Demand Forecasting in The Fashion Industry: A Review

Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

ARTICLE

International Journal of Engineering Business Management


Special Issue on Innovations in Fashion Industry

Demand Forecasting in
the Fashion Industry: A Review
Regular Paper

Maria Elena Nenni1,*, Luca Giustiniano2 and Luca Pirolo2


1 University of Naples Federico II
2 University LUISS Guido Carli
* Corresponding author E-mail: [email protected]

Received 1 June 2013; Accepted 15 July 2013

DOI: 10.5772/56840

© 2013 Nenni et al.; licensee InTech. This is an open access article distributed under the terms of the Creative
Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use,
distribution, and reproduction in any medium, provided the original work is properly cited.

Abstract Forecasting demand is a crucial issue for driving Typically, high performance companies focus on robust
efficient operations management plans. This is especially demand forecasting approaches; however, the challenge
the case in the fashion industry, where demand of demand forecasting varies greatly according to
uncertainty, lack of historical data and seasonal trends company and industry. In the fashion industry, products
usually coexist. Many approaches to this issue have been are usually characterized by long replenishment lead
times, short selling seasons and nearly unpredictable
proposed in the literature over the past few decades. In
demand and therefore, inaccurate forecasts [1]. All these
this paper, forecasting methods are compared with the
features make the issue of forecasting demand
aim of linking approaches to the market features.
particularly challenging. Companies in the fashion
industry have been trying to manage the demand for
Keywords Demand Forecasting, Fashion, Supply many years, which has brought about the development of
a number of specific forecasting methods and techniques.

1. Introduction Much of this earlier work was intended to create insights


and tools for improving the demand forecasting of
Demand forecasting plays an important role in basic fashion products. However, the reality that is now
Operations Management as an input for planning gradually being accepted both by those who work in the
activities. Poor forecasting effects are stock outs or high industry and those who research forecasting is that the
inventory, obsolescence, low service level, rush orders, demand for fashion products cannot be forecast. Instead,
inefficient resource utilization and bullwhip propagating we need to recognize that fashion markets are complex
through the upstream supply chain. As such, demand open systems that frequently demonstrate high levels of
forecasting is a popular research topic and many models ‘chaos’. In such conditions, managerial efforts may be
better expended on devising strategies and structures that
for forecasting fashion products have been proposed in
enable products to be created, manufactured and
the literature over the past few decades.
delivered on the basis of ‘real‐time’ demand [2].

