The Review of Economic Studies, Ltd.

The Theory of Hedging and Speculation in Commodity Futures

Author(s): Leland L. Johnson
Source: The Review of Economic Studies, Vol. 27, No. 3 (Jun., 1960), pp. 139-151
Published by: Oxford University Press
Stable URL: http://www.jstor.org/stable/2296076 .
Accessed: 26/08/2013 11:08

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 [email protected].

Oxford University Press and The Review of Economic Studies, Ltd. are collaborating with JSTOR to digitize,
preserve and extend access to The Review of Economic Studies.

http://www.jstor.org

This content downloaded from 206.212.0.156 on Mon, 26 Aug 2013 11:08:11 AM

All use subject to JSTOR Terms and Conditions
 TheTheoryofHedgingand
Futures
in Commodity
Speculation
Althoughsignificant contributions have appeared in the literaturein recentyears,
the present-day theoryof hedgingand speculationappears to account inadequatelyfor
certainmarketpractices.In particular,themotivation ofthetraderwho undertakes hedging
activities,therole thathedgingplaysin his over-allmarketoperations,and thedistinction
betweena traderwho hedgesand one who speculateshave givenriseto difficulties in the
literature. My purposeshereare (1) to outlinebrieflythe purposesand mechanicsof a
commodityfuturesmarket,(2) to discussand appraisethe theoryof hedgingand specu-
lationas it existstoday,(3) to presenta reformulated conceptof hedging,and (4) to con-
structa modelthat may both assistin clarifying theconceptsof hedgingand speculation
and contributeto a betterunderstanding of certainmarketphenomena.

THE NATUREOF FUTUREsTRADING

Organizedcommodityfutures twokindsofactivity-speculation
tradingfacilitates and
hedging. When futurestradingin a given commodityexists,the speculatorgenerally
findsit advantageousto deal in futurescontractsratherthaneither(1) buyinga quantity
of the commodityat the current" spot" priceand holdingit in the hope of a rise in the
spot price,(2) sellingthe commodityshortby promisingin privatenegotiationswith a
buyerto deliverat a specifiedlaterdate, at a pricewhichhe expectsto be above the spot
pricethatwill prevailat thatdate. Equally important, if not more so, futuresmarkets
are usefulforhedgingoperations. An essentialfeatureof commodityhedgingis thatthe
tradersynchronizes hisactivitiesin twomarkets. One is generallythe cash' or " spot" "

market(the marketfor immediatedelivery); the otheris generallythe futuresmarket.

A futurecontract,beingmerelya promiseof the sellerto deliverwithina specified
monthand a promiseofthebuyerto takedeliveryof a standardquantityand qualityof the
commodityat an agreedpriceis readilyadaptableby its homogeneouscharacterto being
traded on an exchange. The commodityexchangeprovidesa centrallocation where
potentialbuyersand sellersmakebids and offersforcontractscoveringdeliveryin various
later months. Each deliverymonthin which tradingtakes place is referredto as a
" future". Because of the centralizednatureof the marketand the rapidityand ease
withwhichsales and purchasescan be made,thepriceof any sale in a givenfutureduring
of supplyand demand conditionsexistingat
reflection
the tradingday is a near-perfect
thatinstantof timeforcontractsof thatparticularfuture.
In mostfutures markets,onlya smallfractionofcontractssold is closedout bydelivery
of the actual commodity. Because nearlyall participantsare motivatedby the desireto
trade on price movements,they liquidate by undertakingoffsetting transactions. The
buyer(seller)can liquidatehis positionin thefuturesmarketpr-iorto actualdeliveryof the
commodityby merelyselling(buying)on the exchangecontractsof the same future. The

1I am indebted
to RichardR. Nelsonand RichardN. Rosettformanyhelpfulcommentson earlier
ofthemanuscript.
drafts

139

This content downloaded from 206.212.0.156 on Mon, 26 Aug 2013 11:08:11 AM

All use subject to JSTOR Terms and Conditions
 140 REVIEW OF ECONOMIC STUDIES
secondtransactionoffsetsand cancelsthe previouslyexistingcommitment. For each
contractpurchasedor sold and subsequentlyliquidatedthetradertakesa totalprofit
orlossequaltothedifference
between hisbuyingandselling bythenumber
pricemultiplied
ofunitsofthecommodity in thecontract.
specified

THE THEORYOF HEDGING

Most commodity tradingtheorists have visualizedthe hedgeras a dealerin the
"actual " commodity who desires" insurance " againstthe priceriskshe faces. For
example, ifhe purchasesa unitofthecommodity at a givenspotpriceand thepricefalls
it,heis exposedto a capitalloss(gain),inaddition
(rises)priorto hisreselling to whatever
merchandising profit bytheamountofthatpricechange. According
he receives, to these
theorists he wouldtypicallyprotect hisinventoryposition ofx unitsfromtheriskofsuch
pricefluctuation bysimultaneously sellinga sufficient
number offuture contractsto cover
delivery ofx units;1 whenhe resellshisinventory he wouldsimultaneously liquidatehis
positionin futures bypurchasing the samenumberof contracts (of thesamefuture)as
before. Ifthenetchangein spotpriceis equalto thenetchangeinthepriceofhisfuture,
i.e., ifthepricemovements are parallelto each other,thegainhe enjoysin one market
offsets thelossin theotherand he wouldbe leftwithonlyhis" normal"merchandising
profit.2Otherwise he wouldbe leftwitha residualcapitalgainor loss.
assumea hedgecarriedin a future
As an illustration, fromtimet1to timet2 (where
thefuture specifies
deliveryat t3) againstx unitsofinventory purchased at t, and sold at
t2. Let S1 and S2 denotethespotpriceandF1andF2thepriceofthefuture thatexistat
The hedgerwilltakea totalgain(loss) arisingfrompricemove-
t, and t2 respectively.
mentsfromt, to t2 equal to thepositive(negative) valueof x[(S2 - S,) - (F2- F1)].
Thehedgeis perfectly effective
if [(S2 - S) - (F2 F1)] is equal to 0.
-

