我国商业银行住房抵押贷款风险研究

发布时间：2018-06-15 12:37

本文选题：住房抵押贷款 + 信用风险　；参考：《中央财经大学》2015年博士论文

【摘要】：近年来,随着住房抵押贷款(或称“房贷”)所占贷款比重的逐步上升,相应的金融风险也在增加。尤其是房地产行业调控政策的频出,又带来住房抵押贷款新的风险。特别是2014年以来,全国房价普遍下跌,开发商“跑路”事件频发,对商业银行的房贷风险管理提出更高要求。因房贷在我国起步较晚,“风险管理水平落后”已在业界和学术界达成共识,为此,笔者对住房抵押贷款的风险展开了研究,希望能对该领域做出一点微薄的贡献,对我国房贷风险管理水平的提升产生一定的价值。笔者根据在商业银行住房抵押贷款中心多年的工作经验以及对房贷的研究和思考,认为在微观层面,房贷的风险涉及信用风险和操作风险;在宏观层面,则涉及市场风险和“面临房地产泡沫破裂的风险”(简称“泡沫风险”,正文将专门界定)。本文对上述四种风险从理论到实证的分析也由此展开。本文先界定了住房抵押贷款风险以及被细分成的四种风险,具体阐述了这四种风险的区别和联系。接着,从微观和宏观层面阐述了这四种风险的理论基础。在微观层面,房贷信用风险和操作风险的理论基础为信息经济学理论;在宏观层面,市场风险和泡沫风险的理论基础分别为LTV理论和明斯基金融不稳定理论。然后,本文针对这四种风险的特点进行了系统的理论分析。通过博弈论对房贷信用风险的分析结果表明:在商业银行和借款人的动态博弈中,银行对借款人违约概率的判断是整个博弈的关键所在;用委托代理理论分析操作风险的结果表明:在房贷中心,帕累托最优的激励是让对房贷操作风险有贡献的岗位员工(代理人)承担全部风险;通过LTV理论对房价变化的分析,阐明了房价变化和借款者主动违约的关系;用明斯基金融不稳定理论结合中国的现状分析得知:我国目前房价下跌的情形已经符合明斯基意义上的泡沫破裂情形。实证分析的目的,在于解决目前我国房贷风险管理中存在的突出问题,本文通过问卷调查和对现实情况的综合分析,找出商业银行住房抵押贷款四种风险管理中的突出问题,然后针对问题提出解决方案。问卷调研涉及了工行、农行、中行、建行、交行、中信、浦发、平安、招行、华夏、邮储银行的住房金融部门和30位工作五年以上的资深客户经理。通过调研反映出的问题有:信用风险方面,商业银行对该风险的识别缺乏全面性和系统性,目前主要采用的评分卡定量分析法存在主观因素较强的弊端,并且不普及;在市场风险方面,商业银行对该风险的防控普遍处于被动应对状态,市场风险对房贷客户经理影响最大的是“额度收紧风险”。在操作风险方面,人员是操作风险最重要的因素,但商业银行并未量化分析过哪些岗位员工对操作风险的贡献是显著性的,因此也无法预警操作风险,更无法通过科学的奖惩制度解决“委托代理问题”。在泡沫风险方面,商业银行因未曾经历过该风险,对其研究较少。另外,在问卷调查中,对房贷资产证券化转移房贷风险提得较多,该项工程在国内的推行并不顺利,因此本文将单独对其进行分析。根据问题的梳理,信用风险的分析思路为:在对信用风险进一步细分的基础上建立指标体系(解决识别片面的问题),然后根据该指标体系定量分析(解决量化模型主观性强并且不普及的问题),最后提出防范措施。具体内容是:将房贷信用风险分为三类,即:被迫违约风险、理性违约风险、恶意违约风险,然后选取能体现这三种风险的十个指标(包括职业、学历、婚姻状况、家庭收入、年龄等等)建立信用风险指标体系,接着通过分类统计法得到违约率在各指标上的分布。但是,分类统计法的局限性在于其结果是边缘概率,无法让决策人在风险防范时抓住主要矛盾,并且不能联系相关指标来评价风险,需要其他方法来互补。针对这个问题,本文进一步通过r语言编程,采用数据挖掘中的决策树法,再次做实证得知:就整体而言,在众多的分析因素中首付款比率,职业,年龄,家庭收入是对住房抵押贷款违约率影响最大的因素。通过“决策树”的路径,本文还找到了违约概率最大和最小的人群特征。市场风险的分析包括两部分:通过wilson模型做压力测试,分析在三种情景下房价利率与违约率的定量关系;设计额度风险缓释型理财产品(应对“商业银行额度收紧风险”)操作风险因杂乱繁琐的特点,用墨较多。本文先从房贷涉及的所有岗位中分析出15个操作风险诱因,并找到各种诱因的衡量指标即“差错率”以及相关数据,然后通过多重共线性分析,将风险诱因简化为11个。为进一步筛选,本文在matlab环境下运用竞争性神经网络模型(som)将连续的差错率数据离散化,然后用粗糙集约简的实证分析方法筛选出6个关键风险指标,即kdi(粗糙集约简只适用于离散数据,故连续数据需先离散化)。然后,本文通过ahp和变异系数相结合的组合赋权实证分析出了与这6个关键风险指标kdi相对应岗位的权重分布。接着通过bp前馈神经网络模型在matlab环境下设计出了房贷操作风险的预警机制;并根据对操作风险评估时的组合赋权结果以及委托代理的分析结论定量设计出了3个岗位的薪酬激励机制;并提出与保险公司合作开发住房抵押贷款的保险产品的定量方案。对于泡沫风险,通过理论分析得知:我国的房价不会陡然下跌,只会逐步下跌,故可分阶段进行分析。本文基于wilson模型的假设情景、分析过程和结果,将房价下跌过程分为初期、中期和末期。其中,初期是cltv(当前贷款房价比)小于等于100%时,即房贷余额小于等于房价,此时住房抵押贷款的信用风险最显著,中期是指cltv的值处于100%-120%之间,即贷款余额大于房价,但房地产还未崩溃,此时住房抵押贷款的市场风险最显著,末期是指cltv大于120%,即房地产已经崩溃,开发商大量倒闭,此时信用风险和操作风险最显著。然后本文结合我国的恩格尔系数,以cltv(当前贷款房价比)和cdsr(当前收入偿债比率)为指标,设计出了我国房价下跌时的风险分期预警机制。这在一定程度上填补了很多商业银行对泡沫风险研究的空白。为更好的通过资产证券化实现房贷风险的转移,本文建议在银行间债券市场上发行住房抵押贷款资产证券化产品,并引入海外投资者来购买。具体思路为:为达到住房抵押贷款风险转移的效果和目的,资产证券化产品的投资期限需较长,而国内投资者的资产配置久期大部分较短,故考虑引进海外投资者,而通过本文的分析可知银行间债券市场在国内的各种金融市场中较容易被海外投资者接受,因此需根据该市场的特点和海外投资者的偏好特征,以及国内房贷的特殊性来设计房贷资产证券化产品的推进方案。本文设计了此方案,内容包括:(1)房贷资产证券化产品的基本交易结构;(2)如何引入海外投资者(3)如何进行住房抵押贷款资产证券化产品的市场建设。对交易结构,本文通过流程图的方式做了阐述;在海外投资者引进方面,本文根据其偏好设计了“三步走”的策略;在市场建设方面,本文从市场的风险隔离模式建设、信息披露体系建设、增信体系建设三个方面进行了分阶段的设计。本文将其归纳为“一引三建”。其中,在风险隔离上,本文设计了如何从“表内模式”向“表外模式”过渡;在信息披露层面,本文设计了如何从“一般信息披露”走向“全面信息披露”;在增信体系建设方面,如何从传统的担保以及优先劣后分级到引进crma(风险缓释合约)和crmw(风险缓释凭证)。本文可能的创新点包括:第一,首次将住房抵押贷款的信用、市场、操作和泡沫风险作为一个相互联系的整体进行全面的分析研究,以往关于住房抵押贷款的论文一般偏重于写一种风险,较多的研究其信用风险,少部分研究其市场风险,研究其操作风险和泡沫风险的几乎没有。第二,在研究住房抵押贷款信用风险时,使用r语言编程,采用数据挖掘方法中的决策树法来实证分析评估住房抵押贷款的信用风险,并根据评估结果提出相应的风险控制措施。第三,在研究住房抵押贷市场风险时,为防范“额度收紧风险”,本文设计出风险缓释型理财产品进行应对。第四,在研究住房抵押贷款操作风险时,在MATLAB环境下通过竞争性神经网络模型将“差错率”数据进行离散化,再用粗糙集约简的方法通过实证分析从15个房贷操作风险诱因中筛选出6个关键风险指标(KDI);然后通过AHP和变异系数相结合的组合赋权法实证分析出与这6个关键风险指标所对应岗位的风险权重;在此基础上运用BP前馈神经网络模型得到了房贷操作风险的预警机制,并设计出基于委托代理理论分析下的薪酬体制,同时设计出对房贷操作风险进行保险的方案。第五,在研究住房抵押贷款的泡沫风险时,在运用Wilson模型对我国住房抵押贷款风险进行压力测试的基础上,通过对当前贷款房价比CLTV和当前偿债率CDSR的定量和定性分析,得到了在房价下跌过程中住房抵押贷款三种风险爆发的顺序及过程。并针对此过程,结合我国的恩格尔系数,以CLTV和CDSR为指标,设计出了我国房价下跌时的风险预警机制。第六,在考虑运用资产证券化法来转移房贷风险时,本文通过分析我国债券市场的现实情况,创造性的设计出了“一引三建”的方案体系。“一引三建”包括如何逐步引入海外投资者,如何分阶段进行住房抵押贷款资产证券化的风险隔离模式建设,即在“表内模式”下如何向“表外模式”发展;如何分阶段进行信息披露体系建设;如何不断充实增信体系,即如何逐步从传统的抵押、担保、保证以及优先级、中间级、次级分层到引进CRMA(信用风险缓释合约)和CRMW(信用风险缓释凭证)。
