Journal of Banking & Finance 28 (2004) 1–17

www.elsevier.com/locate/econbase

Hedging with forwards and puts in complete

and incomplete markets q
a,b b,*
Simon Z. Benninga , Casper M. Oosterhof
a
Leon Recanati School of Business Administration, Tel Aviv University, P.O. Box 39010, Ramat Aviv,
Tel Aviv 69978, Israel
b
Department of Finance, University of Groningen, P.O. Box 800, 9700 AV Groningen,
The Netherlands

Abstract

We derive general conditions for forward and/or put unbiasedness and show that restric-
tions on the probability distribution suffice for simultaneous unbiasedness of forwards and
puts, even if consumers are assumed to be risk averse. We examine the optimal production
and hedging decisions by a risk-averse producer. If the producerÕs state prices are derived from
his marginal rates of substitution, an unbiased market forward price is perceived as overpriced
and an unbiased market put price as underpriced. Even in this case the full hedging and sepa-
ration theorems still hold and, contrary to previous literature, there is a hedging role for puts.
Ó 2002 Elsevier B.V. All rights reserved.

JEL classification: D21; D81; G10

Keywords: Hedging; Put options; Forward contracts; Unbiasedness

1. Introduction

Income uncertainty and optimal hedging decisions by a competitive commodity

producer have been the object of considerable research. This paper examines two
issues which are not addressed or have caused some confusion in the hedging lit-
erature. We first derive general conditions under which forward and/or put price

q
This paper was presented at the X International Tor Vergata Conference on Banking and Finance held
in Roma on 5/7 Dec. 2001.
*
Corresponding author.
E-mail addresses: [email protected] (S.Z. Benninga), [email protected] (C.M. Ooster-
hof).

0378-4266/$ - see front matter Ó 2002 Elsevier B.V. All rights reserved.
doi:10.1016/S0378-4266(02)00390-4
2 S.Z. Benninga, C.M. Oosterhof / Journal of Banking & Finance 28 (2004) 1–17

unbiasedness occurs. Contrary to the traditional belief that unbiasedness occurs only
under risk-neutrality, we show that restrictions on the probability distribution suffice
for unbiasedness, even if consumers are assumed to be strictly risk averse. Second,
we examine the optimal production and hedging decisions by a risk-averse producer.
Hedging is utility-enhancing for this producer only if his private state prices (derived
from the marginal rates of substitution) differ from the market state prices. If the
producerÕs state prices are derived from his marginal rates of substitution, he will
perceive an unbiased market forward contract to be overpriced and an unbiased
market put price to be underpriced. Contrary to the previous literature we show
there is a hedging role for put options together with forward contracts.
In a pioneering article, McKinnon (1967) presents a model of a commodity pro-
ducer who minimizes income volatility in a mean–variance framework. He shows
that the correlation between stochastic price and production is crucial in the optimal
hedging decision. A missing feature in McKinnonÕs model is that production cannot
be chosen. Baron (1970) and Sandmo (1971) develop a model of optimal production
under price uncertainty, which is extended by Danthine (1978), Holthausen (1979)
and Feder et al. (1980) to incorporate optimal hedging decisions as well. They show
that, when output is non-random, the well-known separation theorem holds. The op-
timal production decision is independent of the producerÕs risk preferences and ex-
pectations and can be separated from the optimal hedging decision. If the forward
price is unbiased, the optimal production decision is to produce until the marginal
costs equal the forward price and the optimal forward position is a full hedge.
The results above – extended by Benninga et al. (1983) and Lapan et al. (1991) –
apply to a competitive producer who faces price risk only. For most commodities,
however, a producer faces multiple sources of risk. Lapan and Moschini (1994)
consider a producer facing price, production, and basis risk. They derive an exact
solution to the optimal hedging problem under the assumption that price, produc-
tion and basis risk are joint-normally distributed and that the producer maximizes
an exponential utility function. An important finding is that the optimal hedge
depends on the degree of risk aversion, even if the forward price is assumed to
be unbiased.
The use of options as a hedging instrument has been examined much less than the
use of futures. Lapan et al. (1991) consider a producer facing price and basis risk and
compare the use of futures to put options as a hedging device. They show that, when
the futures price is unbiased, options are redundant hedging instruments since fu-
tures provide a payoff that is linear in price risk. Moschini and Lapan (1995) study
the problem of a producer facing price, (non-linear) basis, and production risk. They
provide analytical solutions to the use of futures contracts and straddles, assuming
an exponential utility function and joint–normal distributions between the risk fac-
tors. Under the assumption of unbiased forward and straddle prices, they show that
the optimal strategy is to buy straddles along with a short position in futures. Bat-
termann et al. (2000) compare the use of forward contracts and put options within a
one period utility framework. They show that, in case of unbiased put prices, the op-
timal hedging strategy is to overhedge and the optimal output decision is to produce
up to a point where the marginal costs are less than the forward price (assuming
 S.Z. Benninga, C.M. Oosterhof / Journal of Banking & Finance 28 (2004) 1–17 3

unbiasedness of the forward price). Furthermore, forwards will always be preferred

