Spatial Dependence And Influence Factors Of Carbon Productivity Among Provinces In China

Based on the theories of economic growth and the New Economic Geography, this paper gives an estimation of the carbon productivity of29provinces in China from1995to2009, and analyses their productivity differences and the causes of these differences. Firstly, values for carbon production of the targeted provinces are estimated, based on which estimation of carbon productivity is done by deploying the DEA method. Then analys is given to the issue of carbon productivity clubs convergence among China’s provinces, the tools used being the markov chain and space markov chain, and the space pattern and process of the convergence as well. Finally comparisons are made by applying the hybrid Pooled Least Squares, fixed-effects models and random effects models, which enables further research on the influence of such factors as openness level, technological investment, public investment, and industrial structure to carbon productivity.In specific terms, this paper firstly describes the background and significance of the research, and proposes its research framework based on a literature review of domestic and overseas research. As the world’s largest developing country, it is now crucial for China to maintain healthy and sustainable economic development. In the meantime, the country has to base any of its carbon emission reduction schemes on a guarantee of certain economic growth. And only when GDP and carbon emission reduction amount are considered as two interrelated indexes can the emission reduction be achieved with economic growth being guaranteed. Hence research on carbon productivity is both significant and valuable. It is found through literature review that domestic research about carbon productivity is still lacking. Literature is especially rare on topics about carbon productivity of provinces. As a result, this paper analyses productivity of carbon in China’s provinces and its transmission amongst these provinces based on the markov chain and space markov chain.Relevant energy data is used to obtain the value of carbon emission for each province. A DEA model based on fixed investment-oriented returns to scale is then selected, and is adopted to estimate changes of carbon productivity in the29provinces in China from1995to2009, with regional GDP value being the desirable output and capital deposit, labor and carbon emission being the input variables. Common markov chain based analysis shows that carbon productivity is relatively stable in all the provinces and that possibility of transmission is relatively low to either higher or lower level. Meanwhile, provinces located in mid-to-low levels and mid-to-high levels are of little likelihood to move upwards or downwards. Comparatively the possibility to stay still is higher. Among others, moving to mid-to-high levels from mid-to-low level in the early stage is of the highest possibility, proving that upward move from mid-to-low levels is quite possible.Further, space markov chain is applied to research on the influences of geographic environment on productivity transmission. Space markov chain matrix helps to find that elements in the matrix are not equal to those on the same positions of common markov chain. This demonstrates that carbon productivities are not absolutely independent among the provinces, but are correlated spatially. Further research suggests that high level productivity environments would facilitate move-up of mid-to-high levels. So mid-to-high levels and mid-to-low levels do to mid-to-low levels and low levels. That is to say, technical diffusion effects emerge in regions with similar productivity levels.Such methods as LLC, IPS, ADF and PP are used to conduct unit root test to relevant date so as to measure its stability, and Pedroni’s heterogeneous panel cointegration test and the Kao method are deployed to do co-integration test to carbon productivity and its influence factors. Meanwhile, in order for the long-term relationships to be effective identified, the error correction model can help to rule out the impacts of false returns resulting from unsmooth variables and distinguish between short-term volatility and long-term equilibrium relations between the variables. In addition, the granger causality test is done, to do comparative study on the influence factors of carbon productivity based on the hybrid Pooled Least Squares, fixed effects model and random effects model. Finally a conclusion is made. Openness level, technical input and public investment have distinct positive influences to the increase of carbon productivity, and industrial structure has negative relations with the increase of carbon productivity. The ending part of this paper draws a conclusion of the findings, gives relevant policy-level suggestions, and discusses the limitations of this paper and directions for further research.