Informs: INFORMS Is Collaborating With JSTOR To Digitize, Preserve and Extend Access To Operations Research

INFORMS
On the Integration of Production and Financial Hedging Decisions in Global Markets

Author(s): Qing Ding, Lingxiu Dong and Panos Kouvelis
Source: Operations Research, Vol. 55, No. 3 (May - Jun., 2007), pp. 470-489
Published by: INFORMS
Stable URL: http://www.jstor.org/stable/25147094
Accessed: 28-10-2015 10:22 UTC
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.
INFORMS is collaborating with JSTOR to digitize, preserve and extend access to Operations Research.
http://www.jstor.org
This content downloaded from 142.103.160.110 on Wed, 28 Oct 2015 10:22:55 UTC
All use subject to JSTOR Terms and Conditions
Operations Research infflflflTfl.
Vol. 55, No. 3, May-June 2007, pp. 470-489
ISSN0030-364X|eissn 1526-5463 10715503 10470 l 10.1287/opre. 1070.0364
?2007 INFORMS
On the Integration of Production and Financial

Hedging Decisions in Global Markets
Qing Ding
Lee Kong Chian School of Business, Singapore Management University, Singapore 178899, dingqing@smu.edu.sg
Lingxiu Dong, Panos Kouvelis

John M. Olin School of Business, Washington University in St. Louis, St. Louis, Missouri 63130
{dong@wustl.edu, kouvelis@wustl.edu}
We study the integrated operational and financial hedging decisions faced by a global firm who sells to both home and
foreign markets. Production occurs at a single
either facility located in one of the markets or at two facilities, one in each
market. The company has to invest in capacity before the selling season starts when the demand in both markets and the
currency exchange rate are uncertain. The currency exchange rate risk can be hedged by delaying allocation of the capacity
to specific markets until both the currency and demand uncertainties are resolved and/or by buying financial option contracts
on the currency exchange rate when capacity commitment is made. A mean-variance utility function is used to model the
firm's risk aversion in decision making. We derive the joint optimal capacity and financial option decision, and analyze the
of the delayed allocation option and the financial options on capacity commitment and the firm's performance. We
impact
show that the firm's financial hedging strategy ties closely to, and can have both quantitative and qualitative impact on, the
firm's strategy. The use, or lack of use of financial hedges, can go beyond affecting the magnitude of capacity
operational
levels altering chain structural choices, such as the desired location and number of production facilities
by global supply
to be employed to meet global demand.
Subject classifications: inventory/production: capacity, allocation, stochastic; finance: hedging, currency exchange rate;
utility: mean-variance.
Area of review: Manufacturing, Service, and Supply Chain Operations.
History: Received July 2004; revision received August 2005; accepted April 2006.
1. Introduction to use operational strategies as effective hedges against

exchange rate and input price uncertainties. Operational
1.1. Problem Motivation hedging strategies, as clearly defined and illustrated in
As firms locate activities of their supply chain all over the Cohen and Huchzermeier (1999), Cohen and Mallik (1997),
world, and products flow across national boundaries, man
and Kouvelis (1999), can be viewed as real compound
options that are exercised in response to the demand,
agers face the uncertainties and complexities of the global
price, and exchange-rate contingencies faced by firms in a
environment. Exchange rates and price uncertainties in pro
global supply chain context. Such real options include post
duction inputs are two of the complicating factors in the
ponement of assembly and distribution logistics decisions,
global supply chain environment. Exposure to exchange
delaying final commitment of capacity and process technol
rates, in particular, affects the underlying economics of any
ogy investments, and/or switching production locations or
firm dealing with foreign buyers, suppliers, or competitors
sourcing partners contingent on demand and/or exchange
through its impact on input costs, sales prices, and volume. rate scenarios. (We are going to restrict our attention to
Such currency fluctuations can be significant (fluctuations
operational hedges with the real option to postpone the
of 1% in a day or 20% in a year are not unheard of) with
deployment of some of the firm resources in response to
drastic impact on production and sourcing costs (Dornier demand and exchange-rate/price scenarios. For the reader
et al. 1998). interested in switching options, see the work of Kogut
have employed different risk-management
Companies and Kulatilaka 1994, Triantis and Hodder 1990, Li and
approaches to cope with exchange rate and input price Kouvelis 1999.)
uncertainties. The typical way is to use financial markets, Even though substantial literature has been developed
whenever possible, to hedge against such risks. Currency on both the financial hedging 1996, and
(see O'Brien
options are the most frequently used tools for hedging references therein) and the operational hedging practices
currency exposure (O'Brien 1996). Options are financial of price and currency risks (Cohen and Mallik 1997,
instruments that allow a firm to buy the right, but not the Kouvelis 1999, and references therein), very little effort
obligation, to sell or buy currencies at set prices. A some has been spent in developing an all-encompassing risk
times overlooked option, but an effective one, is for firms management approach that effectively integrates financial
470
Ding, Dong, and Kouvelis: On the Integration of Production and Financial Hedging Decisions in Global Markets
Operations Research 55(3), pp. 470-489, ? 2007 INFORMS 471
and operational hedges. This weakness of the literature One could start from the classical papers on
uncertainties.
is clearly pointed out, and outlined as a future research option prices of Black and Scholes (1973), Merton (1973),
direction, in Cohen and Huchzermeier (1999), while the to the more currency option specific papers of Cornell and
potential for its effectiveness is anecdotally exposed via Reinganum (1981), Biger and Hull (1983), Jorion (2001),
examples in Dornier et al. (1998, Chapter 9). Our research Shastri and Tandon (1986), and Bodurtha and Courtadon
takes a few steps in addressing this gap in the literature. (1986). This literature develops creative financial instru
For more detailed positioning of our research within the ments and values them to hedge uncertain magnitude cash
relevant literature, see ?1.2: Literature Review. flows, due to price/exchange rate uncertainties, but without
We study the integration of operational and financial consideration of how production decisions and operational
hedging policies of risk-averse global firms within a styl hedging schemes might be interacting with the magnitude
ized, but representative, modeling setting. We consider and variance of such cash flows.
a firm selling to both home and foreign markets. In a The international finance literature, in particular the re
two-stage decision framework, early capacity/production search stream dealing with defining and measuring differ
commitments and financial hedging are decided in the pres ent types of exchange rate exposure (Hodder 1982, Flood
ence of demand and exchange-rate (price) uncertainty in the and Lessard 1986), clearly recognizes the need for a com
first stage, while in the second stage, and after observing bination of financial and operational in effectively
options
demand and exchange-rate realizations, the firm exercises to exchange rate movements.
managing operating exposure
its production "allocation" option in supplying the domes However, beyond some intuitive and anecdotal level, dis
tic and foreign market demand (i.e., how many units to cussions on the relationships between operational flexibil
"localize" and distribute to the two markets). The emphasis
ity, financial hedging, and exchange risk (see Lessard and
of our analysis is on clearly establishing the value of the 1986, or a textbook-level exposition in Shapiro
Lightstone
joint use of the operational hedge ("allocation" option) and 1988), no structural models or quantitative tools are pro
the financial hedge, and understanding their effects on a vided to aid the integrated operational-financial hedging
risk-averse firm's capacity decisions and performance. Fur decision making in uncertain price/exchange rate environ
thermore, interesting insights are obtained on the nature of ments. Our research addresses this issue by clearly relat
optimal, or whenever possible perfect, financial hedges for ing capacity/production plan choices to financial hedging
our modeled environment. and that for the risk-averse firm the pro
strategies, showing
duction plans it chooses are functions of both the financial
1.2. Literature Review
hedging strategies it employs and the opportunity to exer
The literature on operational hedging practices of price cise a real option ("allocation" option) contingent on price
and exchange rate uncertainty is recent, with the work of and market demand scenarios.
Huchzermeier and Cohen (1996) the most influential from The work of Mello et al. (1995) is the one closest in
a modeling perspective. Their results demonstrate the bene spirit to our research. They present an integrated model
fits of operational hedging practices via excess capacity and of a multinational firm with flexibility in sourcing its pro
production switching options in environments of volatile duction (i.e., a switching option in sourcing from differ
ent countries) and with the use of financial markets to
exchange rates. The work of Kogut and Kulatilaka (1994)
is along the same direction with the emphasis on explicitly hedge exchange-rate risk. The emphasis of Mello et al.
valuing the option of shifting production between two plants (1995) is on valuing dynamically the operating exposure
located in different countries as exchange rates fluctuate. to exchange-rate movements via a
state-contingent model
Kazaz et al. (2005) illustrate that, for an expected profit in continuous time, with the model explicitly accounting
maximizing firm, production hedging policies of excess for the strategic exercise of the switching option (sourc
capacity and postponed allocation are features of robust ing flexibility) of the firm. Numerical results illustrate the
optimal performance under exchange rate uncertainty. Li interdependency of sourcing flexibility and hedging strat
and Kouvelis (1999) explicitly study flexible and risk egy in increasing the value of the firm. Chowdhry and
sharing supply contracts under price uncertainty. Their dis Howe (1999) examine the capacity allocation and finan
cussion clearly illustrates how operational flexibility, sup cial hedging decisions in a setting where the total pro
plier selection, and risk sharing, when carefully exercised, duction capacity is fixed. Our research focus is different,
can effectively reduce the sourcing cost in environments with an emphasis on understanding the implication of the
of price uncertainty. For a thorough coverage of the vast use of the operational hedge (via an "allocation" option)
global supply chain literature, the detailed positioning of and the financial hedge (via currency option contracts) on
the operational hedging research stream within it, and other the capacity investment decision and firm's mean-variance
work peripherally related to it, see Cohen and Mallik (1997), (MV) performance. Our model allows us to obtain closed
Cohen and Huchzermeier (1999), Van Mieghem (2004), and form formulas for both production plans and currency con
Boyabath and Toktay (2004). tract parameters, and thus to quantify the magnitude of the
There is a vast literature on the use of financial hedg interaction of operational and financial hedges in an uncer
ing instruments to better manage price and exchange rate tain currency exchange rate and demand environment.
472 Operations Research 55(3), pp. 470-489, ?2007 INFORMS
Risk-neutral newsvendor models with cost (price) uncer financial hedges in such a context. Building on the intu
tainty as implied in uncertain over-and-under stocking costs ition of these results, in ?5 we proceed to analyze the
are examined in the work of Lowe et al. (1988). Kouvelis same problem for a risk-averse firm selling to both domes
and Gutierrez (1997) consider price uncertainty, due to tic and foreign markets with stochastic demand. We show
uncertain exchange rates, in a newsvendor
problem in a that some of the intuitive findings for the one-market case
global market under risk neutrality and two ordering oppor in ?4 no longer hold for the two-market case. Section 6
tunities. Gurnani and Tang (1999) consider the issue of illustrates that the use of financial hedge can affect supply
demand and forecast updating and uncertain unit costs in a chain structural decisions such as location and the number
instants for a risk-neutral newsvendor. of production facilities to meet global demand. We con
two-ordering setting
clude with a summary of main insights and discussions of
Uncertain prices have been studied also in the setting of
spot market as an alternative source of buying and selling, future research in ?7.
with the focus of finding the optimal balance of long-term
contract and spot sourcing (see Seifert et al. 2004, Yi and 2. Model
Scheller-Wolf 2003 for risk-neutral settings, and Wu and We analyze the production and financial hedging deci
Kleindorfer 2005, Dong and Liu 2007 for risk-averse set sions of a global firm selling to two markets: market 1, its
tings). Our work differs from the above models in its focus "domestic" market (defined for our purpose as the market
on the risk-averse global newsvendor with a single-ordering in the "home the
trading country" currency, i.e., currency
(capacity building) opportunity and exploiting the com the firm uses to report its consolidated financial statements),
bined use of an "allocation" option and financial hedges for and market 2, a "foreign country" market with an uncertain
MV-utility optimization. currency exchange rate.
Risk aversion issues in inventory and capacity manage We use a two-stage stochastic program to model the
ment have been captured in the work of Bouakiz and Sobel firm's decisions. In the first stage, a "capacity-production"
(1992), Eeckhoudt et al. (1995), Agrawal and Seshadri
plan for the production facility is developed, and appro
and Chen et al. Most of these models
(2000), (2004). priate financial hedging contracts on the foreign currency
are variations of the classical newsvendor problem under are purchased in the presence of uncertainty in market
different risk-averse objective functions. While our work demands, exchange rate, or both. the first
During stage,
obviously has similarities with the above work in terms the firm invests in needed technology, equipment, and fac
of incorporating risk aversion and MV trade-offs, within tory space, or modifies existing facilities, in anticipation
a newsvendor network setting, our decision emphasis is of market needs. With capacities in place, commitment
on integrated operational (capacity/inventory) and financial of production resources as part of a first-stage production
hedging decisions, which leads to completely different opti plan may occur, frequently in an effort to provide quick
mization problems and managerial insights. response to foreign market demand by executing, prior to
Research efforts to hedge operational risk via financial demand realization, long lead time production activities
instruments that exploit demand correlation with tradable (such as acquisition of raw materials, production of com
market assets appeared early in Anvari (1987) and Chung plex components and subassemblies, or even nonmarket
(1990), with the study of newsvendor models within a capi specific "vanilla"-middle products). In the remainder of this
tal asset pricing model (CAPM) framework. More recently, we will use to refer to such a
paper, "capacity" "capacity
Gaur and Seshadri (2005) demonstrate the effectiveness production" plan. In the second stage, after observing the
of financial hedging when one can discover tradable mar demand and exchange-rate realization, the firm makes pro
ket assets partially correlated with market demand, while duction "allocation" (e.g., how many units to
decisions
distribute in each
Caldentey and Haugh (2006) provide a modeling frame appropriately configure?"localize"?and
work that allows continuous trading in the financial mar market) with the necessary distribution and logistic costs
ket. Our work differs from the above models because it to optimize its profits. The allocation option represents the
does not emphasize the financial hedging of demand uncer firm's flexibility in choosing ex post between domestic and
tainty via traded market assets, but instead searches for foreign markets to sell its products. Postponement capabil
ity of manufacturing/assembly activities and the access to
integrated risk-management approaches that combine oper
ational hedging (via postponed production allocation) with the worldwide distribution channel are required for such
financial hedging instruments to mitigate the price risk flexibility.
associated with the currency exchange-rate volatility.
Thestructure of this paper is as follows: In ?2, we 2.1. Real Options ("Postponed Allocation")
provide a modeling framework for joint production and Modeling
financial hedging decisions. In ?3, we present the optimal We use the following notation:
financial hedging strategy for a given capacity global sup s: foreign market currency exchange rate in stage 2,
ply chain. In ?4, we focus on a

