全球企业及投资银行业务部风险管理组合.ppt_第1页
全球企业及投资银行业务部风险管理组合.ppt_第2页
全球企业及投资银行业务部风险管理组合.ppt_第3页
全球企业及投资银行业务部风险管理组合.ppt_第4页
全球企业及投资银行业务部风险管理组合.ppt_第5页
已阅读5页,还剩18页未读 继续免费阅读

下载本文档

版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领

文档简介

macroeconomic-based approach to credit loss forecasting 基于宏观经济的信用损失预测方法 gcib-risk management portfolio 全球企业及投资银行业务部风险管理组合,risk rating migration, macroeconomic indicators & applications to portfolio credit risk forecasting 信用评级变动、宏观经济指标以及在 组合信用风险预测中的应用,credit portfolio analytics 信贷组合分析组,credit portfolio analytics group mission 信贷组合分析组的使命 to provide management with an independent, top-down, macroeconomic-driven asset quality forecast range, including a baseline forecast based on most-likely assumption, that is accurate, comprehensive, nimble, and actionable. 向管理层提供以宏观经济为基础、独立、自上而下的资产质量预测,包括根据“最可几”假设作出的、准确、全面、可提供行动依据的基准线预测。 assist in management of earnings volatility associated with risk in the credit portfolio. 协助管理与信用组合风险有关的盈利波动。 to mitigate significant potential losses due to event risk, and 减少因事件风险造成的重大潜在损失;以及 to design and implement proactive strategies for portfolio management. 设计和实施主动性的组合管理战略。 old environmentone forecast opinion 旧的环境预测角度的看法 line-of-business/risk management provided asset quality forecast to senior management 由业务线/风险管理部门向高级管理层提供资产质量预测 “bottoms up” approachaccount level data ”自下而上“的预测方法账户级数据 - how many standard deviations from expected? 距离期望值有多少个标准差? no contrarian view 不采取反向视角 not explicitly linked to future economic environment 没有与未来的经济环境建立明确的关联 one outcome provided to senior management 只向高级管理层提供一个结果 - conservative? reasonable? or a stretch? 保守?合理?或吃紧?,overview 概述,overview 概述,new environmentone forecast opinion 新的环境预测角度的看法 provide ceo, cfo, and chief risk officer with an alternative to line-of-business forecasts 除业务线预测之外,向首席执行官、财务总监和风险总监提供另一个选择 forecasting based on a single view of the economy 根据对经济的单一看法进行预测 ability to run numerous simulation and sensitivity analyses 能够进行大量的模拟和敏感性分析 synchronized loan and interest income forecasts with corporate treasury. 贷款和利息收入预测与集团司库部同步。 objective目标 the main objective is to produce a forecast of the asset quality mix as well as credit loss associated with the commercial loans, taking into account the following factors: 主要目标是在考虑以下因素的前提下,提供与个人信用卡有关的信用损失预测: underlying macroeconomic indicators相关的宏观经济指标 the banking industry credit environment银行业的信用环境 migration of banks loans along the risk rating spectrum (including non- performing status and charge-offs) 银行贷款沿信用评级档自低向高的变动(包括“不良”状态和撇账) models produced for commercial products 开发用于商业产品的模型 macroeconomic-based loss forecast models have been developed (for combined bank of america and fleet legacy portfolios): 已经开发基于宏观经济的损失预测模型(针对美国银行和 fleet 两家公司合并前的组合): legacy gcib portfoliorisk management portfolio 合并前的 gcib 组合风险管理组合 legacy middle market bankingcommercial banking regions 合并前的中间市场金融业务各商业银行业务区 legacy middle market bankingmid capital corporation 合并前的中间市场金融业务中间资本公司 total commercial excl. small business (not in production) 除小型企业外的所有商业银行业务(尚未投入使用),main components of the loss model 损失模型的主要组成部分,step 1 : establishing the future credit environment 第1步:确立远期信用环境 using forecasted economic data, estimate the future credit cycle index (cci). 使用预测的经济数据,估算远期的信用周期指数(cci) step 2 : applying the credit environment to the banks portfolio 第2步:对银行的信用组合应用信用环境 derive future migration matrices based on the forecasted cci. 根据预测的cci导出远期的信用评级变动矩阵 step 3 : forecasting the banks credit quality 第3步:预测银行的信用质量 multiply current rating distribution by forecasted transition matrices to arrive at the future rating distribution. 用预测的变动矩阵乘以当前的评级分布,得出远期的评级分布 add new business, balance changes, maturities, and other funds flows; consistent with corporate treasury loan forecast. 增加新的业务、余额变化、期限和其他资金流;与集团司库部的贷款预测相一致 step 4 : forecast the credit loss 第4步:预测信用损失 for “defaulted” loans, apply “severity of loss” assumptions (or loss given default). 