Improving the Jarrow-Yildirim Inflation Model

Similar documents
Market Models. Practitioner Course: Interest Rate Models. John Dodson. March 29, 2009

The Mathematics Of Stock Option Valuation - Part Four Deriving The Black-Scholes Model Via Partial Differential Equations

INSTITUTE OF ACTUARIES OF INDIA

7 pages 1. Hull and White Generalized model. Ismail Laachir. March 1, Model Presentation 1

Pricing FX Target Redemption Forward under. Regime Switching Model

Hull-White one factor model Version

(1 + Nominal Yield) = (1 + Real Yield) (1 + Expected Inflation Rate) (1 + Inflation Risk Premium)

Pricing formula for power quanto options with each type of payoffs at maturity

Matematisk statistik Tentamen: kl FMS170/MASM19 Prissättning av Derivattillgångar, 9 hp Lunds tekniska högskola. Solution.

INSTITUTE OF ACTUARIES OF INDIA

DYNAMIC ECONOMETRIC MODELS Vol. 7 Nicolaus Copernicus University Toruń Krzysztof Jajuga Wrocław University of Economics

Models of Default Risk

2. Quantity and price measures in macroeconomic statistics 2.1. Long-run deflation? As typical price indexes, Figure 2-1 depicts the GDP deflator,

1 Purpose of the paper

Equivalent Martingale Measure in Asian Geometric Average Option Pricing

Volatility and Hedging Errors

Foreign Exchange, ADR s and Quanto-Securities

Available online at ScienceDirect

Research Article A General Gaussian Interest Rate Model Consistent with the Current Term Structure

Jarrow-Lando-Turnbull model

Agenda. What is an ESG? GIRO Convention September 2008 Hilton Sorrento Palace

Quanto Options. Uwe Wystup. MathFinance AG Waldems, Germany 19 September 2008

Pricing Vulnerable American Options. April 16, Peter Klein. and. Jun (James) Yang. Simon Fraser University. Burnaby, B.C. V5A 1S6.

A pricing model for the Guaranteed Lifelong Withdrawal Benefit Option

Tentamen i 5B1575 Finansiella Derivat. Måndag 27 augusti 2007 kl Answers and suggestions for solutions.

Empirical analysis on China money multiplier

An Analytical Implementation of the Hull and White Model

where lnp(, ) f(, ) = P(, ) = exp { f(, u)du} = exp{q(, )} Q(, ) = f(, u)du Heah, Jarrow, and Moron (1992) claimed ha under risk-neural measure, he dr

DEBT INSTRUMENTS AND MARKETS

MAFS Quantitative Modeling of Derivative Securities

On multicurve models for the term structure.

Exam 1. Econ520. Spring 2017

Option Valuation of Oil & Gas E&P Projects by Futures Term Structure Approach. Hidetaka (Hugh) Nakaoka

Bond Prices and Interest Rates

CURRENCY TRANSLATED OPTIONS

VaR and Low Interest Rates

Research Paper Series. No. 64. Yield Spread Options under the DLG Model. July, 2009

Financial Econometrics (FinMetrics02) Returns, Yields, Compounding, and Horizon

FINAL EXAM EC26102: MONEY, BANKING AND FINANCIAL MARKETS MAY 11, 2004

Synthetic CDO s and Basket Default Swaps in a Fixed Income Credit Portfolio

Appendix B: DETAILS ABOUT THE SIMULATION MODEL. contained in lookup tables that are all calculated on an auxiliary spreadsheet.

Single Premium of Equity-Linked with CRR and CIR Binomial Tree

Standard derivatives pricing theory (see, for example, Hull,

Funding beyond discounting: collateral agreements and derivatives pricing

Proceedings of the 48th European Study Group Mathematics with Industry 1

Term Structure Models: IEOR E4710 Spring 2005 c 2005 by Martin Haugh. Market Models. 1 LIBOR, Swap Rates and Black s Formulae for Caps and Swaptions

Macroeconomics II A dynamic approach to short run economic fluctuations. The DAD/DAS model.

