Currency Risk Hedging: Futures vs. Forward: Abraham Lioui

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 21

Journal of Banking & Finance 22 (1998) 61±81

Currency risk hedging: Futures vs. forward


1
Abraham Lioui
Department of Economics, Bar-Ilan University, 52900 Ramat-Gan, Israel

Received 14 November 1996; accepted 3 July 1997

Abstract

The objective of this paper is to address the issue of choosing between currency for-
ward and currency futures contracts when hedging against currency risk within a sto-
chastic interest rates environment. We compare between the hedging e€ectiveness of
the two derivative assets both within a narrow sense (i.e., volatility minimization) and
within a wide sense (i.e., risk-return trade-o€). When judging hedging e€ectiveness in
the narrow sense, forward and futures contracts give identical results even if they do
not have identical prices. When judging hedging e€ectiveness in the wide sense, the
choice between the two contracts is determined by the correlation between the domestic
and the foreign term structures dynamics. Ó 1998 Elsevier Science B.V. All rights re-
served.

JEL classi®cation: G11; G13; G15

Keywords: Marking-to-market; Hedging e€ectiveness; Continuous time; Martingale


approach

1. Introduction

Brealey and Kaplanis (1995) have recently shown that commonly used strat-
egies to hedge against currency risk, such as one-period cash ¯ow hedges and
long-term ®xed hedges, may leave the ®rm exposed to foreign exchange risk. A

1
Tel.: 972 3 531 8940; fax: 972 3 535 3180; e-mail: liouia@ashur.cc.biu.ac.il.

0378-4266/98/$19.00 Ó 1998 Elsevier Science B.V. All rights reserved.


PII S 0 3 7 8 - 4 2 6 6 ( 9 7 ) 0 0 0 3 9 - 3
62 A. Lioui / Journal of Banking & Finance 22 (1998) 61±81

continuously rebalanced hedge is needed in order to enhance the hedging e€ec-


tiveness of a forward contract. The need to improve upon the above-mentioned
strategies is particularly strong when interest rates, either both domestic and
foreign or only one of the two, are stochastic, as often found in reality. Their
results emphasize the importance of deriving optimal forward strategies for
hedgers and speculators who are operating within a continuous time frame-
work and are endowed with a currency risk sensitive non-traded cash position.
Work along these lines has been done by Briys and Solnik (1992) and Tong
(1996) who also show that the forward strategy, under such circumstances,
can be decomposed into minimum-variance hedging components, Merton/
Breeden hedging components and speculative components. An extension of
the above results can be found in Glen and Jorion (1993) who included in
the investor's strategy not only forward contracts but also primitive assets.
They compared between the risk-return performance of globally diversi®ed
portfolios with and without forward contracts. They show that inclusion of
forward contracts results in statistically signi®cant improvements in the perfor-
mance of the internationally diversi®ed portfolios.
All the above-mentioned research has been carried out on the basis of hedg-
ing with forward contracts for which markets are most developed. However,
there are also futures markets that o€er hedging opportunities. When interest
rates are deterministic and there is no di€erence in price between the two deriv-
ative assets, the choice between the two derivatives is of no consequence. The
risk-return trade-o€ of a portfolio will be identical when using either futures or
forward contracts since the hedger is able to reach a perfect hedge in both
cases. However, the relevant framework when considering the above results
is that of stochastic interest rates, which are likely to introduce complications
to the analysis. Forward and futures contracts are no longer interchangeable
due to di€erences between them which are brought about by stochastic interest
rates (see Cox et al., 1981; Due and Stanton, 1992).
The main di€erence between futures and forward contracts results from the
payment schedule. Forward contracts charge gains/losses only when the hedge
is lifted, while with futures contracts, gains and losses are continuously
marked-to-market in a margin account. The marking-to-market procedure
has however been ignored by most authors who do so on the basis of empirical
results that show no economically signi®cant di€erence in price between the
two derivative assets (see Benninga and Protopapadakis, 1994 2 and the refer-
ences therein). Lately, Dezhbakhsh (1994) has obtained empirical results that
are more compatible with theory and show that the small sample inferences

2
Note that Meulbroek (1992) arrives at a di€erent conclusion than Benninga and
Protopapadakis (1994) since she found a signi®cant di€erence, for interest rates derivatives,
between the forward and futures prices.
A. Lioui / Journal of Banking & Finance 22 (1998) 61±81 63

based on the t-test may be suspect since the di€erence between the prices does
not follow a normal distribution. He therefore performs non-parametric distri-
bution-free tests which lead to signi®cant divergence between the two prices.
He also stresses the role of marking-to-market as the principal reason for this
di€erence.
The di€erence in prices as outlined above not only emphasizes the need to
return to issues that have been analyzed solely on the basis of forward con-
tracts, but also suggests a di€erence in hedging e€ectiveness that should be in-
vestigated. In contrast to recent empirical research that has found signi®cant
di€erences in prices, no such di€erence has been found with regard to hedging
e€ectiveness (see Herbst et al., 1992 and the references therein). Thus support is
lent to initial empirical ®ndings (see Cornell and Reinganum, 1981). However,
this issue has still not been comprehensively tackled by theoretical literature.
The objective of this paper is to address the issue of choosing between cur-
rency forward and currency futures contracts when hedging against currency
risk within a stochastic interest rates environment. Our analysis is conducted
within the widely used Markovian framework for interest rates. We compare
between the hedging e€ectiveness 3 of the two derivative assets both within a
narrow sense (i.e., volatility minimization) and within a wide sense (i.e., risk-re-
turn trade-o€).
Section 2 outlines our framework while Section 3 contains the main results.
Some concluding remarks are o€ered in Section 4. Appendix A contains some
technical derivations.

2. Arbitrage free international ®nancial market

We use a traditional model of a frictionless international ®nancial market


where trading takes place continuously over the time interval [0,s]. We consider
four sources of uncertainty across the two economies, represented by four in-
dependent Brownian motions fZ1 …t†; Z2 …t†; Z3 …t†; Z4 …t†; t 2 ‰0; sŠg on a complete
probability space (X, F, Q), where X is the state space, F is the r-algebra rep-
resenting measurable events, and Q is the historical probability measure. This
will allow us to have both speci®c factors and common factors a€ecting each of
the domestic term structure, the foreign term structure and the exchange rate.
All the processes used below are adapted to the augmented ®ltration generated
by the four Brownian motions. This ®ltration which is denoted by
F  fFt gt2‰0;sŠ satis®es the usual conditions. 4

3
See Howard and d'Antonio (1984, 1987).
4
The r-algebra contains the events whose probability with respect to Q is null (see Karatzas and
Shreve, 1991, p. 89).
64 A. Lioui / Journal of Banking & Finance 22 (1998) 61±81

As usual in the Martingale approach to the term structure, we characterize


the domestic and foreign term structures by specifying the evolution of the in-
stantaneous forward interest rates.