to puts when both instruments are perceived as unbiased predictors of future payoffs.
In all of the above papers hedging is the result of well-specified risks and deriva-
tives markets which allow the complete or partial (through cross-hedging) hedging of
these risks. An innovative paper by Franke et al. (1998) examines the hedging motive
when there are unhedgeable background risks. This hedging behavior has similar or-
igins to that discussed in the current paper: Where individuals disagree strongly with
the market prices (whether through greater background risk or for the unspecified
reasons in this paper), they will be more strongly motivated to use non-linear deriv-
ative instruments to try to complete the market.
This paper has two purposes: First, we examine the conditions under which for-
ward contracts and/or put options are unbiased. It is sometimes argued that unbi-
asedness of derivative instruments only occurs under risk-neutrality. 1 We show
that this is not true, and that restricting the probability distribution is sufficient
for unbiasedness of forward and put prices. Second, we examine the impact of unbi-
asedness on optimal hedging and production decisions. Our model extends previous
research by showing that there is a hedging role for put options even if only price is
stochastic. We also show that, whereas unbiased forward prices do not affect produc-
tion decisions, the use of puts reduces production.
The remainder of this paper is organized as follows. Section 2 introduces the gen-
eral model specification in which the conditions for the forward price and the put
price to be unbiased are derived. In Section 3 we derive the optimal production
and risk management decisions for a risk-averse producer. We examine the possibil-
ity of optimal hedging and production under market completeness and under market
incompleteness. Section 4 concludes the paper.

2. The model

We consider a two-date framework where today is denoted as time 0 and tomor-

row as time 1. Time 1 has N states of the world. We examine an asset S, having a spot
price S0 today and e S ¼ fS1 < S2 <    < SN g prices in the states of the world tomor-
row. The state probabilities are given by p~ ¼ fp1 ; p2 ; . . . ; pN g, and the state prices by
which financial assets are priced are denoted as q~ ¼ fq1 ; q2 ; . . . ; qN g.
Assets are priced by the state prices.
PN For example, the equilibrium risk-free rate of
interest rf is given by 1=ð1 þ rf Þ ¼ j¼1 qj . In general, given the state prices, any as-
~
PN state-dependent payoffs A ¼ fA1 ; A2 ; . . . ; AN g will have a price today
set having
A0 ¼ j¼1 qj Aj . As shown by Beja (1972), we can write the value A0 as a function
of the discounted expected payoff plus a covariance term representing the risk of
the asset:

1
Among others, Chiang and Trinidad (1997), Wu and Zhang (1997) and Baillie and Bollerslev (2000)
argue that forward unbiasedness occurs under the joint assumptions of efficient markets and risk-neutral
consumers.
4 S.Z. Benninga, C.M. Oosterhof / Journal of Banking & Finance 28 (2004) 1–17
h i !
X
N X
N
qj E A~ q~ ~
A0 ¼ q j Aj ¼ p j Aj ¼ þ Cov ;A : ð1Þ
j¼1 j¼1
pj 1 þ rf p~

For future reference we note that in the case of a single representative consumer
with a Von Neumann–Morgenstern time-additive utility function, the state prices are
derived from the consumerÕs marginal rates of substitution qj ¼ dpj ðU 0 ðcj Þ=U 0 ðc0 ÞÞ,
where d is the consumerÕs pure rate of time preference, pj is the probability of state j,
and cj is consumption in state j. At this point we leave open the question of whether
the market state prices are determined by the individual consumerÕs marginal rates of
substitution (see Section 3). Suppose the size of optimal consumption is correlated to
the commodity price so that c1 < c2 <    < cN . Since the utility function is concave,
it follows that for time-additive utility:
q1 U 0 ðc1 Þ q2 U 0 ðc2 Þ qN U 0 ðcN Þ
¼d 0 > ¼d 0 >  > ¼d 0 : ð2Þ
p1 U ðc0 Þ p2 U ðc0 Þ pN U ðc0 Þ
Before deriving the optimal production and hedging decisions in the next section,
we first examine the conditions under which forward contracts and put options will
be unbiased.

2.1. Unbiasedness of the forward price

In this subsection we derive conditions for the forward price to be unbiased. Let F
denote the forward price at date 0 for the delivery of one unit of the asset at date 1.
By definition of the forward price, F is set so that the time 0 cost is zero:
XN
qj ðSj F Þ ¼ 0: ð3Þ
j¼1
P
Solving Eq. (3) for the forward price gives F =ð1 þ rf Þ ¼ Nj¼1 qj Sj ¼ S0 ) F ¼
ð1 þ rf ÞS0 . This forward price is unbiased if F ¼ E½ e
S . As shown above, we can write
h i !
F X N XN
qj E eS q~ e
¼ qj Sj ¼ pj Sj ¼ þ Cov ;S : ð4Þ
1 þ rf j¼1 j¼1
pj 1 þ rf p~
Solving for F gives
!
h i q~
F ¼E e
S þ ð1 þ rf ÞCov ;e
S : ð5Þ
p~

Thus, the forward price is unbiased if and only if Covðpq~~ ; e

S Þ ¼ 0. As the lemma
below shows, the covariance is zero for two cases:

Lemma 1. The forward price is unbiased if and only if one of the following holds:

1. The state prices are derived from a risk-neutral representative consumer.

2. The consumer is risk averse and there is one restriction on the probability distribution.
 S.Z. Benninga, C.M. Oosterhof / Journal of Banking & Finance 28 (2004) 1–17 5