A majorroleofthespeculator in futures,
accordingto thebulkoftheliterature, is to
assumetherisksthathedgersdesireto transfer fromtheirownshoulders ;3 thefutures
marketis visualizedas a convenient mechanism through whichpriceriskcan be trans-
ferredfromone groupto another.The hedgeris oftendescribed as an apparently un-
sophisticated
participantin futuresdealingswho,in thewordsof Hawtrey, " regardsthe
makingof priceas a whole-time occupation forexperts [speculators?], and,in general,
willnotpithis fragmentary information againstthesystematic studyat thedisposalof
theprofessionaldealers."4 Expanding uponthisthesis, J. M. Keynes,in hisA Treatise
on Money,deducedthetheoryof " normalbackwardation" in whichhe assertedthat
hedgersare willingto pay a " riskpremium"to relievethemselves of pricerisk,while
are willingto enterthefutures
speculators marketonlyif theyhavetheexpectation of
a premium.Therefore,
collecting sincehedgersare predominately shortin futures and
speculatorsare predominately long,the current futureor " forward " pricemustfall
belowthefuture priceexpected to prevailat any latertime by the amount of thisrisk
premium.If the expectedfuturepriceis equal to thecurrent spot price,the current
futurepricemustfallbelowthecurrent spotpricebytheamountof thepremium.The

1 This abstractsfromtheproblemof indivisibility whicharisesfromthefactthata contractcovers

a fixedquantityof thecommodity.
2 In a case where stockhe couldalso carrya " perpatual
he carriesa permanent " hedgebyliquidating
hiscontracts and simultaneously
priorto delivery to maintainthehedge.
of a laterfuture
sellingcontracts
Thisprocedure, however, does notalterthebasicprinciplesof hedgingoutlinedabove.
3See, forexample,P. A. Samuelson,Economics, An IntroductoryAnalysis,
4thedition,(New York,
1958),p. 425.
* R. G. Hawtrey," Mr. Kaldor on theForwardMarket,"Reviewof EconomicStudies,June1940,
VII, p. 203.

This content downloaded from 206.212.0.156 on Mon, 26 Aug 2013 11:08:11 AM

All use subject to JSTOR Terms and Conditions
 HEDGING AND SPECULATION IN COMMODITY FUTURES 141

existenceof a discountof a futurebelowthecurrentspot priceby an amountequal to tne

premiumis a conditionof normalbackwardation.1
In recentyearsHolbrook Workinghas writtena seriesof articlesthat run counter
to theolder," traditional" conceptof thenatureof hedgingand thefunctionof a futures
market. le envisagesthe hedgeras one who does not seek primarilyto avoid riskbut
one who hedgesbecause of an expectedreturnarisingfromanticipationsof favourable
relativepricemovements in the spot and futuresmarkets. The traderdoes not somehow
findhimself witha givensize inventorythathas to be hedgedagainst,buthe takespositions
in bothmarketsas a formof " arbitrage".
The role of risk-avoidancein most commercialhedginghas been greatlyover-
emphasizedin economicdiscussions. Most hedgingis done largely,and maybe done
wholly,because information on whichthe merchantor processoracts leads logically
to hedging. He buys the spot commoditybecause the spot price is low relativeto
thefuturespriceand he has reasonto expectthespotpremiumto advance ; therefore
he buysspotand sellsthefuture.2
Sincethehedgeris notmotivatedprimarilybydesireto reduceriskitis also misleading,
of hedgingaccordingto the degreeto which
Workingasserts,to judge the effectiveness
futurespriceand spot price movementsare parallel.
of hedgingdependson parallelismof
. . .the basic idea that completeeffectiveness
movementof spot and futurespricesis false,and an improperstandardby whichto
testthe effectivenessof hedging. The effectiveness
of hedgingintelligently
used with
commoditystorage,depends on inequalitiesbetweenthe movementsof spot and
futuresprices and on reasonablepredictabilityof such inequalities.3

AN APPRAISAL

On the basis of personalinterviewswithrepresentativesof 20 firmsin one particular

commoditymarket-theNew York coffeetrade-I have concludedthatthe present-day
bodyof theoryoutlinedabove does notaccountadequatelyforcertainmarketphenomena.
On the basis of interviewswithseveralrepresentatives of brokeragefirmsdealing in a
wide rangeof commodities, thereis reasonto believethatthissituationextendsto trading
in othercommoditiesas well.
On one hand,it is truethathedgersin thecoffeemarketsometimesat leastact on the
basis of expectedrelativepricemovementsbetweenspot and futuresand adjust boththeir
stockholdingsand theirpositionsin futuressomewhatas describedby Working. The
priceof a givenfutureis almostalwaysat a largediscountrelativeto spotat thetimetrade
in it beginsa yearpriorto its deliveryperiod,and thisdiscounttendsto diminishas the
deliverymonth approaches. Accordingto our previous terminology, (S1 - F1) and
[(S2 - S1) - (F2 F1)] tend to be positiveand negativerespectively.Therefore,the
tradertends to take a loss, aside frommerchandising profit,on his hedged inventory

