Variance and Volatility Swaps in Energy Markets

Anatoliy Swishchuk1 Mathematical and Computational Finance Laboratory Department of Mathematics and Statistics University of Calgary 2500 University Drive NM Calgary, Alberta, Canada, T2N 1N4 E-mail: aswish@math.ucalgary.ca web page: http://www.math.ucalgary.ca/aswish Abstract. This paper is devoted to the pricing of variance and volatility swaps in energy market. We found explicit variance swap formula and closed form volatility swap formula (using Brockhaus-Long approximation) for energy asset with stochastic volatility that follows continuous-time GARCH (1,1) model (mean-reverting) (or Pilipovi one-factor model). Numerical exc ample is presented for AECO Natural Gas Index (1 May 1998-30 April 1999). Keywords: energy market, stochastic mean-reverting volatility, variance and volatility swaps, AECO Natural Gas Index

Introduction

Variance swaps are quite common in commodity, e.g., in energy market, and they are commonly traded. We consider Ornstein-Uhlenbeck process for commodity asset with stochastic volatility following continuous-time GARCH model or Pilipovic (1998) one-factor model. The classical stochastic process for the spot dynamics of commodity prices is given by the Schwartz model (1997). It is dened as the exponential of an Ornstein-Uhlenbeck (OU) process, and has become the standard model for energy prices possessing mean-reverting features. In this paper, we consider a risky asset in energy market with stochastic volatility following a mean-reverting stochastic process satisfying the following SDE (continuous-time GARCH(1,1) model): d 2 (t) = a(L 2 (t))dt + 2 (t)dWt , where a is a speed of mean revertion, L is the mean reverting level (or equilibrium level), is the volatility of volatility (t), Wt is a standard Wiener process. Using a change of time method we nd an explicit solution of this equation and using this solution we are able to nd the variance and volatility swaps pricing formula under the physical measure. Then, using the same argument, we nd the option pricing formula under risk-neutral measure. We
1

Research is supported by NSERC

applied Brockhaus-Long (2000) approximation to nd the value of volatil;ity swap. A numerical example for the AECO Natural Gas Index for the period 1 May 1998 to 30 April 1999 is presented. The continuos-time GARCH model has also been explioted by Javaheri, Wilmott and Haug (2002) to calculate volatility swap for S&P 500 index. They used PDE approach and mentioned (page 8, sec. 3.3) that it would be interesting to use an alternative method to calculate F (v, t and the other above quantities. This paper exactly contains the alternative method, namely, change of time method, to get varinace and volatility swaps. The change of time method was also applied by Swishchuk (2004) for pricing variance, volatility, covariance and correlation swaps for Heston model. The rst paper on pricing of commodity contracts was published by Black (1976).

Mean-Reverting Stochastic Volatility Model (MRSVM)

In this section we introduce MRSVM and study some properties of this model that we can use later for calculating variance and volatility swaps. Let (, F, Ft , P ) be a probability space with a sample space , -algebra of Borel sets F and probability P. The ltration Ft , t [0, T ], is the natural ltration of a standard Brownian motion Wt , t [0, T ], such that FT = F. We consider a risky asset in energy market with stochastic volatility following a mean-reverting stochastic process the following stochastic dierential equation: d 2 (t) = a(L 2 (t))dt + 2 (t)dWt , (1) where a > 0 is a speed (or strength) of mean revertion, L > 0 is the mean reverting level (or equilibrium level, or long-term mean), > 0 is the volatility of volatility (t), Wt is a standard Wiener process.

2.1
Let

Explicit Solution of MRSVM

Then, from (2) and (1) we obtain dVt = aeat ( 2 (t) L)dt + eat d 2 (t) = (Vt + eat L)dWt . (3)

Using change of time approach to the equation (3) (see Ikeda and Watanabe (1981) or Elliott (1982)) we obtain the following solution of the equation (3) t Vt = 2 (0) L + W (1 ), or (see (2)), 2 (t) = eat [ 2 (0) L + W (1 )] + L, t 2 (4)

 where W (t) is an Ft -measurable standard one-dimensional Wiener process, 1 t is an inverse function to t :

t = We note that

2 0

( 2 (0) L + W (s) + eas L)2 ds.

(5)

1 = 2 t
0

t ( 2 (0) L + W (1 ) + eas L)2 ds,

(6)

which follows from (5).

