379
AdvancesinProductionEngineering&Management ISSN1854‐6250
Volume14|Number3|September2019|pp379–390 Journalhome:apem‐journal.org
https://doi.org/10.14743/apem2019.3.335 Originalscientificpaper
Optimal timing of price change with strategic customers under demand uncertainty: A real option approach
Lee, Y.a, Lee, J.P.b, Kim, S.b,* aDepartment of Mechanical Engineering, University of Texas at San Antonio, San Antonio, United States of America (USA) bCollege of Business Administration, Hongik University, Seoul, South Korea
A B S T R A C T A R T I C L E I N F O
Thispaperproposesamodeltodeterminetheoptimalmarkdowntimingforacompanywithstrategiccustomerpurchasingbehaviour. Since strategic cus‐tomers are aware of potentialmarkdownunder the posted pricing scheme,theymaychoosetowaitlongertomaximisetheirutilisationinsteadofbuyingaproductand fulfillingan instant surplus.On theotherhand, thesellercandelaythemarkdowndecisionuntilitisprovedtobeprofitableandhencehasanoptiontodeterminethetiming. Inestimatingthevalueof themarkdowndecision,theseller’soptionneedstobeestimated.However,thevalueoftheoptionishardtobecapturedbytheconventionalnetpresentvalueanalysis.Under market uncertainty where potential customer demand evolves overtime,theseller’srevenuefunctionisintheformofastochasticdynamicpro‐grammingmodel.Applyingarealoptionapproach,weinvestigatetheoptimalprice path and propose the optimal markdown threshold. Given the mark‐downcostsincurred,wefindthattheoptimaldiscounttimingforthefirmisdeterminedbyathresholdpolicy.Furthermore,ourresultsshowthatiffuturemarketbecomesmoreuncertain,thesellerneedstowaitlongerordelaythemarkdown decision. In addition, the optimal threshold of the markdowndecreasesexponentiallyinadecliningmarket,whichexplainstheearlymark‐downpolicyofsomeconsumerproductcompanies.
©2019CPE,UniversityofMaribor.Allrightsreserved.
Keywords:Strategiccustomers;Pricechange;Postedpricing;Markdown;Demanduncertainty;Realoption
*Correspondingauthor:[email protected](Kim,S.)
Articlehistory:Received28February2019Revised10September2019Accepted12September2019
1. Introduction
Demandmanagementbecomesthebasisforthedecisionmakingofthefirms;fromproductionplanningtoinventorymanagement[1,20].Pricingpoliciesarefrequentlyusedtoolswhenfirmsmanagetheirdemand[20].Thepricingpoliciesofafirmareoftencomplexanddiversedepend‐ingonthebusinessenvironmentinwhichthecompanylies[1,5,6,14,20].Inthefashionindus‐try, for instance,simplemarkdownpricing iswidelyusedtoselloutremainingstockaftertheregularsalesseason[21].Somecustomersmaychoosetowaitandpurchasetheproductlateratthemarkdownpriceratherthanbuyingitrightaway.Ontheotherhand,airlinecompaniescon‐tinuouslymarkupthepricesofticketsupondeparture.Whenlookingforanairlineticket,cus‐tomerscanexpectan increase inprices if theydelaytheirpurchase.Therefore,understandingcustomerpurchasingbehaviouriscriticalforthefirmtomakepricingdecisions.
Strategic customers in theoperationsmanagement literature aredefined as thosewhoareawareofthefirm'sdynamicpricingpoliciesandmakeinter‐temporalpurchasingdecisions[20].Sincesuchcustomersareconsciousofpotentialchangesinpricesatalaterpointintime,theyarebeingstrategicratherthanmyopic.Insteadofbuyingaproductandfulfillinganinstantsur‐
Lee, Lee, Kim
380 Advances in Production Engineering & Management 14(3) 2019
plus, they strategicallywait for a future pricemarkdown and thereby seek tomaximise theirutilisation[5,6,18].Assuch,strategiccustomershavebecomeasubstituteformyopiccustom‐erswhosimplymakeabuyingdecisionifthepriceislowerthantheirvaluation[5,6,20].There‐fore, firmsmustcomprehendthestrategicbehaviourofcustomersandfindanoptimalpricingschemebasedonittomaximiserevenue.
Inresponsetothestrategicbehaviourofcustomers, the firm'sdecisionsaregenerallytwo‐fold:thetimingofpricechangesandtheavailabilityoftheproduct[3,20].Thefirmsellsaprod‐uctforadurationoftime,afterwhichitmaydecidetochangethepriceatacertainpointintime.Limited supply could also be used as a marketing strategy to increase the sense of urgencyamongcustomers.Therefore,customersinthemarketchooseeithertopurchaseaproductatitscurrentpriceor to revisit it after theprice goesdown, consideringnotonly the timingof themarkdownbutalsothepossibilityofsellouts.
Inthispaper,weinvestigatethemarkdowndecisionofamonopolistwhowishestomaximiseexpectedrevenuesinthepresenceofstrategiccustomers.Ourmodelcapturesseveralimportantpropertiesofthemarketenvironmentforconsumerproducts.First,thesellercommitstoafixedpath of two prices: itmay sell a product for a duration of time, afterwhich itmay decide tochangetheprice.Themarkdowndecisioncanbemadenomorethanonceoverthesaleshorizonand is hence irreversible. Second, customers show strategic purchasing behaviour towardsfirms:evenifthevaluationoftheproductexceedsthepriceoftheproductduringthefirstpartofthesaleshorizon,customersmaynotsimplypurchaseit.Instead,theirdecisiontopurchaseisbasedonthevaluationthatexceedsacertainlevel,thusfollowingathresholdpolicy.Third,po‐tentialcustomerdemandisstochastic.Inparticular,themarketsizefollowsageometricBrown‐ianmotionthatevolvesdynamicallyovertime.
