“Time Changed Markov Processes in United Credit-Equity Modeling” Rafael Mendoza-Arriaga∗ Northwestern University Department of Industrial Engineering and Management Sciences Rutgers University Feb 2008 ∗ Joint work with Peter Carr and Vadim Linetsky Motivation. 1.- Local Volatility Models • In the Black-Scholes-Merton model, the stock price follows GBM, which is a process with infinite lifetime (i.e. No possibility of bankruptcy) and with constant volatility. Not very realistic assumptions!!! • Newer Stock options pricing models focus on modeling the implied volatility smile/skew by various means, such as local volatility, stochastic volatility, and jumps, but ignore the possibility of default of the firm underlying the option contract. • Real-World firms have positive probability of bankruptcy in finite time. 1 Motivation. 2- Credit Models • Credit risk literature focuses on modeling default, credit spreads, and the pricing of corporate bonds and credit derivatives, • This literature ignores the information in the equity options market • Clear disconnection between equity and credit derivatives, even though evidence shows the usage of equity (put) options as protection vehicles in case of default. 2 Motivation. 3-GM Example • In 2006 financial analysts considered GM to have high default risk. • In particular, GM equity Put Options with Low Strikes had the highest Open Interest: 1. Jan 07 $10 Put @ $1.50 (ImpVol ≈100%) with OI 493,051 contracts. 2. Jan 07 $5 Put @ $0.50 (ImpVol ≈120%) with OI 205,703 contracts. 3. Jan 07 $7.50, @ $1.00 (ImpVol ≈110%) and $2.50, @ $0.15 (ImpVol ≈133%) Puts also had substantial OI • Total outstanding notional for Jan 07 and Jan 08 Puts with strikes $2.50-$10 ≈130 million shares at $21 USD. 3 Motivation. 4- Far from Continuous The GM stock price was $21.19 in Feb. 22, 2006 4 Motivation. 5- Far from Constant Vol. 140 130 Implied Volatility, % 120 110 100 90 80 70 60 50 40 0 5 10 15 20 25 30 Strike Price Implied Volatility Skew for GM Jan 2007 Puts on Feb 22, 2006. Historical Volatility of GM stock price over the previous 12 months ≈ 46% 5 Observations • Put options on the stock provide default protection and can be used to manage default risk. • Deep out-of-the-money put options are essentially credit derivatives. • The possibility of default contributes to the implied volatility skew in stock options: there is a linkage between implied volatilities skew in options markets and credit spreads in credit markets. (Default Prob ⇑↔ S ⇓→ σ ⇑) • Corporate liabilities, credit derivatives, and equity derivatives should be modeled within a unified framework. 6 Our Approach • We develop an analytically tractable unified modeling framework for corporate debt, credit derivatives, and equity derivatives. • In the reduced-form, intensity-based framework, we model Defaultable Stock as the fundamental state variable. • We view corporate debt, equity derivatives, and credit derivatives as contingent claims written on the defaultable stock. • We introduce default into major classes of equity derivatives models, incorporating jumps and stochastic volatility based on time changes of Markov processes with killing. 