Proof
Part 1: In this case the state prices are given by qj ¼ pj =ð1 þ rf Þ and the covari-
ance term is Covðpq~~ ; e 1
S Þ ¼ Covð1þr f
;eS Þ ¼ 0.
Part 2: Given the market state prices q~ and the asset prices e
S , unbiasedness of the
forward price imposes restrictions on the state probabilities.
PN When
PN there arePNN states
of the world, forward unbiasedness occurs if F ¼ ð j¼1 qj Sj = j¼1 qj Þ ¼ j¼1 pj Sj .
Given the state prices and the asset prices, this equation can be solved for any state
probability pk . This means that unbiasedness imposes the following general restric-
tion on pk :
hP i
N 1
F p ðS
j¼1:j6¼k j j S N Þ þ S N
pk ¼ ; ð6Þ
½Sk SN
where state 1 6 k 6 N : 

The first result in Lemma 1 is standard, since under risk-neutrality all asset prices
are unbiased. The second part of the lemma shows that risk-neutrality is not a nec-
essary condition for forward unbiasedness. The restriction in Eq. (6) depends on all
the other probabilities and on the state prices, meaning that both the probability dis-
tribution and the state prices can have any form. Note that if there are only two
states of the world, this restriction implies that p1 ¼ q1 =ðq1 þ q2 Þ and p2 ¼
q2 =ðq1 þ q2 Þ. 2 For three or more states, this restriction is unrelated to the degree
of risk aversion of the consumers.

2.2. Unbiasedness of the put price

Let P ðX Þ denote the date 0 put price with an exercise price equal to X . The put is
unbiased if it equals the discounted expected put payoffs. Since we can write the put
price as
h iþ
E X e
S !
XN
þ q~ h e
iþ
P ðX Þ ¼ qj ½X Sj ¼ þ Cov ; X S ; ð7Þ
j¼1
1 þ rf p~

put price unbiasedness occurs if and only if Covðpq~~ ; ½X e þ

S Þ ¼ 0.

Lemma 2. The put price is unbiased if and only if one of the following holds:

1. The state prices for every state in which the put is in the money are risk-neutral
prices.
2. The forward price is unbiased and the put is either always in or out of the money.
3. A restriction on the probability distribution similar to that derived in Lemma 1 for
forward prices is imposed.

2
We thank an anonymous referee for pointing this out.
6 S.Z. Benninga, C.M. Oosterhof / Journal of Banking & Finance 28 (2004) 1–17

Proof
Part 1: We start by recalling from Eq. (7) that 3
! PN
q~ h iþ þ
XN
j¼1 pj ½X Sj
Cov ; X Se ¼ qj ½X Sj
þ
:
p~ j¼1
1 þ rf

Now suppose that the put option is exercised in states 1; . . . ; k of the world, i.e.,
that Skþ1 > X > Sk > Sk 1 >    > S1 . In this case the put price is unbiased if and
only if
! Pk
q~ h e
iþ X k
j¼1 pj ðX Sj Þ
Cov ; X S ¼ qj ðX Sj Þ
p~ j¼1
1 þ rf
X
k
pj
¼ qj ðX Sj Þ ¼ 0:
j¼1
1 þ rf

Since the factors ðX Sj Þ > 0, j ¼ 1; . . . ; k, it follows that the put is unbiased if and
only if qj =pj ¼ 1=ð1 þ rf Þ for every state in which it is exercised.
Part 2: Suppose
PN that X <þS1 so that the put is always out of the money. In this
case, P ðX Þ ¼ j¼1 qj ½X Sj ¼ 0, so that the put is unbiased. On the other hand,
if the put is always in the money, i.e., X > SN , then
h iþ
E X e
S !
X N
þ q~ h e
iþ
P ðX Þ ¼ qj X Sj ¼ þ Cov ; X S
j¼1
1 þ rf p~
h i ! h i
E X e S q~ E X e S
¼ þ Cov ;X e S ¼ ;
1 þ rf p~ 1 þ rf

since the forward is unbiased.

Part 3: Similar to deriving the restriction imposed on the probability distribution
in case of forward unbiasedness, we can solve for put unbiasedness as well. Given
the market state prices and the asset prices, unbiasedness of the put price imposes
a restriction on the probability distribution. Again, when PN there are Nþstates PN of the
world,
PN put unbiasedness occurs if P ðX Þð1 þ r f Þ ¼ ð j¼1 q j ½X S j = j¼1 qj Þ ¼
j¼1 pj ½X Sj þ . Given the state prices q~ and the asset prices e S , this equation can
be solved for any state probability pk . This means that unbiasedness imposes the fol-
lowing general restriction on pk :
P  
N 1 þ þ þ
PP ðX
N
Þ
j¼1:j6¼k pj ½X S j ½X S N þ ½X S N
qj
pk ¼ j¼1 ð8Þ
½X Sk þ ½X SM þ
where state 1 6 k 6 N : 

3
Since E½A  B ¼ E½A E½B þ CovðA; BÞ, the covariance term CovðA; BÞ ¼ E½A  B ¼ E½A E½B .
 S.Z. Benninga, C.M. Oosterhof / Journal of Banking & Finance 28 (2004) 1–17 7

Note the similarity between equations (6) and (8). Note also that the conditions in
part 1 of Lemma 2 are close to risk-neutrality, which occurs if and only if qj =pj ¼
1=ð1 þ rf Þ for all states j. If the put price is unbiased for all exercise prices X , then
there is risk-neutrality, so that the forward price is also unbiased. 4