www.intechopen.com Maria Elena Nenni, Luca Giustiniano and Luca Pirolo: Demand Forecasting
Int. j.ineng.
the bus.
Fashion Industry:
manag., 2013,A Review
Vol. 5, 1
Special Issue Innovations in Fashion Industry, 37:2013
The difficulty in predicting demand has led companies to fashion [6]. Moreover, in the fashion industry,
focus on the improvement of the supply chain [3, 4] and customer tastes change dynamically and their
the traditional newsvendor‐type overage‐underage trade‐ expectations are varied [7, 8]. Companies are
off. This is one of the factors in the success of brands such expected to meet the requirements of the customer on
as H&M and Zara, which have the shortest market lead both a product and service level. For instance, fast
times. Over the past few years, fashion companies have fashion has increased its share in the apparel market,
worked on strategies and inventory, and the framework as customers expect greater variety and frequent
of the industry has been radically changed. design changes [9]. Retail stores have a critical role in
creating brand image and have an influence on
The research question that therefore arises is: does it customer satisfaction [10]. Finally, service level
make sense to continue to study demand forecasting? quality, as noted by Rayman et al. [11] is one of the
What are the approaches and methods that may be more major parameters for customer satisfaction.
fitting with the actual context?  Technology: has impacted the fashion industry in
many ways and has contributed to the increase in
The purpose of this paper is to discuss the actual state of customer requirements through a dramatic rise in the
the context in the fashion industry compared with the instantaneous knowledge of new trends and brands.
demand forecasting approaches developed over the last Additionally, it has improved the capability of
few years. The most important literature on forecasting retailers, wholesalers and manufacturers for sharing
demand of recent years is confronted with recent findings data and making better business decisions. Out of this
on strategies of the main supply and demand‐oriented came two important advances: (i) Automatic
firms in the fashion industry and on consumer behaviour. replenishment, allowing manufacturers to
The aim is to understand which forecasting methods are automatically ship goods when inventory falls below
more effective under the current conditions. agreed upon thresholds and (ii) Value Chain Initiative
(VCI), where standardized codes and linkage have
The main expected findings from the literature review been developed throughout the industry [12].
will aim to propose a new framework for forecasting
demand. As a result of the above‐mentioned factors, the fashion
industry is now synonymous with rapid change and the
The paper is structured as follows: section 2 is devoted to organization’s flexibility and responsiveness [13]. Recent
analysing the actual supply chain features and firm trends across the fashion industry are:
context and what have been the main motivating factors  Vertical integration and outsourcing: the traditional
of change. Section 3 introduces and discusses in depth the chain of suppliers to manufacturers to retailers to
features of demand. Section 4 yields an analysis of the consumers is blurred. Many companies experience
main works in the literature presented. Section 5 vertical integration in an effort to increase efficiency,
summarizes the findings and provides some final eliminate intermediaries and better understand
remarks. Finally, section 6 proposes a new conceptual consumer needs. Other manufacturers choose to
framework for forecasting demand and discusses some have all production outsourced in order to gain a
open issues. competitive advantage [14]. Manufacturers and
retailers have also recognized that cooperation leads
2. Development in the global fashion industry to quicker product development, production and
distribution and higher profits.
The fashion industry has been in a state of transition over  Agile Supply Chain: an agile organization
the past 20 years [5] due to pressure from many factors: embedded within an agile supply chain performs
 Needs for reducing cost: as with many other better than conventional organizational structures
industries, fashion has needed to reduce its cost base and forecast‐driven supply chains that are not
to increase competitive advantage; the main result adequate for meeting the challenges of the fashion
of this tendency has been to buy materials and move industry [15‐17].
production to developing nations where production  Quick Response: in order to reduce the inventory
is cheaper due to lower labour cost. level, fashion retailers have adopted various
 Globalization: in terms of both production and measures such as the quick response policy [18–21].
retailing, this is a growing trend in many companies Responsiveness is the ability to scale up (or down)
for gaining a substantial cost advantage. However, quickly and the rapid incorporation of consumer
to source product and materials off‐shore has in preferences into the processes of a supply chain.
many cases led to significantly longer lead‐times.
 Increase of customer requirements: the market has Relevant from the last trend is the recognized capability
evolved from mass fashion into the segmented of Quick Response to counter the negative impacts of

