Arbitrage Models of Commodity Prices

Andrea Roncoroni
ESSEC Business School, Paris - Singapore

Arbitrage models of commodity prices are relative pricing devices. They allow to evaluate
commodity-linked securities in terms of quoted commodity prices or indices. Model selection
mainly depends on the level of development characterizing the market under consideration. For
instance, oil markets are rather mature; they provide traders with a large quantity of reliable
quotes for futures as well as vanilla options; these prices may thus be considered as model
primitives from which prices of more complex positions can be derived. On the contrary, several
electricity markets quote spot prices only, which are to be taken as primitives for valuation
purposes. We consider four major classes of commodity price models.
Long-term investments in natural resources can be interpreted as real (vs. …nancial) options
written on the value of the corresponding commodity price(s). This viewpoint leads to the key
notion of Net Convenience Revenue (NCR) de…ned as the net sum of all bene…ts (e.g., saved
costs of shortage, consumption value) and costs (e.g., storage expenses, replacement costs for
perishable goods) stemming from physically holding a storable asset over a time period (Brennan
(1991)).
For the purpose of pricing forward contracts, which represent the simplest class of commodity
derivatives, NCR must be computed net of the risk-free rate of interest on the currency of
denomination, which in turn represents the purely …nancial cost of funding the operation. This
results into a quantity referred to as the Cost of Carry (CoC). The underlying argument goes
as follows. We consider an evaluation time t for a forward contract delivering a single unit of
a commodity S at a future time T for a price x (t). It is assumed that holding the asset leads
to an overall bene…t B and bears a total cost C, both quantities expressed in units of currency
at time T . A short forward position at time t leads to a time T pay-o¤ equal to S (T ) +x (t).
This cash ‡ow can be hedged by borrowing the present value of delivery price x (t), then buying
the underlying asset, and …nally o¤setting the resulting net bene…t B C using the proceeds
from borrowing funds equal to the present value of that amount. Table 1 illustrates the resulting
replicating strategy , which is commonly referred to as Cash-and-Carry. There, Vf denotes the
forward contract value and r is the continuously compounded short rate of interest, which we
assume to be constant over time. The net cash ‡ow V (T ) of this trading strategy at delivery
is zero. Barring arbitrage opportunities, the time t value V (t) of the resulting position is zero
as well. This fact leads to a formula for commodity forward value Vf :
r(T t)
V (t) = 0 , Vf (t) = S (t) [x (t) + B C] e :

The corresponding time t forward price is de…ned as the unique delivery price negotiated at
time t and making the value of the forward contract equal to zero. Solving equation Vf (t) = 0

Electronic copy available at: http://ssrn.com/abstract=1274050

 Strategy value at t Value at T
1 short fwd V f (t) S (T ) +x (t)
Borrowing $ [x (t) +B] e r(T t) x (t) B
Lending $ Ce r(T t) +C
1 long asset +S (t) +S (T ) +B C
r(T t)
Portfolio V (t) = V f (t) [x (t) +B C] e +S (t) V (T ) = 0

Table 1: Replicating strategy for a commodity forward contract.

in the unknown delivery price x (t) leads to a formula for the arbitrage-free forward price:

fT (t) = S (t) er(T t)

B + C: (1)

If side bene…ts and costs are represented using a continuously compounded rate of appreciation
a and, respectively, depreciation d, then we may write:

fT (t) = S (t) e[r (a d)](T t)

: (2)

The di¤erence a d de…nes a constant Convenience Yield (CY).

The Spot-Forward Parity (2) (resp.(1)) results from arbitrage considerations under the as-
sumption that:
1) A spot price is quoted;
2) Forward delivery occurs at one single point in time;
3) The underlying asset is storable.
In contrast to hypothesis 1, some markets (e.g., oil) do not quote prices for spot delivery. In
this case, we may either identify spot with the shortest maturity forwards, or simply avoid
considering spot prices and focus on forward curve models. As opposed to assumption 2, a few
markets (e.g., electricity and shipping freight) quote forwards for delivery over a whole time
period. A possible way out consists of …nding implied forward prices compatible to the existing
market quotes (Fleten and Lemming (2003), Koekebakker and Os Adland (2004), Benth et al.
(2007)). As far as assumption 3 is concerned, some commodities (e.g., electricity and meat)
cannot be stored or bear a perishing process. These facts prevent traders from performing a
cash-and-carry strategy in the way described above, what may lead to deny the validity of the
SF parity. However, asset storability is not actually required to derive this relationship. One
simply needs for being able to enter a commitment whereby the asset is delivered at the standing
spot price. This is the case, for instance, of a trader owning a production unit with available
capacity on the date of interest.
If any between the time value of money and the convenience revenue is stochastic, one has
to rely on a general pricing formula whereby the value process of a forward contract is the
risk-neutral conditional expectation of the discounted future cash ‡ow S (T ) x (t):
RT
Vf (t) = EPt e t
r(u)du
(S (T ) x (s)) , (3)

