Managing ales Return in ual ales Channel: Common Return versus Cross-Channel Return Analysis Erwin Wioo, atsuhiko Takahashi, atsumi Morikawa, I Nyoman Pujawan, an Bui antosa Abstract This work examines the financial benefit in erforming two kin of sales return (online sales claim because of non conformity scenarios uner ual sales channel structure The first scenario mimics a strategy of meeting such claim through one esignate online facility The secon one reresents a re-fulfillment rocess involving a conventional store as channel counterart (cross channel return so that comlaining customer reference might be accommoate better In aition, two kin of ricing ecision making rocesses are evaluate, namely Bertran scheme for simultaneous rocess an tackelberg leaer scheme for leaer-follower consieration one The result shows that central warehouse an its online facility (leaer refer to aly scenario using tackelberg leaer scheme, while conventional store (follower exeriences better rofit uner first scenario an Bertran scheme However, the first scenario always erforms better than the secon one in the view oint of total channel rofit Further fruitful management insights are also rovie in the analysis section of this aer Inex Terms ual sales channel (C, sales return I INTROCTION ual sales channel (hereinafter is shortene as C has been increasingly confiscating the interest of researchers in the corresoning area A number of works have been one to roose managerial insight in ealing with ecision making within such channel collaboration between conventional istribution structure an internet-base (online orer fulfillment facility ome examle of works in C concetual iea eveloment are [], [10], [1] Later on, some works that base on analytical moels to cature secific C structure characteristics are roose for obtaining closer alicability on its channel s coorination an collaboration [1], [6], [8], [9], [11], [14], [17], [0] Manuscrit receive ecember 7, 009 Erwin Wioo is the corresoning author He is a lecturer in Inustrial Engineering, IT, urabaya, Inonesia Currently, he is ursuing his octoral egree in Hiroshima niversity, Jaan Contact number: 81 8 44 7705 (e-mail: erwinwioo@hiroshima-uacj atsuhiko Takahashi is a rofessor in Grauate chool of Engineering, Hiroshima niversity, Jaan His research interest is rouction system an suly chain management (e-mail: takahasi@hiroshima-uacj atsumi Morikawa is an assistant rofessor in Grauate chool of Engineering, Hiroshima niversity, Jaan He has interest in rouction system management (e-mail: mkatsumi@hiroshima-uacj I Nyoman Pujawan is a rofessor in suly chain managementin Inustrial Engineering, IT, urabaya, Inonesia He is also the Eitor-in-Chief of Oerations an uly Chain Management: an International Journal (htt://journaloscm-forumorg (e-mail: ujawan@ieitsaci Bui antosa is an associate rofessor in Inustrial Engineering, IT, urabaya, Inonesia His main research an teaching area is on otimization an machine learning (e-mail: bui_s@ieitsaci To be more articular, sales return is regare as one rominent henomenon in managing online sales within C environment However, its existence is still uner-reresente for further betterment A limite number of works have been one in ealing with this backwar rouct flow, however they were lai in the layer between central warehouse an its sales outlet None of them consiere the layer between this outlet an its final online customer that offers interlay between conventional store an online facility In our revious work [7], to fill in this research ga, we trie to evaluate the financial benefit of erforming rouct substitution in resoning the online customer claim for rouct non conformity In that aer, we evaluate each layer an total channel rofits by comaring cash back strategy as the benchmark scenario an rouct substitution strategy as the online facility sales return scenario ome beneficial managerial insights have been successfully elicite As for the main result, it was shown that erforming rouct substitution leas to better total channel rofit uner Bertran ecision making rocess However, as our revious work rawback, we i not consier the analysis of customers channel-reference in claiming their non conformity In currently roose work, we comose one interesting research question, ie oes accommoating customer reference to han over their non-conforme rouct irectly through conventional store instea of sen it back to its online facility lea to better financial erformance? In searching for its answer, two scenarios are roose The first one is online-facility sales-return scenario, which consiers online sales facility as the efault otion in returning the unesire roucts The secon one is conventional-store sales-return This challenging scenario accommoates the customer reference to eliver their below-exectation rouct irectly to conventional store for better substitution By comaring C iniviual layer an total channel rofits uner Bertran scheme as well as tackelberg leaer scheme, this aer is able to show whether accommoating customer reference is beneficial or not II LITERATRE REVIEW A coule of years after internet boom erio in the late of 0 th century, C concet of arallel selling between reviously establishe conventional channel equie by newly eicate online facility was introuce by several authors [1] is one of them Then, it was followe by a number of works, like [10] who erforme a case stuy analysis for marketing PC in Euroe, or [] who roose a concetual framework that ensures the successfulness of
