Identifying the factors influencing the performance of reverse supply chains (RSC)

Abstract

This paper aims to extract the factors influencing the performance of reverse supply chains (RSCs) based on the structure equation model (SEM). We first introduce the definition of RSC and describe its current status and follow this with a literature review of previous RSC studies and the technology acceptance model . We next develop our research model and 11 hypotheses and then use SEM to test our model and identify those factors that actually influence the success of RSC. Next, we use both questionnaire and web‐based methods to survey five companies which have RSC operation experience in China and Korea. Using the 168 responses, we used measurement modeling test and SEM to validate our proposed hypotheses. As a result, nine hypotheses were accepted while two were rejected. We found that ease of use, perceived usefulness, service quality, channel relationship and RSC cost were the five most important factors which influence the success of RSC. Finally, we conclude by highlighting our research contribution and propose future research.

Keywords:

• reverse supply chains
• critical success factors
• structural equation modelling

1. Introduction

It is estimated that some 26 billion tons of materials are used annually world‐wide. Used materials represent an enormous amount of unclaimed value (Brown 2001). The current industrial production system not only extracts resources from the environment, but also returns pollutants to the same resource reservoir (Brown 2001). Recently, the effects of environmental degradation have gained attention and forced the global community to look towards building a more sustainable economy. Therefore, reverse supply chain (RSC) could play an important role. Modern businesses can combine the forward supply chain (FSC) with RSC to form a whole supply chain that creates improved value. Vlachos et al. (2007) defined RSC management (RSCM) as the effective and efficient management of the series of activities required to retrieve a product from a customer for either disposal or value recovery. The importance of studying RSCs has increased in recent years for several reasons (Kocabasoglu and Suresh 2007):

	RSC costs amounted to approximately $35 billion in 1997. The magnitude and impact of RSC vary by industry and channel position. They also vary depending on the firm's channel choice. However, it is clear that the overall amount of reverse logistics activities in the economy is large and still growing (Rogers and Tibben‐Lembke 2001). The amount of product returns can be very high, with some industries experiencing returns at over 50% of sales (Trebilcock 2002).
	The sales opportunities in secondary and global markets have increased revenue generation from previously discarded products (Corbett and Kleindorfer 2001).
	End‐of‐life take‐back laws have developed over the past decade both in the European Union and in the United States, requiring businesses to effectively manage the entire life of the product (Toffel 2004).
	Consumers have successfully pressured businesses to take responsibility for the disposal of their products that contain hazardous waste (Tony 2003).
	Landfill capacity has become limited and expensive.

Alternatives to disposal, such as repackaging, recycling and remanufacturing have become more prevalent and viable (Thierry et al. 1995). It is easy for us to think of another similar concept called reverse logistics. Generally speaking, reverse logistics is similar to but different from RSC. Reverse logistics could be a dispensable part of RSC considering that it links discarded products acquisition, product inspection and reuse. In most cases, reverse logistics focuses on reducing transportation costs and maximising profit. RSC, on the other hand, focuses on the overall management of the whole supply chain both forward and backward.

Most of the SCM researchers concentrate on the forward movement and transformation of the materials from the suppliers to the end consumer, and on the impact that information has on the bullwhip effect as it transverses upstream. However, the reverse flow of products from consumers to upstream businesses has not received much interest (Rogers and Tibben 2001). The majority of companies do not properly understand the value of returns to their revenues, or to their reputation. Most are distracted by the fact that returns can be expensive and hard to administer (Stock et al. 2002). Therefore, firms face enormous risks (Stock 2001, Kocabasoglu and Suresh 2007). The beneficial factors of RSCM include the improvement of process efficiencies, customer service, supply chain design, product design, after‐market product sales, and after‐sales service. However, most research on managing RSCs has relied on case studies and optimisation models. Therefore, in this paper, an SEM is employed to investigate those factors that actually influence the success of RSC.

2. Literature review

2.1 Research related to factors influencing RSC performance

Researchers have studied RSC and developed its framework and structure. According to research by Kocabasoglu and Suresh (2007), the RSC process can be organised sequentially into five key steps: product acquisition; reverse logistics; inspection and disposition; reconditioning; and distribution and sales.

	Product acquisition: the process of obtaining the product from the customer. Vlachos and Dekker (2003) used the term product collection rather than product acquisition. Product collection refers to the initial transaction by which a company gains possession of the products.
	Reverse logistics: the process of planning, implementing, and controlling the efficiency, cost effective flow or raw materials, in‐process inventory, finished goods, and related information from the point of consumption to the point of origin for the purpose of recapturing or creating value or proper disposal (Rogers and Tibben 2001).
	Inspection and disposition: the sorting of the product stream into fractions of different qualities and their allocation to different reuse options (Vlachos and Dekker 2003).
	Reconditioning: repair, refurbishing, remanufacturing and recycling products when the product upgrade or material recovery option is determined to be the most appropriate disposition strategy (Penev and de Ron 1996).
	Distribution and sales: the delivery to a new market (Vlachos and Dekker 2003).