2.2

t Some Properties of the Process W (1 )

t We note that process W (1 ) is Ft := F1 -measurable and Ft -martingale. t Then E W (1 ) = 0. (7) t t Lets calculate the second moment of W (1 ) (see (6)): t E W 2 (1 ) = E < W (1 ) >= E1 t t t = 2 0 E( 2 (0) L + W (1 ) + eas L)2 ds s 2 at 2 2at = 2 [( 2 (0) L)2 t + 2L( (0)L)(e 1) + L (e2a 1) a t + 0 E W 2 (1 )ds]. s

(8)

From (8), solving this linear ordinary nonhomogeneous dierential equation with respect to E W 2 (1 ), t dE W 2 (1 ) t = 2 [( 2 (0) L)2 + 2L( 2 (0) L)eat + L2 e2at + E W 2 (1 )], t dt we obtain E W 2 (1 ) = 2 [( 2 (0)L)2 t We note, that
s t E W (1 )W (1 ) = 2 [( 2 (0) L)2 e 2 (ts)

e t 1 2L( 2 (0) L)(eat e t ) L2 (e2at e t ) + + ]. 2 a 2 2a 2

1 )

2L( 2 (0)L)(ea(ts) e a 2

2 (ts)

L2 (e2a(t) e 2a 2

2 (ts)

(9) ],

and the second moment for W 2 (1 ) above follows from (9). t

2.3

t Explicit Expression for the Process W (1 )

t It is turns out that we can nd the explicit expression for the process W (1 ). From the expression (see Section 3.1) Vt = 2 (0) L + W (1 ), t t we have the following relationship between W (t) and W (1 ) :
t

t dW (1 ) =
0

s [S(0) L + Leat + W (1 )]dW (t).

t It is a linear SDE with respect to W (1 ) and we can solve it explicitly. The solution has the following look:
t t W (1 ) = 2 (0)(eW (t) 2 1)+L(1eat )+aLeW (t) 2 t 2 2

eas eW (s)+
0

2s 2

ds.

(10) t It is easy to see from (10) that W (1 ) can be presented in the form of a linear combination of two zero-mean martingales m1 (t) and m2 (t) : W (1 ) = m1 (t) + Lm2 (t), t where m1 (t) := 2 (0)(eW (t) and m2 (t) = (1 e ) + ae
2t 2t 2

1)
2s 2

at

W (t) 2 t 0

eas eW (s)+

ds.

t Indeed, process W (1 ) is a martingale (see Section 3.2), also it is well-known that process eW (t) 2 and, hence, process m1 (t) is a martingale. Then the process m2 (t), as the dierence between two martingales, is also martingale.

2.4

Some Properties of the Mean-Reverting Stochastic Volatility 2 (t) : First Two Moments, Variance and Covariation

From (4) we obtain the mean value of the rst moment for mean-reverting stochastic volatility 2 (t) : E 2 (t) = eat [ 2 (0) L] + L. (11)

It means that E 2 (t) L when t +. We need this moment to value the variance swap. Using formulae (4) and (9) we can calculate the second moment of 2 (t) : E 2 (t) = (eat ( 2 (0) L) + L)2 + 2 e2at [( 2 (0) L)2 e
2 t 1

2L( 2 (0)L)(eat e t ) a 2

L2 (e2at e t ) ]. 2a 2

Combining the rst and the second moments we have the variance of 2 (t) : V ar( 2 (t)) = E 2 (t)2 (E 2 (t))2 = 2 e2at [( 2 (0) L)2 e
2 t 1 2 2

2L( 2 (0)L)(eat e t ) a 2

L2 (e2at e t ) ]. 2a 2

t From the expression for W (1 ) (see (10)) and for 2 (t) in (4) we can nd the explicit expression for 2 (t) through W (t) : t 2 (t) = eat [ 2 (0) L + W (1 )] + L = eat [ 2 (0) L + m1 (t) + Lm2 (t)] + L = 2 (0)eat eW (t)
2t 2

+ aLeat eW (t)

2t 2

t as W (s)+ s 2 ds, e e 0

(12) where m1 (t) and m2 (t) are dened as in Section 3.3. From (12) it follows that 2 (t) > 0 as long as 2 (0) > 0. The covariation for 2 (t) may be obtained from (4), (7) and (9): E 2 (t) 2 (s) = ea(t+s) ( 2 (0) L)2 + ea(t+s) { 2 [( 2 (0) L)2 e + + e
2L( 2 (0)L)(ea(ts) e a 2 at 2
2 (ts) 2 (ts) 1

( (0) L)L + e

2 2a(ts) e + L (e 2a 2 as 2

2 (ts)

]} (13)

( (0) L)L + L .

We need this covariance to value the volatility swap.