Optimal Early Exercise of Vulnerable American Options

Origins of currency swaps

HEDGING SYSTEMATIC MORTALITY RISK WITH MORTALITY DERIVATIVES

ANSWER ALL QUESTIONS. CHAPTERS 6-9; (Blanchard)

The Global Factor in Neutral Policy Rates

Valuation and Hedging of Correlation Swaps. Mats Draijer

a) No constraints on import- export, no limit on reservoir, all water in the first period The monopoly optimisation problem is:

CURRENCY CHOICES IN VALUATION AND THE INTEREST PARITY AND PURCHASING POWER PARITY THEORIES DR. GUILLERMO L. DUMRAUF

MORNING SESSION. Date: Wednesday, April 26, 2017 Time: 8:30 a.m. 11:45 a.m. INSTRUCTIONS TO CANDIDATES

MA Advanced Macro, 2016 (Karl Whelan) 1

On Monte Carlo Simulation for the HJM Model Based on Jump

Core issue: there are limits or restrictions that each policy-setting authority places on the actions of the other

Final Exam Answers Exchange Rate Economics

Mathematical methods for finance (preparatory course) Simple numerical examples on bond basics

AMS Q03 Financial Derivatives I

Introduction to Black-Scholes Model

Exotic FX Swap. Analytics. ver 1.0. Exotics Pricing Methodology Trading Credit Risk Pricing

Advanced Tools for Risk Management and Asset Pricing

Pricing Inflation-Indexed Derivatives Using the Extended Vasicek Model of Hull and White

Description of the CBOE Russell 2000 BuyWrite Index (BXR SM )

CENTRO DE ESTUDIOS MONETARIOS Y FINANCIEROS T. J. KEHOE MACROECONOMICS I WINTER 2011 PROBLEM SET #6

May 2007 Exam MFE Solutions 1. Answer = (B)

Fundamental Basic. Fundamentals. Fundamental PV Principle. Time Value of Money. Fundamental. Chapter 2. How to Calculate Present Values

LIBOR MARKET MODEL AND GAUSSIAN HJM EXPLICIT APPROACHES TO OPTION ON COMPOSITION

LIDSTONE IN THE CONTINUOUS CASE by. Ragnar Norberg

UCLA Department of Economics Fall PhD. Qualifying Exam in Macroeconomic Theory

Computations in the Hull-White Model

Problem Set 1 Answers. a. The computer is a final good produced and sold in Hence, 2006 GDP increases by $2,000.

INTEREST RATES AND FX MODELS

Coupling Smiles. November 18, 2006

OPTIMUM FISCAL AND MONETARY POLICY USING THE MONETARY OVERLAPPING GENERATION MODELS

Description of the CBOE S&P 500 2% OTM BuyWrite Index (BXY SM )

Black-Scholes Model and Risk Neutral Pricing

Supplement to Chapter 3

New Acceleration Schemes with the Asymptotic Expansion in Monte Carlo Simulation

Subdivided Research on the Inflation-hedging Ability of Residential Property: A Case of Hong Kong

Alexander L. Baranovski, Carsten von Lieres and André Wilch 18. May 2009/Eurobanking 2009

Interest Rate Products

Misspecification in term structure models of commodity prices: Implications for hedging price risk

CHAPTER CHAPTER26. Fiscal Policy: A Summing Up. Prepared by: Fernando Quijano and Yvonn Quijano

This specification describes the models that are used to forecast

Macro-finance models of the term structure: a review

Principles of Finance CONTENTS

Stock Market Behaviour Around Profit Warning Announcements

Hull & White Convexity Adjustments for Credit Riskless Interest Rate Swaps Under CSA

Markov-Functional Interest Rate Models*

HEDGING VOLATILITY RISK

AMS Computational Finance

IJRSS Volume 2, Issue 2 ISSN:

t=1 C t e δt, and the tc t v t i t=1 C t (1 + i) t = n tc t (1 + i) t C t (1 + i) t = C t vi

Financial Markets And Empirical Regularities An Introduction to Financial Econometrics

Parameters of the IRB Approach. 1. Class of exposures to central governments and central banks, exposures to institutions or corporate exposures

Valuing Real Options on Oil & Gas Exploration & Production Projects

Transcription:

