1、New Game for Financial Market 金融市场新游戏 & New Game Field for Risk Management 风险管理新环境,Yimin Yang (杨一民) Vice President & Sr. Manager Risk Analytics,2,介绍 PNC,It used to be a bank, but now (曾经叫做银行,但现在是),PNC Bank 15th largest US Bank $70 Billion Asset,BlackRock One of the best & fastest growing Investment
2、Company on Wall Street $300 Billion Asset Under Mgnt,PFPC Brokerage & Processing,Capital Market Investment,PNC Financial Service Group (PNC 金融服务集团),3,介绍Risk Analytics,Risk Analytics (Centralized Risk Mngt Team),4,Now, back to our topics,言归正传,5,两个热门话题 Two Current Hot Topics,The Game(新游戏) Credit Risk,
3、 more specifically (信用风险) Credit Derivatives (信用衍生品) The Game Field (新环境) New Regulatory Environment, more specifically (新监管要求)Basel II (巴塞尔协议),6,Credit Risk & Credit Derivatives,信用风险与信用衍生产品,7,What is Credit Derivative 什么是信用衍生品,What is Credit Risk (什么是信用风险) The risk that a company is unable/unwillin
4、g to make promised payment (不能按时支付) What are Credit Derivatives (什么是信用衍生品) Financial products that price & transfer credit risk (用于转移信用风险的金融产品) Credit Default Swap (CDS), Asset Swap Cash Flow Collateral Debt Obligation (CDO), Synthetic CDO Nth-to-default baskets, Single Tranche CDO, CDO Option Stand
5、ardized Index Tranches, CDO Squared,8,Example: Credit Default Swap (CDS) 例一,Most of Credit Derivatives are CDS (70%) (市场主要产品) It is similar to buy/sell insurance for default(类似破产保险),Protection Buyer 买保方,Protection Seller 卖保方,Underlying Asset (GE Bond, for example) 持有风险资产 比如GE债券,Regular % Fee 定期付费(保险
6、费) (Libor + Spread),Payment IF the underlying asset defaults 当GE发生破产时,付损失费,9,Example: (Collateralized Debt Obligation( CDO)例二,CDO is Todays market favorite(市场宠儿) A portfolio is tranched to absorb losses. Investors can invest in different tranches,Equity Tranche $10mm,Mazzanine $15mm,Senior Tranche $
7、35mm,Super Senior $40mm,$100mm Asset (A portfolio of Corp. bond,e.g.) 1亿风险资产,10,Example: CDX.HY 例三,Popular Dow Jones CDX.HY (Dow Jones 推出的CDS指数之一) 100 liquid BB/B-rated CDS ENTITIES, equally weighted. (一百家高风险公司组成) Reset every 6 months 5 year is the most common maturity Over 20 industries Standardize
8、d 6 Loss Tranches (六个标准断) 0-3% Equity, 3%-7%, 7%-10%, 10%-15%, 15%-30%, 30%-100% Investors buy & trade these tranches, just like buying stocks or any other financial products (象一般投资一样买卖) HY stands for High-Yield (there is also CDX.IG for Investment Grade with 125 names) (Dow Jones还有其他指数),11,Example:
9、 CDX.HY 例三,Each Tranche is traded at a different price (风险不同,定价不同) Equity Tranche (0%-3%) has the highest risk, therefore, highest return Super Senior Tranche (30%-100%) offers the lowest return (spread),12,Global Market(全球市场),13,Major Players(主要运动队),14,Key Drivers(主要原因),Hedge Portfolio Risk(风险抵消) T
10、ake Exposure in Credit Risk (投资信用风险) Enhance Profitability (增加回报) Improve Risk Management & Reduce Regulatory Capital (风险与资本管理) Increase Asset & Liability Management Capability (资产管理) Credit is now a TRADABLE asset (已成交易资产),15,Challenges To Financial Practitioners (主要的挑战),Understanding of Default Ri
11、sk (深入了解违约风险) Default Probability Modeling(违约率模型) Default Risk Pricing for various markets(不同市场违约风险定价) Portfolio Management (组合管理) Correlation & Portfolio Loss Distribution(相关性与损失分别) Risk Management & Capital Allocation(风险管理与资本配置) Cross-Market Risk Hedging Techniques (跨越不同市场的风险分化手段),16,Single name D
12、efault Modeling (单个公司违约率模型),Structural Model (结构性模型) Asset Value-Based Type: A firm will default if its asset value falls below certain threshold CreditMetrics KMV Reduced-Form Type Model (简化性模型) Intensity-based first-passage Market Price (Risk Neutral Default Probability)-Based (市场价格性模型) CDS & Bond
