On the multiplicity of option prices under CEV with positive elasticity of variance

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

Introduction to Black-Scholes Model

Pricing FX Target Redemption Forward under. Regime Switching Model

Equivalent Martingale Measure in Asian Geometric Average Option Pricing

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

INSTITUTE OF ACTUARIES OF INDIA

IJRSS Volume 2, Issue 2 ISSN:

Optimal Early Exercise of Vulnerable American Options

INSTITUTE OF ACTUARIES OF INDIA

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

Pricing options on defaultable stocks

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

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

LIDSTONE IN THE CONTINUOUS CASE by. Ragnar Norberg

Models of Default Risk

Available online at ScienceDirect

MAFS Quantitative Modeling of Derivative Securities

Tentamen i 5B1575 Finansiella Derivat. Torsdag 25 augusti 2005 kl

Asymmetry and Leverage in Stochastic Volatility Models: An Exposition

Jarrow-Lando-Turnbull model

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

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

A pricing model for the Guaranteed Lifelong Withdrawal Benefit Option

Final Exam Answers Exchange Rate Economics

Systemic Risk Illustrated

A Note on Missing Data Effects on the Hausman (1978) Simultaneity Test:

Black-Scholes Model and Risk Neutral Pricing

An Analytical Implementation of the Hull and White Model

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

Brownian motion. Since σ is not random, we can conclude from Example sheet 3, Problem 1, that

UvA-DARE (Digital Academic Repository) Currency option pricing in a credible exchange rate target zone Veestraeten, D.J.M. Link to publication

Computations in the Hull-White Model

Extended MAD for Real Option Valuation

PARAMETER ESTIMATION IN A BLACK SCHOLES

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

The Binomial Model and Risk Neutrality: Some Important Details

MA Advanced Macro, 2016 (Karl Whelan) 1

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

Option pricing and hedging in jump diffusion models

VALUATION OF THE AMERICAN-STYLE OF ASIAN OPTION BY A SOLUTION TO AN INTEGRAL EQUATION

R e. Y R, X R, u e, and. Use the attached excel spreadsheets to

A NOTE ON BUSINESS CYCLE NON-LINEARITY IN U.S. CONSUMPTION 247

Reconciling Gross Output TFP Growth with Value Added TFP Growth

Erratic Price, Smooth Dividend. Variance Bounds. Present Value. Ex Post Rational Price. Standard and Poor s Composite Stock-Price Index

A Theory of Tax Effects on Economic Damages. Scott Gilbert Southern Illinois University Carbondale. Comments? Please send to

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

On the Impact of Inflation and Exchange Rate on Conditional Stock Market Volatility: A Re-Assessment

VaR and Low Interest Rates

Black-Scholes and the Volatility Surface

Volatility and Hedging Errors

EXPONENTIAL MARTINGALES AND TIME INTEGRALS OF BROWNIAN MOTION

Hull-White one factor model Version

An Incentive-Based, Multi-Period Decision Model for Hierarchical Systems

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

You should turn in (at least) FOUR bluebooks, one (or more, if needed) bluebook(s) for each question.

Some Remarks on Derivatives Markets (third edition, 2013)

The Investigation of the Mean Reversion Model Containing the G-Brownian Motion

arxiv:math/ v2 [math.pr] 26 Jan 2007

ECON Lecture 5 (OB), Sept. 21, 2010

Foreign Exchange, ADR s and Quanto-Securities

Bond Prices and Interest Rates

STATIONERY REQUIREMENTS SPECIAL REQUIREMENTS 20 Page booklet List of statistical formulae New Cambridge Elementary Statistical Tables

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

Advanced Tools for Risk Management and Asset Pricing

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

Aid, Policies, and Growth

Estimating Earnings Trend Using Unobserved Components Framework

Introduction. Enterprises and background. chapter

Fitting the Heston Stochastic Volatility Model to Chinese Stocks

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

VOLATILITY CLUSTERING, NEW HEAVY-TAILED DISTRIBUTION AND THE STOCK MARKET RETURNS IN SOUTH KOREA

Coupling Smiles. November 18, 2006

Leveraged Stock Portfolios over Long Holding Periods: A Continuous Time Model. Dale L. Domian, Marie D. Racine, and Craig A.