special
case of a risk-averse that is, the value of one unit of the foreign currency mea
firm selling exclusively to a foreign market with stochas sured in the domestic currency;
tic demand and clearly explain the role of operational and X: capacity (vector) reserved in stage 1.
Operations Research 55(3), pp. 470-489, ?2007 INFORMS 473
c: unit capacity reservation cost (vector) in stage 1 be viewed as a portfolio of real options on the exchange
(in home currency). rate, some of them of a call-option nature (see the first term
pt: price per unit in market / in stage 2 (in market i in (1)) and some of a put-option nature (see the second
currency), /= 1,2. term in (1)). Following the obvious analogy to financial
d={dt)t=x: random variable (vector) that represents the options, we often refer to Sc and Sp as the "real-option
demand in two markets. exercise prices." The more concrete interpretation and the
motivation for the use of the various terms in (1) will be
7rop(X, s, d): for a given capacity reservation X, the opti
mal operations profit in stage 2 after the realization of given in the discussion of the specific cases/examples that
demand d and exchange rate s. follow. Let us start with
e{): probability density function (PDF) of the currency 2.1.1. Case 1: One Facility and Localization at the
exchange-rate distribution. Source. The firm has a single production facility located
g{-, ): PDF of the joint distribution of demand in two in market 1 (i.e., the domestic market), and the stage 2
markets. localization operation is also conducted at this facility.
= 1, 2.
gi(-): PDF of demand distribution for market /, i Define
Gt{), Gf(-): cumulative distribution function (CDF) and Tt: relevant unit localization costs for market i shipped
the complementary CDF of demand in market /, respec in stage 2 (in market 1 currency), /= 1,2.
= ?
tively,
i= l,2. rx px tx > 0: incremental profit per unit sold in
/( , ): density function of joint distribution of exchange stage 2 in market 1.
= ?
rate and demand. r2(s) sp2 t2: incremental profit per unit sold in
T: superscript that represents the transpose operator. stage 2 in market 2.
*: superscript that represents optimal decisions and cor Thus,
outcomes.
responding 7T?P(X,S,d)
x+ = max(x, 0). ?
Usually firms do not adjust selling prices immediately at
rxmin((Z d2)+, dx) + r2(s) min(X, d2)
the swing of the currency exchange rate due to its poten if r2(s)^rx,
tial long-term impact on firms' market shares. Thus, for
= -
the short- to medium-term planning problem studied in this rxmin(X, dx) + r2(s) min((X dx)+, d2) (2)
paper, and especially for the second-stage production allo if rx > r2(s) ^0,
cation decisions, we assume that the selling price in the
rxmin(X,if r2(s) < 0.
foreign market, p2, is fixed. Further motivation for this dx)
assumption is a perfectly competitive foreign market with As (2) suggests, when r2(s) ^ rx, we first allocate capac
a host of local and multinational firms contributing to the ity to meet demand in market 2, and the remaining units,
equilibrium price, with our firm having no control over such if any, are localized for the needs of market 1. The allo
prices. cation priorities are reversed when rx > r2(s) ^ 0. For the
The functional form of 7rop(X, s, d) can vary depending case r2(s) < 0, we only allocate the capacity for market l's
on where the global firm locates its production facilities, needs. (2) can be rewritten as
and where the configuration adjustment ("localization") of = - -
7Top(Z, s, d) r2+(s)min((X dx)+, d2) + (r2(s) rx)+
the vanilla products to the market-specific final products is
- -
conducted. We find that the following functional form of (min(X, d2) min((X dx)+, d2))
7Top(X, s, d) holds for a variety of assumptions on local + rxmin(X,dx), (3)
ization activities of stage 2 (as we will explain in detail ?
that is, a0 = 0, ax = p2min((X dx)+, d2), Sf
=
r2/p2,
later): = - - =
a2 p2(mm(X, d2) min((X dx)+, d2)), Sf r2/p2 +
~ =
= - (Pi T\)/Pn and c r\min(X,dx).
7Top(X, s, d) a(X, d){{s SC)+)T a
Similarly, single productionfacility in the foreign mar
+ b(X,d)((S/>-s)+)T + c(X,d) ket with localization conducted there results in
= -
7Top(X, s, d) r2(s) min(X, d2) + (rx(s) r2(s))+
=
J2ai{X,d){s-Sfr - -
(min(Z, dx) min((Z d2)+, dx))
m
+ rx+(s)min((X-d2)+,dx),
+ ?>,(X,d)(Sf-s)+ + c(X,d), (1) ? ?
where rx(s) = =
1=1 px stx, r2(s) s(p2 r2), and rt is the unit
localization cost for market i shipped in stage 2 (in mar
where a(X,d) = (^(X,d))f=0, b(X,d) = (^d))^, ket 2 currency), /= 1,2. That is, a0 = (p2 ? r2) min(X, d2),
Sc = (Sf)iU, and Sp = {Sr)?=x, with 50c= 0, 0 < Sf < = ? - -
Sf b\ (Pi + T\ r2)(min(X,^1) min((X d2)+,dx)),
and 0 < Sf < for 1^ / < j, and n and m are integers. = = - =
Sj S? Pj(p2 + Ti- r2), b2 rxmin((X d2)+, dx), Sp2
(1) suggests that a firm's second-stage operations profit can px/rx, and c= 0.
AIA Operations Research 55(3), pp. 470-489, ?2007 INFORMS
2.1.2. Case 2: One or Two Facilities and Localiza 2.1.3. Case 3: Two Facilities Without Postponement.
tion in the Target Markets. The firm has two production The firm has two production facilities with facility i located
facilities with facility i located in market /, i= 1,2. Both inmarket /, i= 1,2. Both facilities produce market-specific
facilities can produce vanilla products in the first stage, and products and ship to the target market in the first stage.
facility i can configure the vanilla products to market j Define
Xtj
as facility f s stage 1 production shipped tomar
specific products, i,j =1,2. We assume that there are no ket j, i, j =1,2. In this case, the stage 2 operations profit
capacity constraints for the localization operation in stage 2 can be written as
(e.g., local content is added at ample capacity assembly and
distribution facilities). Define: 7rop= px min(Z11 + X2X, dx) + sp2 min(X12 + Z22, d2).
Ty: unit localization cost of configuring facility i vanilla
products to satisfy demand in market j (measured in That is, a0 = p2 min(Z12 -f X22, d2) and c = px min(Xn +
market j currency), /= 1, 2 and j = 1, 2.
X2X,dx).
tomarket j (mea
rtf. unit profit of selling facility / product ? Other possible variations of those cases along the same
sured in market 1 currency). Specifically, =
rxx px lines (e.g., two facilities localization at the source) satisfy
= - =
Tin s(Pi r2X =px- r21, and
(1) but are omitted for brevity.
rn(s) r12), r22{s)
~
S{Pl T2l)'
It is reasonable to assume, without loss of generality,
that txx < r21 and r22 < t12. That is, configuring facility 2.2. Financial Options
f s vanilla products into market j {î) products is more We now describe the type of financial contracts we are
expensive than configuring facility j's to satisfy market j considering, and how they can be valued. We consider
demand. It follows that rxx ^ r2l and r22{s) ^ ru(s). We currency option contracts. We will conduct most of our
can also assume that we have
rtj ^ 0. Then, analysis with call- and put-option currency contracts. Con
sider a portfolio of call- and put-currency contracts h =
7T?P{X,S,d)
- - ((SC,SP),(QC,QP)), where Sc
= and SP =
= (SQ)Zo
rx2{s) rmn{{d2 dx)+)
X2)+, {Xx are row vectors
(Sp,)"!lx the of exercise prices of the call
- and and and
+ r22{s) min(X2, d2) + {rX2{s) rlx)+ options put options, respectively, Qc=((2c.)|!f1
- - - Qp=(Qp.)"!1x are the row vectors of the corresponding con
(min^, dx) min((X1 {d2 X2)+)+, dx)) tract sizes. The payoff of portfolio h in stage 2 is
-
+ {r2i r22{s))+
= + (SP-s)+Ql
- - - Rh(s) (s-Sc)+Ql
(min(Z2, d2) min((X2 {dx Xx)+)+, d2))
- - = - -
+ rxx rmn{Xx,dx) + r2X min((Z2 d2)+, {dx Xx)+). ?(* Sc)+QCi+ Y,(SPi s)+QPr (4)
- - - /=1
That is, a0 = {p2- rX2)min((d2
1=0
X2)+, {Xx dx)+) + {p2
= ? ? ?
r22)min(X2, d2), ax {p2 TX2){min{Xx,dx) min^^ Let C(S) denote the price of a unit call or put option with
- = - - =
{d2 x2yy,dx)), sf - {Px txx)/{P2 r12), bx exercise price S in stage 1, which is decided in the cur
- - -
{Pi r22){min{X2,d2) min((Z2 {dx Xx)+)+,d2)), rency exchange rate market or determined by the option
= ~ - =
S? (Pi T2X)/{p2 r22), and c rxxrmn{Xx,dx) + pricing theory. For example, if the currency exchange rate
? ? The derivation of the
r21min((Z2 d2)+,{dx î)+)- in stage 2, s, follows a lognormal distribution, then C(S)
above expression is given in the online appendix that can can be determined by using Black-Scholes' valuation (see
be found at http://or.journal.informs.org/.
Black and Scholes 1973). Let H(h) be the cost of acquiring
The special case of a single production facility (located
financial contract h incurred in stage 1.We have
in either a domestic or foreign market) with the localiza
tion of shipped products performed in the target market is
= +
given by H(h) t,C(Sc)QCi ?,C(SPi)QPr
i=0 i=l
7T0p(X,S,d)
= - - The risk premium of a unit call (respectively, put) option
r2{s) min((Z dx)+, d2) + {r2{s) rx)+
- - (S, Q) is defined as
(min(X, d2) min((X dx)+, d2)) + rxmin(X, dx),
= ? = ?
and is the unit = - -
where rx px rx, r2{s) s{p2 t2), rf AC(5) C(S)eyT E[(s S)+]
localization cost for market / shipped in stage 2 (in market = - -
/ currency), /= 1, 2. That is, (respectively,AC(S) C(S)eyT E[(5 s)+]), (5)
= ~
0o (Pi r2)min{{X-dx)+, d2), where y is the risk-free interest rate in home currency, and
= T is the time-to-maturity of the option. The risk premium
a\ (P2-T2)(min(^' d2)-rmn{{X-dx)+, d2)),
of a unit portfolio h is defined as a row vector
Sl=(Pi-rx)/{p2-T2),
and c = rxmin(X, = (6)
dx). AC(SC, S?) ((AC(SC())?,, (AC(SP))Zi).
Forward contracts are commonly used to hedge foreign MV approximation to utility functions of constant absolute
currency risk and can be viewed as special cases of call risk aversion (CARA) for the single-newsvendor problem
options. Consider that the firm buys a forward contract and the sequential newsvendor network, and finds that the
hi = {f,Q) in stage 1, where / is the forward exchange difference between utility-optimal paths are small, and the
rate to be used for the delivery of Q in stage 2. Then, the MV optimal resource levels typically underestimate the risk
payoff of hi in stage 2 is adjustments, but the difference is small when risk adjust
ment is small. Justification for the use of total risk measure
= - = "
*?(') (' f)Q (-(/ SL)Qe~yT)eyT + (s- sL)Q, (such as V[-]) instead of measures of systematic risk is
provided in Hodder and Dincer (1986) with the main argu
where sL = infjs}. Observe that RhX{s) can be viewed ment being that managers are typically concerned about
as the sum of the stage 2 payoff of call option h2 = total risk. This can be attributed to a concern with the prob
{sL, Q) and the stage 2 value of a fixed payment of ability of financial distress, or bankruptcy, as well as to
?
(/ sL)Qe~yT made in stage 1. The portfolio replication agency considerations (see Jensen and Meckling 1976 and
argument requires that the cost of the call option hi in stage Markus 1982). Therefore, the firm's production and finan
1 should be H{h2) = (/ - sL)Qe~yT} Thus, the stage 2 cial hedging problem is
payoff of forward contract hi is the same as the net payoff
of call option hi in stage 2. Similarly, we can establish that max \E[7r(X,h,s,d)]-\y[7r(X,h,s,d)]], (8)
a forward contract can be viewed as a special case of a put