I A Treatise on Money (London, 1930), Vol. II, Chap. XXIX. This theoryhas been elaborated by
J. R. Hicks in Value and Capital, 2nd ed. (Oxford, 1946), pp. 137-139,and has been extensivelytreatedby
N. Kaldor in " Speculationand Income Stability,"Reviewof EconomicStudies,October 1939, VIl, pp. 1-27.
An interestingrecentanalysis of the financialresultsof speculation in commodityfuturesis contained in
H. S. Houthakker," Can Speculators Forecast Prices? " Review of Economics and Statistics,May 1957,
XXXIX, pp. 143-151.
2 " Futures Trading and Hedging," AmericanEconomic Review, June 1953, XLIII, p. 325. Italics
in the original.
3 " Hedging Reconsidered,"Journalof Farm Economics,November 1953, XXXV, pp. 547-549. Italics
in the original.

This content downloaded from 206.212.0.156 on Mon, 26 Aug 2013 11:08:11 AM

All use subject to JSTOR Terms and Conditions
 142 REVIEW OF ECONOMIC STUDIES

For thisreasonmostof thetraders contactedwhoclaimto hedgeat all statedthatthey

oftencarryverysmallinventories (" buyhand-to-mouth ") and therebymaintainonly
smallpositionsin futures.At thesametime,however, theyclaimto be motivatedto
hedgeprimarilyin orderto reducepricerisk. The importance oftheprice" insurance
"
factorin coffee
hedgingmoost itselfin thefactthatonegroup.
clearlymranifests oftraders,
who facelittlepriceriskin holdinginventory
the roasters, almostneverhedge,while
another theimporters,
group,, whodofacelargepricerisksmakeextensive useofthefutures
marketforhedging purposes.'
Furthermore, thetradersin thesurvey takecognizance
notonlyofexpected relative
pricemovements butofexpected absolute
pricemovementsas well. Generally,
iftraders
expectspotpricesto risetheytendto remove hedgesandincrease
theirinventory
holdings.
In somecasestheytakelongpositions in boththespotand futures markets
as a more
obvioussI)eculative
venture.On theotherhand,if theyare bearishtheyincreasetheir
shortfuturespositionsin excessof hedgingrequirements. In otherwords,hedging
getmixedin verycloselywithspeculative
activities operations
in theaccountsof thein-
dividualtrader.
In general,hedging activitiesappearto be motivated bythedesireto reducerisk,as
described intraditional
theory, butlevelsof inventory heldappearto be notindependent
ofexpected hedging as emphasized
profits, byWorking.Furthermore, thatan individual
mayholda mixofhedgedand speculative positionsin responseto hisexpectations con-
cerning absolute pricechangesis a practicenotwellexplainedin either traditional
theory
or in Working's theory.In theformer thetendency, by Keynes'" normal
as illustrated
backwardation," is muchmoreto speakofthehedgerand thespeculator as iftheywere
entirely separateindividuals withentirely different
motivations.2 In thelatter,thereis
no treatment of theconditions underwhichthetradermayspeculatein lieu of or in
combination withhedging. In fact,theverydistinction between hedging andspeculation
is fuzzy; whenthetrader takesmarket positions on thebasis of expectations concerning
relativepricechanges, he is speculating
insofar as he is notbetting on a " surething".
In viewof thedisparities
betweentheoryand thesemarketphenomena one could
wellask: " Precisely
whatis a hedge? Can a hedgebe meaningfully
definedin thecase
in whichthetraderacts on thebasisof absolutepriceexpectations?Can lhedging
be
as an activity
treatedtheoretically conductedsimultaneously
withspeculation?"

A REFORMULATION OF THE THEORY OF HEDGING

First,a reformulated definitionand analysisof hedgingin commodity is in

futures
order. Givena positionconsisting of a number, xi, of physical
unitsheldin marketi,
a " hedge" is defined
as a positioninmarket.jofsizex>*unitssuchthatthe" pricerisk"
ofholdingxi andx1*from timet,totimet2isminimized.Thescopeoftheterm" market "
in thisdefinition
is restricted to includetradingin a commodity of sufficiently
exact
so thatitspricemaybe considered
specification a scalarmagnitude.

1 For a detaileddiscussionofthispoint,see myarticle," PriceInstability,

Hedgingand TradeVolume
in theCoffeeMarket,"Journalof PoliticalEconomy, August1957,XLV, pp. 319-321.
' L. G. Telserin " SafetyFirstand Hedging,"Reviewof EconomicStudies,No. 60, 1955-56,XXIII,
pp. 1-16,demonstrates the rationalityof holdinghedgedand unhedgedpositionssimultaneously but
operatesunderthe highlyrestrictive assumption thatthe tradermaximizesexpectedincomeunderthe
contraintthattheprobability of theoccurrenceof a given" disaster
" level-
of incomenotexceeda given
value. M. J.Brennan in" SupplyofStorage,"American Economic Review,March1958,XLVIII, pp. 69-70,
mentions brieflythesimultaneous holdingofbothhedgedand unhedged stockson the basis of riskabove
a " critical" levelbeingtransferredto speculators
via hedging.

