See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/313851557

Fashion retailing: A framework for supply chain optimization

Article  in  Uncertain Supply Chain Management · January 2017

DOI: 10.5267/j.uscm.2016.12.002

CITATIONS READS
13 5,550

5 authors, including:

Giada Martino Raffaele Iannone

Università degli Studi di Salerno Università degli Studi di Salerno
12 PUBLICATIONS   159 CITATIONS    103 PUBLICATIONS   930 CITATIONS   

SEE PROFILE SEE PROFILE

Marcello Fera Salvatore Miranda

Università degli Studi della Campania "Luigi Vanvitelli" Università degli Studi di Salerno
71 PUBLICATIONS   836 CITATIONS    96 PUBLICATIONS   1,102 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Application of Digital Twin Models View project

Additive Manufacturing Planning and Scheduling Model View project

All content following this page was uploaded by Marcello Fera on 16 March 2017.

The user has requested enhancement of the downloaded file.

 Uncertain Supply Chain Management 5 (2017) 243–272

Contents lists available at GrowingScience

Uncertain Supply Chain Management

homepage: www.GrowingScience.com/uscm

Fashion retailing: A framework for supply chain optimization

Giada Martinoa*, Raffaele Iannnonea, Marcello Ferab, Salvatore Mirandaa, Stefano Riemmaa

a
Department of Industrial Engineering, University of Salerno, Via G. Paolo II, Fisciano (SA), Italy
b
Department of Industrial and Information Engineering, Second University of Naples, Via Roma 29, Aversa (CE), Italy
CHRONICLE ABSTRACT

Article history: Fashion and Apparel Supply Chains work in a very fast-changing environment and always
Received July 2, 2016 demand better quality, higher availability of products, broader assortments and shorter delivery
Received in revised format times. An efficient Supply Chain Management can make a difference between success and
September 10, 2016
failure in the market. In this context, the main purposes of the presented work are: (i) to define
Accepted December 15 2016
Available online the physical and informative flows, together with connected cost and revenue items, which
December 17 2016 characterize a Fashion Supply Chain working with a wide network of direct-operated or
Keywords: franchising mono-brand stores and (ii) to optimize Supply Chain performances through a
Supply chain management responsive approach which, during the sales season, analyses actual market demand and
Fashion and apparel industry adjusts operations plans accordingly. The framework aims at becoming a decision support
Retailing system for the optimization of the performances of a process that starts from the development
Framework of the collection by the Styling Office and ends with the withdrawal of unsold items from the
SCOR model stores. In order to analyze the performances under different scenarios, a set of Key Performance
Key performance indicator indicators, partially selected from the SCOR Model, is defined.
Simulation © 2017 Growing Science Ltd. All rights reserved.

Nomenclature

i=1,…,n nr. of stores

j=1,…,m nr. of warehouses
k=1,..,l nr. of clothing items/suppliers
T duration of the sales season
Fki demand forecast for the k-th item in the i-th store
Fk total demand forecast for the k-th item in all the stores
Qk purchase quantity for the k-th item
Q total purchase quantity for all the items
SS percentage of safety stock
QD,kj quantity delivered from the k-th supplier og the k-th item to the j-th warehouse
QD,j total quantity delivered to the j-th warehouse
STj stock level in the j-th warehouse
average stock level in the j-th warehouse
STPOS,ki stock level for the k-th item in the i-th store
, average stock level in the i-th store
QPOS,ji quantity delivered from the j-th warehouse to the i-th store
* Corresponding author
E-mail address: gmartino@unisa.it (G. Martino)

© 2017 Growing Science Ltd. All rights reserved.

doi: 10.5267/j.uscm.2016.12.002

 
 
 
 
244
qkji quantity if the k-th item delivered from the j-th warehouse to the i-th store
ski sales of the k-th item in the i-th store
dki demand of the k-th item in the i-th store
∆ , deviation between demand forecast and actual sales for the k-th item in the i-th store
threshold for stores deviation
∆ , global deviation between demand forecast and actual sales for the k-th item
threshold for demand deviation
u percentage of variation in operations plans
CP purchase cost
cuk unitary purchase cost for the k-th item
CPT primary transport cost
Ctf,k fixed primary transport cost from the k-th supplier
Ctv,k variable primary transport cost from the k-th supplier
DISTkj distance from the k-th supplier to the j-th warehouse
CMW warehouse management cost
Cmf,j fixed management cost for the j-th warehouse
chK holding cost of the k-th item in the warehouse
CST secondary transport cost
ctf,i fixed secondary transport cost to the i-th store
ctv,i variable secondary transport cost to the i-th store
disti distance from the j-th warehouse to the i-th store
OOSki out of stock for the k-th item in the i-th store
CMS stores management cost
fixed management cost of the i-th store
cmf,i fixed management cost for the i-th store
Store holding cost fot the k-th item
COOS Out of stock cost
prk Price of the k-th item
csh Shortage cost
R Revenue
P Profit
SL Service level
FA Forecasting accuracy
ITw,j Inventory turnover for the j-th warehouse
ITPOS,i Inventory turnover for the i-th store
As,ki Shelf availability of the k-th item in the i-th store
Aw,kj Warehouse availability of the k-th item in the j-th warehouse
hki Historical sales data for the k-th item in the i-th store
fdel Frequency of delivery from suppliers to warehouses
ndel Nr. of deliveries from suppliers to warehouses
frep Frequency of replenishment from warehouses to stores
nrep Nr. of replenishments deom suppliers to warehouses
Cpen Penalty cost for changing purchase plans
pen Unitary penalty cost

1. Introduction
In modern Fashion and Apparel Supply Chains, customers are demanding better quality, higher
availability of products, broader assortments and shorter delivery times. In this complex context, the
main challenge is to gain value through supply chain management, allowing to respond quickly,
efficiently and with flexibility to demand fluctuations (Battista & Schiraldi, 2013; Martino et al.,
2016a). This obviously requires (Masson et al., 2007):

• market sensitivity, connection to the customer and capacity to capture trends as they emerge
(Martino et al., 2015);
• integration with all the other Supply Chain actors sharing real-time demand data;
• process alignment, both within the company and externally with upstream and downstream
partners.
 G. Martino et al. /Uncertain Supply Chain Management 5 (2017) 245
 

Based on these considerations, this work presents a framework which defines and formalizes physical
and informative flows that are characteristics of a fashion company operating with a dense network of
direct-operated or franchising mono-brand stores. The model analyses the entire supply chain process,
from the definition of the seasonal collection to the withdrawal of unsold goods from stores. This model
is used to propose a more reactive approach which is based on a real-time updating process in order to
quickly respond to changes in market demand and to optimise a set of Key Performance Indicators.
The paper is organized as follows:

 section 2 analyses the fashion market and its features in order to highlight its complexity and to
define the context in which the framework fits;
 the third section presents the framework and describes each block into details with the related
main cost items;
 section 4 presents a set of Key Performance Indicators (KPI) selected from the Supply Chain
Operations Reference (SCOR) model used to evaluate and optimize performances of the entire
Supply Chain;
 section 5 describes the simulation model implemented in Excel;
 the sixth section presents the case study and all the data used in the simulation model;
 a sensitivity analysis is performed (section 7) before analysing results in section 8;
 in the end, in section 9, the results obtained in this research work are discussed.

2. The fashion market and its characteristics

The Fashion and Apparel (F&A) Industry represents a particular example of the manufacturing Industry
and shows several characteristics that make it difficult to manage its productive and logistic process
using traditional methods. In particular, three main issues were identified for this sector (Christopher
et al., 2004):

 Short Product Life Cycles: the product is designed to capture the mood of the moment, then,
compared to other markets, fashion sales trend has a rapid growth, a peak of popularity and
immediately a stage of decline or even rejection of the product by the market (Fig. 1). Given
that products have a limited time in the market from their introduction to decline, retailers have
to be more efficient in the replenishment process (Barnes, 2009; Martino et al., 2016b).
 Unpredictable and Volatile Demand since it is driven by extremely unstable phenomena, such
as weather, movies, sports, etc. Consumer’s demand could change completely in a short time
because an increasing star leads a new fashion trend (Wang et al., 2012), then nobody can ever
tell if a fashion item will be successful on the market.
 Impulsive Purchasing Behaviour: given the fickle nature of fashion shoppers and the
impulsiveness that surrounds their purchase behaviour, retailers have to arrange layouts and
displays items in an appropriate way in order to manipulate purchasing decisions (Newman &
Foxall, 2003). These considerations point out the need to ensure high availability, not only in
terms of product range but also in terms of sizes and colours.