H1. The domestic instantaneous forward interest rate solves the following
stochastic di€erential equation: 5
dfd …t; T † ˆ ld …t; T ; fd …t; T †† dt ‡ md1 dZ1 …t† ‡ md2 dZ2 …t† …1†
for all x 2 X; t 6 T ; T 2 ‰0; sŠ. ld …t; T ; fd …t; T †† is the drift term that satis®es the
usual conditions 6 such that Eq. (1) has a unique solution and md1 and md2 are
strictly positive constants. The speci®cation Eq. (1) of the domestic term struc-
ture is supported by a general equilibrium model in which two uncorrelated
state variables drive the economy. 7 From a partial equilibrium point of view,
the two factors term structure models are usually motivated by the presence of
a long term factor and a short term factor driving the term structure. We retain
the assumption that the instantaneous volatilities parameters (md1 and md2 ) are
time invariant to avoid heavy notations. Note that had we assumed time-de-
pendent (deterministic) volatility parameters, the results would not have been
a€ected. Of course, conducting our analysis for stochastic volatilities will be
a natural extension of our paper as discussed in Section 4 of this paper.
The price at each time t of a domestic discount bond maturing at time si is
8 s 9
< Zi =
Pd …t; si † ˆ exp ÿ fd …t; T † dT : …2†
: ;
t

Using Eq. (1), the domestic discount bond price dynamics is as follows:
dPd …t; si †
ˆ ‰bd …t; si † ‡ rd …t†Š dt ÿ md1 …si ÿ t† dZ1 …t† ÿ md2 …si ÿ t† dZ2 …t†;
Pd …t; si †
…3†
where bd …t; si † is the instantaneous risk premium which could be found by ap-
plying Ito's lemma to Eq. (2). 8
We assume a similar structure for the foreign economy.

5
The suggested model for the instantaneous forward rate allows this economic variable to take
negative values. Unfortunately, this is one of the few polar cases in Financial Economics which
allows for explicit results. Nevertheless, Amin and Morton (1994), when testing contingent claims
implications of alternative term structure models, show that the Gaussian models do better than
lognormal ones. For a lucid discussion of this issue, see Subrahmanyam (1996).
6
See conditions C.1, p. 80 and C.2, p. 81 of Heath et al. (1992).
7
See Longsta€ and Schwartz (1992).
8
To save space and notation, we do not specify this process. Note that for the remaining, there is
no need to specify the process since it is well known that it will not be present in the prices of the
contingent claims.
A. Lioui / Journal of Banking & Finance 22 (1998) 61±81 65

H2. The foreign economy instantaneous forward interest rate solves the
following stochastic di€erential equation:
dff …t; T † ˆ lf …t; T ; ff …t; T †† dt ‡ mf2 dZ2 …t† ‡ mf3 dZ3 …t† …4†
for all x 2 X; t 6 T ; T 2 ‰0; sŠ. lf …t; T ; ff …t; T †† is the drift term that satis®es the
usual conditions 9 such that Eq. (4) has a unique solution and mf2 and mf3 are
strictly positive constants.
The dynamics in Eqs. (1) and (4) incorporate one common factor (Z2 ) which
accounts for the instantaneous correlation between the term structures of the
two economies. Note that in each economy there is a speci®c factor (Z1 for
the domestic term structure and Z3 for the foreign term structure) driving
the term structure in addition to the common factor.
The price of a foreign discount bond at each time t is
8 s 9
< Zi =
Pf …t; si † ˆ exp ÿ ff …t; T † dT …5†
: ;
t

and its dynamics is as follows:


dPf …t; si †
ˆ ‰bf …t; si † ‡ rf …t†Š dt ÿ mf2 …si ÿ t† dZ2 …t† ÿ mf3 …si ÿ t† dZ3 …t†; …6†
Pf …t; si †
where bf …t; si † is the instantaneous risk premium associated with the foreign
discount bond.
The link between the two economies is guaranteed by the spot exchange
rate. It is assumed to evolve as follows.

H3. The spot exchange rate (in units of the domestic currency) solves the
following stochastic di€erential equation:
dS …t†
ˆ lS …t; S …t†† dt ‡ mS1 dZ1 …t† ‡ mS2 dZ2 …t† ‡ mS3 dZ3 …t†
S …t †
‡ mS4 dZ4 …t†; …7†
10
where lS (t,s(t)) is the drift term that satis®es the usual conditions such that
Eq. (8) has a unique solution; mS: is a strictly positive constant.
The speci®cation in Eq. (7) allows the exchange rate to be driven by a source
of uncertainty (Z4 ) which does not a€ect the two economies under consider-
ation, in addition to the sources of uncertainty a€ecting the two economies
(Z1 , Z2 , Z3 ). This allows us to account for exogenous shocks a€ecting the ex-
change rate coming from the interdependence of the two economies under con-
sideration with other countries.

9
See conditions C.1, p. 80 and C.2, p. 81 of Heath et al. (1992).
10
See conditions C.1, p. 80 and C.2, p. 81 of Heath et al. (1992).
66 A. Lioui / Journal of Banking & Finance 22 (1998) 61±81

From a domestic investor's perspective, one needs to express all the foreign
asset prices in terms of the domestic currency. This is done by multiplying the
price of a foreign asset by the spot exchange rate. We will denote by P^f …t; si † the
price of a foreign discount bond in units of the domestic currency. Therefore,
we have
P^f …t; si † ˆ Pf …t; si †S …t†: …8†
Applying Ito's lemma to Eq. (8) and using Eqs. (6) and (7), one gets the dy-
namics of the foreign discount bond in units of the domestic currency, namely,
dP^f …t; si †
ˆ b^f …t; si † dt ‡ mS1 dZ1 …t† ‡ ‰mS2 ÿ mf2 …si ÿ t†Š dZ2 …t†
P^f …t; si †
‡ ‰mS3 ÿ mf3 …si ÿ t†Š dZ3 …t† ‡ mS4 dZ4 …t†; …9†
^
where bf …t; si †  bf …t; si † ‡ rf …t† ‡ lS …t; s…t†† ÿ mS2 mf2 …si ÿ t† ÿ mS3 mf3 …si ÿ t†:
We assume the international ®nancial market to be complete and arbitrage
free. All the portfolio strategies to be considered in Section 3 are assumed to be
admissible portfolio strategies. 11

3. The main results

Having set out the framework for analysis we can now proceed with the
main objective of the paper. First, we derive the forward contract price and
the corresponding futures settlement price and then we set the hedger's prob-
lem when confronted with a commitment to pay or receive a certain quantity
of the currency. Thirdly, we solve his problem in two separate cases, one when
futures contracts are traded and the other when forward contracts are traded.
Lastly, we perform a comparison between these two strategies to determine
which of the two derivative contracts is preferable from a hedging e€ectiveness
point of view.