This content downloaded from 206.212.0.156 on Mon, 26 Aug 2013 11:08:11 AM

All use subject to JSTOR Terms and Conditions
 HEDGING AND SPECULATION IN COMMODITY FUTURES 143

"Price risk" can be considered of thevariance(or standard

a reflection deviation)
of a subjective distribution
probability (or a subjective density
probability for
function)
pricechangefromt,to t2 thatthetraderholdsat timetl,whereactualpricefromt1to t.
is treatedas a randomvariable. The varianceof pricechange,denotedby aj2 in thei
market is equaltothevarianceofreturnor " pricerisk" ofholdingoneunitinthei market
froml o
toO 2, sincethe(absolute)value of actualreturn to pricechangefrom
attributed
valueoftheactualpricechangeitself. The varianceof
tl to t2is equal to the(absolute)
returnor the priceriskof holdingxi unitsis equal to xf2 aj2. Likewise,the priceriskof
holdingone unitin thej market as thevariance,
can be considered denotedby o9, Ofa
distribution
probability
subjective ofpricechangeinthejmarket. Thevariance ofreturn
of holdingxgunits,whentheseunitsare considered alone,is equal to x12a2. Where
covq denotesthecovarianceofpricechange(or covariance ofreturn dueto pricechange)
betweenmarketi and market j, a of
conmbinationz in
positions i andj has a totalvariance
ofreturnV(R) due to pricechangegivenby:
(1) x 2 aj2 + xi2 a12 + 2xtx1 covqj
V(R) -

The combinationalso has an actual returnR and an expectedreturnE(R) due to price

changegivenrespectivelyby:
(2) R - xi B? + xj Bg
and
(3) E(R) = xi u + xj u1
whereBj, Bj denotestheactual pricechangesfromt1to t2 in i andj, and ui and u1denote
thepricechangesfromt, to t2 expectedat tl, As such,ut and u1arethemeanvaluesof
theprobability ofreturn
distributions inthei andj markets
existing at timetl.'
respectively
Differentiatingequation (1) withrespectto xj and settingthe derivativeequal to 0,
we have the value xj* minimizing the varianceof returnforthe combinationxi, x>*.
x
(4) =* - covi1

thevaluex1*forxj in equation(1) andletting

Substituting V(R)* denoteth-totalvariance
of returnof the combinationxi, x>*, we have:
V(R)* = xj2 q2 + 1Xi2Covt9 2Xi2 CoV112
( ai 2 a2

or V(R)* -Xi (ai ci2)

of correlation,
Since the coefficient by thetraderis equal to c-oy
p, estimated a1
then
at
V(R)* X- x2 at2 (1 - p2). Generallyspeakingthelargerthe(absolute)value of thecoeffi-

1 Equations1, 2, and 3 arebasedon thegeneralequationsinvolving theactualreturn,

expectedreturn,
and varianceof return ofa linearcombination of randomvariables. Givena number n markets
in which
actualreturnBi in each is a randomvariable,totalreturnR is a weighted sumof randomvariablessuch
n
thatR = E xi Bi whereeach weight is thesize ofpositionxi heldin each market. The expectedreturn

E(R) is givenbyE(R) == xi Ui andthevarianceofreturnV(R) is given-byV(R) = Xi 22 + 2E X

x X=1 1=1 >!
xs xs covf..

This content downloaded from 206.212.0.156 on Mon, 26 Aug 2013 11:08:11 AM

All use subject to JSTOR Terms and Conditions
144 REVIEW OF ECONOMIC STUDIES
thegreater
cientofcorrelation, thereduction inpriceriskofholding xi thatcanbe effected
bycarrying thehedgexg*. Ifthetrader believesattimet,thatpricemovements areperfectly
correlatedbetween t,and t2,p is equalto 1 andover-allpriceriskis reducedto 0. If he
believesthatthereis no correlationwhatever, V(R)*is equal to X,2 aj2 - thevarianceof
xi heldalone. The effectiveness e ofthehedgeis measured byconsidering thevariance
ofreturn, V(R)*associatedwiththecombination xi,x>*in a ratiowiththevariance xj2a,2
associatedwiththepositionxi heldaloneso that

e- ( 1- X2) ore p2.