=
option with exercise price su sup{s}. By considering only where n is a given feasible financial hedging set (in this
call and put options, we include forward contracts in the
paper, portfolios of call and put options). In the remainder
feasible set of contracts.
of this paper, we will use the MV objective when referring
to the firm's utility.2
2.3. Integrated Risk-Management
Problem Formulation
3. Optimal Financial Hedging Strategy
We can write the firm's profit in stage 2 as
In this section, we present a general result on the opti
mal hedging sizes for a portfolio of call- and put-option
7r(X, h, s, d)
- currency contracts for a given production capacity X. Let
= + 7Top(X, s, d) + Rh{s),
(-cX H{h))eyT (7) S = (Sc, SP) be the exercise price vector of the portfolio
of call and put options. Define cov(x, y) as the covariance
where T is the time duration between stage 1 and stage 2.
matrix of random vectors x and y, and V[x] as the variance
We assume throughout, unless otherwise explicitly noted, x.
covariance matrix of random vector
that the firm is risk averse. Furthermore, we assume that
(a) the firm's objective is to maximize the expected util Proposition 1. Given capacity X and a portfolio of call
for the 2 as and contracts with exer
ity stage profit expressed by (7), and put-option currency prespecified
cise price vector S = in which Sc and SP do not
(b) the firm's expected utility is represented by (Sc, SP)
have common elements, the firm's MV is concave
objective
= in the hedging size Q, and the optimal size vec
iy(7r) E[ir]-AV[7r], hedge
tor Q*(X, S) is the unique solution to the system of linear
where E[] and V[-] are the expectation and variance oper equations
ators taken on the joint distribution of exchange rate and
demand, respectively, and A ^ 0 represents the rate at which = - -
-(AC(S))T 2A(V[((* Sc)+, (S, *)+)]Q*T
the firm will substitute variance for expected value, also
referred to as the MV ratio. It is well known (see Philip +cov(7r?P(X,*,d),((>-Sc)+,(SP-*)+))).
patos and Gressis 1975 and Jucker and Carlson 1976) that (9)
the use of the MV criterion is consistent with the principle
of maximizing expected utility if Proposition 1 reflects the marginal trade-off of adding a
(i) the firm's utility function can be represented by a unit of option portfolio at Q. First, buying one more unit of
quadratic function of time payoff, or the portfolio will affect the firm's expected profit through
(ii) the probability distribution of time payoff is a two the risk-premium term, AC(S), while the profit from the
parameter distribution (e.g., normal distribution). operations decision in the second stage 7Topwill not be
Van Mieghem (2004) discusses the applicability of the affected. Second, the total variance of the profit will be
MV objective in the real-options context. The MV objective affected in two ways?the variance of the portfolio payoffs
offers a second-order approximation of the true objective - -
Rh will increase by \J[{(s Sc)+, (SP s)+)]QT, and the
for a general utility function, and it provides clear insights covariance between the operations profit tt?p and the portfo
to decision makers. Van Mieghem (2004) also conducts lio profit Rh will increase by cov(7rop(X, s, d), ((s ? Sc)+,
numerical studies to investigate the appropriateness of the ?
(SP s)+)). The optimal portfolio size Q*(X,S) should
balance the marginal impact on the expected profit and the (2) For the independent case, if zero risk premium
total variance of the profit. Proposition 1 provides the opti is assumed, then the optimal hedging policy among
mal hedge size for a given capacity and a portfolio of call all call and put options is a portfolio of call options
and put options with prespecified exercise prices. It applies (Sc, -E[a(X, d)]) and put options {Sp, -E[b(X, d)]).
to all functional forms of 7Top.
When the firm only considers trading currency options
We now proceed to ask several questions of interest:
with exercise prices that are the same as the embed
How many call options do we need for optimal hedg
ded real-option exercise prices, Sc and Sp, Proposition 2
ing? What exercise prices of the financial options are good
choices for a given capacity decision? Applying Proposi provides the optimal hedging quantity. The optimal
hedging quantity can be viewed as consisting of two
tion 1 to (1) yields some interesting results. First, observe
that if demand d is deterministic, then the firm's operations parts (see Proposition 2(1 )(a) and (b)). One part,
aims to reduce the
profit in stage 2 is identical to the payoff of a portfolio -V-1(F(5,Sop))cov(7rop,FT(5,Sop)),
of call and put options with exercise Sc and Sp variance of operations profit through the counterbalanc
prices
(i.e.,
use the "real-option exercise prices"
as the contracted ing effect between cash flows from operations and from
exercise prices for the financial options) and contract sizes exercising the currency exchange options; the other part,
referred to as the hedge
a(X, d) and b(X, d), respectively. Thus, selling call and -V-1(F(Jy,Sop))(AC(Sop))T/2A,
put options in the first stage will completely hedge the risk size deviation, reflects the trade-off between the gain/loss
from the rate, zero-variance in risk premium and the gain in the variance of financial
currency exchange i.e., ("per
fect") hedge can be achieved. When the market demand is trading. In particular, when the foreign market demand is
stochastic, zero-variance hedges cannot be achieved. How independent of the currency exchange rate, which implies
ever, we can show that under reasonable the that abT(X, d) is independent of the unit payoff of the real
assumptions
exercise prices Sc and Sp in the firm's real options are options FT(5*, Sop), the financial hedge can help to reduce
indeed the best choices of exercise prices for financial call the operations profit variance caused by the currency fluc
and put options. tuation but not that by the demand uncertainty. Moreover,
First, without loss of generality, we assume that Sc and when the risk premium is zero (Proposition 2(2)), i.e.,
Sp do not have common elements. We can arrange the ele financial trading does not affect the firm's expected profit,
ments of Sc and Sp in (1) in an increasing order to obtain a the optimal hedge is the minimum-variance hedge. Selling
new vector Sop = (S,?P)?LV-Correspondingly, we can define a portfolio of call and put options with Sc and Sp as exer
vector ab(X, d) = (abf (X, d))?+n as cise prices and with E[a(X, d)] and E[b(X, d)] as hedging
quantities will reduce the variance of operations profit to
the element of a(X, d) that is the coefficient the minimum. Proposition 2(2) also highlights that, to some
of (s - S?p)+in tt?v extent, the effective financial hedging contracts are related
if 5?p is an element of Sc, to (and affected by) the operational strategies that a firm
abt(X,d)
= employs. That is, there is a natural linkage between the
the element of b(X, d) that is the coefficient firm's real options and the financial options that the firm
should employ to efficiently hedge its risk. For example,
of (S?p-s)+in7rop
a firm that has two production facilities, one in each mar
if 5?p is an element of Sp.
ket, and conducts localization in target markets (case (2) in
a random vector ?(s, = ?2.1) will need a forward, a call option, and a put option
Define Sop) (Ft(s, Sop))"+On,where to manage the exchange risk; a firm that has two facili
- if S?p is an element of Sc, ties but with no postponement capability (case (3) in ?2.1)
I(s S?p)+
will only need a forward contract to manage the exchange
- if S?p is an element of Sp. risk. We note that the optimal financial hedging strategies
(S,op s)+
in Propositions 1 and 2 are specific to the MV objective
Proposition 2. (1) Given capacity X and exercise prices because they are the result of balancing marginal expected
Sc for call options and exercise prices Sp for put options, profit and profit variance.
(a) The optimal hedge sizes are We will further specialize our results for an operational
setting that was described previously (case (1) in ?2.1):
=
Q?pt _V-1(F(5,S0P)) The firm has one production facility located in the domes
tic market, and stage 2 localization is also conducted there.
(cov(tt?p,FT(s, Sop))+ (AC(Sop))T/2A). This particular operational setting is instrumental for our
(b) If the demand is independent of the currency presentation of results for the joint production capacity
exchange rate, referred to as the independent case, then the and hedging decision in ??4 and 5. Such practice is quite
optimal hedge sizes are common. For example, in the automotive industry, most
of the European brand autos are manufactured in Europe
= _
qopt _E[abT(X) d)j y~\F(s, Sop))(AC(Sop))T/2A. and exported to the United States. For this setting, the
stage 2 operations profit function irop{X,s,d) is given (b) If the joint demand is independent of the currency
by (3). In ?4, we study in more detail its special case in exchange rate, referred to as the independent case, the opti
which the domestic facility serves exclusively the demand mal hedge sizes are
in the foreign market (market 2). That is,
= Q\{X,S?S2)
7Top(Z, s, d2) r2+{s)min(X, d2)
- = - - -
=
p2{s T2/p2)+ min(X, d2). -/>2E[min((X d,)\ d2)] (AC(S,)V[(S S2)+]
- - -
Section 5 focuses on the general case of the domestic facil AC(S2)cov((s 5'1)+, (s S2)+))/(2Adet V),
ity supplying two markets (i.e., 7rop is described by (3)).
The following two corollaries of Proposition 2 illustrate the
q;(x,sx,s2)
? ? ?
optimal financial hedging policies for these two cases. ?/?2E[min(.Y, d2) min((X d^)+, d2)]
Corollary 1. (1) For a domestic production facility sup -(AC(S2M(s-Sx)+]
plying a foreign market, given capacity X and a call - - -
AC(Sx)cov((s Sx)+, (s S2)+))/(2Adet V).
option currency contract with prespecified exercise price
S = T2/p2,
(2) For the independent case, if zero risk premium is
(a) The optimal hedge size Q*{X, r2/p2) is given by
assumed, then the optimal policy among all call and put
e*(x;r2/p2) options is a portfolio of two call options with exercise
- - = = ?
p2co\{{s T2/p2)+ min(X, d2), (s T2/p2)+) prices Sx r2/p2 and S2 r2/p2 + (px rx)/p2, and
respective contract sizes
V[(j-T2/ft)+]
AC(T2/p2) = -
Q\{X, SX,S2) -p2E[min((X dx)+, d2)],
2\v[(s-T2/p2yy = - -
is independent Q*2(X, Sx, S2) -p2E[mm(X, d2) min((X dx)+, d2)].
(b) If the foreign market demand of
the currency exchange rate, referred to as the independent
Some of the above results are based on the assumption
case, the optimal hedge size is
that the demand and currency exchange rate are indepen
-
G*(X,r2/p2)= -p2E[min{X,
d2)] ^f^*-
dent. This is a reasonable assumption for the short-term
2AV[(s r2/p2)+] production allocation planning problem. Furthermore, this
(2)3 For the independent case, if zero risk premium is assumption leads to clean results and helpful intuition on
assumed, then the optimal policy among all call and put the nature of the optimal financial hedging policy. When
options is a single call option with demand and currency exchange rate are indeed partially
S = T2/p2 and = correlated, the interrelatedness of the operational decision