3.1. Forward price and futures settlement price

Within the framework outlined above, the arbitrage free price of a currency
forward contract maturing at time s1 , denoted by G…t; s1 †, is
P^f …t; s1 † Pf …t; s1 †
G…t; s1 † ˆ ˆ S …t † …10†
Pd …t; s1 † Pd …t; s1 †

11
For an explicit de®nition of admissible strategies in our framework, see Amin and Jarrow
(1991).
A. Lioui / Journal of Banking & Finance 22 (1998) 61±81 67

while the arbitrage free settlement price of the corresponding futures contract
is 12
(
ÿ  …s1 ÿ t†3
H …t; s1 † ˆ G…t; s1 † exp v2dl ‡ v2d2 ‡ …vdl vS1
3
)
…s1 ÿ t†2 …s1 ÿ t†3
‡vd2 vS2 † ÿ vd2 vf 2 : …11†
2 3
As expected we arrive at a di€erence between the futures settlement price
and the forward contract price. Note that a deterministic domestic term struc-
ture leads to identical prices. This is due to the fact that the margin account
yields or charges are in terms of the domestic interest rate. To further investi-
gate the di€erence in prices, let us write Eq. (11) as follows:
(
P^f …t; s1 †
2
…s1 ÿ t †
H …t; s1 † ˆ exp q…t; s1 † ‡ …md1 mS1
Pd …t; s1 † 3
)
…s1 ÿ t †2
‡md2 mS2 † ; …110 †
6
where q…t; s1 †  ‰md1 …mS1 ‡ md1 …s1 ÿ t†† ‡ md2 …mS2 ‡ …md2 ÿ mf2 †…s1 ÿ t††Š:
One possible interpretation of this di€erence in prices could be as follows.
When q…t; s1 † is positive, the futures contract settlement price is greater than
the forward price. Therefore, a buyer of a futures contract is willing to pay
more than for a forward contract and a seller will ask for a higher price. To
understand this, ®rst, note that q…t; s1 † is the instantaneous covariance between
the ¯uctuations of the domestic term structure and the local ¯uctuations of the
futures settlement price. Therefore, q…t; s1 † > 0 means that an increase in (do-
mestic) interest rates is followed by an increase in the futures settlement price.
A buyer of futures contracts could then invest the margin he receives at a high-
er interest rate. Similarly, a seller of futures contracts will ®nance the margin
that he must pay at a higher (domestic) interest rate. Because the seller is ad-
versely a€ected by the (domestic) interest rates ¯uctuations, the settlement
price he will set for the futures contract will be higher than the price for a for-
ward contract. The buyer, who gains from interest rates ¯uctuations, is willing
to pay more for the futures contract than for the forward contract.

12
While deriving Eq. (10) is trivial this is not the case for the futures settlement price. Our
framework is similar to the one assumed by Amin and Jarrow (1991), namely the completeness of
the primitive assets market, and therefore one can recover Eq. (11) using the results of these
authors. Nevertheless, a complete proof of Eq. (11) is given in Appendix A for self-countenance of
the paper.
68 A. Lioui / Journal of Banking & Finance 22 (1998) 61±81

Assume now that q…t; s1 † is suciently 13 negative such that the futures set-
tlement price is less than the forward price. Then the futures ¯uctuations are
opposite to the ¯uctuations of the domestic term structure. An increase in in-
terest rates implies a decrease in the futures settlement price. A seller of futures
contracts will then receive a margin that he can invest at a higher interest rate.
A buyer pays a margin that he must ®nance at a higher interest rate. Therefore,
the buyer who is adversely a€ected by the ¯uctuations of the domestic term
structure is willing to buy the futures contract only at a price which is lower
than that of the corresponding forward contract.
The above analysis is consistent with the way the forward price and the fu-
tures settlement price di€er. An inspection of Eq. (110 ) shows that the forward
price and the futures settlement prices dynamics will have the same instanta-
neous volatilities while their drifts will di€er. Had they had di€erent instanta-
neous volatilities, we would not have been able to say anything (in general)
concerning the relationship between the futures settlement price and the for-
ward price.

3.2. The hedger's problem

The hedger has a commitment to p units (he pays when p < 0 and he re-
ceives when p > 0) of the foreign currency at time s1 < s which is his hedging
horizon. The hedger can hedge against currency risk by using either forward or
futures contracts.
Let bH (t) denote the number of futures contracts held by the hedger at time t
for the purpose of hedging his non-traded position. The futures contract posi-
tions are marked-to-market in a domestic interest rate bearing account. If we
denote by X(t) the value of the margin account at time t, then
8 t 9
Zt <Z =
X …t† ˆ exp rd …l†dl bH …s† dH …s; s1 †: …12†
: ;
0 s

The hedger's wealth at each time t, denoted WH …t†, is as follows:


WH …t† ˆ pS …t† ‡ X …t†: …13†
Despite being allowed to trade the riskless asset (i.e., the domestic money
market account), we do not include it in his portfolio because the correspond-
ing position is indeterminate. To see why this is the case, let us recall the solu-
tion to the hedging problem in the case of deterministic interest rates (domestic
and foreign). One computes the volatility and the return of the hedged portfo-
lio which is constituted of the non-traded position, the margin account and a

