An assessment of operational efficiencies in the UK retail sector

The Authors

Wantao Yu, Jubilee Campus, Nottingham University Business School, Nottingham, UK

Ramakrishnan Ramanathan, Jubilee Campus, Nottingham University Business School, Nottingham, UK

Abstract

Purpose – The paper's aim is to assess performance of firms in the UK retail sector.

Design/methodology/approach – Economic efficiencies of 41 retail companies working in the UK between 2000 and 2005 are examined in this study using three related methodologies: data envelopment analysis (DEA), Malmquist productivity index (MPI), a bootstrapped Tobit regression model. DEA is used to calculate technical and scale efficiencies of companies. Two outputs (turnover, profit before taxation) and three inputs (total assets, shareholders funds, and number of employees) are employed for the efficiency measurement. MPI is used to analyze the patterns of efficiency change over the six year period 2000-2005. DEA efficiencies are then used to test important hypotheses on the impact of environmental variables, namely head office location, type of ownership, years of incorporation, legal form and retail characteristic, on the functioning of the UK retail sector using bootstrapped Tobit regression.

Findings – DEA analysis has shown that only ten retail companies are considered as efficient under CRS assumption, and 16 firms under VRS assumption in 2005. MPI results have indicated that about 50 percent of retail companies have registered progress in terms of MPI during 2000 and 2005. Twenty out of 41 retail companies have adopted advanced and efficient retailing technologies during this period. Three environmental variables, namely the type of ownership, legal form and retail characteristic, have been found to play significant roles influencing retail efficiency using bootstrapped Tobit regression.

Research limitations/implications – Data availability has limited the level of analysis in some parts of this study, especially in the bootstrapped Tobit regression.

Originality/value – This study seems to be the first in applying productivity analysis using DEA for the UK retail sector.

Article Type:

Research paper

Keyword(s):

Business performance; Retailing; Data analysis; United Kingdom.

Journal:

International Journal of Retail & Distribution Management

Volume:

Number:

Year:

2008

pp:

861-882

Copyright ©

Emerald Group Publishing Limited

ISSN:

0959-0552

Introduction

Over the past few decades, efficiency and productivity have become an important issue for managers, both in the manufacturing and the service sector (Sellers-Rubio and Mas-Ruiz, 2007). In the retail industry, retail productivity plays an important role in the control and management of retail companies, and providing vital information for a number of tactical, strategic, and policy related decisions (Dubelaar et al., 2002; Sellers-Rubio and Mas-Ruiz, 2007). The analysis of productivity and efficiency has become an important activity in retailing (Oxford Institute of Retail Management, 2004; Barros and Alves, 2004; Lusch et al., 1995). However, it has been well-recognized that attempts to measure efficiencies and productivities of firms in the retail sector face a number of challenges owing to the difficulties in identifying the level of retail services. Previous studies in this area have presented a number of measures, models and methods to assess retail productivity and efficiency, including regression, stochastic frontier analysis and data envelopment analysis (DEA). Particularly, in this study we estimate the economic efficiency of selected companies in the UK retail sector using DEA. DEA is an operations research based performance evaluation methodology that has been used in assessing managerially useful measure of company/store level retail productivity. DEA allows using multiple measures of inputs and outputs for evaluating the performance of decision-making units (DMUs) within a retail company or among companies in the retail industry (Wen et al., 2003). Over the last few years, efficiency in the retail industry in several countries has been analyzed using DEA by a number of studies (Thomas et al., 1998; Ratchford, 2003; Donthu and Yoo, 1998; Keh and Chu, 2003; Barros and Alves, 2003, 2004; Kamakura et al., 1996). However, to our knowledge, there seems to be no study on UK retail sector using DEA, though there are studies in productivity of UK retail sector (Oxford Institute of Retail Management, 2004).