Handfield and Nichols (1999) reviewed the literature in RSCs and developed 10 research propositions for study using empirical research methods. However, they merely provided research suggestions and did not conduct a survey to verify their propositions. Davis et al. (1997) developed a model for reverse logistics entry by third‐party providers and examined the issues and processes that an organisation has to address to engage in the reverse logistics business. A reverse logistics decision‐making model was developed to guide the process of examining the feasibility of implementing reverse logistics in third‐party providers such as transportation companies. Tsoulfas and Costas (2002) explored managerial efforts in RSCs, where the focus was on the capture and exploitation of used products and materials. They concluded that recycling lead from batteries incurs some environmental impacts associated with their storage, transportation and processing. Kocabasoglu and Suresh (2007) used a survey of plant managers to empirically assess the linkages between supply chain investments, organisational risk propensity (i.e. willingness to take risk) and business uncertainty. They argued that RSC investment had three primary dimensions: reconditioning; recycling; and waste management. Ongoing investment in the FSC was significantly related to investment in recycling and waste management, but not to investment in reconditioning. Moreover, risk propensity was found to mediate the relationship between the external business uncertainty and investment in both the FSC and RSC. They also proposed the following hypotheses:

1. there is a direct positive relationship between FSC investment and RSC investment.

2. there is a direct positive relationship between FSC risk propensity and FSC investment.

Jap (1998) suggested shifting the supply chain into reverse and identified some returns and reverse logistics challenges for B2B and B2C. However, he failed to discuss how those challenges influence RSC performance. De Brito (2003) developed a framework for RSC and presented a content analysis of reverse logistics issues. Moritz et al. (2004) thought that RSC could capture value in the extended supply chain. They discussed reverse logistics opportunities and their challenges and indicated potential methods for companies to master. They also presented the differences between ‘conventional’ FSC and RSC supply chain processes and explained how separate FSC and RSC processes could be integrated into one extended supply chain concept. Krumwiede and Sheu (2002) introduced a new location‐inventory policy with reverse logistics applied to Chinese B2C e‐markets based on the characteristics of consumer purchasing behaviour within business‐to‐consumer (B2C) electronic markets in China. Carter and Kaufmann (2007) developed a system dynamics model for dynamic capacity planning of remanufacturing in closed‐loop supply chains. They argued that capacity planning is a strategic issue of increased complexity and importance. However, they only developed a mathematical model to give a potential solution to the capacity planning problem; and the detailed reasons or influencing factors for RSC adoption, were not discussed. As described above, most research in RSC has relied on case studies and optimisation models, thus there is a need for the identification of the critical success factors (CSFs) for RSC.

2.2 Technology acceptance model and diffusion of theory

The technology acceptance model (TAM), derived from the theory of reasoned action and introduced by Davis (1989), is a well‐established model that has been used broadly to predict and explain human behaviour in the domain of information systems. Technology acceptance model consists of perceived ease of use (PEOU), perceived usefulness (PU), attitude toward using, behavioural intention to use (BI), and actual system use (AU). Perceived usefulness and PEOU are the two most important determinants for system use. Legris et al. (2003) and Lee et al. (2003) reviewed articles related to TAM. Lee et al. (2003) investigated how TAM had made progress over an 18‐year period and divided this progress into four sections: introduction; validation; extension and elaboration. Many researchers have suggested that TAM requires additional variables to provide an even stronger model.

Taylor and Todd (1995) found that attitude to use is a weak predictor of either BI or AU. Venkatesh and Davis (2000) proposed an alternative version, TAM2, in which attitude to use was omitted and social influence and cognitive instrumental processes were included.

Diffusion theory is another prominent line of behavioural research useful to understanding IT use. According to Rogers (1983), an individual's decision on whether to accept or reject is predicted by five key perceptions: relative advantage; compatibility; complexity; ‘trialability’ and ‘observability’. Tornatzky and Klein (1982), who performed a meta‐analysis for innovation characteristics, found that three of 10 innovation characteristics (compatibility, relative advantage and complexity) related consistently to adoption. Moore and Benbasat (1991) expanded the relevant innovation characteristics set and developed an instrument to measure the perceived characteristics of using an innovation and claimed that relative advantage was very similar to the notion of usefulness in TAM and ease of use is similar in definition to Rogers' notion of complexity. Technology acceptance model and innovation diffusion theory (IDT) are similar in some constructs and supplement one another. Technology acceptance model provides theoretical linkages among beliefs, attitude, intention and action. Innovation diffusion theory, on the other hand, involves the formation of a favourable or unfavourable attitude toward an innovation. However, it does not provide further evidence on how the attitude evolves into the accept/reject decision. Plouffe et al. (2001) suggested that the perceived characteristics of innovating (PCI) set for antecedents explained substantially more variance than does TAM.

A key purpose of TAM is to provide a basis for tracing the impact of external variables on internal beliefs, attitudes and intentions. Therefore, this paper indentifies the factors influencing RSC performance based on TAM, PCI and some external factors.

3. Hypothesis and research model

We first reviewed the previous research related to RSC for extracting the hypothesis candidate. Second, the hypothesis was decided based upon an experts' meeting and the TAM. According to the hypothesis, we developed a concept model to identify those factors that influenced the success of RSC. Next, we developed the SEM model to extract factors and measures. Based on the SEM model, a questionnaire to collect data was constructed. The testing hypothesis step was conducted to identify and validate our proposed hypothesis. After the analysis of SEM, the factors of RSC were identified.

3.1 Hypothesis

Based on hypothesis candidate, TAM and literature review, 11 hypotheses were extracted. A key purpose of TAM is to provide a basis for tracing the impact of external variables on internal beliefs, attitudes and intentions. Therefore, TAM was used to develop hypotheses that considered inherent characteristics and some external factors.