(c) Suppose X UF (2, 2), with density f(x) = 1/(1 + x) 2 for x 0 and 0 otherwise. Then. 0 (1 + x) 2 dx (5) { 1, if t = 0,

Risk-Neutral Probabilities Explained

a. If Y is 1,000, M is 100, and the growth rate of nominal money is 1 percent, what must i and P be?

CHRISTOPH MÖHR ABSTRACT

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

Data Mining Anomaly Detection. Lecture Notes for Chapter 10. Introduction to Data Mining

Li Gan Guan Gong Michael Hurd. April, 2006

Analyzing Surplus Appropriation Schemes in Participating Life Insurance from the Insurer s and the Policyholder s Perspective

Data Mining Anomaly Detection. Lecture Notes for Chapter 10. Introduction to Data Mining

Non-Stationary Processes: Part IV. ARCH(m) (Autoregressive Conditional Heteroskedasticity) Models

On Monte Carlo Simulation for the HJM Model Based on Jump

The Impact of Stochastic Volatility on Pricing, Hedging, and Hedge Efficiency of Variable Annuity Guarantees

Output: The Demand for Goods and Services

Credit Spread Option Valuation under GARCH. Working Paper July 2000 ISSN :

Question 1 / 15 Question 2 / 15 Question 3 / 28 Question 4 / 42

EVA NOPAT Capital charges ( = WACC * Invested Capital) = EVA [1 P] each

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

FORECASTING WITH A LINEX LOSS: A MONTE CARLO STUDY

Macroeconomics. Typical macro questions (I) Typical macro questions (II) Methodology of macroeconomics. Tasks carried out by macroeconomists

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

A Method for Estimating the Change in Terminal Value Required to Increase IRR

Changes of Numeraire for Pricing Futures, Forwards, and Options

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

Funding beyond discounting: collateral agreements and derivatives pricing

Web Usage Patterns Using Association Rules and Markov Chains

FAIR VALUATION OF INSURANCE LIABILITIES. Pierre DEVOLDER Université Catholique de Louvain 03/ 09/2004

Transcription:

Rev Deriv Res (207) 20: 3 DOI 0.007/s47-06-922-2 On he mulipliciy of opion prices under CEV wih posiive elasiciy of variance Dirk Veesraeen Published online: 4 April 206 The Auhor(s) 206. This aricle is published wih open access a Springerlink.com Absrac The discouned sock price under he Consan Elasiciy of Variance model is no a maringale when he elasiciy of variance is posiive. Two expressions for he European call price hen arise, namely he price for which pu-call pariy holds and he price ha represens he lowes cos of replicaing he call opion s payoffs. The greeks of European pu and call prices are derived and i is shown ha he greeks of he risk-neural call can subsanially differ from sandard resuls. For insance, he relaion beween he call price and variance may become non-monoonic. Such unfamiliar behavior hen migh yield opion-based ess for he poenial presence of a bubble in he underlying sock price. Keywords Bubbles Consan elasiciy of variance Opion pricing Pu-call pariy Risk-neural valuaion JEL Classificaion G3 Inroducion The elasiciy of he variance of reurns o he sock price under Geomeric Brownian Moion (GBM) is zero and reurns hus are assumed o have consan variance. This assumpion was challenged in, for insance, Black (976) and became he main I am very graeful o he anonymous referee whose consrucive commens allowed me o remove erroneous claims and o grealy enhance he focus of he paper. The usual disclaimer applies. B Dirk Veesraeen d.j.m.veesraeen@uva.nl Amserdam School of Economics, Universiy of Amserdam, Roeerssraa, 08WB Amserdam, The Neherlands 23