Weconsideraseller'sproblemondecidingtheoptimalpricepathandthetimingofamark‐down under demand uncertainty. Specifically,we present a stochastic dynamic programmingmodelwherethesellerhasasingleopportunity todiscountthepriceof theproductatasunkcost.Customersinthemarketareawareofapotentialmarkdownandthelikelihoodofasellout.Basedoncustomers'valuationinregardtothetwoprices,thevalueofthefirmisexpressedasastreamofexpectedrevenues.Solvingtheproblemusingarealoptionapproach,weshowthattheseller'soptimalmarkdowntimingdecision isbasedonthethresholdpolicy.Tothebestofourknowledge,thisisoneofthefirststudiesthatconsidersapostedpricingschemeundermar‐ketuncertainty.
Theremainderofthepaperisorganisedasfollows.Section2outlinespreviousrelatedworkstosummariseextantresearch.Section3proposesarevenuemaximisationmodel,andSection4continueswiththetopicbyanalysingthesolutionofthemodel.Finally,wediscussbroaderfind‐ings,conclusions,andpotentialfutureresearchopportunitiesinSection5.
2. Literature review
Strategiccustomersandfirms'pricingpolicyproblemshavebecomeanincreasinglyproductiveresearch area.Amongothers, studies regarding customerpurchasingbehavior are thoroughlyreviewedbyShenandSu[20].Mostofthepapersintheliteratureconsidertwoimportantele‐ments inmodellingstrategiccustomerbehaviour.Thefirst isthearrivalprocessofcustomers.Whethercustomerspreexistedinthemarket[7,13,18]orsequentiallyarrivedinthemarket[3,14,21,23]isaquestionbasedonthispremise.Thesecondisthedecisionmakingofthecustom‐ersandhowthedecisionmakingultimatelyformsanequilibrium.Regardlessofthemarketsize,thedecisionmakingofanindividualcustomermakesanimpactonthedynamicsofthemarketto some extent. For instance, when many customers purchase goods in the early stages, theproductmayrunoutofstockforsomeofthosewhoinitiallydecidedtodelaythepurchase[13].Sometimescustomersmayhavetopurchasethegoodsatanevenhigherpriceifthefirmorsell‐erchoosestoadoptamarkuppricingpolicy[24].Astrategiccustomertriestomakeanoptimaldecision,foreseeingthesesituations,andthisprocessmay, inturn,compriseanequilibriumindecisionmaking.Theseller,ontheotherhand,setshisorherpricingpolicybasedonthisequi‐libriuminanefforttomaximiseprofit.Therefore,thisgame‐theoreticrelationshipwithconflict‐
Optimal timing of price change with strategic customers under demand uncertainty: A real option approach
Advances in Production Engineering & Management 14(3) 2019 381
ing interestsbetweenthesellerandthestrategicbuyersnecessarily leadstoahighlycomplexmodelinmanystudies.
Thereare twotypesofsimplifications todealwith thecomplexity in themodelling.Firstly,thetimeofthepricechangeisoftenfixed.Aspecificnumberofperiodsarepresumed,andstaticpricingismaintainedforthedurationoftheperiods.Inotherwords,theanalysisofoptimalpric‐ingisbasedonthedefinitenumberofperiodsinwhichafixedpriceisoffered,ratherthanfind‐ingthepricechangingperiodonebyone[5‐7,13].Thesecondcaseisthesizeofthemarket:inapplyingagametheoryapproach,asmallmarketsizeisassumed.Inthissituation,acustomerpredicts the decisionmaking of other consumers tomake his or her optimal decision and anequilibriumofstrategicpurchases isachieved.AvivandPazgal[3] foundthata firm'sbenefitsfrompricedifferentiationmaydecreaseascustomersbecomemorestrategic,andhenceoptimalpricingpoliciesmayresultinpotentialrevenuelossesinthepresenceofstrategiccustomers.
Customers'purchasingdecisionsdependontheinteractionamongthepricingpolicy,availa‐bility,customervaluations,remainingtime,andsoforth.Underthepostedpricing,forinstance,wherethesellerannouncesitspricepathinadvance,theavailabilityoftheproductorthepossi‐bilitytopurchaseitlaterintimewillbethemajorconcernforthecustomer[2,7,13].AsDasuandTong[7]specificallypointedout,theseller'sdynamicpricingdecisionismeaningfulonlyifcustomersareawareofthestock‐outpossibility,whiletheimpactoftheperceptiononstrategiccustomerbehaviourisdifferent inheterogeneouscustomervaluations[23]. Inmanystudies,atwo‐periodpostedpricingschemehasbeenusedduetoitssimplicityandapplicability,althoughthesellercanstillmakeapricechangeatanytime[4,7,13,15].DasuandTong[7],inparticular,foundthat theapproximationclose to themaximumrevenuecanbeachievedbytwoor threepricingchanges.Inthisstudy,ourmodelwillalsobebasedonthetwo‐periodpostedpricingincontinuoustimeperiodstofindtheoptimaltimingofpricechange,whiletheavailabilityoftheitemislimitedafterthemarkdown.