In this study, economic efficiency of 41 retail companies working in the UK between 2000 and 2005 are examined, employing three related methodologies: DEA, Malmquist productivity index (MPI), and a bootstrapped Tobit regression model. DEA is used to calculate technical and scale efficiencies of firms. The use of DEA for the analysis of comparative retail companies' efficiency can be of value in examining the competitiveness of the retail industry as a whole. Competitiveness should be based on benchmarking the retail companies, which compose the sector. The retail companies that achieved the highest efficiency are considered as benchmark and the economy efficiency of the other companies are evaluated relative to this benchmark. Two outputs (turnover, profit before taxation) and three inputs (total assets, shareholders funds, and number of employees) are introduced for the efficiency measurement. MPI is employed to analyze the patterns of efficiency change over the period 2000-2005. DEA efficiencies are then used to test important hypotheses on the impact of environmental variables, including head office location, types of ownership, years of incorporation, legal form, and retail characteristic, on the functioning of the UK retail sector using bootstrapped Tobit regression. To overcome the problem of the inherent dependency of DEA efficiency scores when used in regression analysis, a bootstrapping technique is applied. The aim of this regression is to seek out those best practices that will lead to improved performance throughout the whole retail chain.

The remainder of this paper is organized as follows. We first provide the foundations of DEA methodology. A brief literature review on the DEA applications in retail sector follows. We then discuss the input/output and environmental variables, the proposed DEA model, and data collection issues. Next, we describe an empirical study in which DEA is employed to assess the efficiencies of 41 retail companies in the UK. Managerial implications, limitations and future research are also discussed.

Review of DEA and its applications in retail sector

This section makes a literature review of DEA, MPI, and bootstrapped Tobit regression analysis. For the sake of brevity of this paper, detailed discussions of these tools are not described here. Important references are provided to help the interested readers. Then applications of DEA in retail sector over the last few years are reviewed.

Data envelopment analysis (DEA)

DEA is a mathematical programming technique that calculates the relative efficiencies of organizations (usually refers to a DMUs) based on multiple inputs and outputs (Charnes et al., 1978). To calculate efficiency scores employing DEA, two different assumptions can be made, i.e. constant return to scale (CRS) and variable returns to scale (VRS). The VRS efficiency score measures pure technical efficiency, i.e. a measure of efficiency without scale efficiency. On the other hand, the CRS efficiency score represents technical efficiency, which measures inefficiencies due to the input/output configuration and the size of operations (Cooper et al., 2007). Scale efficiency can be computed by the ratio of CRS efficiency to VRS efficiency. Hence, scale efficiency of a DMU operating in its most productive scale size is one. More details on DEA can be found in Cooper et al. (2007) and Ramanathan (2003).

Malmquist productivity index (MPI)

A special method of time series analysis in DEA is to use the results of DEA in conjunction with the MPI. The MPI was introduced as a theoretical index by Caves et al. (1982) and popularized as an empirical index by Färe et al. (1994). The MPI is defined as the product of the “catching-up” and the “frontier shift” terms. The “catching-up” term relates to the extent by which a company improves its efficiency, while the “frontier-shift” term reflects the change in the efficient frontier surrounding the company between the two periods of time (Sellers-Rubio and Mas-Ruiz, 2006). This index allows changes in productivity to be broken down into changes in efficiency (deviations from the best practice frontier) and technology change (TC) (movements of the frontier), and is defined using distance functions.

Bootstrapped Tobit regression

Tobit regression is often encountered in second stage DEA, i.e. when the relationship between exogenous factors (non-physical inputs) and DEA efficiency scores is assessed (Hoff, 2007). However, the previous DEA studies have shown that the efficient scores obtained in the first stage are correlated with the explanatory variables used in the second term, so that the second-stage estimates will be inconsistent and biased (Xue and Harker, 1999; Simar and Wilson, 2000). Therefore, Simar and Wilson (1999) suggested that a bootstrap procedure should be employed to overcome this problem. The bootstrap is a computer-based method for assigning measures of accuracy to statistical estimates. It was first introduced by Efron (1979) and since then it has become a popular and powerful statistical tool (Casu and Molyneux, 2000). More details on truncated regression with bootstrap can be seen in Simar and Wilson (2007).