2 D. Veesraeen argumen behind he developmen of he Consan Elasiciy of Variance (CEV) opion pricing model in Cox (975). The discouned sock price is a maringale under he CEV process wih negaive or zero elasiciy, bu is no a maringale for posiive values of he elasiciy parameer as was firs noed in Emanuel and MacBeh (982) and Lewis (2000). Cox and Hobson (2005) and Heson e al. (2007) inerpreed his loss of he maringale propery under posiive elasiciy of variance as evidence for he presence of a bubble in he underlying sock price. As a resul, here are (a leas) wo candidae prices for he call opion, namely he price for which he pu-call pariy holds and he price ha represens he lowes cos of replicaing he payoff of he call (see Heson e al. 2007). This paper calculaes he greeks of European pu and call opions under CEV wih posiive elasiciy of variance and inerpres hem as poenial indicaors for he presence of bubbles in he underlying. I is shown ha he greeks for he risk-neural call can reveal behavior ha is in sharp conras wih he sandard resuls as suggesed earlier in Cox and Hobson (2005), Eksröm and Tysk (2009) and Pal and Proer (200). For insance, risk-neural call prices can be non-monoonic funcions of he remaining ime o mauriy in which increasing ime iniially increases he call s price bu afer some poin sars o depress he value of he call. Deecing such paern hen migh poin o he presence of a bubble in he sock price. Indeed, he bubble may in he shor run inflae furher and hus sep up opion prices, bu he likelihood of is collapse grows for longer ime horizons causing he call price hen o decrease in he ime o mauriy. Addiional ess could be based on a similar non-monoonic relaion beween he risk-neural call price and he variance of he underlying as well as he poenial concave relaion beween call and sock prices. The remainder of his paper is organized as follows. Secion 2 discusses he loss of he maringale propery for he discouned sock price under CEV wih posiive elasiciy of variance. Secion 3 calculaes he European call and pu prices. Secion 4 discusses he sensiiviies of he opion prices o heir underlying parameers and inerpres heir behavior as poenial indicaors of bubbles in sock prices. Secion 5 concludes. 2 The loss of he maringale propery under CEV wih posiive elasiciy of variance The sock price is assumed o follow a CEV diffusion process wih risk-neuralized drif rs and diffusion coefficien σ 2 S 2α, respecively ds = rsd + σ S α dw, () wihr being he posiive, consan risk-free ineres rae. The elasiciy of he variance of reurns o he sock price equals α and GBM hus emerges as he limi for α =. This paper focuses on posiive elasiciy of variance, i.e. α>, and for breviy he acronym CEV from now on refers o CEV wih posiive elasiciy of variance. 23

On he mulipliciy of opion prices under CEV wih... 3 The ransiion probabiliy densiy funcion p S T, T ; S, is derived in Emanuel and MacBeh (982) as ( p S T, T ; S, =2 (α ) k 2( α) wih k = z z 4α T r σ 2 ( α)(exp 2r ( α)(t ) ) > 0, z = ks 2( α) exp 2r ( α)(t ), ) exp z z T I ( ) 2 z z T, 2(α ) (2) z T = ks 2( α) T, I q = he modified Bessel funcion of he firs kind of order q. The disribuional relaions in Schroder (989) allow o specify he following wo inegral expressions ha faciliae calculaion of he below resuls and J ;β,β 2 = β 2 β Q J 2;β,β 2 = p S T, T ; S, ds T = Q 2kβ 2( α), 2 + α, 2z β 2 β S T p S T, T ; S, ds T = S exp r (T ) Q 2z, { Q 2z, α, 2kβ2( α) 2 2kβ 2( α) 2, 2 + α, 2z α, 2kβ2( α) }, where Q a, b, c is he complemenary non-cenral chi-square disribuion evaluaed a a wih b degrees of freedom and non-cenraliy parameer c. 2 The condiional expecaion for he CEV process (), E S T, T ; S,, hen is E S T, T ; S, = + 0 S T p S T, T ; S, ds T The expression for he ransiion probabiliy densiy funcion in (7) inheson e al. (2007) conains a misprin as he erm (xz α) 4 α should be (xz 4α) 4( α). Also, he numeraor in he firs fracion in he erm u on p. 367 should be r raher han 2r. 2 The complemenary chi-square disribuion arises if he non-cenraliy parameer equals 0. See Zelen and Severo (964) for more deail on he (non-cenral) chi-square disribuion. 23

