020-0724

Reviewing Supply Chain Strategy - a longitudinal case study

Roy Stratton, NottinghamTrentUniversity, Burton Street, Nottingham, NG14BU, UK

Tel: +441158488689

POMS 22nd Annual Conference

Reno, Nevada, U.S.A.

April 29 to May 2, 2011

Abstract

Markets are increasingly characterised by demand uncertainty and short product life cycles that are often exacerbated by supply shifting to low cost global sources. The effect of these two changes is the growing importance of a cost versus response trade-off, acutely felt in the apparel industry (Fisher, 97; Lowson, 2002; Lee, 2002; Stratton,2003, Sun et al., 2009). However, when management is confronted by such transitions the implications of poor response are often only belatedly acknowledged and addressed. This paper reports on a longitudinal case study where the trade-off implications of such a transition initially resulted in significant supply instability. The study explores how and why stability was re-established through product and process design changes with particular reference to the systematic causes of the management inertia. This case is then used to explore how existing supply chain strategy models (Fisher, 97; Lee, 2002) support such decision making.

1.0Introduction

The term supply chain management (SCM) was originally introduced in the early 1980s (Oliver and Webber, 1992), however, the concept of a supply chain is a natural extension of production operations management and there are common underlying theoretical developments (Schmenner and Swink, 1998). Central to these is the need to control variation (Shewart, 1931)which rapidly progressed into an awareness of the implications of variation on the wider production system (Deming, 1986) in improving overall supply chain performance. The quality revolution that developed in Japan quickly merged with the Toyota Production System (Ohno, 1988) with what we now know as lean supply.This led to a flow focus, level scheduling,close supplier relationships and the reduction of wasteful variationthat drivesinventory and capacity buffering.

Rather than directly reducing the wasteful variation in the supply chain westernattention wasdirected at understanding how to manage the supply dynamics associated with mulit-echelon information and material flows (Forester, 1961). The need to manage such variation and uncertainty resulted in the emergence of the concept of manufacturing strategy (Skinner, 1969) and with it the need to focus (Skinner, 1974; Hill, 1985; Porter, 1987) in making strategic choices. The supply chain offers the opportunity to focus differently across the supply chain through postponement(Zinn and Bowersox, 1988; Van Hoek, 1998; Olhanger, 2003) and mass customization(Pine, 1993; Fitzingler and Lee, 1997). These concepts effectively limit the impact of the demand variation and uncertainty through postponing the introduction of customized design features (Walker et al., 2000). In this way both lean and agile systems can be associated with the same supply chain (Nailor et al., 1999).

1.1Supply chain strategy models

Our understanding of how these concepts and theoretical developments impact supply chains have been influenced by two particular publications (Fisher, 97 and Lee, 2002) which have been validated academically through various hypotheses tests (Selldin and Olhager, 2007; Sun et al., 2009). Such tests verify the general relationships but do not effectively evaluate the practical utility of the conceptual models from a management perspective. This paper aims to explore the practical utility of such models by using a longitudinal case study to identify the managerial issues and then evaluate the support offered by these models in comparison with a further model developed through multi-case research (Stratton, 2008).

2.Theoretical models in support practice

These threeconceptual models are described below.

2.1Fisher’s supply chain management model

Fisher (1997) used empirical case research to clarify the trade-off relationship between different classes of product with efficiency and response in a supply chain (Fisher et al., 1997). Figure 1 highlights the need to align the design of the supply chain with the uncertain nature of the product, embracing the concepts of uncertainty, trade-offs and buffering (capacity, inventory and customer tolerance time).

Fisher uses the model to convey the concept of performance trade-offs and clearly associates the supply chain choice of efficiency (therefore minimal buffering) with minimal demand variation and uncertainty associated with functional products. Whereas, demand uncertainty, associated with innovative products, is allied to the choice of a responsive supply chains. The top right mismatch zone effectively conveys the excess and shortage consequences of not suitably buffering in the case of innovative products. Fisher uses this model to stress the need to adopt three coordinated strategies.

  1. Strive to reduce uncertainty (e.g. timely demand data or common parts)
  2. Avoid uncertainty by cutting lead-times and increasing the supply chain flexibility so that it can produce ideally within the tolerance time of the customer.
  3. Once uncertainty has been reduced or avoided as much as possible, hedge against the remaining residual uncertainty with buffers of inventory or excess capacity.

(Fisher, 1997: 114)

This model and the associated strategies emphasise demand uncertainty rather than wider sources of variation, however, it implicitly encompasses the concept of postponement in the first of the three coordinated strategies. Again, the use of buffering is embodied in the third strategy, with the trade-off implications implicitly if not explicitly conveyed. Fisher’s model, therefore, implicitly links the key concepts of uncertainty, performance trade-offs and buffering mechanisms in a supply chain setting, but does not attempt to integrate this with production and other sources of variation and uncertainty. This model, therefore, is limited in its utility to SCM and the model is not presented in such a way as to support falsification.

