Testing Distributions of Stochastically Generated Yield Curves Gary G Venter AFIR Seminar September 2003.

Slides:



Advertisements
Similar presentations
Value-at-Risk: A Risk Estimating Tool for Management
Advertisements

Arvid Kjellberg- Jakub Lawik - Juan Mojica - Xiaodong Xu.
Modeling of Variance and Volatility Swaps for Financial Markets with Stochastic Volatility Anatoliy Swishchuk Department of Mathematics & Statistics, York.
By Thomas S. Y. Ho And Sang Bin Lee May 2005
VAR METHODS. VAR  Portfolio theory: risk should be measure at the level of the portfolio  not single asset  Financial risk management before 1990 was.
Chapter 21 Value at Risk Options, Futures, and Other Derivatives, 8th Edition, Copyright © John C. Hull 2012.
Cox Model With Intermitten and Error-Prone Covariate Observation Yury Gubman PhD thesis in Statistics Supervisors: Prof. David Zucker, Prof. Orly Manor.
MGT 821/ECON 873 Volatility Smiles & Extension of Models
CAS 1999 Dynamic Financial Analysis Seminar Chicago, Illinois July 19, 1999 Calibrating Stochastic Models for DFA John M. Mulvey - Princeton University.
An Introduction to Stochastic Reserve Analysis Gerald Kirschner, FCAS, MAAA Deloitte Consulting Casualty Loss Reserve Seminar September 2004.
CHAPTER 13 Measurement of Interest-Rate Risk for ALM What is in this Chapter? INTRODUCTION RATE-SHIFT SCENARIOS SIMULATION METHODS.
CF-3 Bank Hapoalim Jun-2001 Zvi Wiener Computational Finance.
Modeling of Economic Series Coordinated with Interest Rate Scenarios Research Sponsored by the Casualty Actuarial Society and the Society of Actuaries.
FRM Zvi Wiener Following P. Jorion, Financial Risk Manager Handbook Financial Risk Management.
Copyright K.Cuthbertson, D. Nitzsche 1 FINANCIAL ENGINEERING: DERIVATIVES AND RISK MANAGEMENT (J. Wiley, 2001) K. Cuthbertson and D. Nitzsche Lecture VaR:
J. K. Dietrich - FBE Fall, 2005 Term Structure: Tests and Models Week 7 -- October 5, 2005.
Introduction to Regression Analysis, Chapter 13,
Chapter 8 Mean-Reverting Processes and Term Structure Modeling.
JUMP DIFFUSION MODELS Karina Mignone Option Pricing under Jump Diffusion.
Financial Risk Management of Insurance Enterprises Stochastic Interest Rate Models.
Lecture 11 Implementation Issues – Part 2. Monte Carlo Simulation An alternative approach to valuing embedded options is simulation Underlying model “simulates”
Statistical Methods For Engineers ChE 477 (UO Lab) Larry Baxter & Stan Harding Brigham Young University.
Inference for regression - Simple linear regression
Lecture 7: Simulations.
Risk Management and Financial Institutions 2e, Chapter 13, Copyright © John C. Hull 2009 Chapter 13 Market Risk VaR: Model- Building Approach 1.
Ewa Lukasik - Jakub Lawik - Juan Mojica - Xiaodong Xu.
Advanced Risk Management I Lecture 6 Non-linear portfolios.
Financial Risk Management of Insurance Enterprises
Alternative Measures of Risk. The Optimal Risk Measure Desirable Properties for Risk Measure A risk measure maps the whole distribution of one dollar.
