经济繁荣期商业银行风险预警研究

【作者】 杨延青；

【导师】 傅志华；

【作者基本信息】 财政部财政科学研究所 ， 财政学， 2010， 博士

【摘要】 2007年下半年以来,世界金融市场经历着20世纪30年代大萧条以来最为严重的危机,对于全球金融市场的发展和金融监管产生重大影响。监管当局认识到,宏观金融风险实质上是微观金融风险不断积聚并最终爆发的结果,要从根本上防范和控制宏观金融风险,必须重视微观风险的预警与防范。为达到这一目标,监管当局需要采用金融预警工具来增强监管的预见性和资源分配的合理性。本文以我国商业银行的风险状况为因变量,以财务比率指标为自变量,采用定性分析法和数理统计法构建我国商业银行风险预警模型。通过研究达到以下几个主要目的：一是建立银行风险预警目标的概念；二是区分出高风险和低风险银行；三是设计和构造风险预警指标体系,选取对银行风险具有显著解释作用的关键指标；四是建立商业银行风险的预警模型,为提高监管能力提供参考框架。为解决以上问题,本文论述和设定了预警目标。只有确定了预警目标,才能确定预警方法和预警指标,各类方法和预警指标围绕预警目标展开。银行最终风险是破产,因此许多研究将破产概率作为预警的研究目标。但实践中,银行破产很少发生,研究者转而将重点放在银行风险特征的识别上,及早识别出这些特征,也能够起到风险预警的作用。本文确定三个目标,即银行脆弱性的方向转变、监管评级降级的识别、银行未来降级概率,分别对应不同的预警方法。采用扩散指数法、合成指数法、百分位排序及Logistic四种方法。银行脆弱性的方向转变使用扩散指数法,对应先行指标；监管评级降级的识别采用合成指数法和百分位排序,使用脆弱指标；银行未来降级概率采用Logistic模型,采用脆弱指标。扩散指数法和合成指数法通过计算预警结果,与实际评级结果进行对比,从而确定各类模型参数,建立完整的预警模型。Logistic模型以样本银行数据计算银行未来降级概率,检验脆弱指标的解释能力。同时,由于预警本身存在一类错误和二类错误,因此需要对预警方法的准确度进行检验。在检验方法上,本文选取了效率指数、噪音信号比等方法,对预警结果进行检验,并验证各类参数。本文的研究结论主要有以下几点：第一,经济繁荣期的风险更为隐蔽和易被忽视,该时期银行风险预警目的在于发现银行的主要风险点,以预警结果为依据提早采取行动。在经济衰退期,银行风险在政府的扩张政策中渐渐消去,其风险隐蔽性显然已不同于经济繁荣期。第二,要充分借鉴各国银行风险预警理论和实践。由于不同的金融发展史和金融市场结构,各国对于监管当局采用了不同的监管方式,技术水平和应用能力参差不齐。各国根据自身不同情况,开发银行风险预警系统。在国内没有经验可供借鉴的情况下,需要充分了解和总结银行风险预警先进国家的经验,然后结合我国现有非现场监管发展情况,确定预警目标和方法。第三,要有明确的预警目标及与之对应的各类预警要素。研究发现,目前我国银行风险预警理论和实践还较少,对于预警的目标及采用的方法、指标之间的逻辑关系尚缺乏研究,要预警银行风险,就需要建立预警目标和各要素之间清晰的关系。预警目标决定预警方法和预警指标,也决定预警系统的实施方法和流程,但同时预警方法、预警指标和实践也对预警目标产生约束。预警系统是一个循环反馈过程。在系统长期运行过程中,预警目标、预警方法和预警指标是否相互适应将反复评价。第四,要对预警结果进行检验和修正。预警是技术,也是艺术,是对两类错误的平衡。对于不同错误率的均衡,取决于对每种错误的容忍程度。无论采取何种预警方法,选取的指标并不非一成不变,需要进行检验和修正,在预警长期实践中不断优化。本文的创新之处在于以下几点：一是对预警的宏观经济环境进行了界定,明确提出经济繁荣期银行风险预警方法和指标体系,使银行风险预警的目标性更强。二是提出了预警目标的概念,指出预警的具体目标多样,需要对应不同的预警方法和指标,建立起预警目标和预警方法的逻辑关系。三是对于所选取的指标的预警意义进行了说明,解释了选取的角度,并运用统计方法对这些指标的具体运用结果进行检验。四是运用四种预警方法,对目前国内银行风险情况进行测算,并对测算结果进行了检验,对方法和指标进行针对性评价。更多还原

