Attribution Non-Commercial (BY-NC)

Als PDF, TXT **herunterladen** oder online auf Scribd lesen

28 Aufrufe

Attribution Non-Commercial (BY-NC)

Als PDF, TXT **herunterladen** oder online auf Scribd lesen

- Inappropriate Use of Cap and Trade
- Hernandez Nievera vs Hernandea
- Intro 2 Opt
- Derivatives Tips for the Week
- 08-24-16 edition
- Sample FRM Part 1 Answer
- Beginners Guide to Voluntary Market
- State of the Voluntary Carbon Markets 2010
- Ch18HullOFOD9thEdition
- Canadas CEO Elite 100FINAL
- Carbon Trading 0109 2
- Log Normal
- Sino Australia Prospectus
- Untitled
- Syllabus
- Final Project Report
- FINANCE 12
- Energy and Markets Newsletter 092311
- Structured
- Tips

Sie sind auf Seite 1von 7

markets suggests the urgent need for the development

of selective hedging techniques such as futures contracts

and option instruments. As a result, a valid price model is

required for pricing nancial instruments or projects whose value

derives from the future carbon dioxide (CO

2

) spot permit price. Due

to the recent introduction in the market of option-like instruments

for hedging purposes, various models were developed to approximate

the dynamics of spot prices for CO

2

emission allowances. Dierent

approaches to the modelling of European Union emission allowances

(EUAs) ourished in the literature, ranging from equilibrium models

considering one trading period, to models derived from empirical

studies based on the rst two phase periods.

1 Modelling the dynamics of CO

2

emissions

Presently, the relevant spot price history is still short and may be

distorted by potential one-time eects due to the markets immature

state. Nonetheless, various authors considered the qualitative and

quantitative properties of the data, based on time-series analysis and

distribution analysis of the time series, to devise pricing models.

However, as opposed to equilibriummodels, all these empirical models

do not yet consider the fundamental properties of permit contracts.

1.1 Abatement opportunities

One of the key understandings in economics is that in an ecient

market, the equilibrium price of emission allowances is equal to the

marginal costs of the cheapest pollution abatement solution. Hence,

at a given time t, a company emitting CO

2

has to decide whether

to invest in infrastructure to reduce fuel production or delay the

investment to a future time, and instead buy emission allowances.

However, the fact that a market-based approach leads to an ecient

allocation of abatement costs across dierent polluters strongly

depends on the assumption that any technological abatement

solution is perceived as a perfect substitute for emission allowances.

Unfortunately, this is no longer true in the presence of uncertainty

and Chao et al (1993) proved that most abatement technologies are

perceived as durable and irreversible investments compared with

emission allowances, which are seen to provide a greater exibility in

adapting to changing conditions.

Consequently, few options are available to the majority of

companies aected by emission trading and even fewer fall into

the list of short-term abatement possibilities. As a result, it is

reasonable to assume that companies optimise their cost function

by continuously adjusting their permit portfolio allocations and by

choosing the optimal permit amount to purchase or sell, considering

the payment of the penalty as an alternative to compliance (see

Chesney et al (2008)).

1.2 An equilibrium approach

In 2006, when insucient price history for the European Unions

emissions trading scheme (EU ETS) existed, Seifert et al (2006)

considered a theoretical equilibriummodel that incorporated the key

properties of the EU ETS, such as abatement and penalty costs, and

deduced the main properties of the resulting spot price process.

ese results should be accounted for when explicitly modelling a

CO

2

spot price process. Fehr et al (2007) showed that the success of

projects including a carbon nance component is determined by the

correct valuation of their returns, whose cashows are equivalent to

derivatives written on future carbon prices.

Further, they showed that the spot price must be positive and

bounded by the penalty cost and the cost of delivering any lacking

allowances. Later, this idea was pursued by other authors such as

Chesney et al (2008) and Carmona et al (2009a) to name just a

few. Assuming the existence of a single representative rm, they

considered a pollution process with an initial pollution level, initial

permit endowment and with corresponding boundary conditions.

In CO

2

equilibrium models, permit

prices are positive and bounded by the

penalty level. To obtain closed-form

solutions to the pricing of CO

2

derivatives,

DanielBloch models the permit price

as a function of a positive unbounded

process, and shows that there is no

equivalent probability measure whereby

the discounted spot price is a martingale

64 EnergyRisk.com March 2011

Cuttingedge

i

S

t

o

c

k

p

h

o

t

o

.

c

o

m

/

CARBON DERIVATIVES PRICING:

AN ARBITRAGEABLE MARKET

March 2011 EnergyRisk.com 65

Cuttingedge

Assuming that the companys pollution dynamics are exogenous

processes, they used that process to derive the price of the tradable

permit by minimising the companys total cost. at is, the tradable

permit is a function of the pollution process, which is the unique

source of risk in that model. It is interesting to note that since the

pollution process is not a tradable asset, its risk can not be hedged

away, and Chesney et al (2008) concluded that there is no need to

construct a risk-neutral probability measure for the pollution process.

ey are therefore implicitly assuming that the underlying asset is not

a tradable asset and that consequently, its discounted price does not

need to be a martingale under the risk-neutral measure.

