MIT Sloan School of Management
MIT Sloan School Working Paper 4656-07
8/2/2007
Inventory Management of a Fast-Fashion Retail Network
Felipe Caro, Jeremie Gallien
© 2007 Felipe Caro, Jeremie Gallien
All rights rerved. Short ctions of text, not to exceed two paragraphs, may be quoted without explicit permission, provided that full credit including © notice is given to the source.语文学习
英语翻译This paper also can be downloaded without charge from the
Social Science Rearch Network Electronic Paper Collection:
cpi是什么意思>spoil
Inventory Management of a Fast-Fashion Retail Network
Felipe Caro∗J´e r´e mie Gallien†
August2,2007
Abstract
Working in collaboration with Spain-bad retailer Zara,we address the problem of dis-tributing over time a limited amount of inventory across all the stores in a fast-fashion retail
network.Challenges specific to that environment include very short product life-cycles,and
store policies whereby a reference is removed from display whenever one of its key sizes stocks
out.Wefirst formulate and analyze a stochastic model predicting the sales of a reference in a
single store during a replenishment period as a function of demand forecasts,the inventory of
each size initially available and the store inventory management policy just stated.Secondly,
we formulate a mixed-integer program embedding a piece-wi linear approximation of thefirst
model applied to every store in the network and allowing to compute store shipment quantities
maximizing overall predicted sales,subject to inventory availability and other constraints.We
beatingreport the implementation of this optimization model by Zara to support its inventory distribu-
tion process,and the ensuing controlledfield experiment performed to asss the impact of that
model relative to the prior procedure ud to determine weekly shipment quantities.The results
of that experiment suggest that the new allocation process tested increas sales,reduces tran-
shipments,and increas the proportion of time that an important category of Zara’s products
spends on display.
handles1.Introduction
The recent impressivefinancial performance of the spanish group Inditex(its2005income-to-sales ratio of12%was among the highest in the retail industry)shows the promi of the fast-fashion model adopted by itsflagship brand Zara,but also other retailers that include Sweden-bad H&M, Japan-bad World Co.,and Spain-bad Mango.The key defining feature of this new retail model lies in novel product development process and supply chain architectures relying more heavily on local cutting,dyeing and/or wing,in contrast with the traditional outsourcing of the activities from developing countries.While such local production obviously increas labor costs,it also provides greater supplyflexibility and market responsiveness.Indeed,fast-fashion retailers offer in each ason a larger number of references produced in smaller ries,continuously changing the assortment of products displayed in their stores(Ghemawat and Nueno2003report that Zara offers on average11,000references in a given ason,compared to only2,000−4,000items for key competitors)in order to increa their appeal to customers(a top Zara executive quoted in ∗UCLA Anderson School of Management,Los Angeles,CA90095,fcaro@anderson.ucla.edu
†MIT Sloan School of Management,Cambridge,MA02142,jgallien@mit.edu
Fraiman et al.2002states that Zara customers in Spain make on average17store visits per year). In addition,products offered by fast fashion retailers during the lling ason may result from design c
hanges decided after the ason has started as a respon to actual sales information,which considerably eas the matching of supply with demand(Ghemawat and Nueno2003report that only15−20%of Zara’s sales are typically generated at marked-down prices compared with30−40% for most of its European peers,with an average percentage discount estimated at roughly half of the30%average for competing European apparel retailers).
The fast-fashion retail model just described gives ri to veral important and novel operational challenges.The work to be described here,which has been conducted in collaboration with Zara, address in particular the problem of distributing over time a limited amount of merchandi inventory between all the stores in a retail network.Note that while the general problem just stated is not specific to fast-fashion retailing,we believe that veral features which are specific to this retail paradigm(short product life cycles,store inventory display policies)do justify new approaches.Indeed,Zara’s interest in this area of collaboration was motivated by its desire to improve the inventory distribution process it was using at the beginning of our interaction for deciding the quantity of each reference to be included in the weekly shipment from the warehou to each store(e Figure1(a)for an illustration).
Figure1:Existing Process and New Process Envisioned to Determine Weekly Shipments to Stores.
According to that process,each store manager would receive a weekly statement of the subt of references available in the central warehou for which he/she may request a shipment to his/her store.Note that this weekly statement(dubbed”the offer”)would thus effectively implement any assortment decision made by Zara’s headquarters for that particular store.It would not mention
迷迷糊糊的意思
however the total quantity of inventory available in the warehou for each reference listed.After considering the inventory remaining in their respective stores,store managers would then transmit back requested shipment quantities(possibly zero)for every size of every one of tho references.A team of employees at the warehou would then reconcile all tho requests by modifying(typically lowering)the shipment quantities so that the overall quantity shipped for each reference and size was feasible in light of the remaining warehou inventory.