DEA applications in retail sector

The analysis of productivity and efficiency has become an important activity in retailing (Barros and Alves, 2004; Lusch et al., 1995; Kamakura et al., 1996). When reviewing contemporary researches in efficiency measurement that were published over the last few years (Thomas et al., 1998; Ratchford, 2003; Donthu and Yoo, 1998; Keh and Chu, 2003; Barros and Alves, 2003, 2004), it is apparent that there is an increase in the use of DEA to evaluate retail efficiency and productivity (Table I).

Sellers-Rubio and Mas-Ruiz (2006) have used DEA to estimate the economic efficiency of supermarket chains in the Spanish retailing industry. The empirical application has been carried out on a sample of 100 supermarket chains between 1995 and 2001. The study has revealed high levels of economic inefficiency in the Spanish retailing sector. Barros (2006) has analyzed a representative sample of 22 hypermarkets and supermarkets working in the Portuguese market, adopting a two-stage procedure to benchmark the retail companies. In the first stage DEA has been used and in the second stage a Tobit model has been employed to estimate the efficiency drivers. The following are important conclusions from this study:

efficiency of hypermarket and supermarket retail companies is high compared to other sectors;
larger retail groups are more efficient than the smaller retailers;
national retailers are on average more efficient than regional retailers;
scale plays an important role in this market;
factors such as market share, number of outlets and location are important efficiency drivers; and
regulation has a negative effect on efficiency.

Donthu and Yoo (1998) have analyzed 24 outlets of a fast-food restaurant chain using DEA.

These previous studies have evaluated cost efficiency (Ratchford, 2003), technical efficiency (Thomas et al., 1998; Donthu and Yoo, 1998; Keh and Chu, 2003; Barros and Alves, 2003) and scale efficiency (Keh and Chu, 2003; Barros and Alves, 2003). The majority of studies adopt a static perspective (Thomas et al., 1998; Ratchford, 2003; Donthu and Yoo, 1998; Keh and Chu, 2003; Barros and Alves, 2003), whereas only Barros and Alves (2004) and Sellers-Rubio and Mas-Ruiz (2007) adopted a dynamic perspective, examining the patterns of changes in efficiency using MPI. For example, Sellers-Rubio and Mas-Ruiz (2007) have used the MPI for a sample of 96 supermarket chains operating in Spain between 1995 and 2003 to estimate total productivity change in these retailing firms and to decompose it into efficiency change and technical change (i.e. the consequence of innovation and adoption of new technologies). They have concluded that information, communication and technology (ICT) have the capacity to alter the productive structures of retail firms, favoring their productivity. Barros and Alves (2004) have estimated total productivity change and decomposed it into technically efficient change and technological change for a Portuguese retail store chain with MPI. The authors have ranked the stores according to their total productivity change for the period 1999-2000, and conclude that some stores experienced productivity growth while others experienced productivity decrease.

Selection of input and output variables

The choice of the input and output variables is vital to the successful application of DEA. Donthu and Yoo (1998) stated that input and output variables for DEA should exactly reflect the retail company's objectives and sales situation. Based on a review of the literature, main inputs and outputs criteria that have been used to examine retail efficiency and productivity are summarised in Table I. Previous studies have proposed different measures of output, both in monetary units (such as sales revenue, profit volume and value added) (Thomas et al., 1998; Donthu and Yoo, 1998; Keh and Chu, 2003) and in non-monetary units (such as customer store loyalty and satisfaction, and service quality) (Donthu and Yoo, 1998; Keh and Chu, 2003).

The literature on productivity assessment in retail sector generally differentiates two different kinds of input–controllable inputs and non-controllable inputs, according to whether the company considers them or not in its management action plans (Donthu and Yoo, 1998; Sellers-Rubio and Mas-Ruiz, 2006). Since controllable inputs can be controlled by companies to gain competitive advantage, it is a common practice to use them as part of efficiency assessment. Examples of controllable inputs considered in the literature include company managerial factors and personnel factors, such as size of company (e.g. square feet of selling space) (Pilling et al., 1995; Lusch and Serpkenci, 1990), the number of outlets in the supermarket chain (Sellers-Rubio and Mas-Ruiz, 2006), current total assets (Doutt, 1984), and the number of employees (Thomas et al., 1998; Sellers-Rubio and Mas-Ruiz, 2006). In contrast, non-controllable inputs are generally considered as environmental variables since they could influence the efficiency of companies but are not directly controllable by the companies. Examples of environmental variables considered in the literature include retail structure (Goldman, 1992), retail sector conditions (Goldman, 1992), location (Donthu and Yoo, 1998), demographics of clientele in the area (Donthu and Yoo, 1998), and national economic development (Pilling et al., 1995). Normally, non-controllable inputs are ignored in the estimation of retail productivity (Donthu and Yoo, 1998). We follow a similar strategy in choosing the controllable and environmental variables for this study and the details are discussed in the next section.