Improving he Jarrow-Yildirim Inflaion Model Rober Hardy May 19, 2013 1 Inroducion The mos liquid inflaion markes are hose of he US, UK, France and Eurozone. Each is suppored by a regular supply of governmen-issued inflaion-linked bonds, and mos rading desks provide swaps and vanilla opions o some exen. The Eurozone marke is he mos developed in erms of derivaive producs. Roughly in order from he mos- o leas-liquid, here are prices in zero-coupon (ZC) swaps, year-on-year (YoY) swaps and opions, zero-coupon opions; i is also possible o ge prices for some firs-order exoics such as YoY digials and YoY range-accruals. Wihin all markes he YoY vanilla opion flows are concenraed on he zero-percen srike, wih he YoY 0% floors being he mos raded produc, mainly because YoY swap rades wih cliens end o have he coupons of he inflaion leg floored a 0%. Oher YoY srikes will rade less frequenly, perhaps as hedges o srucured producs such as inflaion-linked MTNs, and in periods where he marke anicipaes higher levels of inflaion here may be increased amouns of rading of he highersrike (eg 4% or 5%) caps. All hings considered, i is fair o say ha smile modelling is less imporan for inflaion markes han i is for he raes markes, say. In fac i is only he UK inflaion marke wih is LPI produc for pension funds which migh presen he need for a smile-enabled mone-carlo model, and even here one can find a reasonable workaround wih approximaion formulas. In he auhor s experience he main prioriy for inflaion modelling has been o produce a single model which he rading desk can use o: generae he convexiy adjusmens for YoY swap raes, calibrae o he erm srucure of YoY 0% srike vols, calibrae o he erm srucure of ZC 0% srike vols, generae paymen-delay adjusmens for producs like pay-as-you-go swaps. The benefi of having a single model o generae hese volailiy-dependen prices or price adjusmens is wofold: rading and risk-conrol eams prefer o have a single model which can saisfacorily explain he prices seen in he marke, and secondly i offers he rading desk more hedging sraegies such as hedging ZC opions wih YoY or hedging he YoY swap adjusmens wih YoY opions. I goes wihou saying ha a single model which achieves all hese requiremens can be used as he basis for pricing and risk managing he few pah-dependen producs ha ge requesed from ime o ime even hough i is no smile enabled we can be confiden ha i correcly reflecs he core volailiy and correlaion levels of he marke (raders end o see he spread beween ZC and YoY opion volailiies as reflecing he correlaions amongs he erm srucure of YoY opions). 1

The earlies arbirage-free model for inflaion was presened in an aricle by Jarrow and Yildirim, and is based on he FX analogy. Since i is in fac nohing oher han he well known HJM crosscurrency model, i was easy for rading houses o code up a new inflaion wrapper for he FX model and voila! he JY model became he sandard approach for dealing wih inflaion-linked derivaive producs. In recen years he JY model has seen is populariy fade, as marke-model approaches have become developed, and i is easy o undersand why: he JY model seems o suffer from over-paramerizaion, i diffuses he raher esoeric real-yield process, i is no obvious how o calibrae. In conras, he marke models ake a more inuiive slimmed-down approach and diffuse he inflaion process direcly, and i ends o be obvious how hey should be calibraed. The fac is however ha wih a small amoun of work, he JY model can be modified o produce a new model ha is perfecly able o acheive all of he requiremens lised above. In his aricle we presen he mahemaics behind a re-facoring of he JY model which produces a very saisfacory inflaion model. Furhermore, we show ha a marginally reduced version of his new model can be implemened very quickly as a wrapper around an exising implemenaion of he JY model, which means ha you can have his beer version up and running wih a minimal amoun of effor. 2 The JY model The JY model is based on real and nominal economies, each wih is own yield curve, which are conneced by a spo process for he inflaion index. The inflaion index is analogous o he FX spo process and dicaes he curren nominal price of real asses. The JY model specifies he dynamics of he real and nominal discoun facors and he spo index process (respecively B r (; T ), B n (; T ) and I()), under he risk-neural measure P n, as follows: db n (; T ) B n (; T ) = r n() d + σ Bn (; T ) dw n, db r (; T ) B r (; T ) = [r r() σ I ()σ Br (; T )ρ ri ()] d + σ Br (; T ) dw r, di() I() = [r n() r r ()] d + σ I () dw I, where ( ) W n, W r, W I is a Brownian moion under Pn wih correlaion marix 1 ρ nr ρ ni ρ nr 1 ρ ri ρ ni ρ ri 1 Gaussian dynamics for he raes are specified: T σ Bk (; T ) = σ k () e s λ k(u) du ds, k = n, r wih σ n, σ r, λ n and λ r being deerminisic funcions (he shor-rae vols and he mean reversions). I is no obvious how o fully calibrae he JY model, bu below is an ouline of he approach his auhor found o be mos pracicable and useful (paricularly because i gives good-qualiy risks): 1. calibrae he erm srucure of nominal volailies, he σ n, in order o correcly price libor caps a a given srike (depending on he nominal vol hedge o be used), 2