However, the CO

2

permit is a tradable asset and to avoid

arbitrage when estimating its value one should construct an

equivalent probability measure such that the discounted spot price

is a martingale under that measure. In that setting, one can expect

either a shortage or a surplus situation at the end of the trading

period, meaning the company will either be holding worthless

permits or paying the penalty costs. Hence, in a wait-and-see

situation, the payout at maturity is that of an Asian call option

with a oating strike price so that emission allowances are option

contracts. e expected future permits net position is valued by

minimising the total cost of the rm, computed as the sum of the

cashows at the initial time and the potential penalties at the end

of the period. As a result, the spot price is a function of the penalty

level and the probability of a permit-shortage situation

S t,T ( ) = pe

r Tt ( )

P c

e

0,T ( ) > N F

t ( )

where c

e

(0, T) models the cumulative emissions in the trading

period [0, T], N is the amount of emission allowances handed out

by the regulator and p is the penalty cost. As the pollution process

is uncertain, the emission allowance price lies between zero and

the discounted penalty level S

t

[0, e

r(Tt)

p]. Each author species

a dierent process for the method of approximation of cumulative

emissions c

e

(0, t). In order to compute the probability P(.), each

author species the cumulative emissions with a dierent process

c

e

(0, t) = j

t

0

Y(s)ds where Y(.) is the emission rate following a particular

process. erefore, given the drift and diusion terms of the emission

rate one can deduce the CO

2

spot price. For ease of computation,

the emission rate is usually normally or lognormally distributed so

that the probability of a permit-shortage situation is the cumulative

distribution function of a standard normal random variable. However,

permit prices are inherently prone to jumps therefore, to enable

equilibrium models to reproduce these jumps, the distribution of the

emission rate must be modied accordingly (see Grull et al (2010)).

1.3 An empirical approach

Various authors have presented empirical studies using spot and

futures prices suggesting that CO

2

emission allowances price levels

are non-normally distributed with fat tails, and that the price

dynamics are nonstationary and exhibit abrupt discontinuous shifts.

For instance, Daskalakis et al (2006) considered various popular

jump-diusion processes from the equity market to approximate the

dynamics of CO

2

allowance contracts, using a maximum likelihood

(ML) approach to estimate the model parameters. ey showed

that the Merton model (1976) had the best performance in terms of

likelihood and parsimony. Further, Gagliardi (2009) stressed that

volatility exhibits clustering over time and proposed a Heston model

combined with jump components. eir ndings imply that the

EUAs have a proportional, non-meanreverting structure with jumps.

In their articles, they did not check for no-arbitrage conditions,

and considered the dynamics of the spot price under the historical

measure without compensating the drift for the jump process. As a

result, they performed analysis with an incorrect expected value of the

spot price. Unfortunately, these models do not take into consideration

the properties of the CO

2

spot price described in section (1.2). More

precisely, the processes involved do not consider the penalty function

and as a result are not bounded, which is in contradiction with the

characteristics obtained from the equilibrium models.

In addition, the analysis performed by the authors on futures

prices were done under the historical measure and the market

prices of risk were not considered. As we saw earlier, it can have

tremendous implications in the dynamics of the spot process,

resulting in misleading interpretation of market behaviour. To

illustrate our comments, we refer the readers to Paul et al (2010) who

presented another empirical study on the CO

2

spot prices where they

considered jump-diusion models and in particular the normal-

inverse Gaussian model. ese models were calibrated to EUA prices

ranging from January 2006 to March 2009 and model parameters

summarised in a table. In view of the market behaviour of the spot

prices during that period of time, the spot prices are clearly bounded

by the penalty costs, and the excess of permits allocated, combined

with a slower global economy, pushed prices down. It is therefore not

surprising that an unbounded process such as a geometric Brownian

motion exhibits a negative drift, with = 0.29.

Similarly, in the case of geometric Brownian motion with mean

reversion, the alpha term is negative so that the drift becomes

0.009( lnS), and for = 1, the drift is negative for most values

of the spot price. To conclude, in the geometric Brownian motion

with jumps, the jump size is given by

saut

= 2.28, so that in

all cases, the model parameters of the unbounded processes are

articially constrained so that their dynamics resemble those of

a bounded process. To remedy these drawbacks, we are going to

consider a spot process that take into consideration the properties of

the CO

2

emission allowances.

1.4 A risk-neutral approach

To ensure no-arbitrage in order to compute option prices, we need to

derive the dynamics of the spot price or that of the future price under

a risk-neutral measure. Carmona et al (2009b) addressed the problem

of risk-neutral modelling of emission permits and considered models

based on diusion martingales ending up with two values. ey

assumed a one-period setting with no banking allowed, and let the

carbon price at the compliance date be a random variable taking only

the values zero and p. Because of the digital nature of the terminal

allowance price, they focused on the event of non-compliance

and modelled the future price to match the terminal condition. It

amounts to modelling a hypothetical positive-valued random variable

I exceeding the boundary condition one at the end of the period.

To match the recent allowance price and the observed instantaneous

uctuation intensity, they assumed a deterministic volatility function

for the process I

t

, leading to a local volatility diusion process for

the allowance price. Further, they devised a non-compliance process

not hitting zero or one in a nite time with probability one, and as

a result they identied classes of martingales taking values in the

interval (0,1) for the normatised future price process. en, they

extended their model to a two-period setting by assuming that the

cap-and-trade system is terminated at the end of the second period.