13
Assume for example that jq…t; s1 †j > …md1 mS1 ‡ md2 mS2 †…s1 ÿ t†2 =2:
A. Lioui / Journal of Banking & Finance 22 (1998) 61±81 69

riskless asset position. Since interest rates are deterministic, a perfect hedge of
the non-traded position is possible and the futures position is determined such
that the volatility of the hedged portfolio is zero. However, since the portfolio
is riskless, it should yield the riskless interest rate. Therefore, one must set the
return on the portfolio to be equal to the riskless rate and then solve for the
riskless asset's position. Interest rates are stochastic, therefore our investor is
able only to achieve a minimum-variance hedge and therefore the return on
the hedged portfolio will di€er from the spot interest rate. Finally, had we in-
cluded a position in a riskless asset, we would not have been able to determine
it since one cannot say anything concerning the return on the hedged portfolio.
Turning now to an investor who uses forward contracts to hedge against
currency risk, the investor's wealth at each time t is as follows:
Zt
WG …t† ˆ pS …t† ‡ Pd …s; s1 †bG …s† dG…s; s1 †; …14†
0

where bG …t† is the number of forward contracts held by the hedger in his port-
folio at time t. The second term on the right-hand side of Eq. (14) represents
the present value of the gains and losses on the forward positions.
The hedger's problem is to construct the optimal derivatives strategy so as to
minimize the instantaneous volatility of his wealth. As a result he would have
made a decision as to whether he prefers forward or futures contracts.

3.3. The solutions to the hedger's problem


14
The forward and the futures strategies, respectively, are as follows:
r
S …t † m…t; s1 † mS
bG …t† ˆ ÿp ; …15†
Pd …t; s1 †G…t; s1 † m…t; s1 †r m…t; s1 †
r
S …t† m…t; s1 † mS
bH …t† ˆ ÿp ; …16†
H …t; s1 † m…t; s1 †r m…t; s1 †
where
0 1 0 1
mS1 ‡ md1 …s1 ÿ t† mS1
B m ÿ m …s ÿ t † ‡ m …s ÿ t † C Bm C
B S2 f2 1 d2 1 C B S2 C
m…t; s1 †  B C and mS  B C:
@ mS3 ÿ mf3 …s1 ÿ t† A @ mS3 A
mS4 mS4
Eqs. (15) and (16) give the number of forward contracts and futures contracts,
respectively, held by the investor at each time t. To get the number of contracts

14
Both strategies are derived in Appendix A.
70 A. Lioui / Journal of Banking & Finance 22 (1998) 61±81

exchanged between t and t ‡ dt it suces to di€erentiate Eqs. (15) and (16)


using Ito's lemma.
These two strategies call for the following remarks.
(i) The two strategies have an identical structure which comprises the size
of the non-traded position multiplied by a price adjusting factor
…1=Pf …t; s1 † or S…t†=H …t; s1 †† and by a volatility adjusting factor …m…t; s1 †r mS
=m…t; s1 †r m…t; s1 ††. The volatility adjusting factor is the usual instantaneous
minimum-variance hedge ratio of the ¯uctuations of the exchange rate using
the derivative contracts. It is important to emphasize that the futures con-
tract and the forward contract strategies share the same minimum-variance
hedge ratio component because stochastic interest rates a€ect only the drift
and not the volatility of the futures contract, which remains equal to the
volatility of the forward contract (see the preceding paragraph for the eco-
nomic intuition of this result). The minimum-variance hedging factor adds a
time dimension to the two strategies. However, unlike previous research, the
source of this time dimension is not the volatility of the underlying asset
(Kroner and Sultan, 1993), but rather the stochastic interest rates that gen-
erate a need to continuously rebalance. The price adjusting factors are also
a source of a time dimension that strengthen the need to continuously re-
balance.
(ii) An interesting characteristic of the two strategies is that the hedger will
not necessarily choose a derivative contract's position which is opposite in sign
to the non-traded position. To see this, it suces to remark that
r
X4
m…t; s1 † mS …t† ˆ m2Si ‡ ‰…mS1 md1 ‡ mS2 md2 † ÿ …mS2 mf2 ‡ mS3 mf3 †Š…s1 ÿ t†: …17†
iˆ1

A sucient condition for the investor to take a position in the derivative


assets market which is opposite in sign to the non-traded position is that
‰…mS1 md1 ‡ mS2 md2 † ÿ …mS2 mf2 ‡ mS3 mf3 †Š > 0. This condition can be given an intu-
itive explanation since ‰…mS1 md1 ‡ mS2 md2 † ÿ …mS2 mf2 ‡ mS3 mf3 †Š is the di€erence
between (a) the instantaneous covariance between the exchange rate dynamics
and the domestic term structure dynamics and (b) the instantaneous covari-
ance between the exchange rate dynamics and the foreign term structure dy-
namics.

3.4. Futures vs. forwards

Now that we have derived the hedging strategies chosen by the investor
when hedging with forward or futures contracts, we are able to compare be-
tween the two strategies and decide which of the two contracts is preferable.
The framework of our analysis dictates a di€erence in price between the two
derivative assets, which plays an important role in our analysis. The ®rst result
is as follows.
A. Lioui / Journal of Banking & Finance 22 (1998) 61±81 71

3.4.1. Rule 1
When judging hedging e€ectiveness in the narrow sense, i.e., measuring the
volatility of a hedged portfolio, forward and futures contracts give identical re-
sults.
This ®nding can be explained by the fact that the instantaneous volatility of
the forward contract is equal to that of the futures contract as explained in Sec-
tion 3.1. Our results ®ll the gap left by previous theoretic research that has not
addressed the issue of hedging e€ectiveness. Furthermore, our conclusion is
compatible with empirical ®ndings in the ®eld of hedging e€ectiveness (see
Herbst et al., 1992). However, the above result is not sucient for a compre-
hensive analysis, because the hedger's portfolio is not risk-free, thus we are
compelled to measure hedging e€ectiveness in a wider sense, i.e., the risk-return
trade-o€. For such an analysis we need only compare between the drifts of the
two hedged portfolios, due to the equality in instantaneous volatility.
When comparing between the two drifts we obtain the following result: 15
r
m…t; s1 † mS
lH …t† ÿ lG …t† ˆ rd …t†X …t† ‡ q…t; s1 †…s1 ÿ t†pS …t†; …18†
m…t; s1 †r m…t; s1 †
where lH …t† is the instantaneous drift of the hedger's portfolio when futures are
used, lG …t† is the instantaneous drift of the hedger's portfolio when forward
contracts are used.
The ®rst term on the right-hand side of Eq. (18) depicts the accrued interest
on the margin account due to marking-to-market of the hedger's position. The
second term on the right-hand side of Eq. (18) re¯ects the contribution of in-
terest rates risk to the di€erence in the drifts. This is so because it is a function
(through q( )) of the di€erence in the drifts between the futures settlement price
dynamic and the forward price dynamic. The intuition behind the second term
on the right-hand side of Eq. (18) is best captured by rewriting Eq. (18) using
Eq. (16) as follows:
lH …t† ÿ lG …t† ˆ rd …t†X …t† ÿ bH …t†H …t; s1 †q…t; s1 †…s1 ÿ t†: …180 †
The second term on the right-hand side of Eq. (180 ) can now be interpreted
as the ``premium'' paid or received depending on the sign of the futures posi-
tion and on the di€erence between the forward price and the futures settlement
price.
An immediate result is, that in general one cannot determine which of the
two derivatives is preferable when the domestic interest rates are stochastic.
The reason for this is that outcome depends on the signs taken on by bH …t†
and q…t; s1 †. The following table gives all the possible outcomes for di€erent pa-
rameters of the model.