2.3. Both forward and put unbiasedness

Up to now we have only considered the possibility of forward unbiasedness or put

unbiasedness. Forward and put unbiasedness occurs if both Covðpq~~ ; e S Þ ¼ 0 and
Covðpq~~ ; ½X eS þ Þ ¼ 0. If we abstract from risk-neutrality, this only occurs if there
are two restrictions
PN onP
the probability
PN distribution, and we simultaneously
PN have to
N þ
solve F ¼ ð j¼1 qj Sj = j¼1 qj Þ ¼ j¼1 pj Sj and P ðX Þð1 þ rf Þ ¼ ð j¼1 qj ½X Sj =
PN PN
j¼1 qj Þ ¼ j¼1 pj ½X Sj þ . This occurs if there are two states k and l (with
k < l) for which the following restrictions hold:
!!
PN  PN   PN
þ þ
1
Sk j¼1:j6¼k;l p j Sj þ F Sl j¼1:j6¼k;l pj Sj þ F ½X Sk þ j¼1:j6¼k;l pj ½X S PP ðX
N
Þ
qj
j¼1
pk ¼ þ þ ;
Sl ½X Sk Sk ½X Sl

0 P  1
N þ
pj Sj þF ½ X Sk PN þ
Sk @ A
j¼1:j6¼k;l P ðX Þ
Sk
þ j¼1:j6¼k;l pj X Sj PN
qj
j¼1:j6¼k;l
pl ¼ þ þ ; ð9Þ
Sl ½ X Sk Sk ½ X Sl

where state 1 6 k < l 6 N .

Given these restrictions on the probability distribution both forward contracts as
well as put options are unbiased predictors of the expected payoff. Contrary to the
traditional belief that unbiasedness of derivative instruments only occurs under
risk-neutrality, it can also occur for any probability distribution. 5

3. Optimal production and risk management by a risk-averse producer

Up to this point we have proved some general statements about state price math-
ematics. We now introduce a producer who maximizes a utility function which in-
cludes production and hedging. 6 We assume that the producerÕs state prices
(denoted fqPj g) are not necessarily the same as the market state prices which deter-
mine the prices of the forward contract and the put contract (from now on we denote
these prices by fqMj g). In order to derive optimal production and risk management
decisions we must first say something about the market conditions. We say that

4
Of course, in risk-neutrality all asset prices are unbiased.
5
The technical restrictions derived in this section ensure that forwards and/or put price unbiasedness
occurs. The numerical effect of unbiasedness on the probability distribution will, of course, depend on the
form of the probability distribution in relation to the risk-free rate of interest.
6
We assume that both the market and the producer have the same subjective state probabilities p.
8 S.Z. Benninga, C.M. Oosterhof / Journal of Banking & Finance 28 (2004) 1–17

markets are complete if the producerÕs state prices and the market state prices are the
same. 7 In Theorem 1 we prove that in complete markets there is no advantage to the
producer in hedging, either by using puts or by using forward contracts.
The market is incomplete if the producerÕs and the market state prices are not the
same. In this case, the producer disagrees with the market about the pricing of for-
ward contracts and put options, and the producerÕs valuation of forward and put
contracts (by using his private state prices) differs from that of the market.

3.1. The case of complete markets

The producer faces uncertainty because the future price of goods sold is random.
At time 0 the producer, with initial wealth W0 , chooses the output y to his production
function; these inputs cost CðyÞ, where C is a strictly convex cost function. At time 1,
uncertainty regarding the commodity price is resolved; in each state of the world, the
producer produces output y and realizes the proceeds from his sales given the sto-
chastic commodity price. If the producer has access to forward and put markets,
he has to choose how many forward contracts (nF ) and how many puts (nP ) to
buy or sell in order to solve the following problem:
h i PN
max E U ð~ cÞ ¼ U ðc0 Þ þ d pj U ðcj Þ;
j¼1
s:t: ð10Þ
c0 ¼ W0 CðyÞ nP P ðX Þ;
cj ¼ Sj y þ nP ½X Sj þ þ nF ðF Sj Þ:

Before turning to the optimal production and risk management decisions we first
P P
examine the covariance factors Covðq~p~ ; e
S Þ and Covðq~p~ ; ½X e
S þ Þ.

Lemma 3. Suppose the producer does not have access to either put or forward markets.
If the producer’s state prices are derived from his marginal rates of substitution, then
P P
the covariance term Covðq~p~ ; e
S Þ < 0 and Covðq~p~ ; ½X e
þ
S Þ > 0. Thus for this case both
the put and the forward prices are biased.

Proof. When the cost function is strictly convex there is a unique solution for optimal
production y  and to state prices qPj ¼ dðpj U 0 ðSj y  Þ=U 0 ðW0 Cðy  ÞÞÞ. This implies
that the covariance factor can be written as
0 1
pj U 0 ðcj Þ !
qPj d
B U ðc0 Þ
0 
C U 0 ðcj Þ
Cov ; Sj ¼ Cov@ ; Sj A ¼ Cov d 0  ; Sj ;
pj pj U ðc0 Þ

where c denotes consumption given optimal production. Given a strictly concave

utility function, U 0 ðcj Þ=U 0 ðc0 Þ is a decreasing function in S, which results in a neg-

7
This will happen, for example, if the producer is also the representative consumer.
 S.Z. Benninga, C.M. Oosterhof / Journal of Banking & Finance 28 (2004) 1–17 9

P
ative covariance factor Covðq~p~ ; e
SÞ and a positive covariance term
P
Covðq~p~ ; ½X e
þ
S Þ: 

3.2. Hedging with puts and forwards

The first-order conditions of (10) with respect to output, forwards, and puts are
given by

dE½U ð~
cÞ X
N
¼ C 0 ðyÞU 0 ðc0 Þ þ d pj Sj U 0 ðcj Þ ¼ 0;
dy j¼1

dE½U ð~
cÞ X
N
¼d pj U 0 ðcj ÞðSj F Þ ¼ 0; ð11Þ
dnF j¼1

dE½U ð~
cÞ X
N
  þ
¼ U 0 ðc0 ÞP ðX Þ þ d pj U 0 cj ½X Sj ¼ 0:
dnP j¼1

Theorem 1. If the producer is representative in the sense that his marginal rates of
substitution are equal to the market state prices, then an incremental purchase of the
forward or the put contract does not increase the producer’s welfare. In this case neither
hedging with puts or forwards is preferable one over the other.

