使用供应商管理库存提高预测【外文翻译】.doc

1、外文翻译原文USEVMITOIMPROVEFORECASTINGIVEALWAYSBEENABIGBELIEVERTHATUSINGDOWNSTREAMSUPPLYCHAININFORMATIONCANIMPROVEDEMANDFORECASTINGANDPLANNINGIVEWRITTENSEVERALCOLUMNSTHATDISCUSSEDVARIOUSASPECTSOFTHISDURINGMY12YEARTENUREINWRITINGFORTHEJOURNALOFBUSINESSFORECASTINGJBFTHESEDISCUSSEDTHEPOTENTIALUSEOFDOWNSTREAM

2、DATAINCLUDINGASIVEDEFINEDTHEMCUSTOMERSORDERSANDREPLENISHMENTFORECASTS,RETAILERWAREHOUSEWITHDRAWALS,POINTOFSALEPOSDATAANDFORECASTSVENDORMANAGEDINVENTORYVMI,ANDCOLLABORATIVE,PLANNING,FORECASTINGANDREPLENISHMENTCPFRINFORMATIONINTHESEARTICLESIPROPOSEDTHATONESHOULDBEABLETOIMPROVEFORECASTINGBYLEVERAGINGTH

3、ESETYPESOFINFORMATIONSINCETHEYAREGENERATEDCLOSERTOTHECONSUMERASSUCH,DEMANDCHANGESOUGHTTOBEDETECTEDSOONERANDBESUBJECTTOLESSDISTORTIONFROMTHE“BULLWHIP“EFFECTTHATADDSMORENOISETOCONSUMPTIONDEMANDASITMOVESUPTHESUPPLYCHAINGENERALLY,IVEBEENPREACHINGTHATTHESEDOWNSTREAMDATASOURCESOUGHTTOPROVIDEMUCHBETTER“DEM

4、ANDSIGNALS“THANAMANUFACTURERSORDERSANDSHIPMENTSVMIDATAOUGHTTOBEUSEFULALMOSTSEVENYEARSAGOIWROTEANARTICLEINTHEWINTER2001/2002ISSUEOFTHEJBF,TITLED“VMICANBEGOODFORYOURFORECASTINGHEALTH“INITITALKEDABOUTTHEFACTTHATTHEHISTORYOFVMIHADBEENMIXEDBECAUSEOFTENCUSTOMERS,SUCHASRETAILERS,HADBEENCOERCINGTHEIRSUPPLIE

5、RSTOMANAGETHEIRINVENTORIESFORTHEMFORFREE,WITHLITTLEPERCEIVEDGAINTOTHESUPPLIERSHOWEVER,IMAINTAINEDTHATWHENDONEWELL,VMIPROGRAMSCANIMPROVEINTERCOMPANYSUPPLYCHAINOPERATIONSBECAUSETHEYHELPASUPPLIERBETTERCOORDINATEPRODUCTFLOWTOCUSTOMERSINTHISREGARDTHEBENEFITSOFVMIINCLUDEENSURINGTHATAMANUFACTURINGCUSTOMERS

6、PRODUCTIONOPERATIONSARENOTDISRUPTEDBYMATERIALSHORTAGESADISTRIBUTORORRETAILERCUSTOMERISNOTCAUGHTSHORTONPRODUCTAVAILABILITYNEITHERACUSTOMERNORITSSUPPLIERISHOLDINGMOREINVENTORYTHANABSOLUTELYNECESSARYTOMEETCONSUMERDEMANDWHILETHESEBENEFITSAREWORTHY,THEYDONOTINCLUDEWHATIBELIEVETOBETHEMOSTIMPORTANTBENEFITO

7、FVMIIMPROVINGTHEPLANNINGPROCESSESTHATINDIRECTLYHELPLEADTOTHESEBENEFITS,ASWELLASOTHERSITSEEMEDTOMETHATIFASUPPLIERISMANAGINGACUSTOMERSINVENTORIESANDPLACINGINVENTORYREPLENISHMENTORDERSFORIT,THESUPPLIERSHOULDHAVEAGOODIDEAWHATTHATCUSTOMERWILLNEEDOVERTHENEXTSEVERALWEEKSAFTERALL,TOPLACETHESEORDERS,ASUPPLIE

8、RSVMIACCOUNTTEAMHASTOBEWORKINGCLOSELYWITHTHECUSTOMERTOENSUREINVENTORYISAVAILABLEFORREGULAR“TURN“BUSINESS,ASWELLFORANYINCREMENTALPROMOTIONALBUSINESSTHATISGENERATEDFROMPROMOTIONSTHATTHECUSTOMERWILLBERUNNINGINTHEFUTUREDISJOINTPROCESSESANDSYSTEMSHAMPERVMIUSEIBELIEVESUPPLIERFORECASTERSHAVENOTBEENABLETOLE