emboying conventional channel with internet-base one Comare to our iea, all of those aers were focusing on the creation of C basic concet In contrast, our iea focuses on how to normatively imlement sales return mechanism in C structure We focus on how to analytically rovie some betterment for C structure After concetual erio, then escritive an normative analytical erio took lace The aer by [5] was one of the ioneers in channel structure analysis of C This aer introuce some substitution inventory-scenarios The result showe that this kin of integration erforme financially better Quite similar inventory relate roosition work was unertaken by [17] Their originality was on the assumtion that the system receives stochastic emans an customer may switch channel when a stock-out occurs As for the result, mixe strategy of ual-channel outerforms two-single strategies Furthermore, very interesting iea was resente by [14] They roose ynamic assignment olicies by taking benefits of: 1 traversal eman e-fulfillment an; still in transit inventory The numerical showe that total sales increase about 8% over the otimal static olicy Our roose iea is to rovie extension research by emboying the C structure with sales return function Our work rovies following u action unertaken by C layers after sales take lace In essence, our research creates new research stream in the area of C On the other sie, rouct/sales return aers mainly resie on manufacturer-retailer layer Paer ublishe by [15] can be consiere as one of the ioneers in rouct return work They consiere marketing ersective to examine strategic effect of using incentive for return olicies on retail cometition Other aer, such as [4] roose otimal orer quantity in relation to return hanling otions They evaluate several moels about selling roucts to seconary market, artial reuse, artial recovery, an its cost They obtaine a close form analytical exression for otimal orer quantity Those rouct/sales return works exist in the environment between sulier/warehouse an retailer To be comare to our iea, those works exerience big value but relatively small frequency of return In contrast, within our iea, the sales value is relatively small but the frequency is consierably high In aition, none of them was evote to overcome managerial roblems in C environment that has to eal with online an in-store eman simultaneously Hence, we are keen to state that our roose iea is original III YTEM ECRIPTION & IT MOEL A ystem uner iscussion Fig1 comrehensively shows the basic structure of system uner iscussion This figure reresents a single rouct flow which is available to be sol both through online an conventional (in-store ways Within this C structure, there is one central warehouse which gets rouct suly from a manufacturer This central warehouse then istributes the roucts to several conventional stores ubsequently, this store fulfills in-store eman in its marketing region Fig1 C structure with roose sales return flows Asie of this conventional istribution, as a secial characteristic of C, there is an internet equie facility which is calle online facility to fulfill online customer eman In short, there are two cometing arties in selling the rouct, the first one is central warehouse with its online facility as newly establishe channel, an the secon one is conventional store as the original sales channel Both channels work simultaneously Online customer may ask for rouct substitution in two ways The first way is to re-sen the non-conforme rouct to online facility It will be elaborate in scenario 1 The other way is to carry the unesire rouct irectly to conventional store This will be reresente by scenario In aition, there are several system-assumtions, namely: 1 ales return is allowe only for online sales There is no return service for in-store (conventional sales In return hanling, online facility (in scenario 1 or conventional store (in scenario receives the returne rouct, checks its sales return aroriateness In case it is aroriate, the customer may get rouct substitution Otherwise, the receiving facility or store rejects the claim 3 The meaning of following u action is a set of categorize actions erforme by central warehouse to utilize the value of returne rouct In case of there is: a No efect; rouct is ut back into its inventory b A slight efect; rouct is sol in iscounte rice c A serious efect; rouct is liquiate to salvager 4 Profit comonents to be consiere are total channel rofit, central warehouse an its online facility rofit, as well as conventional store rofit, uner eterministic eman setting for both conventional an online customers B Mathematical moel comonents Prior to etail escrition of our roose mathematical moel, let us consier three ecision variables, namely for wholesale rice ecie by central warehouse, for retailer rice ecie by conventional store, an for retailer rice ecie by online facility Alongsie these ecision variables, a set of inirect variables an arameters are emloye in this work (ee aenix The first comonent to be reare is conventional store eman Basically, we emloy commonly use eman function is β [13], [18], [19], where is the largest amount of in-store eman, an β is the eman elasticity ratio on rice Without losing its generality we assume β 1 To reresent in-store an online emans