2 Int. j. eng. bus. manag., 2013, Vol. 5, www.intechopen.com


Special Issue Innovations in Fashion Industry, 37:2013
uncertainty. Ideally, a Quick Response system would Obviously, the categorization of alternative demand
enable the manufacturer to adjust the production of patterns facilitates the selection of a forecasting method.
different styles, colours and sizes in response to retail In addition to the general features of the products,
sales during the season; responsiveness can then be used Varghese and Rossetti [27] have proposed classifying
to effectively substitute forecasting ability and/or demand according to the following attributes:
inventory required for operating under uncertainty.  Smoothness: demand is quite stable.
 Intermittence: intermittent demand is generally
3. Analysis of the products defined as demand occurring randomly with many
time periods with zero demands. Johnston et al. [28]
Many authors [22, 23] agree on characteristics typically propose that if the mean interval between non‐zero
exhibited by products in the fashion industry. demands is 1.25 times greater than the inventory
 Short life‐cycles: the product is often ephemeral, review period, the demand series could be
designed to capture the mood of the moment; considered as intermittent.
consequently, the period in which it will be saleable  Lumpiness: the feature of tending to have periods of
is likely to be very short and seasonal, measured in very low or zero demand and then spikes of
months or even weeks. demand. A lumpy demand is variable, sporadic and
 Short selling season: today’s fashion market place is nervous [29].
highly competitive and the constant need to  Erraticness: demand is described as patterns with
‘refresh’ product ranges means that there is an high variability in non‐zero demands [30].
inevitable move by many retailers to extend the Erraticness relates to the demand size rather than
number of ‘seasons’, i.e., the frequency with which demand per unit time period.
the entire merchandise within a store is changed. In  Slow‐moving: demand is usually defined as having
extreme cases, typified by the successful fashion infrequent demands, which occur in very few units
retailer Zara, there might be twenty seasons in a [30]. Slow demands are usually intermittent demands.
year. The implications of this trend for supply chain
management are clearly profound. Most of the authors agree on the features of products or
 Long replenishment lead times. supply chain in the fashion industry. On the contrary
there is no largely recognized link between demand in
Regarding the features of the demand: fashion and a specific attribute or pattern. Only
 High impulse purchasing: many buying decisions Bartezzaghi [31] hazards a guess on the basis of
by consumers for these products are made at the correlation as a cause of lumpiness. Correlation may be,
point of purchase. In other words, the shopper, for example, due to imitation in fashion, which will lead
when confronted with the product is stimulated to to sudden peaks in demand.
buy it; hence the critical need for ‘availability’.
 High volatility: demand for these products is rarely A DI=1,32
stable or linear. It may be influenced by the vagaries
of weather, films, or even by pop stars and
footballers. There are numerous sources of Erratic but not
uncertainty in a fashion supply pipeline, starting very intermi ent v
Lumpy
with demand through to the reliability on the part of
CV 2=0,49
suppliers and shippers, etc.
 Low predictability: because of the volatility of Intermi ent but
Smooth not very erratic
demand it is extremely difficult to forecast with any
accuracy even total demand within a period, let
alone week‐by‐week or item‐by‐item demand.
 Tremendous product variety: demand is now more
Figure 1. Items demand attributes.
fragmented and the consumer more discerning
about quality and choice.
As a result, many authors [30, 32, 33] propose different
 Large variance in demand and high number of stock
sets of indicators in order to classify demand. The most
keeping units: as a result, the volume of sales per
popular is based on two cut‐off parameters:
SKU is very low [24] and demand for SKUs within
 Average inter‐Demand Interval (ADI): measures the
the same product line can vary significantly [25].
average number of time periods between two
Thus, even if aggregate demand can be predicted
successive demands.
with some certainty, it is very difficult to predict
 Coefficient of Variation (CV): represents the
how that demand will be distributed over the many
standard deviation of period requirements divided
products that are offered [26].
by the average period requirements.

www.intechopen.com Maria Elena Nenni, Luca Giustiniano and Luca Pirolo: Demand Forecasting in the Fashion Industry: A Review 3
ADI and CV distinguish the different attributes as in Though somewhat dated, the most interesting
Figure 1. Recognizing the demand pattern is useful in contribution comes from Bartezzaghi [48]. In his paper,
order to select the forecasting technique. the author includes the main causes of demand
lumpiness as:
4. Analysis of forecasting approaches  High numerousness of potential customers
 High heterogeneity of customers
Demand forecasting is one of the biggest challenges for  Low frequency of customer requests
retailers, wholesalers and manufacturers in any industry,  High variety of customer requests
and this topic has received a great deal of attention from  High correlation between customer requests
both researchers and practitioners. The question is
whether the forecasting approaches are applicable and It is easy to verify that the above‐mentioned features are
useful within the fashion industry. common in the fashion industry and it is logical to
presume that the main attribute of demand in the fashion
Traditional forecasting methods, such as exponential industry is in fact lumpiness.
smoothing [34], are designed for smooth, high‐volume
demand and don’t work well with intermittent, erratic or Bartezzaghi [49] even proposes two approaches for
lumpy demand. managing uncertainty typically present in lumpy demand:
 Early sales: this method exploits information from
There are many papers that propose the use of statistical actual orders that have already been received for
methods in order to forecast demand [35]. This first future delivery. Making Bayesian use of information
group includes the extension of standard methods and from actual orders already received provides some
variants of the Poisson model [36], a model based on degree of correlation between the unknown and
binomial distribution [37], as well as Croston’s model and known portions of the demand.
its variants [28, 38, 39] and bootstrap methods [27]. Many  Order over‐planning: another approach for
authors [40, 41] have compared models and the general anticipating future lumpy requirements is to exploit
consensus is that performance should vary significantly the early information that a customer generates
according to the level of attributes. Particularly, if the during his purchasing process before he places his
demand pattern has a high level of lumpiness or actual order. Order over‐planning uses as
erractiness, which is likely in fashion demand and often forecasting unit each single customer order instead
causes poor performance with statistical methods. of the overall demand.