for issuing time = s evaluation time t T = delivery time. By setting the expression (3)

Electronic copy available at: http://ssrn.com/abstract=1274050

equal to zero, we have a general spot-forward parity:
RT
r(s)ds
EPt e t S (T )
fT (t) = ; (4)
PT (t)

where PT (t) stands for the time t zero-coupon bond price expiring at time T . The underlying
convenience revenue appears implicitly via the risk-neutral drift of spot price dynamics, which is
given by the instantaneous CoC r (t) c (t) (Du¢ e (2001)). Here the CY need not be constant.
We refer to this quantity at the Instantaneous Convenience Yield (ICY). A more explicit ex-
pression is derived in Jamshidian (1993) under deterministic covariance between the spot price
and an exponentially integrated convenience yield processes.
Arbitrage pricing theory implies that drift must equal its historical counterpart plus the
spot price volatility times a market price of risk (Björk (2003)). The latter explicitly appears,
for instance, in the risk-neutral dynamics of the convenience yield and can be estimated using
methods proposed by Benth et al.(2008) and Kolos and Ronn (2008), among others. We also
note that formula (4) implies that forward prices are martingales under the forward-neutral
probability, namely the one making gain processes martingales once they are expressed in units
of zero-coupon bonds (Jamshidian (1987), Jarrow (1987), and, for general numéraires, Geman
et al.(1995)). It can also be shown that futures prices are martingales under the risk-neutral
probability which, in the case of deterministic time value of money, matches the forward-neutral
probability.
Gibson and Schwartz (1990) propose a model where commodity prices are driven by two
factors: one factor is idiosyncratic to spot price returns; the other a¤ects price evolution through
the ICY process. The resulting P -dynamics read as:

dS(t) = [r c(t)]S(t)dt + 1 S(t)dW1 (t) ; (5)

dc(t) = [ c(t)]dt + 2 dW2 (t) ;

where 1 (resp., 2 ) represents the price return (resp., ICY) volatility, denotes the ICY mean
1
reversion frequency, = + 2 (resp. ) indicates the risk-neutral (resp. historical) long
term ICY level, and the two P -Brownian motions W1 and W2 are assumed to be correlated
with coe¢ cient .
This model represents the prototype of the Spot-Convenience Yield Model class (SC models).
Jamshidian and Fein (1990) and Bjerksund (1991) show that forward prices in this model are
exponentially a¢ ne functions of the pair (ln S (t) ; c (t)). This feature is shared by most SC
models, a fact leading Bjork and Landen (2001) to pursue a systematic study of a¢ ne models
for commodity prices in the general setting of jump-di¤usions.
SC models have been extensively studied in a number of papers, such as Schwartz (1997),
Deng (1998), Eydeland and Geman (1998), Schwartz (1998), Schwartz and Smith (2000), Sorensen
(2002), Nielsen and Schwartz (2004), Casassus and Collin-Dufresne (2005), and Korn (2005),
among others. Extensions to jump-di¤usions can be found in Hilliard and Reiss (1998), Lien and
Strom (1999), and Crosby (2008), whereas Richter and Sorensen (2002) and Trolle and Schwartz
(2006) examine the e¤ect of introducing a stochastic volatility component.
SC models are intuitive in terms of economic interpretation and deliver analytically tractable
expressions. However, they experience all consequences stemming from limited observation of
their primitives. First, commodity markets do not quote ICY’s. Consequently, …ltering tech-
niques (e.g., Kalman) are required to estimate the corresponding process. However, these proce-
dures may deliver an unstable assessment of model parameters. This fact that can undermines
the signi…cance of hedging prescriptions resulting from a …tted model. Second, …tting quoted
forward curves or term structures of volatility and correlation requires calibration procedures
whose e¤ectiveness and numerical stability must be evaluated case by case. These issues lead
some authors to propose an alternative modelling framework, the so-called Forward Model class
(FD models).
Reisman (1991) and, independently, Jamshidian (1992) directly model futures prices under
the risk-neutral probability as:
(
dFT (t)
FT (t) = T (t)dW (t) ; (6)
FT (t0 ) = g (T ) ;