2.2 Lee’s supply chain management model

Lee (2002) extended Fisher’s model to accommodate supply as well as demand uncertainty, claiming that the uncertainty associated with the supply chain also needs to be strategically managed. In addition to the demand uncertainty Lee maps the extremes of supply uncertainty identifying ‘stable processes’ and ‘evolving processes’ with low and high levels respectively. Lee identifies the continuous improvement need to adopt uncertainty reduction strategies as opposed to the trade-off management emphasized in Fisher’s model. To accommodate these different sources of uncertainty Lee (2002) proposes four viable supply chain strategies: efficient, responsive, risk-hedging and agile.

The efficient and responsive strategies are closely allied to Fisher’s model as there is low supply uncertainty. A risk-hedging strategy is allied to low demand and high supply uncertainty and in this situation he advocates managing the disruption through pooling and sharing resources in the form of inventory or capacity. The final strategy, agile supply, aims at being responsive and flexible to customer demand, while hedging the risk of supply uncertainties. Agile strategies are, therefore, for the most unstable environments, high demand and supply uncertainty. Lee leaves these strategies at this conceptual level although he discusses with examples the need to postpone and buffer with inventory and capacity. He finishes by stressing the need to devise the right strategy whilst also identifying the need to dynamically adjust and adapt. However this is not specifically addressed.

2.3 Variation & Uncertainty Buffering (VUB)Model

This model (Figure 3)was developed through multi-case research (Stratton, 2008) of which this case is one of six. This research identified key concepts of variation and uncertainty, performance trade-off and buffering mechanisms (capacity, inventory and forward load). Therefore, avoiding the use of complex constructs, such as flexible and agile, used in Lee’s model. ThisVUB model,in a similar way to Fisher (1997),incorporates threecoordinated generic strategies, however the definition of these strategies is more tightly defined around seven observed propositions linking the key constructs.

Figure 3 Variation and Uncertainty Buffering Model

The generic strategies were found to be present in all the cases investigated and the GS definitions are as follow.

GS 1Buffer variationand uncertainty

Variation and uncertainty in a supply chain drives the need for buffering and the mix of buffering mechanism (forward load, capacity and inventory) determines the performance trade-off.

GS 2Reduce variation and uncertainty

The reduction of variation and uncertainty reduces the need for buffering and the associated waste. The variation and uncertainty may be in demand, supply or internal processes (including set-up time).

GS 3Separate or postpone variation and uncertainty

The buffering requirements in size and form may be limited by separating or postponing the impact of variation and uncertainty on the supply chain.

The model shows these generic strategies need to be focused on limiting operations trade-offs or market priorities with the intention of reducing /aligning the buffering mechanisms, as illustrated in Figure 3.

The model embraces the concept of variation as well as uncertainty but emphasizes the buffering implications, therefore, the need to reduce and align the buffer choices with the market priorities. These generic strategies can be aligned to the classic paradigms discussed earlier but as the model illustrates they need to be uniquely coordinated.

3.0Research methodology

This apparel case study (Stevensons) comprises two echelons in the supply chain with the focal company supplying direct to a major high street retailer in the UK. The case study covered 3 years which involved 6 site visits together with a retail customer survey and customer interviews.In total nine senior level management interviews took place together with gathering a collection of multiple sources of data. Repeated visits enabled the collection of contemporary evidence and the development and testing of the causal relationships. The study concluded with the relocation of the garment dyeing facilities to a lower cost country in a joint venture.

This case formed part of a wider research project that gave rise of the VUB model being evaluated. The research questions centred on clarifying the key concepts/variables and their relationship to each other. Such research is clearly suited to focused multiple case research (Melnyk and Handfield, 1998; Voss et al., 2002). The approach adopted embraces analytic induction which is now a well established research approach (Yin, 1994) that Eisenhardt (1989) has developed into a staged theory building process to encourage rigor. This process was followed and commences with tentative propositions/hypotheses that are progressively developed through the focused analysis of cases. These cases were selected (Yin, 1994: 48) to both replicate and extend the boundaries of the emergent theory. The automotive, grocery and apparel sectors were chosen to reflect a cross section of supply chains that encompass distinct forms of variation and uncertainty.