Fig. 1. Fashion product life cycle compared to other markets (Bandinelli et al., 2011)
246

Other characteristics highlighted by several studies in this sector are:

• Extremely Wide Product Variety (Vaagen & Wallace, 2008): it comes to thousands of Stock Keeping
Units (SKUs) considering variety of sizes and colours, then production lines have to manage
highly variable small batches (De Carlo et al. 2013); 

• Demand-Driven Supply Chains (Walters, 2006): consumer expectation has increased asking for
speed, variety and style at low prices;
• Long and Complex Supply Chains, which often include suppliers located in several different
Countries (Bruce et al., 2007). In fact, offshore transfer of the labor intensive stages of the
supply chain to low-cost Countries has been one of the favourite options of companies in the
industrialised Countries to offset some of the risks in the volatile world of fashion (Fernie &
Azuma, 2004). This contributes to highly increase lead time even reducing labor cost;
• Long time‐to‐market, there is almost a year lapse from the definition of the clothing item to its
introduction in the stores. It means that wholesalers and stores define their orders before the
previous season is over and therefore the level of unsold stocks is not known yet (Forza and
Vinelli, 2000) thus contributing to enlarge demand uncertainty (Xiao & Jiao, 2011) (Fera et al.,
2017); 

• New Product Development (NPD) Process is long and not always successful. This activity usually
begins two years before production (Bandinelli et al., 2013) and not all items that come out
from this process are introduced into the market. This generates costs of development and
prototyping that cannot be recovered by sales. In fact, pattern book only represents the
collection idea but items which are not successful during presentation will not be produced.
From all these considerations, it becomes clear that non-value added phases represent an
important part of the production process, making logistic costs significantly impact on the total
cost of the product. If this cannot be a critical issue for companies working in the luxury market,
given the very high contribution margins that they can achieve, for large mass productions,
instead, these costs may critically influence on profits defining company’s economic success or
failure.

3. Definition of the framework

In this section, the traditional approach, solely based on sales forecasting, is analysed (As-Is - ref.
section 3.1). Then, in section 3.2 the proposed approach (To-Be) is described, which is based on a
deviation analysis and an adjusting process of the operations plans with the main purpose of making
the Supply Chain more reactive to market changes. The models are developed according to the
following hypotheses:

a. the seasonal collection is unique, developed almost a year before its introduction in the market;

b. the production is launched based on sales forecasts and not on consolidated orders;

c. Points of Sale (POS) are replenished more than once during the sales season.

3.1. As-Is model

The process flow in the traditional approach is shown in Fig. 2 and starts from the development of the
New Collection (A) by the styling office and the definition of the Demand Forecasts (C). While the
New Collection is considered as a simple input for the framework, forecasting is one of the pillars on
which all further planning activities are based. In the F&A industry this process is crucial and
particularly complex due to high volatility and unpredictability of demand and is based on historical
sales data and characteristics of the new collection (Classification - B) and stores. Next step is the
drafting of Merchandise Orders (D), which define purchasing quantities for each item, and Delivery
 G. Martino et al. /Uncertain Supply Chain Management 5 (2017) 247
 

Orders (E), which define time and place for products deliveries from Suppliers (F). For simplicity, we
suppose that the k-th supplier produces the k-th item and delivers it to the j-th area warehouse in
quantity Qkj. The supply process ends with the delivery of goods to the Area Warehouses (G) according
to the Delivery Orders. At this point, warehouse staff has the task of preparing personalized kits of
items to send to the Stores (I) according to the Replenishment Orders (H). The j-th warehouse supplies
only a specific set of nj stores pertaining to its area. The process described so far defines the material
and informative flow that characterizes the PRE-SEASON phase that, as the name implies, is performed
before the beginning of the sales season. For example, in the case of Fall/Winter (FW) Season this
phase starts in October, with the development of the new collection before the most important sector
fairs, and ends in September with the first deliveries to the stores (Table 2).

The IN SEASON phase, instead, starts with the first sales recorded in the stores. We suppose that both
deliveries from the suppliers and replenishments to the stores are also performed during the selling
season even if they are scheduled before it. This phase ends with the collection of unsold goods from
stores and warehouses. These un- sold items will be then delivered to Factory Outlet stores (K) and
disposed during the following seasons (POST SEASON). It is clear that each item will have a fall in
price in proportion to the time of storage in the outlet store, which results in a reduction of the
contribution margin. In addition, during this phase, all sales data recorded will be collected and used to
draft demand forecasts for the following season (T+1).The different steps of this framework will be
described in detail in the following paragraphs and the nomenclature is summarised in Appendix.

Fig. 2. Framework for the definition of the As-Is model

3.1.1. Development of the New Collection (A)

The definition of the collection primarily concerns to the Styling Office and is per- formed down line
of a long phase of studies of trends, fair attendance, etc. This activity will not be part of this study but
it will be just considered as an input to the model. It is, in fact, responsibility of fashion designers to
define materials, design clothing items and decide in which colours, sizes and variations to replicate
them. Table 1 shows an example of the information items needed for the definition of the collection. In
more complex cases of worldwide companies, the collection could also be substantially modified
according to the reference market (European, Asian or American). Generally, after its development and
before starting procurement and production, the collection is presented to buyers during the most
important sector fairs (Milano Fashion Week, Premiere Vision in Paris, etc.); from this moment on,
sales to wholesalers begin. Starting production after having acquired orders from the market obviously
increases time to market; this can only be accepted for high level brands or for collection with an
important stylistic content.
248

Table 1
An example of characteristics of collection items
Item code Materials Colors Sizes Process Accessories
Blue Buttons
001 Wool Black 40-46 Knitting Brooch
Brown Belt
Blue Cutting
002 Denim 38-44 -
Black Scouring
… … … … … …

Table 2
Example of activity schedule for a Fall/Winter Season

PRE SEASON IN SEASON POST SEASON
Steps/Months 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 1 2 3
New Collection
(A)

Classification
(B)

Demand
Forecast (C)

Merchandise
Plan (D)

Production

Del.to
Warehouse (G)

First Delivery
to Stores (I)

Sales season

Replenishment
(H)

Collection of
unsold goods

Del. To outlet
stores (K)

Outlet selling
season

traditional flow
update process

Sophisticated and expensive items, in fact, cannot be produced without reliable sales data; wrong
demand forecast can lead to heavy losses and capitals fixed in stocks. Companies that work with a
direct retail distribution, as the case described in this paper, could also avoid this phase of presentation
to buyers, or perform it but without time constraints, just to acquire image on the market and to promote
the company to the press. In this case, there are no long waits for acquiring final orders and no specific
delivery commitments since the company issues internal orders for its Retailers and delivers goods
according to its internal plans. On the other hand, procurement and purchase orders are performed based
on demand forecasts, forcing the company to take the risk connected to forecast errors and to be sure
that products are available in stores when the market demands them, so as not to lose sales.

3.1.2. Classification (B)

For specific business needs, each item is generally identified with a unique code. For our purposes and
given the aim of the proposed framework, instead, each clothing item is classified according to Product
Category, Price Range and Seasonality. Researches on these past sales data show that, depending on
the position of the store (in a Shopping Mall, on the Street or in an Airport), customers purchase
behaviour is different both for Product Category and for Price Range (Iannone et al., 2013). In
particular, for products we identified three macro-categories:

 Clothing: products that can be quickly purchased without trying them on in the dressing room;
this aspect allows to reduce the time spent by the customers for the purchasing activities and by
 G. Martino et al. /Uncertain Supply Chain Management 5 (2017) 249
 

the shop assistants for serving clients;

 Clothing to Try on: trousers, dresses and all other items that require the use of the dressing room
and a longer stay of customers in the stores;

 Accessories: handbags, scarves, jewellery, etc. For this articles customers do not have the
choose size which best fits them but just color.