Moreover, Gutierrez et al. 2008 [42] clearly demonstrated 5. Final remarks and future research
that traditional time‐series methods may not always
capture a nonlinear pattern in data. Expert systems, such Lesson learned from both the context analysis and
as an Artificial Neural Network (ANN), are a logical literature review proposed in this paper is that there are
choice for overcoming these limitations. many different methods for and approaches to forecasting.
However, product and supply chain features of the fashion
Many authors have obtained impressive results through industry remain dominant factors. It is therefore not
ANN [43, 44] and we can even count some interesting surprising that the most famous brands have decided to
applications to fashion demand among these [45‐47]. focus on improving their supply chain performance. This
However, the same authors found that while the ANN does not mean that Zara or H&M do not forecast demand;
model can yield accurate forecasts, the required more likely, they instead rely on marketing approaches.
forecasting time can be a large barrier to its real‐world Main barriers in forecasting demand are:
applications. This is because the training time required by  Short selling seasons
ANN strongly increases according to the complexity or  Level of uncertainty (lumpiness)
variety of the data. This limitation renders it impractical  Lack of historical data
with the feature of the short selling season in the fashion
industry and the requirement of responsiveness, too. The last barrier results from the level of product
innovation that can be found at t each season in fashion.
The last group of papers we have analysed discusses That is the main reason for the considering the order
various techniques in managing the level of uncertainty overlapping method. The idea to use the customer as a
[48, 49, 30]. Such papers focused on the development of a forecasting unit has the power to overcome this problem.
single algorithm or framework and attempted to measure Under the hypothesis of a relatively stable set of
the performance of such a framework against existing customers, the historical series are then populated.
ones, often through a simulative approach.