9、VERAGEVMITOIMPROVETHEIRFORECASTINGBECAUSEOFTHEDEARTHOFVMIDATAAVAILABLETOTHEMTHISISPRIMARILYDUETOORGANIZATIONALASWELLASSYSTEMCONSTRAINTSREGARDINGTHEFORMERVMIPROGRAMSARETYPICALLYSUPPORTEDBYASUPPLIERSACCOUNTTEAM,COMPRISEDOFCUSTOMERFACINGSALESANDLOGISTICSPERSONNELTHETEAMISCHARTEREDWITHMANAGINGACUSTOMERS

10、INVENTORIESANDISPRIMARILYRESPONSIBLEFORENSURINGPRODUCTSUPPLYCUSTOMERSTOCKINGLOCATIONSARETREATEDBYTHETEAMASJUSTANOTHERLOCATIONWHEREINVENTORIESNEEDTOBEMANAGEDWHENINITIATINGAVMIPROCESSWITHACUSTOMER,ASUPPLIERSMAINFOCUSISEXTERNALLYORIENTEDTHATIS,ANACCOUNTTEAMWILLMAINLYBECONCERNEDWITHGETTINGVMITOWORKWELLF

11、ORTHECUSTOMERFROMATRANSACTIONALVIEWPOINT,ANDHAVEMINIMALPLANSTOPOSSIBLYTIEVMITOTHESUPPLIERSINTERNALPLANNINGPROCESSESATALATERDATETHUS,VMIPROCESSESANDSYSTEMSARESETUPTOFOCUSSOLELYONEXECUTION,NOTPLANNINGPROCESSESOVERTIME,BUSYCUSTOMERFACINGACCOUNTTEAMSDONOTCOMMUNICATEACUSTOMERSREPLENISHMENTNEEDSTOTHEIRFOR

12、ECASTINGANDPLANNINGORGANIZATIONSREGARDINGTHESYSTEMCONSTRAINTS,SINCETHEVMISYSTEMSAREINITIALLYSETUPTOBEINDEPENDENTOFTHEOTHERSUPPLIERSYSTEMS,OVERTIME,ANYPLANSTOCONNECTTHEMUSUALLYFALLBYTHEWAYSIDEASTHEITDEPARTMENTMOVESONTOOTHERPRIORITYPROJECTSITBECOMESHARDERANDHARDERTOINTEGRATETHEVMISYSTEMSINTOTHEINTERNA

13、LPLANNINGSYSTEMSEVENTUALLYITWINDSUPTOBETHECASETHATUNLESSSOMEONEBECOMESPASSIONATEABOUTABELIEFTHATVMIDATACANYIELDSIGNIFICANTBENEFITS,NOONEWANTSTOBOTHERTODOTHEBUSINESSCASEANALYSISNEEDEDTOSTARTANINTEGRATIONPROJECTTHISLEAVESFORECASTERSANDPLANNERSWITHLITTLEINFORMATIONABOUTTHECUSTOMERSFUTURENEEDS,EXCEPTTHO

14、SETHEYCANGLEANFROMHISTORICAL“SELLIN“SALESTOTHATCUSTOMERACASESTUDYINUSINGVMISOWHILEIVEBEENDISCUSSINGHOWTHEUSEOFVMIINFORMATIONOUGHTTOBEABLETOIMPROVEDEMANDFORECASTINGANDPLANNINGSINCEIWROTETHATARTICLEALMOSTSEVENYEARSAGO,IVEBEENATALOSSTOFINDACOMPANYTHATHASPROVEDITTOBETHECASE,OREVENTRIEDTODOSOMOSTFORECAST

15、ERSIVEDISCUSSEDITWITHDIDNTTHINKITWASWORTHTHEEFFORTTHATWASUNTILTHISPASTACADEMICYEARWHENALARGECONSUMERPRODUCTSCOMPANYDECIDEDTOSPONSORANMITGRADUATESTUDENTTOLOOKINTOITFORTHEM,ASAPARTOFHISTHESISRESEARCHIADVISEDTHISSTUDENTANDIAMPLEASEDTOREPORTTHATHISFINDINGSCONVINCEDTHESPONSORTHATTHEREWERESIGNIFICANTGAINS

