Evaluation of Simultaneous and Simultaneous Spatial Autoregressive Equation with Three Stage Least Square Method (Case Study on GRDP, Poverty and Unemployment Data in Papua in 2018)

Abstract

In the simultaneous equation (SM), the dependent variable in an equation could be an independent variable in another equation. Dependent variables are called endogenous variables and independent variables are called exogenous variables. If the observations in SM are locations and contain spatial autocorrelation, then a spatial dependency model can be added. Adding dependencies to endogenous variables is modeled as spatial autoregressive regression (SAR). Addition of SAR to the SM model hereinafter referred to as simultaneous spatial autoregressive regression (SM SAR). Estimation of parameters in the SM SAR can use the Three Stage Least Square (3SLS) method as used in SM. This study aims to predict this SM SAR model on GRDP, poverty and unemployment data in Papua in 2018. SM consists of three equations. The weight matrix used in each equation can be the same or a combination of inverse distance, rook contiguity, exponential weight and K-NN matrices. The results showed the addition of SAR in SM could reduce MSE by 0.31and an increase in R² by 4.68% compared to the SM model.