Q*{X, r2/p2) -p2E[min{X, d2)].
(e.g., capacity) and the financial hedge is stronger because
Corollary 2. (1) For a domestic production facility sup a natural hedge of the demand risk might arise by using
plying both foreign and domestic markets, given capacity X the appropriate financial hedge. A firm may need a more
and two contracts at prespecified exer
call-option currency complex optimal financial hedging strategy to achieve the
cise price Sx = r2/p2 and S2 = r2/p2 4- {px ? rx)/p2, minimum variance than what is suggested in Propo
hedge
(a) The optimal hedge sizes are sition 2(2), and the form of the optimal financial hedging
Qi(X,SltS2) strategy will depend on the specific assumption on the cor

relation between demand and exchange rate. For example,
=
-(cov(7t?p,(5-51)+)V[(.-52)+] for the one-market case in ?4, if the foreign demand d2 =
?
where a and b are constants, we can show that
-cov(7rop,(5-52)+)cov((j-5,)+,(5-S2)+))/detV (a bs)+,
the firm will need infinitely many call options to achieve a
zero-variance hedge.4
- - - A few observations about the effect of the risk premium:
AC(S2)cov((s S,)+> (s 52)+))/(2Adet V),
In the exchange-rate market, the risk premium of options
Q*2(X,S{,S2) is typically positive and very small (see Biger and Hull
= 1983). The firm cannot make a lot of money from selling
-(cOV(ir?*,(s-S2)+)\/[(s-Sl)+]
a call option, and the hedge-size deviation is small when
-cov(irop,(5-5l)+)cov((5-5l)+,(i-52)+))/detV the MV ratio is in a reasonable range. From an operations
-(AC(S2)V[(s-S,)+] manager perspective, the main goal of participating in the
- - - currency trading market is hedging the risk in the opera
AC(5!)cov((5 5,)+, (s S2)+))/(2Adet V), tions due to the
profit currency exchange-rate uncertainty,
where det V = V[(s - S,)+]V[(.s -
S2)+]
-
cov2((s
-
S,)+, not gaining profit from the currency exchange market. Such
(s-S2)+). profits often disappear when relevant transaction costs are
applied. Furthermore, from a mathematical tractability per Corollary 3. The optimal risk-neutral capacity X^ is the
spective, assuming zero risk premium offers clean results solution to
and clear insights for the problem. Thus, for the above rea p OO p. oo
=
sons, we will present most of the results in this paper under /
Jr2/P2 J**n
/ r2{s)f{s,d2)dd2ds ceyT.
a zero
risk-premium assumption.
Moreover, for the independent case, X^ is the solution to

4. Domestic Production Facility = ce?r.
Supplying a Foreign Market E[r2+(s)]G2(X*) (11)
We will now study a simplified model in which the produc For a risk-neutral firm, the optimal capacity is expressed
tion facility is located in market 1 (domestic market) and by a news vendor-type formula (11). For a risk-averse firm,
serves exclusively demand in market 2 (foreign market). the newsvendor formula ismodified by a risk-aversion term
This model applies to the situation where either market 1 (see (10)). Comparing (10) and (11), we have the following
demand is zero or the firm uses a separate production facil corollary.
ity to supply market 1. Again, the firm's stage 2 operations Corollary 4. For a domestic production facility supply
profit is a
ing foreign market, if theforeign market demand is inde
= pendent of the currency exchange rate and the risk premium
7Top(Z, s, d2) r+(s) min(X, d2) is zero, then the optimal capacity of a risk-averse firm is
= - less than that a risk-neutral i.e., X* ^ X^.
p2(s r2/p2)+ min(Z, d2). of firm',
4.1. Basic Results 4.2. Analysis of Operational and

Financial Hedges
Proposition 3. For a domestic production facility supply
We will now discuss how operational and financial hedges
ing a foreign market where the foreign market demand is
the a
can help mitigate the firm's exposure to the exchange rate
independent of currency exchange rate, given call
and demand uncertainties. Operationally, the firm might
option currency contract with prespecified exercise price
S = r2/p2, thefirm's MV objective is optimized by a pair of have committed to the foreign market demand and does not
have the option to allocate the capacity contingent upon
unique values (X*, Q*) that solves thefollowing equations:
the currency exchange rate, i.e., 7rop =
r2{s)rmn{X,d2).5
We refer to this case as "without allocation option," as
(E[r2+(s)]+p2AC(r2/p2)
opposed to the case with allocation option where the firm
- - = ceyT,
2AE[r+2(s)]E[(X* d2)+])G2(X*) is not bound to satisfy the foreign market demand and
77op= r2{s)rmn{X,d2) (as we have analyzed in ?4.1).
and a firm can choose between or not
Financially, using using
currency options. Thus, we compare four cases:
exchange
-
Q*= -;>2E[min(X*,
d2)) - (a) without allocation option and without financial hedge
^^f^ T2lp2y],+y
2A\/[(s (?A, ?H), (b) with allocation option and without finan
cial hedge (A, ? H), (c) without allocation option and with
Moreover, if zero risk premium is assumed, then the optimal
financial hedge (?A, H), and (d) with allocation option and
(X\ ?*) solves with financial hedge (A, H). (A graphical representation of
- - comparisons among the four cases is given in Figure 1.)

(E[r2+(*)] 2AE[r+2(s)]E[(X* d2)+]) Proposition 4 shows the effect of the allocation option
= c^r on the firm's optimal capacity, expected profit, and MV
G2(Z*) (10)
objective.
and Proposition 4. Assuming a domestic production facility
supplying a foreign market when the demand and exchange
= are
Q* -p2E[min(X*,d2)]. rate independent.
(a) By implementing an allocation option, a risk-averse