4 D. Veesraeen E S T, T ; S, = J 2;0,+ { E S T, T ; S, = S exp r (T ) Q 2z, E S T, T ; S, = S exp r (T ) { Q } α, + Q 2z, α, 0 } 2z,, (3) α, 0 given ha Q a, b, + = as can be inferred from definiion 26.4.25 in Zelen and Severo (964). The condiional expecaion (3) hus falls below S exp r (T ) since Q 2z, α, 0 is a disribuion. As a resul, he discouned sock price is no a maringale. 3 Cox and Hobson (2005) and Heson e al. (2007) inerpreed his loss of he maringale propery as evidence ha he sock price has a bubble and as a resul is expeced o decrease in he fuure. 3 European call and pu opion prices The risk-neural value of a European pu opion wih exercise price K and ime o mauriy T, P, is given by P = exp r (T ) K 0 K S T p S T, T ; S, ds T P = exp r (T ) { } K J ;0,K J 2;0,K { ( P = exp r (T ) K Q 2kK 2( α), 2 + α, 2z Q +, 2 + )} α, 2z { ( exp r (T ) S exp r (T ) Q 2z, α, + )} Q 2z, α, 2kK2( α) P = K exp r (T ) Q 2kK 2( α), 2 + α, 2z ) S ( Q 2z, α, 2kK2( α), (4) which corresponds wih he pu price ha is derived in Hull (2006). 3 Noe ha resul 26.4. in Zelen and Severo (964) implies ha Q 2z, α, 0 = 0forα =. The discouned sock price hen is a maringale in line wih he fac ha α = yields GBM. 23

On he mulipliciy of opion prices under CEV wih... 5 The risk-neural expeced discouned value of he payoff o he European call opion, in shor he risk-neural call price C,is C = exp r (T ) + K S T K p S T, T ; S, ds T C = exp r (T ) { } J 2;K,+ K J ;K,+ { C = exp r (T ) S exp r (T ) )} Q 2z, α, 0 { ( exp r (T ) K Q 0, 2 + Q C = S (Q 2z, 2kK 2( α), 2 + K exp r (T ) α, 2z α, 2kK2( α) Q ( Q ( Q 2z, α, 2z )} 2z, α, 2kK2( α) ) α, 0 2kK 2( α), 2 + α, 2z ). (5) The call price (5) corresponds wih he price G 2 (S, ) ha was obained in Heson e al. (2007). Emanuel and MacBeh (982) repored a second expression for he call price, denoed by C EM, ha can be obained via he pu-call pariy. Plugging he risk-neural pu price (4) ino he pu-call pariy yields C EM C EM = P + S K exp r (T ) = S Q 2z, α, 2kK2( α) ( K exp r (T ) Q 2kK 2( α), 2 + ) α, 2z. (6) This price is documened in, amongs ohers, Schroder (989), Davydov and Linesky (200), Hull (2006) and is specified as G (S, ) in Heson e al. (2007). The relaion beween he call prices (5) and (6)is C EM C = S Q 2z, α, 0. (7) 23