At the beginning of our interaction,Zara expresd some concerns about the process just de-scribed,stating that while it had worked well for the distribution network for which it had been originally designed,the growth of its network to more than a thousand stores(and recent expansion at a pace of more than a hundred stores per year)may justify a more scalable process.Afirst issue centered on the incentives of store managers,who were primarily rewarded for the total sales achieved in their stores.We believe that as a conquence store managers would frequently request quantities exceeding their true needs,particularly when suspecting that the warehou may not hold enough inventory of a top-lling reference to satisfy all stores(among others,Cachon and Lariviere1999study a stock rationing model capturing this behavior).Another issue was that store managers are responsible for a large t of responsibilities beyond determining shipment quantities,
including building,sustaining and managing a team of veral dozen sales associates in environ-ments with high employee turnover.Finally,we also believe that the very large amount of data that the warehou allocation team was responsible for reviewing(shipments of veral hundred references offered in veral sizes to more than a thousand stores)made it challenging to balance inventory allocations across stores and references in a consistent way,let alone one that would globally maximize sales.Motivated by the obrvations,we started discussing with Zara the alternative process for determining the weekly shipment quantities that is illustrated in Figure1 (b).The new process envisioned consists of using some input from store managers along with past historical sales to build demand forecasts,then u the forecasts,the inventory of each reference and size remaining both in the warehou and each store,and the assortment decisions as inputs to an optimization model having shipment quantities as its main decision variables.
The remainder of this paper discuss the work we have performed in order to develop and test the optimization model supporting the new inventory allocation process just described–we do not further discuss the associated forecasting model here,details of which may be found in Correa(2007).After a discussion of the relevant literature in§2,we discuss in§3the successive steps we followed to develop that optimization model,specifically the analysis of a stochastic model predicting the weekly
ddmsales to be expected from a single store with given starting inventory profile and merchandi display policy(in§3.1)and the formulation of an optimization model for
the distribution of a single reference over the entire network embedding an approximation of the stochastic model just described(in§3.2).Section4discuss afield experiment we have conducted with Zara in order to asss the likely impact of a potential large-scale implementation of our propod inventory allocation process.Finally,we offer concluding remarks in§5.The Appendix contains a technical proof(§5.1),an extension of the model just mentioned to the ca of references offered in multiple colors(§5.2),and some material related to the software implementation of this work(§5.3).
vrio
2.Literature Review
The fast-fashion retail paradigm described in the previous ction gives ri to many novel and interesting operational challenges,as highlighted in the ca studies on Zara by Ghemawat and Nueno(2003),Ferdows et al.(2003),McAfee et al.(2004)and Fraiman et al.(2002).However,we are aware of only one paper besides the prent one describing an analytical model formulated to address an operational problem that is specific to fast-fashion companies.Namely,Caro and Gallien (
sqa2007)study the problem of dynamically optimizing the assortment of a which products it carries)as more information becomes available during the lling ason,which is motivated by the frequent assortment changes en in fast-fashion outlets.In the prent paper,the product assortment of each store is considered an exogenous input to the inventory allocation problem. This is justified by Zara’s current operations,whereby inventory shipments are subordinated to assortment decisions in a hierarchical manner(e Figure1).In that n,the prent paper constitutes a logical continuation to Caro and Gallien(2007).
The generic problem of allocating inventory from a central warehou to veral locations sat-isfying parate demand streams has received much attention in the literature.Remarkably,the optimal allotment of limited stock over time in common models of such a distribution system is still an open question.When demand is assumed to be deterministic however,there are very effective heuristics with data-independent worst ca performance bounds for tting reorder intervals(e Muckstadt and Roundy1993for a survey).For the arguably more realistic ca of stochastic de-mand that we consider here,inventory policies described in the literature are bad on approximate analysis,and bounds on their performance,when they are available,depend on problem data.Fo-cusing on stochastic periodic-review models(store inventory replenishment occurs on afixed weekly schedule a
t Zara),Table1summarizes the main features of reprentative existing studies along with that of the prent one–for a more exhaustive description of this body of literature,e the recent paper by Axs¨a ter,Marklund and Silver(2002)or the earlier survey by Federgruen(1993).