When theMV ratio A = 0, the model reduces to the tradi firm not hedging financially increases its optimal capac
tional risk-neutral problem. Note that if the risk premium is ity, expected profit, and MV objective. That is, X* A _H ^
not zero, the risk-neutral firm will be able to make infinite < and
XI _?, ?!A> _H El _H, WK _H ^ UI _H.
expected profit by trading an infinite number of options in (b) If the risk premium is zero, then by implementing
the currency exchange market. To exclude this pathological an allocation option, a risk-averse firm using the opti
case, we always
assume zero risk premium when discussing mal financial hedge increases its optimal capacity, expected
the risk-neutral problem. Corollary 3 provides the optimal profit, and MV objective. That is, X*_h H ^ X*A H, E*_A H ^
production capacity decision for a risk-neutral firm. ElH,andUlAHÛlH.
Figure 1. Summary of the effects of operational and from operations transactions to reduce the total variance of
financial hedges. the profit. At a given capacity, the optimal financial contract
always reduces the profit variance and thus creates incen
tives for the firm to invest in more capacity. As a result,
Without allocation With allocation
with the financial hedge, the firm always invests more than
I Ioption (-A) option (A)
without the financial hedge. Thus, although the financial
Without financial ? J. A, N
hedge does not increase profit directly, indirectly through
h^rW( mProposition 4(a) the reduction of the profit variance and the increase of the
hedge(-H) R _- _A,-H capacity investment, the firm's expected profit increases.
Because the profit variance increases in capacity invest
-O--L o
ment, the higher MV objective might be at the cost of
With financial g g higher profit variance. This insight applies to both cases of
hedge (H) & ^ with and without allocation option.
The above analysis reveals intricate relationships among
-A,H -1- A,H
operational and financial hedges, capacity, and currency
Proposition 4(b)*
exchange and demand risks. At a given capacity, the cur
rency exchange rate risk is reduced by using either the
operational or the financial hedge; the expected profit can
*
Represents the case of zero risk premium. be increased only by the operational hedge. However, the
reduction of the currency exchange risk allows the firm to
be more aggressive on capacity investment and allows the
For a given capacity, the allocation option allows the
firm to take on more of the demand risk, and thus increases
firm to avoid unfavorable currency exchange-rate realiza
the firm's expected profit. As a result, the firm that uses
tions and to capitalize on the positive part of the unit profit,
both hedging instruments is better off in expected profit
r2(s). This leads to a twofold benefit: the increase of the and MV objective than the firm that uses none or only
unit expected operations profit and the decrease of the unit
one instrument, and achieves closer-to-maximum-expected
profit variance. The benefits of such value enhancement and
profit performance with a lower risk than a risk-neutral
risk reduction entice the firm to invest more capacity in the
firm. The study of Mello et al. (1995) on switching options
first stage than without allocation option and consequently
in a dynamic setting offers similar insights on the inter
bring a higher expected profit. However, a larger capacity action of the operational hedge (flexibility) and financial
also leads to a higher profit variance. Thus, the improved can
hedge. That is, the increased operational flexibility
MV objective might be at the cost of higher profit variance increase the firm's first-best value, and an efficient financial
(the numerical example in ?4.3 will illustrate such a case). hedge increases the firm's value, but only by creating an
Corollary 5 and Proposition 5 summarize the effect of incentive to choose the optimal operational policy.
the financial hedge.
4.3. Numerical Examples
Corollary 5. For a domestic production facility sup
a In this subsection, we will first give an example to illustrate
plying foreign market without allocation option, given
capacity X, if the foreign market demand is independent the results of Propositions 4 and 5. Then, we will study
the rate and zero risk is the effects of the currency exchange-rate demand
of currency exchange premium volatility,
assumed, then the optimal policy among all call and put volatility, and MV ratio on the firm's performance. The
options is a single forward contract. following example (Example 1) data will be extensively
used in this section.
Example 1. Assume that the exchange rate s follows a
ing a foreign market with the demand and exchange rate
lognormal distribution with Ins ^ N{l,l). The foreign
being independent, if the risk premium of call options and market demand is independent of the exchange rate and
contracts are then
forward nonnegative,
follows a normal distribution of d2 ~ Af(100,30). Let
(a) With an allocation option, the use of the call
ceyT = 2, p2 = 1, t2 = 1.5, and A = 0.0002. Assume zero
option with exercise price r2/p2 increases the firm's opti risk premium.
mal capacity, expected profit, and MV objective. That is,
** H, Table 1 verifies the results of Propositions 4 and 5. We
XI,-u < ?a,-h ^ E*Ali, and I/?_H ^ [/?H;
Without an allocation option, if E[r2(s)] > 0, then can see that both the allocation option and the financial
(b)
the use of the forward contract increases the firm's opti hedge increase the optimal capacity, the expected profit,
mal capacity, expected profit, and MV objective. That is, and the MV objective. In this particular example, the allo
cation option also increases the profit variance while the
^ *-A,H> ^ ?-A,H> and ^ ^-A, H'
**A,-H E-A,-U U~A,-U
financial hedge decreases the profit variance significantly.
The main impact of a financial hedge is to introduce As we discussed in ?4.2, the allocation option provides the
a financial transaction that counterbalances the cash flow benefits of value enhancement and risk reduction, and the
Table 1. An example of the effect of allocation option and financial hedge.

Without allocation With allocation
-
option (-A) option (A) ((A, ) (-A, -))/(-A,.)
Without financial hedge (?H)
(X*, U*)(58.203,31.323) (65.039,42.277) (11.75%, 34.97%)
(?*, AV*)(53.818,22.496) (68.957,26.679) (28.13%, 18.59%)
With financial hedge (H)
(X*, U*)(84.961,64.457) (87.695,77.71) (3.22%, 20.56%)
(?*, AV*) (65.758,1.301) (79.209,1.499) (20.46%, 15.22%)
- -
((A,H)-(-A,-H))/(-A,-H)
((-, H) (-, -H))/(., -H)
(X*, U*) (45.97%, 105.78%) (34.83%, 83.81%) (50.67%, 148.09%)
(?*, AV*) (22.19%, -94.22%) (14.87%, -94.38%) (47.18%, -93.34%)
financial hedge provides the benefit of risk reduction. When without financial hedge, while the insensitivity of optimal
only one of the two instruments is considered by the firm, capacity leads to a slight increase of the profit variance for
the relative effectiveness of the two instruments depends on firms using financial hedge. Thus, the MV objectives of
the relative magnitudes of the value enhancement and the firms using financial hedge decrease slower than firms not
risk reduction which, in turn, are determined by the mag using financial hedge as shown in Figure 2(d). For a firm
nitude of the localization cost and the degree of the firm's that uses both allocation and financial options, the advan
risk aversion. However, the joint use of both instruments tage of the increased expected profit outweighs the disad
improves the firms's MV objective significantly, compared vantage of the increased variance; the MV objective might
to the use of a single instrument. This observation is consis increase with an increasing exchange-rate volatility.
tent with the empirical findings of Allayannis et al. (2001), To study the effect of demand volatility, we let the stan
use different measures of dard deviation of the demand vary from 30 to 50, i.e., the
who, however, operational hedg
coefficient of variation changes from 30% to 50%. Fig
ing strategies.
We then study how changes in the currency exchange ure 3(a) shows that the optimal capacities decrease as the
rate volatility affect the optimal capacities and the firm's volatility of the foreign demand increases. Interestingly,
firms using financial hedge are slightly less sensitive to the
corresponding performance for the four cases. When adjust
the currency we let the stan change in demand volatility. This is because a financial
ing exchange-rate volatility,
dard deviation of In s vary from 1 to 2 while keeping the hedge reduces the risk caused by the currency exchange
of s at a constant level of eL5 by adjusting rate, i.e., the risk associated with the firm's unit profit.
expectation
the mean of Ins accordingly, where e is the base of the Therefore, the demand volatility will have less of an effect
on the firm's total profit risk. Figure 3(b) shows that the
natural logarithm. Figures 2(a)-(d) illustrate the effect of
expected profits decrease due to the decrease of optimal
an of the currency rate. Fig
increasing volatility exchange
ure 2(a) shows that the optimal decrease as the capacities. Similar to the observation in Figure 2(c), the
capacities
rate becomes more volatile. For firms
insensitivity of optimal capacities to the demand volatility
currency exchange for firms using financial hedge leads to a slight increase
without financial hedge, capacity is the only instrument
of the total profit variance. The net effect, as Figure 3(d)
for variance control, and the optimal capacities decrease
shows, is that the MV objectives decrease as demand
as the exchange-rate increases. Firms with
volatility increases. Firms using financial hedge see a faster
quickly volatility
the optimal financial hedge in place, on the other hand, are decrease in theMV objective than firms not using financial
affected much less by the exchange-rate volatility because
hedge.
the risk associated with the exchange rate is reduced signif
Figure 4 shows the effect of the MV ratio. For this
icantly through the financial hedge. Figure 2(b) shows that particular set of parameters, firms using financial hedge
the expected profit decreases in the exchange-rate volatil reduce the profit variance more significantly than firms not
ity except for the firm using both the allocation option and using financial hedge. Thus, Figure 4(a) shows that as firms
financial hedge. This is because the allocation option is become risk averse, firms not using financial hedge have
equivalent to a call option (on the exchange rate) whose to reduce capacity faster than firms using financial hedge.
value increases as the increases, Reflected on the corresponding
exchange-rate volatility expected profit, variance,
and with financial hedge, the optimal capacity is not sen and MV objective, as shown in Figures 4(b), (c), and (d),
sitive to the increase of the exchange-rate volatility. The respectively, financial hedge makes firms less sensitive to
increase of the expected unit profit can have a more dom the degree of their risk aversion.
inating effect than the slight decrease of optimal capacity As we have shown in Corollary 4, the optimal capac
and thus leads to an increasing expected profit. Figure 2(c) ity for case (A, H) of the risk-averse firm is less than that
shows that the relatively fast decrease of the optimal capac of the risk-neutral firm. This result holds for the other
ity leads to a fast decrease of profit variance for firms three cases as well and can be observed in Figure 4(a)
Figure 2. Effect of the currency exchange-rate volatility on (a) optimal capacity, (b) expected profit, (c) profit variance,
and (d) MV objective.
(a) 100 p (b) E*

160
r_
""" -"
-
120 ^---"-""""
*?
teo--
|100-^.^_
340- .-. ^
I"'",
* ? ** *. 40 -
*
-^ *
20- ^* - ^. * * *
*. ^.
*" - - - -^ *
-1 ni 2?
**--*-. /?""**" " " "
nl-1-1-1-1_t 0'_'_'_'_ "_'
1.4 1.2
1.0 1.6 1.8 1.0 1.21.4 1.6
1.8
Exchange rate standard deviation Exchange rate standard deviation
(C) . ^(d) ? ?
** " I (A,H)
3?f _ 12?f_ (-A,H)|
25 \ 100
^ I-?(A.-H) ?-(-A,-H)| __
^
20 - ^ I 80- ..-'-^
| ^
%
M*N> aS ? ? _ _ _
-
*> 15** =? 60 - -? ? ? _
? ^. o ??
?* io*- 40" ^"
% -^
^~~~*^
ol_i-1-1_I_i o I_i_i_i""""i---_I
1.61.01.2
1.4 1.8 1.0 1.2 1.4
1.61.8
Figure 3. Effect of the demand standard deviation on (a) optimal capacity, (b) expected profit, (c) profit variance, and
(d) MV objective.
x 10?r
(a)
100 r
- ?
80
- ""
70 .
| * I .-_?..'-" ------
? . _1,40-
- . . 20
50
40I-1-1-1-1-1
0I-?-'-'-'-'
4230 3438 46 30 42 4634
38
Demand standard deviation Demand standard deviation
(o (? u*
30f 100 r- r_,
-(AH)-(_AH)
- ^* *
25 ^ |??
? (A,-H)-(-Af-H)
* *" *^*"
v - - - ?*?, ""
"^ ^
- ? ?
? 15 "----*. ^ __
|
1 5 40--_
?
? ? --
io- ? * ^
2
".--?--. Cu
- "
5 20
I_I_I_-_I_I_i 0 0 I_I_I_i_i_i
4230 3438 46 30 3842 4634
Demand standard deviation Demand standard deviation
Figure 4. Effect of the MV ratio on (a) optimal capacity, (b) expected profit, (c) profit variance, and (d) MV objective.
X* (b) E*
(a) 1W
100 r
- - 80-??
so
^=lv
^
? 60 - ** / -. ?
60" *-
. ^
a-- -.j^. ...