对“违约”贷款,应用“损失严重度”假设(或违约损失) produce a projection of “potential credit loss” for future time-periods. 得出远期“潜在信用损失”的预测值,loss forecast process 损失预测流程,step 1 : establishing the future credit environment 第1步:确立远期信用环境 step 2 : applying the credit environment to the banks portfolio 第2步:对银行的信用组合应用信用环境 step 3 : forecasting the banks credit quality 第3步:预测银行的信用质量 step 4 : forecast the credit loss 第4步:预测信用损失,credit cycle index (cci) 信用周期指数 (cci),cci indicates the credit state of the financial market as a whole cci反映金融市场的总体信用状况 the index is designed to be: 指数的设计原则 “positive”, for good times, indicating lower levels of downgrading and defaults, and a higher upgrading probability, than average 在景气好的时期,指数为“正”,表明与平均水平相比,信用等级降低和违约处于较低水平,信用等级升高的可能性较高 “negative”, for bad times, implying higher levels of downgrading and defaults, and a lower upgrading probability, than average 在景气不好的时期,指数为“负”,表明与平均水平相比,信用等级降低和违约处于较高水平,信用等级升高的可能性较低,what is the cci? 什么是 cci?,how do we construct it? 如何构造这个指数?,a simple way to construct the cci is to calculate the default probabilities of all credit rated bonds. 计算所有获得信用评级债券的违约概率,是构造cci的一个简单方法。 highly rated bonds were excluded since they have very low default probabilities. 高评级债券的违约概率非常低,因此不包括在内。 a normal distribution transformation of us speculative default probability, sdp (rated equal to or lower than moodys ba rating) is used. 采用按正态分布形式表示的美国投机级债券的违约概率,简称sdp(等于或低于穆迪的ba评级)。,how do we model it? 如何模拟这个指数?,as a result of their relationship to sdp, and after testing economic data, the following variables are included in the cci model: 考虑到这个指数与sdp的关系,在测试经济数据后,cci模型中包含以下变量: the gdp growth gdp增长 s&p 500 returns 标准普尔500指数的回报率 corporate bond spread (baa-aaa) 企业债券利差(baa-aaa) treasury bond spread (10yr. tr. bond minus 3 month tr. bill) 国债利差(10年期国债利率减去3月期短期国债利率) the initial regression line was corrected for the serial correlation of the error term and a first-order autoregressive process, denoted ar (1), was included in the model. 得出最初的回归线后,按误差项的序列相关性进行修整,并在模型中包括一阶自回归过程,记为ar(1),credit cycle index 信用周期指数,model 模型,a comparison between the actual value of the us speculative grade default rate and the model fitted value was illustrated 美国投机级债券违约率实际值与模型拟合值的对比,the predicted value of the us speculative grade default rate was converted into a credit cycle index value. the bar chart graph is an illustrative representation of this credit cycle index. 美国投机级债券违约率的预测值转换为信用周期指数值。条状图代表这个信用周期指数,示意性,for illustrative purposes only 仅用于示意目的,投机级债券违约概率(sdp)ar(1)过程,cci与美国投机级债券违约概率(sdp),credit cycle index cci & macroeconomic indicators 信用周期指数ccci与宏观经济指标,gdp growth below 3% indicates a negative credit environment gdp增长低于3%,表示信用环境不佳 gdp is the primary driver in the model gdp是这个模型中的主要决定因素,equity markets are a leading indicator of credit conditions 股票市场是信用状况的先行指标,lower spread between bbb and aaa corporate yields indicate stronger credit environments bbb与aaa级企业债券的利差越低,信用环境越好,steeper treasury yield curve indicates stronger credit environment 国债收益率曲线越陡峭,说明信用环境越好,for illustrative purposes only 仅用于示意目的,gdp增长,baa-aaa利差,标准普尔500指数回报率,国债利差(10年期减3月期),loss forecast process 损失预测流程,step 1 : establishing the future credit environment 第1步:确立远期信用环境 step 2 : applying the credit environment to the banks portfolio 第2步:对银行的信用组合应用信用环境 step 3 : forecasting the banks credit quality 第3步:预测银行的信用质量 step 4 : forecast the credit loss 第4步:预测信用损失,modeling credit rating migration 模拟信用评级变动,default behavior of commercial portfolios can be modeled using credit rating migration; a migration matrix also known as a transition matrix. 可使用信用评级变动法来模拟商业组合的违约行为;变动矩阵也称为变化矩阵。 