The spatial autocorrelation model studied in the framework of structural equations is the spatial error regression model. The results of this study are applied to South Sulawesi's Gross Regional Domestic Product (GRDP) data. For parameter estimation using open source software Mx. To implement the spatial error model in SEM, two new sets of weighted spatial variables need to be formed, namely W based on the dependent variable (PW) and ηW based on the independent variable (PW) and ξW based on the independent variable (QW). Since in the case of the latent model, the variables P and Q cannot be observed directly, then ηW and ξW are directly defined by the observation variables (indicators) Y yW and Y xW which are related to each other as Yy and Yx to η and ξ. obtained a model that represents the spatial error in SEM. By using South Sulawesi GRDP data where y represents the per capita GRDP in the Regency/City, x1 and x2 respectively represent the value of the Mining sector and the building sector in the Regency/City. XW1 represents first-order contiguity spatially lagged for trade and XW2 represents first-order contiguity spatially lagged for agriculture. yW denotes spatially lagged first-order contiguity for GRDP. (1−λ)γ0 represents the unit variable coefficient. From the model it can be stated that GRDP (y) is influenced by several sectors in the economy such as mining (x1) and building (x2). In addition, there is a location effect (Spatial Effect) that affects the GRDP in South Sulawesi. Based on the final results obtained, it is known that λ = 0,16 which indicates that there is a dependency on the GRDP data in South Sulawesi in 2008 between one district/city and another district/city based on the spatial correction. Areas that are centers of mining and construction in South Sulawesi are mutually dependent, causing dependence on GRDP data, this can be seen in the positive covariance value between mining lagged, and building lagged, and lagged GRDPKeywords: Effect Spatial, Error Spatial, SEM, GRDP
Copyrights © 2021