Asforthefirm'spointofview,ontheotherhand,marketsizeisthemainsourceofuncertain‐ty.Giventhepriceandthetimingofthepricechange,thefirm'srevenuemustbesignificantlydifferentdependingonchangesindemandatthemoment.Undermarketuncertainty,thesellercaneithermakean immediateprice changeor intentionallydelay thedecision toobserve theactualdemandmovement.Thissituationisverycommoninmanyoperationalpractices:compa‐nieshaveanopportunitytoinvestbuttheycanstillwaitfornewinformation.Inotherwords,afirmwith theability topostponeadecisionhas theoption,not theobligation, toexercise it –making it analogous to holding a financial call option. Since first proposed by Pindyck [19],McDonaldandSiegel[17],DixitandPindyck[8]andothers,thisrealoptionapproachhasbeenwidelyborrowedintheareasofmarketingandoperationsmanagementbecauseithelpsustobetterunderstandthetruevalueoftheinvestmentopportunity.
Adoptingtherealoptionconceptisnotcompletelynewinrevenuemanagementliterature.Innumerouspapers,thedynamicpricingdecisionisdeterminedbyconsideringtheoptionvalueofunsoldproducts[11,16].Sincethisoptionvaluedecreasestowardstheendofthetimehorizon,theoptimalpricepathalsodecreasesovertime.Inanotherpaper,GallegoandSahin[10]usedthe real option approach tomodel uncertain customer valuations. In this paper, however,weassumethatpotentialcustomerdemandevolvesovertimeandfollowsthegeometricBrownianmotion(GBM).Assumingtheknowndistributiononcustomervaluationsandthelevelofavaila‐bility,weexploretheoptimalmarkdowntimingproblembasedonthenetpresentvalueoftheseller'sexpectedrevenue.Tothebestofourknowledge,intheliteratureonstrategiccustomers,there areonly ahandful of studies thatdealwith theoptimal timingof price change, andyetfewerstillthatatthesametimeaddressoptimalpricingwithstrategiccustomersundermarketuncertainty.
3. Model description
Inthispaper,weconsideramonopolisticfirmthatsellsasingleitemtopotentialcustomersovertwoperiods.Thefirmwantstomaximiseitsnetpresentvalueofexpectedrevenue.Below,weexplainfurtherassumptionsbeforebuildingourmodel.
Lee, Lee, Kim
382 Advances in Production Engineering & Management 14(3) 2019
Assumption1.Themonopolistic firm followsatwo‐periodmarkdownpricingschemeandcom‐mitstothepricepathinbothphases.
Assumption2.Theoriginal( )andmarkdownprices( )arepre‐announcedandthe in‐stockprobability(π)inthesecondperiodisalsogiveninformation.
Assumption 3. Customersare strategic rather thanmyopicandareawareofmarkdownsandpossibilitiesofstock‐outs.
Assumption4.Thedistributionofcustomervaluations( ⋅ )isknown.
3.1 Valuation of a strategic customer
Supposethatthereisamonopolistwhohasasufficientlylargenumberofanitem.Untiltime ,theitemisinitiallysoldatprice ,andafter theitemissoldatthemarkdownprice .Thetwoprices, and ,arepre‐announced.CustomerdemandfollowsageometricBrownianmotion(GBM),andeachcustomerissupposedtopurchaseonlyoneunitoftheitem.Whentheselleroffersamarkdownprice,weassumethatthesellercancontrolthelevelofproductavaila‐bility, ,toinducescarcity.Inotherwords,inthesecondperiod,thein‐stockprobabilitydecidedbythesellerwillbesetto 1.Controllingtheavailabilityofservicesoritemsofdifferentclas‐sesisprevalentinrevenuemanagement[24]andinducingalevelofscarcityisalsooneofthemostcommonstrategiesinmarketing[9,22].
Let denotethecustomer'ssurplus.Thentheutilisationofthecustomerwhopurchasestheitemrightnowisasfollows:
(1)
where isthesurplusofthecustomerand isthecurrentpriceoftheproduct.Similarly,theutilisationofthecustomerwhodecidestowaitforthediscountisasfollows:
(2)
where is the customer'svaluationof theproduct, is the currentpriceof theproduct, isthe future price of the product, and is the service level of the product at the lowerprice .Thus, the stock‐out probability is1 . stands for the customer's preference for risk; 0indicates risk‐averse, 0risk‐taking, and 0risk‐neutral attitude. Furthermore, we as‐sumethatthecustomersareeitherrisk‐averseorrisk‐neutral,whichisaprevalentassumptionmadebymanyresearchers[13,15].
Inthissetting,thestrategiccustomersdecidetopurchaseinthefirstperiodiftheirvaluationisgreaterthanorequaltothethresholdvalue.Thepurchasingdecisionofthestrategiccustomerisdeterminedbythefollowinglemma.
Lemma1.Thethresholdofastrategiccustomer'svaluationisgivenby:
1 (3)
Proof.Thetwochoices,purchasingrightnoworwaitingforadiscount,generatethesamesur‐pluswhen .Solvingtheequation,astrategiccustomerwillhavethefollowingthreshold.Thatis,
. (4)Thiswouldfinishtheproof.