7 GENERAL PANORAMA We start off with… … which we modify under a time change… Continuous Continuous Markov Markov Process Process w/killing w/killing Jump Jump Diffusion Diffusion Process Process w/Stochastic w/Stochastic Volatility Volatility & & Killing Killing Time Changes Bochner’s Bochner’sLevy Levy Subordination Subordination Bochner’s Bochner’sLevy Levy Subordination Subordination Abs. Abs.Cts. Cts.Time Time Changes Changes Abs. Abs.Cts. Cts.Time Time Change Change …to move on from here…. …and get a model for options and credit Analytical Option and Credit Pricing f ( x) ∈ L2 ((0, ∞), m) Spectral SpectralExp. Exp. Approach Approach f ( x) ∉ L2 ((0, ∞), m) Laplace LaplaceTransform Transform Approach 8 Approach -1- Jump-to-Default Extended Diffusion (JDD) • Model the pre-default stock dynamics under an EMM Q as a diffusion: dSt = [µ + λ (St )] Stdt + σ (St ) St dBt, S0 = S > 0 where µ = r − q, r, q, σ and λ are the short rate, dividend yield, volatility, and default intensity. • If the diffusion can hit zero, we kill it at the first hitting time of zero, T0 , and send it to a cemetery (bankruptcy) state ∆, where it remains forever. • Prior to T0, a jump-to-default arrives at the first jump time of a doubly stochastic Poisson process with intensity λ(St). The time of default ¯Z t ¾ ½ is: ¯ λ (Su ) du ≥ e , e ≈ Exp(1) ζ = inf t ∈ [0, T0 ] ¯¯ 0 • Assume stock holders do not receive any recovery in the event of default. Addition of λ in the drift r − q + λ compensates for default to insure that the discounted gain process to the stock holders is a martingale under the EMM. 9 Corporate Bonds and European Options under JDD • The time-t price of a defaultable zero coupon bond with face value of $1 and recovery R ∈ [0, 1] at maturity: B (S, t; T ) = e−r(T −t) Q (S, t; T ) + e−r(T −t) R [1 − Q (S, t; T )] where Q (S, t; T ) is the risk neutral · R T survival probability: ¸ λ(Su )du − Q (S, t; T ) = E e t 1{T0>T } • Likewise, the time-t price of a Call Option with strike price K > 0 is given by: · RT ¸ − λ(Su )du C (S, t; K, T ) = e−r(T −t) E e t (ST − K)+ 1{T0 >T } 10 Corporate Bonds and European Options under JDD • The payoff Put Option with strike price K > 0 can be decomposed as: (K − ST )+ 1{ζ>T } + K 1{ζ≤T } Where the first part is the payoff given no default by time T and a recovery amount K at time T in case of default occurs prior maturity. • Therefore, we price the European Put option as: · RT ¸ − λ(Su )du P (S, t; K, T ) = e−r(T −t) E e t (K − ST )+ 1{T0>T } +Ke−r(T −t) [1 − Q (S, t; T )] NOTICE. A Credit Derivative is embedded in the Put Option!!! 11 A Jump-to-Default Extended Constant Elasticity of Variance (JDCEV) Model • The JDCEV process is a 1D diffusion introduced by Carr and Linetsky (2006): dSt = [µ + λ(St)]St dt + σ(St )St dBt , S0 = S > 0, where the local volatility is CEV and default intensity is linear in instantaneous variance: σ(S) = aS β , λ(S) = b + c σ 2 (S) = b + c a2S 2β , where a is the volatility scale parameter (fixing ATM volatility), β < 0 is the volatility elasticity parameter, b is the constant part of the default intensity, and c controls the contribution of the stock return variance in default intensity. • This specification is consistent with the leverage effect St ⇓→ σ ⇑ and the linkage between stock volatility and credit spreads σ ⇑↔ λ ⇑. 12 Inducing Jumps and Stochastic Volatility Via Time Changes • JDD models and in particular the JDCEV process assume that the prior to default process evolves continuously, with a single jump to default. • Even though models like the JDCEV process provide enough structure to capture the leverage effect and volatility skews, these can be enhanced by allowing jumps and stochastic volatility that better represent reality. • This current work allows a more general class of models via Time Changes: pre-default stock price is a process with jumps and stochastic volatility + jump to default 13 Introduction to Time Changes Tt • {Tt, t ≥ 0} is an increasing RCLL process starting at T0 = 0 and E [Tt ] < ∞. • We are interested in time changes with analytically tractable Laplace Transform (LT): £ −λTt ¤ L(t, λ) = E e <∞ – Lévy Subordinators: increasing Lévy processes (stationary and independent increments) with positive jumps and nonnegative drift. – Absolutely Continuous (A.C.) time changes: Z t Tt = Vu du 0 with such activity rate process Vu that the LT. is known in closed form. – Composed Time Changes, e.g., subordinate first, then do an A.C. time change. These time changes induce both Jumps + Stoch. Volatility 14 The Expectation Operator • To compute expectations, we can condition on the time change (recall that X and T are independent): ¯ £ ¤¤ ¤ £ £ ¯ Ex 1{ζ>Tt}f (XTt ) = Ex Ex 1{ζ>s}f (Xs) Tt = s • We need to calculate the expectation operator of the original diffusion process Xt : £ ¤ Ex 1{ζ>t}f (Xt) • To gain tractability we choose two important approaches: 1. Resolvent Operator, which is the LT of the expectation operator. Generalized methodology. 2. Spectral Repressentation, which is particular case when the infinitesimal generator is symmetric and the payoff is in the space of functions f ∈ L2 (D, m) 15 The Resolvent Operator Approach • For any Markov process X, the Resolvent Operator Rs is defined as the Laplace Transform of the Expectation Operator: Z ∞ £ ¤ −st (Rs f )(x) := e Ex 1{ζ>t} f (Xt ) dt 0 • Conversely, the Bromwich Laplace inversion formula reads: Z ²+i∞ £ ¤ 1 Ex 1{ζ>t}f (Xt) = est (Rs f )(x)ds 2πi ²−i∞ • Therefore, if we know the Resolvent we can solve for the Expectation and vice versa. • NOTICE. The time t enters in this expression only through the exponential es t .This is crucial for time changes! 16 The Spectral Approach • When X is a 1D diffusion, the generator G is a self-adjoint 2 operator (i.e., (Gu, R v) = (u, Gv)) in L (D, m) with the inner product (f, g) = D f (x)g(x)m(x)dx, where m(x) is the speed density of the 1D diffusion (σ is the diffusion coefficient and µ is the drift): R 2µ(x) 2 dx m(x) = 2 e σ2(x) . σ (x) • If f ∈ L2 (D, m), we can apply the Spectral Theorem to write down the spectral representation for the expectation operator: ∞ X ¤ £ Ex 1{ζ>t}f (Xt) = e−λn tcn ϕn (x), n=1 where λn and ϕn are the eigenvalues and eigenfunctions of −G: −Gϕ(x) = λn ϕ(x), and the expansion coefficients are: cn = (f, ϕn ) . If the spectrum is continuous, the integral takes place of the sum. 17 Time Changing in the Resolvent Approach • The expectation operator for the time changed process can now be expressed in terms of the resolvent of the original process X and the Laplace transform of the time change T : ¯ ¤¤ £ ¤ £ £ Ex 1{ζ>Tt}f (XTt ) = Ex Ex 1{ζ>s}f (Xs)¯ Tt = s ·Z ²+i∞ ¸ 1 = E esTt (Rs f )(x)ds 2πi ²−i∞ Z ²+i∞ £ ¤ 1 = E esTt (Rsf )(x)ds 2πi ²−i∞ Z ²+i∞ 1 = L (t, −s) (Rs f )(x)ds. 2πi ²−i∞ 18 Time Changing in the Spectral Approach • If f ∈ L2 (D, m) and the spectrum of G is discrete, we can write: Ex £ ∞ X ¤ £ ¤ 1{ζ>Tt}f (XTt ) = E e−λnTt cnϕn(x) n=1 = ∞ X L (t, λn ) cn ϕn (x), n=1 where L (t, λn ) is the Laplace Transform of Tt. If the spectrum is continuous, the integral takes place of the sum. • For f ∈ / L2 (D, m) one needs to use the Laplace transform approach. 