66 EnergyRisk.com March 2011

Cuttingedge

2 An alternative approach

We propose to bridge the gap between theory and observed market

price behaviours by considering a stochastic dierential equation for

the carbon permit price that satises the fundamental properties of

the permit contracts. Rather than working directly with a bounded

process and focusing on the event of non-compliance, we are going

to consider a positive exogenous process bounded by the penalty

cost. Similarly to Carmona et al (2009b), the dynamics of the

normalised spot price take values in the interval (0, 1) but it is not

forced by a digital payout at the end of the trading period. In general,

the valuation of forward and futures contracts in the commodity

market can be divided into two groups. e rst group considers

a risk premium to derive a model relating short-term and long-

term prices, while the second group is closely linked to the cost and

convenience of holding inventories. In order to dene our model in a

non-arbitrage market, we are going to consider both approaches and

try to compute an equivalent probability measure.

2.1 Accounting for the penalty cost

We let [T

i 1

, T

i

] be the ith trading period and consider the penalty

costs K(T

i

) for lacking EUAs during the entire ith trading period.

Assuming multi-period trading, allowing for unlimited banking and

forbidding borrowing, we let the strike K(t) be piecewise constant

given by K(t) =_

n

i=1

k

i

I

[T

i

1, T

i

]

(t) where k

i

is a positive constant in the ith

trading period. Taking into consideration the properties of the spot

price described in section (1.2), we let ( X

t

)

t ~ 0

be a positive process

describing the unbounded spot price of emission allowances in the

range [0, ), and dene the CO

2

spot price S

t

as

S

t

= min X

t

, K t ( ) ( )

= X

t

X

t

K t ( ) ( )

+

= K t ( ) K t ( ) X

t ( )

+

(2.1)

where, the spot price S

t

is a positive process taking values in the interval

[0, K(t)]. We rst concentrate on modelling the spot price within the

single period of time [0, T]. If constant penalty costs p are paid at the

end of the trading period, their values at time t are K(t) =e

r(Tt)

p, while

if they are paid at any time t [0, T], their values remain constant.

Assuming the latter, the dynamics of the spot price are given by

dS

t

= d K X

t

( )

+

which is equivalent to the dynamics of a put option on the unbounded

process. e process X

t

has the predictable representation property,

which is not the case of the discontinuous process S

t

, so that the

market for options on the spot price is incomplete.

2.2 The dynamics of the spot price

Given the unbounded spot price X=(X

t

)

t [0, T]

, we consider the

derivative price S

t

= f (t, X

t

) written on the underlying X. We choose

to work with semimartingales, since in that framework stochastic

integration and nonlinear transformations are stable. Without loss of

generality, and with the aim of presenting our idea clearly, we let the

dynamics of the unbounded spot price X under the historical measure

P be given by this stochastic dierential equation (SDE)

dX

t

X

t

=

X

dt +

X

d

W

X

t ( ) (2.2)

where

X

is the historical drift and o

X

is the volatility of the

process, and satisfying the usual conditions for the SDE to have

a unique solution. We consider the convex function C(X

t

) of the

underlying price given by the call payout C(X

t

) =(X

t

K)

+

and apply

Tanakas formula to get its dynamic

dC X

t

( ) = I

X

t

K { }

X

X

t

dt +

+ I

X

t

K { }

X

X

t

d

W

X

t ( ) +

1

2

X

t

K ( )

X

2

X

t

2

dt

where o(.) is the Dirac function. In the case where the strike is a

function of time, we get the extra term I

{X

t

~K}

(dK(t)/dt)/dt. So, given

Equation (2.1), the dynamics of the spot price are

dS

t

= dX

t

dC X

t

( )

=

X

X

t

I

X

t

< K { }

dt +

X

X

t

I

X

t

< K { }

d

W

X

t ( )

1

2

X

t

K ( )

X

2

X

t

2

dt

where I

{X

t

<K}

=1 I

{X

t

~K}

. e volatility of the spot price o

S

=o

X

X

t

I

{X

t

<K}

is bounded and equal to zero when the unbounded process X is either

equal to zero or greater or equal to the strike, which is in accordance

with the properties of the theoretical spot price given in section

(1.2). However, it contradicts one of the fundamental properties of a

tradable asset, which states that the volatility must be almost surely

not zero to make the hedge possible (see Shreve (2004)). at is, if the

volatility vanishes, then the randomness of the Brownian motion does

not enter the spot, but it may still enter the derivative security, making

the spot price no-longer an eective hedging instrument. Hence, we

can directly conclude that the model for carbon trading is incomplete.