for t0 t T . Here W is an n-dimensional standard Brownian motion and g denotes the

standing curve. If interest rates are deterministic, model (6) can be used to describe forward
price dynamics as well, a fact of particular importance in view of the observation that interest
rate volatility is strongly dominated by commodity price volatility (Schwartz (1997)). Any
particular instance of this class results from assigning a value to each of the following three
quantities:
1) A futures/forward curve g quoted at time t0 ;
2) A number n of statistically independent driving factors;
3) A volatility structure ( T (t); t0 t T ).
FD models have been studied in Cortazar and Schwartz (1994), Clewlow and Strickland (2000),
Audet et al.(2002), and Andersen (2008), among others. Benth and Koekebakker (2008) directly
model electricity prices referring to forward contracts with delivery spanning a whole time period.
Id Brik and Roncoroni (2008) provide a statistically robust estimation procedure for FD models.
The main bene…t from using FD models is that primitives are all observable quantities,
possibly up to deterministic transformation of market price data. Moreover, FD models are
very intuitive and ‡exible from a trader’s perspective. For instance, they easily allow the user
to identify noise terms with a number of forward curve deformations resulting from a statistical
analysis of historical market moves (Guiotto and Roncoroni (2001) and Galluccio and Roncoroni
(2006)).
Spot dynamics implied by FD models as S (t) := Ft (t) need not be Markovian processes. A
simpler version of this problem arises in the Heath, Jarrow and Morton (1992) theory (HJM) of
interest rates and has been investigated by several authors since the pioneering work of Jamshid-
ian (1991) and Carverhill (1992). In the context of commodity models, Jamshidian (1992) de-
rives conditions on the volatility structure ensuring that the resulting spot price dynamics are
Markovian. Unfortunately, this and subsequent results pursued within the HJM setting cannot
be applied to most multi-dimensional commodity price models, including SC models. Roncoroni
and Id Brik (2008) …rst pursue a systematic study on the correspondence between SC models and
FD models, obtaining necessary and su¢ cient conditions for Markovianity of the pair (ln S; c)
and a su¢ cient condition for a general class of multi-dimensional FD models. Alternatively,
by using a technique introduced in Cheyette (1991) for interest rates models, Andersen (2008)
disentangles abstract Markovian factors underlying forward price dynamics.
 A third class of models has been proposed by Miltersen and Schwartz (1998). These authors
consider the notion of Instantaneous Forward Convenience Yield (IFCY) cT (t) as implicitly
de…ned through: RT
FT (t) = S (t) e t cs (t)ds ; (7)
for all T t. Clearly, IFCY represents NCR according to a continuously compounded accrual
rule. It can be shown that the process ct (t) := limT #t cT (t) is actually the ICY as de…ned as the
formal “instantaneous dividend rate” process c (t) entering the risk-neutral drift in expression
(5). More precisely, Bjork and Landen (2001) show that:
RT
CovPt e t
r(u)du
S (T ) ; c (T )
cT (t) = EPt (c (T )) + RT ;
r(u)du
EPt e t S (T )