19 Lévy Subordinators • A Lévy subordinator is a non-decreasing Lévy process {Tt , t ≥ 0} with positive jumps and non-negative drift with Laplace Transform (LT): L (t, λ) = E[e−λTt ] = e−tφ(λ) • with the Laplce exponent given by the Lévy-Khintchine formula: Z ¡ ¢ −λs φ (λ) = γλ + 1−e ν (ds) (0,∞) • Where the drift γ ≥ 0, the Lévy measure ν (ds) satisfies R (s ∧ 1) ν (ds) < ∞, and the transition probability πt (ds) (0,∞) is: Z e−λs π (ds) = e−tφ(λ) [0,∞) 20 Lévy Subordinators (Cont.) • T starts at T0 , drifts at rate γ ≥ 0, and experiences positive jumps controlled by ν (ds). • ν (ds) describes the arrival rates of jumps so that jumps of sizes in some Borel set A bounded away from zero R occur according to a Poisson process with intensity ν (A) = A ν (ds). R • If + ν (ds) < ∞, the subordinator is a compound Poisson proR R cess with intensity α = + ν (ds) and the jump size distribution R −1 is given by α ν. • Otherwise, it is an infinite activity jump process (infinite number of jumps in any finite time interval). 21 Examples of Subordinators • Lévy measure (3 parameter class): ν(ds) = Cs−Y −1 e−ηs ds C > 0, η > 0, Y < 1 Where Y controls the short term jumps, η the long term jumps and C is a general scaling factor. • For Y ∈ (0, 1): Tempered Stable subordinators – Special case Y = 1/2: Inverse Gaussian process – Limiting case Y = 0: Gamma process 22 Examples of Subordinators (Cont.) • For Y 6= 0 the Laplace exponent is given by: φ(λ) = γλ − CΓ(−Y )[(λ + η)Y − η Y ] • For Y = 0: φ(λ) = γλ + C ln(1 + λ/η) • The processes with Y < 0 are compound Poisson processes with gamma distributed jump sizes – Special case with Y = −1: compound Poisson process with exponential jumps: αλ ν(ds) = αηe−ηs ds, φ(λ) = γλ + λ+η • The processes with Y ∈ [0, 1) are infinite activity. 23 Inducing Jumps by Lévy Subordination • If the original process Xt is a diffusion and Tt is a subordinator with Lévy measure ν, the time changed process Yt = XTt acquires jumps with the state-dependent Lévy density: Z ∞ π Y (x, y) = p(t; x, y)ν(dt) 0 where p(t; x, y) is the transition probability density of the original process X from x to y in time t. • The default intensity for the process Y is: Z ∞ λY (x) = γλ(x) + Pd (x, t)ν(dt) 0 where Pd (x, t) is the probability of default by time t if started at x for the original process X. 24 Lévy Subordination Yt = XTt (Xt = Bt and Tt = t + CP P ) Background Process X(t) 0.6 Time Process 2 0.5 1.8 X(t) 0.4 0.3 1.6 0.2 0.1 1.4 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 Time T(t) Time Changed Process Y(t)=X(T(t)) Time T(t) 0 1.2 1 0.8 0.6 0.6 Y(t)=X (T(t)) 0.5 0.4 0.4 0.3 0.2 0.2 0.1 0 0 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 0 1.7 0.2 0.4 0.6 0.8 Time t Time (t) CPP parameters α = 4,1/η = 0.1 25 1 Absolutely Continuous Time Changes • Let {Zt , t ≥ 0} be a n-dimensional Markov process independent Rt of X. Define Tt = 0 V (Zu )du, where V (z) is some positive function. We are interested in analytically tractable LT: Rt −λ V (Zu )du 0 Lz (t, λ) = Ez [e ] • Example: The Cox-Ingersoll-Ross (CIR) process: √ dVt = κ(θ − Vt)dt + σV VtdWt with V0 = v > 0, rate of mean reversion κ > 0, long-run level θ > 0, and volatility σV > 0. The CIR Laplace Transform (CIR zero-coupon bond): Rt −λ V (Zu )du 0 Lv (t, λ) = Ez [e ] = A(t, λ)e−B(t,λ)v µ A(t, λ) = 2$e($+κ)t/2 ¶ 2κθ σ2 V , ($ + κ)(e$t − 1) + 2$ q 2λ(e$t − 1) 2 2 B(t, λ) = 2σ λ + κ , $ = V ($ + κ)(e$t − 1) + 2$ 26 Absolutely Continuous Time Changes √ κ = 4, θ = V0 = 2 and σv = 2 CIR Process 3 Time Process V(t) 2.5 2 1 1.5 0.9 1 0.8 0.5 0.7 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 Time (t) Processes X(t) vs Y(t)=X(T(t)) 0.5 Time T(t) 0 0.6 0.5 0.4 X(t) 0.3 X(T(t)) 0.2 0.4 0.1 0.3 0 0 0.2 0.1 0.2 0.3 0.4 0.5 Time t 0.1 0 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 Time (t) 27 0.6 0.7 0.8 0.9 1 Composite Time Changes • Consider a composite time change process: Tt = TT1t2 where Tt1 is a subordinator with Laplace exponent φ and Tt2 is an integral of some positive function of a Markov process with LT Lz (t, λ). • T is obtained by time changing a Lévy subordinator T 1 with an AC time change T 2 . • Conditioning on Tt2 , we have: E[e−λTt ] = E[e−Tt φ(λ)] = Lz (t, φ(λ)) 2 • Composing Levy and A.C. time changes induces both jumps and stochastic volatility! 