Variance Swap for MRSVM

To calculate the variance swap for 2 (t) we need E 2 (t) (see Chapter 2). From (11) it follows that E 2 (t) = eat [ 2 (0) L] + L.
2 Then ER := EV takes the following form: 2 ER := EV :=

1 T
T 0

E 2 (t)dt =
0

( 2 (0) L) (1 eaT ) + L. aT

(14)

Recall, that V :=

1 T

2 (t)dt.

Volatility Swap for MRSVM

To calculate the volatility swap for 2 (t) we need E V = E R and it means that we more than just E 2 (t) (see Chapter 2), because the realized

volatility R := get

V =

2 R . Using Brockhaus-Long approximation we then

V ar(V ) E V EV . 8(EV )3/2 V ar(V ) = EV 2 (EV )2 .

(15)

We have EV calculated in (14). We need (16)

From (14) it follows that (EV )2 has the form: (EV )2 = ( 2 (0) L) ( 2 (0) L)2 (1 eaT )L + L2 . (1 eaT )2 + 2 a2 T 2 aT (17)

Let us calculate EV 2 using (9) and (13): EV 2 = = +

(18)

+ + e

( (0) L)L + e

( (0) L)L + L ]dtds

After calculating the interals in the second, forth and fth lines in (18) we have: EV 2 = = + + +
T T 1 E 2 (t) 2 (s)dtds T2 0 0 2 (0)L)2 ( (1 eaT )2 a2 T 2 2 (ts) T T a(t+s) 1 e { 2 [( 2 (0) L)2 e 2 1 2 T 0 0 2 2 2a(ts) e 2 (ts) ) 2L( 2 (0)L)(ea(ts) e (ts) ) + L (e 2a 2 ]}dtds a 2 ( 2 (0)L)L ( 2 (0)L)L (1 eaT ) + (1 eaT ) + L2 . aT aT

(19)

Taking into account (16), (17) and (19) we arrive at the following expression for V ar(V ) : V ar(V ) = EV 2 (EV )2 = + = + +
2 (ts) T T a(t+s) 1 e { 2 [( 2 (0) L)2 e 2 1 T2 0 0 2 2 2a(ts) e 2 (ts) ) 2L( 2 (0)L)(ea(ts) e (ts) ) + L (e 2a 2 ]}dtds 2 a 2 (0)L T T a(t+s) 2 (ts) e (e 1)dtds T2 0 0 2 2L 2 ( 2 (0)L) T T a(t+s) a(ts) e (e e (ts) )dtds (a2 2 )T 2 0 0 2 T T a(t+s) 2a(ts) 2 L2 e (e e (ts) )dtds. (2a 2 )T 2 0 0

(20)

After calculating the three integrals in (20) we obtain: V ar(V ) = EV 2 (EV )2 = +

(21)

From (15) and (21) we get the volatility swap: V ar(V ) E V EV 8(EV )3/2 .

(22)

Mean-Reverting Risk-Neutral Stochastic Volatility Model

In this section, we are going to obtain the values of variance and volatility swaps under risk-neutral measure P , using the same arguments as in sections 3-7, where in place of a and L we are going to take a and L a a := a + , L L := aL , a +

where is a market price of risk (See section 3).

5.1

Risk Neutral Stochastic Volatility Model (SVM)

Consider our model (1) d 2 (t) = a(L 2 (t))dt + 2 (t)dWt . (23)

We want to nd a probability P equivalent to P, under which the process ert 2 (t) is a martingale, where r > 0 is a constant interest rate. The hypothesis we made on the ltration (Ft )t[0,T ] allows us to express the density of the probability P with respect to P. We denote this density by LT . It is well-known (see Lamperton and Lapeyre (1996), Proposition 6.1.1, p. 123), that there is an adopted process (q(t))t[0,T ] such that, for all t [0, T ], t 1 t 2 Lt = e 0 q(s)dWs 2 0 q (s)ds a.s. In this case,
T 1 T 2 dP = e 0 q(s)dWs 2 0 q (s)ds = LT . dP In our case, with model (17), the process q(t) is equal to

q(t) = 2 (t), where is the market price of risk and R. Hence, for our model LT = e
T 0

(24)

2 (u)dWu 1 2

T 0

4 (u)du

Under probability P , the process (Wt ) dened by

Wt := Wt +
0

2 (u)du

(25)

is a standard Brownian motion (Girsanov theorem) (see Elliott and Kopp (1999)). Therefore, in a risk-neutral world our model (23) takes the following look: d 2 (t) = (aL (a + ) 2 (t))dt + 2 (t)dWt , or, equivalently, d 2 (t) = a (L 2 (t))dt + 2 (t)dWt , where a := a + , L := aL , a + (26)

(27)

and Wt is dened in (25). Now, we have the same model in (26) as in (1), and we are going to apply our change of time method to this model (26) to obtain the values of variance and volatility swaps.

