Study On Mortgage Lending Patterns In China

Introduction

- Objective

- Why study this issue

- organisation of the dissertation

With China's rapid economic development, urbanization advance, driven by domestic demand and improved living standards, the residents’ demand for housing improvement has been increasing. In the foreseeable future, China will have a thriving real estate development space and potential.

Financial support is necessary for real estate development, especially housing mortgage loans. For Commercial Banks in China, this is a huge cake and cheese. The one with best risk-management abilities and tools will be able to gain more market share.

This dissertation take China Merchants Bank as case study on the design, methods, processes, data, development and verification of housing mortgage loans scoring card, propose home mortgage loans scorecard model, and be used effective and widely.

Literature Review – Credit Scoring/Sub-prime crisis

Mortgage contracts/types of mortgages

Credit Scoring technologies are used to control levels of default within american consumer credit. And such technologies have been involved with its own methodological, procedural and temporal risks. And periodic renewal of models and constant reappraisal of methods and procedures are required in accessing the risk. Through testing a diffusion model of techbology adoption for the financial services industry, we find that banking organizations that are more centralized in their organizational structure and those located in the New York Federal Reserve district adpted small business credit scoring before their peers. The former is due to the influence of organizational structure on technology adoption, and the latter is consistent with theories of geography-dependent innovation diffusion. Race is influential to the mortgage lending disicion in USA. However, it is not important in China. Profession and income level have greater effect on the probability of getting the loan. I will develop another critiria to understand the pattern in China based on the study of “Understanding Mortgage Lending Patterns? by Issac F. Megbolugbe.

Background of mortgage lending in China

Strong growth in China's housing markets. The housing mortgage in China is not very old in comparison to others.

Credit Scoring system in China Merchants Bank

Comparison with UK/US credit scoring system just for one bank

招商银行是中国最早研究和应用信用评分模型的商业银行之一。从2004年开始,招行就开始了信用评分模型的研究,2007年开?完?了全国范围内的?房贷款申请评分模型并于2008年投入实际应用。2009年,伴?新巴塞尔??议实施对银行信用风险计?和管?的?求,招商银行全?开?了?房按?贷款﹑汽车消费贷款﹑个人消费贷款等??零售贷款业务的申请评分模型﹑行为评分模型和催收评分模型,并已覆盖95%以上的零售贷款业务。

招行的?房按?贷款申请评分模型采用了欧美商业银行普??采用的信用模型建模方法论,通过收集样本债务人风险特?信?,确定好??客户定义和表现期,根?因??0/1分布的特点进行Logistic回归,以?使用KS统计方法检验模型的预测能力等确定最终模型。

由于所处政治??济?文化环境的??,中国银行业的信用评分体系与欧美银行评分体系在模型设计的具体处?上还存在一些差异。相比西方先进商业银行,中国银行业建立和应用信用评分模型的历??然比较短暂,在数?收集等方?还存在一定缺陷。这些差异和缺陷将在入模??的结果上予以??映。

比如,在模型设计方?,招商银行个人?房贷款申请评分模型使用地区进行划分(segmentation),这是由于中国幅员辽阔,?地区?济?展??平衡,??地区的人?特??个人?房贷款的风险特?也?尽相?。而在欧美,较少有银行个人?房贷款申请评分模型会以地区进行划分。

?如,在样本数?方?,由于缺少正?的验?渠?,中国银行业获得的?房贷款部分申请数??能?够完整和准确。招商银行的?房贷款数?,尤其是人?信?,主???于客户自己填写的资料,并且无法从其他地方,尤其是官方的?信机构得到验?,因此真实性无法完全??。有时借款人?能会虚报收入﹑婚姻状?等??信?以满足获得贷款的政策?求。因此,招商银行在使用这些数?时会比较谨慎。而西方银行会有渠?例如?信局?验?这些信?,因此数?比较?信。

对美国花旗银行与招商银行的信用评分系统加以比较,?以?现两者在评分模型的方法论?逻辑回归?统计检验方?比较一致,但在样本数?方?有以下一些明显的差异:

首先,花旗银行的信用评分系统广泛使用?信局的数?进行建模,其中的个人公告信?和信用历?记录的??性较高。美国有1000多家?信机构,但基本隶属于三家信用报告公?:Experian, TransUnion和Equifax。而中国央行下设的?信中心2008年正?开始??供?信?务,许多贷款申请人尚无信用记录。而且由于?信中心???供电?化的批?数?,因此在信用评级模型开?中也无法使用。