13、 Spread Statistical/Actuarial (统计性模型) Moodys RiskCalc,17,Structural Model (结构性模型),Asset Value-Based Model CreditMetrics Asset follows a normal distribution with mean and standard deviation . Debt is assumed to be deterministic. The default probability is thenKMV Asset market value (not book value) f
14、ollows a (lognormal) diffusion process (Brownian process ),18,Structural Model (结构性模型),KMV This can be easily solvedTo estimate the market value, one uses Mertons view: Equity is a Call option on Asset with Debt as the strike,19,Structural Model (结构性模型),KMV By Itos Lemma, we obtainAnother condition
15、is added to solve the Call price (Equity) KMV is very successful (it was sold to Moodys last year for $220mm),20,Reduced-Form Model (简化性模型),First Passage Default Model: Default occurs when the Asset hits the Debt for the 1st time. That is, the default time and default probabilityIntensity-based mode
16、l: Conditional on any realization of the default intensity , the default process follows a Poisson arrival. If let be the information available up to time t, then,21,Reduced-Form Model (简化性模型),Intensity-based model Mean reverting with jumpsCox-Ingersoll-Ross Process (Affine process),22,Market Price
17、Model (市场价格性模型),Financial products such as Bond, CDS, Asset Swap, etc. contain substantial amount of information about (Risk Neutral) default probability of a company Various Theoretical & Empirical models are trying to estimate Implied Default Probability by a market price (or vice versa) CreditGra
18、de, CreditEdge Spread = Risk Neutral Default Prob. + Liquidity(Size) Premium= Function of Empirical & Market Price of Risk + Liquidity(Size) Premium,23,Statistical Model (统计模型),Moodys RiskCalc Based on Moodys History A logistic model using 7-10 financial transformed ratios,Profitability,Capital Stru
19、cture,Liquidity,24,Default Correlation Modeling (违约率相关度),Default Correlation Is Crucial To CDO,25,Default Correlation Modeling (违约率相关度),Correlation of default events of two firms isHere and are default probabilities for the two firms Asset Value-Based Model: through asset correlationTo obtain asset
20、correlation, it is very popular to use factor model,26,Default Correlation Modeling (违约率相关度),Factor model example: ( and are standard normal dist).The asset correlation = , the joint default probIntensity-Based Model: through correlation of intensities Intensity model example: the default times are
21、given by,27,Default Correlation Modeling (违约率相关度),Here is a Poisson Process with intensity The default correlation is thenCopula is becoming one of the most important methods for modeling correlation What is a Copula? Suppose that and are marginal distributions for and . Let be the joint distributio
22、n, then there is a function (copula) C such that,28,Default Correlation Modeling (违约率相关度),Using the copula for default time, the default correlation is,29,Default Correlation Modeling (违约率相关度),In the pricing of CDX, correlation is the key The model is assumed to be homogenous, and the correlation is
23、 assumed to be constant between all companies Market observed price for each tranche is inputted to the model to solve an implied correlation. Depending on the way of estimating the accumulative losses for each tranche, one obtain Implied Correlation: attempt to mimic the success of Implied Volatili
24、ty for Option pricing Base Correlation: is getting the markets attention,30,The New Game Field,金融业面临一个新的环境,31,Major Reasons(产生的主因),Significant Financial Losses & Failures (许多重大损失) Complex Financial Products, Operation & Organization (复杂的金融产品,运作与机构) Investors Concerns (投资者的顾虑) Regulatory Disciplines(