Proof. As shown in Lemma 3, there is a unique solution to the optimal production

decision which gives state prices qj ¼ dðpj U 0 ðSj y  Þ=U 0 ðW0 Cðy  ÞÞÞ. Now suppose
the producer is trying to decide whether to add a small quantity nF forwards or nP
puts to his position. His expected utility will now be

cðnF ; nP ÞÞ ¼ U ðW0 Cðy  Þ nP P ðX ÞÞ

E½U ð~
XN
 þ
þd pj U Sj y  þ nF ðSj F Þ þ nP ½X Sj :
j¼1

Consider the choice of forward contracts first. Using a standard first-order Taylor
series expansion, it is clear that
X
N
E½U ð~
cðnF ; 0ÞÞ ¼ U ðW0 Cðy  ÞÞ þ d pj U ðSj y  þ nF ðSj F ÞÞ
j¼1

X
N
 
¼ E½U ð~
cð0; 0ÞÞ þ dnF pj U 0 Sj y  ðSj F Þ;
j¼1

cð0; 0ÞÞ is the expected utility and y  is the optimal production in case the
where E½U ð~
firmPdoes not hedge. It is obvious that hedging adds value if and only if
N
dnF j¼1 pj U 0 ðSj y  ÞðSj FPÞ > 0. However, dividing the previousP equation by
N N
U ðcð0; 0ÞÞ yields dnF j¼1 pj ðU 0 ðSj y  Þ=U 0 ðcð0; 0ÞÞÞðSj F Þ ¼ nF j¼1 qj ðSj F Þ,
0
10 S.Z. Benninga, C.M. Oosterhof / Journal of Banking & Finance 28 (2004) 1–17

which equals zero by definition of the forward price. This proves the theorem for
forwards. The proof for puts is similar. 

Thus, if the producer is the representative agent, hedging with forward contracts
or put options does not improve his personal utility of wealth, since buying or selling
financial instruments is always a zero-NPV investment. There will be welfare im-
provements only if the producerÕs implicit state prices differ from the pricing for
the forward/put contracts. Thus, hedging can only add value if there is some kind
of market incompleteness.

3.3. The case of incomplete markets

When markets are incomplete, the producerÕs marginal rates of substitution are
different from the market prices. In this case the use of forwards and put options
can lead to welfare improvements. Suppose that the producerÕs state prices are de-
rived from his marginal rates of substitution from maximization problem (10), lead-
ing to state prices qPj ¼ dðpj U 0 ðSj y  Þ=U 0 ðW0 Cðy  ÞÞÞ. Furthermore, suppose that the
market forward price is unbiased. Then
PN
j¼1 qM
j Sj
X
N
F ¼ PN ¼ pj Sj : ð12Þ
j¼1 qM
j j¼1

If the producer agrees with the market valuation, then

PN
j¼1 qPj Sj X
N
F ¼ PN ¼ pj Sj : ð13Þ
j¼1 qPj j¼1
PN
However, by Lemma 3, F < j¼1 pj Sj . From this we conclude that if the forward
price is unbiased (using market state prices), then the producer views the market
forward price as being overpriced. A similar statement is true for puts.
Thus, if the market forward price is unbiased, then the producer thinks that the
market forward price is too high and the put price is too low. Therefore the optimal
forward position for the producer is to short the forward contract and to go long the
put contract.

Theorem 2. If the market forward price is unbiased and markets are incomplete, the
producer will engage in a full hedge. There will be separation between the production
and hedging decision even though the producer perceives the forward price to be
overpriced. 8

8
Note that we explicitly distinguish between the market forward price, which is given to the producer,
and his private valuation which he would be willing to pay for the forward contract given his private state
prices.
 S.Z. Benninga, C.M. Oosterhof / Journal of Banking & Finance 28 (2004) 1–17 11

Proof We consider a producer who maximizes his expected utility using forwards
only:
PN
max E½U ð~ cÞ ¼ U ðc0 Þ þ d j¼1 pj U ðcj Þ;
s:t:
c0 ¼ W0 CðyÞ; ð14Þ
M
cj ¼ Sj y þ nF ðSj F Þ;
P PN M PN PN P PN P
where F M ¼ ð Nj¼1 qM
j Sj = j¼1 qj Þ ¼
P
j¼1 pj Sj > F ¼ ð j¼1 qj Sj = j¼1 qj Þ imply-
ing that the producer perceives the market forward price to be overpriced (by
Lemma 3). The superscripts M and P are added to distinguish between the market
forward price and the producerÕs private valuation of the forward price. The first-
order conditions for the producer are given by
dE½U ð~
cÞ X
N
¼ C 0 ðyÞU 0 ðc0 Þ þ d pj Sj U 0 ðcj Þ ¼ 0
dy j¼1
ð15Þ
dE½U ð~
cÞ X
N
0 M
¼d pj U ðcj ÞðSj F Þ¼0
dnF j¼1