Integrating Reserve Risk Models into Economic Capital Models Stuart White, Corporate Actuary Casualty Loss Reserve Seminar, Washington D.C September.
BPS - 3rd Ed. Chapter 211 Inference for Regression.
CH12- WIENER PROCESSES AND ITÔ'S LEMMA
HJM Models.
© 2011 Neil D. Pearson A Simulation Implementation of the Hull- White Model Neil D. Pearson.
Chapter 13 Wiener Processes and Itô’s Lemma
Two Approaches to Calculating Correlated Reserve Indications Across Multiple Lines of Business Gerald Kirschner Classic Solutions Casualty Loss Reserve.
Options, Futures, and Other Derivatives, 5th edition © 2002 by John C. Hull 23.1 Interest Rate Derivatives: Models of the Short Rate Chapter 23.
The Effective Duration of Property- Liability Insurance Liabilities with Stochastic Interest Rates Stephen P. D’Arcy, FCAS, Ph.D. Richard W. Gorvett, FCAS,
Interest Rate Derivatives: Model of the Short Rate Chapter 30 1 Options, Futures, and Other Derivatives, 7th Edition, Copyright © John C. Hull 2008.
Financial Risk Management of Insurance Enterprises Financial Scenario Generators.
Chapter 12 Modeling the Yield Curve Dynamics FIXED-INCOME SECURITIES.
Chapter 5: Regression Analysis Part 1: Simple Linear Regression.
Actuarial Science Meets Financial Economics Buhlmann’s classifications of actuaries Actuaries of the first kind - Life Deterministic calculations Actuaries.
Modeling of Economic Series Coordinated with Interest Rate Scenarios Research Sponsored by the Casualty Actuarial Society and the Society of Actuaries.
Fixed Income Analysis Week 4 Measuring Price Risk
CIA Annual Meeting LOOKING BACK…focused on the future.
Lotter Actuarial Partners 1 Pricing and Managing Derivative Risk Risk Measurement and Modeling Howard Zail, Partner AVW
Chapter 24 Interest Rate Models.
Chapter 30 Interest Rate Derivatives: Model of the Short Rate
Options, Futures, and Other Derivatives, 4th edition © 1999 by John C. Hull 14.1 Value at Risk Chapter 14.
Development of Asset Models: Calibration Issues Chris Madsen, ASA, CFA, MAAA American Re-Insurance Company CAS DFA Forum, Chicago July 19th-20th, 1999.
Applications of Stochastic Processes in Asset Price Modeling Preetam D’Souza.
BPS - 5th Ed. Chapter 231 Inference for Regression.
Chapter 13 Wiener Processes and Itô’s Lemma 1. Stochastic Processes Describes the way in which a variable such as a stock price, exchange rate or interest.
Stats Methods at IC Lecture 3: Regression.
Financial Risk Management of Insurance Enterprises
Oliver Schulte Machine Learning 726
Financial Risk Management of Insurance Enterprises
Actuarial Science Meets Financial Economics
3.1 Examples of Demand Functions
The Term Structure of Interest Rates
Financial Risk Management of Insurance Enterprises
CHAPTER 29: Multiple Regression*
Probability & Statistics Probability Theory Mathematical Probability Models Event Relationships Distributions of Random Variables Continuous Random.
Baosheng Yuan and Kan Chen
Basic Training for Statistical Process Control
Basic Training for Statistical Process Control
Chapter 14 Wiener Processes and Itô’s Lemma
Financial Risk Management of Insurance Enterprises
Presentation transcript:

Testing Distributions of Stochastically Generated Yield Curves Gary G Venter AFIR Seminar September 2003

Guy Carpenter 2 Advantages of Stochastic Generators Deterministic scenarios allow checking risk against specific outcomes Deterministic scenarios allow checking risk against specific outcomes Stochastic generators add dimension of probability of scenarios Stochastic generators add dimension of probability of scenarios Can incorporate full range of reasonably possible outcomes Can incorporate full range of reasonably possible outcomes Each scenario can be a time series of outcomes Each scenario can be a time series of outcomes

Guy Carpenter 3 Testing for Potential Problems of Stochastic Generators Model could miss possible scenarios Model could miss possible scenarios Model could overweight some unlikely scenarios and underweight others – giving unrealistic distribution of results Model could overweight some unlikely scenarios and underweight others – giving unrealistic distribution of results Traditional tests look at time series properties of individual scenarios – like autocorrelations, shapes of curves compared to historical, correlation of short and long term rates and their comparative volatility, and mean reversion Traditional tests look at time series properties of individual scenarios – like autocorrelations, shapes of curves compared to historical, correlation of short and long term rates and their comparative volatility, and mean reversion Options pricing models test distributions across scenarios by their impacts on option prices Options pricing models test distributions across scenarios by their impacts on option prices For insurer risk models, we propose testing generators by comparing distributions of yield curves against historical For insurer risk models, we propose testing generators by comparing distributions of yield curves against historical Look for aspects of historical distributions that do not change too much over time Look for aspects of historical distributions that do not change too much over time

Some Models of the Yield Curve ( Then we’ll look at testing)

Guy Carpenter 5 Example Short-Term Rate Models Usually defined using Brownian motion z t. After time t, z t is normal with mean zero and variance t. Usually defined using Brownian motion z t. After time t, z t is normal with mean zero and variance t. Cox, Ingersoll, Ross (CIR): Cox, Ingersoll, Ross (CIR): dr = a(b - r)dt + sr 1/2 dz In discrete form for a short period: r t – r t–1 = a(b – r t–1 ) + sr t –1 1/2  CIR change in interest rate has two components: CIR change in interest rate has two components: – A trend which is mean reverting to b, i.e., is negative if r>b and positive if r b and positive if r<b  Speed of mean reversion given by a – A random component proportional to r 1/2, so variance rts 2 in time t

Guy Carpenter 6 Adding Effects to CIR Mean that is reverted to can be stochastic: Mean that is reverted to can be stochastic: d b = j(q - b)dt + wb 1/2 dz 1 This postulates same dynamics for reverting mean as for r This postulates same dynamics for reverting mean as for r Volatility can be stochastic as well: Volatility can be stochastic as well: d ln s 2 = c(p - ln s 2 )dt + vdz 2 Here Brownian motion in log Here Brownian motion in log Power on r in dz term might not be ½ : dr = a(b - r)dt + sr q dz Power on r in dz term might not be ½ : dr = a(b - r)dt + sr q dz CIR with these two added factors fit by Andersen and Lund, working paper 214, Northwestern University Department of Finance, who also estimate the power of r (1/2 for CIR). CIR with these two added factors fit by Andersen and Lund, working paper 214, Northwestern University Department of Finance, who also estimate the power of r (1/2 for CIR).

Guy Carpenter 7 Fitting Stochastic Generators If you can integrate out to resulting observed periods you can fit by MLE If you can integrate out to resulting observed periods you can fit by MLE – CIR distribution of r t+T given r t is non-central chi-sq. – f(r t+T |r t ) = ce -u-v (v/u) q/2 I q (2(uv) 1/2 ), where – c = 2as -2 /(1-e -aT ), q=-1+2abs -2, u=cr t e -aT, v=cr t+T I q is modified Bessel function of the first kind, order q I q is modified Bessel function of the first kind, order q – I q (2z)=  k=0  z 2k+q /[k!(q+k)!], where factorial off integers is defined by the gamma function Can use this for mle estimates of a, b, and s Can use this for mle estimates of a, b, and s

Guy Carpenter 8 Fitting Stochastic Generators If cannot integrate distribution, some other methods used: If cannot integrate distribution, some other methods used: – Quasi-likelihood – Generalized method of moments (GMM)  E[(3/x) ln x] is a generalized moment, for example  Or anything else that you can take an expected value of  Need to decide which moments to match

Guy Carpenter 9 Which Moments to Match? Title of paper developing efficient method of moments (EMM) Title of paper developing efficient method of moments (EMM) Suggests finding the best fitting time-series model to the time-series data, called the auxiliary model Suggests finding the best fitting time-series model to the time-series data, called the auxiliary model Scores (partial derivates of log-likelihood of auxiliary model) are zero for the data at the MLE parameters Scores (partial derivates of log-likelihood of auxiliary model) are zero for the data at the MLE parameters EMM considers these scores, with the fitted parameters of the auxiliary model fixed, to be the generalized moments, and seeks the parameters of the stochastic model that when used to simulate data, gives data with zero scores EMM considers these scores, with the fitted parameters of the auxiliary model fixed, to be the generalized moments, and seeks the parameters of the stochastic model that when used to simulate data, gives data with zero scores Actually minimizes distance from zero Actually minimizes distance from zero