Although thisreformulated conceptresembles thetraditional

theoryinsofaras the
hedgeis considered as a mechanism to reducepricerisk,theconceptof priceriskitself
is quitedifferent.In traditionaltheory priceriskis measured bythesize of actual gain
or loss due to pricechangethatthe traderincurs. The effectiveness of thehedgeis
measured byconsidering thegainor loss due to pricechangesincurred in an unhedged
positionrelativeto thatincurred in a hedgedposition.If pricemovements in i and j
fromt1to t2wereequalto eachother,anylossin onemarket wouldbe exactlyoffset by
thegainintheother;priceriskwouldtherefore be reducedto zero,andthehedgewould
be considered perfectlyeffective.However,in thereformulated framework bothprice
riskand theeffectiveness of thehedgeare treatedquiteapartfromtheeffects of actual
pricechange. Priceriskis something thatexistsin themindofthetraderat t1; namely
it is thevarianceof his probabilitydistributionof returndue to pricechange. Hence,
onlyinsubjective
itis treated terms-notinterms ofexpostpricechanges. Theeffective-
nessof thehedgeis likewise considered onlyin subjectiveterms-itis measured by the
extent to whichthetrader believesat t1thatthevariance ofreturnofholding xi is reduced
by simultaneously holdingx>*.
Parallelismof actualpricemovements wouldbe indicative of a perfectlyeffective
hedgeonlyifthetrader withcertainty
believed at t,thatanychangein onepricewouldbe
equalto thechangein theother; butthecondition ofactualparallelismitselfwouldbe
neither a necessary nora sufficient
condition to makepossiblea perfectly effectivehedge.
It is nota necessaryconditionbecauseifthetrader believedwithcertainty
thatanyrelation-
shipheldperfectly between pricemovements in i andj he couldtakea perfectly effective
hedge. If,forexample, he believedwithcertainty thatthepriceinthejmarketwouldfall
fromt1to t2byc centsrelative to thepriceinthei market, hecouldtakea longpositionxi
in combination withshortpositioninj of thesamenumberof units,and be certainof
achieving equalto c(xi),and no priceriskas suchwouldbe involved. In other
a return
words,neither theexpected returnnortheactualreturn ofthecombination needbe equal
to zeroto makepossiblea perfectly effectivehedge.
The reformulated theoryis similarto thetraditional theoryin thata " primary
market the
is postulated, primary marketvisualized hereas oneinwhichthetrader insome
sense" makeshisliving". More specifically, theprimary marketis definedas one in
whichthetrader(assumedto be normally is ableto obtaina netmerchandising
efficient)
positionsin thatmarket. As suchhe can mostgeneralybe considereda
profitbycarrying
middleman in thehandlingof a commodity.References to themiddleman roleplayed
by the hedgerare frequently foundin theliterature, althoughthe "primary"market
conditionis generallynotdefined Hereit is essential
explicitly. to postulate
theprimary
market within of hedging
thedefinition becausethemarket in whichtheriskof a given
position mustbespecified
is to be minimized inadvance. Onlyifthe(subjective) coefficieni
ofcorrelation wereequalto 1 wouldthexJ*thatminimizes theriskofholdingxi be ofa

This content downloaded from 206.212.0.156 on Mon, 26 Aug 2013 11:08:11 AM

All use subject to JSTOR Terms and Conditions
 HEDGING AND SPECULATION IN COMMODITY FUTURES 145

valuesuchthatthexi*whichwouldinreverse theriskofholding
minimize xj* be equalto
x0.

A MODEL OF HEDGING AND SPECULATION

conceptofhedging
Withthisgeneral in mindweshallconsider a modelwithin which
maybe examined.Givena framework
a widerangeof tradingactivities of particular
assumptionsaboutthetraderand thenatureof hisworld,it willbe possibleto demon-
strateunderwhatconditionsvariousmarketphenomena arise.

w W"

K "t

0 E(R)
Fig. I
As to thenatureof thetraderhimself, he is postulatedto havean indifference map
in Figure1. The totalexpected
illustrated netreturn E(R) generated byall ofhismarket
positionstogetherfromt,to_t2iS plotted on theX-axis. As a directmeasureofpricerisk,
thestandard deviationV/V(R)ofreturn is plottedontheY-axis. Curves1 and2 represent
curvesamongwhichthetrader
indifference hasrankedall possiblecombinations ofV(VR)
andE(R). Theshapeandpositions ofthecurvesindicate thattoremain ona givenindiffer-
enceleveltheincremental ratioof VV(R) to E(R) mustfallas he movesto a higher E(R).
Witha givenlevelof E(R) he movesto a higher(lower)indifference curveas VIV(R)
decreases(increases).His optimum over-all market positionis defined as thatpositionor
combination ofpositionswhichgenerates an E(R) and VfV(R)suchthatheattainsthehigh-
estindifferencecurvethathe is able to attainundergivenconstraints.
I shallassumefurther thatthetradercan engagein one or moreof thefollowing
activities
(1) He can takea longpositionin thespotor i market at t, bypurchasing a stockof
thecommodity at thespotpriceprevailing at t1 and resellingit at t2 at thespotprice

1
The valueof x>* is XSi2 By substitution
in equation4 thevalue of xi, denotedherebyxi*

theriskof holdingxi* is givenby (ci

thatminimizes ). xi* is equal to xi onlyif !'Z is equal
to 1.

This content downloaded from 206.212.0.156 on Mon, 26 Aug 2013 11:08:11 AM

All use subject to JSTOR Terms and Conditions
146 REVIEW OF ECONOMIC STUDIES

prevailingat t2 ; he receivesforthis operation,in addition,a net merchandising profit

(m) equal to a fixedmarkupper unitheld fromt1to t2. This profitis simplya netreturn
for servicesrenderedas a middleman. The institutional environment is assumed to be
suchthatthesize of thisreturnis knownwithcertainty. The totalriskfacedbythetrader
is confinedto " pricerisk". In thiscase equation (3) mustbe modifiedto

(3a) E(R) = xi(ui + in) + Xj uj wherexi > 0.

regardedas the primarymarket.

The spot marketis, therefore,

(2) He can take a long positionin a givenfutureof thej marketat t, by purchasing

contractsof a givenfutureat thepriceprevailingat t, and liquidatethemat t2at theprice
prevailingat t2.

(3) He can take a shortpositionin the givenfutureby sellingcontractsat t1 and

liquidatingthemat t2,also at therelevantprevailingprices. In all cases futurecontracts
are assumed to be completelydivisible-he can take positionsin a futurerepresenting
any numberof units of the commodityhe chooses.