If 1,nocustomerswouldbuyinthefirstperiod.Furthermore,weassumethatcustomersdonotpurchaseiftheutilisationislessthanzerowithoutlossofgenerality.Thatmeans, isnotlessthan .Therefore,itissufficienttoconsideronlythecasewherethethresholdisbetweenthefirstperiod'spricingandone.Thatis,
1 (5)
Thisalsodecidestheupperandthelowerboundsof accordingly.Followingtheliterature,potentialcustomerdemandisassumedtobeamultiplicationofthe
customervaluefunctionandthetime‐varyingpotentialdemand.Thatis,thedemandfunctionisgivenby
Optimal timing of price change with strategic customers under demand uncertainty: A real option approach
Advances in Production Engineering & Management 14(3) 2019 383
1 (6)
where is a knowndistribution function of the product at customer valuation , and isthemultiplicativedemandshockprocess.Thismaybethoughtasdemandinwhichtheproducthasaunitprice.
In the firstperiod,a strategiccustomerwouldpurchase theproduct, if and 0.Therefore,fromtheconditions,
and (7)
wehave
. (8)
Since forthecustomerspurchasinginthefirstperiod,thecurrentdemandfunctionis
1 11
. (9)
Ontheotherhand,aproportionofcustomerswouldwaitandpurchaselateratalowerprice,if and 0.Thevaluationofsuchcustomersisasfollows:
. (10)
Hence,thedemandfunctionofthecustomerswhocomebacklaterinthesecondperiodtopur‐chasewillbe:
1. (11)
Finally,whentheproductstartsbeingsoldatamarkdownprice ,anycustomerwhoseval‐uationisatleastgreaterthanthepricewouldpurchaseit.Namely,thedemandfunctionwillbe:
1 . (12)
Without loss of generality, let the valuation of customers, , be uniformly distributed over[0,1].Thenthedemandforeachcaseisgivenasfollows:
1 11
1 (13)
1 1 1 (14)
1 . (15)
3.2 Customer demand
Inthispaper,weuseageometricBrownianmotion(GBM)toformulatethemulti‐plicativede‐mandshock attime .Thatmeanstherelativechangeindemand, / , withinashorttimeinterval, , ,canvarywithtime .Thedynamicsofdemandarerepresentedbythefollow‐ingformula:
, (16)
where isthegrowthrateordriftrateindemand, isthevolatilityoftheprocess,and isastandardWienerprocess. If 0,market size is increasingover time. If 0,market size isdecreasing.
Thiscontinuousrandomvariable issaidtohavelognormaldistributionbecausetheinte‐gralofEq.16givesthefollowingdemandfunction(seeAppendixAforthederivation):
⁄ , (17)
where istheinitialdemand.Whilethebell‐shapedpatternofdemandisexpectedbyEq.17,therealisationofdemandwillsubstantiallydeviatefromit,dependingonthemarketvolatility,asshowninFig.1.
Lee, Lee, Kim
384 Advances in Production Engineering & Management 14(3) 2019
Fig.1 SamplepathofpotentialdemandfromthegeometricBrownianmotioninadecreasingmarket.
Note that the threesamplepaths inFig.1aredrawn fromEq.16withameandrift rateof
0.1and three standard deviations of 0.05, 0.1, and 0.2. As shown in the figure, thesamplepathwithalargerstandarddeviationtendstofluctuatesignificantly,whileallthreetra‐jectorieshaveadecreasingtrendincommonduetothenegativemeandriftrate.
3.3 Optimal timing of price discount
Next,weconsidertheoptimaltimingproblemconditionedonthecustomer'spurchasingstrat‐egy.Wedevelopamodelforanoptimaldiscounttimingdecisionusingarealoptionmodel.Inpractice,thecompanyhasan"option"todelaythediscountandhenceneedstodeterminewhenthepriceshouldbediscounted.Aftermarkdown,thecompanywouldmakerevenue ,withanirreversiblesunkcost beingincurredfromsalespromotion,inventorymanagement,andsoforth.
Hereinweformulatethevaluefunctionofthefirmwithanopportunityofthediscounttiming.Whentheproductissoldattheoriginalprice ,aproportionofcustomers, ,whosevalua‐tionisfarhigherthan ,orgreaterthan ,willdecidetopurchasetheitem.Agroupofstrategiccustomers, , whose valuation is between and would like to wait and see if the price ismarkeddown.Once the firmdecides todiscount theoriginalprice to themarkdownprice, ,theycomebacktopurchasetheproductbutonly ofthemwillbeabletogetone.Weassumethat suchdemand is instantaneous,meaning that customerdemandaccumulatedup to time willberealisedattime [3].Fromtime ,anycustomerswhosevaluationisatleasthigherthanwouldliketopurchasetheproductbut,again,only ofthemwouldgetone.Webeginwith thevalue functionof the firm for theoptimaldiscount timingproblem.The
valueofthefirm, ,isthestreamofrevenue,whichconsistsofthreecasesstatedearlier.Weusedynamicprogramming,stipulatinganexogenousdiscountrate .Then istheexpectedpresentvalue
max 1 (18)
where
1 . (19)
Sincethedemandofstrategiccustomerswillberealisedattime ,werearrangetheformulasothattherevenueisincludedintheterminalpayoff .Then
Optimal timing of price change with strategic customers under demand uncertainty: A real option approach
Advances in Production Engineering & Management 14(3) 2019 385
max 1 (20)
where
1 (21)
1 (22)
1 (23)
Since 1 / (See Appendix A) and ⁄ , we
finallyhave:
1 1 (24)
Substituting inEq.3intotheformula,thevaluefunctionofthefirmissummarisedasfollows.