20 - ^ |4o-20 -
0I-1-1-1-1-1 0I-1-1-1-1-1
0 0.0001 0.0002 0.0003 0.0004 0 0.0001 0.0002 0.0003 0.0004
Mean-variance ratio Mean-variance ratio
VW w
300 r ioo r- -(A,H)-(-A.H)
~ 250.
|-(A,-H)-(-A.-H)
80 _
g200- >. ? +-
s *;>, 60" -*._^
2 150- I X % ^
*.^ ?
50- 20~ .
%%**^;-
-. . . '
ol o'-'-'-'-L~-'
0 0.0001 0.0002 0.0003 0.0004 0 0.0001 0.0002 0.0003 0.0004
Mean-variance ratio Mean-variance ratio
by comparing the values of X* at any A > 0 with those The allocation to the foreign market starts when s ^ r2/p2
at A = 0. In particular, the optimal capacity associated (i.e., r2{s) ^ 0), and the allocation priority is given to
with risk-neutral firms that use the allocation option, i.e., the foreign market when s ^ T2/p2 + {px ? Tx)/p2 (i.e.,
cases (A, ) at A = 0, maximizes the firm's expected profit r2{s) > rx).
and can be used as a benchmark. Taking any A > 0, we The counterpart of Proposition 3, i.e., the unimodality
start from case (?A, ? H), which has the smallest optimal of the objective function in X, does not hold anymore. We
capacity, expected profit, and MV objective among the four will use h* to represent the portfolio of the two call options
cases. Adding the allocation option, case (A, ?H) brings
specified in Corollary 2.
the optimal capacity and expected profit up closer to those
Proposition 6. For the independent case, if zero risk pre
of the benchmark. Then, adding the financial hedge, case
mium is assumed, then
(A, H) brings the optimal capacity and expected profit fur
ther up to be very close to those of the benchmark. A sim (1) E[7Top(X, h, s, d)] increases concavely in X.
ilar pattern can be observed for the sequence (?A, ? H), (2)
miX-ddl^^^] < E[(X-</,)+]Pr(</1 +
d2 > X) for i= 1,2, then \I[tt{X, h*, s, d)] increases in X.
(?A, H), and then (A, H). Thus, the allocation option and
financial hedge are instruments through which the risk
Proposition 6 states that, although the firm's operations
averse firm can achieve the close-to-optimal expected profit in capacity, the profit
profit always increases concavely
at lower risk. variance is not always monotone in capacity. An extra unit
of capacity increases the expected sales in both markets
5. Domestic Production Facility and in general increases the profit variance in both mar
Supplying Both Foreign and kets. However, the covariance between the sales of the two
Domestic Markets markets tends to be reduced by the extra unit of capac
ity for the following reason. Consider a scenario of given
5.1. Basic Results exchange rate.Without loss of generality, assume that mar
ket 1 is the first-priority market (i.e., rx^ r2{s)). An extra
Recall that for the two-market case, the firm's operations
in unit of capacity (a) can satisfy one more unit of demand
profit stage 2 is defined by
in the first-priority market when the first-priority market
= - - demand was not completely satisfied yet, i.e., dx > X, or
7Top(X, s, d) r+(s) min((X dx)+, d2) + (r2(s) rx)+
. - - (b) can satisfy one more unit of demand in the second
(min(X, d2) min((Z dx)+, d2))
priority market when the first-priority market demand was
+ rxmin(X,dx). (12) completely satisfied already, i.e., dx^X and dx + d2 > X,
or (c) is useless if demand in both markets was completely In the two-market case, as the monotonicity of the vari
satisfied already, i.e., dx + d2 ^ X. The covariance between ance in capacity no longer holds, the comparison of optimal
sales in the two markets becomes more negative in case (b), capacities between the risk-averse and the risk-neutral firms
but is not affected in cases (a) and (c) by the extra unit of becomes complicated. In general, the straight dominance of
capacity. Whether the total profit variance will decrease in the risk-neutral capacity over the risk-averse capacity in the
capacity is determined by whether the negative part in the one-market case is no longer true in the two-market case, as
covariance of the two market sales is significant enough. we shall see in numerical
examples in ?5.2. For results on
Proposition 6(2) gives the sufficient condition of the profit the optimal risk-neutral capacity for the two-market case,
variance increasing in capacity, which roughly represents we refer interested readers to Ding et al. (2004) (our pre
the condition under which the effect of case (b) is out vious working paper) and Kazaz et al. (2005).
weighed by the effects of cases (a) and (c). When this con
dition fails, the profit variance may decrease in capacity. 5.2. of Operational and
Analysis
The following example shows that as a result of the non Financial Hedges
monotonicity of profit variance in capacity, the firm's MV
Recall that the main findings from the analysis of oper
objective might be multimodal in capacity.
ational and financial hedges in the one-market case are:
Example 2. Assume that ceyT =
3, rx
=
r2
=
0, px
=
(1) The concerns about risks lead risk-averse firms to invest
~
10, p2 = 1, Ins iV(1.5,0.5), and the risk premium is less in capacity than risk-neutral firms. However, risk
zero, A = 0.005. The two-market demand follows a joint averse firms can use allocation option or financial hedge
~ - to increase capacity and expected profit. Using both instru
two-point distribution, (dx,d2) (1005,200(1 8)) and
= = = = 0.5. We can think of 8 as a loca ments will achieve the biggest increase of capacity and
Pr(5 0) Pr(S 1)
tion random variable that indicates a particular event will expected profit. (2) Using the financial hedge makes a risk
take place in either home or foreign country and will create averse firm's capacity investment less sensitive to volatili
demand in the corresponding market. Note that the total ties in exchange rate and foreign demand and less sensitive
demand of the two markets is between 100 and 200. Thus, to its degree of risk aversion.
when X < 100, the firm is always in case (a) regardless of Finding (1) is no longer true in the two-market case.
the exchange rate. As shown in Figure 5, the profit vari Specifically, it is because the addition of the domestic mar
ance first increases for X ^ 100. When X e [100,200], all ket introduces the covariance between the sales in two
three cases could happen, and the profit variance decreases markets. Consider the case without allocation option. An
for X g [100,150) then increases again for X increase of capacity for one market could increase or
[150,200].
This is because the effect of case (b) first dominates but decrease the covariance between the sales in two mar
is then dominated by the effects of cases (a) and (c) kets (i.e., cov(min(X1, dx),min{X2, d2))), depending on
of case the demand correlation between the two markets. The rel
(also reflected by the first-increase-then-decrease
ative sizes of the two markets also play a role in deter
(b) probability in X, i.e., Pr(dx ^ X, dx + d2 > X)). As a
result, the profit variance ismultimodal in X and there exist mining whether this marginal impact on covariance has a
multiple local optimal capacity decisions. dominating effect on the change in total profit variance.
Thus, unlike the one-market case in which the profit vari
ance always increases in capacity, profit variance can be
Lemma 1. Assume that the risk premium is zero. If
nonmonotone in capacity. Consider the case with allocation
dV/dX ÔforXÔ, then X* < X*.
option, as we discussed in ?5.1, that the marginal effect of
an extra unit of capacity will have an even stronger and
Figure 5. An example of multiple local optimal capac more complicated effect on the covariance. Thus, a risk
ity decisions. averse firm might invest more than a risk-neutral firm in
seeking the benefit of a lower profit variance. Moreover,
500r
E
" """' * "- * adopting allocation does not necessarily lead to an increase
400-
+^ of the total capacity, even for a risk-neutral firm. The rea
3oo- y son is similar to that for the general risk-pooling cases.
*
-
200 y An allocation option allows the firm to address the com
_V
bined demand of two markets by using a common capacity.
The combined demand can be much more skewed than the
0?"5-'-'-'-'-'*
10 * 50 90 130
170 individual market demand (Yang and Schrage 2006), which
- * *
-100
can lead to a higher or lower total capacity investment.
-200" For a risk-averse firm, the impact of the allocation option
-XV
*\ *
- on profit variance further complicates the impact on capac
-300 ^ ?--""*-%
-400L
* ity investment. Example 3 shows, in the two-market case,
X
Capacity how relationships between no-allocation and allocation, and
Figure 6. Effect of the allocation option in the two-market case in Example 3. (a) Capacity comparison and (b) variance
comparison between (A, ?H) and (?A, ?H). (c) Capacity comparison and (d) variance comparison between
(A,H) and (-A,H).
(a) X*
2001 (b) v*
-
250
15?-
m\ 200
-
15?
1loo- -1-
I /\ \
\ u \ S ico - S . \
^ *.:v
-: 50-s
-'-'
0J-1-1-1-1-1
o'-'-
0.25 0.15 0.35 0.45 0.55 0.15 0.25 0.35 0.45
0.55
W x* (d) v*_200
-(A, H) or (A, risk neutral)
[" y
-(A.-H) /
" -_^
- "*
150 ? (_A,H)
XX X"
g " I-(-A.-H)_|
.t? -S 90
y^
? 100-* V^
UI 60-S
-
_ 50 - ^^^
30
^^^^
0I-1-1-1-1-1
0I-1-1-1-1-1
0.25 0.15 0.35 0.45 0.55 0.15 0.25 0.35
0.45 0.55
between risk-averse and risk-neutral, can be different from markets, and the domestic market capacity might need to
those in the one-market case. be adjusted to control that effect on the total profit vari
= = = =
ance. Thus, for firms not having an allocation option, the
Example 3. Assume that ceyT 1, rx r2 0, px eL5,
p2
= 1, and the risk premium is zero. The joint demand capacity investment in the domestic market might increase
or decrease when financial hedge is used. Corollary 6
follows a joint uniform distribution: dx = 100 ? d2 and
~ Assume that In s ~ N{p,, a) and A = 0.02. and Proposition 7 are the counterparts to Corollary 5 and
d2 U[0,100].
We increase a from 0.15 to 0.6 and adjust /x so that E[s] Proposition 5.
is kept at a constant level of exs. Figure 6(a) compares
? ? Corollary 6. For a domestic production facility supply
capacities among risk averse (A, H), (?A, H), and risk
ing both foreign and domestic markets without the alloca
neutral with allocation. We can see that X* A _H ^ XA _H ^
tion option, given capacity X, if the demand and exchange
f?r ? G I0-15' ?'4]- FigUre 60) Sn0WSmat the
*A,risk neutral rate are independent and the risk premium is assumed zero,
allocation option could increase or decrease the total profit
then the optimal financial hedge among all call and put
variance. Similar observations can be made for (A, H) and
options is a single forward contract.
(-A,H) (Figures6(c) and (d)).
The impact of the financial hedge on capacity is quite
both foreign and domestic markets with the demand
interesting in the two-market case. Note that under the same ing
and exchange rate being independent and the risk premium
capacity, the financial hedge only reduces the foreign mar
ket profit variance and has no effect on the domestic mar assumed zero:
ket profit variance or the covariance between two-market (1) With the allocation option, the use of the opti
profits. For a firm that already uses the allocation option,
mal financial hedging strategy (specified in Corollary 2)
this benefit will motivate the firm to invest more in over increases the optimal capacity and theMV objective. That
all capacity. A firm that does not have the allocation option is,Xl_H<XlHandUl_HÛlH.
will be motivated to invest more dedicated capacity for the (2) Without the allocation option, the use of the opti
foreign market. However, as we discussed, the increase of mal forward contract increases the optimal capacity dedi
the dedicated foreign market capacity might have a non cated to the foreign market and theMV objective. That is,
monotonic effect on the covariance between sales in the two < *-A.H.2 and < ^-A. H> ***/*
*-A,-H,2 ^-A.-H X-A,-H,2
Operations Research 55(3), pp. 470-^89, ?2007 INFORMS 485
Figure 7. An example illustrating a firm's facility location decision can be affected by employing a financial hedge.
(a> 0>)
200- ? - 5001
->
" "
?160- ?300-
\
* T3
1
a 140- \ ?
s ^ 200
8
120-
\ \ &
10?
100- _-*-*-
80
-\-r-t-1-1-1-1-
0H-1-1-1-1-1-1
0.55 0.67 0.76 0.83 0.89 0.94 0.99 0.55 0.67 0.76 0.83 0.89 0.94 0.99
(c) (d) so-. I-1

-(1,-H)-(2,-H)
-?< ?
'--^?^-(1,H) (2,H) _.'
60-
^00* siQ,^****^
? no- >* 40"