15
See the derivation of this result in Appendix A.
72 A. Lioui / Journal of Banking & Finance 22 (1998) 61±81

3.4.2. Rule 2

q…t; s1 † > 0 q…t; s1 † < 0


r
m…t; s1 † mS > 0 p > 0 ) futures p > 0 ) forward
p < 0 ) forward p < 0 ) futures
r
m…t; s1 † mS < 0 p > 0 ) forward p > 0 ) futures
p < 0 ) futures p < 0 ) forward

The above table describes the cases in which a hedger who has a long posi-
tion (p > 0, the long-hedger hereafter) will prefer one derivative asset over the
other, and the corresponding results for a hedger who has a short position
(p < 0, the short-hedger hereafter). Remember that the sign of m…t; s1 †r mS
which is the instantaneous correlation between the ¯uctuations of the deriva-
tive contracts and the spot exchange rate, determines the sign of the derivative
contracts position (b given by Eqs. (15) and (16)) in relation to the sign of the
non-traded position. Furthermore, q( ) is the instantaneous correlation be-
tween the price dynamics of the derivative contracts and the dynamics of the
domestic term structure.

3.4.2.1. Analysis of the top left-hand cell of the table. When m…t; s1 †r mS is posi-
tive, a hedger will always choose a position in the derivative contract that is
opposite in sign to the non-traded position. Thus, the long-hedger (p > 0) will
short the derivative contract while the short-hedger will be long in derivative
contract.
When q…t; s1 † is positive, then the table shows that the long-hedger chooses
to short the futures contract and the short-hedger chooses to be long in the for-
ward contract. The choice between futures and forward contracts can be given
the following intuitive explanation. 16 When q…t; s1 † is positive, a long-hedger
will always bene®t from an increment in the instantaneous drift of his wealth,
i.e., ÿbH …t†H …t; s1 †q…t; s1 †…s1 ÿ t† > 0. The short-hedger is a€ected adversely by

16
At ®rst glance a simpler intuition for this could be that when q(t,s1 ) is positive, the futures
settlement price is greater than the forward price and therefore the long-hedger prefers to short the
most expensive contract. Similarly, the short-hedger must buy derivative contracts to hedge his
non-traded position and therefore prefers to buy the forward contract which is cheaper. However,
this kind of explanation is not satisfactory since it does not explain the right-hand column of the
table. When q(t,s1 ) is non-positive, there is still a range of values for q(t,s1 ) such that the futures
settlement price is greater than the forward price and the long-hedger will nevertheless prefer to
hedge using forward contracts!
A. Lioui / Journal of Banking & Finance 22 (1998) 61±81 73

a positive q…t; s1 † since ÿbH …t†H …t; s1 †q…t; s1 †…s1 ÿ t† is negative and he will
therefore choose the forward contract for hedging against the risk that stems
from his non-traded position.
The perceptive reader will notice that the explanations so far, as well as the
results that appear in Rule 2, relate only to the e€ect that changes in interest
rates have on the di€erence between the drifts of the two derivative assets
(see the second term on the right-hand side of Eq. (18)). However, we can show
that the ®rst term on the right-hand side of Eq. (18), namely, the traditional
component which consists of interest accrued in the margin account, will not
a€ect our results. To understand why this is true, consider an extremely bad
case, from a long-hedger's point of view, in which interest rates are constantly
increasing. The long-hedger who is short in the futures contract, will ®nd him-
self constantly paying a margin. This is so because a positive q( ) implies a pos-
itive correlation between changes in the interest rates and changes in the futures
settlement price. Thus, any increase in interest rates brings about an increase in
the settlement price of the futures contract thereby creating a need to pay a
margin. Moreover, the newly created margins must be ®nanced by ever grow-
ing interest rates. However, the increase in expense incurred by the long-hedger
is likely to be more than o€set by the premium from trading in futures con-
tracts, (ÿbH …t†H …t; s1 †q…t; s1 †…s1 ÿ t††. Note that the premium is a function of
the value at time t of the non-traded position (see Eq. (18)) while the interest
rate is paid only on margins, i.e., on the changes in the futures settlement
prices.
Turning now to the short-hedger a similar analysis can be made. Consider
the best situation a short-hedger can ®nd himself in, namely, increasing interest
rates. Assume that this short-hedger has chosen a long position in futures con-
tracts. The increase in interest rates creates a new margin in his favor, which he
can invest at a higher interest rate. Furthermore, this increase in income is un-
likely to o€set the adverse e€ect from trading in futures contracts brought
about by the negative premium …ÿbH …t†H …t; s1 †q…t; s1 †…s1 ÿ t† is negative).
Thus, the short-hedger will prefer to trade forward contracts.
The remaining three cells of Rule 2 can be explained in a similar manner.
The analysis of these three cells is omitted to avoid repetitions that do not
add intuition to the paper. Furthermore, note that from a practical point of
view, the currency and the derivative contracts are likely to be very highly cor-
related, i.e., m…t; s1 †r mS > 0. On the other hand, q…t; s1 † > 0 unless vf2  0.
Thus, it appears that the only cell in Rule 2 that is relevant is the top left-hand
side one. q…t; s1 † also has an impact on the size of the derivative contract's po-
sition, since 17