G ZN'

A C

I
Y4 r~~

,Y1 \Lz1Z

7t->
-Y2

B ~~D
Fig. 2
The co-ordinatesystemin Figure 2 illustratesgeometricallythe market positions
he can takeat tl. The numberof unitspurchasedin thespotor i marketis measuredalong
the positiveX-axis,the numbersold in the givenfutureor i marketalong the negative
Y-axisand the numberpurchasedin the same futurealong the positive Y-axis.

This content downloaded from 206.212.0.156 on Mon, 26 Aug 2013 11:08:11 AM

All use subject to JSTOR Terms and Conditions
 HEDGING AND SPECULATION IN COMMODITY FUTURES 147

Iso-expectedreturnlinesof slope i + appear in Figure2 to denotecombinations

U'
of marketpositionsfor whichtotal expectedreturnremainsconstant.' Suppose, first,
thatthetraderexpectsno pricechangesin eitherthespot or thefuturemarkets,i.e., usand
u1are bothequal to 0. Verticaliso-expectedreturncurvessuchas AB and CD in Figure2
would be appropriatein thiscase to denotethe factthatpositiveexpectedreturnforany
combinationis generatedsolely by merchandising profit,m, frompositionsin the spot
marketon the X-axis.
Iso-varianceof returnellipses(whichare centeredat 0) are drawn on the basis of
equation (1) to connect combinationsfor which variance of returnremainsconstant.
Because shortpositionsin the spot marketare not postulated,only quadrants 1 and 4
of theco-ordinatesystemare relevant. The locus of tangenciesOZ indicatesthecombin-
ations of positionssuch that for any givenlevel of total expectedreturnE(R), variance
(or standarddeviation)of returnV(R) is minimized. Each such combinationof E(R)
and standarddeviationof returna/V(R) is in turnplottedin Figure 1; the opportunity
line OW which is necessarilylinear, connects these points. The trader's optimum
combinationof E(R) and v/V(R)is at point K wherehe attainshis highestindifference
curve; therefore he wouldtakethemarketcombinationin Figure2, letus say X1and -Y
read frompointL, whichgeneratesthiscombinationof E(R) and aVV(R).
Accordingto the earlierdefinitionof hedging,- Y, representsa hedge against X1
becausethisis theshortpositionin thefuturewhichminimizesthevarianceof holdingX1.
It can be seen thatany combinationread fromthe line OZ would represent
a positionin
spot againstwhicha hedge in the futureis held. By substitutionin equation (4), the
position-Y, is equal to- 1coj2. Therefore,the slope of OZ is equal to cov2

or to p -. OZ, then,indicatestheappropriatevalue along the Y-axisthatconstitutes

a hedge againstany givenvalue along the X-axis.
The empiricalobservationthat the individualtradermay assume marketpositions
representing a mixtureof hedgingand speculativeactivitycan be analyzedin thismodel
bycomparingvariousalternative marketcombinationswiththe one thatthetraderwould
take if he engagedin no speculativeactivity. Letting%odenotea positionheld in either
thespotor futuremarketunderconditionsin whichbothuiand uj are equal to 0, and letting
ocadenotethepositionin thesame marketwheneitherus or u1is non'-zero, I shall definea
" speculative element " of a marketposition as oca- co. The value of x, - xowill be called
a directspeculativeelementif it arises because of an expectednon-zeropricechangein
thesame marketto whichthevalue al - ocorefers;it will be called an indirectspeculative
elementif it arises because of an expectednon-zeroprice change in the othermarket.
Underthesedefinitions thecombinationX,, - Y, read frompointL in Fig. 2 involvesno
speculativeelementsincebothui and uj are equal to 0. The traderin thissituationmaybe
considereda " pure" hedgerbecausehe takesa positionin spoton thebasis ofan expected
returncomposed entirelyof merchandising profitand he takes a positionin the future
only to minimizethe priceriskof holdinghis positionin spot. Suppose, however,that

to equation(3a)
1 According E(R) = xi(ui + m) + xj uj. Therefore,
xj xi= j (u -+ m) + E(R)
lii
and _ _
E(R)

This content downloaded from 206.212.0.156 on Mon, 26 Aug 2013 11:08:11 AM

All use subject to JSTOR Terms and Conditions
148 REVIEW OF ECONOMIC STUDIES

usispositive, andu1remains equalto0. Theiso-expected returnlineswouldstillbevertical