28 A Credit-Equity Model • Model defaultable stock: St = 1{t<τd } eρt XTt . • X is a JDCEV: dXt = [µ + λ(Xt)]Xt dt + σ(Xt )Xt dBt, X0 = x > 0, σ(x) = axβ , λ(x) = b + c σ 2 (x) = b + c a2 x2β . ¯R o ¯ t • ζ is the lifetime of X. ζ = inf t ∈ [0, T0 ] ¯ 0 λ (Xu ) du ≥ e , e ≈ Exp(1) n • Tt is either a subordinator, an A.C., or a composite time change. • The default time τd is: τd = inf{t ≥ 0 : ζ ≤ Tt}. At τd the stock jumps to zero and stays there (default). 29 A Credit-Equity Model (Cont.) • Intensity λ has to be added in the drift of X to compensate for jump to zero, and ρ and µ are parameters to be selected to make the discounted process into a martingale: E[St2 |Ft1 ] = e(r−q)(t2−t1)St1 , t1 ≤ t2, where r and q are the risk-free rate and dividend yield. • If Tt is a subordinator, then µ can be arbitrary and ρ = r − q + φ(−µ). • If Tt is an A.C. time change, then µ = 0, ρ = r − q. 30 Pricing Corporate Bonds • Survival probability: 1 Q(τd > t) = Q(ζ > Tt) = 2πi Z ε+i∞ L(t, −s)(Rs 1)(x)ds, ε−i∞ where Q(ζ > t) is the survival probability of X with lifetime ζ. • A defaultable zero-coupon bond with unit face value, maturity t > 0, and recovery R ∈ [0, 1]: BR (x, t) = e−rt Q(τd > t) + Re−rt [1 − Q(τd > t)] = e−rt R + e−rt (1 − R)Q(τd > t). • In this case the function f (x) = 1 is NOT in L2 (D, m) thus we need to use the Resolvent Approach! 31 Resolvent • In this model the resolvent is available in closed form: Z ∞ Rs f (x) = Gs (x, y)f (y)dy 0 with the resolvent kernel (for µ + b > 0): Γ(ν/2 + 1/2 − k(s)) Gs (x, y) = (µ + b)Γ(1 + ν)y µ ¶c+1/2−β x −2β −2β eA(y −x ) y ×Mk(s), 2ν (A(x ∧ y)−2β )Wk(s), 2ν (A(x ∨ y)−2β ) where 1 + 2c ν−1 s µ+b ν= , k(s) = − , A= 2 , 2|β| 2 2|β|(µ + b) a |β| and Mk,m (z) and Wk,m (z) are the first and second Whittaker functions (related to the Kummer and Tricomi confluent hypergeometric functions). • Analytical tractability is due to a connection of the JDCEV process to Bessel processes. 32 Computing the Survival Probability • In this case the Laplace transform can be inverted analytically using the Cauchy Residue Theorem, yielding the survival probability for the process X: Q(ζ > t) = ∞ X e−(b+ωn)t n=0 1 ×A 2|β| xe−Ax −2β 1 F1 (1 Γ(1 + c/|β|)Γ(n + 1/(2|β|)) Γ(ν + 1)Γ(1/(2|β|))n! − n + c/|β|, ν + 1, Ax−2β ), where 1 F1 (a; b; x) is the Kummer confluent hypergeometric function and ω = 2|β|(µ + b). • To obtain the survival probability for the time changed process XTt , just replace e−(b+ωn)t with L(t, b + ωn)! 33 Pricing Stock Options • A call option with payoff (St − K)+ at t has no recovery if the firm defaults: C(x; K, t) = e−rt Ex[(eρt XTt − K)+ 1{τd >t} ]. • A put option with payoff (K − St)+ can be decomposed into two parts: the put payoff (K − St)+1{τd >t} , given no default by t, and a recovery payment equal to the strike K in the event of default τd ≤ t: P (x; K, t) = P0 (x; K, t) + PD (x; K, t), P0 (x; K, t) = e−rt Ex[(K − eρt XTt )+1{τd >t} ], PD (x; K, t) = Ke−rt [1 − Q(τd > t)]. • The recovery part of the put is a credit derivative that pays a fixed cash amount K at maturity t if and only if the underlying firm has defaulted by t. The put option contains an embedded credit derivative ⇒ UNIFIED CREDIT-EQUITY! 