【作者】 韩东安；

【导师】 王雅林；

【作者基本信息】 哈尔滨工业大学 ， 技术经济及管理， 2008， 博士

【摘要】 高科技企业在全球范围内已迅速发展起来,成为重要的经济增长推动力量。由于高科技企业在技术、运作模式等方面所具备的特殊性,使得运用传统财务监控模式对高科技企业经营者财务监控的效果并不理想。一些资料显示高科技企业的发展除面临技术、市场、资金等外部风险之外,还面临着由经营者产生的内部风险。同时由于企业内部制度不健全、经营者的不善管理、以及监控不到位等因素,造成高科技企业核心技术人员流失,从而导致企业的核心竞争力下降,直接损害了股东、政府以及其他利益相关者的利益。本文从高科技企业经营者财务监控问题出发,探索财务监控主体监控模式、监控模型的构建以及监控环境的评价等领域内容,为高科技企业经营者舞弊行为监控提供核心技术和支撑环境,以便推动企业的健康稳定发展。首先,本文对经营者财务舞弊问题的监控进行了框架性研究,把高科技企业经营者财务监控系统划分为主体系统、运行系统、环境系统三个子系统,并以此为基础构建了本文的研究框架。其次,对三个子系统进行深入研究。主体系统方面,依据财务收益索取权,本文确定高科技企业的监控主体。通过对高科技企业发展过程中的融资方式、人力资本特征、以及政府功能比较分析,按照资金来源不同,提出了高科技企业监控主体系统的新型分阶段形态,即股权持有者、政府、人力资本所有者组成的的无借贷资本形态和由股权持有者、政府、人力资本持有者、债权人组成的有借贷资本形态。同时,引入不确定环境下的博弈模型作为监控主体间博弈分析的基本模型,提出确定政府代表作为博弈过程的调解员,由博弈分析得出使总体利益最大的最优策略。最后依据各个主体的特征构建了各个主体的监控模式。再次,本文将环境系统定义为支撑主体系统与运行系统发挥作用的制度保障体系。在对前人研究分析的基础上,设计一组问券调查,并对问券调查结果进行了的分析,针对高科技企业的特征、选择有代表性的指标建立环境系统的评价指标体系,采用层次分析法确定指标的权重并以可拓学的多级物元作为基本概念,构建环境系统评价模型。最后以哈尔滨科力电力有限公司作为实证样本,对模型进行验证。验证发现,本文所建立的模型对高科技企业经营者有较好的评价效果,能有效的评价高科技企业经营者财务监控的环境系统。最后,运行系统解决的是监控方法的问题,本文除了对现有的监控手段进行比较分析之外,构建了基于神经网络的适时监控系统。选择反映财务舞弊的指标,以大量的上市公司数据作为训练样本,培植出具有高分辨率的适时财务监控神经网络。通过对已有数据的验证,发现网络的有效分辨率达到94.3%。能较好的通过财务数据反映财务舞弊情况,提高监控主体的事前预判能力。更多还原

【Abstract】 As the rapid development of high-tech enterprises in the word, it has become the main engine of economic increasing . Because of the uniqueness of high-tech enterprises’technology and operational mode, the traditional financial supervision mode is not suitable. Besides the external risks of technology, market and assets, high-tech enterprises alse face to internal risks caused by the operators. Datas show that due to the imperfections of the institutions, the poor management of operators and the lack of supervision, the core technical person leave frequently, which decreased the core competitiveness. And this damaged the benefits of shareholders, the govement and other stakeholders. This paper concentrated on studying the fiancial supervision system model, the model construction and supervision environment evaluation to improve the enterprise development.There are four parts in this paper.Firstly, this paper research on the financial fraud supervising framework, and divides the financial supervision system into main system, operating system and environment system.Secondly, the paper takes the right of benefits as the standard to build up the main system of high-tech enterprise. By analyzing of financing capital, human capital and govement function during the develop process of high-tech enterprise, the paper develops the stage model of the main system includeing non-loan capital—contains of shareholders, govement and human capital holders, and loan capital—contains of shareholders, govement, human capital holders and creditors. By introduction of the game under uncertain environment as a basic model, the paper points out that the operators determine the mediation process to achieve strategy of maximize profits. Then we construct the main system supervising model based on main characteristic of each stage.Thirdly, in the environment system part, the paper defines the environment syetem as instituations security system to support the main system and operating syetem. According to previous research and the high-tech enterprises characteristics, making a set of questionnaire to environment system of the high-tech enterprises, we select some respective indexes to construct the evluation indexes system of environment system. Then this paper uses AHP to fix the weights and build up the environment evluation system model based on the multilevel matter element of comprehensive extension evaluation.we choose Harbin Keli electrical company as a demonstration to verify the model. The results show that the model has a good effect on the environment system estimation of high-tech enterprise.Finally, the operating syetem can resolve the problem of supervision method. By compared analysis of existing financial supervising, this paper constructs the real time supervision system based on neural network. Then we selected financial fraud indicator, based on the training samples of listed companies, and constructed a timely financial supervision neural nerwork which was effective. Past data show that the network differentiates the financial fraud from the financial date at an effective rate of 94.3%. And the nueral network can also improve the prior predict capacity of supervision main.更多还原