andthelocusoftangencies wouldremain alongOZ, butAB andAC wouldindicate higher
levelsof expectedreturnforgivenlevelsof variancethanbefore. A newopportunity
line,OW',inFigure1indicates thatforanygivenlevelofVV(R),E(R) ishigher thanbefore.
The trader'soptimum combination of V/V(R)and E(R) is at K'; the corresponding
market combination is, say,at X2and -Y2. Becauseofan expected spotpricerisethe
tradercarriesa largerpositionin spotthanifno pricerisehad beenexpected.The in-
creaseinspotfromX; to X2is a direct speculative element.He maintains a hedgeagainst
X2byincreasing hisshortposition in thefuture to - Y2. He is motivated to increasehis
shortpositiononlyto minimize theriskofholdinga nowlargerpositionin spot,and not
becauseof anyreturn expectedin thefuture itself; nevertheless, thegap between-Y1
and-Y2 is insomesensea speculative element becauseitarisesinresponse to an expected
pricerisealthoughit is one in theothermarket.Therefore, thechangein thefuture
maybe termed an indirect speculative element.In thecase of an expected pricefall in
spotbylessthantheamountofmerchandising profit (us+ n > 0) andan expected constant
future price,iso-expected return linesremainvertical andthelocusof tangencies remain
alongOZ. For any VV(R),E(R) is lowerthanin thepurehedgecase and a newoppor-
tunity lineOW" in Figure1 is generated.The market combination is calculatedfrom
K" inthesamemanner as before, anddeviations fromX1and Y1arerespectively regarded
as directand indirect speculative elements.
If,instead, he expectsa pricefallin thefuture andno pricechangein spot,theiso-
expected return linessuchas EF wouldhavea positive slopeanda newlocusoftangencies
OZ' wouldbe generated.On thebasisofa corresponding opportunity linein Figure1,
thetrader wouldpicka market combination, sayX3,- Y3indicated in Figure2. Because
oftheexpected pricefallinthefuture, thetrader increases hisposition inthefuture beyond
thatrequired fora hedgeagainstX3 bythedistance - Y3' to - Y3. Themovement from
X, to X3 is an indirect speculative effect;themovement from- Y1to - Y3 is a direct
speculative effect.Or supposehe expectsa pricerisein thefuture and a constant price
in spot. In thiscase theiso-expected return linessuchas GH are negatively sloped,a
newlocusof tangencies OZ" is generated, and a marketcombination, say X4,- Y4 iS
taken. ln tlis casethetraderwouldgo shortin thefuture, eventhoughhe wouldexpect
a negative return inthefuture in so doing,forthesakeofreducing theriskofholdingX'
in spot. Thissituation is illustrative of Keynes'" normalbackwardation "-the trader
is willing to paya riskpremium (intheformhereofan expected lossinthefuture) forthe
sakeof reducing theriskof holdinga positionin anothermarket.However, he would
notcarrya fullhedgeagainstX4 whichwouldrequirea positionof-Y4' in thefuture;
dueto theeffect oftheexpected pricerisein thefuture thetrader carriesa hedgedposition
X,,'in spot and leaves X4- X4' unhedged. In this casethe hedge is partially withdrawn.
The reduction in spot fromXAto X4 is an indirect speculativeelement, thereduction in
theshortfuture from- Y1 to - Y4 a directspeculative element.If theexpectedprice
nse in thefuture is of sufficient magnitude, theslopeof theiso-expected return curves
willmovethelocusoftangencies, sayOZ"' to quadrant1. In thiscase thetraderwould
withdraw hishedgecompletely and go longin thefuture.
Bothexpected priccs could be allowedto varysimultaneously from0, butin thiscase
it wouldbe impossible to distinguish in Figure2 between directand indirect speculative
elements. Marketcombinations couldbe derivedas before,however, and deviations
fromXA1 and Y1couldbe considered simply as a mixture ofdirectand indirect speculative
elements.
In general, thismodelappearstoaccounttosomedegree foxthephenomena mentionied
earlierthatI haveobserved in theNewYorkcoffee market-traders maywellundertake

This content downloaded from 206.212.0.156 on Mon, 26 Aug 2013 11:08:11 AM

All use subject to JSTOR Terms and Conditions
 HEDGING AND SPECULATION IN COMMODITY FUTURES 149

hedgingactivitiesbut theseactivitiesare not independentof expectedpricechanges. A

hedge may be lifted,a long positiontakenin the future,inventories adjusted,all on the
basis of priceexpectations. This modelexplainshow priceexpectations can affectmarket
positiorisin a likemanner.

HEDGING AND RELATIVE PRICE CHANGES

What about thecase of Working,however,in whichthetradermayhave expectations

onlyabout relativepricechanges? Suppose it is trueat least in some instancesthatthe
tradertakes positionin one marketto offseta positionin anothermarketbecause of the
expectedgain in so doingand not primarily in orderto avoid risk. So long as it can be
assumed that the traderhas an indifference map betweenoverallE(R) and V V(R) as
illustratedin Figure 1, theconceptof hedgingas a formof arbitragefitsinto the present
modelas a specialcase-a case in whichthemodelcollapsesfroma two-market to whatis
in effecta one-market analysis. Since thetraderhas no priceexpectationsin theseparate
markets,he has no subjectiveprobability distributionof returnin each of thetwo markets
and, therefore, facesno priceriskas definedpreviouslyin termsof a varianceof return.
But sincehe takesa unit-for-unit hedge,his expectationsconcerningrelativepricechanges
can formthe basis of a probabilitydistribution whose mean Uhand varianceah2 measure
the expectedreturnand priceriskrespectively of lholdingthe hedgedcommodity. Both
expectedreturnand variancecan be definedonlyin termsof a unitof hedgedcommodity
consistingof one unit in spot and one unit in a futuretakentogetherand inseparablv.
Where Xh denotesthe numberof unitsheld long in spot and hedged,E(R) is equal to
Xh (uh + m) and V(R) is equal to X2 ah2. The hedger,sincehe is dealingherein a one-to-
one ratio in two markets,can be consideredto be dealing in one mnarket-the market
forthehedgedcommodity.A non-zeromeanand varianceoftheprobability distribution
of
returnhavethesamesiginificance hereas theywouldforanycommodity normally considered

0T

-Y, ~~L
H

Fig.3
in one market.Figure3 represents a co-ordinatesystemidenticalto thatin Figure2. There
are no iso-varianceor iso-expectedreturncurves,however,since expectedreturnand
varianceof returnare not computedfromspot and futurespositionstaken singly. The
onlycombinationsthetradercan take fallalong OH whichnecessarilyhas a slope of -1.
Each pointalong OH has a E(R) and a V/V(R) combinationwhichcan be plottedin Figure
1, say along O W. The optiim-um combinationat K correspondsto the combinationat L
in Figure3. If uhis equal to 0, the combinationat L would representa " pure hedge.
If Uh is greaterthan0, thetradermighttake a combinationotherthanL on the basis of a
newopportuLnity line. The difference,
measuredalongtheX-axis,betweenthiscombination
and theone at L is whatI wouldcall thedirectspeculativeeffect
on stocksheldand liedged.