3.1.1 Hypothesis 1: ease of use of RSC positively influences the RSC performance

Prahinski and Kocabasoglu (2006) explained that RSC is not popular because it is difficult to use. Typically, an item returned by a customer does not just retrace its steps back home, rather it is treated as a special item that can travel along numerous routes depending on a variety of factors, including return reason (for instance, defect versus ‘did not like’) and timing (such as during versus after season). According to Reyes and Meade (2006), the application of RSC should focus on the ease of use. This is because any technology or system that is complex and requires the implementation of a sophisticated infrastructure at all participating locations in the RSC has minimal chance of success. According to the statistics provided by Tony (2003), 37% of online buyers and 54% of online browsers were deterred from purchasing online because of return and exchange processes that were too difficult. Thus, ease of use of RSC is critical, which is the fundamental factor of our hypothesis.

3.1.2 Hypothesis 2: RSC cost negatively influences the RSC performance

According to statistics provided by Johnson (1998), reverse logistics costs account for between 0.5% and 1% of the total US gross domestic product. The cost of processing a return can be two to three times that of an outbound shipment. According to research by Reyes and Meade (2006), RSC is estimated to consume approximately 4% of overall logistic costs in the USA and to reduce profits by up to 35%. Many companies are now beginning to consider the cost and impact of reverse logistics when designing products, rather than mere forward considerations such as manufacturing costs, marketability and sales margin. Zeng and Zhou (2006) explained that the RSC cost heavily influences the RSC performance. They proposed a model to minimise the total reverse logistics cost. The primary objectives of third party logistics providers are to lower the total cost of logistics for the supplier. Hence, based on the literature review, we can deduce that if the RSC cost is high, then the economic performance of RSC will be low. Finally the overall satisfaction of RSC will be low.

3.1.3 Hypothesis 3: service quality positively influences RSC performance

Handfield and Nichols (1999) proposed that the service quality controls the variability in the RSC transactional processes through the design and management of tasks to be performed. Furthermore, it is tangibles in the environment like the treatment of the customer. They proposed that improved RSC service quality positively influences RSC performance. The management of customer returns of previous purchases, a common type of reverse logistics, may be the final frontier of competitive advantage. Here, enhanced customer service quality will enhance the customer trust and finally influence the overall satisfaction of the RSC process.

3.1.4 Hypothesis 4: perceived usefulness positively influences RSC performance

Reverse supply chain practices have often been environmentally driven, particularly in the European Union where environmental regulations are more stringent than in the USA. Many companies first focused on RSC issues because of environmental concerns. Today, some are concerned only with reverse logistics as they are related to returning product to their suppliers. However, in the future, environmental considerations will have a greater impact on many logistics decisions. Since RSC is very useful in that it can protect the environment; reduce overall cost and improve customer satisfaction. If RSC participants could be made fully aware of these benefits they would be more willing to invest in RSC and make a greater effort to improve RSC performance. Furthermore, in TAM (Davis et al. 1997), perceived usefulness and ease of use are the two most important factors in determining whether a certain system is popular and easily accepted.

3.1.5 Hypothesis 5: channel relationship positively influences RSC performance

The relationship between channel relationship and RSC performance has been implied in several studies (Stock et al. 2002). Increased complexity and changing channel relationships have dramatically increased the need for information exchange compatibility across organisations (Stuart et al. 1998, Daugherty 2001). Given that most buying organisations have communication arrangements with multiple suppliers, system compatibility between specific channel partners can be difficult to arrange. However, a good channel relationship is necessary to increase the efficiency of RSC efforts. Bagozzi and Edwards (1998) insisted that relationship management affects all areas of RSC and has a dramatic impact on performance. Based on the literature review, we conclude that a good channel relationship could positively affect the customer's opinion of the retailer and supplier, thus the product quality may be improved initially, and finally greatly improve the RSC partner's satisfaction of the RSC as well as its economic performance.

3.1.6 Hypothesis 6a: IT capacity negatively influences RSC cost

Increasingly, firms are finding it necessary to develop capabilities related to the use of technology, especially IT. An IT capability scale was thus employed to assess the effectiveness of a firm's use of technology in the management of the RSC process to reduce the RSC cost (Autry et al. 2001). IT could help reduce costs while meeting industry requirements (Lu et al. 2000). Reverse supply chain IT capabilities have the potential to make transactions more fluid and transparent for customers compared to paper based methods (Klassen and Vachon 2003). Based on IT capacity, RSC participants can easily communicate with each other in real time. By doing so, they can reduce labour cost, information cost, management cost, and other, invisible costs.

3.1.7 Hypothesis 6b: IT capacity positively influences service quality

As Dawe (1995) noted, many managers believe that IT is the single most important contributing factor influencing RSC performance improvement by improving the service quality. Toffel (2004) identified 10 differentiators between leading edge logistics organisations and average firms. One of those differentiators is the ability and willingness to invest in state‐of‐the‐art IT. IT capabilities significantly influence overall logistics competence (Daugherty et al. 2001). The current trend in supply chain management is toward a greater use of IT systems across all of the logistics functions (Parasuraman et al. 1988). When applied to reverse logistics, these IT capabilities may assist in all the improvements discussed above including the improvement of the service quality. Recent advances in supply chains have come about through the development and use of technology and the focus of using technology to improve the service quality (Vlachos et al. 2007).