其次,花旗银行的信用评分系统中,?些人?信?,例如???性别等,?能在评分模型中使用,?则会被作为歧视而引起法律诉讼,但是中国没有此类?制。

第三﹑花旗银行的信用评分系统中,将拒?的贷款申请也纳入模型开?样本进行分?,从而??供了新的评价维度。而招商银行对于拒?的贷款并没有录入系统,因此在建模时没有使用任何拒?数?进行模型的校正。

总之,招行已采用国际通行的建模方法建立了包括按?贷款评分模型在内的一系列应用模型,但在模型实际和数?收集方?还存在一些差异和?足。

China Merchants Bank (hereinafter referred to as “CMB?) is one of the first bank engaged in studying and adopting credit scoring model in China since 2004. In 2007, CMB developed its mortgage loan application scoring model, and the model was applied countrywide to all its 700 branches next year. In 2009, with the progress of Basel II Accord implementing in Chinese banking industry, CMB developed its application scoring model, behavior scoring model, collection scoring model of mortgage loan, auto loan, personal consumption loan in succession to measure and manage its risk in a more accurate way. As for now, scoring model systems are used in the bank to cover over 95 percent of loan business in retail banking sector through various process including loan application, marketing, pricing, after sale management, default loan collection, etc.

CMB’s mortgage loan application scoring model uses the same methodology as that in those active European and American banks. By collecting debtor’s risk characteristics information, shaping its default definition and performance period, utilizing logistic regression based on 0/1 distribution of dependent variable, using KS statistical analysis to test the model’s predictive capability, the scoring model was finally established.

Due to the difference in China’s political, economic and cultural environment with western countries, CMB’s credit scoring model also differs in its design details from European and American banks. With the relatively short history in studying and adopting credit scoring model in Chinese banking industry, some shortfalls in date collection was also in existence and will be reflected in the independent risk variables as the result of the credit scoring model.

For example, CMB generated different scoring model for different regional segmentation in its model designing period. The reason relies on the unbalanced economical development and different demographical characteristics and risk characteristics through over the country. The risk characteristics like age, income can have totally different impact to the credit scoring model. In some cities, elder applicant represents high income level and stable economic status, thus the possibility of loan default is low. While in some other cities, younger applicant can get financial support from the whole family, thus means lower risk of this type of applicants. However, European banks usually apply only one consistent model in one country. For some small banks, one consistent model will be applied even in several countries due to the lack of adequate sample date and similar economic and cultural environment.

Another example is that the demographical information in CMB sample data is not so credible as that for Western Banks’. Most of the demographical information in CMB data comes from the application form which is filled out by applicants themselves. Some demographical information, like income, marital status, could not be verified through any official credit agency. In some occasions, applicants may exaggerate their income level or distort their marital status to meet the credit approval requirement of the bank. So the authenticity of the date could not guaranteed, and the bank must use those information cautiously. The relationship manager of the bank always need to assess their customers’ risk level by using the tools bank provided or with their own experience, but the assessment result are usually not quantitative output and can not be used in building the credit scoring models. While for European and American banks, the authoritative credit agencies provided a transparent and credible information environment. Banks can cheaply and conveniently access to the credit agencies for demographical information of applicants, including age, dwelling address, telephone number, period in current employment, household income, other outstanding loan, default record and tax information, and those electric information can be downloaded and integrated into the banks’ database system for the purpose of building the credit scoring model.

To compare the credit scoring model development process in CMB and CITI Bank of USA, (hereinafter referred to as “CITI?), we found the same basic methodology are adopted, both banks using logistic regression to create their mortgage application scoring models, the default definition and performance period are quite close, and KS testing, out-of-sample testing and cross-time testing are all applied to validate the model’s effectiveness. The main discrepancy rest with the date sample date sector and can be sorted as follows:

Firstly, the sample date is easily accessible from credit agencies for Citi bank, and the bank use extensively those date in its model building process. There are more than 1000 credit agencies in USA and mostly of them are subjected to three most authoritative credit agencies: Experian, TransUnion and Equifax. Citi bank receives credit information both directly from Experian and from their outsourcing credit enquiry companies in Georgia and South Dakota. But for Chinese banks, including CMB, the one and only credit information channel is the PBOC(People’s Bank of China) Credit Center established few years ago. This official credit agency started to provide its credit query service to commercial banks only in 2008. Due to its limited history, more than half of the nation’s population are still excluded from outside the system. For many applicants, Credit Center could neither verify their demographical information nor find credit record for them. Furthermore, the credit query service do not provide mass electronic date but only printable date, thus made the integration of query results with banks database an impossible mission.