17
This term is known in the literature as the ``tailing factor''. See Figlewski et al. (1991).
74 A. Lioui / Journal of Banking & Finance 22 (1998) 61±81
( )
2 2
bH …t† …s1 ÿ t† …s1 ÿ t†
ˆ Pd …t; s1 † exp ÿ q…t; s1 † ÿ …vd1 vS1 ‡ vd2 vS2 † :
bG …t† 3 6
…19†
When q…t; s1 † > 0, the futures position is always smaller than the forward
position. To understand why this is the case it suces to rewrite Eqs. (15)
and (16) as follows:
5
S…t† v…t; s1 † vS
bG …t†G…t; s1 † ˆ ÿp ; …150 †
Pd …t; s1 † v…t; s1 †5 v…t; s1 †
5
v…t; s1 † vS
bH …t†H …t; s1 † ˆ ÿpS…t† 5 : …160 †
v…t; s1 † v…t; s1 †

Therefore, the value of the forward position is always greater than the val-
ue of the futures position at each time t due to the presence of the discount
factor in Eq. (150 ). Moreover, when q…t; s1 † > 0, the futures settlement price is
greater than the forward price. Consequently, the hedger must always hold
less futures contracts than forward contracts. Note that when interest rates
are deterministic, it is well known that the marking-to-market of the futures
position requires a continuous rebalancing of the futures position in order to
achieve a perfect hedge. The ratio of the futures position to the forward po-
sition is called the tailing factor. A similar result is found in Eq. (19), i.e., the
discount factor. In the presence of stochastic interest rates, a second factor is
added which is brought about by the di€erence in prices between the two
contracts.
Note that q…t; s1 † > 0 is only a sucient condition but this case is the one
which is the most likely to prevail in the international ®nancial markets.

4. Concluding remarks

The main message of this paper is that currency forward contracts and cur-
rency futures contracts are not interchangeable when interest rate risk exists.
Therefore, conclusions obtained in the ®eld of International Hedging and Syn-
thetic International Diversi®cation are sensitive to the nature of the derivative
contract included in the portfolios. For example, it is not clear that the results
of Glen and Jorion (1993) will still hold when using futures contracts instead of
forward contracts. The dynamic hedging strategy derived by Briys and Solnik
(1992) and used by Tong (1996) is also likely to be a€ected by the marking-to-
market procedure.
One important question from a practical point of view is to what extent the
introduction of frictions in the international ®nancial market (such as transac-
A. Lioui / Journal of Banking & Finance 22 (1998) 61±81 75

tion costs) could change our results. Fortunately, our results are likely to be
robust even if one takes into account the presence of transaction costs. The rea-
son for this is simply because strategies Eqs. (15) and (16) need a continuous
rebalancing of the hedger's portfolio. Moreover, the futures position is always
smaller than the forward positions. Therefore, transactions costs, should they
be taken into account, would probably not change the results in the table.
Our results are sensitive to the dynamic of the domestic term structure pos-
tulated a priori. While the assumption of deterministic instantaneous volatility
of the domestic term structure has been made for convenience, there is no
doubt that a more realistic framework should account for the possibility of sto-
chastic volatility. Moreover, because interest rates are the main tool of mone-
tary policies, a more realistic modeling of the term structure dynamics should
include the possibility of jumps in the dynamic of the domestic interest rates.
However, such properties, although important, would have made our analysis
almost entirely untractable since it is well known that in the case of jumps, for
example, even explicit discount bond prices are hard to obtain (see El-Jahel et
al., 1996).
Our results could be extended in several other directions. While only the
case of a hedger has been addressed in this paper, the e€ect on the behavior
of a speculator (expected utility maximizer) could also be examined. Di€eren-
ces in the welfare reached in the case of forward trading and futures contracts
trading are likely to exist since the two contracts generate di€erent opportu-
nity sets. However the task of comparing between the welfare levels could be
dicult since the speculator has access to an incomplete market in both
cases. 18
Several currencies are subject to a realignment risk since they belong to a
target zone. Dumas et al. (1995) analyzed the e€ect of this risk on currency op-
tion pricing. It is likely that, due to stochastic interest rates and the marking-to-
market procedure, the futures settlement price will incorporate a premium to
compensate for this risk. How this risk a€ects the hedging strategy and the
hedging e€ectiveness of both contracts is an important question both from a
theoretical and an empirical point of view.
The presence of the Siegel Paradox in currency options pricing had been
widely studied (see Bardhan, 1995). The marking-to-market procedure could
be a source of a Siegel Paradox in pricing futures contracts since the margin
account bears the domestic spot rate.

18
The paper by Lioui and Poncet (1996) is an attempt to apply the martingale approach to
address the issue of intertemporal hedging in complete markets. Such an analysis could be extended
to the suggested problem based on the results of He and Pearson (1991) and Karatzas et al. (1991).
76 A. Lioui / Journal of Banking & Finance 22 (1998) 61±81

Acknowledgements

I would like to thank Rafael Eldor, Miriam Krausz and especially Patrice
Poncet for very insightful discussions and comments. Two anonymous referees
provided extensive comments and suggestions which greatly improved the pre-
sentation of the paper. All remaining errors are mine.

Appendix A

A.1. Derivation of Eqs. (10) and (11)

The derivative contracts represent the right to get one unit of the foreign
currency at the contract's maturity date, namely s1 . Therefore, it is straightfor-
ward to show that the forward contract price in units of domestic currency, de-
noted G…t; s1 †, is
P^f …t; s1 † Pf …t; s1 †
G…t; s1 † ˆ ˆ S…t†: …A:1†
Pd …t; s1 † Pd …t; s1 †
The international ®nancial market is complete and arbitrage free. This im-
plies that there exists a (unique) probability measure associated to the numera-
ire Bd ( ), equivalent to Q denoted Q, ~ such that the dynamics of the domestic
discount bond and the foreign discount bond price processes could be written
as follows:
dPd …t; s†
ˆ rd …t† dt ÿ vd1 …s ÿ t† dZ~1 …t† ÿ vd2 …s ÿ t† dZ~2 …t†; …A:2†
Pd …t; s†

dP^f …t; s†
ˆ rd …t† dt ‡ vs1 dZ~1 …t† ‡ ‰vS2 ÿ vf2 …s ÿ t†Š dZ~2 …t†
P^f …t; s† …A:3†
‡ ‰vS3 ÿ vf3 …s ÿ t†Š dZ~3 …t† ‡ vS4 dZ~4 …t†;

where Z~1 …t†; Z~2 …t† Z~3 …t†; Z~4 …t† are standard Brownian motions with respect to
~ Moreover, following Heath et al. (1992), 19 one obtains that
Q.
 t
fd …t; T † ˆ fd …0; T † ‡ …v2d1 ‡ v2d2 †t T ÿ ‡ vd1 Z~1 …t† ‡ vd2 Z~2 …t† …A:4†
2
and then
t2
rd …t† ˆ fd …t; t† ˆ fd …0; t† ‡ …v2d1 ‡ v2d2 † ‡ vd1 Z~1 …t† ‡ vd2 Z~2 …t†: …A:5†
2

