【作者】 滕焕钦；

【导师】 胡金焱；

【作者基本信息】 山东大学 ， 金融学， 2011， 博士

【摘要】 我国保险业从恢复国内业务至今,经历了30多年的快速发展,保险公司由1979年的1家发展到2009年的121家,保费收入年均增长20%以上,2009年,保费收入首次突破万亿元,成为全球第7大保险市场,2010年保费收入达到1.45万亿元,从业人员超过300万人。2000-2009年,我国保险密度从15.2美元/人上升到121.2美元/人,保险深度从1.79%上升到3.4%,保险与人们的关系越来越密切,在国民经济中的地位越来越重要。但同世界水平相比,我国保险业仍处较低水平,2009年,世界保险密度595.1美元,我国仅为其1/5,居世界第64位,世界保险深度7%,我国尚不及其1/2,列世界第44位,这也意味着我国保险市场存在巨大发展潜力。但是保险业快速发展了30年,不可避免地积累了一些深层次矛盾和问题,这次百年一遇的金融危机使这些矛盾和问题凸显出来。为了保障我国保险业的良性持续发展,确保保险业发挥现代经济“助推器”和社会“稳定器”的独特功能,对保险公司风险预警机制进行研究具有重要的理论意义和现实意义。鉴于寿险公司与财险公司风险差异明显,限于时间、能力,本文只针对我国财产保险公司风险预警系统的构建进行研究,以期对我国保险企业风险预警机制的建立与完善有所裨益。本文根据全面风险管理(ERM)框架,并参照国内多数研究者所接受的“明确警义、寻找警源、分析警兆和预报警度”的逻辑框架,围绕“界定预警对象—识别风险因素—设计指标体系—建立预警模型”的基本柜架,来构建我国财产保险公司风险预警系统。首先明确预警对象,解决“对谁预警”的问题。借鉴国内外理论研究与应用实践,结合我国当前财产保险行业发展和监管现状,确定以偿付能力是否充足作为判定问题公司的标准。因为偿付能力是保险公司生存与发展的前提,是维护被保险人利益的基本保证。保险公司只有具备充足的偿付能力,保证保险事件发生时被保险人能够获得保险合同约定的经济赔偿或给付,才能有效地保护被保险人利益,实现保险公司乃至整个保险业的稳健发展,维护社会稳定。偿付能力也是世界各国保险监管核心内容之一,国际保险监督官协会(IAIS)财务监管、公司治理监管、市场行为监管的“三支柱”中,偿付能力是第一支柱的核心内容,中国保监会监则直接将偿付能力监管与公司治理监管、市场行为监管作为监管“三支柱”之一。因此,将偿付能力是否充足作为判定问题公司的标准较为合适,在国内外研究中较为普遍。在我国,随着《保险公司信息披露管理办法》的实施,2011年开始,各保险公司必须公布偿付能力等关键指标,国内研究开始具备以偿付能力为判定问题公司标准的条件。其次,进行风险识别,分析财险公司面临的风险因素,解决“对什么预警”的问题。基于全面风险管理(ERM)理念和方法,根据财险公司行业特征,本文对保险公司面临的风险进行了全面分析,识别风险因素即寻找警源,并根据我国财险公司的制度性、阶段性特征进行具体分析,勾画出现阶段我国财险公司面临的主要风险,回答“对什么预警”的问题。第三,选择预警指标,设计指标体系,解决“以什么预警”的问题。选择预警指标即进行警兆分析,筛选可揭示风险因素、预示警情的代表性变量。基于全面性和系统性的原则,借鉴国内外理论研究中有关风险指标体系的设计方法及金融监管部门设定的风险预警指标体系,本文选取并建立10类51项有关财险公司风险综合预警指标体系参照国内外理论研究成果和监管实践标准,设计了我国财产保险公司风险预警指标体系,在此基础上,根据相关公司实际经营数据的可得性,进一步筛选出5类18项财务指标,并对其设定了相应的风险预警阀值,以此进行客观的风险评价预警,解决“以什么预警”的问题。在以上明确预警对象、分析风险因素、设计指标体系的基础上,本文建立了预警模型,以实现预报警度的最终目标,解决“怎么预警”的问题。这一部分是本文的核心与重点,包括两个基本环节。第一个环节是数据的获得与处理,本文兼顾实践者的主观感知结果和客观财务报表数据,尽可能地避免理论脱离实践的问题。通过调查问卷获得主观数据,对其进行有效处理,计算出指标体系中各指标(包括18个财务指标)的权重,由定性转化为定量；通过《中国保险年鉴》,获得各财险公司2004-2009年年度财务数据并对其进行初步处理,进而以调查问卷指标体系中计算得出的18个财务指标体系的权重,对各公司的各项年度指标进行归一化处理,为风险预警系统的构建提供有效的数据基础。第二个环节是风险预警方法的选择。本文曾选择试用了多种分析方法进行风险预警测量,结果发现,有的方法在国外对保险预警有效,或者对国内银行证券及其它企业进行预警也有效,却不适用我国财险公司风险预警。究其原因在于我国保险业发展时间相对较短,保险数据不完善,而且保险会计标准经常变化。经过论证,本文最终选择了三种相对有效的方法——离差分析、主成分分析和模糊综合评价,分别对相应的财险公司年度风险状况进行测量评价,并检验本文预警指标体系设计及预警阀值设计的有效性,为今后财险公司风险预警提供理论参照。本文选取占我国财产保险市场份额90%以上的21家保险公司作为研究对象,以其近6年来的数据资料进行预警分析,并以2011年初次公布的各财险公司2010年底实际偿付能力状况指标为参照,将运用各种方法得到的预警结果与之对比,进行实践检验,检验本文预警指标体系设计及预警阀值设计的有效性,实现了理论与实践的密切结合,从而证明了本文的理论价值及实践意义。三种预警方法的基本原理本文选择离差方法作为初步预警手段对各公司各年度进行单指标因素的风险测量计算,效果不太理想。单指标风险预警是测量最近几个不同年度内各公司同一指标的离差,并将离差大小分为A、B、C、D四个依次从小到大的等级。对存在问题的公司,对其一次“记录”。然后对所有公司的所有问题记录汇总,以“报警次数”衡量相应指标的风险大小,并对其进行预警排序。离差越大,意味着风险越高；一方面可以计量各公司各年度各个指标的风险值,对公司的风险值大小进行测量排序；另一方面还可以对各年度各指标的风险值大小进行测度。然后,根据各个公司在每年度各个指标的衡量结果,计算各公司相应的指标值。研究结果表明,以2006-2009共4个年度财务预警指标对2010年底偿付能力不足的财险公司进行预警,总体预警效果不太理想。其中数据年度越早,预警效果越差。本文选用主成分分析方法对相关各公司各年度指标进行综合实证研究。由于建立的指标体系中可能存在信息耦合和非独立性的状况。为了避免在指标体系中存在的这种指标间线性相关现象,得到一个综合反映财险公司风险状况的综合指标体系,利用主成分分析方法进行处理,并且分别对各个年度数据进行分析。检验结果显示,采用主成分分析方法对财险公司风险状况进行预警分析的结果并不理想,以2008、2009两个年度的数据进行预警,Ⅰ类、Ⅱ类错误均高达75%,噪音信号比也高达92.3%。结果表明,针对2010年末偿付能力不足的财险公司,采用主成分分析方法进行的分析,效果不太理想。本文采用模糊综合评价法,以相应的指标为预警因素,以21个财产保险公司2004-2009共6个年度(部分指标只有5个年度)的数据为预警因素值；根据保监会监管指标范围和数据分布,确定预警阀值；以预警阀值为标杆,运用各公司各年度数据进行预警,得到预警值；最后按照各指标的权重进行加权平均,得到各公司各年度的预警结果。结果与实践基本吻合。比较三种分析方法,离差分析和主成分分析方法效果不太理想。究其原因,不在于预警方法和指标选择,而主要在于这两种方法完全依据量化指标分析,而国内保险统计数据的质量不高、数据不全的问题较为突出,特别是统计口径和会计制度的频繁变化,进一步影响了统计数据的质量。因此,目前国内采用纯粹定量手段进行财险公司风险预警缺乏基础,不适宜采用纯粹量化指标和方法进行保险公司风险预警。而模糊综合评价法则被证明比较有效,因为它适应当前国内数据质量和数量的现状。该方法通过专家调查法确定指标权重,经过技术处理,把定量分析与定性分析结合起来,弥补了数据质量和数量上的缺陷。因此,在财产保险公司风险预警系统构建中,模糊综合评价方法可以作为优选方法。最后,为使预警系统发挥实效并获得不断改进,本文对建立财险公司风险预警体系提出了政策建议,对建立风险预警系统的原则与框架做了阐述,并对系统有效运行及逐步完善所需要的配套环境做了分析。本文可能的创新之处,一是设计了一套较为适应当前我国财险公司经营与监管水平的预警指标体系并设定了相应的预警阀值。本文在设计指标体系和设定预警阀值时,广泛参考了国内外理论研究和应用实践成果,但更重要的是适应数据可得性,紧密结合国内财险业实际,提高指标体系和预警阀值的有效性。二是本文对相关数据的获得处理,首次兼顾实践者的主观感知结果和客观财务报表数据,尽可能地避免了理论脱离实践的问题。通过专家调查法获得适当的主观数据,对其进行有效处理,获得相应的结果(权重),并结合客户财务数据,对数据进行规范处理,为后面的风险预警提供有效的基础。三是本文首次运用多种实证技术模型并使用主客观结合的面板数据,建立了针对我国财产保险公司的风险预警模型。四是本文运用国内首次公布的各财险公司偿付能力实际状况指标为参照,将本文的研究结果与之对比,进行实践检验。更多还原