6 D. Veesraeen Hence, C EM exceeds C. 4 The call price C EM hus has a bubble given ha i exceeds he cos of a sraegy ha replicaes he payoffs of he call opion (see Heson e al. 2007 for more deail). 4 The greeks of European pu and call opion prices I is possible o calculae he greeks for he above opion prices for general values of α as definiion 26.4.25 in Zelen and Severo (964) expresses he non-cenral chi-square disribuion as an infinie sum of chi-square disribuions. The required derivaives hen be compued via relaion 26.4.9 in Zelen and Severo (964) and relaion 6.5.25 in Davis (964). Bu, he resuling expressions become oo elaborae for a meaningful analysis. However, opion prices and heir greeks can be expressed in (surprisingly) compac and easy-o-inerpre form for paricular values of α. Forα = 2, he densiy 2 funcion (2) can be simplified via he ideniy I (x) = sinh x (see Polyanin 2 π x 2002) ino 5 p α=2 S T, T ; S, = S k exp r (T ) exp 2 ST 3 {exp π { 2kS T The condiional expecaion hen is { 2kS T 2 + } } 2 2kS exp r (T ). } 2 2kS exp r (T ) E α=2 S T, T ; S, = S exp r (T ) { 2 q }, where q = 2kS exp r (T ) and q = 2π sandard normal disribuion funcion. The risk-neural pu price, P α=2, emerges as P α=2 q exp 2 x2 dx is he = K exp r (T ) σ S exp 2r (T ) {φ q 2 φ q 3 } 2r + K exp r (T ) { q 2 + q 3 } S { q 3 q 2 }, (8) 4 Boh expressions for he call price saisfy he valuaion parial differenial equaion of Black and Scholes (973) and is boundary condiions as discussed in Heson e al. (2007) andeksröm and Tysk (2009). Heson e al. (2007) and Eksröm and Tysk (2009) furhermore argued ha he muliples of he difference (7) also saisfy he valuaion equaion such ha acually an infinie number of soluions arises. 5 The ideniies in Polyanin (2002) andschroder (989) can be combined o yield similar simplificaions for various oher values of α. Cox and Hobson (2005) andeksröm and Tysk (2009) also focused on he case of α = 2 bu wih he added resricion of r = 0 in order o express he sock price as he reciprocal of he radial par of a 3-dimensional Brownian moion. 23

On he mulipliciy of opion prices under CEV wih... 7 where q 2 = 2k { S exp r ( T ) K }, q 3 = 2k { S exp r ( T ) + K } and φ q = exp 2π 2 q2. The risk-neural value of he call opion, C α=2, is given by C α=2 = S {2 q q 2 q 3 } K exp r (T ) { q 3 q 2 } K exp r (T ) σ S exp 2r (T ) {φ q 3 φ q 2 } (9) 2r and he relaion beween he risk-neural pu and call prices, i.e. he risk-neural pu-call pariy, is C α=2 = P α=2 + S { 2 q } K exp r (T ). (0) The call price of Emanuel and MacBeh (982), C EM,α=2, can be obained by plugging ino he pu-call pariy P α=2 C EM,α=2 = P α=2 + S K exp r (T ) () C EM,α=2 = S {2 q 2 q 3 } K exp r (T ) { q 3 q 2 } K exp r (T ) σ S exp 2r (T ) {φ q 3 φ q 2 }. 2r (2) The call price of Emanuel and MacBeh (982) exceeds he risk-neural call price given ha C EM,α=2 = C α=2 + 2S q (3) 4. Dela The dela of he risk-neural pu price (8) can be calculaed as P α=2 S = q 2 q 3 + K 2k {φ q 2 φ q 3 }. I is no possible o analyically show ha he dela of he pu has he familiar negaive value. However, an exensive grid search confirmed ha he dela is always negaive. 6 The dela of he pu ogeher wih he risk-neural pu-call pariy (0) hen specifies 6 The grid search used values for S beween 0.0 and 0, K ranged beween 0.0 and 5, r moved beween 0.02 and 0., σ beween 0.5and 0.45and T ranged from0. o 5. The oal number of cases ha was evaluaed was a,048,576. 23