As regards the price, instead, we defined three different ranges: Cheap, Intermediate and Expensive.

According to seasonality characteristics of items, we can distinguish (Nuttle et al., 1991; Sen, 2008):

 Basic products, sold throughout the year, which amount approximately 25% of all apparel;
 Seasonal products, with a 20-week product life, which represent almost 45% of apparel. Those
products are replaced by new lines twice or three times per year;
 Fashion products, with 10-week product life and more than four seasons per year. Fashion items
are steadily increasing their share of the market at the expense of both seasonal and basic goods
since consumers always demand greater variety and more frequent changes. Some major
retailers argue that customers visit their store, on average, once every two months and they want
their customers to see 
new lines of products during each visit.

This decision to deal with product categories and price ranges instead of single item codes is primarily
due to the need of having forecasts as accurate as possible; a lower detail level, that is aggregated
forecast for few groups of products instead of detailed forecast for thousands of codes, in fact, allows
committing a lower error. Furthermore, we observed that, depending on the store location, customer’s
purchasing choices can be substantially different toward the different categories indicated (refer to
paragraph 3.1.3).

3.1.3. Demand Forecasts (C)

Next step of the model is demand forecasting. In the fashion industry this process is crucial and
particularly complex due to high volatility and unpredictability of demand. In the proposed model,
Demand Forecasts provide the basis in which all operations plans are compiled and are drawn up
according to historical data collected from stores. In order to compile reliable forecasts, it is useful to
distinguish whether product is basic/continuing or seasonal/fashionable since forecasting methods can
be substantially different in these two cases. In the first case, in fact, historical data are quite reliable
and stable; this means that forecasts can nearly be only based on them. In the second case, instead,
historical data are not available since the product is new and it is difficult to anticipate sales results.
Inputs required for this step fall into three classes: the first one is related to physical characteristics of
stores, the second one concerns historical data and the last one regards characteristics of the new
collection.

1. Physical characteristics of stores are defined just once, when opening the stores, and remain
unchanged over time. They are:
 Dimension (Dim), both of the exhibition area and of the internal warehouse. This factor is
important to define maximum level of stocks that stores are able to accept and manage
without overloading;
 Location (Loc), which can be on the Street (ST), in a Shopping Mall (SM) or in an Airport
(ARP). As already introduced in the previous paragraph, customers purchasing behaviour is
considerably different for different cases. Initial analysis of data coming from all the stores
highlighted that, depending on the position, the three product categories - clothing, clothing
to try on and accessories - record different sales levels. Accessories, for example, are highly
sold in airports because customers are passing by and the purchasing must be very quick,
250

while in shopping malls and in stores on the street, accessories have very little success.
Opposite behaviour is shown for clothing to try on, while clothing which do not require the
use of the dressing room are equally sold in all stores.
 Geographical Area (Geo), in which stores are located. In this paper, since we are referring
to a company that works nationwide in Italy, we consider three different areas: North, Centre
and South; but it can be simply adjusted considering regions, counties, etc., in other cases.
This parameter is used to evaluate socio-economic factors that influence purchasing
behaviour mainly toward different price ranges;
 Competitor (Comp), which defines how many nearby competitors are located in the same
area.

2. Historical sales data, instead, are processed at the end of each season thank to Electronic Data
Interchange (EDI) and Electronic Point of Sale (EPOS). They are:


 Turnover (Turn), recorded during the previous comparable season. It is meaningless, in fact,
to evaluate the previous season; it means considering turnover of Spring/Summer season for
drafting forecasts for Fall/Winter season and vice versa. For stores located in tourist areas
this can lead to significant errors if we do not consider the alternation between off-season
and tourist season. 

 Sales percentage (%Sales) related to deliveries, defined as: 

% (1)

It is clear that sales will increase with the availability of goods in stores (ref. paragraph
4.7), thus making this parameter more relevant than the pure sales data.
Hence, given l the number of handled items and n the number of active stores during the season, demand
forecasts for the k-th item and the i-th store ( ) will be a complex function of all the above-mentioned
parameters:

, , , , ,% (2)
3. New Collection parameters are the last ones that have to be considered; they are expressed as
a percentage deviation from the previous season and are defined as (Bini, 2011):
 Attractiveness (∆att): which has a positive value if the Styling Office thinks that the new
collection will be more successful than the previous one, negative in the opposite case. This
impact is estimated on the ratings of customers to which the collection is shown during
sector fairs. 

 Investment in marketing and communication (∆mrk): if the company intends to carry out a
powerful advertising and communication campaign, this will increase brand awareness and
directly influence final client’s purchasing. Unlike the previous factor that can only be
subjectively assessed, the budget devoted to advertising is defined in advance by the
company and ∆mrk can be easily calculated as:

∆ , (3)

where T is the index related to the season and Mkt is the budget assigned to the marketing campaign.

Then ultimately, sales forecasts previously defined must be adjusted as:

∗ 1 ∆ ∆ , (4)
 G. Martino et al. /Uncertain Supply Chain Management 5 (2017) 251
 

All remarks made so far do not include either size or color of the clothing item. For sizes, generally
demand trend follows a Gaussian curve centered on a particular size which varies according to the
customer target and the reference market. All colors, instead, are treated in the same way: equal quantity
for each variation of color is purchased. Nevertheless, it may be defined the “trendy” color for the
season under exam and increase its forecasts of a small percentage. All forecasting factors and their
characteristics are summarised in Table 3.
The Purchasing Office will, then, receive sales forecasts for
each k-th clothing item defined as:

(5)

These quantities will represent a kind of internal order for the company whereby launching orders to
suppliers. It is important to underline that, if the company plans to open new stores, you have to evaluate
its impact by considering past sales of a real store which is equivalent in terms of dimension and
location. In this case, forecasts will be partially reduced due to the necessary start up.

Table 3
Summary of the forecasting factors and their characteristics
Degree of Level of detail Parameters Unit of Forecasting characteristics
detail measure
Impact
COLLECTION % Deviation from previous season
Marketing
Dimension m2
ST
Position SM
Non changing physical
ARP
POS characteristics
North
Geogr. Area Centre
South
Turnover € Historical Data
Basic Consolidated Historical Data
SEASONALITY Seasonal % Historical Data
Fashion NO Historical Data
% Sales Clothing
PRODUCT
% Sales Cl. To try on % Historical Data
CATEGORY
% Sales Accessories
Historical Data from trendy color
Trendy
COLOR - of previous season
Basic Consolidated historical data
+ Gaussian
SIZE - Target Market
Distribution

3.1.4. Definition of the Merchandise Orders (D)

The main Operations Plan that a fashion company must issue is the Merchandise Plan which defined
purchase quantities for each item code. For its definition the Purchasing Office evaluates, besides sales
forecasts, possible volume discounts granted by suppliers, company’s economic and financial capacity
and a need to ensure a certain unsold stock in order to conveniently manage factory outlets, defined as
Outlet Compensation Stock (OS). Therefore, the purchase quantity Qk will be defined as:
∗ 1
(6)

where:
252

 ASk is the Assortment Stock and represents the quantity to be added to the forecasts in order
to guarantee availability of all sizes and colors for each item. Then:
0 ,
(7)
0 ,
 OS is the Outlet Compensation Stock and is expressed as a percentage of the quantities
defined by forecasts and assortments. As described following (ref. paragraph 3.1.10),
factory outlet stores are used to absorb the risk of overestimation of the demand which can
be important when managing hundreds of stores. At the same time, therefore, to be attractive
for customers, outlet stores as well must guarantee a fixed assortment of sizes and colours,
at least at the beginning of the season. 

 SS is the Safety Stock defined according to the service level that you want to ensure and to
the demand rate. 
Then the total quantity will be: 


. (8)

Before issuing these orders to suppliers, the Purchasing Office must evaluate: 


 Purchasing Budget, which is fixed when defining the collection;

 Discounts on quantities that we can obtain from suppliers. 