The case analysis involved inductive analysis of transitions in the level of stability, with a particular focus on the trade-off performance associated with changes in the choice and level of buffering mechanism. Data was collected in line with a protocol using multiple sources of evidence. The data collection methods included plant observation, semi-structured interviews, archival records and documents, with due attention being given to triangulation and subsequent analysis (Miles and Huberman, 1994). Case data was collected in line with pre prepared semi structured interviews, followed up by archival data and observations allied to changes in transition.

The interview process was designed to take an overview of the immediate supply chain from the perspective of the company concerned before centering on any sustained transitions in variation, uncertainty and buffering mechanism. Such transitions provided the focus for data collection with reference to customer order winners and qualifiers, together with the nature and location of associated buffering mechanisms. Emphasis was placed on exploring changing trade-offs associated with the transition and the drivers and actions involved in reducing, mitigating and managing the trade-off implications.

4.0Stevensons case analysis

4.1Overview

Stevensons is a garment dying business that was originally vertically integrated with the Coats Viyella Group (CVG) in the UK providing the capability to postpone the colour choice in garment manufacture. The transition in stability occurred in 2001 when the Group dissolved and much of the work moved offshore, considerably extending the colour choice lead times with the move to yarn rather than garment dyeing. Stevensons downsized and developed its capability to both garment dye and finish within 10 days. The capability was, however, more expensive and the tension between fast response and low cost resulted in uncertain demand over the years, finally resulting the plant being closed and moved to Sri Lanka and Bangladesh in 2006. The case outlines the issues and the factors underlying the transitions concerning sourcing choices.

4.2Supply chain issues and analysis

Stevensons garment dyers occupies a 17-acre site in Derbyshire and is one of the largest garment dyeing facilities in Europe, combining expertise with the economies of a semi-automated plant which was part of a multi-million pound investment installed in the late 1990s. For several decades it was part of the vertically integrated Coats Viyella Group (CVG). Typical garment dyed products included knitwear, hosiery, and woven fabrics, but the technical requirements of knitwear had become their main focus of activity in recent years as this area of CVG business had proved more difficult to move offshore. Stevensons worked alongside local knitting and finishing factories in the design and development of garments as well as volume production. As CVG was a major supplier to high street retailers Stevensons had generally been made responsible for developing the seasonal colour pallet and recipes across their product ranges as well as supporting product development.

In 2001 CVGwithdrew from garment manufacture, closing or selling off its interests in knitwear. Stevensons annual sales quickly fell from £15M to £5M and their full time workforce was subsequently reduced from over 500 employees to a little over 100, in their fight to remain in business under a management buyout. By 2002 their major retail customers still depended on them for the colour pallet and recipe specification, but few regular production orders remained.

4.3Typical product and dyeing process

Typical garment dyed products are often solid (single) colours, but the garment dyeing process can be very sophisticated. For example, by pattern knitting ecru (undyed) yarns with different dye resist properties the cross dyeing process enables complex coloured patterns to be dyed into a garment. A wide range of fibres may be dyed in this way, including: lambs wool, acrylic, cashmere, cotton, and mohair.

In the case of knitwear, as the name suggests, garment or piece dyeing involves dyeing the garment after it has been knitted in the natural colour of the fibres, referred to as ecru. The alternative is to use pre-dyed yarn to knit the garment. The garment dyeing facility at Stevensons comprises a wide range of chemical treatment processes, but the main capability is embedded in the semi-automated dye mixing and dispensing to the dyeing machines.

4.4Original knitwear process route

Garment dyeing had advantages over yarn dyeing for both the supplier and the retailer and therefore had become a popular option when CVG was developing new products for their customers. Almost the first decision in the yarn dye manufacturing process is the colour choice, whereas with garment dyeing it is close to the last. Committing to a colour and quantity is a risk that worsens the further ahead you forecast therefore, it is important to minimise the lost sales or excess stock by postponing this decision as long as possible. This need to delay colour choice also puts pressure on the seasonal workload for yarn dyeing and subsequent knitting in an attempt to delay the colour choice and subsequent start of production. Therefore, there has been a constant tension over the years between the manufacturer attempting to efficiently smooth his capacity demand by insisting on longer lead-times and the retailer compressing the lead-times in order to postpone the colour decision. CVG was also commercially aware of this conflict as it was common practice for them to share the losses associated with excess stock at the end of a season with their retail customers. The neat advantage offered by garment dyeing was the partial separation of these two conflicting requirements. The knitting factory was able to start manufacturing garments as soon as the styles were agreed, typically much earlier, without having to wait for colouring decisions and dyed yarn to be supplied. This use of ecru yarn also enabled longer production runs and fewer set-ups, which improved the reliability of the process as changing to different coloured yarns introduces process variation and associated quality issues and delays. Also, with ecru yarn there is no risk of running out of specific colours and therefore yarn wastage and shortages are reduced.