3.1.8 Hypothesis 6c: IT capacity positively influences ease of use of RSC

Much of the IT used in RSC today emphasises improving service to partners and customers. Thus, many firms are focused on developing IT capabilities as an enabler of customer orientation and customer commitment (Kent and Mentzer 2003). Through effective combination of interactive dual direction technologies like point of sale (POS), EDI and inventory management systems, firms may improve RSC ease of use. Offering an effective and easy‐to‐use returns procedure can give consumers the confidence to make a purchase in the knowledge that they can send back any items that fall short of their expectations. Information and IT have long been recognised as offering the potential to serve as ‘competitive weapons’ critical to the support of overall strategic initiatives (Parasuraman et al. 1985). It is easy to understand how IT can help the ease of use of RSC by providing a better communication tool and good information.

3.1.9 Hypothesis 7: government policy positively influences perceived usefulness

Returns policies deal with the degree of difficulty involved in returning products (Autry et al. 2001). A liberal policy implies that it is easy to receive authorisation (and credit) for returns. In some instances, no authorisation may be needed. In contrast, a restrictive policy may involve considerable documentation, pre‐authorisation, a cap or limit on the amount that can be returned or at the extreme, a no returns policy. Tony (2003) investigated a potential relationship between environmental regulations and product recovery operations, which many contributors have assumed as a given fact. He examined whether companies have established and maintained their remanufacturing activities because of environmental product take‐back and recovery legislation. An important factor that influences the RSC performance is a convenient returns policy (Reyes and Meade 2006). According to Monahan (2004), returns policies deal with the degree of difficulty involved in returning products. In addition, Tony (2003) noted that the returns environment in the United States is out of control due to a ‘take it back’ culture propagated by retailers with liberal and almost unlimited return policies.

3.1.10 Hypothesis 8a: organisational commitment positively influences RSC performance

Organisational commitment has been extensively studied in organisational behaviour literature. Patricia et al. (2005) defined organisational commitment as investments in the trading partner's business; the definition of organisational commitment has been expanded to include loyalty and longevity (Bloemhof‐Ruwaard 1995). Rogers and Tibben (2001) indicated that investment is critical to the success of the RSC effort. Although the study's focus was on investments, the authors' argument could easily be extended to include other aspects of commitment. Daugherty et al. (2001) studied the impact of managerial and financial resource commitment in the RSC on the achievement of operational performance objectives. Both managerial and financial resource commitment were measured by single items that assessed the extent (level) of resource commitments. They indicated that managerial resource commitment in RSC was positively related to all of the performance measures except for improved customer relations and cost containment. They concluded that the commitment of managerial resources in the RSC had a greater level of influence on operational performance than on financial resource commitment. Research by Kent and Parker (1999) has shown that as managers commit more financial, human and physical resources to a programme or process then that programme or process may show superior financial performance. Although the relationship between organisational commitment and RSC performance is implied in several studies, research has not addressed how organisational commitment and the form of the investments influence operational performance. We anticipate that perceived organisational commitment, as measured through various structural and infrastructural investments, would influence different aspects of operational performance.

3.1.11 Hypothesis 8b: organisational commitment positively influences channel relationship

Stock et al. (2002) stated that ‘returns handling, if done right, can enhance relationships with consumers and supply chain partners.’ Developing this conviction, we state that organisational commitment to RSCs has a positive influence on the satisfaction with RSC channel relationships. In the marketing literature, Jap (2003) found that both the supplier and retailer transaction‐specific investments influenced the retailer's perceptions of the supplier's commitment to the relationship, and this commitment directly influenced relationship satisfaction. Increased commitment tends to make firms resist attractive short‐term alternatives in favour of expected long‐term benefits, and the emphasis is on preserving and promoting the ongoing relationship (Morgan and Hunt 1994).

3.2 Research model

Based on the literature review and interviews with experts, we developed the 11 hypotheses summarised in Figure 1. We have determined nine factors and 27 measurements. The arrows show the hypothesis between different factors.

Figure 1 Proposed SEM model for extracting CSF of RSC.

4. Research methods

4.1. Measures

Based on the literature review, we identified and developed the following measures shown in Table 1. The factors are developed based on the hypothesis.

Table 1. Definition of factors and measures.

Factor	Definition	Measure	Reference
RSC cost	Transportation cost, operating cost, fixed cost for new facilities, disposal and resale cost (Zeng and Zhou 2006)	Equipment cost	(Zeng and Zhou 2006)
		Management cost	(Toffel 2004)
		Invisible cost	(Souza and Ketzenberg 2002)
Technology capacity	IT and manufacturing technology capacity (Kent 2003)	IT technology	(Dawe 1995)
		Technology innovation	(Tushman and Nadler 1986)
		Technology investment	(Autry et al. 2001)
Government policy	Deal with the degree of difficulty involved in returning products (Richey 2005)	Encouragement	(Wang et al. 2007)
		Penalty	(Wang and Huang 2004)
		Law	(Wang et al. 2007)
Organisational commitment	Investments in the trading partner's business, loyalty and longevity (Cannon and Perreault 1999)	Investment commitment	(Toffel 2004)
		loyalty and longevity	(Schneider 2006)
		Finance resource	(Daugherty et al. 2001)
Channel relationship	Relationship between consumers and supply chain partners (Stock 2002)	Customer and retailer	(Williams et al. 1997)
		Supplier and retailer	(Amaldoss et al. 2000)
		Customer and supplier	(Stock et al. 2002)
Service quality	Tasks, tangibles and treatment (Schneider 2006)	Repaired product quality	(Handfield and Nichols 1999)
		Reliable and trustful	(Closs et al. 1997)
		Timely and properly	(Dawe 1995)
Ease of use	The degree to which RSC participants feel easy and convenient (Monahan et al. 2004)	Operation/adoption	(Amini and Retzalf‐Roberts 1999)
		Learn and understand	(Amini and Retzalf‐Roberts 1999)
		Control and manipulate	(Prahinski and Kocabasoglu 2006)
Perceived usefulness	Feeling of useful of RSC in environment protection and value‐adding (Retzalf‐Roberts 1998)	Environment protection	(Bagozzi and Edwards 1998)
		Overall benefit	(Davis et al. 1997)
		Usefulness	(Amini and Retzalf‐Roberts 1999)
RSC performance	Economic performance and overall satisfaction (Cannon 1999, Heide and John 1990)	Economic performance	(Krumwiede and Sheu 2002)
		Overall satisfaction	(Krumwiede and Sheu 2002)
		Investment and adoption	(Daugherty et al. 2001)