19
Proposition 3, p. 86, of Heath et al. (1992).
A. Lioui / Journal of Banking & Finance 22 (1998) 61±81 77

From Due and Stanton (1992), the futures settlement R s price at time t
is equal to the price at time t of a cash ¯ow S…s1 † exp t 1 rd …u† du at the
instant s1 . Therefore, if H(t,s1 ) denotes the futures settlement price, it is such
that
  R s1 
H …t; s1 † Q~ S…s1 † exp 1
rd …u† du
ˆE Ft …A:6†
Bd …t† Bd …s1 †
and
~
H …t; s1 † ˆ EQ ‰S…s1 †jFt Š: …A:7†

Using Eq. (8), this can also be written as follows:


~
H …t; s1 † ˆ EQ ‰P^f …s1 ; s1 †jFt Š: …A:8†

Applying Ito's lemma to Ln P^f …t; s1 †, one obtains


8 t
<Z  1

P^f …t; s1 † ˆ P^f …0; s1 † exp
5
rd …u† ÿ n…u† n…u† du
: 2
0
9
Zt =
‡ ~5
n…u† dZ…u† ; …A:9†
;
0

where n…t†  …vS1 ; vS2 ÿ vf2 …s1 ÿ t†; vS3 ÿ vf3 …s1 ÿ t†; vS4 †5 . Then,
8s
<Z1  1

^ ^
Pf …s1 ; s1 † ˆ Pf …t; s1 † exp
5
rd …u† ÿ n…u† n…u† du
: 2
t
9
Zs1 =
5 ~
‡ n…u† dZ…u† : …A:10†
;
t

Using Eqs. (A.4) and (A.5), one can show that


Zs1 Zs1 3
…s1 ÿ t†
rd …u† du ˆ fd …t; T † dT ‡ …v2d1 ‡ v2d2 †
6
t t

Zs1 Zs1
‡ vd1 …s1 ÿ u† dZ~1 …u† ‡ vd2 …s1 ÿ u† dZ~2 …u†: …A:11†
t t

Substituting for Eq. (A.11) in Eq. (A.10) and using Eq. (2), Eq. (A.10) be-
comes
78 A. Lioui / Journal of Banking & Finance 22 (1998) 61±81
8
^f …t1 ; s1 † < 3 Zs1
^ P 2 2 …s1 ÿ t† 1 5
Pf …t1 ; s1 † ˆ exp …vd1 ‡ vd2 † ÿ n…u† n…u† du
Pd …t; s1 † : 6 2
9 t
Zs1 =
5
‡ …n…u† ‡ v…u†† ‡ dZ…u† ~ ; …A:12†
;
t
5
where v…t†  …vd1 …s1 ÿ t† vd2 …s1 ÿ t† 0 0† . The futures settlement price is thus
2 8
^
Pf …t; s1 † <ÿ  …s1 ÿ t†3
~
H …t; s1 † ˆ  EQ 4 exp m2d1 ‡ m2d2
Pd …t; s1 † : 6
9 3
1
Zs1 Zs1 =
ÿ n…u† n…u† du ‡ …n…u† ‡ v…u†† dZ …u† Ft 5:
r r ~ …A:13†
2 ;
t t

With respect to the ®ltration generated by the Brownian motion Z( ), the


stochastic integral in the exponential term on the right-hand side of
Eq. (A.13) is independent of this ®ltration (thanks to Novikov's criterion),
and is distributed as a Gaussian random variable. Result Eq. (11) follows.

A.2. Derivation of Eqs. (15) and (16)

Applying Ito's lemma to Eqs. (10) and (11), one obtains the dynamics of the
forward contract price and the futures contract settlement price as follows:
dG…t; s1 †
ˆ lG …t; s1 † dt ‡ m…t; s1 † dZ~…t†
r
…A:14†
G…t; s1 †
and
dH …t; s1 †
ˆ ‰lG …t; s1 † ÿ q…t; s1 †…s1 ÿ t†Š dt ‡ m…t; s1 †r dZ~…t†: …A:15†
H …t; s1 †
Now applying Ito's lemma to Eq. (14) we get
dWG …t† ˆ p dS …t† ‡ Pd …t; s1 †bG …t† dG…t; s1 †: …A:16†
Substituting for Eqs. (7) and (A.14) in Eq. (A.16), one obtains
dWG …t† ˆ lG …t† dt ‡ ‰pS …t†mS ‡ Pd …t; s1 †bG …t†G…t; s1 †m…t; s1 †Š dZ~…t†;
r

…A:17†
where
lG …t†  pS…t†lS …t† ‡ Pd …t; s1 †bG …t†G…t; s1 †lG …t; s1 †: …A:18†
The instantaneous variance of this wealth is
‰pS …t†mS ‡ Pd …t; s1 †bG …t†G…t; s1 †m…t; s1 †Šr ‰pS …t†mS
‡Pd …t; s1 †bG …t†G…t; s1 †m…t; s1 †Š …A:19†
A. Lioui / Journal of Banking & Finance 22 (1998) 61±81 79

and Eq. (15) follows.


To derive the dynamics of Eq. (13), apply Ito's lemma to Eqs. (12) and (13)
to get:
dX …t† ˆ rd …t†X …t† dt ‡ bH …t† dH …t; s1 †; …A:20†

dWH …t† ˆ p dS …t† ‡ rd …t†X …t† dt ‡ bH …t† dH …t; s1 †: …A:21†


Substituting for Eqs. (7), (A.15) and (A.20) in Eq. (A.21), one gets
dWH …t† ˆ lH …t† dt ‡ ‰pS …t†mS ‡ bH …t†H …t; s1 †m…t; s1 †Š dZ~…t†;
r
…A:22†
where
lH …t†  pS …t†lS …t† ‡ rd …t†X …t† ‡ bH …t†H …t; s1 †‰lG …t; s1 †
‡q…t; s1 †…s1 ÿ t†Š: …A:23†
The instantaneous variance is
‰pS …t†mS ‡ bH …t†H …t; s1 †m…t; s1 †Šr ‰pS …t†mS ‡ bH …t†H …t; s1 †m…t; s1 †Š …A:24†
and Eq. (16) follows.