However, since both the producer and the market face the same probability distri-
bution, the producer perceives the market forward price as being unbiased and the
optimization problem becomes a standard decision. Look at the choice of the
number of forward contracts first. The first-order condition can be rewritten as
dE½U ð~
cÞ XN
  
¼d pj U 0 cj Sj F M ¼ dE½U 0 ðcj ÞðSj F MÞ
dnF j¼1
    
¼ dE½U 0 ðcj Þ E ðSj F M Þ þ Cov U 0 cj ; Sj FM :
Since E½ðSj F M Þ is zero by definition of an unbiased market forward price, the first
order condition is zero if and only if CovðU 0 ðcj Þ; ðSj F M ÞÞ ¼ 0. If the producer
engages in a full hedge, i.e., nF ¼ y, marginal utility is constant and the covariance
term will be zero. Thus, even though the producer perceives the forward to be
overpriced, he still engages in a full hedge. Now consider the optimal production
decision. Dividing the first-order condition through U 0 ðc0 Þ yields
" #
dE½U ð~
cÞ XN
U 0 ðcj Þ U 0
ð~
c Þ
0
¼ C ðyÞ þ d p j Sj 0 ¼ C ðyÞ þ E Se d 0
0
dy j¼1
U ðc0 Þ U ðc0 Þ
" # !
h i U 0 ð~
cÞ U 0 ð~
cÞ
0 e
¼ C ðyÞ þ E S E d 0 e
þ Cov S ; d 0 :
U ðc0 Þ U ðc0 Þ

However since the producer engages in a full hedge, U 0 ð~

cÞ is constant which allows us
to rewrite the first-order condition as 9

9
A bar (–) is used to denote a non-random variable.
12 S.Z. Benninga, C.M. Oosterhof / Journal of Banking & Finance 28 (2004) 1–17
" # !
dE½U ð~
cÞ h i U 0 ðcÞ U 0 ðcÞ
¼ 0 e
C ðyÞ þ E S E d 0 e
þ Cov S ; d 0
dy U ðc0 Þ U ðc0 Þ
1
¼ C 0 ðyÞ þ F M ¼ C 0 ðyÞ þ S0 :
1 þ rf

This completes the proof. 

By Theorem 2, production takes place up to the point where the marginal costs
equal the current spot price (or alternatively production takes place up to the point
where the forward marginal costs equal the market forward price). The optimal
hedge is a full hedge and the optimal production decision is to produce until the mar-
ginal costs of production equal the current spot price. Thus, the producerÕs risk pref-
erences do not influence his optimal production decision, nor his optimal forward
hedge. This proof differs from the traditional papers on optimal hedging and produc-
tion, in which the producer agrees with the market valuation of the forward. 10 Even
though the producer does not agree with the market valuation of the forward con-
tract (according to his private valuation the forward is overpriced) he still engages in
a full hedge and there is separation between the production and hedging decision.

Lemma 4. If the market valuation of the put is unbiased and if the producer and the
market agree on the interest rate, then the producer will view the put as underpriced.
Thus the optimal hedging position for the producer will be long in the put. 11

Proof. If the market valuation of the put is unbiased we must have the general re-
striction on the probability distribution, as given in Eq. (11). 12 We will now show,
by contradiction, that the producer cannot agree with the market valuation P of the
put. IfPthe marketPput price is unbiased then we have P ðX Þð1 þ rf Þ ¼ ð Nj¼1 qM j ½X
Sj þ = Nj¼1 qMj Þ ¼
N
j¼1 pj ½X Sj þ . Now we will show that the producer does not
agree with the market valuation of the put. To see this, we proceed by contradiction
by saying that if the producer would agree with the market put price, then
PN P þ
j¼1 qj ½X Sj XN
þ
P ðX Þð1 þ rf Þ ¼ PN P ¼ pj ½X Sj :
q
j¼1 j j¼1

PN þ
However, from Lemma 3, P ðX Þð1 þ rf Þ > j¼1 pj ½X Sj , from which we can
conclude that if the market put price is unbiased, then the producer will view the put
option as being underpriced. This implies that the optimal put position will be
long. 

10
See e.g., Benninga et al. (1983, 1985).
11
Remember that an unbiased put price means that the forward put price equals the expected payoff.
12
We assume that at least one state of the world generates a positive payoff.
 S.Z. Benninga, C.M. Oosterhof / Journal of Banking & Finance 28 (2004) 1–17 13

Theorem 3. If the market put price is unbiased, then the optimal put position depends
on producer access to the bond market. This can result in a full hedge, in an overhedge,
and in an underhedge.

Proof. If the producer can use put options only, his maximization problem becomes
P
max E½U ð~ cÞ ¼ U ðc0 Þ þ d Nj¼1 pj U ðcj Þ;
s:t:
ð16Þ
c0 ¼ W0 CðyÞ nP P M ;
þ
cj ¼ Sj y þ nP ½X Sj ;
which results in the following first order conditions:
dE½U ð~
cÞ XN
¼ C 0 ðyÞU 0 ðc0 Þ þ d pj Sj U 0 ðcj Þ ¼ 0;
dy j¼1
ð17Þ
dE½U ð~
cÞ X
N
  þ
0 M 0
¼ U ðc0 ÞP þ d pj U cj ½X Sj ¼ 0:
dnP j¼1