2.3 A no-arbitrage model

2.3.1 The risk premium approach

To avoid arbitrage, we want to define an equivalent probability

measure Q to the real-world probability P such that the discounted

stock price is a martingale under Q. Hence, we let S

_

t

=e

rt

S

t

be the

discounted spot price, which we dierentiate with respect to time to

get its dynamics under the historical measure as

dS

t

= re

rt

X

t

I

X

t

K { }

dt re

rt

KI

X

t

K { }

dt +e

rt

I

X

t

K { }

X

X

t

dt +

+e

rt

I

X

t

K { }

X

X

t

d

W

X

t ( )

1

2

e

rt

X

t

K ( )

X

2

X

t

2

dt

e standard approach would be to compare the annualised rate of

return of the spot price per unit of time

X

to a risk-free investment

and consider

X

r as the reference parameter. However, this is

no-longer the case, and instead we must consider

X

= r +

S

X

+r

K

X

t

I

X

t

K { }

I

X

t

K { }

+

1

2

X

t

K ( )

I

X

t

K { }

X

2

X

t

so that the market price of risk becomes

S

=

X

r r

K

X

t

I

X

t

K { }

I

X

t

K { }

1

2

X

t

K ( )

I

X

t

K { }

X

2

X

t

X

which is only dened if o

X

0 and X [r, K r] where r > 0, that is,

o

S

0. However, we saw in section (2.2) that o

S

vanishes on some

part of the domain. In other words, for the no-arbitrage condition

to apply, the market price of risk must satisfy the Novikov condition

(see Shreve, (2004)). Hence, the no-arbitrage condition is not

satised when the unbounded spot price X

t

is greater or equal to the

penalty cost K. As a result, we cannot construct the dynamics for

the spot price and use the change of measure theory to express it

under an equivalent probability measure. Hence, in that setting there

is no equivalent probability measure such that the discounted spot

price S

_

t

is a martingale and we must therefore consider an alternative

approach. In the next section we are going to see if we can apply the

convenience yield approach to solve the problem.

March 2011 EnergyRisk.com 67

Cuttingedge

2.3.2 The convenience yield approach

Assuming the existence of a benet or a cost attached to holding

one unit of the spot price, the dynamics of the spot price under the

historical measure P are

dS

t

= I

X

t

K { }

X

X

t

dt + I

X

t

K { }

X

X

t

d

W

X

t ( )

1

2

X

t

K ( )

X

2

X

t

2

dt

where q(t, T) = -

1

2

o

2

X

X

2

t

is a convenience yield paid at all time t when

X

t

= K. is is consistent with the results obtained by Borak et al

(2006) where they found that a high fraction of the yields could be

explained by the price level and volatility of the spot prices. Since

the extra drift term q

~

(t, T) =-

1

2

o(X

t

K)o

2

X

X

2

t

is just a function of the

unbounded process X

t

, there is no additional source of risk to that

of the Brownian motion

X

(t). Hence, a standard approach would

be to compare the annualised rate of return of the spot price per

unit of time

X

to a risk-free investment and consider

X

r as the

reference parameter. Once again, considering the weak form of the

no-arbitrage condition, we get

I

X

t

K { }

X

X

t

S

= I

X

t

K { }

X

X

t

rX

t

I

X

t

K { }

rKI

X

t

K { }

so that the market price of risk becomes

S

=

X

r r

K

X

t

I

X

t

K { }

I

X

t

K { }

X

which is only dened if o

X

0 and X[r, Kr] where r > 0. erefore,

the no-arbitrage conditions do not apply, since the Novikov condition

is not satised. One possibility would be to modify the dynamics in

Equation (2.2) by adding the drift term

r

K

X

t

I

X

t

K { }

I

X

t

K { }

.

However, even though the market price of risk would be bounded, the

dynamics of the unbounded process would no longer be dened.

2.4 An arbitrage model

Various pricing methods exist, some of which are based on hedging

arguments, on the law of large numbers or as actuaries know it

on the standard deviation principle

1

, to name but a few. Recognising

that the discounted CO

2

stock price is not a martingale under an

equivalent probability measure that is, assuming the existence

of arbitrage opportunities in the carbon market we can therefore

arbitrarily dene the drift in the CO

2

spot price. We can either

assume a market price of risk and price in the corresponding

measure or we can directly price in the historical measure.

Accordingly, one must rely on the actuarial pricing approach of

marked-to-model. In that sense, the pricing of carbon permits

derivatives is very sensitive to the assumptions made and the choice

of a model for the underlying process.

2.4.1 The Black-Scholes measure

Given the dynamics of the spot price in Section (2.3.2), together with

its associated market price of risk, we assume that the growth rate of the

spot price is not far fromthe risk-free rate. erefore, following Black &

Scholes (1973), the market price of risk of the spot price becomes

S

=

X

r

X

Hence, the dynamics of the spot price under the Q

-measure are

dS

t

= I

X

t

K { }

rX

t

dt + I

X

t

K { }

X

X

t

dt

+ I

X

t

K { }

X

X

t

d

W

X

t ( )

1

2

X

t

K ( )

X

2

X

t

2

dt .

Using the change of measure, the Brownian motion W

t

given by

dW

t

= d

W

t

+

S

dt

is a Q

the measure Q

become

dS

t

= I

X

t

K { }

rX

t

dt

1

2

X

t

K ( )

X

2

X

t

2

dt + I

X

t

K { }

X

X

t

dW

X

t ( ) (2.3)

One of the key assumptions in mathematical nance is that the market

price of risk is not specic to the traded asset but to its source of noise.