5.2

Variance and Volatility Swaps for Risk-Neutral SVM

Using the same arguments as in the previous section (where inplace of (4) we have to take (26)) we get the following expressions for variance and volatility swaps taking into account (27). For the variance swaps we have (see (14) and (27)): 1 := EV := T
T 0

2 E R

( 2 (0) L ) E (t)dt = (1 ea T ) + L . T a
2

(28)

For the volatility swap we obtain (see (22) and (27)) V ar E V E V 8(E V(V ) )3/2

(29)

5.9. Numerical Example: AECO Natural GAS Index (1 May 1998-30 April 1999)
We shall calculate the value of variance and volatility swaps prices of a daily natural gas contract. To apply our formula for calculating these values we 2 need to calibrate the parameters a, L, 0 and (T is monthly). These parameters may be obtained from futures prices for the AECO Natural Gas Index for the period 1 May 1998 to 30 April 1999 (see Bos, Ware and Pavlov (2002)). The parameters are the following: a 4.6488 Parameters L 1.5116 2.7264 0.18 8

For variance swap we use formula (14) and for volatility swap we use formula (22). From this table we can calculate the values for risk adjusted parameters a and L : a = a + = 4.9337, aL = 2.5690. a + For the value of 2 (0) we can take 2 (0) = 2.25. For variance swap and for volatility swap with risk adjusted papameters we use formula (28) and (29), respectively. L = and

References
L. P. Bos, A. F. Ware and B. S. Pavlov (2002): On a semi-spectral method for pricing an option on a mean-reverting asset, Quantitative Finance, Volume 2, 337-345. F. Black (1976): The pricing of commodity contracts, J. Financial Economics, 3, 167-179. O. Brockhaus and D. Long (2000): Volatility swaps made simple, Risk Magazine, January, 2000. R. Elliott (1982): Stochastic Calculus and Applications, Springer-Verlag, New York. N. Ikeda and S. Watanabe (1981): Stochastic Dierential Equations and Diusion Processes, Kodansha Ltd., Tokyo. A. Javaheri, P. Wilmott and E. Haug (2002): GARCH and volatility swaps, Wilmott Magazine, January. D. Lamperton and B. Lapeyre (1996): Introduction to Stochastic Calculus Applied to Finance, Chapmann & Hall. D. Pilipovic (1998): Energy Risk. Valuing and Managing Energy Derivatives, New York, McGraw-Hill. E. Schwartz (1997): The stochastic behaviour of commodity prices: implications for pricing and hedging, J. Finance, 52, 923-973. A. Swishchuk (2004): Modeling and valuing of variance and volatility swaps for nancial markets with stochastic volatilities, Wilmott Magazine, Technical Article No2, September Issue, 64-72. P. Wilmott, S. Howison and J. Dewynne (1995): The Mathematics of Financial Derivatives, Cambridge, Cambridge University Press. 9

P. Wilmott (2000): Paul Wilmott on Quantitative Finance, New York, Wiley.

Appendix: Figures
Figure 1 depicts variance swap (price vs. maturity) for AECO Natural Gas Index (1 May 1998 to 30 April 1999), using formula (14). Figure 2 depicts volatility swap (price vs. maturity) for AECO Natural Gas Index (1 May 1998 to 30 April 1999), using formula (22). Figure 3 depicts varinace swap with risk adjusted parameters (price vs. maturity) for AECO Natural Gas Index (1 May 1998 to 30 April 1999), using formula (28). Figure 4 depicts volatility swap with risk adjusted parameters (price vs. maturity) for AECO Natural Gas Index (1 May 1998 to 30 April 1999), using formula (29). Figure 5 depicts comparison of adjusted (green line) and non-adjusted price (red line) (naive strike vs. adjusted strike). Figure 6 depicts convexity adjustment. Its decreasing with swap maturity (the volatility of volatility over a long period of time is low).

10

Fig. 1: Variance Swap

Fig. 2: Volatility Swap

Fig. 3: Variance Swap (Risk Adjusted Parameters)

Fig. 4: Volatility Swap (Risk Adjusted Parameters)

Fig. 5: Comparison: Adjusted and Non-Adjusted Price

Fig. 6: Convexity Adjustment

11