16、TOBEGOTTENFROMLEVERAGINGVMIINSHORTTERMFORECASTINGFORUPTOFOURWEEKSOUTXIHANGKOUSTHESISISREFERENCEDBELOWFORMOREDETAILTHEANALYSISTHESTUDENTCONDUCTEDUSEDWEEKLYHISTORICALDEMANDDATAANDFORECASTSFORTWOPRODUCTLINES,ADRYANDATEMPERATURESENSITIVEPRODUCTLINEDEMANDDATAWASCOLLECTEDFOR29PRODUCTSSOLDTOTWOGROCERYRETAI

17、LERSTHATDIFFEREDINTERMSOFTHEAMOUNTOFPROMOTIONSTHEYCONDUCTEDTHEFIRSTCUSTOMERA,EXTENSIVELYLEVERAGESWEEKLYPROMOTIONSTOMARKETMERCHANDIZE,WHILETHESECOND,CUSTOMERB,ISANEVERYDAYLOWPRICEMARKETERTHATRUNSFEWPROMOTIONSTHESEWERESPECIFICALLYCHOSENTOASSESSWHETHERVMIDATAWASMOREVALUABLEFORPROMOTIONALFORECASTING,THA

18、NTOFORECASTINGFORREGULAR“TURN“BUSINESSFOREACHCUSTOMERFOUR4TYPESOFDATAWERECOLLECTED1DEMANDPLANNINGFORECASTSGENERATEDBYTHEDEMANDPLANNINGDEPARTMENTOFTHESUPPLIER2SHIPMENTSTOTHEGROCERFROMTHESUPPLIER3VMIFORECASTSOFTHEGROCERSSTOREDEMANDGENERATEDBYTHEGROCER4DCINVENTORIESINAGROCERSDISTRIBUTIONCENTERSDCSUSING

19、THISDATATOGOBACKINTOHISTORY,AGROCERSWEEKLYDCREPLENISHMENTPLANRPFOREACHPRODUCTWASESTIMATEDFORONETOFOURWEEKSOUTTHEPLANSWEREGENERATEDUSINGTHREEINPUTSSTARTINGDCINVENTORIES,VMIFORECASTS,ANDTHEVMIPLANNERSINVENTORYREPLENISHMENTPOLICIESFORADCTHECRUXOFTHEANALYSISINVOLVEDCOMPARINGTHEFORECASTACCURACIESOFUSINGT

20、HEDEMANDPLANNINGDPFORECASTANDTHEREPLENISHMENTPLANRPTOFORECASTACTUALSHIPMENTSTOEACHGROCERSDCFORFOURWEEKSOUTMEANABSOLUTEPERCENTERRORMAPEWASUSEDASTHEMEASUREOFFORECASTACCURACYTABLE1REPRESENTSSUMMARYFINDINGSFROMTHESTUDENTSTHESISITDEPICTSWEEKLYFORECASTACCURACYIE,MAPEFROMONETOFOURWEEKSOUTFOREACHCUSTOMER,AN

21、DISBROKENDOWNFORCUSTOMERABYPROMOTIONALVERSUSNONPROMOTIONALWEEKSASANASIDE,NOTETHATWHILETHEMOPESMAYSEEMHIGH,RECALLTHATFORECASTERRORHEREISMEASUREDBYWEEK,BYSKU,ANDBYACUSTOMERSDCIE,SHIPTOLOCATION,ANDTHEREARENOBENCHMARKSAROUNDFORTHISGRANULARITYOFFORECASTING,THATIKNOWOFASCANBENOTEDFORCUSTOMERA,THEGROCERTHA

22、TPROMOTESWEEKLY,THERPFORECASTEXPERIENCESLESSTHANHALFTHEWEEKLYERROROFTHEDPFORECASTFORALLWEEKSONE,TWO,THREE,ANDFOURWEEKSOUTINADDITION,WHILETHISHOLDSFORBOTHPROMOTIONALASWELLASNONPROMOTIONALWEEKS,THERPISEVENMOREACCURATETHANTHEDPFORECASTDURINGPROMOTIONALWEEKS,WITHMAPESOVER66LOWERSINCETHEGROCERPROMOTESHEA

23、VILY,WEEKLYVARIATIONINITSSHIPMENTSISVERYHIGHASNOTEDBYTHEMAGNITUDEOFITSFORECASTERRORSCOMPAREDTOCUSTOMERBSSOTHERPISEVENMOREACCURATETHANTHEDPFORECASTBECAUSETHEVMIACCOUNTTEAMSCANINCORPORATEMOREREALISTICVMIIMPACTSINTOTHEFORECASTSDURINGPROMOTIONALWEEKS,WHICHTHECORPORATEDEMANDPLANNERSLACKKNOWLEDGEOFSIMILAR