Each order must contain the following information:

 Suppliers: they are chosen with a precise procedure that evaluates stylistic and technical
requirements (ref. paragraph 3.1.6); 

 Quantities Qk: we are supposing that each item is produced by one and only supplier, then
the index k refers both to the item and to the supplier.

3.1.5. Definition of the Delivery Orders (E)

These orders define, for each supplier, quantities and times for deliveries to the central warehouse. It is
clear that, in the simplest case, there is a single central warehouse, for example when companies operate
only at national level. For global companies, instead, warehouses will be numerous and located all over
the world. In this case, defining delivery plans will be more complex since it will be necessary to
optimize those plans according to sales forecasts for the different geographical areas served by the
warehouse. The orders that define deliveries to the central warehouse are issued by the Purchase Office
in agreement with the suppliers. These plans, in fact, must combine two conflicting requirements. On
one hand, the company requires frequent deliveries for medium-small lots so as not to overload
warehouse and its resources. In addition, for more complex collections, which involve many clothing
items of different warmness, the company prefers to have at stock items appropriate to the temperature
of the moment in order to be able to respond quickly to customers’ requests. During the Spring/Summer
season, warmer items are generally shipped at the beginning of the season (late January/early February)
because weather is still cold and those items are then more requested, while lighter ones are delivered
in early spring when milder weather justifies their purchasing. In the Fall/Winter season the opposite
case occurs. In August/September it is better to have lighter clothing in stores for still high
temperatures, while only later, when weather gets colder, stores will receive winter clothes. On the
other hand, suppliers would prefer less frequent deliveries for larger lots in order to reduce shipping
costs. Hence, it is possible to define, together with the suppliers, the total quantities to be delivered at
time t to the j-th warehouse as:

, , (9)
 G. Martino et al. /Uncertain Supply Chain Management 5 (2017) 253
 

With ∑ , ∀ ∈ 1, … ,
(10)

Eq. (10) means that the purchased quantity for each k-th item has to be totally delivered to the
warehouse during the considered time range T.

3.1.6. Suppliers (F)

Outsourcing is becoming an unavoidable trend in cost cutting for F&A companies, especially when
purchasing activities involve multiple international suppliers overseas located. This means that
establishing an efficient relationship between buyers and suppliers is a critical factor. Two are the
unavoidable features that suppliers must respect:

 Qualification and Quality Assurance: the initial hypothesis on which the described model
is based is that collection is already defined by stylists and consequently all materials and
manufacturing processes requested are already known. It is clear that suppliers must possess
all the skills and equipment needed to produce a clothing item respecting all the company’s
specifications;

 Productive capacity: selected suppliers must be able to produce the required quantities on
schedule. Then it is necessary to verify that they are not already saturated due to other
client’s orders.

Other discriminant factors are:

 Cost: a factor of paramount importance is clearly the purchase cost of the clothing item and
is one of the main elements of suppliers choice; 

 Location: the geographical position of the supplier is important to evaluate shipping time
and cost and any custom duties;
 Flexibility: which represents the capability of the supplier to respond to unexpected requests
or variations in terms of volume and mix of products; 

 Reliability: which evaluates the supplier’s ability of respecting contractual requirements in
terms of delivery dates, quality etc. 

In the proposed model we consider the case of complete outsourcing of the production process.
Actually, some companies outsource only a part of the entire process while internally performing
finishing steps. In this case we should also consider a suppliers’ network, a warehouse for semi-finished
products and the resources needed for final phases of production. Several studies were proposed in this
context for evaluating and selecting suppliers; in the apparel industry most used models are based on
AHP (Tend & Jaramillo, 2005; Chan & Chan, 2010; Fera & Macchiaroli, 2010) and use several areas
of evaluation such as delivery, quality, assurance of supply, flexibility, cost and reliability.


3.1.7. Central Warehouses (G)

The central warehouse is responsible for receiving goods from different suppliers, storing and inventory
managing (Fera & Macchiaroli, 2009). Then, through the picking and sorting process, personalized kits
of items will be prepared and shipped to the stores network. Besides material flows, warehouses must
manage the information flow concerning delivery and replenishment plans provided by the Department.
The stock level (ST) at the time t for the j-th warehouse can be calculated as:

1 , , (11)
254

where QPOS,ji is the quantity delivered from the j-th warehouse to the i-th Point of Sale (POS). We
suppose that each warehouse supplies only a specific set of nj stores pertaining to its area. Then QPOS,ji
is null for the i-th stores not pertaining to the j-th warehouse. We also need to consider physical limits
of the central warehouse concerning storage capacity:

, (12)
and concerning materials handling capacity:

, , , (13)

where HCmax is the maximum materials handling capacity.

It is important to underline that some items may require additional treatments (such as for example
ironing) before delivery to the stores; this obviously creates another time constraints that has to be
respected before shipment to stores.

3.1.8. Definition of the Replenishment Orders (H)

Against the delivery plan to the central warehouse, the company must define personalized kits of items
to deliver to the i-th store for the k-th item (d’Avolio et al., 2015). Given n the total number of stores
directly managed by the company and qkji(t) the quantity of the k-th item delivered from the j-th
warehouse to the i-th store at the time t, we can say that the total quantity delivered to the i-th store is:

, (14)

where ∑ , ∑ ,
(15)

Eq. (15) means that, for each k-th item, quantity delivered to the i-th store in the considered time range
T cannot obviously overcome total quantity delivered to the warehouse.
It is important not leave stores lacking in some requested items or sizes but neither to overload them,
thus respecting the capacity constraint:

, ,,
(16)

where STPOS,i is the Stock Level in the i-th POS and is defined as:

, , 1 (17)

where ski(t) are the sales of the k-th item in the i-th store at time t. Even if a first schedule is defined
during the In Season phase, stores are not replenished in a unique solution before the sales season. In
the As-Is case plans drafting is a process which is performed only once according to forecasts, while in
he proposed approach (To-Be case - ref. section 3.2) it is performed before each replenishment
according to actual market demand.

3.1.9. Stores (I)

Stores are the final ring of the supply chain and represent the point of direct interface with customers
 G. Martino et al. /Uncertain Supply Chain Management 5 (2017) 255
 

offering not only a mix of goods but also services. In this study they are considered as passive recipients
of products allocated by the company but have the essential role of recording actual sales through EDI
and EPOS systems. The only way of implementing an effective demand driven approach is, in fact, real
time information sharing between retailers and the Logistic Department.
Real sales recorded in the stores can be evaluated as:

min ; , , (18)
where dki(t) is the demand of the k-th item in the i-th store. Actually, this parameter is difficult to
measure; in most cases, in fact there is no awareness of the entity of the potential lost sales since the
only available information on customer demand derives from sales data (Battista et al., 2011).

3.1.10. Factory Outlet stores (K)

The primary role of outlet stores is to absorb the risk of incorrect demand forecasts, especially in case
of demand overestimation. Factory Outlet stores sell at significantly discounted prices thus greatly
reducing contribution margins for each item and involving additional costs for the management of other
stores and for the withdrawal of unsold goods from main stores and delivery to outlets. Therefore, when
it comes to hundreds of main POS, unsold items reach such high volumes that it is unavoidable the use
of an effective system for their disposal. On the other hand, to conveniently run an outlet, we cannot
simply offer unsold items from main stores; it would be a poor and not assorted offer that would not
attract customers. Many companies, then, complete this offer by producing additional items to replace
missing colors and sizes or just to offer new maybe simpler and cheaper items.

3.2. To-Be model

As discussed in previous sections, the As-Is model is solely based on forecasts thus making it difficult
to adjust purchasing and operations plans according to high fluctuations in demand during the sales
season. To overcome this limit, the proposed model defines plans according to actuals sales data rather
than only forecasts. Thank to sales data recorded in the stores, in fact, it is possible to assess deviations
between real sales and forecasts. This analysis represents the core of the proposed approach (To-Be)
and defines in real-time how much the demand was under-estimated or over-estimated. If this deviation
is higher than a fixed threshold, the model will update all the Merchandise Orders and possibly cancel
some orders or issue new ones; otherwise it will simply update Replenishment Orders, increasing or
reducing quantities to be delivered to stores. This adjusting procedure is called Update Process.