2. calibrae he erm srucure of real volailies, he σ r, in order o produce he correc convexiy adjusmen for YoY swap raes, 3. calibrae he erm srucure of CPI volailies, he σ I, in order o correcly price a chosen se of YoY opions (usually being he 0% floors). This calibraion recipe sill leaves a number of parameers unconsrained: he wo mean reversions λ n and λ r and he hree correlaions ρ ni, ρ ri and ρ nr. I is possible o use hisorical series o pu a figure on ρ nr and in principle he same applies o ρ ni and ρ ri bu esimaion is raher more difficul because of he low number of hisorical index daa poins. Insead i was preferred o se ρ ni = ρ ri = 0 because i means ha he sep where σ I is calibraed o YoY opions will no disurb he previous calibraion o he convexiy adjusmens (which depend on σ r, σ n, ρ ni and ρ ri ). For he mean reversions a pragmaic soluion works bes, a compromise based on hisorical analysis and mahemaical simpliciy, which for he Eurozone marke mean choosing λ n = λ r = 10%. 3 Refacoring he JY model The key o improving he calibraion and he dynamics of he JY model is o no diffuse he real-yield curve bu insead o diffuse an inflaion curve, as we show in his secion. To moivae he new erms we inroduce, we recall ha in he JY model he ime- value of he inflaion index wih mauriy T is given by: I(; T ) = I() B r(; T ) B n (; T ) = I() T e fn(;s) fr(;s) ds, where f n (; s) and f r (; s) represen he ime- values of he insananeous forward raes wih mauriy s in he nominal and real economies. Clearly he spread f n (; s) f r (; s) defines an implied curve of insananeous inflaion forward raes, which we wrie as f i (; s) and which our new model diffuses (raher han f r (; s) in he JY model). To his end we define a process ZC(; T ) which represens he coninuously-compounded inflaionary growh beween imes and T : ZC(; T ) := e T fi(;s) ds. The new model keeps he same processes as JY for he nominal discoun facors and he inflaion spo index, bu replaces he process of real discoun facors wih he ZC process. The specificaion is: db n (; T ) B n (; T ) = r n() d + σ Bn (; T ) dw n, dzc(; T ) ZC(; T ) = r i() d + µ(; T ) d + σ ZC (; T ) dw i, di() I() = r i() d + σ I () dw I, where i can be checked ha for no-arbirage he drif erm mus be given as µ(; T ) = σ ZC (; T ) 2 ρ ii σ I ()σ ZC (; T ) + [ρ ni σ I () ρ ni σ ZC (; T )] σ Bn (; T ) and where ( ) W n, W i, W I is a Brownian moion under Pn wih correlaion marix 1 ρ ni ρ ni ρ ni 1 ρ ii ρ ni ρ ii 1 3

Gaussian dynamics for he nominal and inflaion raes are specified: T σ Bn (; T ) = σ n () σ ZC (; T ) = σ i () T e s λn(u) du ds, e s λi(u) du ds, wih σ n, σ i, λ n and λ i being deerminisic funcions (he shor-rae vols and he mean reversions). In his new model we have I(; T ) = I()ZC(; T ), from which i follows ha he forward-index erms I(; T ) have dynamics given by: di(; T ) I(; T ) = [ρ niσ I () ρ ni σ ZC (; T )] d + σ I () dw I () + σ ZC (; T ) dw i (), and herefore he Black-Scholes volaiilies of inflaion opions (boh YoY and ZC) are only dependen on he parameers σ I and σ i and he correlaion ρ ii. The convexiy adjumen is given by: E ( ) I(T2 ) = I(; T 2) I(T 1 ) I(; T 1 ) ( ) T1 exp [ρ ii σ I (s) σ ZC (s; T 1 )] [σ ZC (s; T 2 ) σ ZC (s; T 1 )] ds ( ) T1 exp [ρ ni σ I (s) ρ ni σ ZC (s; T 1 )] [σ Bn (s; T 2 ) σ Bn (s; T 1 )] ds The convexiy adjusmen erms facor nealy ino wo pars: one depending only on he inflaion parameers, and he oher which also depends on he nominal vol erms. This second erm arises from here being a paymen delay on he denominaor erm. 4 Calibraion of he new model The new specificaion gives a much beer relaionship beween he parameers of he model and he prices of he asses ha are raded in he marke and his immediaely improves he prospecs for a beer calibraion. The correspondence is: he prices of nominal opions (eg libor caps or floors) are deermined by σ n and λ n (as hey are in he JY model), he Black-Scholes volailiies of inflaion opions are deermined by σ I, σ i, ρ ii and λ i, he YoY convexiy adjusmens can be weaked wih ρ ni and ρ ni. Imporanly, he way ha σ i and σ I affec he Black-Scholes volailiies of ZC and YoY opions is quie differen, so we now have a mechanism for adjusing he spread beween ZC and YoY opions: if we pu more inflaion volailiy ino he model wih σ i we will end o increase he ZC-YoY volailiy spread, whereas if we use σ I o increase he inflaion volailiy we will end o decrease he ZC-YoY volailiy spread. Taking all his ino consideraion, he following scheme for calibraion of his new model has been found o work very well in pracice i will fi he YoY volailiies exacly and has been able o generae a good qualiy fi o ZC volailiies and YoY convexiy adjusmens. 1. sar he calibraion loop wih λ i = 0.1, ρ ni = 0, ρ ii = 0 and ρ ni = 0, and σ i = 0.005, 4