In our particular example, the unique source of noise of the spot price

is given by the Brownian motion

X

of the unbounded spot price. Even

though the unbounded spot price is not a tradable, one can not freely

specify its market price of risk with respect to a particular risk aversion,

as it has already been dened for the traded spot price. Hence, the

dynamics of the unbounded spot price under the measure Q

are

dX

t

X

t

= rdt +

X

dW

X

t ( )

which is the classical geometric Brownian motion used within the

Black-Scholes formula. In our setting, the dynamics of the discounted

spot price S

t

under the user-dened probability measure are

dS

t

= re

rt

KI

X

t

K { }

dt

1

2

e

rt

X

t

K ( )

X

2

X

t

2

dt

+e

rt

I

X

t

K { }

X

X

t

dW

X

t ( )

with extra drift term re

rt

KI

{X

t

~K}

dt. Since the spot rate and the

strike are positive, in this model, the growth rate is lower than the

risk-free rate. Using a deterministic setting, Rubin (1996) provided

a continuous time trading model for carbon permits and found that

the prices grow in equilibrium with risk-free interest rates. Later, as a

result of introducing uncertainty in Rubins model, Schennach (2000)

showed that the expected permit price growth rate was reduced, which

is also what we get in our model with the extra drift term q

~

(t, T) .

2.5 The forward price

Given the denition of the spot price in Equation (2.1), the forward

price under either the historical measure P or a measure Q

becomes

F t,T ( ) = E X

T

F

t

E X

T

K T ( ) ( )

+

F

t

= K T ( ) E K T ( ) X

T ( )

+

F

t

2

spot price has a xed upper bound given by

the penalty costs, the resulting forward price is an embedded option

on the unbounded price process equivalent to the strike minus a

discounted put option, and is therefore model-dependent. at is, the

forward price is a function of a European option on the unbounded

spot price and its associated volatility, and as such is no-longer linear.

Note, when the strike K is constant or time-dependent, as expected

in a single trading period, the option is a call or put option. However,

when the strike is stochastic, which could happen in multi-periods,

the option becomes an exchange option. Hence, given a process

for X

t

, we can compute the forward price at a given time, infer the

dynamics of the call or put option on X, and as a result, determine

1.

P(X) = E[X] +o[X], where o[X] corresponds to a risk premium

68 EnergyRisk.com March 2011

Cuttingedge

the dynamics of the forward price. One can choose the model of his

choice, and for simplicity of exposition, following the example given

in section (2.3.2) where X

t

is a geometric Brownian motion, we can

apply Its lemma to the function V(t, X

t

) and obtain the dynamics

dV t, X

t

( ) =

t

V t, X

t

( ) dt +

1

2

X

2

X

t

2

xx

V t, X

t

( ) dt +

x

V t, X

t

( ) dX

t

with the instantaneous volatility of the European option (the

forwardprice on S) being

x

V t, X

t

( )

V t, X

t

( )

X

t

X

which is bigger than the volatility o

X

of the spot price. Hence, in

that model, the holder of a forward contract must manage extra

volatility risk compared with empirical models that do not take

into consideration the penalty costs. Further, call option prices are

positive, convex functions of the underlying and are bounded by

V t, T; x, K ( ) x Ke

r T t ( )

( )

+

.

As a result, the forward prices in our model also exhibit these

properties. To conclude, in the forward market on carbon permits,

there is a single xed strike (the penalty cost) leading to an implied

term structure of volatility to be calibrated.

2.6 The option price

Since 2005, a CO

2

option market is slowly growing and attracting

a wide variety of industrials, utilities and nancial institutions of

various nature. is market satises the primary need of risk transfer

from those wishing to reduce the risk of permit shortage situation,

to those willing to accept it. As the tradable permit is an option

in disguise, an option on emission allowances should resemble a

compound option. at is, on the rst expiration date T

1

, the holder

has the right to buy a new call using the strike price K

1

where the

new call has expiration date T

2

and strike price K

2

. If we let the

current time be 0, the spot price is S, and C(S, ; K) denotes the

value of a call with time to expiry and strike price K, on the rst

expiration date T

1

, the value of a call on a call is

max K

1

, C S, T

2

T

1

; K

2

( )

= max C S, T

2

T

1

; K

2

( ) K

1

, 0

+ K

1

Letting P

_

(S, ; K) be the discounted put price and setting K

1

< K

2

,

the payout at maturity T

1

of a call option on the forward price is

max F T

1

, T

2

( ) K

1

, 0

( )

= max K

2

K

1

P X, T

2

T

1

; K

2

( ), 0

K min K, P X, T

2

T

1

; K

2

( )

where K

_

=K

2

K

1

. Similarly, in the case of a call option on the permit

price, as the option strike cannot be higher than the penalty cost, we

must have the constraint K

1

< K

2

, and the payout at maturity T is

max S

T

K

1

, 0 ( ) = max X

T

X

T

K

2

( )

+

K

1

, 0

= max K

2

K

1

K

2

X

T

( )

+

, 0

which simplies to

max S

T

K

1

, 0 ( ) = max X

T

K

1

, 0 ( ) max X

T

K

2

, 0 ( )

is is the payo of a call spread on the unbounded process X.

erefore, in our model, the call price on the CO

2

spot price is lower

than that of an empirical model, not taking into consideration the

penalty costs. In the case of a put option the payout at maturity is

max K

1

S

T

, 0 ( ) = K

1

X

T

( )

+

which is equivalent to a put option on the unbounded process. Again,

given the dynamics of the unbounded process X we can then price

the European options in the risk-neutral measure. In the call-option

case, the payo is a convex and concave function such that its price

depends on two dierent volatility levels and such that the notion of

skew becomes important. It is interesting to note that vanilla options

on emission allowances become exotic options when considering

the unbounded process. As a result, the choice of a model for the

unbounded process X

t

is important, and one should consider empirical

results to infer its dynamics. All price series analysed in the literature

present non-zero skewness and excess kurtosis with summary

statistics of the data revealing fat-tailed leptokurtic distributions and

non-normal returns. Hence, for tractability reasons, an ane jump-

diusion model could be a plausible candidate.