simultaneously, this function has to be moifie by inutting into the equation In aition, customer accetance ratio of online rouct arameter, is introuce It exresses the rouct value ecrease because the online rouct is only virtually insecte rior to urchase by customer Hence, the term: is introuce to relace The numerator reresents the eviation between the in-store rice an online rice (customer saving an the enumerator shows customer sacrifice in acceting the ecreasing value of online rouct [16] Accoringly, the new in-store eman function eman is (1 econly, the online eman function is reare This function is efine as the total amount of rouct orere by customer who refers to sho through internet-enable system facility Mathematically, this comonent is obtaine from subtracting (1 from moifie with ajoining to as ( ( The next comonent is hanling cost HC Hanling cost incurre in online facility is simly calculate by multilying hanling cost/rouct H with online sales return r By ( this way, we get Hr (3 A grou of return value comonents which eal with hanling the sales return by customer are also reare 1 Let ki to be the roortion of sales return being ut back in to inventory, then, the amount of sales return in this ( treatment is R I kir Consequently, central ( warehouse gets VRI kir ( c c 0 (4 uose k to be the roortion of sales return being sol in iscounte rice an α is the amount of unsol rouct in secon sales, then, the amount of sales return in ( this treatment is R kr α (5 3 It is assume that equation (5 is a tyical ecreasing eman function of seconary market for sales return sol in Hence, the value of selling the roucts is ( VR ( kr α( c (6 4 Let α reresents the elasticity ratio of secon market eman on iscounte rice It is also assume that: a <, therefore l < l, where l is the salvage ratio for sales return being sol in iscounte rice an l is the salvage ratio for sales return being liquiate b ntil the en of erio, unsol R will be ae to R 5 uose k is the roortion of sales return being liquiate through salvager, then, the amount of sales ( return in this treatment is R k r α (7 6 sing similar logic, the value of selling liquiate ( roucts is VR ( k r α ( c (8 C ales return scenario 1 cenario 1: ales return to online facility This scenario is eveloe base on iea of erforming rouct substitution for online customer return unertaken by online facility Base on reviously reare moel comonents, the objective function for scenario 1 is: max max (,, max (,,,, ( ub HC IC V R V (( ( ( max,, (( ( ( c (1 r( ( r( ( ( ( Hr Ir ( kr ( l c αl ( kr ( l c αll (9 The first term shows the rofit gaine by the conventional store consiering its in-store eman (1 an store rofit margin, ( - The secon term is central warehouse rofit for fulfilling in-store eman base on the its rofit margin ( - c Next, the thir term is online facility rofit by consiering of net online ratio, online eman, an the online facility rofit margin ( - Fourth is the rofit gaine by rouct substitution The negative contribution is given by total hanling cost for sales return consiering hanling cost er unit for sales return H by return ratio r, an online eman Another negative effect is inventory cost for anticiating rouct substitution involving unit inventory cost I The last two items are the obtaine values of erforming following u actions by selling in iscounte rice an through salvager as elaborate in equation (6 an (8 resectively Besies this objective function, the corresoning constraints are also necessary to be comose They are: 1,, > c, rices are higher than its unit cost c, to ensure eman interlay (threshol value 3, to let online facility taking rofit 4, to let conventional store taking rofit 5 (, the highest value of 6 < (, threshol value is reasonable R