credit migration analysis tries to answer questions like 信用变动分析试图回答以下这类问题 - “what is the probability that 以下事件的概率是多大 . aa rated company gets downgraded to a over the next year? aa级公司明年降级为a? . bbb rated company defaults during the next 5 years? bbb级公司未来5年内违约? - answer: all migration information is contained in a transition matrix 答案: 所有变动信息都报告在变化矩阵内 average annual moodys migration matrix 穆迪的平均年度变动矩阵,85.14% of ba rated loans remain ba, while 1.28% will migrate to the default state ba级贷款当中,有85.14%依然是 ba级,有1.28%将变动至违约状态,for illustrative purposes only 仅用于示意目的,起点状态,目标状态,违约,违约,economic index based migration 基于经济指数的变动,modifying transition and default probabilities along the modeled economic scenarios根据模拟的经济情境,修正变动概率和违约概率 - an index which reflects economic performance is constructed so that a value of “0” represents a neutral economic performance. (corresponding to mean asset return=0) 构造一个反映经济表现的指数,经济表现为中性时这个指数的值为“0”。(对应资产回报率均值=0) - an average transition matrix, corresponding to the “neutral economy” scenario was constructed. 与所构造的“中性经济”情境相对应的是平均变化矩阵。 - relative to this average, in 相对于这个平均值 - good scenarios: the value of the index is positive (greater than zero) there are more upgrades and fewer defaults. 好情境: 指数为正值(大于零);信用升级增加,违约减少。 - bad scenarios: the value of the index is negative (less than zero) there are more downgrades and more defaults. 坏情境: 指数为负值(小于零);信用降级增加,违约增加 - any change in economic index values reflects changes in the mean asset return along the scenarios. 经济指数值的任何变化都反映资产回报率情境分布均值的变化。,what we want to do here 我们要达到什么目的,traditionally, an average of historical data has been used to construct the migration matrix. 传统上,使用历史数据的平均值来构造信用评级变动矩阵。 the procedure outlined herein incorporates an important step beyond this. 这里介绍的方法包括一个传统方法里没有的重要步骤。 the likelihood of migration from one credit rating to another is defined as a function of the state of the economy, as reflected by the cci. 从一个信用评级进入变为另一个信用评级的似然率被定义为cci所代表的经济状况的函数。,approach for calibrating the banks avg. transition matrix 银行平均迁移矩阵的校正方法,step 1 第1步,full-period bond matrix全期间债券矩阵,1q81,truncated bond matrix 截断的债券矩阵,2q06,1q99,step 2 第2步,step 3 第3步,truncated bank matrix 截断的银行矩阵,step 4 第4步,full-period bank matrix 全期间银行矩阵,2q06,1q99,2q06,2q06,1q81,performance across multiple cycles 多个周期的表现 . internal risk rating migration data over multiple business/credit cycles do not exist 不存在跨越多个商业/信用周期的内部风险评价迁移数据 . the following approach was used to establish a bank-specific average risk rating transition matrix: 采用以下方法来建立针对具体银行的平均信用评级迁移矩阵 1. using available internal bank data over the period 1q99-2q06, the average migration matrix was calculated. this matrix was considered to be the truncated average migration matrix. 使用1999年1季度至2006年2季度这个期间内现有的银行内部数据,计算平均迁移矩阵。这个矩 阵被视作截断的平均迁移矩阵 2. using s&p bond data, an average bond matrix for each quarter was calculated for the 1q99-2q06 and 1q81-2q06 time periods. 使用标准普尔的债券数据,计算1999年1季度至2006年2季度和1981年1季度至2006年2季度这 两个期间内每个季度的平均债券矩阵 3. the calculated average bond matrices were then mapped to the corresponding bank risk ratings. 然后,在计算得出的平均债券矩阵与相应的银行风险评级之间建立对应关系 4. the banks truncated average matrix was then extrapolated to 1q81 based on the relationship between the full and truncated bond matrices to produce a full-period average bank matrix. 然后,根据全期间矩阵与截断债券矩阵之间的关系,将银行的截断平均矩阵外推 至1981年1季度,获得银行的全期间平均矩阵,banks average migration matrix 银行的平均变动矩阵,the bank adjusted” average transition matrix, representing performance across multiple cycles, was presented here. 此表为美国银行的“调整后”的平均变动矩阵,反映在多个周期内的表现。,for illustrative purposes only 仅用于示意目的,美国银行的平均变动矩阵(组合管理),债券,信用评级,schematic: index based migration 简要流程:基于指数的变动,process allows time dependent migration matrices to be represented by an index 这个流程允许使用一个指数来代表具有时间依赖性的变动矩阵,index has an intuitive representation - “0” = average migration - “+” = more upgrades than average - “-” = more downgrades than average,+,=,index has an intuitive representation 指数的直观表述 - “0” = average migration 平均变动 - “+” = more upgrades than average 升级超过平均值 - “-” = more downgrades than average 降级超过平均值,平均变动矩阵,第2季度,预测指数,第n季度,based on the historical data, the bank (gcib) average migration matrix was calculated and mapped to s&p migration data. 