Proposition1.Thevaluefunctionofthefirmforoptimaldiscounttiming isgivenby:
max1
1 (25)
where
11 1 (26)
Bysolvingthisstochasticdynamicprogrammingproblem,wecanobtaintheoptimaltimingforamarkdown.Theoption‐likeapproachshownin[17]and[19]isusedtosolvethedynamicstochasticproblem.Asthepotentialdemand evolvesstochastically,theoptimalstrategyistoexercise(markdown)sothatthevalueisatleastgreaterthanthecriticalvalue ∗.Afirm’sop‐timalmarkdowntimingsolutionisrepresentedinthefollowingproposition.Proposition2.Acompanyconsideringmarkdownoftheretailpricewillhaveavaluefunctionasfollows:
1 ∗
1 1 ∗ (27)
where∗
1⋅
1 11 (28)
1 1⋅
∗
(29)
and12
12
2 (30)
Proof.Let denotethetimingatwhichthefirmdiscountstheoriginalpriceoftheitem.Asde‐scribedearlier,thefirmmakesrevenueflowsof 1 beforethemarkdown.Attime ,thefirmwouldmakerevenueflow withtheirreversiblesunkcost .Therefore,asshownin[8],theBellmanequationinthecontinuationregion,wherevaluesof arenotoptimaltomark‐down,isgivenby:
1 , (31)
whichimpliesthatoveratimeinterval ,thetotalexpectedreturnonthemarkdownopportu‐nityisequaltoitsexpectedrateofcapitalappreciation.
Lee, Lee, Kim
386 Advances in Production Engineering & Management 14(3) 2019
ApplyingIto'slemma,wehave
12
, (32)
where ⁄ and ⁄ .SubstitutingEq.16anddividingthroughby ,wehavethefollowingBellmanequation(see
AppendixBforproof):
12
1 0(33)
Toensuretheexistenceoftheoptimalsolution,weassumethat .Thedifferentialequa‐tion mustsatisfythefollowingthreeboundaryconditions:
0 0 (34)
∗ 1 1∗
(35)
∗ 1 (36)
Eq.34holdsbasedon theobservation that itwill stayzero if thestochasticprocess goes tozero.Theothertwoequationsaretoimposecontinuityandsmoothnessatthecriticalpoint ∗,thepotentialdemandatwhichitisoptimaltodiscount.Eq.35isthevalue‐matchingcondition,indicatingtherevenuethefirmmakesuponmarkdown.Eq.36isthesmooth‐pastingconditionatthepoint.
Therefore,thesolutionofthedifferentialEq.33musttaketheform
1, (37)
where isaconstanttobedeterminedand isoneofthesolutionsofthefollowingquadratureequation:
12
1 0. (38)
Solvingtheequationandtake
12
12
21 (39)
toensuretheboundarycondition.Fromthesmoothpastingandthevalue‐matchingconditions,wehave
∗ ∗ 1 ∗ 1 1∗
(40)
and
∗ ∗ 1 1. (41)
Solvingtheseequationsresultsin:
∗1
⋅1 1
1 (42)
1 1⋅
∗
(43)
and12
12
2. (44)
Herein, only if 1 / and 1 1 ,we have a positive
Optimal timing of price change with strategic customers under demand uncertainty: A real option approach
Advances in Production Engineering & Management 14(3) 2019 387
threshold ∗ 0.ThisresultleadstoProposition3.Again,thethreshold ∗determinestheoptimalmarkdowntimingforafirm.Whentheactual
customerdemandofthefirmattime islowerthanthethreshold ∗,itisbeneficialtoselltheproductattheoriginalretailprice ,makingtherevenuestreamof 1 / aswellasgiving the flexibility that the firmcanhold for thepricemarkdown,measuredbyαX .On theotherhand,whentheactualdemandisgreaterthanthethresholdX∗,thefirmwilldecidetodis‐
countthepriceandtakethebenefitofmarkdown 1 byspend‐
inginvestmentcost .
4. Analysis and discussion
Thissectionexplainssomeof the importantcharacteristics foroptimalmarkdownapproachessuggestedearlier.First,thefollowingpropositionillustratesthatthereexistsapositivethresh‐oldforthefirmatanygiventime underspecificconditionsfor and .
Proposition3. Let ∗denotetheoptimaltimingofmarkdowntomaximisethe firmvalue.Thentheoptimalmarkdowntimeisfinite ∗ ∞[12],andthefirstepochthatdemandexceedsthethre‐sholdisestimatedatthefollowingtime:
∗ inf 0 | ∗ , (45)
where there exists a positive threshold ∗ ⋅ 1
attime if 1 / and 1 1 / .Proof.ByProposition2.
Notethatweassumeadecreasingmarketsize( 0).Astime increases,therefore,wecanobserve that the threshold ∗decreases exponentially,while theminimumvalue for the fixedcost thatisrequiredforthisapproachtobefeasibleincreasesexponentiallybeforehittingthelowerbound asshowninthefollowingproposition.
Proposition 4. As → ∞, we have a threshold ∗ → 0and the lower bound of the fixed cost→ / .Again,as the threshold ∗for themarkdowndecreasesexponentially,a firmis likely tode‐
cideonapricediscountintherelativelyearlystages.Italsoindicatesthatnosignificantrevenueis expected after a certain amountof timebecauseof the reduction in customerdemand, andhencethefirmnolongerneedstoinvestmoreinlaterstagesunderthisapproach.