7 L E, lower limit of online eman, which is efine as at least there is a ortion (exresse by existence lower limit E L of in-store eman as a minimum amount of online eman 8 E, uer limit of online eman, which has a meaning of at most there is a ortion (exresse by existence uer limit E of in-store eman as a maximum amount of online eman cenario : ales return to conventional store This oosite scenario is base on the iea of giving more freeom to online customer When a claim occurs, the customer is aske to bring the rouct to the conventional store instea of initial online facility server (cross-channel return The benefit are:1customer has oortunity to irectly insect the substitution rouct so that it guarantees the next claim will not haen, an by knowing this olicy, customer has higher accetance ratio on online rouct offere Then, the ifferences to scenario 1 are: 1 Increase in customer accetance ratio on online rouct, from in scenario 1 to ub ub in scenario ( < Inventory cost shift from the resonsibility of online facility (so that IC is introuce to resonsibility of conventional store (hence IC is introuce 3 is roose This notation means the roortion of comensation fee rovie by central warehouse to conventional store for erforming rouct substitution The objective function for this scenario is: max max (,,,, max ((,, max,, r( ub ub IC ( ub HC V ub ((1 ( ( ub ub ( ( ub ub Ir (1 ub ub (1 ub ((1 ( r( ub ( ub (1 ub ( ub ub (1 r( ( ub ub (1 ub ( k r ( l ub ub (1 ( c ( Hr c αl ub ub (1 ub R V ub ( k r ( l c αl l ub ub (10 (1 The constraints remain the same as use in scenario 1 ecision making scheme In rototying the C ecision makers behavior of, game theory can be use as note by [3] Let us consier two kins of ecision making rocess for,, an : 1 Bertran scheme: a simultaneous rocess of etermining the value of ecision variables It means that all managers R of central warehouse, conventional store, an online facility ecie their own unit s rouct rice at the same time The imlication is the search of otimality for total channel rofit can be one only in one hase otimization tackelberg leaer scheme: a sequential rocess in establishing the value of ecision variables This rocess requires three rouns of ecision rocess The first roun is one by the leaer, central warehouse, by releasing arbitrarily In resoning to this value, the follower, conventional store tries to maximize its own rofit by consiering its objective function This secon roun rovies an otimum After knowing this value, the central warehouse together with its online facility can otimize their rofits an yiels otimum an IV NMERICAL ANALYI To evaluate the erformance of two sets of C with return consieration scenario, first, let us have the following ub arameter values: 100; c 10; 07; 075; r 01; H 1; k I 07; k 0; k 01; l 08; l 07; α 003; I 3; an 05 The comlete notation list for setting is rovie in Aenix Then, the objective functions (9 an (10 for scenario 1 an resectively can be solve uner equential Quaratic Programming (QP by using Quasi-Newton algorithm roviing the otimum ecision variables,, an an corresoning rofit,, an as shown in table 1 A cenario an ecision making analysis Table 1 shows that from the leaer (central warehouse an its online facility oint of view, scenario of conventional store sales return scenario (C-R is referable since this scenario gives better rofits (from 8109 to 64818an from 13936 to 19847 uner Bertran an tackelberg leaer schemes resectively The reason is the increase of customer accetance ratio on online rouct 07 to ub 075 This ifference increases online eman significantly, from 940 to 539 uner Bertran scheme an from 639 to 58 uner tackelberg leaer scheme Because of this online sales increase, the leaer exeriences better rofit by emloying scenario In contrast, in the follower (conventional store oint of view, scenario 1 of online facility sales return scenario (F-R is referable since this scenario rovies better rofits (from 116743 to 1685 an from 5909 to 6458 uner Bertran an tackelberg leaer schemes resectively The reason is in scenario 1 is set to be lower than the one in scenario It leas to more conventional store eman in scenario 1, ie 3758 comare to 599 uner Bertran scheme an 553 comare to 1064 uner tackelberg leaer scheme This results in better rofit of for conventional store as the follower
Table 1 Otimization result for,, an an their corresoning rofit,, an r I cenario cheme 1:F-R Bertran 01 3 07 n/a 110000 37117 558431 93951 375803 1685186 810915 19663101 tackelberg 01 3 07 n/a 44767 476596 700000 6389 5530 645779 1393595 18846374 :C-R Bertran 01 3 075 05 110000 364667 549695 53889 59888 1167465 6481800 18156064 tackelberg 01 3 075 05 45141 476596 700000 58155 106384 590943 1984661 15575604 Consiering the total rofit of the whole channel, it is shown that scenario 1 outerforms scenario in both Bertran an tackelberg schemes (196631 against 181560 an 188464 against 155756 resectively This inicates that the increase in as the strong contributor of scenario 1 is greater than the ecrease in as the weak one, hence, the scenario 1 exeriences ositive total rofit margin The reason for high increase in is the oosite fact ( ecrease of the reason for ecreasing as exlaine before A ensitivity analysis The first sensitivity analysis is to examine the behavior of leaer an follower rofit on the change of sales return ratio, r This arameter is chosen because r is the reresentation of sales return consieration in this aer However, the result in Fig shows insignificant shifts on the observe rofits 1500 C-O rofit on r Moreover, when