【作者】 孙智勇；

【导师】 刘星；

【作者基本信息】 重庆大学 ， 技术经济及管理， 2010， 博士

【摘要】 税收是一个内容丰富而又充满变化的话题。从历史上看,各国政府都曾尝试过使用各种手段保证国家财政收入的主要构成部分——税收收入的稳定增长,或出台各种税收政策作为有效实施国家财政政策的重要工具。而税收制度更是从美索不达米亚和古埃及文明的出现之日起,就伴随着一系列历史事件演化至今。当前不同国家的税制虽然不尽相同,但是总有共同之处,即税收体制不仅牵涉到千家万户的实际经济利益,而且成为国家宏观经济调控体系的重要组成部分。同时,税收体制的演变也势必遵循一定社会经济活动的路径影响。因此,随着中国经济与全球经济的相互依赖不断加深,市场经济体制的改革范围不断拓宽,如何理解存在于税收研究领域的重要问题,分析税收政策出台所导致的税收收入的变化形势,把握发生在税收体制上的最新变革方向,就成为我们研究税收问题时面临的紧迫课题。本文在回顾税收理论发展的基础上,对税收与经济增长的关系作了简要介绍,阐述了税收政策的实施及其制度改革与经济社会同步发展的重大意义。同时,对税收预测问题作了详尽的规范分析,建立了基于灰色理论的单一税收预测模型以及代表性地选择了Elman回归神经网络模型、含政策虚拟变量的自回归模型、ARIMA(1,1,1)的时间序列模型和多因素SVM回归模型等四种模型作为组合预测模型,系统、深入地研究了税收收入预测问题。研究结果表明,这些模型虽然有不同程度的缺陷,但均可以较好模拟出税收总量的变化趋势,给出较满意的预测结果,而组合预测模型比单一预测模型能更好地减少预测误差。并且两者在不同程度上都验证了在影响税收收入结果的因素中,政策变化因素对税收收入的影响是非常显著的。在上述分析的基础上,本文先是结合我国税收结构变化作为背景描述,对改革开放之后的中国结构性减税政策演变进程进行了归纳总结,并依据众多税收结构实证文献,创新性地通过关联规则的数据挖掘方法,对最近的一轮结构性减税政策出台前后的各主要税种数据进行关联度的定量分析,得到几个主要税种对税收收入的影响因子,从数据分析的角度探讨了结构性减税政策对主要税种以及总的税收收入的影响。其次,本文运用模糊软集合方法将支持向量回归机、灰色预测模型以及非线性最小二乘回归分析等三种预测方法结合形成组合预测模型,通过该模型对税收增长、国民生产总值(GDP)、单位GDP能耗以及单位税收能耗数据之间的变化关系进行预测研究,并特别针对第二产业中的高污染、高能耗行业以及第三产业中的新兴行业的税收收入与能耗数据进行分析对比,从转变经济增长方式的角度出发,预测结构性减税政策未来的新趋势,进而改革和完善税收制度。本文最后总结了几个重要论点。更多还原

【Abstract】 Tax revenue is a topic rich in content and full of changes. In history, all governments had been employing every means to ensure the steady growth of the main component of national fiscal revenue, i.e. tax revenue or issuing various tax policies as an important tool to effectively bring national fiscal revenue policy into effect. Since its birth in Mesopotamia and ancient Egypt, tax system has evolved with a string of historical events into what it is now. Today, different tax systems are adopted in different countries, but they have something in common, i.e. they not only concern the practical economic benefits of each household, but also have become a critical part of the National System of Macro-economic Control. Meanwhile, the evolvement of tax system is influenced by some social and economic activities. Therefore, with the further interdependence between Chinese economy and the world economy and the expansion of the market-oriented economic system reform with socialist characteristics, it is imperative to study how to deal with the key issues occurring in the taxation field and how to properly understand the new changes in tax system.After a review of the development in taxation theory, this study briefly introduces the relationship between tax revenue and economic growth and clearly states the importance of keeping the tax system reform in pace with the economic and social development. In addition, the study gives a detailed analysis on the prediction of tax revenue and sets up a single forecast model based on the Grey Theory and a combined forecast model made up of the Elman Recursive Neural Network Model, the Auto-regressive Model with policy as the dummy variable, the ARIMA(1, 1, 1) Model of Time Series and the Multi-factor Regression Model. Results show that these models can simulate the varying tendency of the total tax revenue and make a satisfactory forecast despite their own flaws to some degree; the combined forecast model can better reduce forecast errors than the single forecast model; the two models prove to one degree or another that among all factors affecting taxation results, policy has a very significant effect on tax revenue.Based on the above analysis and an overview of changes in tax structure, the study first summarizes the evolving process of the post-reform-and-opening-up structural tax reduction policies in China. Grounded on numerous empirical literatures and by the creative use of the association rules data mining analytical method, the quantitative analysis of correlation degree is conducted on the data of the main taxes before and after the release of the latest structural tax reduction policy and thus the impact factors are gained and the effects of the tax reduction policy on the main taxes and the total tax revenue are explored. Second, the study employs fussy soft sets to form a combined forecast model made up of three forecast methods, i.e. the Support Vector Regression Machine (SVRM), the Grey Model (GM) and the Nonlinear Least-squares Regression (NLSR). Based on the combined forecast model, the relationships among tax revenue growth, gross domestic product (GDP), energy consumption per unit of GDP and energy consumption per unit of tax revenue are analyzed and the contrastive analysis on tax revenue and energy consumption is conducted between the secondary industry featuring high pollution and high energy consumption and the third industry featuring new industries. Thus, from the perspective of transforming mode of economic growth, the new trend in tax policy can be predicted, which can better serve the perfection of tax system. Finally, the major points in this study are summarized.更多还原