A single equation model that is often used ignores the interdependence between response variables. Frequently encountered variables that have a two-way relationship. These interrelated two-way relationships can be summarized in a simultaneous equation model system. There is a relationship between variables which are in fact dynamic. In the system model of simultaneous equations with dynamic panel data, each structural equation is a dynamic panel data regression equation. The estimation of using Ordinary Least Square (OLS) in the dynamic panel data model results in biased and inconsistent predictors because there is a lag of the dependent variable that correlates with the error. First difference in dynamic panel models is used to eliminate individual effects. Instrument variables are needed, namely variables that do not correlate with errors. Therefore, dynamic panel data models are more suitable to be used in analyzing poverty and social change. From the simultaneous equation model obtained, the dominant factors affecting the level of poverty in Central Java Province are the unemployment rate, Human Development Index, labor force participation rate, population, and Gross Regional Domestic Product.