【Abstract】 Since the recovery of domestic business, the insurance industry has experienced 30 years of rapid development, the number of insurance companises grew from one in 1979 to 121 in 2009, the averaged growth rate of annual premium income reaches more than 20% during the period. In 2009, for the first time premium income exceeded one trillion yuan, became the world’s 7th largest insurance market, and in 2010 premium income reached 1.45 trillion yuan, employing more than 3 million people.2000-2009, China’s insurance density increased from 15.2 U.S. dollars per capita rose to $121.2 per capita, insurance penetration increased from 1.79% to 3.4%, insurance and people are getting closer, in the national economy is becoming increasingly important. However, compared with the world level, China’s insurance industry is still in relatively low level, in 2009, world insurance density is 595.1 U.S. dollars, China is only 20% of it, ranked 64th in the world, and the world’s insurance penetration is 7%, and China is only about 50% of it. On the other hand, it also means that China’s insurance market is huge potential for development. However, the rapid development of the insurance industry for 30 years, inevitably have some deep-seated contradictions and problems, all of which highlight in the financial crisis.In order to protect China’s insurance industry to develop continuously, early-warning of insurance companies’risks has become more urgent. Limited time and capacity, this thesis only study on property insurance companies.Based on framework of the Enterprise Risk Management (ERM), focusing on "Define early-warning object-Identify business risk-Design index system Establishing an early-warning model", this thesis try to build the basic proposition of property insurance company risk early-warning system.The thesis first defines the "troubled" companies. Refering to domestic and foreign companies and the company’s definition of normal standards and combining the current property insurance industry and the regulatory status quo, the thesis chooses the adequacy of the solvency standards as a decision problem. Because the solvency is a prerequisite for the survival of the insurance company, and it can maintain the basic interests of the insured guarantee. As a special risk management industry, insurance industry has many stakeholders, with a wide range of social nature. Insurance companies only have enough capacity for compensation and benefits, can effectively protect the interests of the insured, ensure that access to insurance contracts when the economy promised damages, achieve the steady development of the insurance industry, maintain social stability. Meanwhile, the solvency is need to for insurance industry to achieve sustainable development, and also is the core of the world insurance regulation. Therefore, the solvency problem of the adequacy as a judgement of the company’s standards is appropriate.Secondly, based on the Enterprise Risk Management concepts and methods, with reference to theoretical research results and international regulatory standards of practice, design the property and casualty insurance company risk early-warning indicator system, and set the corresponding threshold. This thesis draws on theoretical study of domestic and international risk index system design and financial regulators to set the risk early-warning indicator system. With reference to comprehensive risk management (ERM) concept, the thesis try to find and analyze the risk factors answer "warning what" t;followed by the selection of the warning signs of early-warning indicators that can reveal the risk factors screening, indicating the representativeness of the risk variables. Based on the principles of comprehensive and systematic, the thesis selects 10 categories and 51 indicators to build early-warning system. On this basis, according to the company’s actual operating data related to availability, the thesis further categories 5 of 18 financial indicators and sets corresponding risk warning t threshold as an objective risk assessment of early-warning.Following, the thesis is to solve the problem on "how to early warning" by