Astheoptimaltiming ∗isrepresentedbysomeexogenousfactors,weexploittheimpactoftheparametersonthethreshold.
Proposition5.Theoptimaltimingthresholdincreaseswithrespecttodemandvolatility.Thatis,
∗
0 (46)
Proof.Notingthefractionalvalue 0,wetakethederivativeof fromEq.16withrespecttothedemandvolatility, .Weknow 11isoneofthesolutionstothefollowingquadraticfunction
0(theothersolutionis 0),where12
1 .(47)
Takingaderivativeoftheequation,wehave
0. (48)
Since ⁄ 0and ⁄ 0,wehave
0. (49)
Furthermore,
Lee, Lee, Kim
388 Advances in Production Engineering & Management 14(3) 2019
111
0. (50)
Finally,thederivativeoftheoptimalthreshold ∗withrespecttoσis,∗
1 1 11 0 (51)
Thispropositionindicatesthattheoptimalmarkdownthresholdincreasesasthevariationindemandincreases.Simplyput,itismorebeneficialforthefirmtowaitanddelaythemarkdown,therebyavoidingtheriskofmakingtheinstantaneousdecisionwhenthemarketishighlyuncer‐tain. The firm iswilling tomake themarkdown decision, onlywhen excessive revenue is ex‐pectedwheretheamountofuncertaintyregardingfuturedemandislarger.
5. Conclusion
Inthispaper,theoptimalpricingpolicyofamonopolisticfirmisinvestigatedwithstrategiccus‐tomerbehaviour.Whencustomersstrategicallywait foradiscount, themonopolisthasanop‐tiontoofferamarkdowntomaximiseitsrevenue.Assumingthattheunderlyingcustomerde‐mandisstochastic,evolvingdynamicallyovertime,wedevelopavaluefunctionforthefirmtofindtheoptimaltimeforthediscount.Usingarealoptionapproach,thestochasticdynamicpro‐grammingmodelissolved.Giventhefixedcostofthemarkdown,servicelevel,andaknowndis‐countedprice,theoptimalpolicyforthefirmistofollowthethresholdpolicy.Thesellermax‐imisesitsrevenuebydiscountingthepriceoftheproductwhenthepotentialcustomerdemandisgreaterthanthethresholdvalue.
Thecontributionofthispaperisasfollows:Consideringtheoptimalmarkdowndecisionforamonopolistic sellerwith strategic customers,we address the gap in other literature on thesecustomerswithproblemsundermarketuncertainty.Astochasticdynamicoptimisationmodelisproposedtofindtheoptimalmarkdownstrategyoftheseller.Arealoptionapproachisappliedtoobtainaclosed‐formsolutionofthefirm’sdemandthreshold.Theanalysisoftheoptimaltim‐ingrevealstherelationshipbetweenthedegreeofmarketuncertaintyandthemarkdowndeci‐sion‐making.
Althoughtheoptimalthresholdpolicyisfound,carefulinterpretationsoftheresultareneed‐ed.First,customersareawareofpotentialmarkdownswhilethediscountedpriceisknown.Theseller may not exercise the option to markdown if the potential demand never exceeds thethreshold. Second,we found that there is an exponential decrease in the threshold value in adecliningmarket,whichjustifiestheearlymarkdowninsomeindustries.Ontheotherhand,theoptimalmarkdownthresholdincreasesasthevariationindemandincreases.Thisindicatesthata firmneeds to avoid the riskof committingmarkdownpricing too earlywhen themarket ishighlyuncertain.Thecompany’smanufacturingandproductionplanningmustbealignedwiththisstrategicdecisiononthemarkdowntiming.
There aremany challenges involved in the proposed study for future research. Discussionoverpotentialdemandisrecommended.Furtherinvestigationonthepostedpricingschemeofdemanddiffusioncanbedevelopedwherethenewproductgetsadoptedinthepopulationovertime.Anotherpotentialareaofresearchwouldbethepredictionofstrategiccustomerdemandbyapplyingdata‐drivenapproaches,suchasmeta‐heuristicsandmachinelearningalgorithms.Finally,aninterestingextensionwouldbetoimplementtheproposedframeworkonreal‐worldproblemstodemonstratethepracticalimplicationsofourmodel.
Conflict of interests
Theauthors thank theeditorand tworeviewers for their constructivecomments,whichhelpedus to improve thispaper.Theauthorsdeclarethatthereisnoconflictofinterestsregardingthepublicationofthispaper.
Optimal timing of price change with strategic customers under demand uncertainty: A real option approach
Acknowledgement This work was supported by the Ministry of Education of the Republic of Korea and the National Research Foundation of Korea (NRF-2016S1A5A8019542).
References [1] Albey, E., Norouzi, A., Kempf, K.G., Uzsoy, R. (2015). Demand modeling with forecast evolution: An application to
production planning, IEEE Transactions on Semiconductor Manufacturing, Vol. 28, No. 3, 374-384, doi: 10.1109/ TSM.2015.2453792.
[2] Arnold, M.A., Lippman, S.A. (2001). Analytics of search with posted prices, Economic Theory, Vol. 17, No. 2, 447-466, doi: 10.1007/pl00004113.
