Chinese-Style Decentralization and Environmental Pollution:An Empirical Study of Air Quality Based on Provincial Panel Data in China
- 2015年 
【中文摘要】在当前工业化和城市化髙速发展时期，我国的环境形势严峻。我国的环境问题是由粗放型经济发展模式导致的，而这种发展模式又源于中国式分权下的政府行为。利用1994一2012年省级面板数据，采用双向固定效应模型，可检验中国式分权对我国空气质量的影响。研究发现，中国式分权体制下，经济发展仍是以牺牲环境为代价，分权提髙了环境库茨涅茨曲线（Environmental Kuznets Curve，简称EKC)转折点所要求的平均收入水平。支出责任过度下放会导致空气质量恶化，而充足的财政自主权能够抑制空气污染。此外，中国式分权对空气质量的影响存在显著的地区差异。因此，合理划分政府间事权、提髙地方政府的财政自主度、将环境质量纳入官员考核机制、激励地方政府间良性竞争和科学发展，是改善地方环境质量、提髙公共服务供给效率的必要举措。 【Abstract】In the context of accelerating industrialization and urbanization, China is facing serious challenges in environmental protection. Environmental problems in China mainly result from the extensive pattern of economic development, which originates from the government's behavior owing to Chinese-style decentralization. This study uses the provincial panel data from 1994 to 2012 and the two-way fixed-effect model to look into the impact of Chinese-style decentralization on air quality in China. Our results show that China's economy still develops at the expense of environmental sustainability within the framework of Chinese-style decentralization. Decentralization raises the average income level required by the turning point of Environmental Kuznets Curve in China. Excessive decentralization regarding expenditure leads to deterioration in air quality while sufficient fiscal autonomy can slow down or curb air pollution. In addition, it is found that the impact of Chinese-style decentralization on air quality varies significantly from province to province. Therefore, it is concluded, in order to improve the quality of local environmental protection and enhance the efficiency of public services, it is necessary to assign intergovernmental responsibilities reasonably, increase local governments’ fiscal autonomy, bring environmental quality into the assessment of government officials, establish an incentive mechanism to promote healthy competition among different governments, and take a scientific approach to development.