early-warning model to achieve the ultimate goal of pre-alarm level. This is the core and focus of the thesis. First, data acquisition and processing. The thesis dealt with related data, for the first time take into account the results of the practitioner’s subjective perception and objective financial statement data, as much as possible to combine theory with practice. Then, calculate the indexes’(including the 18 financial indicators) weight by treating the subjective questionnaire data appropriately by. Based on the China Insurance Yearbook (mainly 2004-2010), the thesis obtain the corresponding data and process them initially, then normalize the corresponding data according to the 18 indicator’s weight, and it provides an effective data base for the risk early-warning system. Secondly, choice of methods of risk warning. This thesis has been selected to trial a variety of risk warning measurement methods and found that some method are effective on early-warning of insurance in foreign countries, or in securities of domestic banks, or in other companies, but not in the risk early-warning of China’s insurance company. The reason is that China’s insurance industry relatively short development time, the insurance data is imperfect, and insurance accounting standards often change. After argument, the thesis chose three relatively effective ways-from the deviation analysis, principal component analysis and fuzzy comprehensive evaluation to assess the annual risk of the corresponding property insurance companies respectively, and test the effectiveness both of index system and the threshold, which would provide a theoretical reference to early-warning risk for property insurance companies.This thesis selects 21 insurance companies which sharing the market more than 90 percent as a research goal, based on their nearly 6 years annual data, with the risk of deviation analysis to measure single index, the principal component analysis and fuzzy comprehensive evaluation methods on a comprehensive evaluation of the risk status, and test the effectiveness. With the CIRC solvency index of property insurers as a reference, the thesis will use the results obtained by various methods contrasting the practice, to achieve a close integration of theory and practice. Thus proving the theoretical value and practical significance.The effect by choosing deviation method as an initial means of early-warning to measure the risk of the companies each year is not ideal. Single index is a measure of risk early-warning within the last several years away from the bad company of the same indicator, and the deviation were divided into A, B, C, D four levels in order from small to large. On the problems of the company will get a "record". Then all the problems of all records of the company’s summary, with "the number of alarm" to measure the size of the corresponding risk indicators and early-warning of its sort. The larger deviation means higher risk; one can measure both the various indicators for each year of the company’s risk and the size of the company’s risk measurement order; the other hand, can also be indicators of risk for each year the size of each size measure. Then, according to all the various indicators in the measurement of annual results of companies calculate the corresponding index. It shows with the results of the four early-warning indicators of 2006-2009 to early warn insolvency of the companies in the end of 2010, the overall effect is not very satisfactory. One year earlier the data, early-warning effect worse.The thesis chooses principal component analysis to study the relevant risk indicator for each year of the each company empirically. The creation of the index system of information may exist in the coupling and non-independence situation. In order to avoid such indicators exist linear correlation between the phenomena and get a comprehensive risk profile reflects a comprehensive index system, the thesis apply principal component analysis for processing, and data were