^n*000^''' ^-?"^"""^ "T-V^
** ' "" -* ? 20- '^

|1TO'XJS><><' j a<*
^
- -
_ ?2 - a 0 -|-^-1-1-1-1-1
fc
H 0.55 X0.67 0.76 0.83 0.89 0.94 0.99
- - - ,,--*"
40" _ ? - X
-20
10-1-1-1-1-1-1-1
0.55 0.67 0.76 0.83 0.89 0.94 0.99 _40_
and X* A H 2 are the optimal capacities dedicated to the as the exchange rate standard deviation a changes from
market in (?A, ?H) and (A, ? H), respectively.
foreign 0.55 to 1 while keeping the mean exchange rate constant.
In particular, Figure 7(d) plots the MV utility difference
Extensive numerical studies of a similar nature were also
between the domestic location and the foreign location for
conducted for the two-market case (with one or two facili
the without-hedging and with-hedging cases. That is, a pos
ties), and finding (2) remains valid in those studies as well. itive (respectively, negative) value indicates that the firm
would prefer the domestic (respectively, foreign) location.
6. Effect of Financial Hedging on We can see that when a is less than 0.67, without finan
Location and Supply Chain Structure cial hedge the firm will choose the foreign location, but
We will now show through numerical examples that in with financial hedge the firm will choose the domestic loca
other operational settings, the profit variance reduction ef tion. The domestic location has the obvious advantage of
fect of the financial hedge can have an impact on a firm's being able to serve demands in both markets. However,
the fact that demands are perfectly correlated between two
strategic decisions such as the location and the number of
its production facilities. markets makes the domestic location riskier than the for
eign location, especially when the domestic facility has
Example 4. We consider a firm that has the choice of than
capacity higher 100. Thus, without financial hedge,
setting up one facility in either the domestic or the for the compound demand and exchange-rate risk makes the
eign market and the initial production and localization are domestic location less attractive than the foreign location.
conducted at the same facility. We assume that px = 10, As exchange-rate volatility increases, the domestic capacity
= 1, = = 1005 with = = = =
p2 dx d2 Pr(S 0) Pr(5 1) is reduced to and stays at 100, the benefit from the alloca
0.5, A = 0.0002. The rate s follows a
exchange lognor tion option outweighs exposure to the exchange-rate risk,
mal distribution with Ins ~ N(p, a) with p = 2 and a ? and thus the domestic location is preferred. With financial
0.5. For the domestic we have = =
facility, cxeyT 2.5, rxx hedge, the exchange-rate risk is significantly reduced, the
= and for the = =
t12 0, foreign facility, c2eyT 1, r2X oo, firm always prefers the domestic location, and it can afford
= 0. the allocation option, we use (i, ?H)
r22 Assuming to invest full capacity of 200 for a large range of exchange
and (i, H), i = 1,2, to denote the cases where the pro rate volatility.
duction facility is located in market /, without and with
financial hedge, respectively. Figure 7 provides compar Example 5. We consider a firm that can set up two pro
isons on capacity, expected profit, variance, and MV utility duction facilities, one in the domestic market and the other
Figure 8. An example illustrating a firm's decision on with financial hedge, the firm uses both facilities when a e
the number of production facilities can be (0.85,1.11), and uses only the foreign facility for a outside
affected by employing a financial hedge. that range. In this example, the use of financial hedge
affects the number of production facilities employed by the
<a> I i-1
100 ht ?.
^ ?(Xi, -H)-(X{, H) global firm to meet demand. To understand the changes in
* * the firm's capacity decision, we start from the case without
x \L..(X2,-H)-...(X2,H)
i-1
hedging. Because the demand in two markets is perfectly
so \ negatively correlated (with a total demand of 100) and
only the foreign facility can serve both markets, the firm
\ * prefers foreign capacity. For small exchange-rate volatil
60 - firm invests 100 at the
& \ ity, the capacity foreign facility and
a V * zero at the domestic facility. As the exchange rate becomes
* v *
more volatile, the firm starts to decrease the foreign capac
ity and increase domestic capacity. When the exchange rate
standard deviation exceeds 0.85, the need for controlling
20 ^ the total profit variance makes the firm also start to cut
the capacity at the domestic facility and to stop using the
/ a c domestic capacity when a exceeds 1.01. The use of finan
' ^-i cial hedge allows the firm to keep the foreign capacity at
pi /--4-lv-1-^
0.05 0.61 0.80 0.92 1.01 1.08 1.14 full 100 for a larger range of exchange-rate volatility and
Exchange rate standard deviation use the domestic facility for a smaller range of exchange
rate volatility. Finally, the solutions with the financial hedge
r
(b) 300
dominate across the whole range of exchange rates, both in
|?? -h.TT]
terms of profit and MV utility (and have decreased profit
250 r variance for all exchange rates above 0.85; see Figure 8(b)).
/"\
200
[\ ; 7. Summary
I V Global firms selling to multiple countries face demand
uncertainties in different countries and currency exchange
% 150 L / >.
'"i-.^ rate uncertainties. From the operations perspective, a firm
I 100L ? \
could exploit the capacity allocation option by delaying the
commitment of capacity to specific markets until demand
\^
and rate uncertainties are realized. From the
exchange
finance perspective, a firm could use financial instruments
50L
such as option contracts in the currency exchange market to
hedge the exchange rate uncertainty. Often, the allocation
I_I_I_I_I_I_
0I
0.05 0.61 0.80 0.92 1.01 1.08 1.14 option is viewed as an instrument for profit improvement
rate standard deviation and the financial hedge is viewed as a tool for profit vari
Exchange
ance control, while in fact both have an impact on the firm's
optimal capacity decision, expected profit, and profit vari
in the foreign market. We assume that px=%, = 1, ance. In this paper, we studied the impact of the simultane
p2
~ - = 0) = Pr(5 = 1)= ous use of these two instruments in a stylized MV model
(dx,d2) (100S, 100(1 5)), Pr(S
0.5, and A = 0.0002. exchange rate s follows a logThe in which a firm sells to markets in their home country as
normal distribution with Ins ~ N(p, a) with p = 1.5 and well as a foreign country through a production facility in
a = 0.4. For the domestic facility, we have cxeyT =4.5, the home country.
= = the optimal finan
tl1 =0, t12 oo, and for the foreign facility, c2eyT Our analysis started from developing
= =
0.5, r21 r22 0. Assuming the allocation option,
we cial hedging policy for the operations profit of general func
compare the without-financial-hedge case with the with tional form. When a portfolio of call and put options is
financial-hedge case. Figure 8 compares capacity and profit considered, we derived the financial hedging quantities for
variance for the two cases as the exchange rate standard given exercise prices. This result illustrates the tight link
deviation a changes from 0.05 to 1.16, while keeping the age between the firm's operational strategy and financial
mean exchange rate constant. Figure 8(a) shows that with strategy. To study the optimal joint capacity and financial
out financial hedge, the firm uses both the domestic and hedge decisions, we focus on a simpler case of the home
foreign facilities for a e (0.05,1.01), but uses only the for country production facility supplying the foreign market,
eign facility for cr outside that range. On the other hand, in which case the optimal capacity can be obtained from
the first-order condition. We then decomposed the effect of But this is exactly what we found in our numerical experi
the joint use of the allocation option and financial hedge. ments with two-facility-two-market models in the presence
As expected, given capacity, the allocation option enables of demand and exchange-rate uncertainty (see ?6).
the firm to avoid unprofitable exchange-rate scenarios and Our research makes an important first step in closing an
thus improves expected profit and reduces profit variance; apparent gap in the international operation and finance liter
financial hedge reduces the impact of exchange rates on its ature on quantifying the simultaneous setting of operational
profit variance. Hence, the firm will increase capacity and and financial hedging policy parameters and explaining the
improve the expected profit and MV objective by adopting nature of implications of such practices for capacity deci
either risk-hedging instrument. Naturally, the firm achieves sions of global firms. Further research that understands the
its best performance when both instruments are utilized. competitive implication in the presence of multiple firms
Although the risk-averse firm's capacity and expected profit serving global markets from the use, or lack of use, of
are always below the risk-neutral firm, the allocation option integrated risk-management approaches will be fruitful. In
and financial hedge help firms get control of the profit vari particular, a concrete understanding of the effectiveness of
ance and thus allows them to reach a capacity level that is operational and financial hedges in mitigating the risk of
closer to (however, not exceeding) the capacity level of a competitive exposure to exchange-rate fluctuation will be a
risk-neutral firm. Numerical studies showed that firms using valuable next step. We have started exploring some of these
financial hedge are less sensitive to demand and exchange issues in our ongoing research efforts.
rate volatilities and their own risk attitude than firms not
using financial hedge. 8. Electronic Companion
As the analysis extends to the case of the home-country
An electronic companion to this paper is available as part
production facility supplying both domestic and foreign
of the online version that can be found at http://or.journal.
markets, some from the one-market case remain
insights
informs.org/.
true, while others no longer hold. This is mainly because
the addition of the domestic market introduces covariance
between sales in the two markets, and the effect of the Endnotes
covariance on the profit variance is not necessarily mono 1. If H(h2) <(f- sL)Qe~yT, then one can borrow H(h2)
tone in capacity. The main differences in results are: (1) the from the bank at a risk-free rate of y, buy call option hi,
firm's MV objective can be multimodal in capacity, i.e., the and sell forward h\ at stage 1. The stage 2 payoff would
firm can have multiple local optimal capacities. This is a be (s ? sL)Q from the call option, ?H(h2)eyT for paying
direct result of the nonmonotonicity of covariance in capac off the loan, (f ? s)Q from the forward, and the net pay
ity. (2) A risk-averse firm might invest in more capacity off is (/ - sL)Q - H(h2)eyT > 0, arbitrage! If H(h2) >
than the risk-neutral firm. The possible negative covariance ? one can
(f sL)Qe~yT, then sell call option hi, deposit
between sales in the two markets may drive risk-averse in the bank, and buy forward hi at stage 1. The
H(h2)
firms to overinvest in capacity than a risk-neutral firm to ? ?
stage 2 payoff would be (s sL)Q from the call option,
seek a lower profit variance. (3) Adopting an allocation ?
H(h2)eyT from the bank, (s f)Q from the forward, and
option might decrease the firm's capacity. (4) The firm that the net payoff is H(h2)eyT + (sL - f)Q > 0, arbitrage!
does not use the allocation option will always increase the 2. One widely used measure of financial risk is value of
dedicated capacity for the foreign market after the adop risk (VAR), i.e., the quantile of the projected distribution
tion of the financial hedge but could decrease or increase of gains and losses by holding a portfolio over the tar
the dedicated capacity for the domestic market. Results that get horizon (Jorion 2001). The computation of VaR can be
remain valid in the two-market case are: (1) The firm that in operational settings where the distributions
challenging
uses the allocation will increase the overall capac of profits are often functions of truncated demand/price dis
option
ity after adopting financial hedge. (2) Firms using financial tributions. Caldentey and Haugh (2005) show that a more
hedge are less sensitive to volatilities and risk attitude than tractable mean and standard deviation (MStd) constraint
firms not using financial hedge. provides a lower bound to the VaR constraint, making it
In our paper, we also dealt with how the presence of mul more tractable to study the problem of maximizing a firm's
tiple production facilities and multiple markets can affect expected value subject to the downside risk constraint.
the nature of operational and financial hedging practices. 3. Wong and Yick (2004) show the optimal call-option size
While it is intuitively expected that the integrated use of of a given call option for a general utility function under
operational and financial hedges will have an effect on the the assumption of deterministic demand.
magnitude of capacity investments and production output, 4. Finite many call options can reduce the profit variance
it is less intuitive to anticipate that the use, or lack of use, significantly and thus constitute a more practical finan
of financial hedges can affect global supply chain structural cial hedging strategy. See Ding et al. (2004) (our previous
choices, such as the desired location and number of pro working paper, which is available from the authors upon
duction facilities to be employed to meet global demand. request).
488 Operations Research 55(3), pp. 470-489, ? 2007 INFORMS
5. Wong (2003 and references therein) uses "export flexi Dornier, P., R. Ernst, M. Fender, P. Kouvelis. 1998. Global Operations
to to this allocation option, and gives examples and Logistics: Text and Cases. Wiley, New York.
bility" refer
Eeckhoudt, L., C. Gollier, H. Schlessinger. 1995. The risk averse (and
of multinationals that may or may not have it.
prudent) newsboy. Management Sci. 41(5) 786-794.
Flood, E., Jr., D. R. Lessard. 1986. On the measurement of operational