Secondly, for Citi bank, some demographical information, like race, gender, can’t be used in credit scoring model development due to the restriction of laws and regulations. Otherwise, the discrimination between loan applicants with different races or genders may cause serious consequences such as lawsuit against the bank. But there is not such limitation in both China’s banking regulations and traditional morality, the bank could use such information to discriminate their customers in the credit scoring model if only they have enough predictive ability and business sense.

Thirdly, in Citi bank’s credit scoring model, the information of refusal debtors’ are also deemed as sample date, thus provide a new dimension to analysis the customers’ risk characteristics. The bank may summarize rules from the risk characteristics of those rejected applicants and adopted into the credit approval guidelines for further practice operations. CMB did not reserve any information about rejected applicant in past so that its credit scoring system did not involve any reject inference activity. Recently, CMB has started to collect information about rejected applicant, but the information can’t be used in adjust the model only when they accumulated to a certain amount after about five years.

Generally speaking, CMB has already established a series of credit scoring model to cover its retail banking business including mortgage loan by using universal methodology. However, the date collection is one of its main shortages for the bank to improve the quality of its modelling practice.

Empirical Analysis of Homeowners’ data

Using Cross-sectional Analysis modelling and Probit/Logit modelling

Banking regulation in China

中国?房市场化的改?始于1988年,但直到1995年,中国央行??布了第一部特定的指导法规《商业银行自??房贷款管?暂行办法》,其中规定首付30%,最高贷款期?10年,利率?执行?期固定资产贷款??利率?档优惠。1999年人行?下?《关于鼓励消费贷款的若干??》,将?房贷款与房价款比例从70%??高到80%,?年?将房贷最长期?延长到30年,贷款利率按基准利率?档执行?下调10%。2003年人行《关于进一步加强房地产信贷业务管?的通知》??出一套普通房首付20%,利率按个人?房贷款利率执行,而对购买高档商?房?别墅或第二套以上(?第二套)商?房的借款人,商业银行?以适当??高个人?房贷款首付款比例,并按照中国人民银行公布的?期?档次贷款利率执行,??执行优惠?房利率规定。

近年?,??中国房价的?续快速上涨,对资产价格泡沫的担忧也与日俱增。中国的银行监管部门多次制定政策,对首付比例?按?利率优惠幅度等??出调整。总体而言,对于居民购买首套?房,首付比例?求在20%以上,利率优惠较为宽?;对于购买第二套?房,首付比例?求在30%以上,利率优惠较?。

2010年4月国务院《关于?决??制部分城市房价过快上涨的通知》,对购买首套自?房且套型建筑?积在90平方米以上的家庭,贷款首付款比例?得低于30%;对贷款购买第二套?房的家庭,贷款首付款比例?得低于50%,贷款利率?得低于基准利率的1.1?。??银监会对第二套?房的认定标准进行了比以外更严格的?求。

对于信用评分体系的监管政策,直到2007年中国银监会推动巴塞尔新??议在中国银行业的实施,?有了一些原则性的规定。其中最??的规定?自于《商业银行内部评级体系监管指引》,其中对银行应用信用评级模型??出了以下?求:

略(详?英文部分)

尽管以上针对信用评级体系的监管?求比较原则性,但中国银监会对包括招商银行在内的中国?六大商业银行的风险评级模型始终紧密关注,对模型从数?收集?设计方法到最终的?数结果都进行了审慎而严格的检验评估。

此外,2010年6月,在中国银监会的统一?求和组织下,中国主?商业银行对?房贷款进行了压力测试,结果显示在?度压力的情况下(房价下跌30%),?房按?贷款的?良率上?幅度?常有?。普??的分?认为,这与中国较高的首付比率和与较传统的信用文化有关。

China’s housing market reform started in year 1988, but it was not until 1995 that China’s central bank PBOC issued its first regulation in individual household loan area. This Provisional Code of Housing Loan on Commercial Banks’ Account set the minimum mortgage down payment ratio at 30%, the maximum term of mortgage loan at 10 years, the mortgage interest rate is a preferential rate to the benchmark of fixed assets loan. In 1999, PBOC issued A Number of Views on Encouraging the Development of Consumption Loans, in which the loan to value rate loosened to 80%, the maximum term of mortgage loan to 30 years, and the lowest preferential interest rate can be 10% discount on benchmark. In 2003, PBOC issued Notice on Further Enhance Real Estate Loan Management, in which reaffirmed the 20% minimum down payment, raise the interest rate of mortgage loan to benchmark. The Notice brings forward the new concept of “second set of house?, to which the bank should raise the down payment and interest rate level to the debtor.