Guy Carpenter 10 Andersen-Lund Results Power on r in r-equation volatility somewhat above ½ Power on r in r-equation volatility somewhat above ½ Stochastic volatility and stochastic mean reversion are statistically significant, and so are needed to capture dynamics of short-term rate Stochastic volatility and stochastic mean reversion are statistically significant, and so are needed to capture dynamics of short-term rate Used US data from 1950’s through 1990’s Used US data from 1950’s through 1990’s

Guy Carpenter 11 Getting Yield Curves from Short Rate Dynamics P(T) is price now of a bond paying €1 at time T P(T) is price now of a bond paying €1 at time T This is risk-adjusted expected value of €1 discounted continuously over all paths: This is risk-adjusted expected value of €1 discounted continuously over all paths: P(T) = E * [exp(-  r t dt)] P(T) = E * [exp(-  r t dt)] Risk adjustment is to add something to the trend terms of the generating processes Risk adjustment is to add something to the trend terms of the generating processes The added element is called the market price of risk for the process The added element is called the market price of risk for the process

Guy Carpenter 12 Testing Generated Yield Curves Want distributions to be reasonable in comparison to history Want distributions to be reasonable in comparison to history Distributions of yield curves can be measured by looking at distributions of the various yield spreads Distributions of yield curves can be measured by looking at distributions of the various yield spreads Yield spread distributions differ depending on the short-term rate: spreads compacted when short rates are high Yield spread distributions differ depending on the short-term rate: spreads compacted when short rates are high Look at conditional distributions of spreads given short-term rate Look at conditional distributions of spreads given short-term rate

Now for Testing ( Proposed Distributional Test)

Guy Carpenter 14 Three Month Rate and 10 – 3 Year Spread Clear inverse relationship Mathematical form changes Five periods selected

Guy Carpenter 15 Ten – Three Year Spreads vs Short Rate Slope constant but intercept changes each period

Guy Carpenter 16 Possible Tests of Generated Curves Individual scenarios Individual scenarios – Could look at different time points simulated and see if slope and spread around line is consistent with historical pattern – For longer projections – 10 years + – expect some shift – For 20 year + projections a flatter line would be expected with greater spread, as in combining periods Looking across scenarios at a single time Looking across scenarios at a single time – Observing points over time can be viewed as taking samples from the conditional distribution of spreads given short rate – Alternative scenarios can be considered as providing draws from the same conditional distribution – Distribution of spreads at a time point could reasonably be expected to have the recent inverse relationship to the short rate – same slope and spread

Guy Carpenter 17 Five - Year to Three - Year Spreads

Guy Carpenter 18 Spreads in Generated Scenarios 5 – 3 spreads from Andersen-Lund with a selected market-price of risk Slope ok, spread too narrow Same problem for CIR – even worse in fact

Guy Carpenter 19 Add Stochastic Market Price of Risk Better match on spread

Can also test distribution around the line ( Shape of distribution – not just spread)

Guy Carpenter 21 Distributions Around Trend Line Distributions Around Trend Line Percentiles plotted against t with 33 df Variable Fixed Historical Variable looks more like data But fitted distribution misses in tails for all cases Test only partially successful

Guy Carpenter 22 Summary Treasury yield scenarios should be arbitrage-free, and be consistent with the history of both dynamics of interest rates and distributions of yield curves Treasury yield scenarios should be arbitrage-free, and be consistent with the history of both dynamics of interest rates and distributions of yield curves Short-rate dynamics can be tested by fitting models Short-rate dynamics can be tested by fitting models Yield curve dynamics can be tested with individual generate series Yield curve dynamics can be tested with individual generate series Yield curve distributions tested by conditional distributions of yield spreads given short rate Yield curve distributions tested by conditional distributions of yield spreads given short rate