from which we deduce that ct (t) = c (t). Although formula (7) de…nes cT (t) is terms of forward
and spot prices, for modelling purposes it is actually used the other way around: processes
S and cT are de…ned as primitives for a set of maturities T , then forward prices are derived
using expression (7). Also, drift restrictions on IFCY dynamics can be derived. One of the few
applications of this setting is the calibration procedure developed in Miltersen (2003).
FCY models su¤er from similar drawbacks as those a¤ecting SC models. Again, spot prices
may be unavailable and FCY is more a mental abstraction than an observable quantity. Also, this
setting may experience the Sydney Opera House e¤ ect described in Rebonato (1996): computing
a quoted FCY by di¤erentiating a term structure of forward prices as …tted to observed quotes
may generate a curve exhibiting pronounced spikes due to the instability of the di¤erential
operator.
A …nal modelling framework is perhaps the simplest and widest spread one. The idea is to
focus on the …ne properties of a particular class of commodity spot prices without assuming
any speci…c structure about the underlying convenience yield. This restrictive assumption is
partly o¤set by the ‡exibility introduced upon selecting an appropriate spot price process for
modelling purposes. Spot models for energy price dynamics include Black (1976), Jarrow (1987),
Deng (1999), Barlow (2002), Escribano et al.(2002), Lucia and Schwartz (2002), Roncoroni
(2002), Benth et al. (2003), Huisman and Mahieu (2003), Burger et al.(2004), Cartea and
Figueroa (2005), Ribeiro and Hodges (2005), de Jong (2006), Geman and Roncoroni (2006),
Andersen (2007), and Fusai et al.(2008), among others. Eydeland and Wolyniec (2003), Fusai
and Roncoroni (2008), and Pilipovic (2007) report a large number of references on one-factor
commodity price modelling.
At the time of writing, there seems to be no general consensus on a benchmark setting for
the purpose of pricing commodity-linked derivatives. As a rule, the …nal user should focus on
the following items:
1) Primitives. They should be selected within the set of market quantities for which reliable
data are available.
2) Structural elements. A proper form for price drift, volatility, and jump components, if
any, should be identi…ed using statistical analysis of historical data and then …tted to observed
prices.
3) Driving noise terms. Number and nature of underlying noise terms should be assessed
based on historical price analysis (e.g., Principal Components Analysis) and jump …ltering (e.g.,
Roncoroni (2002) and Duan and Fulop (2007)).
 References
Andersen, L. (2008). Markov Models for Commodity Futures: Theory and Practice. WP, Banc
of America Securities.
Audet, N., Heiskanen, P, Keppo, J., Vehvilainen, I. (2002). Modelling of Electricity Forward
Curve Dynamics. In: Bunn D.W. (Ed.), Modelling Prices in Competitive Electricity Markets,
Wiley and Sons, 251-265.
Barlow, M.T. (2002). A Di¤usion Model for Electricity Prices. Mathematical Finance 12, 287-
298.
Benth, F.E., Cartea, À., Kiesel, R. (2008). Pricing Forward Contracts in Power Markets by the
Certainty Equivalence Principle: Explaining the Sign of the Market Risk Premium. Journal of
Banking and Finance 32(10), 2006-2021.
Benth, F.E., Koekebakker, S. (2008). Stochastic Modeling of Financial Electricity Contracts.
Energy Economics 30(3), 1116-1157.
Benth, F.E., Ekeland, L., Hauge, R., Nielsen, B.F. (2003). A Note of Arbitrage-Free Pricing of
Forward Contracts in Energy Markets. Applied Mathematical Finance 10, 325-336.
Benth, F.E., Koekebakker, S., Ollmar, F. (2007). Extracting and Applying Smooth Forward
Curves from Average-based Commodity Contracts with Seasonal Variation. Journal of Deriva-
tives 15(1), 52-66.
Bjerksund, P. (1991). Contingent Claims Evaluation when the Convenience Yield is Stochastic:
Analytical Results. Unpublished Manuscript.