[Abstract]:In recent years, with the gradual increase of the proportion of loans in housing mortgage loans (or "housing loans"), the corresponding financial risks are also increasing. In particular, the frequency of the regulation policy of the real estate industry brings new risks to housing mortgage loans. Especially since 2014, the price of house prices in the whole country has been generally falling, and the events of the developers "running" are frequent, and the business is on business. The bank's mortgage risk management put forward higher requirements. Because of the late start of the housing loan in China, "the risk management level is backward" has reached a consensus between the industry and the academia. For this reason, the author has studied the risk of housing mortgage loan, hoping to make a little contribution to this field and improve the level of the risk management of the housing loan in our country. At the micro level, the risk of housing loan involves credit risk and operational risk, and at the macro level, it involves market risk and "the risk of real estate bubble rupture" ("bubble wind" for short). The text will be specifically defined in the text. In this paper, the analysis of the above four risks from the theoretical to the empirical analysis is also carried out. This paper first defines the risk of housing mortgage loans and the four kinds of risks that are subdivided, concretely expounds the differences and relations between the four risks. Then, the theoretical basis of these four risks is expounded from the micro and macro level. At the micro level, the theoretical basis of the credit risk and operational risk of the housing loan is the theory of information economics. At the macro level, the theoretical basis of the market risk and the bubble risk is LTV theory and Minsky's financial instability theory respectively. Then, this paper makes a systematic theoretical analysis on the characteristics of these four kinds of risks. The result of the risk analysis shows that in the dynamic game of commercial banks and borrowers, the judgment of the default probability of the borrower is the key to the whole game. The result of the operation risk analysis by the principal-agent theory shows that in the housing loan center, Pareto's best incentive is to make the job employees who have contributed to the risk of the mortgage operation. According to the analysis of the change of house prices by LTV theory, the relationship between the change of house prices and the active default of the borrowers is clarified. In order to solve the outstanding problems in the current risk management of housing loan in China, through the questionnaire survey and the comprehensive analysis of the actual situation, this paper finds out the outstanding problems in the four risk management of the commercial bank housing mortgage, and then puts forward the solutions to the problems. The housing finance department of the postal savings bank and 30 senior customer managers who work