25、监管政策),32,Basel II(巴塞尔协议),Bank for International Settlement is the Central Bank of Central Banks(国际清算银行是中央银行的中央银行) Its Basel Committee Sets Up a Framework/Standard for Risk Capital Requirement for All Financial Institutions BASEL II Accord (其BASEL委员会为所有金融机构制定了资本适足率框架与标准) Many European Banks & Banks i
26、n Developed Countries have decided to Follow Advanced Approach (大多数发达国家银行将采用最高级标准) Major US Banks Will also Follow Advanced Approach(美国的大型银行会采用最高级标准) Non Basel-compliant banks will be disadvantaged(不遵守者处于不利地位),33,Key Requirements (巴塞尔协议主旨),Risk-Based Capital for Financial Risks(风险为依据的资本要求) Credit Ri
27、sk (1988 Accord) (信用风险) Market Risk (1996 Amendment) (市场风险) Operational Risk (New Basel Accord) (运营风险) Three Pillars(三大支柱) Minimum Capital Requirement (最低资本要求) Supervisory Review Process (监管与审阅程序) Market Disciplines & Risk Disclosures (市场纪律与风险公开),34,Key Requirements (巴塞尔协议主旨),Capital Ratio(资本比率)Type
28、 of Capital(三类资本) Tier 1 Core(核心资本) Tier 2 Supplementary(次级资本) Tier 3 For market risk only(只对部分市场风险可用) On-Balance Sheet & Off-Balance Sheet Risk Charges(表内及表外业务风险折算),35,Basel Approaches (巴塞尔协议方法),Goal(目的) To quantify all risks using appropriate models, with recognition of risk diversification & miti
29、gation techniques (用模型来量化一切风险,同时接纳风险分散及转移技术) Credit Risk(信用风险) Standardized Approach(标准方法) Standard Risk Weight External Rating Based Foundation Internal Rating Based Approach (基本内部评级方法) Internal Rating System Use of Probability of Default (PD) Advanced Internal Rating Based Approach (高级内部评级方法) Inte
30、rnal Rating System Use of PD, Loss Given Default (LGD) and Exposure At Default (EAD),36,Basel Approaches (巴塞尔协议方法),Market Risk(市场风险) Standardized Approach (标准方法) Market Charges (Risk Weights) for Interest Rate Risk Maturity Zones Netting within bands Market Charges for Equity Risk Net Position Marke
31、t Charges for Currency Risk Use of maximum of total long or short positions Market Charges for Commodity Netting but not complete offset Market Charges for Option Simplified Approach Intermediate Approach Internal models Approach,37,Basel Approaches (巴塞尔协议方法),Market Risk(市场风险) Internal Models Approa
32、ch (内部模型高级方法) Relies on Internal Risk Management System Qualitative Requirements Independent Risk Control Unit Back-testing Senior Management Involvement Integration of risk models with day-to-day management Use of Limits Stress Testing Compliance,38,Basel Approaches (巴塞尔协议方法),Operational Risk(运营风险)
33、 Basic Indicator Approach (基本指标方法) A single indicator as a proxy for firms exposure to operational risk Gross Income Standardized Approach (标准方法) Distinguishes between business lines and business volume Standardized loss factors (beta) for business lines Internal Measurement Approach (内部度量高级方法) Bott
34、om-Up Approach Key Issue: Internal Loss Data Requirement (5 Year of History) Probability of Loss Event Loss Severity Exposure Indicator,39,Basel Implementation (巴塞尔协议落实),Deadline(截止日) Jan. 1, 2007 (2007年一月一日) Or a later date (或许更晚) Cost(费用) $100mm for each large US Bank (平均一亿美元) Human Resources (人力资
35、源) Risk Managers & Internal Controls Risk Research & Analytics Information Systems (信息系统) Data Support Hardware & Software,40,Math Tools For Op. Risk (运营风险的数学方法),A Significant Challenge(巨大的挑战) In Particular when less/no data available Methods Considered(试用中的方法) Frequency Models Poisson,Binomial, Neg
36、ative Binomial,Geometric, Hypergeometric, etc. Correlation is a problem? Severity Models Heavy tail distributions Lognormal,Pareto, Weibull, etc Extreme Value Theory GEV Distribution,41,Math Tools For Op. Risk (运营风险的数学方法),Value-at-Risk (VaR) Tail distribution Expected Shortfall Stochastic Process Markov Process,Renewal Process,etc. Bayesian Techniques Linear Causal Models Regression Models Factor Models Discriminant Analysis Risk Scores Non-linear Models Neural Networks/Bayesian Networks Fuzzy Logic,