A.3. Proof of Rule 1

Using bG …t† given in Eqs. (15) and (A.17), one can write the hedger's wealth
volatility as follows:
pS …t†mS ‡ Pd …t; s1 †bG …t†G…t; s1 †m…t; s1 †
r
m…t; s1 † mS
ˆ pS …t†mS ÿ pS …t† r
m…t; s1 †: …A:25†
m…t; s1 † m…t; s1 †
Similarly, using Eqs. (16) and (A.22), the hedger's wealth volatility when fu-
tures are used is
pS …t†mS ‡ bH …t†H …t; s1 †m…t; s1 †
r
m…t; s1 † mS
ˆ pS …t†mS ÿ pS …t† r
m…t; s1 † …A:26†
m…t; s1 † m…t; s1 †
which yields the desired result.

A.4. Proof of Rule 2

Using Eqs. (A.18) and (15), the drift of the hedger's wealth when forward
contracts are used as hedging instruments is
lG …t† ˆ pS …t†lS …t† ‡ Pd …t; s1 †bG …t†G…t; s1 †lG …t; s1 †
r
m…t; s1 † mS
ˆ pS …t†lS …t† ÿ pS …t† r
lG …t; s1 † …A:27†
m…t; s1 † m…t; s1 †
80 A. Lioui / Journal of Banking & Finance 22 (1998) 61±81

and using Eqs. (A.23) and (16), one shows that


lH …t† ˆ pS …t†lS …t† ‡ rd …t†X …t† ‡ bH …t†H …t; s1 †‰lG …t; s1 † ÿ q…t; s1 †…s1 ÿ t†Š
r
m…t; s1 † mS
ˆ pS …t†lS …t† ‡ rd …t†X …t† ÿ pS …t† r
‰lG …t; s1 †
m…t; s1 † m…t; s1 †
ÿq…t; s1 †…s1 ÿ t†Š: …A:28†
By comparing Eqs. (A.27) and (A.28) one gets Eq. (18).

References

Amin, K., Jarrow, R., 1991. Pricing foreign currency options under stochastic interest rates.
Journal of International Money and Finance 10, 310±329.
Amin, K., Morton, A., 1994. Implied volatility functions in arbitrage free term structure models.
Journal of Financial Economics 35, 141±180.
Bakshi, G., Chen, Z., 1996. Equilibrium valuation of foreign exchange claims. Journal of Finance
52, 799±826.
Bardhan, I., 1995. Exchange rate schocks, currency options and the Siegel Paradox. Journal of
International Money and Finance 14, 441±458.
Benninga, S., Protopapadakis, A., 1994. Forward and Futures prices with markovian interest rates
processes. Journal of Business 67, 401±421.
Brealey, R., Kaplanis, E., 1995. Discrete exchange rate hedging strategies. Journal of Banking and
Finance 19, 765±784.
Briys, E., Solnik, B., 1992. Optimal currency hedge ratios and interest rates risk. Journal of
International Money and Finance 11, 431±445.
Cornell, B., Reinganum, M., 1981. Forward and futures prices: evidence from the foreign exchange
market. Journal of Finance 36, 1035±1045.
Cox, J., Ingersoll, J., Ross, S., 1981. The relation between forward and futures prices. Journal of
Financial Economics 9, 321±346.
Dezhbakhsh, H., 1994. Foreign exchange forward and futures prices: are they equal. Journal of
Financial and Quantitative Analysis 29, 75±87.
Due, D., Stanton, R., 1992. Pricing Continuously resettled contingent claim. Journal of
Economic Dynamics and Control 16, 561±573.
Dumas, B., Jennergren, P., Naslund, B., 1995. Realignment risk and currency option pricing in
target zones. European Economic Review 39, 1523±1544.
El-Jahel, L., Lindberg, H., Perraudin, W., 1996. Yield curves with jump short rates, The Institute
for Financial Research (Working Paper).
Figlewski, S., Landskroner, Y., Silber, W., 1991. Tailing the hedge: Why and How. Journal of
Futures Markets 11, 201±212.
Glen, J., Jorion, P., 1993. Currency hedging for international portfolios. Journal of Finance
XLVIII, 1865±1886.
He, H., Pearson, N., 1991. Consumption and portfolio policies with incomplete markets and short
sale constraints. Journal of Economic Theory 54, 259±304.
Heath, D., Jarrow, R., Morton, A., 1992. Bond pricing and the term structure of interest rates: a
new methodology for contingent claims valuation. Econometrica 60, 77±105.
Herbst, A., Swanson, P., Caples, S., 1992. A redetermination of hedging strategies using foreign
currency futures contracts and forward markets. Journal of Futures Markets 12, 93±104.
A. Lioui / Journal of Banking & Finance 22 (1998) 61±81 81

Howard, C., d'Antonio, L., 1984. A risk-return measure of hedging e€ectiveness. Journal of
Financial and Quantitative Analysis 19, 101±112.
Howard, C., d'Antonio, L., 1987. A risk-return measure of hedging e€ectiveness: a reply. Journal of
Financial and Quantitative Analysis 22, 377±381.
Karatzas, I., Lehoczky, J., Shreve, S., Xu, G., 1991. Martingale and duality methods for utility
maximization in an incomplete market. SIAM Journal Control and Optimization 29, 702±730.
Karatzas, I., Shreve, S., 1991. Brownian motion and stochastic calculus, 2nd ed. Springer, Berlin.
Kroner, K., Sultan, J., 1993. Time-varying distributions and dynamic hedging with foreign
currency futures. Journal of Financial and Quantitative Analysis 28, 535±551.
Lioui, A., Poncet, P., 1996. Optimal hedging in a dynamic futures market with a non-negative
constraint on wealth. Journal of Economic Dynamics and Control 20, 1101±1113.
Longsta€, F., Schwartz, E., 1992. Interest Rate Volatility and the term structure: A Two ± Factor
General equilibrium model. Journal of Finance XLVII, 1259±1282.
Meulbroek, L., 1992. A comparison of forward and futures prices of an interest rate sensitive
®nancial asset. Journal of Finance XLVII, 381±396.
Subrahmanyam, M., 1996. The term structure of Interest rates: Alternative Paradigms and
Implications for Financial Risk Management. Geneva Papers on Risk and Insurance Theory
21, 7±29.
Tong, W., 1996. An examination of dynamic hedging. Journal of International Money and Finance
15, 19±35.

You might also like