This content downloaded from 206.212.0.156 on Mon, 26 Aug 2013 11:08:11 AM

All use subject to JSTOR Terms and Conditions
150 REVIEW OF ECONOMIC STUDIES

For a uh less than0 the analysisrunssymmetrically in reverseas long as m+ Uh > 0.

If m + uh < 0, the traderwould take no positionat all. Since thereis onlyone market
to consider,no indirectspeculativeeffectscan be analyzed.

This one-market thereasonwhythetheoryof hedginghas not been

modelillustrates
satisfactorily
integrated withthe generaltheoryof speculationand wlhythe tendencyhas
been to speak of the hedgeras one kind of traderand the speculatoras another. Since
thetraderhas no expectationsof absolutepricelevel movements, E(R) and V(R) can be
expressedin quantitativetermsonly along OH. Althoughwe may speak of the direct
speculativeeffectthatmay existin a hedgeas deviationsfromL in Figure3, thereis no
wayto demonstrate underwhatconditionsthetraderwould alterhis positionsin the two
marketsin otherthan a one-to-oneratio,or underwhatconditionshe would move to a
combinationin Quadrant1.

CONCLUDING REMARKS

In conclusion,the major pointsof thisanalysiscan be summarizedas follows: If a

traderhas expectationsregardingonly relativepricechanges,he necessarilytakes a unit-
for-unitpositionin the two markets. However,this hedge may contain a speculative
element,dependingon his reactionto expectednon-zerorelativepricechanges. If he has
expectationsregardingabsolutepricechanges,his activitiescan be analyzedon a general
iso-variance,iso-expectedreturnco-ordinatesystem; however,a hedgecan be meaning-
fullydefinedonlyin termsof " givena positionin one market,"i.e., one marketmustbe
regardedas his primarymarket. Withthisdefinition of hedging,the pure hedgeand the
deviationsfromit generatedby expectednon-zeroprice changes can be geometrically
analyzed. There is no distinctionbetweenthe hedgerand the " ordinary" speculator
insofaras both are motivatedby a desireto obtainia for-them optimumcombinationof
E(R) and V(R) as determinedby theirrespectiveutilityfunctions. The only essential
distinctionbetweenthemis that the hedgerhas a primarymarketwhichin this model
givesriseto a merchandising profit. A necessaryconditionfora fullhedgeto be taken
againstthepositionin theprimary marketis thatUjbe equal to 0. The ordinaryspeculator
could also workwithinthe framework of Figure2. But unlesswe postulatethe primary
marketcondition(conceptuallyit could be eitherthe spot or futuresmarket)nothingcan
be said about a hedgingelementwithinhis combinationof marketpositions.

Thismodelis expressedunderquitesimpleassumptions. Onlytheeffects of expected

pricechangesand a fixedrate merchandising profitenterinto (net) expectedreturn; all
businessriskis confinedto a pricerisk. No mnoney budgetconstraints, imputedinterest
costs or brokeragecommissionsare postulated.

Analysisis confinedto two narrowlydefinedmarkets. If a budgetconstraintwere

postulated,the principaleffectwould be thatthe tradercould be pushedoffthe locus-of-
tangencycurvein Figure2 because of a preference forsome othercombinationalong the
budgetline. For example,lettingthe budgetconstraintbe represented in quadrant4 by
EF, and consideringveiticaliso-expectedreturncurves(ui and ufbothequal to 0) we can
see thathe mightpreferL' to Jl.

1 The opportunity line in Figure1 wouldbe kinkedupwardat thepointcorrespondingto market

combination at J in Figure2, and the uppersegmentwould represent of E(R), VV(R),
combinations
alongtheconstraint
attainable segmentJP.

This content downloaded from 206.212.0.156 on Mon, 26 Aug 2013 11:08:11 AM

All use subject to JSTOR Terms and Conditions
 HEDGING AND SPECULATION IN COMMODITY FUTURES 151

Several markets,both primaryand futuresmarkets,could be includedin a multi-

dimensionalanalysisin whichthe traderselectsan optimumcombinationon the basis
map again in termsof E(R) and V(R). In thiscase thehedgewould be
of his indifference
definedas " giventhecombinationfoundin theprimarymarkets,whlatis thecombination
in the futureswhichwould minimizeoverallvariance."
Finally,the" marginalconvenienceyield,"a conceptwhichhas appearedon occasion
could be introducedinto the model.2 This yieldis a " convenience"
in the literature,
returnenjoyedby the traderwho holds an inventory.The marginalyieldis a declining
held. The effectof its inclusionherewould be to make
functionof the size of inventory
convex to the originnon-verticaliso-expectedreturnlines.
LELAND L. JOHNSON.

2 N. Kaldorand M. J. Brennan,op. cit.

This content downloaded from 206.212.0.156 on Mon, 26 Aug 2013 11:08:11 AM

All use subject to JSTOR Terms and Conditions