Björk, T., Landen, C. (2001). On the Term Structure of Futures and Forward Prices. In:
Geman et al. (Eds.), Mathematical Finance - Bachelier Congress 2000, Springer Verlag, Berlin-
Heidelberg-New York.
Björk, T. (2003). Arbitrage Pricing in Continuous Time. Oxford University Press.
Black, F. (1976). The Pricing of Commodity Contracts. Journal of Financial Economics 3,
167-179.
Brennan, M. (1991). The Price of Convenience and the Valuation of Commodity Contingent
Claims. In: Lund D., Øksendal, B. (eds.). Stochastic Models and Option Values. Elsevier
Science.
Burger, M., Klar, B., Müller, A., Schindlmayr, G. (2004). A Spot Market Model for Pricing
Derivatives in Electricity Markets. Quantitative Finance 4(1), 109-122.
Cartea, Á., Figueroa, M.G. (2005). Pricing in Electricity Markets: A Mean Reverting Jump
Di¤usion Model with Seasonality. Applied Mathematical Finance 12(4), 313-335.
Carverhill, A. (1992). When is the Short Rate Markovian? Mathematical Finance 4(4), 305-312.
Casassus, J., Collin-Dufresne, P. (2005). Stochastic Convenience Yield Implied from Commodity
Futures and Interest Rates. Journal of Finance 60, 2283-2331.
Cheyette, O. (1991). Markov Representation of the Heath-Jarrow-Morton Model,”WP, BARRA.
Clewlow L., Strickland, C. (2000). Energy Derivatives: Pricing and Risk Management. Lacima
Group, Sydney.
Cortazar, G., Schwartz, E.S. (1994). The Valuation of Commodity Contingent Claims. Journal
of Derivatives 1, 27-39.
Crosby, J. (2008). A Multi-Factor Jump-Di¤usion Model for Commodities. Quantitative Fi-
nance 8, 181-200.
de Jong, C. (2006). The Nature of Power Spikes: A Regime-Switch Approach. Studies in
Nonlinear Dynamics & Econometrics 10, article 3.
Deng, S. (1998). Stochastic Models of Energy Commodity Prices and Their Applications: Mean
Reversion with Jumps and Spikes. WP, Georgia Institute of Technology.
Duan, J-C, Fulop, A. (2007). How Frequently Does the Stock Price Jump? An Analysis of
High-Frequency Data with Microstructure Noises. WP, ESSEC.
Du¢ e, D. (2001). Dynamics Asset Pricing Theory. Princeton University Press.
Eydeland, A., Geman, H. (1998). Pricing Power Derivatives. RISK, September.
Eydeland, A., Wolyniec, K. (2002). Energy and Power Risk Management: New Developments
in Modeling, Pricing and Hedging. Wiley&Sons.
Escribano, Á., Peña, J.I., Villaplana, P. (2002). Modeling Electricity Prices: International
Evidence. WP 02-27(8), Universidad Carlos III, Madrid.
Fleten, S.E., Lemming, J. (2003). Constructing Forward Price Curves in Electricity. Energy
Economics 25, 409–424.
Fusai, G., Marena, M., Roncoroni, A. (2008). Analytical Pricing of Discretely Monitored Asian-
Style Options: Theory and Application to Commodity Markets. Journal of Banking and Finance
32(10), 2033-2045.
Fusai, G., Roncoroni, A. (2008). Implementing Models in Quantitative Finance: Methods and
Cases. Springer Finance.
Galluccio, S., Roncoroni, A. (2006). A New Measure of Cross-Sectional Risk and Its Empirical
Implications for Portfolio Risk Management. Journal of Banking and Finance 30(8), 2387-2408.
Geman, H., El Karoui, N., Rochet, J.C. (1995). Changes of Numéraire, Changes of Probability
Measure and Option Pricing. Journal of Applied Probability 32, 443-458.
Geman, H., Roncoroni, A. (2006). Understanding the Fine Structure of Electricity Prices.
Journal of Business 79, 1225-1262.
Gibson, R., Schwartz, E.S. (1990). Stochastic Convenience Yield and the Pricing of Oil Contin-
gent Claims. Journal of Finance 45(3), 959-976.
Guiotto, P., Roncoroni, A. (2001). Theory and Calibration of HJM with Shape Factors. In:
Geman et al. (Eds.), Mathematical Finance - Bachelier Congress 2000, Springer Verlag, Berlin-
Heidelberg-New York, 407-426.
Heath D., Jarrow R.A., Morton, A.J. (1992). Bond Pricing and the Term Structure of Interest
Rates: A New Methodology for Contingent Claims Valuation. Econometrica 60, 77–105.
Hilliard, J.E., Reiss, J. (1998). Valuation of Commodity Futures and Options under Stochastic