3.2.1. Deviation analysis (J)

In order to update in real time all merchandise, delivery and replenishment plans, it is necessary to
analyse the deviation between actual recorded sales and forecasts. This process represents the core of
the proposed approach (To-Be case). Two different analysis must be performed at different time
intervals. The first one, Stores Demand & Inventory Analysis, allows us to evaluate in real-time how
much stores demand was under-estimated or over-estimated.

∆ , (19)

It is performed at time intervals τ and evaluates not only actual sales but also any return, defective or
stolen item. If this deviation is higher than a fixed threshold (δST), the model will update the
replenishment plans adjusting them in order to meet customers’ requests. The replenishment quantities
are updated according to the following equation:
256

∆ , ∗ 1 ∆ , ∗
(20)
∆ ,

where is the updated quantity and u is a parameters which defines how much the replenishment
plans can change and has to be appropriately set. It is important to underline that if ∆ST is negative, it
means that demand was under-estimated then the replenishment quantity ( ) must be incremented,
and vice versa.

Fig. 3. Framework for the definition of the To-Be proposed approach

The second analysis, Supply Chain Demand & Inventory Analysis, evaluates in aggregate data from
stores and warehouses at each time interval θ (with θ > τ).

∆ , (21)

If this deviation is higher than threshold δSC, it means that we have to revise Merchandise Orders and
possibly cancel some orders or issue new ones according to the following equation:
∆ , ∗ 1 ∆ , ∗
(22)
∆ ,

4. Key Performance Indicators (KPI)

After the definition and description of all the Supply Chain processes, it is necessary to define a set of
Key Performance Indicators (KPIs) in order to analyse and optimise performances of the entire Supply
Chain. These KPIs are selected from the SCOR Reference Model 11.0 (Supply Chain Council, 2012)
which defines six primary management processes (Plan, Source, Make, Deliver, Return and Enable)
and five Performance Attributes (Reliability, Responsiveness, Agility, Costs and Asset Management
Efficiency). According to the most used industrial indicators and to (Lanzilotto et al., 2015), the KPI
are selected and listed in Fig. 4; each of them is related to one Performance Attribute and metric defined
by the SCOR model. In the following sections the above-mentioned KPIs will be better described and
 G. Martino et al. /Uncertain Supply Chain Management 5 (2017) 257
 

other two additional KPIs, specifically defined for the application in the fashion industry, will be
proposed.

4.1. Service Level (SL)

It is usually defined as the ratio between orders fulfilled and total orders received. In this context, for
the i-th POS, it is expressed as the ratio between actuals sales recorded (ski) and demand received (dki):

∑
(23)
∑

4.2. Forecasting Accuracy

Forecasting accuracy is calculated for each product category and is defined as the percentage of error
compared to actual sales:

(24)

It is important to control this indicators in order to always improve sales forecasts over time.

4.3. Costs

The main cost items identified for this particular problem are (Iannone et al., 2015):

 Warehouse Management Cost (CM W ):

∗
, ∗ , (25)

where chk is the holding cost expressed as a percentage of the unitary purchase cost (cuk) of the
products in stock at time t and Cmf,j is the fixed management cost of the j-th warehouse.


 Primary Transport Cost (CPT ) (from suppliers to central warehouse):

Given Ctf,k and Ctv,k fixed and variable primary transport costs for the k- th product (we are supposing
that product k is manufactured by one specific supplier) at the time t and given DISTk the distance from
the supplier to the central warehouse, primary transport cost will be:
(26)
, , ∗ , ∗

 Secondary Transport Cost (CST ) (from central warehouse to POS):

(26)
, , ∗ ∗

This is the cost that the company supports for the delivery of quantity qkij to the stores network.
258

Accepting a small approximation, we will suppose that this cost also includes the cost for the collection
of unsold products. Given ctf,i and ctv,i(t) the fixed and variable costs for secondary transport to the i-
th store and distij the distance from the j-th warehouse to the i-th POS, we have that the secondary
transport cost is equal to:

 POS management cost (CMS), including both main stores and Factory Outlet stores:
1
, ∗ ∗ (28)

where:
 is the holding cost for products in the POS. It is higher than the same chk for the central
warehouse since products stored in the POS can not be used anymore for the replenishment
of other POS;
 cmf,i is the fixed management cost for the i-th POS; it is greater than zero only if POS is
owned by the company, otherwise, in case of a franchising store, cmf,i is null since all the
fixed costs are supported by the franchisee.
 Purchase Cost (CP ): Defined as the product of the quantities purchased Q and the unit cost
cu.

∗ (29)

Fig. 4. KPIs identification from SCOR 11.0 model4.4. Inventory Turnover

 G. Martino et al. /Uncertain Supply Chain Management 5 (2017) 259
 

It represents the number of times that inventory is sold or used in a fixed time period and is expressed
by the ratio between quantity outgoing the warehouse and average stock ( ). We evaluate this
indicator both for the central warehouse (in this case outgoing quantities are the item delivered to POS):

∑ ∑ ∑
(30)

and for POS’ internal warehouses (in this case outgoing quantities are sales):

∑
, (31)
,
4.5. Revenues

Besides costs, it is also important to evaluate revenues, which are time depending since products suffer
a depreciation according to the time of permanence in the store. In general, the price will follow a step
function decreasing over time (see an example in Fig. 5).

Fig. 5. Price trend over the sales season

Then, given prk(t) the selling price of the product k at time t, total revenues can be expressed as:

∗ (32)

4.6. Sales Percentage

For the k-th product and for the i-th store, it is defined as the ratio between actual sales and quantities
delivered to each POS:

% (33)

This indicator is not derived from the SCOR model but it is specifically defined for the application in
the fashion industry (ref. paragraph 3.1.3). In this sector, in fact, it is meaningless to evaluate the pure
data on actual sales since, given the impulsive purchasing behavior of customers, sales will increase
with the availability of product in stores.

4.7. Availability

The concept of availability in the retail industry refers to three different aspects (LIUC Centro di
260

Ricerca sulla Logistica, 2011):

 Shelf availability: it is a measure of the event whereby the requested item is not available
on the shelf and not accessible to the customers, although it may available in a different
position of the store (for example in the internal warehouse); 

 Store availability: it is a measure of the event whereby the requested item is not available in
the store but it can be available in the central warehouse/distribution centre or incoming in
the store;
 Warehouse availability: it is a measure of the event whereby there are no available stocks
for the requested item either in the central warehouse. 


In this context we consider that Store Availability coincides with Shelf availability, since we imagine
that shop assistants can replace sold out items on the shelves in a very short time, if obviously they are
available in the store i.e. in their internal warehouse.
This indicator is defined for each item code, and
also for each POS for the shelf availability, by the following relations: 

,

(34)
, ,

4.8. Out of Stock (OOS)

This parameter is strictly connected to the availability (ref. paragraph 4.7) and to the service level (ref.
paragraph 4.1). It is defined as the number of orders that cannot be fulfilled and is given by:

, (35)

In case of out of stock, the customer can act in different ways (Slot et al., 2005; Zeppetella et al., 2016):

a. Buy the same item in a different store;

b. Buy a different item in the same store;
c. Wait until the product is available in the store;
d. Not buy.

In order to reduce the percentage of lost sales, i.e. the percentage of customers which choose the last
alternative (not buy), in recent years, companies have been trying to implement a combination of
“bricks-and-clicks” (Agatz, 2008; Lanzilotto et al., 2014; Lanzilotto et al. 2015). This solution is
commonly called “Multi-Channel Retailing” and involves the integration between physical, mobile and
on-line channel, thus increasing sales and profit-making opportunities (Bermn & Thelen, 2004; Elia et
al., 2014).
The damage caused by OOS to the company (in terms of customer dissatisfaction, image
damage, etc.) can be economically evaluated through the following equation:

∗ ∗ (36)

where csh is the unitary shortage cost expressed as a percentage of the difference between price and
unitary purchase cost.
 G. Martino et al. /Uncertain Supply Chain Management 5 (2017) 261
 

5. Definition of the simulation model

In order to evaluate Supply Chain performances of the proposed framework, a simulation model was
developed and implemented in Excel. Its general diagram is represented in Fig. 6 and has the main
purpose of comparing the previously defined KPIs (ref. section 4) in the AS-Is and To-Be cases. While
the To-Be case involves the In Season deviation analysis and consequent adjusting of the operations
plans, in this case instead, those plans are defined before the sales season and kept fixed throughout the
time range. A summary of the nomenclature already used in previous sections and that will be used
following is reported in Table 4.