we combine several values of r with a set of value shifts, a number of interesting sensitivity results are gaine is a ercentage which reflects a art of online-sales rofit er unit rovie by online facility to conventional store as a comensation for erforming rouct substitution through conventional store Nonetheless, since is not involve in scenario 1, this - r combinatory sensitivity analysis is one exclusively in scenario Fig3 shows the sensitivity of central warehouse an online facility rofit to the change of value uner 3 values of r When increases, will ecrease uner both ecision making rocess schemes In Bertran scheme (Fig 3(a, the reason of this ecrease is there is 1-sum effect for as inicate in the objective function (11 for n an 5 th items This effect results in the stable value of 3 ecision variables When the total rofit is istribute to leaer an follower, increase gives benefit to, then, the leaer exeriences ecrease In tackelberg leaer scheme (Fig 3(b, the reason is that when increases, otimization results will give higher This situation leas to less ecrease to (from aroun 130 to 195 than the one in Bertran scheme (from aroun 670 to 560 C-O rofit 1000 500 700 C-O rofit on uner Bertran tore rofit 0 004 006 008 01 01 014 016 1-Bertran 1-tackelberg r -Bertran -tackelberg (a 000 1500 1000 500 0 tore rofit on r 004 006 008 01 01 014 016 1-Bertran 1-tackelberg r -Bertran -tackelberg (b Fig an sensitivity to r C-O rofit C-O rofit 650 600 550 500 130 1315 1310 1305 1300 195 190 r 01 r 0 r 03 0 03 04 05 06 07 08 (a C-O rofit on uner tackelberg r01 r0 r03 0 03 04 05 06 07 08 (b Fig3 sensitivity to r
Fig 4 is about the oosite result of rofit shift comare to the ones in figure 3 This figure illustrates the increase of follower rofit when increases uner several value of r This is the oosite effect burene by conventional store Consequently, the inverse reason, comare to sensitivity analysis is alie tore rofit tore rofit 180 160 140 10 100 1180 1160 1140 65 60 55 50 45 40 35 30 tore rofit on uner Bertran r01 r0 r03 0 03 04 05 06 07 08 (a tore rofit on uner tackelberg r01 r0 r03 0 03 04 05 06 07 08 (b Fig4 sensitivity to r Fig 5 reresents the sensitivity of channel total rofit to the shift of value uner 3 values of r In Bertran scheme, stable total rofits are shown The reason is 1-sum effect This fact leas to the stable otimization result of ecision variable which is eventually gives the same value even when an r values shift u or own In tackelberg leaer shceme, increase in yiels higher value otimization results This kin of increase yiels the lower value of channel total rofit al rofit 184 18 180 1818 1816 1814 al rofit on uner Bertran r01 r0 r03 0 03 04 05 06 07 08 (a al rofit al rofit on uner tackelberg 1560 1558 1556 1554 155 1550 1548 1546 r01 r0 r03 0 03 04 05 06 07 08 (b Fig5 sensitivity to r V CONCLION A number of revious works have contribute some imortant insights in the area of C Nonetheless, for the time being, online sales return management is still uner-reresente Our current work gives one significant contribution on briging this lack Our research rooses two ifferent scenarios in ortraying customer references in ealing with returning non-conforme online urchase In aition, besies aying attention to customer sie, our roose moel also ortrays ecision making rocess for ricing which is erforme by C managers We evaluate formal-form Bertran an strategic-form tackelberg leaer schemes erformance in fining the equilibrium solutions Base on our analysis, the imortant finings are: 1 Central warehouse an its online facility refer to aly conventional store (cross channel sales return scenario In contrast, conventional store exeriences better financial erformance uner online facility sales return scenario 3 However, in the whole channel oint of view, online facility sales return scenario outerforms conventional store one In resoning to the research question, we are keen to state that allowing conventional store to erform rouct substitution is not financially beneficial when total channel rofit is consiere Nonetheless, when the coorination an collaboration is still far away from reality, the leaer an follower uner the reresente system structure coul take benefit by referring to our quantitative measures There are some ossibilities regaring the future works for our research continuations In rototying real situation, it will be more interesting to evelo a moel which accommoate stochastic or uncertain eman situation In aition, in treating the returne rouct by online customer, some new iea on reselling strategies will be helful for increasing entire channel rofitability