2. calibrae he erm srucure of nominal volailies, he σ n, and adjus he level of mean reversion λ n in order o correcly price he nominal hedge insrumens (eg libor caps and swapions a a given srike), 3. calibrae he erm srucure of he σ I, in order o correcly hi he Black-Scholes volailiies of he YoY opions a a given srike (again, depending on he inflaion vol hedge o be used), 4. increase (decrease) σ i in order o ge generally higher (lower) levels of BS volailiies for he inflaion ZC opions, and reurn o sep 3 5. increase (decrease) λ i in order o pu more (less) curvaure ino he shape of he BS volailiies of he ZC opions, and reurn o sep 3. 6. increase (decrease) ρ ni in order o widen (narrow) he convexiy adjusmen, and reurn o sep 3 Seps 4, 5 and 6 can obviously be encapsulaed in a minimizaion rouine. The wo remaining parameers which we have no ye addressed, ρ ii and ρ ni have a more suble effec on he shapes of he calibraed insrumens, bu wihin he calibraion loop we are suggesing here hey can be used o change he way marke movemens in he YoY opion volailiies generae moves in he ZC volailiies and in he convexiy adjusmens. In oher words hey give some degree of conrol o he rader o choose how marke moves in YoY ge carried across ino he ZC and convexiy markes. 5 A quick implemenaion of (a slighly-resriced version of) he new model A he cos of losing one degree of freedom, i is possible o map he new model back ono he JY model. This allows us o obain an almos immediae implemenaion of he new scheme wihin an implemenaion of JY. Namely, we insis ha we mus always have λ n = λ i and herefore lose he flexibiliy o calibrae o nominal swapions as well as caps, for example; his is a reasonable compromise. In oher words, if we have a se {λ n, λ i ; σ n (), σ i (), σ I (), ρ ni (), ρ ni (), ρ ii ()} of mean reversions and erm-srucure values for he parameers of he new model and furhermore have λ n = λ i, hen wih he following definiions: λ r := λ i, σ r () := σ n () 2 + σ i () 2 2ρ ni ()σ n ()σ i (), ρ nr () := 1 σ r () (σ n() ρ ni ()σ i ()), ρ ri () := 1 σ r () (ρ ni()σ n () ρ ii ()σ i ()), we have anoher se {λ n, λ r ; σ n (), σ r (), σ I (), ρ nr (), ρ ni (), ρ ri ()} of mean reversions and ermsrucure parameers which we can use in a JY model o generae exacly he same volailiy disribuions. This means ha by wriing a simple wrapper a he fron and back of an exising implemenaion of JY, we can very quickly build an implemenaion of his improved model. A lile more work on a basic spreadshee calibraion rouine will hen be enough o have a workable model which he rading desk can experimen wih. None of he inernals of he pricing engines needs o be re-plumbed. 5

6 Beer PnL Explain In he JY model a bump on he nominal vol parameer σ n will impac he valuaion of a book of inflaion opions poenially in wo ways: 1. i will affec Black-Scholes volailiy, 2. i will affec he convexiy adjusmen. The new model is in a much beer siuaion since a bump in he nominal vol will affex only he convexiy adjusmen, and here i only causes a change hrough he paymen delay componen of he adjusmen; in he JY model he change in he convexiy adjusmen will be due boh o he change of paymen delay and he change in implied inflaion volailiy. 6