for more than five years have shown that the credit risk is lack of comprehensiveness and systematicness in the recognition of the risk, and the main shortcomings of the subjective factors are present in the quantitative analysis of the score card. And it is not universal; in the market risk, commercial banks are generally in a passive response to the risk, and the market risk has the greatest impact on the mortgage customer manager. In the operation risk, the personnel are the most important factor in the operation risk, but the commercial bank has not quantified the analysis of the post employees. The contribution of operational risk is significant, so it can not warn the operation risk, but can not solve the "principal-agent problem" through the scientific reward and punishment system. In the case of bubble risk, the commercial bank has not experienced the risk and has less research on it. In addition, in the questionnaire survey, the risk of housing loan asset securitization is more than that of the mortgage loan. The implementation of the project in China is not smooth, so this paper will analyze it alone. According to the problem, the analysis of credit risk is to establish an index system on the basis of further subdivision of credit risk (solve the problem of one side), and then take root according to the quantitative analysis of the index system (to solve the subjectivity of quantitative model. " The specific content is that the credit risk of housing loan is divided into three categories, namely: forced default risk, rational default risk, malicious default risk, and then select ten indicators (including career, study calendar, marital status, family income, age and so on) which can reflect the three risks. However, the limitation of the classification statistical method is that the result is the marginal probability, which can not make the decision-makers seize the main contradiction in the risk prevention, and can not contact the relevant indicators to evaluate the risk and need other methods to complement each other. One step through the R language programming, using the decision tree method in data mining, and again to do empirical study that, as a whole, the first payment ratio, occupation, age and family income are the most influential factors in the default rate of housing mortgage. The characteristics of small crowd. The analysis of market risk includes two parts: using the Wilson model to do the pressure test, analyzing the quantitative relationship between the rate of house price and the rate of default under the three scenarios; design the operational risk of the sustained release type of the amount risk (to deal with the "commercial bank limit risk") the operation risk is complicated and cumbersome. In all the jobs involved, 15 operational risk factors are analyzed, and the "error rate" and related data are found. Then the risk inducement is simplified to 11 by multiple collinearity analysis. In order to further screen, this paper uses the competitive neural network model (SOM) to make continuous errors in the MATLAB environment. The rate data is discretized, and then 6 key risk indexes are screened out by the empirical analysis method of rough intensive simplex. That is, KDI (Rough simplex is only applicable to discrete data, so the continuous data must be discretized first). Then, this paper analyzes the combination of AHP and variation coefficient combining