34 Pricing Put Options The put payoff is in L2 , and the spectral expansion is (k := e−ρt K): ∞ X P0 (x; K, t) = e(ρ−r)tEx [(k − XTt )+ 1{τd >t} ] = L(t, λn )cn ϕn (x), n=1 with the eigenvalues, eigenfunctions, and expansion coefficients (for µ + b > 0): λn = ωn + 2c(µ + b) + b, ω = 2|β|(µ + b), s (n − 1)!(µ + b)(2c + 1) −Ax−2β (ν) xe Ln−1 (Ax−2β ), Γ(ν + n) p ν/2+1 2c+1−2β A k Γ(ν + n) p cn = Γ(ν + 1) (µ + b)(2c + 1)(n − 1)! ¶ ¾ ½ µ c ¢ ¡ 1 − n, |β| +1 Γ(ν + 1)(n − 1)! |β| , × ; Ak−2β − Ak−2β Lν+1 2 F2 c n−1 ν + 1, |β| + 2 c + |β| Γ(ν + n + 1) ν ϕn (x) = A 2 where 2F2 is the generalized hypergeometric function and Ln(ν) are generalized Laguerre polynomials. 35 Numerical Results • Assume a background process {Xt, t > 0} following a JDCEV, and a composite time change Inverse Gaussian Process + CIR with the following parameters: JDCEV S a β c b r q 50 10 −1 0.5 0.01 0.05 0 CIR IG V θ σV κ γ η C 1 1 1 4 0 p8 2 2/π Note that γ = 0, thus the time changed process is a pure jump process!. 36 Implied Volatility Implied Volatility 61% 1/4 Implied Volatility 1/2 1 49% 2 3 37% 25% 13% 30 35 40 45 50 55 45 26.19 26.39 27.53 29.61 31.34 50 21.41 22.64 24.30 26.72 28.64 60 65 Strike Time/Strike 1/4 1/2 1 2 3 30 62.04 51.94 45.74 43.03 42.80 35 47.94 41.47 38.24 37.68 38.34 40 35.52 32.72 32.14 33.23 34.55 55 20.09 20.72 22.12 24.45 26.39 60 20.28 19.84 20.65 22.66 24.52 65 20.88 19.46 19.64 21.25 22.96 Implied volatility smile/skew curves as functions of the strike price. Current stock price level is 50, local volatility σ = aS β = 20% 37 Credit Spreads and Default Probability Credit Spreads Default Probability 80.0% 8.5% 70.0% 30 7.5% Probability of default Credit spreads 40 6.5% 50 60 5.5% 70 4.5% 3.5% 60.0% 50.0% 40.0% 30 40 30.0% 50 20.0% 2.5% 60 70 10.0% 1.5% 0.0% 0 5 10 15 20 25 30 Time to maturity (years) 35 40 45 50 0 5 10 15 20 25 30 Time to maturity (years) 35 40 45 Credit spreads and default probabilities as functions of time to maturity for current stock price levels S = 30, 40, 50, 60, 70. 38 50 Killing Rate Killing Rate (k) Killing Rate (k) 1.2 28 1 25 22 0.8 JDCEV 16 k(S,V) k(S,V) 19 CEV JDCEV 0.6 CEV 13 0.4 10 7 0.2 4 1 0 1 1.5 2 2.5 3 3.5 4 Stock Price (S) 4.5 5 5.5 6 6.5 7 7 11 15 19 23 27 31 35 39 43 Stock Price (S) Killing rate (default intensity) k(S,V) as a function of the stock price S when the activity rate is fixed at V = 1. The solid line is the default intensity in the model obtained by time changing the JDCEV process. The dashed line is the default intensity obtained by time changing the standard CEV process with b = c = 0 without the jump to default 39 47 Conclusions • Starting from an analytically tractable diffusion Xt and an increasing process Tt with known Laplace transform, the time changed process XTt is also analytically tractable. • Time changing diffusions with Lévy subordinators induces jumps, while time changing with A.C. time changes induces stochastic volatility and composite time changes induce both. • By pairing analytically tractable diffusions with killing with analytically tractable time changes we are able to generate analytically tractable Credit-Equity models with jumps, stochastic volatility, and default. • These Hybrid credit-equity models feature linkages between corporate credit spreads in the credit markets and implied volatility skews in the stock options markets. 40 THANK YOU! 41