A model to minimise the total RSC cost consists of: transportation cost; operating cost; fixed cost for new facilities; final disposal cost and landfill cost as well as the sale revenue of reclaimed materials (Zeng and Zhou 2006). A problem with RSC was that the management cost was too high to be affordable (Toffel 2004). Reverse supply chain also has some hidden or invisible cost that may easily be neglected (Souza and Kerzenberg 2002). Therefore, we further classified the RSC cost into: equipment cost; management cost and invisible cost.

Bello et al. (2004) specifically examined supply chain related innovations. They characterise such innovations as: combining developments in information and related technologies with new logistics and marketing procedures to improve operational efficiency and enhance service responsiveness; efficient consumer response; and continuous replenishment to provide examples of supply chain innovations. Innovation refers to the creative process through which new products, services, or production processes are developed (Tushman and Nadler 1986). The aim is to find a better way to handle a process or approach. Autry et al. (2001) asserted that a good and continuous investment in technology provided a good insurance for technology capacity. Here, we employed the following measurements for technology capacity: IT capacity; technology innovation and technology investment.

Returns policies determine the degree of difficulty involved in returning products. A liberal policy implies that it is easy to receive authorisation for returns. Research by Richey et al. (2004) discovered that a favourable government policy, especially some form of encouragement to conduct product recycling in the RSC is very useful and effective. Punishment of actions against the environment can enhance the adoption of product recycling and finally support RSC (Wang and Huang 2004). Furthermore, a clear law should be formed to regulate the product recycling issues (Richey et al. 2007). Based on these reports, we decided to use the following measurements: government encouragement; government penalty and government law and policy.

Organisational commitment is defined as loyalty, longevity and investments (Handfield and Nichols 1999). Financial resource commitment and managerial resource commitment were two important aspects of organisational commitment and were positively related only to environmental regulatory compliance (Daugherty et al. 2001). In conclusion, we decided to use the following measurements for organisational commitment: investment commitment; loyalty; longevity and financial resource.

Interfering relationships have historically been characterised by actions of self‐interest and opportunistic behaviour. Many firms today have attempted to replace such adverse approaches with one of cooperation (Williams 1997). More trading partners are aware of the mutual benefits to be gained and the importance of supporting the relationships with adequate information. Relationship commitment is required on both sides of the buyer‐seller dyad. Stock (2002) discussed the importance of the relationship between customer and supplier in the RSC. Finally, we decided to use the following relationships among RSC participants: relationship between customer and retailer; relationship between supplier and retailer; and relationship between customer and supplier.

Usually, RSC managers use service quality as a proxy of, or link to, overall performance. Moreover, superior service quality is considered to be one of the keys to the development and retention of long‐term partnerships. The service quality scale used here examined the components of reverse logistics service quality. Specifically, the scale focused on the process of issuing return authorisations and crediting returns. The recycled product is as good as new and has a good quality, that is, it is easy for different partners to accept and adopt RSC (Handfield and Nichols 1999). Dawe (1995) suggested some reasons for product return, and asserted that shortening returns processing time is important for handling returns well. If a large number of unauthorised or unidentified items are being discovered, repeatedly, then there must be a significant problem with the return process. Moreover, customers usually prefer a reliable, responsible and trustful recycling process in which they can easily handle the product return in a convenient way (Gerard and Paul 1999). Therefore, we specially selected the following three suitable measurements: repaired product quality; responsibility and reliability; and good communication with customers.

Davis et al. (1997) defined ease of use as ‘the degree to which a person believes that using a particular system would be free of effort’, and RSC programmes should only be developed primarily for ‘uncontrollable returns’, and not all returns, which also include ‘controllable returns’ (for example, wrong product or quantity shipped, out‐of‐date products, damaged products). Treating difficulties in the successful operation and adoption of RSC is a core factor (Amini and Retzalf‐Roberts 2001) and a proper education in the principles of RSC is necessary (Prahinski and Kocabasoglu 2006). Among the many candidate measurements for ease of use, we selected: operation and adoption; learn and understand; and control and manipulate.

