【作者】 周刚；

【导师】 杨招军；

【作者基本信息】 湖南大学 ， 数量经济学， 2005， 硕士

【摘要】 税收是基本的财政收入形式之一。随着市场经济的不断发展,在征管改革逐步深入,税务内部分工越来越走向精细的今天,税收管理和税源正常管理的日常监控出现了一定的真空,这在一定程度上影响到国家税款的及时足额入库。相对于国外一些先进、完善的纳税评估系统而言,我国在纳税评估方面的工作有待于深一步加强。峰值分析法是当前我国国税部门用得比较多的方法,它有一定的理论依据,而且在实际应用中取得了一定的效果。不足的是,当前税务部门在使用峰值分析法中,所考虑的指标偏向于经验,而且考虑的面比较有限。从Allingham和Sandmo首次建立规范性计量经济模型研究纳税评估问题以来,很多学者从各个角度在纳税评估模型方面都进行了的研究。本文在经过实地调查后,在当前税务部门所使用的峰值分析法的基础上,综合考虑了纳税评估类的各项指标,进一步改进峰值分析法,而且从多角度对纳税人的申报数据进行初步审核。在此基础上,再通过人工神经网络方法对识别为非诚信纳税人进行二次识别,最终确定非诚信纳税嫌疑人。本文基于实用和有效的目的,考察了当前实际部门所使用的纳税评估方法和很多学者在纳税评估方面所做的理论研究,分析实际部门所急需解决的问题,结合很多学者在其它领域所采用的有效方法,在原实际部门使用方法的基础上对其加以改进。同时,运用非线性方法对初次识别的结果进行确认,有效的提高了纳税评估工作的效率。最后,为使本文的研究能更好、更快的取得实际的效果,在本文的最后,运用面向对象的程序设计语言将所做的研究通过操作简便、界面友好的软件实现。更多还原

【Abstract】 The tax revenue is one of the basic financial revenue forms. With the unceasing development of the market economy, gradual profound reforms in the collection reforms management, and the accurating internal division of the labor in tax affairs today, daily monitoring for the tax administration and normal management of the tax source appeared a certain vacuum, which affecting the prompt storage of the national tax revenue with full amount. Compared with the advanced tax compliance system overseas, our country needs to further strengthen in this field. The peak value analysis, has a certain theory basis and has obtained a certain effect in practice, which our national tax departments used commonly at present. What insufficient is that the index considered by the tax department at present when using the peak value analysis is partial to experience and the scope of consideration is quite limited. Since the first establishment of the normative metrology model which researches into tax compliance problems by Allingham and Sandmo, many a scholar have carried on research from many aspects to the tax compliance model. The peak value analysis has tried to be further improved and the taxpayer’s declaration data has tried to be preliminary audited from multiple aspects in this article, which were done after the on-the-spot investigation, on the basis of the peak value analysis which tax departments use at present and under the synthetic consideration on every index of the tax compliance. On this foundation, it discriminates the ones who have discriminated as un-faith taxpersons by the means of artificial neural network for the second time and determine the un-faith tax payment suspects at last. Based on practical and effective goal, inspected the tax compliance method before used by the tax department at present and the theoretic achievements by many scholars, analyzed the tax department’s problems which need urgent solution, and combined the effective methods which many scholars used in other fields, this article has tried to advanced the method used by the original tax department. At the same time, the article has effectively improved the efficiency of the tax compliance work through affirming the result of the elementary discrimination by handling the nonlinear method. Finally, to enable the research in this article better and quicker to obtain the actual effect, at the end of this article. The research which was done by utilizing the object-oriented programming language is realized through an easy-operating and friendly contacting surface software.更多还原