Consider the choice of the optimal number of put options first. The question is,
whether the producer – as in the case of hedging with forwards discussed above – will
engage in a full hedge. Dividing the first-order condition by U 0 ðc0 Þ and rewriting
shows us
   0 
dE½U ð~
cÞ X
N
U 0 cj þ U cj
¼ M
P þd pj 0 M
X Sj ¼ P þ E d 0 ½X Sj þ
dnP j¼1
U ðc0 Þ U ðc0 Þ
  
U 0 ðcj Þ h þ
i U 0 cj
¼ PM þ E d 0 E X Sj þ Cov d 0 ; ½X Sj þ
U ðc0 Þ U ðc0 Þ
"  # h
E½½X Sj þ U 0 cj i U 0 ðcj Þ
þ þ
¼ þE d 0 E X Sj þ Cov d 0 ; ½X Sj
1 þ rf U ðc0 Þ U ðc0 Þ
 0   
U cj 1 U 0 cj
¼ E ½X Sj þ E d 0 þ Cov d 0 ; ½X Sj þ :
U ðc0 Þ 1 þ rf U ðc0 Þ
Case 1: If the producer has full access to the bond market then E½dðU 0 ðcj Þ=
U ðc0 ÞÞ ¼ 1=ð1 þ rf Þ. This means that the covariance term CovðdðU 0 ðcj Þ= U 0 ðc0 ÞÞ;
0
þ
½X Sj Þ must be zero in order to have an optimum. As shown by Battermann
et al. (2000) this results in overhedging.
Case 2: If the producer would like to borrow but cannot (e.g., because he has re-
stricted borrowing) the term E½dðU 0 ðcj Þ=U 0 ðc0 ÞÞ < 1=ð1 þ rf Þ. This means that at a
producer optimum
h  i
dE U c~ 
þ U 0 ðcj Þ 1
¼ E ½X Sj E d 0
dnP |fflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflffl} U ðc0 Þ 1 þ rf
>0 |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
<0
0
U ðcj Þ þ
þ Cov d 0 ; ½X Sj ¼ 0:
U ðc0 Þ
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
>0
14 S.Z. Benninga, C.M. Oosterhof / Journal of Banking & Finance 28 (2004) 1–17

In order to set the marginal utility ¼ 0, the Cov > 0; this means that for this case the
producer buys fewer puts than in Case 1, but still has a positive put position by
Lemma 4. Since setting the covariance equal to zero (as in Case 1) leads to over-
hedging we know that choosing the optimal number of put options can lead to the
possibility of full hedging, underhedging, and overhedging.
Case 3: If the producer would faces a binding lending constraint, then the term
E½dðU 0 ðcj Þ=U 0 ðc0 ÞÞ > 1=ð1 þ rf Þ. In this case

dE½U ð~
cÞ þ U 0 ðcj Þ 1
¼ E ½X Sj E d 0
dnP |fflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflffl} U ðc0 Þ 1 þ rf
>0 |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
>0
0
U ðcj Þ þ
þ Cov d 0 ; ½X Sj ¼ 0:
U ðc0 Þ
|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
<0

Thus at an optimum, the covariance term must be negative, which means that the
producer overhedges. Note that in this case, the producer overhedges even more then
in Case 1. 

The results are summarized in Fig. 1. The benchmark case where the producer has
full access to the bond market is given by the dot and results in overhedging. In this
þ
case the covariance term CovðdðU 0 ðcj Þ=U 0 ðc0 ÞÞ; ½X Sj Þ is zero. The thin part of
the line is the case where the producer has restricted borrowing, which implies a pos-
itive covariance term. Depending on the size of the covariance, this will result in the
possibility of underhedging, full hedging and overhedging. If the producer faces a
restricted lending constraint (i.e., CovðdðU 0 ðcj Þ=U 0 ðc0 ÞÞ; ½X Sj þ Þ < 0), his optimal

Fig. 1. Optimal hedging in case of put options.

 S.Z. Benninga, C.M. Oosterhof / Journal of Banking & Finance 28 (2004) 1–17 15

put position is given by the thick part of the line. In this case, the producer over-
hedges, and even more than in the benchmark Case 1.

Theorem 4. If the producer uses put options in his optimization problem he will de-
crease total production.

Proof. A long position in puts increases future consumption in the states of the world
where the put ends up in the money. This means that buying puts (weakly) decreases
the producerÕs implicit state prices because of two reasons. Recall that the producerÕs
implicit state prices are given by

U 0 ðcj Þ
qPj ¼ dp :
U 0 ðc0 Þ

Since buying puts increases future consumption in the ‘‘bad’’ states of the world,
the numerator for these states of the world decreases because of declining marginal
utility. Since buying puts also decreases current consumption the denominator in-
creases which will make the state prices go down for every state of the world. Thus,
buying more puts decreases all implicit state prices. Now suppose we have an equi-
librium in which the producer does not use puts, so that nP ¼ 0:

dE½U ð~
cÞ X
N X
N
U 0 ðcj Þ
¼ C 0 ðyÞU 0 ðc0 Þ þ d pj Sj U 0 ðcj Þ ¼ 0 ) C 0 ðyÞ ¼ dpj Sj :
dy j¼1 j¼1
U 0 ðc0 Þ

If we now add put options by increasing nP , this decreases the implicit state prices
and therefore decreases y. Puts therefore lead to a reduction in output. 

Previous theorems have discussed the (separate) use of puts and forwards in the
producerÕs decision. The next theorem proves properties of the combined use of these
instruments.