Davis (1997) defined perceived usefulness as the degree to which a person believes that using a particular system would enhance his or her job performance. He proposed six measurements for perceived usefulness: working more quickly; job performance; increasing productivity; effectiveness; making job easier and usefulness. Some of those measurements are not suitable for our model. Hence we only selected the suitable ones. One of the contributions of RSC is in the recycling of hazardous products (Bagozzi and Edwards 1998). The usefulness of RSC will become clearer in the near future as many companies are aware of the powerful profit‐creating capability of the RSC (Amini and Retzalf‐Roberts 1999).

The most frequently‐used performance metrics include: disassembly time; disassembly costs and parts revenue (Wang et al. 2002). Reverse supply chain programme formalisation is positively related to performance as defined by cost effectiveness, processing effectiveness and operating level effectiveness (Richey et al. 2007). Performance has been examined in the form of profitability and sales, and efficiency, and satisfaction (Closs et al. 1997). Daugherty et al. (2001) defined RSC performance objectives as environmental regulatory compliance, improved customer relations, recovery of assets, cost containment, improved profitability and reduced inventory investment. Among all the measurements provided by previous research, we finally selected the following three most important factors: economic performance; overall satisfaction and investment and adoption.

4.2 Participants characteristics

The survey stage is very important to our research because most of the results will be based on the survey. Thus we carefully selected the survey respondents and only companies with RSC or reverse logistics experience were selected. In our survey, we distributed 253 questionnaires in three languages: English, Chinese and Korean. We used both questionnaire and website to conduct our survey (the developed website: http://RSC.speedsurvey.com). We investigated two Korean companies: Dongguk Steel and SK Gas; three Chinese steel‐making companies: China Baogang Steel, China Shougang Steel and China Taiyuan Steel. We received 168 returned surveys of which 32 were excluded because they were incomplete. Therefore, we had 136 valid responses, and Tables 2, 3 and 4 show the characteristics and descriptive statistics of the sample respectively.

Table 2. Descriptive statistics of respondent's characteristics.

Company name	Company type	Responses	Percentage
Dongguk Steel	Steel making	22	16.18%
SK Gas	Manufacturing	29	21.32%
China Baogang Steel	Steel making	23	16.92%
China Shougang Steel	Steel making	33	24.26%
China Taiyuan Steel	Steel making	29	21.32%
Total		136	100%

Table 3. Demographic statistics of respondent's characteristics.

Gender			Age
Gender	Number	Percentage	Age range	Number	Percentage
Male	95	69.85%	20–30	33	24.26%
Female	41	30.15%	30–40	61	44.85%
			40–50	42	30.89%

Table 4. Sample characteristics.

Nationality			Business type
Nation	Number	Percentage	Type	Number	Percentage
China	85	62.50%	Steel making	107	78.68%
Korean	51	37.50%	Manufacturing	29	21.32%

5. Results

Scanning electron microscopy is a statistical technique for building and testing statistical models, which are sometimes called causal models. It is a hybrid technique that encompasses the aspects of confirmatory factor analysis, path analysis and regression, which can be seen as special cases of SEM. Advantages of SEM compared with multiple regression include more flexible assumptions (particularly allowing interpretation even in the face of multi‐collinearity), use of confirmatory factor analysis to reduce measurement error by having multiple indicators per latent variable, the attraction of SEM's graphical modelling interface, the desirability of testing models overall rather than coefficients individually, the ability to test models with multiple dependents, the ability to model mediating variables, the ability to model error terms, the ability to test coefficients across multiple between‐subjects groups, and the ability to handle difficult data (time series with auto correlated error, non‐normal data, incomplete data). Therefore, the proposed models for expressing hypothesis are validated based on SEM. This section describes the results of the SEM analysis, particularly; the sample size and its characteristics. The results of measurements and structural models tests are also presented.

5.1 Measurement modelling testing

Once the necessary data was collected, the next step was to analyse the data. In our survey, we used AMOS 7.0 and SPSS 14.0 to test the measurement and structural model. Before testing the structural model, we first examined the measurement model in terms of the construct reliability, because the internal consistency reliability indicates the stability of individual measurement items across replications from the same source of information. According to (Bagozzi and Edwards 1998), values greater than 0.5 usually indicate a good construct reliability. According to Hair et al. (1998), if all Cronbach's alpha values were greater than the benchmark of 0.6, then it is safe to say that there is a reasonable level of internal consistency among all the items. According to Cuieford (1973)<1965 or 1956?>, if the Cronbach's alpha value is <0.35, then the reliability is low, if Cronbach's alpha value is ⩾0.35 but <0.7, then the result is partly reliable, and if the Cronbach's alpha value is ⩾0.7, that means the result is highly reliable. In order to know whether the variables belong to the factors or not, the analysis of the reliability of measurements was conducted by assessing Cronbach's alpha scores.

Furthermore, we also used the confirmatory factor analysis (CFA) to test convergent validity of each construct, and a single factor model was specified for each construct. Our measurement model has 27 items to describe nine latent constructs: technology capacity (TC), government policy (GP), organisational commitment (GC), RSC cost (RSCC), service quality (SQ), EOU, PU, channel relationship (CR), and RSC performance (RSCP). Based on our analysis, all factors loading exceeded 0.4 on their own constructs, hence the convergent validity of the constructs was verified, which means all the measures had both strong reliability and discriminate validity. Table 5 shows the factor loadings and the values of Cronbach's alpha for the measurement model. According to the result, the model satisfied the requirements of construct reliability.

Table 5. Internal consistency reliability and convergent validity.