8 D. Veesraeen he dela of he risk-neural call as C α=2 S =2 q q 2 q 3 2S 2k exp r (T ) φ q + K {φ q 2 φ q 3 }. 2k The grid search showed ha he dela of he risk-neural call always had he familiar posiive sign. The dela for he call price of Emanuel and MacBeh (982) can mos easily be obained via he pu-call pariy () as C EM,α=2 S = q 2 q 3 + + K 2k {φ q 2 φ q 3 } > 0. The laer expression is always posiive given ha he sum of he firs hree erms on he righ-hand side is posiive and his also holds for he remaining erm. Boh call prices increase in he value of he underlying sock price. However, he dela of he call price of Emanuel and MacBeh (982) exceeds he dela of he riskneural call price as, for insance, relaion (3) implies C EM,α=2 S Cα=2 S = 2 q + 2S 2k exp r (T ) φ q > 0. 4.2 Gamma The risk-neural pu price (8) is characerized by he familiar convexiy in he sock price given ha 2 P α=2 S 2 = S 3 2kK exp 2r (T ) {φ q2 φ q 3 } > 0. Combining he laer resul wih he pu-call pariy () implies ha he call price of Emanuel and MacBeh (982) is a convex funcion of he sock price 2 C EM,α=2 S 2 = 2 P α=2 S 2 > 0. The gamma for he risk-neural call price follows from he risk-neural pu-call pariy (0) as 2 C α=2 S 2 = 2 P α=2 S 2 2 5 2 S 4 k 3 2 exp 3r (T ) φ q. The gamma of he risk-neural call hus falls below ha of he pu in view of k > 0. Moreover, he gamma will be negaive for K = 0 in view of he zero value of he 23

On he mulipliciy of opion prices under CEV wih... 9 0.7 0.9 call price 0.35 call price 0.45 0 0 0 4 8 0 0.5 0.3 sock price vola liy (a) K = 5, r = 0.03, T- = 0.75 and =0.2 (b) K = 5, S = 5, r = 0.03 and T- = 0.75 3.26 0.8 call price 3.2 call price 0.4 3.4 0 0 0. 0.2 0 0.5 ineres rae me o mauriy (c) K = 0. 5, S = 5, T- = 0.75 and =0.2 (d) K = 5, S = 5, r = 0.03 and = 0.2 Fig. Sensiiviy of he risk-neural call price o is underlying parameers for α =2 gamma of he pu and he posiive value of k. The limiing case of K = 0 hus produces a concave relaion beween he risk-neural call price and he sock price as noed in Cox and Hobson (2005) and Eksröm and Tysk (2009). However, larger values for K may creae more inricae paerns in which for insance he familiar convex relaion urns ino a concave relaionship for larger sock prices. Panel (a) in Fig. illusraes his for he parameer values K = 5, r = 0.03, T = 0.75 and σ = 0.2. This unfamiliar behavior of he gamma of he risk-neural call could be of ineres for he idenificaion of bubbles. In fac, deecing concaviy in he call opion price migh indicae ha he underlying sock price has a bubble. Furhermore, observing a larger gamma for he pu han for an idenical call migh also poin o he presence of a bubble in he sock price. 4.3 Vega The sandard resul of a posiive relaion beween opion prices and σ is presen for he risk-neural pu price as well as for he call price of Emanuel and MacBeh (982) 23

0 D. Veesraeen since P α=2 σ = KS σ 2k {φ q 2 φ q 3 } > 0 and C EM,α=2 σ = Pα=2 σ > 0. The vega of C α=2 hen can be obained via he risk-neural pu-call pariy (0) as C α=2 σ = Pα=2 σ 2 2 3 k 2 exp r (T ) φ q. σ The vega of he risk-neural call hus falls below he vega of he idenical pu. Also, a sricly negaive vega for he risk-neural call arises for K = 0. However, he relaion beween C α=2 and σ acually can grow non-monoonic as is illusraed in Panel (b) of Fig. for he parameer values S = 5, K = 5, r = 0.03 and T = 0.75. This non-monoonic relaion beween he risk-neural call price and σ can be inerpreed agains he background of he presence of a bubble in he underlying sock price. Iniially, a rise in σ for low values of he laer may furher inflae he bubble and hus boos he call price. However, afer some criical level of σ is reached, furher rises in σ (sharply) increase he likelihood ha he bubble will burs hrough which he call price mus decrease. Eksröm and Tysk (2009) used an advanced reamen of parial differenial equaions o show ha convexiy (concaviy) of he call in he sock price auomaically implies a posiive (negaive) relaion beween he call opion price and volailiy. This direc link is confirmed for he presen CEV process given ha he vega of he above hree prices can always be obained by muliplying he corresponding gamma wih he posiive erm S4 exp 2r (T ). 2kσ The analysis of he vega hus may offer an addiional opion-based es for he poenial presence of a bubble in he underlying sock price. Such bubble migh be presen if he call price decreases in response o an increase in volailiy or increases o a smaller degree when compared wih he reacion in he price of an idenical pu. 4.4 Rho The rho of he risk-neural pu price is given by P α=2 r = KS ( ) 2kσ 2 (T ) exp 2r (T ) {φ q 2 2 3 k 2 3 φ q 2 } r + K (T ) exp r (T ) { q 3 q 2 }. 23