Theorem 5. If the market forward price and the market put price are unbiased, the
producer will decrease his total production, buy put options and fully hedge his total
output with forward contracts.

Proof. If the producer can use unbiased forwards and puts he has to solve the fol-
lowing maximization problem:

P
N
max E½U ð~
cÞ ¼ U ðc0 Þ þ d pj U ðcj Þ;
j¼1
s:t: ð18Þ
c0 ¼ W0 CðyÞ nP P M ðX Þ;
þ
cj ¼ Sj y þ nP ½X Sj þ nF ðF M Sj Þ;
16 S.Z. Benninga, C.M. Oosterhof / Journal of Banking & Finance 28 (2004) 1–17

resulting in the following first-order conditions:

dE½U ð~
cÞ X
N
¼ C 0 ðyÞU 0 ðc0 Þ þ d pj Sj U 0 ðcj Þ ¼ 0;
dy j¼1

dE½U ð~
cÞ XN
¼d pj U 0 ðcj ÞðSj F M Þ ¼ 0; ð19Þ
dnF j¼1

dE½U ð~
cÞ X
N
þ
¼ U 0 ðc0 ÞP M ðX Þ þ d pj U 0 ðcj Þ½X Sj ¼ 0:
dnP j¼1

Consider the choice of the number of puts first. As shown in Theorem 3, the optimal
number of put options depends on the producerÕs access to the bond market, which
can result in underhedging, overhedging, and full hedging. Furthermore, similar to
Theorem 4, this results in the producer lowering his output related to the no-hedging
case. Finally, the second part of Eq. (19) is solved if the producer, as can be expected,
fully hedges his total output. 

Contrary to the traditional papers on optimal hedging and producing, there is a

hedging role for both forwards and options, even if there is only one stochastic fac-
tor.

4. Conclusions

In this paper we examine two issues which have caused some confusion in the op-
timal hedging literature. We derive general conditions under which market prices for
forward and put contracts can both be unbiased; this is to hold under risk neutrality
or if a technical condition related on the state probabilities holds.
Our second line of research relates to optimal hedging by a producer who can use
both puts and forwards to hedge production. This problem has interest mainly in the
case where the producerÕs state prices are different from the market state prices; if
both market and private state prices are identical, then hedging by producers is
not an issue.
When the market state prices differ from the producerÕs private state prices, we
show that unbiasedness of the forward price will lead to a full hedge, even though
the producer will consider the forward contract to be overpriced. Unbiasedness of
the put price will lead to the producer taking a long position in the put. The optimal
put position depends on the producerÕs access to the bond market. This can lead to
underhedging, full hedging and overhedging. Furthermore, if the producer uses puts
to hedge his price exposure, optimal production will decrease. Finally, if both the
prices of forward contracts as well as put options are unbiased there is a hedging role
for put options.
 S.Z. Benninga, C.M. Oosterhof / Journal of Banking & Finance 28 (2004) 1–17 17

References

Baillie, R.T., Bollerslev, T., 2000. The forward premium anomaly is not as bad as you think. Journal of
International Money and Finance 19, 471–488.
Baron, D.P., 1970. Price uncertainty, utility and industry equilibrium in pure competition. International
Economic Review 11, 463–480.
Battermann, H.L., Braulke, M., Broll, U., Schimmelpfennig, J., 2000. The preferred hedge instrument.
Economics Letters 66, 85–91.
Beja, A., 1972. On systematic and unsystematic components of financial risk. Journal of Finance 27, 37–
45.
Benninga, S.Z., Eldor, R., Zilcha, I., 1983. Optimal hedging in the futures market under price uncertainty.
Economics Letters 13, 141–145.
Benninga, S.Z., Eldor, R., Zilcha, I., 1985. Optimal international hedging and output policies in
commodity and currency forward markets. Journal of International Money and Finance 4, 537–552.
Chiang, T.C., Trinidad, J.A., 1997. Risk and international parity conditions: A synthesis from
consumption-based models. International Economic Journal 11, 73–87.
Danthine, J.P., 1978. Information, futures prices and stabilizing speculation. Journal of Economic Theory
17, 79–98.
Feder, G., Just, R.E., Schmitz, A., 1980. Futures markets and the theory of the producer under price
uncertainty. Quarterly Journal of Economics 97, 317–328.
Franke, G., Stapleton, R.C., Subrahmanyam, M.G., 1998. Who buys and who sells options: The role and
pricing of options in an economy with background risk. Journal of Economic Theory 82, 89–109.
Holthausen, D.M., 1979. Hedging and the competitive firm under price uncertainty. American Economic
Review 69, 989–995.
Lapan, H., Moschini, G., 1994. Futures hedging under price, basis, and production risk. American Journal
of Agricultural Economics 76, 465–477.
Lapan, H., Moschini, G., Hanson, S.D., 1991. Production, hedging, and speculative decisions with options
and futures markets. American Journal of Agricultural Economics 73, 66–74.
McKinnon, R.I., 1967. Futures markets, buffer stocks, and income stability for primary producers.
Journal of Political Economy 75, 844–861.
Moschini, G., Lapan, H., 1995. The hedging role of options and futures under joint price, basis, and
production risk. International Economic Review 36, 1025–1049.
Sandmo, A., 1971. On the theory of the competitive firm under price uncertainty. American Economic
Review 61, 65–73.
Wu, Y., Zhang, H., 1997. Forward premiums as unbiased predictors of future currency depreciation: A
non-parametric analysis. Journal of International Money and Finance 16, 609–623.