Latent construct	Measurement	Mean value	Standard deviation	Factor loading	Cronbach's alpha
Technology capacity (TC)	IT technology	3.75	1.34	0.72	0.74
	Technology innovation	3.78	1.27	0.83
	Technology investment	3.51	1.27	0.82
Government policy (GP)	Encouragement	5.09	0.99	0.89	0.77
	Penalty	4.64	1.03	0.64
	Law/policy	4.90	0.89	0.86
Organisational commitment (GC)	Investment commitment	4.78	0.83	0.67	0.72
	loyal and longevity	4.37	1.04	0.78
	Finance resource	4.11	1.02	0.70
RSC cost (RSCC)	Equipment cost	5.58	1.00	0.60	0.64
	Management cost	5.49	0.95	0.67
	Invisible cost	5.35	0.99	0.78
Service quality (SQ)	Repaired product quality	3.32	1.15	0.74	0.74
	Reliable and trustful	3.52	1.19	0.86
	Timely and properly	3.56	1.15	0.76
Ease of use (EOU)	Operation/adoption	3.54	1.29	0.85	0.72
	Learn and understand	3.46	1.38	0.72
	Control and manipulate	3.28	1.25	0.79
Perceived usefulness (PU)	Environment protection	5.38	0.78	0.80	0.81
	Overall benefit	5.47	0.78	0.87
	Useful	5.39	0.67	0.77
Channel relationship (CR)	Customer and retailer	5.18	1.07	0.70	0.76
	Supplier and retailer	5.43	1.04	0.79
	Customer and supplier	5.42	1.37	0.74
RSC performance (RSCP)	Economic performance	5.03	0.99	0.89	0.73
	Overall satisfaction	4.02	1.14	0.72
	Investment and adoption	4.89	1.04	0.71

5.2 Structural modelling testing

5.2.1 Model fit

After testing the measurement model, we used statistics to validate the structural model. Previous research suggested several good fit statistics such as: goodnesss of fit index (GFI); adjusted goodness of fit index (AGFI); normed fit index (NFI); comparative fit index (CFI); root mean square residual (RMR); and root mean square error of approximation (RMSEA).

For GFI, the value usually varies from 0 to 1, and if the value is close to 1, then it indicates a good fit. Also, it is suggested by previous researchers that in order to accept the model, the GFI value should be equal or more than 0.9. For NFI, the value also varies from 0 to 1, usually, the closer the value is to 1, the better the model fit, and the cut‐off value is also 0.9. The RMSEA value should be less than 0.08 to ensure a good fit, and in our survey, the goodness‐of‐fit statistics for the structural model are given in Table 6. It shows that the AGFI value is lower than the acceptable value, but the AGFI, CFI, RMR and RMSEA values all meet the acceptable threshold values. Thus, in general, the measurement model fits the data well.

Table 6. Goodness‐of‐fit statistics for model evaluation.

Goodness‐of‐fit statistics	Acceptable value	Actual value
Goodness of fit index (GFI)	>0.9	0.903
Adjusted goodness of fit index (AGFI)	>0.9	0.845
Comparative fit index (CFI)	>0.9	0.905
Root mean square residual (RMR)	<0.1	0.078
Root mean square error of approximation (RMSEA)	<0.08	0.057

5.2.2. Hypothesis test

Figure 2 shows the result of the structural model testing and the significant structural relationship between variables and standardised path coefficients.

Figure 2 Result of the structural model testing and structural relationship.

The overall results of the hypotheses testing are given in Table 7. Here, the critical ratio (CR) is the parameter estimate divided by an estimate of its standard error. If the appropriate distributional assumptions are met, then this statistic has a standard normal distribution under the null hypothesis that the parameter has a population value of zero. For example, if an estimate has a CR>2 (in absolute value), then the estimate is significantly different from zero at the 0.05 level. Even without distributional assumptions, the CRs have the following interpretation. For any unconstrained parameter, the square of its critical ratio is, approximately, the amount by which the chi‐square statistic would increase if the analysis were repeated with that parameter fixed at zero. Moreover, CR can be calculated as ‘CR = estimate/standard errors’.

Table 7. Result of hypotheses testing.

Hypothesis	Causal path	Hypothesis direction	Critical ratio	Assessment result
H1	EOU‐>RSCP	+	1.98**	Accepted
H2	RSCC‐>RSCP	−	−2.64**	Accepted
H3	SQ‐>RSCP	+	5.64***	Accepted
H4	PU‐>RSCP	+	4.83***	Accepted
H5	CR‐>RSCP	+	2.03**	Accepted
H6a	TC‐>RSCC	−	−2.09**	Accepted
H6b	TC‐>SQ	+	4.51***	Accepted
H6c	TC‐>EOU	+	2.45**	Accepted
H7	GP‐>PU	+	0.81	Rejected
H8a	OC‐>RSCP	+	3.25***	Accepted
H8b	OC‐>CR	+	1.08	Rejected
Critical ratio = Estimate/standard errors*** P<0.01, ** P<0.05

6. Discussion of the results

The survey was restricted to a small sample of five companies, because it proved difficult to find other companies that had the required experience of RSC. While this is a small sample and the nature of the businesses is different, it was our intention to collect and analyse data that would allow us to develop an initial understanding of the most critical factors of success for the introduction and management of RSC. According to the structural modelling testing described above, the proposed model is relatively acceptable and satisfactory. Even though the GFI value of 0.845 is a little bit lower than the expected value of 0.9, all the other values fit well. Based on the result of our proposed hypotheses, we feel confident in making the following statements.