Methodology and variables

Methodology

In this paper, three methodologies, namely DEA, MPI, and bootstrapped Tobit regression are used to study the operational efficiencies of retail sector in the UK. The conceptual framework is proposed in Figure 1. We use the two-stage method in this study (Coelli et al. (2005) for more details). In the first-stage analysis, DEA is used to calculate technical and scale efficiencies of retail companies, which includes only the conventional inputs and outputs. Two outputs (turnover, profit before taxation) and three inputs (total assets, shareholders funds, and number of employees) are employed for the efficiency measurement. In general, any DEA study considers performance analysis at a given point time. However, extensions to the standard DEA procedures such as MPI approach have been reported to provide performance assessment over a period time (Ramanathan, 2003). Thus, MPI is used to examine the patterns of efficiency change over the period 2000-2005. In the second-stage, the DEA efficiency scores from the first-stage are used to test important hypotheses on the impact of environmental variables, including head office location, types of ownership, years of incorporation, legal form, and retail characteristic, on the functioning of the UK retail sector using bootstrapped Tobit regression (Figure 1). The regression aims to investigate those best practices that will generate improved performance throughout the whole retail chain. More details on the implementation of the conceptual framework in our analysis, such as the choice of inputs, outputs and environmental variables, are discussed in the next subsection.

Selection of inputs, outputs, and environmental variables

Inputs and outputs

As can be seen from Table I, different authors have employed different measures of output, such as sales revenue, customer satisfaction, and service quality. In our study, we use two monetary outputs. Sales revenue (Donthu and Yoo, 1998; Barros and Alves, 2003, 2004; Zhu, 2000; Sellers-Rubio and Mas-Ruiz, 2006) is the first output. Justification for this selection is that retail companies work with a large range of products and services, which hinders the collection of disintegrated information on outputs produced (Sellers-Rubio and Mas-Ruiz, 2006). Moreover, retail companies can achieve typical income apart from their main activity, which is not included in their sales volume figures; apart from sales volumes, retailers pay special attention to results as they guarantee the viability of the company; and considering the volume of profits allows for inclusion of the influence of other types of costs not considered as inputs (Sellers-Rubio and Mas-Ruiz, 2006). Therefore, the second output used is the profit volume of the company (Barros and Alves, 2003, 2004; Zhu, 2000; Sellers-Rubio and Mas-Ruiz, 2006).

With regard to inputs, as above mentioned literature review, the factors used to produce goods/services can be divided into controllable and non-controllable. In our study, we apply three controllable productive factors, namely, the number of employees (Thomas et al., 1998; Barros and Alves, 2003, 2004; Sellers-Rubio and Mas-Ruiz, 2006), total assets (Doutt, 1984), and shareholders' funds (Sellers-Rubio and Mas-Ruiz, 2006, 2007; Ratchford, 2003) as well. Here, shareholders' funds are equal to the total of share capital plus reserves. Some details of the outputs and controllable inputs used in this study are shown in Table II.

Environmental variables

In the second-stage, the technical efficiency variable return to scale index of the retail companies is regressed using bootstrapped Tobit regression methodology to identify the impact of the environmental variables listed in Table II. The Tobit model used in our study is presented as follows: Equation 1

Here, is VRS efficiency scores of retail companies in 2005, we use five environmental variables (Table II) that could influence a retail company's operational efficiency. These factors are not the conventional inputs and output in the DEA model and are assumed not under the control of business management (Boame, 2004; Casu and Molyneux, 2000). The explanatory factors can include legal form (public/private), location characteristics, company size, age, industry group, and government regulations (Barros, 2006; Casu and Molyneux, 2000; Wincent, 2005; Temtime and Pansiri, 2005; Pansiri, 2007; Glancey, 1998). In our study, the following environmental variables are considered: head office location, types of ownership, years of incorporation, legal form, and retail characteristic. Data on these variables were independently selected and care was taken to separate controllable input factors that determine efficiencies and environmental variables that characterise the management practices of retail companies.