Convenience Yields, Interest Rates, and Jump Di¤usion in the Spot. Journal of Financial and
Quantitative Analysis 33(1), 61-86.
Huisman, R., Mahieu, R. (2003). Regime Jumps in Electricity Prices. Energy Economics 25,
425-434.
Id Brik, R, Roncoroni, A. (2008). On the Estimation of Commodity Forward Price Models.
WP, ESSEC Business School.
Jamshidian, F. (1987). Pricing of Contingent Claims in the One-Factor Term Structure Model.
Working Paper, Merryl Linch, New York. In: Vasicek and Beyond (1996), Risk Publications.
Jamshidian, F. (1991). Bond and Options Evaluation in the Gaussian Interest Rate Model.
Research in Finance 9, 131-170.
Jamshidian, F. (1992). Commodity Option Valuation in the Gaussian Futures Term Structure
Model. Review of Futures Markets 10, 61-86.
Jamshidian, F., Fein, M. (1990). Closed-Form Solutions for Oil Futures and European Options
in the Gibson-Schwartz Model: A Comment. WP, Financial Strategies Group, Merrill Lynch
Capital Markets, New York.
Jamshidian, F. (1993). Option and Futures Evaluation With Deterministic Volatilities. Mathe-
matical Finance 3(2), 149-159.
Jarrow, R.A. (1987). The Pricing of Commodity Options with Stochastic Interest Rates. Ad-
vances in Futures and Options Research 2, 19-45.
Koekebakker, S., Os Adland, R. (2004). Modelling Forward Freight Rate Dynamics - Empirical
Evidence from Time Charter Rates. Maritime Political Management 31(4), 319-335.
Kolos, S.P., Ronn, E.I. (2008). Estimating the Commodity Market Price of Risk for Energy
Prices. Energy Economics 30, 621–641.
Korn, O. (2005). Drift Matters: An Analysis of Commodity Derivatives. Journal of Futures
Markets 25, 211-241.
Lien, G., Strom, O. (1999). Modelling Jumps in Commodity Prices. WP, Norwegian Agricul-
tural Economics Research Institute, Oslo.
Lucia, J.J., Schwartz, E.S. (2002). Electricity Prices and Power Derivatives: Evidence from the
Nordic Power Exchange. Review of Derivatives Research 5(1), 5-50.
Miltersen, K. (2003). Commodity Price Modelling that Matches Current Observables. Quanti-
tative Finance 3, 51-58.
Miltersen, K., Schwartz, E.S. (1998). Pricing of Options on Commodity Futures with Stochastic
Term Structures of Convenience Yield and Interest Rates. Journal of Financial and Quantitative
Analysis 33, 33-59.
Nielsen, M.J., Schwartz, E.S. (2004). Theory of Storage and the Pricing of Commodity Claims.
Review of Derivatives Research 7, 5–24.
Pilipovic, D. (2007). Energy Risk: Valuing and Managing Energy Derivatives (2nd ed). McGraw-
Hill.
Reisman, H. (1991). Movements of the Terms Structure of Commodity Futures and Pricing of
Commodity Claims. WP, Technion - Israel Institute of Technology.
Rebonato, R. (1998). Interest-Rate Option Models. Wiley & Sons, London.
Ribeiro, D., Hodges, S. (2005). A Contango-Constrained Model for Storable Commodity Prices.
The Journal of Futures Markets 25(11), 1025–1044.
Richter, M.C., Sorensen, C. (2002). Stochastic Volatility and Seasonality in Commodity Futures
and Options: The Case of Soybeans. WP, Copenhagen Business School.
Roncoroni, A. (2002). Essays in Quantitative Finance: Modelling and Calibration in Interest
Rate and Electricity Markets. Ph.D. Dissertation, Université Paris IX Dauphine, France.
Roncoroni, A., Id Brik, R. (2008). On the Relationship between Commodity Forward Models
and Spot-Convenience Yield Model. WP, ESSEC Business School.
Schwartz, E.S. (1997). The Stochastic Behavior of Commodity Prices: Implications for Valuation
and Hedging. Journal of Finance 52(3), 923–973.
Schwartz, E.S. (1998). Valuing Long Term Commodity Assets. Financial Management 27(1),
57-66.
Schwartz, E.S., Smith, J. (2000). Short-Term Variations and Long-Term Dynamics in Commod-
ity Prices. Management Science 46, 893-911.
Sorensen, C. (2002). Modeling Seasonality in Agricultural Commodity Futures. Journal of
Futures Markets 22, 393–426.
Trolle, A.B., Schwartz, E.S. (2006). Unspanned Stochastic Volatility and the Pricing of Com-
modity Derivatives. WP, Copenhagen Business School.