	Five factors influence the performance of RSC, EOU, RSCC, SQ, PU and CR. Of these five factors, service quality and perceived usefulness are the most important. The service quality factor can be interpreted as incorporating repaired product quality, reliability, trustfulness, timeliness and proper recycling time. Perceived usefulness can be exemplified as environment protection, overall benefit and sense of usefulness. According to Amini and Retzalf‐Roberts (1999), many companies have first focused on RSC issues due to environmental concerns. Today, some are concerned only with reverse logistics as it relates to returning products to their suppliers. Hence, in order to improve RSC performance and its acceptance, all the participants along the supply chain should consider how to improve the recycled product quality, and how to recycle the used products at the right time. Also, the government or environmental protection organisations should actively promote the importance and usefulness of RSC and encourage companies to voluntarily adopt RSC, which not only protects the environment, and improves the customer satisfaction, but also creates profits for the company.
	Ease of use, RSCC and CR have important roles in improving the RSC cost. In our survey, some factories told us that even though they wanted to adopt the RSC theory and recycle the used products, the initial cost of recycling equipment was very high and even if they had bought the recycling equipment, the management costs and invisible costs in the later period made them give up. Thus, a good strategy for reducing the RSC cost is an important issue. The use of IT could be a good solution. For some low level companies, there were certain difficulties in understanding and operating the RSC, thus a systematic and large‐scale education of RSC is needed which could be sponsored by either government or non‐government‐organisation. In addition, a good channel relationship could bring many benefits for the RSC operators.
	Nowadays, technology is a key factor in almost every industrial field, including RSC. We found that TC not only helps reduce the RSCC but also improves the SQ and EOU. Reverse supply chain participants could save labour cost and time, collect information more accurately and timely, and also make the operation of RSC much easier. According to Lu et al. (2000), IT could help reduce costs while meeting industry requirements. Also, according to Gerard and Paul (1999), IT capabilities of RSC have the potential to make transactions more fluid and transparent for customers compared to paper‐based methods. From our investigation, most of the respondents said that the technology used in the RSC in their companies was not advanced enough and, what is more serious, was that the investment in the developing of technology, especially IT, (such as RSC management information system) used in RSC was not sufficient.
	According to the test result, organisational commitment directly influences the RSC performance positively, which is easy to understand. Patricia et al. (2005) proposed a hypothesis that commitment of resources to a firm's reverse logistics programme will have a direct positive impact on RSC performance. Amaldoss et al. (2000) proposed that a major challenge in today's business environment is to direct the focus and level of resource commitment. Barney (1986) also pointed out that firms that are able to correctly match and then commit resources to specific programmes and events are more likely to enjoy superior performance. In our survey, most of the respondents thought that a stable finance resource for RSC would determine its success or failure. Furthermore, investment commitment, loyalty and longevity of different participants in the RSC were also very important.
	The proposed hypothesis that government policy positively influences the perceived usefulness was rejected, even though some previous research has pointed to the importance of relationships between government policy and perceived usefulness. One of the possible reasons may be because we used different respondents from different countries, in which the government policies and laws were totally different. For example, China is a socialist country and in most cases, the market is heavily influenced by the government macroscopic readjustment and control: on the other hand, Korea is a capitalist country, where the market is heavily influenced by the market itself. Thus, due to the different social systems, this hypothesis was rejected.
	Finally, the hypothesis that organisational commitment positively influences the channel relationship was also rejected, even though Daughertyet et al. (2001) proposed that ‘the greater the relationship commitment of the buying firm to the supplier, the stronger the positive relationships between information systems support and reverse logistics product return performance’. Considering all of this, we think that the reason why this hypothesis was rejected was because the Chinese companies (China Baogang Steel, China Shougang Steel and China Taiyuan Steel) we interviewed were all state‐owned companies, while Dongbu Steel and SK Gas were private companies in Korea. Chinese companies have monopolised the steel‐making field in the Chinese market. Hence, they seldom need to worry about the channel relationship and simply produce steel and sell it. The other things would be handled by the Chinese government. Thus, the relationship between the organisational commitment and channel relationship in the Chinese market was not very strong and clear. Therefore, this hypothesis was not acceptable.

7. Conclusions and suggestions for future studies

Companies should care about what happens to products after they reach consumers because there are significant opportunities to develop an environmentally compatible advantage through the development of a RSC. Utilisation of the reversed supply stream as an informative tool and as a way to reclaim value can aid companies in designing and presenting appealing products that meet the needs of societies worldwide, while reclaiming profits that would otherwise be lost. Even though our research has pointed out some useful relationships and successful factors of RSC, there is still some room left to improve our research: first, we proposed 11 hypotheses but two were rejected because the characteristics of the Chinese and Korean markets are different. Thus, future research needs to validate whether those two hypotheses were correct. Second, our survey sample size of 136 was limited by our access to respondents who had RSC experience and this consequently limited the robustness of our research. Thus, future research effort must be directed at identifying more RSC practitioners to generate more data that could verify or contest our conclusions. Third, our data were insufficient to describe every aspect of RSC. Further research is required, in particular, to identify those intangible factors that might influence the success of RSC. Finally, some real world failure cases would be useful to explore the causes for rejection of hypotheses H7 and H8b.