Head office location is a dummy variable. It is classified into eight different areas in accordance with data from FAME (Table II). A head office is like a large laboratory in business management, which accumulates knowledge on personnel management, new product development, quality control and operations strategy (Kono, 1999). Ownership is also a dummy variable (one for national firms and two for foreign firms), which is used to measure the advantages achieved through knowledge and experience of the local market. Evidence for the variable has been found by a number of previous studies. For example, as mentioned above, Barros (2006) has found that national retailers in Portugal are on average more efficient than local retailers. Years of incorporation means the time of company forming a legal corporation. This variable is designed to evaluate operations experience, which a retail company has. Experience is a major factor shaping strategic directions and collective knowledge. The impact of company's experience on business performance has been widely discussed. A positive relationship between company age and efficiency may be expected if older companies benefit from dynamic economies of scale by learning from experience. Older companies may also benefit from reputation effects, which allow them to achieve a higher margin on sales (Glancey, 1998). Legal form stands for the legal nature of a company, here one for private limited, two for public quoted, and three for public not quoted. There are advantages and disadvantages to both structures of private company and public company. An advantage is that public companies are able to more easily raise funds and capital through the sale of their securities. Because they are answerable to shareholders, publicly quoted companies face greater pressure to achieve acceptable returns on investment over shorter periods of time than do private companies (White, 1995). As such, private companies have the ability to focus more on the achievement of long-term returns. Finally, retail characteristic is another dummy variable (one for food retailing, two for home appliances retailing, three for DIY and home improvement retailing, four for fashion retailing, and five for others), which is carried out to assess market coverage of both food and non-food retail sectors. Grocery retailing is the largest sector in the UK's retail market economy. According to IGD Research (2007), the UK grocery retail market will continue to grow at an average rate of 2.9 percent over the next five years and will be worth £138.2bn (at current prices) by 2010. It can be argued that distinguishing between retailers on the basis of sector groups (food and non-food retailing) is critical, since different retail sector groups have experienced different degrees of success. However, the relationships between retail sector groups and organizational performance remain speculation only.

Data

To estimate the production frontier, we used panel data on retail companies for the years 2000-2005. All the data required for this study are obtained from the FAME database (Fame database, Copyright © 2007 Bureau van Dijk Electronic Publishing https://fame.bvdep.com/). Since, we use profits and shareholders funds as two of three inputs, 13 retail companies made negative profits and shareholders funds are eliminated in this study, for example, Sainsbury group, Somerfield group, B&Q Plc, Comet group Plc, Phones4u Ltd, Iceland foods Ltd, etc. In addition, due to non-availability of appropriate archival data, some more retail companies could not be analyzed in this analysis, for example, Next Plc, First queen retailing Ltd, WH Smith Plc, and Burberry group Plc, etc. The final sample consists of 41 different retail companies (Table III) operating continually between 2000 and 2005 in the UK retail market, and these retail firms operate their business in several different sectors, such as food retailing, home appliances, DIY and home improvement, and fashion retailing.

Results

The results of data analysis are presented in the following three sections. The first subsection analyzes the results of DEA analysis. Second, the trends in efficiency and productivity over time are discussed in the second subsection. Finally, test important hypotheses on the impact of variables on the functioning of the UK retail sector through regression analysis.