2.7 Results

To illustrate our purpose, we consider the underlying given in section

(2.4.1) where X

t

is a geometric Brownian motion in the Q

measure

that is, the Black-Scholes model. We compute both a call option

and a call spread option on X

t

with maturity T=1, where the latter

is a call option on S

t

in our model. As emission allowance prices are

characterised in the literature by high historical volatility, we let the

volatility and drift be respectively o

X

= 0.4 and r = 0.05. is example

conrms that, assuming the same drift, the call prices obtained

with a bounded spot process are lower than those obtained with an

unbounded process. Hence, to recover the same option prices behaviour

with a model, not taking into consideration the penalty costs, one must

decrease the drift term, possibly obtaining negative drift.

2.7.1 Phase I

As a way of demonstrating the eect of neglecting the penalty cost,

K

2

= 40 in Phase I, we price a series of call options on S

t

and use the

results to infer the historical drift that calibrate a call option on X

t

in the historical measure around the strike K

1

= 32. We present the

results in table 1 where we let the strike vary in the range [20, 40],

assume the spot price to be S(0) =20 and obtain the historical drift

= 0.05, which is in line

with the results found by

Paul et al (2010).

In the Black-Scholes

model, the probability

P(X

t

~ K

2

) depends on

the initial spot price,

the maturity, the drift

and the volatility, so

that depending on the

parametrisation of the

model, the dierences

between the call prices

and the call spread prices

vary in magnitude.

Fixing the model

parameters and allowing

for the initial spot price

to vary in the range

[5; 35], we repeat the

T1. Call options &call spread options

in Phase I with varying strike

Strike Call Call spread Histor. call

20 3.6045 3.3565 2.7639

22 2.8008 2.5527 2.0739

24 2.1611 1.9131 1.5468

26 1.6589 1.4109 1.1490

28 1.2687 1.0207 0.8512

30 0.9679 0.7198 0.6298

32 0.7373 0.4892 0.4658

34 0.5612 0.3132 0.3447

36 0.4272 0.1792 0.2553

38 0.3254 0.0773 0.1894

40 0.2480 0.0 0.1407

Historical drift is = 0.05 Source: Author

March 2011 EnergyRisk.com 69

experiment by computing dierent call options in both models that

is, on X

t

and S

t

under the Q

call prices on X

t

in the historical measure with drift = 0.01.

On one hand we compute at-the-money call options, while on the

other, call options with xed strike K

1

= 20 are considered. e results

are presented in table 2 where in both cases the dierences between

the call prices and the call-spread prices increase as the initial spot

price tends to the boundary of the domain that is, the penalty costs.

is is an important result when pricing options on carbon, but more

importantly, it is crucial when hedging call options, as observed CO

2

spot prices are presently low and the discrepancy small, but in the

event of an increase in spot prices, the hedging strategy constructed

with an unbounded process will become misleading.

2.7.2 Phase II

In Phase II, the penalty cost has been increased to K

2

= 100 and

from 2006 to 2009 we saw that observed carbon prices decreased

due to an excess of permits allocated and a slower global economy.

erefore, keeping the model parametrisation as in Phase I but with

the new penalty cost, the probability P(X

T

~ K

2

) in Phase II is smaller

than that in Phase I, and the dierences between the call and call-

spread options are greatly reduced. However, the latest report from

the World Meteorological Organization (2010) stated that the main

greenhouse gases have reached their highest concentration levels since

pre-industrial times despite the economic slowdown, increasing the

likelihood of a reduction in the permits allocated by governments,

which in turn would result in an increase in CO

2

spot prices.

Conclusion

We proposed a CO

2

permit price model consistent with the features

exhibited by the CO

2

equilibrium models. To remain close to

classical option pricing theory and obtain closed-form solutions, we

directly modelled CO

2

permit prices as a function of an exogenous

and positive unbounded process and introduced one-period

contingent claims in terms of such a dynamics. In that setting, the

permit price is not a martingale under an equivalent probability

measure, which is consistent with empirical ndings, but implies the

existence of arbitrage opportunities. Following a marked-to-model

approach, we considered arbitrarily chosen growth rates for the CO

2

spot permit price, and computed European call option prices.