24、LYFORCUSTOMERB,THEGROCERTHATISANEVERYDAYLOWPRICINGMARKETER,THERPFORECASTEXPERIENCESSIGNIFICANTLYLESSWEEKLYERRORTHANTHEDPFORECASTFORTWO,THREE,ANDFOURWEEKSOUTTHERPFORECASTMAPESARE37,42,AND47LOWER,RESPECTIVELYALSONOTETHATSINCETHISGROCERDOESNOTPROMOTEHEAVILY,MAPESAREALOTLOWERTHANCUSTOMERASSOSUMMARIZINGT

25、HESTUDENTSFINDINGS,THEANALYSISSHOWEDTHATUSINGTHERPFORECASTRATHERTHANTHEDPFORECASTTOFORECASTSHIPMENTSTOGROCERSWOULDSUBSTANTIALLYINCREASETHEFORECASTACCURACIESFORBOTHREGULAR“TURN“ANDPROMOTIONALDEMAND,MORESOFORTHELATTERTHEFINDINGSSPURREDTHESPONSORINTODOINGAMORETHOROUGHANALYSISTOHELPJUSTIFYSTARTINGAPROJE

26、CTTOINTEGRATETHEIRVMISYSTEMSWITHTHEIRINTERNALFORECASTINGANDPLANNINGSYSTEMSTHEFINDINGSALSOHAPPILYSUPPORTAPREMISEIVEHELDFORALMOSTSEVENYEARS“VMICANBEGOODFORYOURFORECASTINGHEALTH“YOUMIGHTWANTTODOYOUROWNANALYSISTOPROVEITANDADDMOREREASONTOHELPYOURCUSTOMERSMANAGETHEIRINVENTORIESOFYOURCOMPANYSPRODUCTSREFERE

27、NCEXIHANGKOU,“VENDORMANAGEDINVENTORYFORECASTOPTIMIZATIONANDINTEGRATION,“MITMASTERSOFENGINEERINGINLOGISTICSTHESIS,JUNE2008SOURCELARRYLAPIDE，USEVMITOIMPROVEFORECASTING，THEJOURNALOFBUSSINESSFORECASTING，2008译文使用供应商管理库存提高预测我一直很相信利用下游供应链的信息可以提高需求预测和规划。我写了几个专栏，讨论了我12年任期中为杂志的商业预测（JBF）写本的各个方面。这些讨论包括下游数据的潜在使用

28、（因为我已经定义它们）客户订单的预测和补货、零售商仓库提取、定点、销售终端（POS）数据和预测，供应商管理库存（VMI）和协同规划、预测与补货（CPFR）的信息。在这些文章中，我认为能够提高利用这些信息类型，因为它们生成的预测更接近消费者。因此，需求的变化应该早被检测，并接受来自“牛鞭效应”较少的失真，更多的顾客消费需求的声音是因为移动了供应链。一般来说，我一直在鼓吹这些下游数据源应该提供比制造商的订单和出货要好得多的“需求信号”。供应商管理库存数据应该是有用的差不多七年前我写了一篇文章，出版在2001/2002冬季的JBF上，题目是“供应商管理库存为预测健康提供好处”。在文章中,我谈到了供应

29、商管理库存的历史事实被混淆。因为通常情况下，顾客如零售商、被他们的供应商强迫免费管理他们的库存,很少有供应商认为增益。但是我认为只要做得好，供应商管理库存方案可以改善公司间的供应链运作，因为它们可以帮助供应商更好地协调产品流向客户。在这方面的VMI的好处包括确保一个制造业客户的生产经营不受材料短缺的影响。分销商或零售商客户对产品的可用性没有不满。无论是客户还是其供应商没有必要持有更多的库存，以满足消费者的需求。虽然这些好处是值得的，我相信VMI的最重要的好处是改善规划过程，间接地帮助导致这些好处以及其他更多。在我看来，如果一个供应商管理客户的存货，并管理它的库存补货订单，供应商对客户将需要在今

30、后几个星期应该有一个好建议。毕竟，使用这些订单，供应商的VMI帐户团队与客户密切合作,确保库存可供定期“转”业务,对于任何增加的促销业务是由提升客户将在今后的发展前景。妨碍流程和系统的最优供应商管理库存使用我相信供应商预测不能使VMI得到改善，因为缺乏VMI的数据提供给他们的预测。这主要是由于组织和制度约束。对于前者。支持VMI的程序通常由供应商的客户团队面向客户的销售和后勤人员组成。该小组是包车管理一个客户的存货和负责保证产品供应。顾客库存位置将作为存货需要被处理的另一个地点。当与客户启动一个VMI流程，供应商的主要目标是便利化。也就是说，一个客户团队将主要关注从更多相互作用的观点为使客户的

