Evaluation of Panel Regression Models in Estimating the Relationship between Water and Countries' Agricultural Added Value in the Aras River Basin

Countries in trans-boundary river basin have always been competing in use of common water source. The geographical location and the number of countries included as well as economic performance of these countries escalate the complexities of shared waters. Therefore, proper analysis of the economic performance of neighboring countries will assist future studies. In this research, Aras River between Iran, Turkey, Armenia and Azerbaijan is chosen. First, the basin is extracted using 30 meters cell size Digital Elevation Model in ArcMap and then it is divided into 10 sub-basins to obtain physiographic properties in terms of average slope, average height and area. Using rainfall data of synoptic stations and surface runoff of hydrometric stations, average annual stream-flow of each sub-basin is estimated. Then, Seemingly Unrelated Regression model as well as Panel Data regression model are used to model the relationship between annual added value data of agricultural sector as dependent variable on available annual water, labor force and capital of each country. Statistical interpretation of the results from different models, proves that Fixed Effect model describes the relationship better and can explain the cross sectional heterogeneity up to 73 percent in added value. Moreover, squared water coefficient term, illustrates the nonlinear relationship between water and added value and is statistically significant.