[3] Aviv, Y., Pazgal, A. (2008). Optimal pricing of seasonal products in the presence of forward-looking consumers, Manufacturing & Service Operations Management, Vol. 10, No. 3, 339-359, doi: 10.1287/msom.1070.0183.
[4] Cachon, G.P., Swinney, R. (2009). Purchasing, pricing, and quick response in the presence of strategic consumers, Management Science, Vol. 55, No. 3, 497-511, doi: 10.1287/mnsc.1080.0948.
[5] Chen, Y., Farias, V.F. (2018). Robust dynamic pricing with strategic customers, Mathematics of Operations Re-search, Vol. 43, No. 4, 1119-1142, doi: 10.1287/moor.2017.0897.
[6] Chen, Y., Farias, V.F., Trichakis, N. (2019). On the efficacy of static prices for revenue management in the face of strategic customers, Management Science, doi: 10.1287/mnsc.2018.3203.
[7] Dasu, S., Tong, C. (2010). Dynamic pricing when consumers are strategic: Analysis of posted and contingent pricing schemes, European Journal of Operational Research, Vol. 204, No. 3, 662-671, doi: 10.1016/j.ejor.2009. 11.018.
[8] Dixit, A.K., Pindyck, R.S. (1994). Investment under uncertainty, Princeton University Press, Princeton, New Jersey, USA.
[9] Dye, R. (2000). The buzz on buzz, Harvard Business Review, Vol. 78, 139-146. [10] Gallego, G., Sahin, O. (2006). Inter-temporal valuations, product design and revenue management, Airline Group
of the International Federation of Operational Research Societies (AGIFORS). [11] Gallego, G., Van Ryzin, G. (1994). Optimal dynamic pricing of inventories with stochastic demand over finite
horizons, Management Science, Vol. 40, No. 8, 999-1020, doi: 10.1287/mnsc.40.8.999. [12] Harrison, J.M. (2013). Brownian models of performance and control, Cambridge University Press, Cambridge,
United Kingdom, doi: 10.1017/CBO9781139087698. [13] Kim, S., Huh, W.T., Dasu, S. (2015). Pre-announced posted pricing scheme: Existence and uniqueness of equilib-
rium bidding strategy, Operations Research Letters, Vol. 43, No. 2, 151-160, doi: 10.1016/j.orl.2014.12.010. [14] Liu, M., Bi, W., Chen, X., Li, G. (2014). Dynamic pricing of fashion-like multiproducts with customers’ reference
effect and limited memory, Mathematical Problems in Engineering, Vol. 2014, Article ID 157865, 10 pages, doi: 10.1155/2014/157865.
[15] Liu, Q., Van Ryzin, G.J. (2008). Strategic capacity rationing to induce early purchases, Management Science, Vol. 54, No. 6, 1115-1131, doi: 10.1287/mnsc.1070.0832.
[16] McAfee, R.P., Te Velde, V. (2007). Dynamic pricing in the airline industry, In: Hendershott T.J. (ed.), Handbook of Economics and Information Systems, Vol. 1, Elsevier Science, New York, USA, doi: 10.1016/S1574-0145(06) 01011-7.
[17] McDonald, R., Siegel, D. (1986). The value of waiting to invest, The Quarterly Journal of Economics, Vol. 101, No. 4, 707-727, doi: 10.2307/1884175.
[18] Özer, Ö., Zheng, Y. (2015). Markdown or everyday low price? The role of behavioral motives, Management Sci-ence, Vol. 62, No. 2, 326-346, doi: 10.1287/mnsc.2014.2147.
[19] Pindyck, R. (1990). Irreversibility, uncertainty, and investment, Working Paper No. 3307, National Bureau of Eco-nomic Research, Cambridge, USA, doi: 10.3386/w3307.
[20] Shen, Z.-J.M., Su, X. (2007). Customer behaviour modelling in revenue management and auctions: A review and new research opportunities, Production and Operations Management, Vol. 16, No. 6, 713-728, doi: 10.1111/ j.1937-5956.2007.tb00291.x.
[21] Smith, S.A., Agrawal, N. (2017). Optimal markdown pricing and inventory allocation for retail chains with inven-tory dependent demand, Manufacturing & Service Operations Management, Vol. 19, No. 2, 290-304, doi: 10.1287/msom.2016.0609.
[22] Stock, A., Balachander, S. (2005). The making of a “hot product": A signalling explanation of marketers' scarcity strategy, Management Science, Vol. 51, No. 8, 1181-1192, doi: 10.1287/mnsc.1050.0381.
[23] Su, X. (2007). Intertemporal pricing with strategic customer behaviour, Management Science, Vol. 53, No. 5, 726-741, doi: 10.1287/mnsc.1060.0667.
[24] Talluri, K.T., Van Ryzin, G.J. (2006). The theory and practice of revenue management, Springer-Verlag, New York, USA.