APPENIX ecision variable wholesale rice (ecision variable conventional store rice (ecision variable online rice (ecision variable Function of ecision variable in-store eman (f(, customer online eman (f(, R online customer eman after return (f(,, r HC hanling cost incurre in central warehouse, online facility or store for unertaking sales return IC rouct substitution inventory cost in online facility IC rouct substitution inventory cost in convt store reselling rice in iscounte rice (f(, l Liquiation rice (f(, l R I number of sales return being ut back into inventory V RI value of sales return being ut back into inventory R number of sales return sol in iscounte rice V R value of sales return being sol in iscounte rice R number of sales return liquiate through salvager V R value of sales return liquiate through salvager rofit for central warehouse conventional store rofit for unertaking conventional sales only conventional store rofit for unertaking conventional sales an sales return online facility rofit ub online facility rofit gaine by erforming sales return substitution ub central warehouse rofit gaine by erforming sales return substitution through conventional store ub conventional store rofit gaine by erforming sales return substitution through conventional store rofit of central warehouse an online facility rofit for the whole suly chain ystem arameters r k I k k maximum value of when is set to be in lower limit value customer accetance ratio of an online rouct comare to the conventional one roortion of sales return from sales return ration of being ut back into inventory sales return ratio of being sol in iscounte rice sales return ratio of liquiation through salvager α unsolness ratio of n market sales to iscounte rice l salvage ratio for sales return being sol in iscounte rice l salvage ratio for sales return being liquiate through salvager E L existence lower limit level for online eman E existence uer limit level for online eman roortion of comensation fee rovie to conventional store for erforming substitution Cost arameters c unit cost in sulying a single rouct H hanling cost er unit for sales return I inventory cost er unit for sales return substitution REFERENCE [1] A A Tsay, an N Agrawal, "Channel conflict an coorination in the e-commerce age," Prouction an Oerations Management, 13, 004, 93-110 [] C teinfiel, H Bouwman, an T Aelaar, "The ynamics of click-an-mortar electronic commerce: oortunities an management strategies," International Journal of Electronic Commerce, 7, 00, 93-119 [3] Fuenberg, an J Tirole, Game Theory, The MIT Press, Massachusetts, 1991 [4] Vlachos, an R ekker, "Return hanling otions an orer quantities for single erio roucts," Euroean Journal of Oerational Research, 151, 003, 38-5 [5] E Benoly, "Integrate inventory ooling for firms servicing both on-line an store eman," Comuters & Oerations Research, 31, 004, 1465-1480 [6] E Benoly, Blocher, M Bretthauer, an M A Venkataramanan, "ervice an cost benefits through clicks-an-mortar integration: Imlications for the centralization/ecentralization ebate," Euroean Journal of Oerational Research, 180, 007, 46-44 [7] E Wioo, Takahashi, Morikawa, IN Pujawan, an B antosa, "Managing sales return in ual sales channel: An analysis of its rouct substitution," Proceeing of Asia Pacific Inustrial Engineering an Management ystem Conference, itakyushu, Jaan, ecember 009, 83-835 [8] G Z Cai, G Zhang, an M Zhang, "Game theoretical ersectives on ual-channel suly chain cometition with rice iscounts an ricing schemes," International Journal of Prouction Economics, 117, 009, 80-96 [9] -Y Chen, M aya, an O Ozer, "ual sales channel management with service cometition," Manufacturing ervice Oerations Management, 10, 008, 654-675 [10] M Gabrielsson, V H M iralani, an R Luostarinen, "Multile channel strategies in the euroean ersonal comuter inustry," Journal of International Marketing, 10, 00, 73-95 [11] N umar, an Ruan, R, "On manufacturers comlementing the traitional retail channel with a irect online channel," Quantitative Marketing an Economics, 4, 006, 89-33 [1] N P Mols, "ual channels of istribution: a transaction cost analysis an roositions," International Review of Retail, istribution & Consumer Research, 10, 000, 7-46 [13] R Yan, "Profit sharing an firm erformance in the manufacturer-retailer ual-channel suly chain," Electronic Commerce Research 8, 008, 155-17 [14] Mahar, M Bretthauer, an M A Venkataramanan, "The value of virtual ooling in ual sales channel suly chains," Euroean Journal of Oerational Research, 19, 009, 561-575 [15] V Pamanabhan, an I P L Png, "Manufacturer's return olicies an retail cometition," Marketing cience, 16, 1997, 81-94 [16] W-y Chiang, Chhaje, an J Hess, "irect marketing, inirect rofits: a strategic analysis of ual-channel suly-chain esign," Management cience, 49, 003, 1-0 [17] W-y Chiang, an G E Monahan, "Managing inventories in a two-echelon ual-channel suly chain," Euroean Journal of Oerational Research, 16, 005, 35-341 [18] W Zhao, an Y Wang, "Coorination of joint ricing-rouction ecisions in a suly chain," IIE Transactions 34, 00, 701-715 [19] X Chen, an imchi-levi, "Coorinating inventory control an ricing strategies with ranom eman an fixe orering cost: The finite horizon case," Oerations Research 5, 004, 887-896 [0] X Yue, an J Liu, "eman forecast sharing in a ual-channel suly chain," Euroean Journal of Oerational Research, 174, 006, 646-667