Spatial Panel Data Model And Its Application

Spatial econometrics experienced a budding, growth and steady development of the three stages in the past three decades. One of the classic assumptions that observations should be independence is relaxed. After accounting into the spatial effects of observations, the equation on the right side can include the spatial lag dependent variable, the spatial lag explanatory variables, and errors can also include spatial lag items. These changes contribute to the research and development of spatial econometrics. The purpose of this paper is to review the models specification and estimation methods of spatial econometric, and apply these models to the economic research. in China.(1) Theoretically, I reviewed the process of development of spatial econometrics; models of the static and dynamic spatial panel data and their specification, estimation methods and hypothesis testing. I introduce some recent developments in spatial panel data models.(2) Then, I apply these models to Chinese economic researches as follows:①Knowledge capital to be divided into measurable intellectual capital and immeasurable intellectual capital. The measurable knowledge capital stock is be indicated by the numbers of patent applications. It is assumed that all the measurable and immeasurable knowledge capital obeys spatial autoregressive process. Through C-D production function, I deduce a spatial econometric model that links the total factor productivity with the measurable knowledge capital stock. Existing literatures on the relationship between knowledge capital stock and total factor productivity use more data about companies, universities and other micro-individual data. Using the panel data about Chinese30districts from1996to2010, this paper utilizes SDM model to analyze the spillover effects of the knowledge capital stock on total factor productivity. Results show that the spillover effects between provincial regions exist. Therefore, we should encourage cooperation between the regions, to encourage the flow of talent and technology between the regions and increase the absorption capacity of the region’s learning in order to reduce the gap of knowledge capital stock between regions, which is helpful for narrowing the income gap between regions.②Factors affecting the regional unemployment rate and unemployment rate show spatial clustering phenomenon, which means that the spatial evolution of regional unemployment rate is caused by urbanization characteristics of the various regions because the region’s economic, socio-demographic attributes behave the cluster effect in space. Using the panel data about Chinese30districts from1999to2010, this paper utilizes spatial econometric methods to analyze the determinants of the regional unemployment rate differences. The results show that employment growth in the primary and secondly industrial employment, educational attainment are important factors affecting the differences in unemployment areas. Changes in these factors not only affect the local labor market, also affect the adjacent areas through spillover effects.③The primary objective of this study is to test the EKC for CO2in China over the period of1995-2009. The CO2EKC hypothesis is tested under the spatial econometric framework. LM (robust LM) tests suggest spatial lag model is appropriate. Furthermore, the Hausman test indicates fixed effect is suitable. Then spatial lag panel data model with fixed effects is used to investigate whether the EKC for CO2exists. The empirical results suggest the existence of a Environmental Kuznets Curve for CO2in China because of the positive sign with per capita GDP and the negative sign with the quadratic term of per capita GDP. Finally, calculations indicate the turning point of Carbon emissions occurs when per capita GDP equals73630yuan.④This paper measures the economic impacts of climate change and non-climate factors on China’s agriculture based on the spatial panel data model. By using province-level data on agricultural net revenue, climate, and other non-climate data for31provinces from1997to2009, The results suggest that the provincial agricultural net revenues behave positive spatial autocorrelation. Among the climate variables, except other seasons, higher summer temperature would reduce the agricultural net revenue, but the agricultural net revenue would benefit from the interaction between summer temperature and precipitation.This shows that under the extreme summer climate, only temperature and rainfall amounts are coordinated, would the agricultural net income be improved. In addition, non-climate factors (fertilizer application, the effective irrigation area, extent of mechanization and population density) are positively correlated with the and agricultural net income. Based on these results, the article finally puts forward some suggestions on water infrastructure, agricultural production capacity and the adjustment of agricultural structure.