empowerment with the 6 key risk indicators, KDI. Then, the BP feedforward neural network model is used to design the early warning mechanism of the risk of the mortgage operation under the MATLAB environment, and the compensation incentive mechanism of the 3 posts is designed according to the combined empowerment results of the operational risk assessment and the analysis conclusion of the principal agent; and the cooperative development and development of the insurance company is proposed. A quantitative scheme for the insurance products of a mortgage loan. For the risk of the bubble, the theoretical analysis shows that the house price in China will not fall abruptly and will only fall gradually, so it can be analyzed in stages. Based on the hypothesis of the Wilson model, the process and the results are analyzed and the process of falling house prices is divided into early, medium and late stages. Cltv (the current loan to house ratio) is less than 100%, that is, the mortgage balance is less than equal to the house price, and the credit risk of housing mortgage is most significant at this time. In the medium term, the value of cltv is between 100%-120%, that is, the loan balance is larger than the house price, but the real estate has not collapsed, and the market risk of housing mortgage loan is the most significant at the end of the period is that cltv is greater than 1. 20%, the real estate has collapsed and the developers have closed down a lot. At this time, the credit risk and the operation risk are most significant. Then this paper, based on the Engel coefficient of our country, designs the risk stage early warning mechanism when the price of our country falls under the cltv (current loan and house price ratio) and CDSR (current income debt repayment ratio). This is filled in a certain extent. In order to better transfer the risk of housing loans through asset securitization, this paper proposes to issue mortgage asset securitization products in the inter-bank bond market and introduce overseas investors to purchase. The specific idea is to achieve the effect of the risk transfer of housing mortgage loan and the effect of the transfer of the risk of housing mortgage loan. Objective, the investment period of the asset securitization product is long, and the duration of the domestic investor's asset allocation is most short, so consider introducing overseas investors, and through this analysis we can see that the interbank bond market is easier to be accepted by overseas investors in various financial markets in China. Therefore, the characteristics of the market and overseas should be based on the characteristics of the market. The characteristics of the investor's preference and the particularity of domestic mortgage are designed to design the promotion scheme of the mortgage asset securitization products. This paper designs this scheme, which includes: (1) the basic transaction structure of the mortgage asset securitization products; (2) how to introduce the overseas investors (3) to the market construction of the housing mortgage loan asset securitization products. In terms of the transaction structure, this paper expounds the way through the flow chart; in the introduction of overseas investors, this article designs the "three step" strategy according to its preference; in the market construction, this article has carried out a phased design from the construction of the market