4 Int. j. eng. bus. manag., 2013, Vol. 5, www.intechopen.com


Special Issue Innovations in Fashion Industry, 37:2013
From this observation arises the starting point for future [8] Battistoni E., Fronzetti Colladon A., Mercorelli G.
research. Having a more populated historical series is a (2013). Prominent determinants of consumer based
key point for facilitating the use of a range of effective brand equity. International Journal of Engineering
forecasting methods. The basic idea is then to use the Business Management.
features of products (colour, size, etc.) that are repetitive [9] Chan F.T.S., Chan H.K., (2010). An AHP model for
in each season as forecasting units. In fact, Zara has just selection of suppliers in the fast changing fashion
demonstrated that colour is a more important feature market. International Journal of Advanced
than model or type of clothes [50]. Expected advantages Manufacturing Technology, 51, pp.1195–1207.
of this idea are: [10] Shubhapriya B., Byoungho J., (2012). A conceptual
 Decrease the level of lumpiness through an effective process of implementing quality apparel retail store
choice of product features attributes: An application of Kano’s model and the
 Improve performance of statistical forecasting methods quality function deployment approach. International
Journal of Business, Humanities and Technology, 2
Main limits are likely to be: (1), pp.174‐183.
 After forecasting a specific feature, how can we [11] Rayman D., Burns D.J., Nelson C.N., (2011). Apparel
translate it into a product forecast? The problem product quality: its nature and measurement. Journal of
then becomes distributing an expected demand of Global Academy of Marketing Science, 21 (1), pp.66‐75.
features among products that possess them. One [12] D’Amico S., Giustiniano L., Nenni M.E., Pirolo L.,
answer could be a Bayesian approach. (2013). Product Lifecycle Management as a tool to
 The performance should vary significantly on a case by create value in the fashion system. International
case basis. Therefore, an extensive validation campaign Journal of Engineering Business Management.
could be conducted in order to fine‐tune the approach. [13] De Felice F., Petrillo A., Autorino C., (2013). Key
The risk is to fall into generalizing the approach. success factors for organizational innovation in the
fashion industry. International Journal of
The last issue should be to focus analysis on trend and Engineering Business Management.
cyclic nature of demand, using again the product features [14] Marchegiani L., Giustiniano L., Peruffo E., Pirolo L.,
in order to have more populated historical series. (2012). Revitalising the Outsourcing Discourse within
the Boundaries of Firms Debate. Business Systems
6. References Review, pp. 157‐177.
[15] Christopher M., Towill D., (2001). An integrated
[1] Minner S., Kiesmuller G.P., (2012). Dynamic Product model for the design of agile supply chains,
Aquisition in Closed Loop Supply Chains. International International Journal of Physical Distribution &
Journal of Production Research, 50, pp. 2836‐2851. Logistics Management, 31 (4), pp.235 – 246.
[2] Christopher M., (2004). Mitigating Supply Chain Risk [16] Battista C., Schiraldi M.M., (2013). The Logistic Maturity
Through Improved Confidence. International Journal Model: application to a fashion firm. International
of Physical Distribution & Logistics Management, 34 Journal of Engineering Business Management
(5), pp. 388‐396. [17] Iannone R., Pepe C., Ingenito A., Riemma S., Martino
[3] De Carlo F., Tucci M., Borgia O. (2013) Bucket G., Miranda S., (2013). Merchandise and
brigades to increase productivity in a luxury replenishment planning optimization for fashion
assembly line. International Journal of Engineering retail. International Journal of Engineering Business
Business Management. Management.
[4] De Carlo F., Arleo M.A., Tucci M., Borgia O. Layout [18] Iyer A.V., Bergen M.E., (1997). Quick Response in
design for a low capacity manufacturing line: a case Manufacturer‐Retailer Channels. Management
study. International Journal of Engineering Business Science, 43 (4), pp. 559‐570.
Management. [19] Lowson R., King R.E., Hunter N.A., (1999). Quick
[5] Frings G.S., (2005). Fashion: From Concept to response: Managing the supply chain to meet consumer
Consumer, Pearson Education. demand. Sussex, England: John Wiley and Sons.
[6] Sekozawa T., Mitsuashi H., Ozawa Y., (2011). One‐to‐ [20] Au K.F., Chan N.Y., (2002). Quick response for Hong
One recommendation system in apparel online Kong clothing suppliers: a total system approach.
shopping. Electronics and Communications in Japan, Proceedings of the 13th Annual Conference of the
94 (1), pp. 51‐60. Production and Operations Management Society, San
[7] Marufuzzaman M., Ahsan K.B., Xing K., (2009). Francisco, USA.
Supplier selection and evaluation method using [21] Choi T.M., (2006). Quick response in fashion supply
Analytical Hierarchy Process (AHP): a case study on chains with dual information updating. Journal of
an apparel manufacturing organisation. Int. J. Value Industrial and Management Optimization, 2, pp. 255‐
Chain Management, 3 (2), pp.224–240. 268.

