding the spread of this disease in this province.