【Abstract】 Since the second half of 2007, the world financial markets are suffering from the worst crisis since the Great Depression in 1930s,bringing significant impact on the global financial markets and financial regulation. Regulatory authorities recognize that macro-financial risk is essentially an accumulation process from micro-financial risks,and, ultimately, will result to the outbreak。For prevention and control of financial risks,it must be to pay attention to early-warning and prevention of micro-risks. To achieve this goal, the regulatory authorities need to adopt an early warning tool to enhance the monitoring of predictability and rationality of resource allocation. In this paper, the author uses the qualitative analysis and mathematical statistics to construct commercial bank risk early warning model while setting the risks of commercial banks to be dependent variable and financial ratios to be independent variables. Through research it will achieve the following main objectives:One is to establish the concept of bank early-warning target, and the second is to distinguish high-risk and low risk banks.Third,the author designs and constructs risk system of early-warning indicator as well as significant explanatory variable key indicators.Fourth,the author establishes an early warning model of commercial bank risk, and provides a reference framework for strengthening supervision.In order to solve the problem mentioned above, this paper discusses and sets the objectives of early warning. Only after set early-warning goals, early-warning tools and indicators can be determined.And the early-warning methods and indicators are set around the objectives. Bank ’s final risk is bankruptcy, so many of the research objectives take it as a target. But in practice bank failures are rare.Thus, the researchers turn to focus on the identification of bank risk characteristics, because the early identification of these features also can play a role in the bank risk-warning. This paper identifies three objectives, namely, the direction change of bank fragility, downgrade of supervisory rating, the probability of bank downgrade in future, corresponding to the different early-warning methods by four methods such as diffusion index, composite index, percentile ranking, and Logistic model.The direction change of bank fragility uses diffusion index, corresponding to leading Indicators; the downgrade of supervisory rating employs synthetic index and percentile ranking, corresponding to the use of vulnerability indicators;The probability of bank downgrade in future makes use of Logistic model, corresponding to the indicators of vulnerability. By calculating the warning results,diffusion index and the composite index can be compared with the actual ratings, which determine the various model parameters for the establishment of a complete early warning model.The Logistic model calculates the probability of bank downgrade in future by using the sample bank data to test the explanatory power of vulnerability indicators.At the same time, there are TypeⅠerror and TypeⅡerror in bank risk early-warning, so it is need to test the accuracy of early-warning method. In the test method, the author selects efficiency-index, the noise-signal and other methods to test the results of early warning, and correct all kinds of parameters.This paper makes the following main points.First, the risk of economic prosperity is more hidden and neglected. In this period, the bank risk early-warning system is to find the major risk points, to take action based on results-based early-warning. During the period of economic recession, the bank risk is gradually eliminated by the impact of the government expansion policy, and is clearly different from the economic boom. Second, we must learn from countries which have rich theory and practice in banking risk warning field. As the different financial history and financial market structure, national regulatory authorities tend to adopt different approaches as varying skill levels and application methods. According to their different situations, these countries develop their own bank risk early-warning system. There is no experience to draw on in the domestic context, the regulatory authority needs to fully understand and summarize the situation in advanced countries about risk of early-warning, and then combined with our existing off-site supervision of development, determine the goals and methods of early-warning. Third, there must be clear objectives and corresponding various eary-warning elements.Study found that the current risk early-warning of bank in theory and practice is still quite weak, for the early-warning objectives’ logical relationships with the approach, and the indicators are still not so clear. It is need to establish the relationships between the early-warning objectives and various elements.Early-warning objectives determine early-warning methods and indicators, and also the implementation approach and process, but on the other hand, early-warning methods, indicators and practices also are the constraints on the early-warning objectives.The early-warning system is a feedback loop process. In long-running process,whether early-warning objectives are compatible with methods and indicators will be repeatedly assessed. Fourth, it should be tested and corrected to the results of the early-warning. The early-warning, to be technology, as well as to be art,has to balance two types of errors.The balance of different errors depends on the degree of tolerance for each error. No matter what kind of early warning methods,the selected indicators are not the final result, they need to be inspected and corrected in the continuous optimization of long-term practice of early-warning.The innovations of this paper are as following:Firstly, the author defines the macroeconomic environment on the early-warning and clearly puts forward the view about early-warning methods and indicators in economic-prosperity period so that it is stronger for bank’s target of risk-warning. Secondly, the author proposes the concept of early-warning objectives, and points out that there are different early-warning objectives and various early-warning objectives need to correspond to different methods and indicators.Thus, the author establishes the logic between the objectives and methods of early-warning. Third, the selected indicators for early warning of significance are described, explained from the view of there meanings for early-warning, and the author uses statistical methods to test the concrete results from these methods and indicators. Fourth, the author uses these four kinds of early-warning methods measures current situation of domestic banks’ risk, and tests the calculated results, and comments on the evaluation methods and indicators.更多还原