risk isolation mode, the construction of information disclosure system, and the construction of the augmented credit system in three aspects. It is summed up as "one cited three construction". Among them, in risk isolation, this paper designs how to transition from "intra table mode" to "out of watch" mode. At the level of information disclosure, this paper designs how to move from "general information disclosure" to "comprehensive information disclosure", and how to guarantee and prioritization in the construction of augmented system. Post classification to the introduction of CRMA (risk sustained release contract) and crmw (risk release voucher). The possible innovation points of this article include: first, a comprehensive analysis of the credit, market, operation and bubble risk of housing mortgages as an interconnected whole is made for the first time. Risk, more research its credit risk, few research its market risk, research its operational risk and the risk of bubble almost. Second, in the study of housing mortgage credit risk, the use of R language programming, the use of data mining method of decision tree to empirical analysis and evaluation of housing mortgage credit risk, and based on the risk of housing mortgage. The evaluation results put forward the corresponding risk control measures. Third, in the study of the risk of housing mortgage loan market, in order to prevent the "quota tightening risk", this paper designs a risk release type financial product to deal with. Fourth, in the study of the operational risk of housing mortgage loan, the "error rate" is adopted by the competitive neural network model under the MATLAB environment. "The data is discretized, and then 6 key risk indicators (KDI) are selected from 15 mortgage operational risk factors through empirical analysis. Then the risk weight of the posts corresponding to these 6 key risk indicators is empirically analyzed by combining the combination empowerment method of AHP and variation coefficient. On this basis, BP is used. The feedforward neural network model gets the early warning mechanism of the risk of the mortgage operation, and designs the compensation system based on the principal-agent theory, and designs the insurance scheme for the risk of the mortgage operation. Fifth, in the study of the risk of housing mortgage loan, the risk of housing mortgage loan in China is carried out by using the Wilson model. On the basis of the pressure test, through the quantitative and qualitative analysis of the current loan price ratio CLTV and the current debt repayment rate CDSR, the order and process of three kinds of risk of housing mortgage loan in the process of housing price fall are obtained. And this process, combining the Engel coefficient of our country and the index of CLTV and CDSR, designs the decline of the house price in our country. Sixth, when considering the use of asset securitization to transfer the risk of housing loan, this paper creatively designs a "one cited three" scheme through the analysis of the reality of the bond market in China. "One cited three" includes how to introduce overseas investors gradually and how to carry out mortgage loan in stages. How to build the risk isolation mode of asset securitization, that is, how to develop the "off balance sheet mode" under the "in table mode", and how to carry out the letter in stages.
【学位授予单位】：中央财经大学
【学位级别】：博士
【学位授予年份】：2015
【分类号】：F832.4

【相似文献】