Advances in Production Engineering & Management 14(3) 2019 389
Lee, Lee, Kim
Appendix A: Proof of Proposition 1
Since 𝐸𝐸 �∫ 𝑋𝑋𝑡𝑡𝑇𝑇0 𝑑𝑑𝑑𝑑� = ∫ 𝐸𝐸 𝑇𝑇
0 [𝑋𝑋𝑡𝑡]𝑑𝑑𝑑𝑑 by Fubini's Theorem, we first apply Ito's lemma to 𝑑𝑑 ln𝑋𝑋𝑡𝑡 , to find 𝐸𝐸[𝑋𝑋𝑡𝑡]:
𝑑𝑑 ln𝑋𝑋𝑡𝑡 =1𝑋𝑋𝑡𝑡𝑑𝑑𝑋𝑋𝑡𝑡 −
12
1𝑋𝑋𝑡𝑡2
(𝑑𝑑𝑋𝑋𝑡𝑡)2 (52)
=1𝑋𝑋𝑡𝑡
(𝜇𝜇𝑋𝑋𝑡𝑡 + 𝜎𝜎𝑋𝑋𝑡𝑡𝑑𝑑𝑑𝑑)−12
1𝑋𝑋𝑡𝑡2
(𝑋𝑋𝑡𝑡2𝜎𝜎2𝑑𝑑2) (53)
= 𝜇𝜇 𝑑𝑑𝑑𝑑 + 𝜎𝜎 𝑑𝑑𝑑𝑑 −12𝜎𝜎2 𝑑𝑑𝑑𝑑 (54)
After integrating and applying the fundamental theorem of calculus, we obtain:
ln𝑋𝑋𝑡𝑡 − ln𝑋𝑋0 = �𝜇𝜇 −12𝜎𝜎2� 𝑑𝑑 + 𝜎𝜎𝑊𝑊𝑡𝑡 (55)
𝑋𝑋𝑡𝑡 = 𝑋𝑋0𝑒𝑒�𝜇𝜇−12𝜎𝜎
2�𝑡𝑡+𝜎𝜎𝑊𝑊𝑡𝑡 (56)
The general form of expectation for Gaussian random variable is 𝐸𝐸[𝑒𝑒𝑋𝑋] = 𝐸𝐸 �𝑒𝑒𝜇𝜇+12𝜎𝜎
2�, where 𝑋𝑋
has the law of a normal random variable with mean 𝜇𝜇 and variance 𝜎𝜎2. Since we know the stand-ard Brownian motion 𝑊𝑊𝑡𝑡~𝑁𝑁(0, 𝑑𝑑), taking expectation on both sides yields the following [9]:
𝐸𝐸[𝑋𝑋𝑡𝑡] = 𝑋𝑋0𝑒𝑒�𝜇𝜇−12𝜎𝜎
2�𝑡𝑡𝐸𝐸[𝑒𝑒𝜎𝜎𝑊𝑊𝑡𝑡] (57)
= 𝑋𝑋0𝑒𝑒�𝜇𝜇−12𝜎𝜎
2�𝑡𝑡𝑒𝑒0+12𝜎𝜎
2𝑡𝑡 (58)
= 𝑋𝑋0𝑒𝑒𝜇𝜇𝑡𝑡 (59) Finally taking integral produces the following results:
� 𝐸𝐸 𝑇𝑇
0[𝑋𝑋𝑑𝑑]𝑑𝑑𝑑𝑑 = � 𝑋𝑋0𝑒𝑒𝜇𝜇𝑡𝑡𝑑𝑑𝑑𝑑
𝑇𝑇
0=𝑋𝑋0𝜇𝜇
(𝑒𝑒𝜇𝜇𝑇𝑇 − 1) (60)
Appendix B: Proof of Theorem 1 Substituting Eq. 16 into Eq. 32, we have the following equation:
𝑑𝑑𝑑𝑑 = 𝑑𝑑′(𝜇𝜇𝑋𝑋 𝑑𝑑𝑑𝑑 + 𝜎𝜎𝑋𝑋 𝑑𝑑𝑑𝑑) +12𝑑𝑑′′(𝜇𝜇𝑋𝑋𝑑𝑑𝑑𝑑 + 𝜎𝜎𝑋𝑋 𝑑𝑑𝑊𝑊)2 (61)
= 𝜇𝜇𝑋𝑋𝑑𝑑′𝑑𝑑𝑑𝑑 + 𝜎𝜎𝑋𝑋 𝑑𝑑′𝑑𝑑𝑊𝑊 +12𝜇𝜇2𝑋𝑋2𝑑𝑑′′(𝑑𝑑𝑑𝑑)2 + 𝜇𝜇𝜎𝜎𝑋𝑋2𝑑𝑑′′(𝑑𝑑𝑑𝑑)(𝑑𝑑𝑊𝑊) +
12𝜎𝜎2𝑋𝑋2𝑑𝑑′′(𝑑𝑑𝑊𝑊)2 (62)
Taking expectations on both sides to apply some properties of GBM and discarding all terms involving dt to a power higher than 1, we have
E[𝑑𝑑𝑑𝑑] = �𝜇𝜇𝑋𝑋𝑑𝑑′ +12𝑑𝑑′′𝜎𝜎2𝑋𝑋2�dt = [𝑟𝑟𝑑𝑑 − 𝑝𝑝𝑜𝑜(1 − 𝜏𝜏)]𝑑𝑑𝑑𝑑. (63)
Note that the term (𝑑𝑑𝑑𝑑)(𝑑𝑑𝑊𝑊) has magnitude (𝑑𝑑𝑑𝑑)3/2, 𝐸𝐸[𝑑𝑑𝑊𝑊] = 0, 𝐸𝐸[(𝑑𝑑𝑊𝑊)2] = 𝑑𝑑𝑑𝑑, and 𝐸𝐸[𝑑𝑑𝑑𝑑] = 0. After dividing through by 𝑑𝑑𝑑𝑑, we have the Bellman Eq. 33.
390 Advances in Production Engineering & Management 14(3) 2019