Table 4
Nomenclature
i=1,…,n nr. of stores Qk Purchase quantity
j=1,…,m nr. of warehouses QD,kj Quantity delivered to warehouse
k=1,…,l nr. of items qkij Quantity delivered to the store
hski Historical sales data dki demand
SS Percentage of safety stock ski sales
fdel Delivery frequency ST Stock level in central warehouses
del=1,…,ndel nr. of deliveries STPOS Stock level in stores
frep replenishment frequency ΔST,ki Stores deviation
rep=1,…,nrep nr. of replenishments δST Stores deviation threshold
Fki Sales forecasts ΔSC,k Supply Chain deviation
δSC Supply Chain deviation threshold

Fig. 6. General Diagram of the simulation model

The model uses as inputs:

- historical sales data (hki);

- percentage of Safety Stocks (SS);

- frequency of deliveries to the warehouse (del) and of replenishment to stores (rep);
- deviation thresholds, both for stores (δST ) and for Supply Chain (δSC );

and defines:

262

- demand forecasts Fki: for simplicity we suppose that forecasts are equal to historical sales data
hski:

(37)

- merchandise orders Qk: they are given by the total quantity that we are supposing to sell during
the season (Fk) adding safety stocks:

∗ 1 (38)

- Delivery orders QD,kj : we are supposing that each delivery is performed at time:

1 ∗ 1 1, … , (39)

where ndel = T/fdel and the delivered quantity is given by the total quantity that we are
supposing to sell until the next delivery:

, ∗ 1 (40)

- Replenishment orders qkij : as for the delivery orders, we are supposing that each replenishment
os performed at time:

1 ∗ 1 1, … , (41)

where nrep = T/frep and the replenishment quantity is given by the total quantity that we are
supposing to sell until the next replenishment:

(42)

These steps reproduce the Pre Season phase, while for the In Season phase the model generates a casual
demand (dki(t)). If the demand is lower than the stocks available, the item can be sold otherwise we
will have an Out of Stock. The stocks available in the stores are given by:

, (43)

The process described so far defines the As-Is case. For the To-Be case we have to include the Deviation
Analysis, described in paragraph 3.2.1. In summary, the To- Be model analyses how much demand was
under or over-estimated, both at Supply Chain level and at Stores level. If this deviation (∆SC,k, ∆ST,ki)
is higher than a fixed threshold (δSC , δST ) the model updates the replenishment and/or merchandise
orders by increasing or decreasing quantity by a defined percentage (u).

5.1. Key performance Indicators

A summary of the KPIs already described in section 4 is reported in Table 5. They will be used for
comparing AS-Is and To-Be cases.
 G. Martino et al. /Uncertain Supply Chain Management 5 (2017) 263
 

It is important to underline that whenever one of the operations plans undergoes a variation, the
company will have to incur a cost. In the To Be case, this Penalty Cost (Cpen) is added to Purchase
Cost (CP ) and calculated as:

∗ |∆ | (44)

where pen is the unitary penalty cost ([e/pcs]) and ∆Q is the variation ([pcs]) of the updated purchase
plan compared to the one defined before the sales season.
All the cost items and revenues can also be
globally evaluated through the Profit.

6. Introduction to the case study

In order to compare Supply Chain performances of the proposed model we used the real case of an
Italian Fashion Company which works in the national territory with hundreds of franchising and direct
operated mono-brand stores and just a single central warehouse (m=1). The data collected from the
above-mentioned company concern characteristics of 10 selected clothing items (k=10 - Table 6) and
10 selected stores (i=10 - Table 7).

Table 5
Key Performance Indicators
Technical KPIs
∑
Service Level
∑

Forecasting Accuracy

∑ ∑ ∑

Inventory Turnover ∑
,
,

Sales Percentage %

Shelf Availability ,

Out of Stock ,

Economic KPIs

Purchase Cost ∗

Primary Transport Cost , , ∗ , ∗

∗
Warehouse Mng Cost , ∗ ,

Secondary Transport Cost , , ∗ ∗

1
Stores Mng Cost , ∗ ∗

Out of Stock Cost ∗ ∗

Revenues ∗

Profit
264

Table 6
Clothing items characteristics
Price (pr)
Item Category Description Cost (cu)
€ range
1 Clothing to try on Trousers 28 Cheap 8
2 Clothing to try on Shirt 26 Cheap 8
3 Clothing to try on Dress 125 Expensive 40
4 Clothing to try on Denim Trousers 45 Cheap 14
5 Clothing to try on Denim Trousers 60 Intermediate 16
6 Clothing Cotton Cardigan 27 Cheap 8
7 Clothing Jacket 72 Intermediate 20
8 Accessories Necklace 65 Intermediate 18
9 Accessories Handbag 85 Intermediate 25
10 Accessories Foulard 18 Cheap 4

Table 7
Stores characteristics
Dimension
Store Geogr. Area Location
m2 category
1 South Airport 66 Small
2 South Shopping Mall 113 Medium
3 South Street 180 Medium
4 South Street 58 Small
5 South Shopping Mall 62 Small
6 Centre Shopping Mall 343 Large
7 Centre Street 82 Small
8 North Shopping Mall 100 Small
9 North Street 84 Small
10 North Street 41 Small

Parameters related to transport are reported in Table 9 for the primary transport (from supplier to central
warehouse) and in Table 8 for the secondary transport (from central warehouse to stores). For
simplicity, we suppose that each clothing item is produced and delivered by one single supplier.

Table 8
Stores characteristics related to secondary transport
Store 1 2 3 4 5 6 7 8 9 10
dist [km] 50 90 70 30 60 120 150 400 450 500
ctf [€] 10 10 10 10 10 15 15 20 20 20
ctv [€/Km*pcs] 0,02 0,02 0,02 0,02 0,02 0,02 0,02 0,02 0,02 0,02

Table 9
Suppliers characteristics related to primary transport
Supplier 1 2 3 4 5 6 7 8 9 10
DIST [km] 500 1700 5000 2500 700 1000 1500 2000 5000 900
Ctf [€] 50 70 150 130 50 50 70 100 150 50
Ctv [€/Km*pcs] 0,01 0,001 0,001 0,0015 0,01 0,0025 0,005 0,005 0,005 0,003

Fig. 7. Case study historical sales data for the whole sales season
 G. Martino et al. /Uncertain Supply Chain Management 5 (2017) 265
 

The related Historical sales data were collected over a time range of 6 months (24 weeks) corresponding
to the whole Fall/Winter season (from September to February) and are reported in Fig. 7 and Table 10.

Table 10
Case study historical sales data for each item and each store
Store
Item 1 2 3 4 5 6 7 8 9 10
1 23 14 40 10 25 17 12 9 5 1
2 76 38 59 26 62 27 17 16 24 18
3 27 25 69 26 18 27 28 32 22 19
4 1 1 5 0 3 0 0 0 1 1
5 27 47 54 17 39 48 15 40 20 29
6 15 15 16 3 18 18 4 11 10 4
7 5 2 42 4 9 7 3 0 2 0
8 0 0 1 0 1 0 1 1 0 1
9 121 32 50 15 55 42 40 80 34 23
10 272 78 190 52 80 90 57 103 85 67

A summary of other parameters used in the simulation model is reported in Table 11.