On he mulipliciy of opion prices under CEV wih... Combining he laer resul wih he pu-call pariy () allows o express he rho of he call price of Emanuel and MacBeh (982) as C EM,α=2 r = Pα=2 r + K (T ) exp r (T ). A grid search showed ha he familiar posiive (negaive) rho for calls (pus) also holds for hese wo opion prices. The rho for he risk-neural call is given by C α=2 r = C EM,α=2 r + 2k exp r (T ) r ( ) 2kσ 2 (T ) φ q. The laer expression ypically reurns a posiive value. However, i can urn negaive for opions ha are deep in-he-money as is illusraed in Panel (c) of Fig. where K = 0.5, S = 5,σ = 0.2 and T = 0.75. 4.5 Thea The expression for he sensiiviy of he risk-neural pu o he remaining ime unil mauriy, hea, is Pα=2 (T ) = σ 2 k 2 KS 2 2 exp 2r (T ) {φ q 3 φ q 2 } rk exp r (T ) { q 3 q 2 }, and his derivaive, as in he sandard case of GBM, may be posiive or negaive. The pu-call pariy () hen specifies he hea of he call price of Emanuel and MacBeh (982) as EM,α=2 C (T ) Pα=2 = rk exp r (T ). (T ) The erm rk exp r (T ) guaranees ha he hea of he call of Emanuel and MacBeh (982) is negaive given ha he erms φ q 2 φ q 3 and q 3 q 2 boh are posiive. The hea for he risk-neural call is Cα=2 (T ) EM,α=2 = C (T ) + 2 3 2 k 2 exp r (T ) (kσ 2 exp 2r (T ) + r ) φ q. Adding he second erm on he righ-hand side hen acually can generae a posiive value for he hea of he risk-neural call. The possibiliy of encounering a posiive 23