Daniel Bloch, Universit Paris VI Pierre et Marie Curie, France

Email: daniel.bloch@mizuho-sc.com

The author is grateful to Nicole El Karoui, Monique Jeanblanc, Mark Davis, Paul Mills as

well as to the referee for their useful comments and suggestions

Cuttingedge

T2. Call options &call spreadoptions inPhaseI withvaryingspot

Spot ATMcall Call spread Histo. call Call K=20 Callspread Histo. call

5 0.9011 0.9011 0.7716 0.0004 0.0004 0.0002

10 1.8022 1.8014 1.5433 0.1240 0.1231 0.0888

15 2.7034 2.6716 2.3150 1.1382 1.1064 0.9119

20 3.6045 3.3565 3.0867 3.6045 3.3565 3.0867

25 4.5057 3.5840 3.8584 7.2441 6.3224 6.4701

30 5.4068 3.1303 4.6301 11.5745 9.2980 10.6267

35 6.3080 1.9187 5.4018 16.2575 11.8682 15.2060

Historical drift is = 0.01 Source: Author

References

Black F and M Scholes, 1973

The pricing of options and corporate liabilities

Journal of Political Economics, 81, pages 637659

Borak S, W Hardle, S Truck and R Weron, 2006

Convenience yields for CO

2

emission allowance futures contracts

Discussion Paper 2006076, Economic Risk, Berlin

Carmona R, M Fehr, J Hinz, A Porchet, 2009a

Market design for emission trading schemes

SIAM Review

Carmona R, J Hinz, 2009b

Risk-neutral modeling of emission allowance prices and option valuation

Preprint

Chao H, R Wilson, 1993

Option value of emission allowances

Journal of Regulatory Economics, 5, pages 233249

Chesney M, L Taschini, 2008

The endogenous price dynamics of emissionallowances: anapplicationtoCO

2

optionpricing

Working Paper 449, National Centre of Competence in Research Financial Valuation

and Risk Management, January

Daskalakis G, D Psychoyios and R Markellos, 2006

Modeling CO

2

emission allowance prices and derivatives: Evidence from the EEX

Working Paper, Athens University of Economics and Business

Fehr M and J Hinz, 2007

A quantitative approach to carbon price risk modeling

Working Paper, Institute for Operations Research, ETH Zentrum

Gagliardi V, 2009

European Union emission trading scheme: A model for valuation and hedging of

emission unit allowances derivatives

Master of Science in Economics & Business Administration, Copenhagen Business School

Grull G, R Kiesel, 2010

Pricing CO

2

permits using approximation approaches

Working paper. Institute of Energy Trading and Financial Services,

University of Duisburg-Essen

Jarraud M, 2010

WMO 2009 greenhouse gas bulletin

World Meteorological Organization: November

Merton R, 1976

Option pricing when underlying stock returns are discontinuous

Journal of Financial Economics, 3, pages 125144

Paul E, M Frunza and D Guegan, 2010

Derivative pricing and hedging on carbon market

CES working paper, Documents de Travail du Centre dEconomie de la Sorbonne

Rubin J, 1996

A model of international emission trading, banking and borrowing

Journal of Environmental Economics and Management, 31, pages 269286

Schennach S, 2000

The economics of pollution permit banking in the context of Title IV of the

1990 Clean Air Act Amendments

Journal of Environmental Economics and Management, 40, pages 189210

Seifert J, M Uhrig-Homburg and M Wagner, 2006

Dynamic behavior of CO

2

spot prices: A stochastic equilibrium model

Working paper, Universitt Karlsruhe, Karlsruhe

Shreve S, 2004

Stochastic calculus for nance

Springer-Verlag, New York

Reproducedwith permissionof thecopyright owner. Further reproductionprohibited without permission.

- Inappropriate Use of Cap and TradeHochgeladen vonjoejoe48
- Hernandez Nievera vs HernandeaHochgeladen vondemsanpedro
- Intro 2 OptHochgeladen vonSumeetGoel
- Derivatives Tips for the WeekHochgeladen vonDasher_No_1
- 08-24-16 editionHochgeladen vonSan Mateo Daily Journal
- Sample FRM Part 1 AnswerHochgeladen vonAditya Bajoria
- Beginners Guide to Voluntary MarketHochgeladen vonDenys Freitas Martins
- State of the Voluntary Carbon Markets 2010Hochgeladen vonPablo Alarcón
- Ch18HullOFOD9thEditionHochgeladen vonseanwu95
- Canadas CEO Elite 100FINALHochgeladen voncanadafcusa
- Carbon Trading 0109 2Hochgeladen vonprashantsingh0911
- Log NormalHochgeladen vonAlexir86
- Sino Australia ProspectusHochgeladen vonkkwlok
- UntitledHochgeladen vonGenability
- SyllabusHochgeladen vonnikhil_bajaj2001
- Final Project ReportHochgeladen vonbinupandey2020
- FINANCE 12Hochgeladen vonShayne Constantino
- Energy and Markets Newsletter 092311Hochgeladen vonchoiceenergy
- StructuredHochgeladen vonMark Ambrin
- TipsHochgeladen vonSravanthi Reddy
- Earnings Theory PaperHochgeladen vonPrateek Sabharwal
- Don’t Bet on Wall Street: The Financialization of Nature and the Risk to Our Common ResourcesHochgeladen vonFood and Water Watch
- Planting a Web 3Hochgeladen vonWajid Khan
- Derivatives Model 1Hochgeladen vonsbdhshrm146
- soeren_skov_hansen.pdfHochgeladen vonjojo
- 081 E KfW ZEW CO2 Barometer 2012 !!!Hochgeladen vonRemus Samoila
- OPTIONS PRICING-S.pptHochgeladen vonFamous Fanta
- 06 Nov Monday Stock Market Derivative ReportsHochgeladen vonElite Investment Advisory Services
- VDZ Activity Report 2003-2005Hochgeladen vonAnonymous Cxriyx9HIX
- RiskHochgeladen vonEsteban