Table 11
Parameters of the simulation model
General parameters Deviation Analysis parameters
i 10 T 24 weeks δ 0,3
j 1 SS 10% δ 0,2
k 10 csh 50% u 10%
fde 8 weeks frep 4 weeks pen 0,01 [€/pcs]
Cmf 100 € ch 5%
cmf 150 € c 10%

6.1. Definition of the scenarios

The primary objective of the simulation model is to evaluate if the proposed model is able to quickly
follow changes in demand trend and to consequently adapt operations plans. With this purpose different
demand profiles were defined that may summarise any possible real case: highly variable demand and
increasing or decreasing trends. Six different scenarios were defined and simulated by varying the
demand profile:

Scenario 1: the demand follows a Gaussian distribution with mean equal to the demand forecast
and standard deviation equal to 1% of the mean (Fig. 8(a));
Scenario 2: it is equal to the previous case with a higher standard deviation: 100% of the mean
value (Fig. 8(b));

Scenario 3: from week 10 to 17 the demand has a peak (3 times greater than forecasts) while during
the other weeks it follows Scenario 1 (Fig. 9(a));
Scenario 4 : from week 10 onwards, the demand has a growing trend equal to 1% (Fig. 9(b));

Scenario 5 : from week 10 onwards, the demand has a decreasing trend equal to 1% (Fig. 10(a));
Scenario 6 : from week 10 onwards, stores 1 to 5 have a growing trend - as Scenario 4 - while store
6 to 10 have a decreasing trend - as Scenario 5 - (Fig. 10(b)).
266

(a) (b)
Fig. 8. Demand Profile of Scenario 1 and 2

(a) (b)
Fig. 9. Demand Profile of Scenario 3 and 4

(a) (b)

Fig. 10. Demand Profile of Scenario 5 and 6

Table 12
One-factor-at-a-time Sensitivity Analysis
Parameter→ u
Base case 0,3 10% 0,2
+ 50% 0,45 10% 0,2
- 50% 0,15 10% 0,2
Cases

u +50% 0,3 15% 0,2

u -50% 0,3 5% 0,2
+ 50% 0,3 10% 0,3
- 50% 0,3 10% 0,1

7. Sensitivity Analysis

In order to set appropriate values for the deviation analysis parameters (δST , δSC and u), a sensitivity
 G. Martino et al. /Uncertain Supply Chain Management 5 (2017) 267
 

analysis was performed by changing one-factor-at-a-time. Table 12 shows the values used in the
analysis while Table 13 shows minimum, maximum and mean value of the percentage variation of the
KPIs. While technical KPIs were individually analysed, the economic performance is globally
evaluated through the Profit value.

Table 13
Results of the Sensitivity Analysis
SL %Sales OOS FA ITPOS ITW AS P
min 0% 0% 0% 0% 0% -1% 0% -5%
+
50%

max 0% 0% 1% 0% 0% 1% 0% 0%
mean 0% 0% 0% 0% 0% 0% 0% -1%
min 0% 0% -1% 0% 0% 0% 0% 0%
-
50%

max 0% 0% 0% 0% 0% 0% 0% 1%
mean 0% 0% 0% 0% 0% 0% 0% 0%
min 0% -1% -7% 0% -3% 0% 0% -1%
+50

max 2% 0% 1% 2% 0% 10% 5% 75%

u

mean 0% 0% -2% 1% -1% 4% 1% 15%

min -2% -1% -2% -3% -1% -9% -4% -73%
50%

max 0% 1% 7% 0% 3% -1% 1% 0%
u-

mean -1% 0% 1% -1% 1% -4% -1% -14%

min 0% 0% 0% 0% 0% 0% 0% 0%
+
50%

max 0% 0% 0% 0% 0% 0% 0% 0%
mean 0% 0% 0% 0% 0% 0% 0% 0%
min 0% 0% 0% 0% 0% 0% 0% 0%
-
50%

max 0% 0% 0% 0% 0% 0% 0% 0%
mean 0% 0% 0% 0% 0% 0% 0% 0%

It is clear from the result that the most critical value is u which causes the highest percentage variation
in all the KPIs. The detailed results of the analysis for the percentage of variation in the operations
plans (u) are shown in Fig. 11: increasing u by 50% will improve Supply Chain performances and in
particular will increase Profit by approximately 70%. This value (u = 15%) will then be used for all
the next simulations.

Fig. 11. Results of the sensitivity analysis for the parameter u (percentage of variation of the operations plans)

8. Analysis of the results

From the analysis of the KPIs in the six different scenarios it emerged that:

 for all the six scenarios in the As Is case, costs for Primary Transport (Fig. 12(a)), Secondary
Transport (Fig. 12(b)), Purchase (Fig. 13(a) a) and Warehouse Management (Fig. 18(b) a)
are constant. This is due to the fact that operations plans are defined according to demand
forecasts and do not change in the different simulated scenarios; 

268

 Transport Costs (Fig. 12(a) and 12(b)) may result higher than the As-is case in some
scenarios (in particular where a peak in demand - Scenario 2 - or a higher uncertainty -
Scenario 3 - occur). This is due to the fact that, in order to cope with market demand, the
company needs to buy (Fig. 13(a) - Purchase Cost) and consequently deliver greater
quantity;

(a) (b)

Fig. 12. Comparison of the results for the Primary and Secondary Transport Costs

(a) (b)
Fig. 13. Comparison of the results for Purchase Cost and Service Level

 despite the number of Out of Stock and related cost (Fig. 14) are very high for Scenario 2,
3 and 4, the Service Level is always higher than the As-Is case; 


(a) (b)

Fig. 14. Comparison of the results for Out of Stock and OOS Cost

 from the stores perspective, Fig. shows that Shelf Availability (Fig. 16(a) a) increases up to 14% in
Scenario 2 even if Stores Management Cost stays almost constant (Fig. 15(a)). This means that the
average stock in the stores is higher on average, as demonstrated by the slightly lower values in the
inventory Turnover (Fig. 15(b));
 G. Martino et al. /Uncertain Supply Chain Management 5 (2017) 269
 

(a) (b)
Fig. 15. Comparison of the results for Management Cost and Inventory Turnover for Stores

 Sales Percentage (Fig. 16(b)) is strictly connected to the stores performance and Revenues
(Fig. 17(a)). This value is, in fact, given by the ratio between sales and quantities delivered
to the stores. As demonstrated by the higher value in Revenues (Scenario 2 and 3), sales
increase in the To Be case. This implies that the quantity delivered to the stores are higher,
as confirmed by the lower value in the Stores Inventory Turnover;

(a) (b)
Fig. 16. Comparison of the results for Shelf Availability and Sales Percentage

 thank to the adjusting process, also the Forecasting Accuracy (Fig. 17(b)) results 
slightly
higher in the proposed model, since it is able to follow market changes; 


(a) (b)
Fig. 17. Comparison of the results for Revenues and Forecasting Accuracy

 both technical and economic performances of the central warehouse improve when
introducing the proposed model: Inventory Turnover (Fig. 18(a)) increases at least of 5%
compared to the As Is case and Warehouse Management Cost (Fig. 18(b)) is always lower;
270

(a) (b)
Fig. 18. Comparison of the results for Inventory Turnover and Management Cost for Warehouse

 from a global economic perspective, the proposed model (To Be) guarantees better

performances. The Profit is, in fact, always higher than the As-Is case and in particular it
is more than 5 times greater in the Scenario 2, that represents an unexpected peak in demand
(Fig. 19). This implies that the proposed model is able to follow demand variations during
the sales season by adjusting operations plans according to that. This behaviour is also
confirmed by the number of Out of Stock (Fig. 14 a), that is on average lower than the As-
Is case. 
9. Conclusions 
The coordination and management of the Supply Chain has always
been a challenging task especially in case of retailing and in such fast changing conditions
as the fashion and apparel Industry. In this context, this paper presents a comprehensive and
flexible model for the definition of the physical and informative flows from the development
of the collection by the styling office until the withdrawal of unsold items from stores
network (As-Is case). The model also proposes a responsive approach which, during the
sales season, analyses actual market demand and adjust operations plans according to it. The
proposed model was implemented in Excel and used to compare As-Is and To-Be cases for
an Italian Fashion company working with a dense network of direct- operated and
franchising mono-brand stores. Results show that the proposed adjusting process guarantees
improved performances both in economic terms (Profit is always higher than the As-Is case
especially when a peak in demand occurs) and in terms of customer satisfaction, with an
increased Service level. This work, then, may represent a useful Decision Support System
for the optimisation of performances of all supply chain’s actors thank to the ability of
quickly following changes in market demand.