根据历史数据,计算出美国银行(gcib)的平均变动矩阵,然后与标准普尔的变动数据建立对应关系,using forecasted cci as calculated based on internal economic projections, the average migration matrix is then adjusted and the future periods rr migration was produced 使用根据内部经济预测计算出的 cci 预测值,对平均变动矩阵进行调整, 得出远期的信用评级变动数据,future migration matrices and default rates based on the cci 基于cci 的远期变动矩阵和违约率,migration & default rates 变动率及违约率,tilted to reflect “recessionary” outcome 对角数据序列 反映出“衰退型经济”,tilted to reflect “growth economy” outcome 对角数据序列 反映出 “增长型经济”,impact loss forecast process alternatives 影响损失预测流程的选择,forecast view of the credit environment 对信用环境的预测看法,economy-adjusted migration & default rates 经济调整后的变动率与违约率,for illustrative purposes only 仅用于示意目的,loss forecast process损失预测流程,step 1 : establishing the future credit environment 第1步:确立远期信用环境 step 2 : applying the credit environment to the banks portfolio 第2步:对银行的信用组合应用信用环境 step 3 : forecasting the banks credit quality 第3步:预测银行的信用质量 step 4 : forecast the credit loss 第4步:预测信用损失,projecting future exposure distribution by risk rating 预测风险暴露风险评级的远期分布,future periods dollar exposure 远期风险暴露额 multiply current dollar risk-rating distribution by the forecasted economy-adjusted transition matrix to arrive at future dollar risk-rating distribution. 用当期风险暴露分布乘以预测的经济调整后变化矩阵,得出远期的风险暴露风险评级分布 using the conditioned transition matrix, along with the beginning-of the-period data, the end-of the-period exposure distribution of each risk rating was then forecasted. 使用条件变化矩阵,以期初数据为初值,可预测每个风险评级的期末风险暴露分布。 adjust exposure level for fund flows 根据资金流调整风险暴露水平 the end-of the-period exposures are then adjusted for net flows: 根据净流量调整期末风险暴露: exiting business (pay-offs, charge-offs) 退出业务(全额还款、撇账) balance changes (draw-downs, increases, partial payments) 余额变化(动用额度、余额增加、部分还款) sales/purchases 出售/收购 new business 新业务,loss forecast process 损失预测流程,step 1 : establishing the future credit environment 第1步:确立远期信用环境 step 2 : applying the credit environment to the banks portfolio 第2步:对银行的信用组合应用信用环境 step 3 : forecasting the banks credit quality 第3步:预测银行的信用质量 step 4 : forecast the credit loss 第4步:预测信用损失,net charge-off projection 预测净撇账,based on the historical relationship between net-charge-offs (nco) and non-performing loans (npl), future ncos are forecasted. 根据净撇账(nco)与不良贷款(npl)之间的历史关系,预测nco。 as a proxy for loss given default (lgd), the ratio of ncos as a percentage of npls is used. 使用nco占npl的比例来代替违约损失(lgd)。 a model for severity of loss assumption 损失严重度模型的一个假设 motivation:假设的出发点: recovery rates are sensitive to the state of the economy and differ across the portfolios. 回收率对经济状况具有敏感性,并且组合间存在差别 recovery rates tend to be higher during the better cycle of the economy and lower during the worse cycle of the economy. 在经济周期的景气期,回收率往往较高;在经济周期的不景气期,回收率往往较低 to produce a model for lgd, the historical relationships between the realized lgd ratios and macroeconomic variables were investigated and a predicted value of lgd for subsequent quarters was produced. 为了构造lgd模型,对已实现lgd比例和宏观经济变量之间的历史关系进行了研究,得出了后续季度的lgd预测值 projection of net charge-offs 净撇账的预测 the appropriate period projected lgd along with the projected non-performing balances are used to forecast the future net charge-offs. 相应期间的的预测lgd和预测不良贷款余额用于预测远期的净撇账。,other applications of the loss forecast 损失预测的其他应用,stress-testing scenarios 压力测试情境 produce the projection of losses under different scenarios of the economy: 得出不同经济情境下的损失预测 - most-likely (base) 最可几(基准) - high gdp growth of the economy (bull) 高gdp增长的经济(繁荣) - low gdp growth of the economy (bear) 低gdp增长的经济(萧条) - any alternative scenario of the economy 任何替代的经济情境 - high energy prices (and inflation) 能源价格(和通货膨胀)高企 - consumer-lead recession 消费主导型衰退 - etc. 其他 - “what-

温馨提示

  • 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
  • 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
  • 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
  • 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
  • 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
  • 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
  • 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。

评论

0/150

提交评论