31、VMI做的更好。并具有最小的计划将在之后配合供应商管理库存对供应商的内部规划过程。因此，VMI的流程和系统都设置为重点完全执行，而不是规划进程。随着时间的推移，他们的预测和规划组织使得顾客的补货需要不再忙于与客户团队的面对面沟通。对于系统的限制，自从供应商管理库存系统的初始设置与其他供应商系统独立开来，随着时间的推移,任何把它们连接在一起的计划通常失败在IT部门移到其他重点项目建设。它变得越来越难融入供应商管理库存内在规划系统的整合。最终,会出现的情况通常是这样除非有人有坚定的信念相信供应商管理库存数据对产量有显著的好处，没有人愿意费心启动一个一体化工程来做案分析。这使得有关客户未来的需求预测

32、和规划的资料很少，除了那些可以收集的“抛售”客户的销售历史。在使用的VMI案例研究虽然我一直在讨论怎样使用供应商管理库存信息能提高需求预测和规划因为近7年前我写了那篇文章,但我一直无法找到一个公司证明它的情况,也无法试图这么做。大多数我讨论过的预言不被认为是值得付出努力的。直到去年，一个大型日用消费产品公司决定赞助麻省理工学院的研究生，作为他论文研究的一部分专题。我建议这个学生并很高兴地向大家报告，赞助商相信他的发现，即从最多VMI的四个星期内利用中短期预报，得到明显的进步。（西航KOU的论文，更多引用更细节见下文）。学生每周进行历史需求数据的分析和预测包括两条产品线干燥的和对温度敏感度的。需

33、求数据是从两个开展不同促销数量的杂货店的29个产品中收集的。第一，A客户广泛利用市场的每周促销活动使商品出口；而第二，客户B利用每日低价，的营销活动进行少量促销。这些都是评估VMI的数据预测比定期“转”的经营业务是否有价值的特别选择。对于每个客户的四种收集数据的类型1、需求计划预测（由供应商处产生的需求规划）2、出货量（从供应商到杂货店）3、VMI的预测（又杂货店产生的杂货商需求）4、直流存货（杂货店的分销中心分布式控制系统）使用这些数据回到历史，一个杂货店内每个产品的每周直流补充计划（RP）估计在1至4星期内。这些计划有三种输入法产生直流库存产生，VMI的预测，规划师在VMI的库存直流补充政

34、策。参与分析的关键包括比较需求规划（DP）和补货计划（RP）的预测精度，四星期中对每个杂货店的直流实际出货量。表1是学生论文中得出的结果摘要。它描述了每一位客户一至四个星期每周预报准确率，并细分客户A为推广周与非推广周。表1DP和RP前景预测的MAPES（顾客A与顾客B的比较）需求计划预测补货计划预测产品直流周的MAPE1周2周3周4周1周2周3周4周顾客A20521420819097808688非促销周16417516816187727684促销周456452458368154132148114顾客B5973818857464747参考表来自XIHANGKOU的“供应商管理库存的优化和整合”

35、麻省理工学院物流工程的硕士论文。2008年6月。由于杂货店促销可用于客户A，周刊指出，RP预测的经验误差不到DP每周预测误差的一半一，二，三，四个星期以外。此外，虽然这两个推广与非推广周同时举行，RP预测甚至比宣传周期间的DP预测更准确准确，MAPES大概降低66。由于杂货商促销资金大，与客户B的大小相比，每周出货量的预测误差幅度非常大。因此，RP的预测准确度甚至超过了DP预测，因为在宣传周期间，VMI的客户团队可以纳入更切合实际情况的VMI的预测影响，该公司的需求规划缺乏知识。同样，对于客户B，杂货店开始每日低价，以RP的预测经验明显比DP的预测误差减少两到四个星期内。RP的MAPES预测分

36、别降低37、42和47。还要注意，因为杂货商没有更大的促销活动，会比客户A的MAPES低很多。总结了学生的调查结果，分析表明，使用RP的预测，而不是DP的预测，更能准确预测杂货店的出货量大幅增加，促销需求预测的准确度也大幅增加，后者更是这样。这一发现促使赞助商做更深入的分析，有助于证明开始一个项目，从内部预测和规划系统方面对供应商管理库存系统进行整合。调查结果支持我提出的将近7年举行的假设“VMI对预测大有好处”。你可能想自己做分析来证明这一点，并添加更多的理由来帮助你的客户管理贵公司的产品库存。参考XIHANGKOU供应商管理库存预测的优化整合麻省理工学院物流工程的硕士论文，2008年6月出处拉里拉普德，使用供应商管理库存提高预测，商务预测杂志，2008