Fig. 19. Comparison of the results for the Profit

 G. Martino et al. /Uncertain Supply Chain Management 5 (2017) 271
 

References

Agatz, N. A., Fleischmann, M., & Van Nunen, J. A. (2008). E-fulfillment and multi-channel
distribution–A review. European Journal of Operational Research, 187(2), 339-356.
Bandinelli, R., Codrino, A. & Terzi, S. (2011). Il plm nel settore della moda.
Bandinelli, R., Rinaldi, R., Rossi, M., & Terzi, S. (2013). New product development in the fashion
industry: an empirical investigation of Italian firms. International Journal of Engineering Business
Management, 5, 91-99.
Barnes, L., & Lea-Greenwood, G. (2010). Fast fashion in the retail store environment. International
Journal of Retail & Distribution Management, 38(10), 760-772.
Battista, C., Falsini, D., Scarabotti, I., & Schiraldi, M. M. (2011, September). Quantifying shelf-out-
of-stock in fashion & apparel retail stores. In Proceedings of the Conference “Breaking down the
Barriers between Research and Industry.
Battista, C., & Schiraldi, M. M. (2013). The logistic maturity model: Application to a fashion
company. International Journal of Engineering Business Management, 5.
Berman, B., & Thelen, S. (2004). A guide to developing and managing a well-integrated multi-channel
retail strategy. International Journal of Retail & Distribution Management, 32(3), 147-156.
Bini,V. (2011). La Supply Chain della moda - Strumenti per la gestione globale dell’impresa: dallo
sviluppo del prodotto al negozio.
Bruce, M., Daly, L., & Towers, N. (2004). Lean or agile: a solution for supply chain management in
the textiles and clothing industry?. International journal of operations & production
management, 24(2), 151-170.
Universita’ Carlo Cattaneo LIUC Centro di Ricerca sulla Logistica (2011). On shelf availability: la
nuova sfida del settore del largo consumo. Logistica Management.
Chan, F. T., & Chan, H. K. (2010). An AHP model for selection of suppliers in the fast changing fashion
market. The International Journal of Advanced Manufacturing Technology, 51(9-12), 1195-1207..
Christopher, M., Lowson, R., & Peck, H. (2004). Creating agile supply chains in the fashion
industry. International Journal of Retail & Distribution Management,32(8), 367-376.
Council, S. C. (2008). Supply-chain operations reference-model. Overview of SCOR version, 11(0).
d'Avolio, E., Bandinelli, R., Pero, M., & Rinaldi, R. (2015). Exploring replenishment in the luxury
fashion Italian firms: evidence from case studies. International Journal of Retail & Distribution
Management, 43(10/11), 967-987.
De Carlo, F., Borgia, O., & Tucci, M. (2013). Bucket brigades to increase productivity in a luxury
assembly line. International Journal of Engineering Business Management, 5.
Elia, V., Gnoni, M.G. & Lanzilotto, A. (2014) Designing and managing operations in innovative multi-
channel retailing systems. XVI Summer School Francesco Turco Proceedings - Senigallia (AN) -
Italy.
Fera, M., & Macchiaroli, R. (2009). Proposal of a quali-quantitative assessment model for health and
safety in small and medium enterprises. WIT Transactions on the Built Environment, 108.
Fera, M., & Macchiaroli, R. (2010). Use of analytic hierarchy process and fire dynamics simulator to
assess the fire protection systems in a tunnel on fire.International Journal of Risk Assessment and
Management, 14(6), 504-529.
Fera, M., Fruggiero, F., Lambiase, A., Macchiaroli, R., & Miranda, S. (2017). The role of uncertainty
in supply chains under dynamic modeling. International Journal of Industrial Engineering
Computations, 8(1), 119-140.
Fernie, J., & Azuma, N. (2004). The changing nature of Japanese fashion: can quick response improve
supply chain efficiency?. European Journal of Marketing, 38(7), 790-808.
Forza, C., & Vinelli, A. (2000). Time compression in production and distribution within the textile-
apparel chain. Integrated manufacturing systems, 11(2), 138-146.
Iannone, R., Ingenito, A., Martino, G., Miranda, S., Pepe, C. & Riemma,S. (2013) Merchandise and
replenishment planning optimisation for fashion retail. International Journal of Engineering and
Business management, 5.
 272

Iannone, R., Martino, G., Miranda, S., & Riemma, S. (2015). Modeling Fashion Retail Supply Chain
through Causal Loop Diagram. IFAC-PapersOnLine, 48(3), 1290-1295.
Lanzilotto, A., Martino, G., Gnoni, M.G. & Iannone,R. (2014) Impact analysis of a cross-channel
strategy in the fashion retail industry: a conceptual framework. XVI Summer School Francesco
Turco Proceedings - Senigallia (AN) - Italy.
Lanzilotto, A., Martino, G., Gnoni, M.G. & Iannone,R. (2015) Impact analysis of a cross-channel
retailing system in the fashion industry by a simulation approach. International Conference on
Modeling and Applied Simulation 2015.
Martino, G., Fera, M., Iannone, R. & Miranda,S. (2016) Proposal of a multi-method decision support
system for the fashion retail industry. IT4Fashion 2016 Proceedings Firenze - Italy.
Martino, G., Fera, M., Iannone, R., Sarno, D. & Miranda, S. (2015) Risk identification map for a fashion
retail supply chain. XIX Summer School Francesco Turco Proceedings Napoli - Italy.
Martino, G., Yuce, B., Iannone, R., & Packianather, M. S. (2016). Optimisation of the replenishment
problem in the Fashion Retail Industry using Tabu-Bees algorithm. IFAC-PapersOnLine, 49(12),
1685-1690.
Masson, R., Iosif, L., MacKerron, G., & Fernie, J. (2007). Managing complexity in agile global fashion
industry supply chains. The International Journal of Logistics Management, 18(2), 238-254.
Newman, A. J., & Foxall, G. R. (2003). In-store customer behaviour in the fashion sector: some
emerging methodological and theoretical directions.International Journal of Retail & Distribution
Management, 31(11), 591-600.
Nuttle, H. L. W., King, R. E., & Hunter, N. A. (1991). A stochastic model of the apparel-retailing
process for seasonal apparel. Journal of the Textile Institute,82(2), 247-259.
Şen, A. (2008). The US fashion industry: a supply chain review. International Journal of Production
Economics, 114(2), 571-593.
Sloot, L. M., Verhoef, P. C., & Franses, P. H. (2005). The impact of brand equity and the hedonic level
of products on consumer stock-out reactions.Journal of Retailing, 81(1), 15-34.
Gary Teng, S., & Jaramillo, H. (2005). A model for evaluation and selection of suppliers in global
textile and apparel supply chains. International Journal of Physical Distribution & Logistics
Management, 35(7), 503-523.
Xiao, T., & Jin, J. (2011). Coordination of a fashion apparel supply chain under lead-time-dependent
demand uncertainty. Production Planning & Control, 22(3), 257-268.
Vaagen, H., & Wallace, S. W. (2008). Product variety arising from hedging in the fashion supply
chains. International Journal of Production Economics,114(2), 431-455.
Walters, D. (2006). Effectiveness and efficiency: the role of demand chain management. The
International Journal of Logistics Management, 17(1), 75-94.
Wang, K., Gou, Q., Sun, J., & Yue, X. (2012). Coordination of a fashion and textile supply chain with
demand variations. Journal of Systems Science and Systems Engineering, 21(4), 461-479.
Zeppetella, L., Gebennini, E., Grassi, A., & Rimini, B. (2016). Optimal production scheduling with
customer-driven demand substitution. International Journal of Production Research, 1-15.

© 2016 by the authors; licensee Growing Science, Canada. This is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC-BY) license (http://creativecommons.org/licenses/by/4.0/).

View publication stats