exposure to exchange rates: A conceptual approach. Financial Man
Acknowledgments agement (Spring) 25-36.
The authors thank Vishal Gaur for insightful discussions. Gaur, V., S. Seshadri. 2005. Hedging inventory risk through market instru
ments. Manufacturing Service Oper. Management 7(2) 103-120.
Useful feedback was provided by participants of the Mini
Gurnani, H., C. S. Tang. 1999. Note: Optimal ordering decisions with
conference on Integrated Risk Management in Operations uncertain cost and demand forecast Sci.
updating. Management
and Global Supply Chain Management 2004, Boeing Cen 45(10) 1456-1462.
ter on Technology, Information, and Manufacturing, Wash Hodder, J. E. 1982. Exposure to exchange rate movements. J. Internat.
Econom. 13 375-386.
ington University, St. Louis, Missouri. Part of Qing Ding's to facility location
Hodder, J. E. 1984. Financial market approaches under
research was funded by LEE Foundation Research Fel
uncertainty. Oper. Res. 32 1374-1380.
lowship 05-C207-LEE-003, Singapore Management Uni 1986. A multifactor for international
Hodder, J. E., M. C. Dincer. model
versity. plant location and financing under uncertainty. Comput. Oper. Res.
13(5) 601-609.
Huchzermeier, A., M. A. Cohen. 1996. Valuing operational flexibility
under exchange rate uncertainty. Oper. Res. 44(1) 100-113.
References
Jensen, M. C, W. H. Meckling. 1976. Theory of the firm: Managerial
Agrawal, V., S. Seshadri. 2000. Impact of uncertainty and risk aversion on
behavior, agency costs and ownership structure. J. Financial Econom.
price and order quantity in the newsvendor problem. Manufacturing 3 305-360.
Service Oper. Management 2(4) 410-423.
Jorion, P. 2001. Value at Risk: The New Benchmark for Managing Finan
Allayannis, G., J. Ihrig, J. R Weston. 2001. Exchange-rate hedging: Finan cial Risk, 2nd ed. McGraw-Hill, New York.
cial vs. operational strategies. Amer. Econom. Rev. Papers Proc. 91(2)
Jucker, J. V., R. C. Carlson. 1976. The simple plant-location problem
391-395.
under uncertainty. Oper. Res. 24 1045-1055.
Anvari, M. 1987. Optimality criteria and risk in inventory models: The
Kazaz, B., M. Dada, H. Moskowitz. 2005. Global production planning
case of the newsboy problem. J. Oper. Res. Soc. 38 625-632.
under exchange-rate uncertainty. Management Sci. 51(7) 1101-1119.
Biger, N., J. Hull. 1983. The valuation of currency options. Financial
Kogut, B., N. Kulatilaka. 1994. Operating flexibility, global manufacturing
Management 12(1) 24-28. and the option value of a multinational network. Management Sci.
F., M. Scholes. 1973. The pricing of options and corporate
Black, liabili
40(1) 123-139.
ties. J. Political Econom. 81 637-659. rate uncer
Kouvelis, P. 1999. Global sourcing strategies under exchange
J. N.,
Jr., G. R. Courtadon. 1986. Efficiency tests of the foreign
Bodurtha, tainty. S. Tayur, R. Ganeshan, M. Magazine, eds. Quantitative Mod
currency options market. J. Finance 41(1) 151-162. els for Supply Chain Management. Kluwer Academic Publishers,
1992. Inventory control with an exponential 625-668.
Bouakiz, M., M. J. Sobel.
utility criterion. Oper. Res. 40 603-608. Kouvelis, P., G. Gutierrez. 1997. The newsvendor problem in a global
market: Optimal centralized and decentralized control policies for
Boyabath, O., L. B. Toktay. 2006. Operational hedging: A review with
a two-market stochastic inventory system. Management Sci. 43(5)
discussion. Working paper, INSEAD, Fontainebleu, France.
571-585.
Caldentey, R., M. Haugh. 2005. The martingale approach to operational
Lessard, D. R., J. B. Lightstone. 1986. Volatile exchange rates can put
and financial hedging. Research presentation, Washington University,
operations at risk. Harvard Bus. Rev. 64(4) 107-114.
St. Louis, MO.
Li, C.-L., P. Kouvelis. 1999. Flexible and risk-sharing supply contracts.
Caldentey, R., M. Haugh. 2006. Optimal control and hedging of operations Sci. 45(10) 1378-1398.
in the presence of financial markets. Math. Oper. Res. 31(2) 285-304. Management
Lowe, T. J., E. J. McGavin, L. B. Schwarz. 1988. The determination of
Chen, X., M. Sim, D. Simchi-Levi, R Sun. 2004. Risk aversion in inven
optimal base-stock inventory policy when the costs of under- and
tory management. Oper. Res. Forthcoming. are uncertain. Naval Res. Logist. 35(4) 539-554.
over-stocking
Chowdhry, B., J. T. B. Howe. 1999. Corporate risk management for multi
Markus, A. J. 1982. Risk sharing and theory of the firm. Bell J. Econom.
national corporations: Financial and operational hedging policies. 13 369-378.
Eur. Finance Rev. 2 229-246.
Mello, A. S., J. E. Parsons, A. J. Triantis. 1995. An integrated model of
Chung, K. 1990. Risk in inventory models: The case of the newsboy multinational and financial J. Internat. Econom.
flexibility hedging.
problem, optimality conditions. J. Oper. Res. Soc. 41 173-176. 39 27-51.
Cohen, M. A., A. Huchzermeier. 1999. Global supply chain manage Merton, R. C. 1973. Theory of rational option pricing. Bell J. Econom.
ment: A survey of research and applications. S. Tayur, R. Ganeshan, Sci. 4 141-183.
Management
M. Magazine, eds. Quantitative Models for Supply Chain Manage O'Brien, T. J. 1996. Global Financial Management. Wiley, New York.
ment. Kluwer Academic Publishers, Norwell, MA, 670-702.
Philippatos, G. C, N. Gressis. 1975. Conditions of equivalence among
Cohen, M. A., S. Mallik. 1997. Global supply chains: Research and appli
E-V, SSD and E-H portfolio selection criteria: The case for uniform,
cations. Production Oper. Management 6(2) 193-210. normal and lognormal distributions. Management Sci. 21 617-625.
Cornell, B., M. Reinganum. 1981. Forward and future prices: Evidence Seifert, R. W., U. W. Thonemann, W. H. Hausman. 2004. Optimal pro
from the foreign exchange markets. J. Finance 36(12) 1035-1045. curement strategies for online spot markets. Eur. J. Oper. Res. 152(3)
Q., L. Dong, R Kouvelis. 2004. On the integration of production 781-799.

Ding,
and financial hedging decisions in global markets. Working paper, Shapiro, A. 1988. International Financial Management. Allyn and Bacon,
Washington University, Saint Louis, MO. Boston, MA.
Dong, L., H. Liu. 2007. Equilibrium forward contracts on nonstorable Shastri, K., K. Tandon. 1986. On the use of European models to price
commodities in the presence of market power. Oper. Res. 55(1) American options on foreign currency. J. Futures Markets 6(1)
128-145. 93-108.
Triantis, A. J., J. E. Hodder. 1990. Valuing flexibility as a complex option. Wong, K. P. 2003. Export flexibility and currency hedging. Internat.
J. Finance. 45(2) 549-565. Econom. Rev. 44(4) 1295-1312.
Van Mieghem, J. A. 1998. Investment strategies for flexible resource. Wong, K. P., H. Y. Yick. 2004. Currency options and export-flexible firms.
Management Sci. 44(8) 1071-1078. Bull. Econom. Res. 56(4) 379-394.

Van Mieghem, J. A. 1999. Coordinating investment, production, and sub Wu, D. J., P. R. Kleindorfer. 2005. Competitive options, supply contract
contracting. Management Sci. 45(7) 954-971. ing, and electronic markets. Management Sci. 51(3) 452-466.
Van Mieghem, J. A. 2004. Risk mitigation in newsvendor networks: Yang, H., L. Schrage. 2006. Conditions that cause risk pooling to increase
Resource diversification, flexibility, sharing, and hedging. Manage inventory. Working paper, University of Utah, Salt Lake City, UT.
ment Sci. Forthcoming. 2003. Dual sourcing from a regular supplier and a
Yi, J., A. Scheller-Wolf.
Van Mieghem, J. A., M. Dada. 1999. Price versus production postpone spot market. Working paper, Carnegie Mellon University, Pittsburgh,
ment: Capacity and competition. Management Sci. 45(12) 1631-1649. PA.

Informs: INFORMS Is Collaborating With JSTOR To Digitize, Preserve and Extend Access To Operations Research

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Informs: INFORMS Is Collaborating With JSTOR To Digitize, Preserve and Extend Access To Operations Research

Uploaded by

Copyright:

Available Formats

INFORMS

On the Integration of Production and Financial Hedging Decisions in Global Markets

On the Integration of Production and Financial

Lingxiu Dong, Panos Kouvelis

1. Introduction to use operational strategies as effective hedges against

ply chain. In ?4, we focus on a

S = T2/p2 and = correlated, the interrelatedness of the operational decision

Qi(X,SltS2) strategy will depend on the specific assumption on the cor

Moreover, for the independent case, X^ is the solution to

4.1. Basic Results 4.2. Analysis of Operational and

- - comparisons among the four cases is given in Figure 1.)

(a) By implementing an allocation option, a risk-averse

Table 1. An example of the effect of allocation option and financial hedge.

(a) 100 p (b) E*

Exchange rate standard deviation Exchange rate standard deviation

Exchange rate standard deviation Exchange rate standard deviation

a-- -.j^. ...

Exchange rate standard deviation Exchange rate standard deviation

Exchange rate standard deviation Exchange rate standard deviation

Exchange rate standard deviation Exchange rate standard deviation

(c) (d) so-. I-1

? no- >* 40"

** ' "" -* ? 20- '^

Flood, E., Jr., D. R. Lessard. 1986. On the measurement of operational

Q., L. Dong, R Kouvelis. 2004. On the integration of production 781-799.

Management Sci. 44(8) 1071-1078. Bull. Econom. Res. 56(4) 379-394.

You might also like