- Cover Letters InstructionsHochgeladen vonObi Wana
- Localization UWB in BANHochgeladen vonMinh Hoang
- Kalman FilterHochgeladen vonDebi Prosad Dogra
- Satnav Toolbox2288 w1Hochgeladen vonMinh Hoang
- Homework Solutions 5Hochgeladen vonMinh Hoang
- Projects 1 2012Hochgeladen vonMinh Hoang
- eth-27335-01Hochgeladen vonMinh Hoang
- Cramer Raoh and Out 08Hochgeladen vonWaranda Anutaraampai
- Evb1000 Product Brief 3Hochgeladen vonMinh Hoang
- Thesis PhD Giovanni BellusciHochgeladen vonMinh Hoang
- Computation Time Comparison Between Matlab and C++ Using Launch WHochgeladen vonMinh Hoang
- Sweat ShopHochgeladen vonMinh Hoang
- Global Warming Prefinal HandoutHochgeladen vonMinh Hoang
- TP MarkovHochgeladen vonMinh Hoang
- Getting StartedHochgeladen vonXenogenic Medusa
- Cover LetterHochgeladen vonklumer_x
- CommSystemDesign_Midterm2012Hochgeladen vonMinh Hoang
- Csd 2012 Midtermexam(Hcmut)Hochgeladen vonMinh Hoang
- CSD 2010 FinalExamHochgeladen vonMinh Hoang
- ESD-Ch4Hochgeladen vonMinh Hoang
- ESD-Ch3Hochgeladen vonMinh Hoang
- ESD-Ch2-p2Hochgeladen vonMinh Hoang
- ESD-Ch2-p1Hochgeladen vonMinh Hoang
- ESD-Ch1Hochgeladen vonMinh Hoang
- ESD-Ch0Hochgeladen voniric911
- Midterm Exam SolutionsHochgeladen vonMinh Hoang
- Homework Solutions 6Hochgeladen vonMinh Hoang

- 49450139 Nokia Brand Equity AnalysisHochgeladen vonbrojas35
- omtimizaetion.pdfHochgeladen vonMalik Saqib
- Analysis of Incidents Reported to PNGRB From July 2013 to Dec 2014Hochgeladen vonsathish_iyengar
- IX-1.Post-Earthquake Pipeline Leak Detection TechnologiesHochgeladen vonImtiaz Ahmed
- Resolution of Vectors Using Triangle LawHochgeladen vonOteng Richard Selasie
- Dot Net Developer / .Net Developer - Sample Resume - CVHochgeladen vonsample.resumes.cv
- Csr HistoryHochgeladen vonaditya jain
- What Are Some Current Issues and Controversies About IronHochgeladen vonDinda Savira
- Saif La Ith Khalid m Fka 2009Hochgeladen vonSANTHA
- A Short Introduction BuddhismHochgeladen vongacungtu2000
- A Strategic Plan for Namibia Youth Credit Scheme - AuneHochgeladen vonIshmael Mc Gazza Mubwandarikwa
- Cognitive Radio Cognitive Network SimulatorHochgeladen vonChandra Mohanty
- List of Area Moments of Inertia - Wikipedia, The Free EncyclopediaHochgeladen vonalvia3
- Assignment NcmHochgeladen vonRoey Rubian
- the benefits of paideia seminar in the literacy classroomHochgeladen vonapi-348933956
- THE IMPACT OF POLITICAL SOCIALIZATION ON POLITICAL PARTICIPATION – A NIGERIAN VIEW POINTHochgeladen vonSteven Jones
- IIT JEEHochgeladen vonAneesh Dhamodaran
- Causes of the First World War TextHochgeladen vonGabriel Entwistle
- Feelings and FollowersHochgeladen vonIsabella Vell
- Pulp and Paper IndustryHochgeladen vonMushahid Ali
- Aerospace SelectorGuideHochgeladen vonjuanpalomo74
- Instruaire BrochureHochgeladen vonNestrami
- TJKM Transpotation Consultants-Traffic Engineering_RedactedHochgeladen vonL. A. Paterson
- GS2011 QP BiologyHochgeladen vonvishnukesavieam1
- Simmons & Armstrong Knowing Animals 2007Hochgeladen vonBlythe Tom
- Material RequiredHochgeladen vonNikhil
- MathematicsHochgeladen vonengrmairy
- Time Management by Ronald PartinHochgeladen vonStephanus Sigit Dwi Prasetyo
- Peoplelinkvc.comPeopleLink Cover Page Story 2016 - CIOReview-PeopleLink Cover Page Story 2016 - CIOReviewHochgeladen vonchakradhar
- Talent Management 2Hochgeladen vonOmkar Bapat

## Viel mehr als nur Dokumente.

Entdecken, was Scribd alles zu bieten hat, inklusive Bücher und Hörbücher von großen Verlagen.

Jederzeit kündbar.