In recent years, the skyrocketing real estate prices in China bring worries on assets bubbles in public and accumulated risks to mortgage loans and real estate loan. The central bank and CBRC (China Banking Regulatory Commission) constituted and enacted several regulations to guide the mortgage business with most of the measures hooked on down payment ratio and interest rate. Generally speaking, for purchase of the first set of house, the down payment requirement is at least 20% and the interest rate are relatively flexible; for purchase of the second set of house, the down payment requirement is at least 30% and the interest rate at benchmark. The relatively loosen policy was deems as one of the important drives to boost the house price to an irrational high level now in many Chinese big cities.

To curb housing market speculation, the State Council announced in April, 2010 that: 1. The minimum down payment ratio is lifted to 30% for the first set of house purchase if the area is more than 90 square meters; 2. Second set of house buyers must pay at least 50% of the value for mortgage down payment, and the interest rate are at least 1.1 times of benchmark; 3. Higher down payment ratio and interest rates are required for “three of more set? of house purchase. Analyst believe the mortgage loan demand should be negatively impacted but within a limited range, and the lifted down payment ratio would provide better buffer for housing pricing decline, which would benefit loan quality.

As for the banking regulations specific in the credit scoring system, it was not until 1997 CBRC released several regulatory guidelines successively along with the Basel II Accord implementation progress in Chinese banking industry. One of the most important is the Internal Rating System Regulatory Guideline, in which principle guidelines for building internal rating system are provided for commercial banks. Being a vital part of internal rating system, credit scoring system is also regulated by these guidelines, and the main principles on credit scoring system in the guidelines are as follows:

The burden is on the bank to satisfy its supervisor that a model or procedure has good predictive power and that regulatory capital requirements will not be distorted as a result of its use. The variables that are input to the model must form a reasonable set of predictors. The model must be accurate on average across the range of borrowers or facilities to which the bank is exposed and there must be no known material biases.

The bank must have in place a process for vetting data inputs into a statistical prediction model which includes an assessment of the accuracy, completeness and appropriateness of the data specific to the assignment of an approved rating.

The bank must demonstrate that the data used to build the model are representative of the population of the bank’s actual borrowers or facilities.

When combining model results with human judgement, the judgement must take into account all relevant and material information not considered by the model. The bank must have written guidance describing how human judgement and model results are to be combined.

The bank must have procedures for human review of model-based rating assignments. Such procedures should focus on finding and limiting errors associated with known model weaknesses and must also include credible ongoing efforts to improve the model’s performance.

The bank must have a regular cycle of model validation that includes monitoring of model performance and stability; review of model relationships; and testing of model outputs against outcomes.

The bank should clearly know the basic assumptions of the model and assess the consistency between the assumptions and current economic or market conditions. The bank must demonstrate that economic or market conditions that underlie the data are relevant to current and foreseeable conditions. If current conditions change, the bank should ensure that the model can adapt the changed conditions. If model can not meet the request, the bank must adjust model conservatively.

The above are all regulation items about credit scoring system. We can see those regulation items are mostly principle-based. The process and method of building credit scoring system are in compliance with the guideline on the whole.

Although the banking regulations on credit scoring system are roughly and fundamentally, CBRC has always kept a close watch on the credit scoring model construction in the country’s biggest six commercial banks including CMB. Stringent and strict on-site assessment has been carried out by CBRC throughout the whole process of date collection, model design to parameter results to ensure the appropriate adoption of the models.

In June 2010, CBRC urgently request various state-owned commercial banks, joint-stock commercial banks as well as the city commercial banks with asset size over 500 billion yuan to start self-examination for real estate loans, and conduct stress testings especially in mortgage loan business. The result shows that even under highest stress scenario that the house price drop 30% from present level, the real estate related loans will maintain good quality, and the NPL (non-performing loan) ratio for mortgage loans will increase less than 1% for the whole banking industry. It is believed that the relatively high down payment ratio and the traditional credit culture in China contributed to such a released consolable result.

Comparison of China/US/UK banking regulation and default risk

Mortgage lending in China is safer and more stable than that in US.

Conclusions