analyzed for each year. Test results showed that principal component analysis carried out on the insurance company risk profile analysis of the results of early-warning is not ideal. Early-warning on 2008 and 2009 data, errors of type I and type II are up to 75%, the noise signal ratio as high as 92.3%. The results show that the effect of principal component analysis method is less than ideal for the insolvency of property insurance company in end of 2010.In this thesis, fuzzy comprehensive evaluation method, with the corresponding factor as early-warning indicators, with the data of the 21 property insurance companies during 6-year period of 2004-2009 (some indicators only of 5 years) as the value of early-warning factors, taking the monitoring indicators of the China Insurance Regulatory Commission(CIRC) as the warning threshold; then get early-warning value of each indicator by comparing every company’s data with warning threshold; finally, get the warning result of every year, every company from indicators of the weighted average. The results are consistent with the reality.Comparing the above three methods, the deviation analysis and principal component analysis results are not very satisfactory, the reason is not the methods and indicators themselves, but the two methods completely based on quantitative indicators, while domestic insurance is not in high quality of statistical data, incomplete data problems are more prominent, moreover, the statistical standards and accounting system frequent change, further affecting the quality of statistical data. So, at present, purely quantitative indicators and methods for insurance company risk warning are in absence of sufficient base of data. Nevertheless, fuzzy comprehensive evaluation is proved to be more effective, because it was adapted to the current data quality and quantity of the domestic status quo. The method is determined by expert survey index weight, quantitative analysis and qualitative analyses are combined after technical processing to make up for deficiencies on the quality and quantity. Therefore, fuzzy comprehensive evaluation method can serve as a preferred method to set up property insurance company risk warning system.Finally, in order to play an effective early-warning systems and access to continuous improvement, this thesis makes policy recommendations for establishment of early-warning system; principles, framework are described in detail, and the effective operation of the system and the gradual improvement of the required supporting environment are analyzed.The innovation of this paper may as follows:First of all, it may be the first in China to design a system of early-warning indicators and set the appropriate warning threshold adapt to the current level of management and supervision. This article set in the design of early warning indicators and thresholds, the extensive reference to the domestic and international results of theoretical research and practical application, but more important is to adapt to data availability, in close connection with domestic property insurance industry practice, improving early warning indicator system and valve Value of effectiveness. Second, this paper dealt with related data, for the first time take into account the results of the practitioner’s subjective perception and objective financial statement data, as much as possible to avoid the problems of theory standing far away from practice. Appropriate by experts subjective survey data, its effective treatment, to obtain the corresponding results (weight), combined with customer financial data, for dealing with the data, the risk for later provide an effective basis for early warning. Third, with the use of a variety of empirical techniques and panel data, the thesis establishes an early-warning model against the risks of property insurers. Fourth, with the first published data of actual situation of the property insurance companies’ solvency as a reference, this thesis makes test with practice.更多还原