2 D. Veesraeen hea was discussed in Pal and Proer (200) wihin he conex of he inverse Bessel process. Noe ha he relaionship beween C α=2 and T can be non-monoonic as is illusraed in Panel (d) of Fig. for K = 5, S = 5, r = 0.03 and σ = 0.2. Such sequence of an iniial rise of he call price in he remaining ime unil mauriy followed by a decline migh be compaible wih he presence of a bubble in he underlying. In fac, he bubble may grow furher in he shor run causing he call o increase in value. However, longer ime horizons make i more likely ha he bubble will collapse which hen negaively impacs upon he price of he call. The above findings for he hea of he call may be useful wihin he search for bubbles in he underlying sock on wo accouns. Firs, a negaive link beween European call prices and he remaining ime unil mauriy migh poin o he presence of a bubble in he sock price as noed earlier. Second, a negaive relaion beween he European call price and is ime o mauriy immediaely implies ha he early-exercise feaure of he idenical American call opion will now have posiive value even when no dividends are paid (Cox and Hobson 2005; Pal and Proer 200). Hence, observing prices for American calls on socks ha pay no dividends being in excess of prices of oherwise idenical European call opions migh likewise signal he presence of a bubble in he underlying. 5 Conclusions The Consan Elasiciy of Variance (CEV) model allows for a more elaborae specificaion of he volailiy of reurns. Opion pricing under he CEV process wih posiive elasiciy of variance, however, is complicaed by he fac ha he discouned sock price no longer is a maringale as was firs noed in Emanuel and MacBeh (982) and Lewis (2000). This loss of he maringale propery under posiive elasiciy of variance was inerpreed as evidence of sock-price bubbles in Cox and Hobson (2005) and Heson e al. (2007).Heson e al. (2007) proceeded by showing ha such bubbles generae (a leas) wo candidae prices for he call opion, namely he price for which pu-call pariy holds and he price ha represens he lowes cos of replicaing he opion s payoffs. This paper calculaes and illusraes he greeks of European pu and call opions under CEV wih posiive elasiciy of variance and discusses heir poenial role in esing for he presence of bubbles in he underlying sock price. I was found ha he greeks of he risk-neural call may exhibi behavior ha deviaes considerably from he sandard resuls. For insance, call prices may become non-monoonic in he variance of he underlying sock. Deecing such paerns hen migh indicae he presence of a bubble in he underlying as low levels of variance may furher inflae he bubble and sep up opion prices whereas he likelihood of a collapse in he bubble and he opion price sharply increases for larger dispersion. Open Access This aricle is disribued under he erms of he Creaive Commons Aribuion 4.0 Inernaional License (hp://creaivecommons.org/licenses/by/4.0/), which permis unresriced use, disribuion, and reproducion in any medium, provided you give appropriae credi o he original auhor(s) and he source, provide a link o he Creaive Commons license, and indicae if changes were made. 23

On he mulipliciy of opion prices under CEV wih... 3 References Black, F. (976). Sudies of sock price volailiy changes. In Proceedings of he 976 meeings of he American saisical associaion, business and economic saisics secion, pp. 77 8. Black, F., & Scholes, M. (973). The pricing of opions and corporae liabiliies. Journal of Poliical Economy, 8, 637 654. Cox, J. C. (975). Noes on opion pricing I: Consan elasiciy of variance diffusions. Unpublished manuscrip, Sanford Universiy. Cox, A. M. G., & Hobson, D. G. (2005). Local maringales, bubbles and opion prices. Finance and Sochasics, 9, 477 492. Davis, P. J. (964). Gamma funcion and relaed funcions. In M. Abramowiz & I. A. Segun (Eds.), Handbook of mahemaical funcions (pp. 253 294). New York: Dover Publicaions. Davydov, D., & Linesky, V. (200). Pricing and hedging pah-dependen opions under he CEV process. Managemen Science, 47, 949 965. Eksröm, E., & Tysk, J. (2009). Bubbles, convexiy and he Black Scholes equaion. Annals of Applied Probabiliy, 9, 369 384. Emanuel, D. C., & MacBeh, J. D. (982). Furher resuls on he consan elasiciy of variance call opion pricing model. Journal of Financial and Quaniaive Analysis, 7, 533 554. Heson, S. L., Loewensein, M., & Willard, G. A. (2007). Opions and bubbles. Review of Financial Sudies, 20, 359 390. Hull, J. C. (2006). Opions, fuures, and oher derivaives. Upper Saddle River: Prenice Hall. Lewis, A. L. (2000). Opion valuaion under sochasic volailiy. Newpor Beach: Finance Press. Pal, S., & Proer, P. (200). Analysis of coninuous sric local maringales via h-ransforms. Sochasic Processes and heir Applicaions, 20, 424 443. Polyanin, A. D. (2002). Handbook of linear parial differenial equaions for engineers and scieniss.boca Raon: Chapman & Hall/CRC. Schroder, M. (989). Compuing he consan elasiciy of variance opion pricing formula. Journal of Finance, 44, 2 29. Zelen, M., & Severo, N. C. (964). Probabiliy funcions. In M. Abramowiz & I. A. Segun (Eds.), Handbook of mahemaical funcions (pp. 925 995). New York: Dover Publicaions. 23