www.intechopen.com Maria Elena Nenni, Luca Giustiniano and Luca Pirolo: Demand Forecasting in the Fashion Industry: A Review 5
[22] Lee H.L., (2002). Aligning Supply Chain Strategies [37] Cachon G., Fisher M., (2000). Supply chain inventory
with Product Uncertainties. California Management management and the value of shared information,
Review, 44 (3), pp. 105‐119. Management Science, 46(8), pp. 1032–1048.
[23] Soni G., Kodali R., (2010). Internal benchmarking for [38] Croston J.D., (1972). Forecasting and stock control for
assessment of supply chain performance. intermittent demands. Operational Research
Benchmarking: An International Journal, 17 (1), pp. Quarterly 23(3), pp. 289–303.
44 – 76. [39] Snyder R., (2002). Forecasting sales of slow and fast
[24] Gutgeld Y., Beyer D., (1995). Are you going out of moving inventories. European Journal of Operational
fashion? The McKinsey Quarterly 3, pp. 55–65. Research, 140, pp. 684–699.
[25] Abernathy F.H., Dunlop J.T, Hammond J.H., Weil D., [40] Willemain T.R., Smart C.N., Shocker J.H., DeSautels
(2000). Retailing and Supply Chains in the P.A., (1994). Forecasting intermittent demand in
information age. Technology in society, 22, pp. 5‐31. manufacturing: a comparative evaluation of
[26] Mostard J., Teunter R., de Koster R., (2011). Croston’s method. International Journal of
Forecasting demand for single‐period products: A Forecasting, 10, pp. 529–538.
case study in the apparel industry. European Journal [41] Gutierrez R.S., Solis A.O., Bendore N.R., (2004).
of Operational Research, 211(1), pp. 139‐147. Lumpy Demand Characterization and Forecasting
[27] Varghese V., Rossetti M.D., (2008). A Parametric Performance: An Exploratory Case Study, WDIS 2004
Bootstrapping Approach to Forecast Intermittent Proceedings.
Demand. Industrial Engineering Research Conference [42] Gutierrez R. S., Solis A., Mukhopadhyay S. (2008).
Proceedings, May 17‐21, 2008, Vancouver, Canada. Lumpy Demand Forecasting Using Neural
[28] Johnston F.R., Boylan J. E., (1996). Forecasting for Networks. International Journal of Production
Items with Intermittent Demand. The Journal of the Economics, 111, pp. 409‐420.
Operational Research Society, 47 (1), pp. 113‐121. [43] Chang P.C., Wang Y.W., Liu C.H., (2007). The
[29] Syntetos A.A., Boylan J.E., (2005). The accuracy of development of a weighted evolving fuzzy neural
intermittent demand estimates. International journal network for PCB sales forecasting. Expert Systems
production economics, 21, pp. 303‐314. with Applications, 32(1), pp. 86 96.
[30] Syntetos A.A., Boylan J.E., (2001). On the bias of [44] Ling S.H., (2010). Genetic Algorithm and Variable
intermittent demand estimates. International Journal Neural Networks: Theory and Application, Lambert
of Production Economics, 71, pp. 457–466. Academic Publishing, German.
[31] Bartezzaghi E., Kalchschmidt M., (2011). The impact [45] Yu Y., Choi T., Hui C., (2011). An intelligent fast sales
of aggregation level on lumpy demand management. forecasting model for fashion products. Expert
In: Altay N., Litteral L. A., Service Parts Management Systems with Applications, 38(6), pp. 7373–7379.
Demand Forecasting and Inventory Control, pp. 89‐ [46] Au K.F., Choi T.M., Yu Y., (2008). Fashion retail
104. Springer‐Verlag. forecasting by evolutionary neural networks.
[32] Eaves A, (2002). The forecasting for the ordering and International Journal of Production Economics,
stock holding of consumable spare parts. 114(2), pp. 615–630.
Unpublished PhD thesis, Lancaster University, UK. [47] Sun Z.L., Choi T. M., Au K. F., Yu Y. (2008). Sales
[33] Ghobbar A.A., Friend C.H., (2003). Evaluation of forecasting using extreme learning machine with
forecasting methods for intermittent parts demand in applications in fashion retailing. Decision Support
the field of aviation: a predictive model. Computers Systems, 46(1), pp. 411–419.
& OR 30(14), pp. 2097‐2114. [48] Bartezzaghi E., Verganti R., (1995). Managing
[34] Brown R.G., (1959). Statistical Forecasting for demand uncertainty through order overplanning.
Inventory Control. McGraw‐Hill, New York. International Journal of Production Economics, 40 (2–
[35] Fumi A., Pepe A., Scarabotti L., Schiraldi M.M., 3), pp. 107–120.
(2013). Fourier analysis for demand forecasting in [49] Bartezzaghi E., Verganti R., Zotteri G., (1999). A
fashion company. International Journal of simulation framework for forecasting uncertain
Engineering Business Management. lumpy demand. International Journal of Production
[36] Wang H.J., Chien C., Liu C., (2005). Demand Economics, 59 (1–3), pp. 499–510.
Forecasting Using Bayesian Experiment with Non‐ [50] McAfee A., Dessain V., Sjoman A., (2007). Zara: IT for
homogenous Poisson Process Model. International Fast Fashion. Harvard Business School, 9, pp. 1‐23
Journal of Operations Research, 2 (1), pp. 21−29.

6 Int. j. eng. bus. manag., 2013, Vol. 5, www.intechopen.com


Special Issue Innovations in Fashion Industry, 37:2013

You might also like