Efficiency of retail companies in 2005 using DEA

The DEA index can be computed in several ways. In this section, we evaluate the economic efficiency of retail companies and divide it into CRS efficiency, VRS efficiency, and scale efficiency. Table III shows that the efficiency scores of the 41 retail companies in the year of 2005. It can be seen that under the CRS assumption the most efficient firms are Asda Group Ltd (1.00); BHS Group Ltd (1.00); Boots Company Plc (The) (1.00); Farmfoods Ltd (1.00); HMV Music Ltd (1.00); Jewson Ltd (1.00); Lidl Ltd (1.00); Netto Foodstores Ltd (1.00); River Island Clothing Co. Ltd (1.00); and Robert Wiseman & Sons Ltd (1.00). These ten retail firms' CRS efficiencies are all 1.00, and located on the efficient frontier. It implies that these companies have produced the maximum possible outputs (turnover, profit before the taxation) for the given level of inputs (total assets, shareholders funds, and the number of employees). We can see that all these ten companies are also efficient when VRS is assumed. The average efficiency score under CRS assumption is equal to 0.73. If we include all source of inefficiency, retail companies could operation on average at 73 percent of their current outputs level using their given input value. The other several companies, such as Allied Domecq Ltd (0.963); Boots The Chemists Ltd (0.919); DSG Retail Ltd (0.969); Tesco Stores Ltd (0.988); and The Carphone Warehouse Ltd (0.978) also have good relative CRS efficiency.

With regard to VRS efficiency that measures pure technical efficiencies are larger than CRS efficiencies. The CRS efficiencies of Allied Domecq Ltd (0.963); DSG Retail Ltd (0.969); Kellogg UK Holding Company Ltd (0.706); Marks and Spencer Plc (0.506); Tesco Plc (0.734); and Tesco Stores Ltd (0.988) are all less than 1.00, but their VRS efficiencies are 1.00. This could mean that these firms could not achieve 100 percent CRS efficiencies since they do not operate at their most productive scale size. Scale efficiency can be calculated as the ratio of CRS and VRS efficiency. Beside the ten CRS efficient retail companies, some firms have higher scale efficiency near to 1.00, including Allied Domecq Ltd (0.963); Argos Ltd (0.99); Boots The Chemists Ltd (0.942); DSG Retail Ltd (0.969); House of Fraser (0.995); IKEA Ltd (0.98); JJB Sports Plc (0.992); John Lewis Partnership Plc (0.938); John Lewis (0.938); Kingfisher Plc (0.939); New Look Retailers Ltd (0.998); Primark Stores Ltd (0.976); Sports World International Ltd (0.998); Tesco Stores Ltd (0.988); The Carphone Warehouse Ltd (0.998); The Game Group Plc (0.999); TK Maxx Group Ltd (0.922); and Wal-Mart Stores (UK) Ltd (0.955).

Trends of changes in efficiency scores over time using MPI

In this section, patterns of changes in efficiency of the retail companies during the period of 2000-2005 employing the MPI approach are presented.

Based on the above analysis of efficiency scores of 41 retail companies in 2005, we selected top ten retail companies whose CRS efficiency was 1.00 in 2005 and then calculated their CRS efficiency during the previous six years, from 2000 to 2005. The CRS efficiency scores are presented in Figure 2. The figure demonstrates the CRS efficiency of Lidl Ltd kept the same scores of 1.00 in the period considered. Asda Group Ltd, Boots Company Plc (The), Netto Foodstores Ltd, River Island Clothing Co. Ltd, and Robert Wiseman & Sons Ltd experienced a significant progress over the years. Particularly, the CRS efficiency of Netto Foodstores Ltd was 0.049 in 2000, it increase to 1.00 in 2004, the next year of 2005 maintained the same score of 1.00. Boots Company Plc (The) also improved from 0.546 in 2000 to 1.00 in 2002; the next three years (2003, 2004, and 2005) kept the score of 1.00. However, some companies experienced a fluctuation during the period of 2000-2005. For example, The CRS efficiency of Farmfoods Ltd was 1.00 in 2000, but it decreased to 0.565 in 2002, and then increased to 1.00 in 2003, it retained the same score in 2004 and 2005. BHS Group Ltd improved from 0.541 in 2000 to 1.00 in 2001, and then dropped to 0.854 in 2003, over the next three years maintain the higher scores of 1.00.

The MPI over the time of six years from 2000 to 2005 is presented in Table IV, the values are geometric means of MPI for the five period 2000-2001, 2001-2002, 2002-2003, 2003-2004, and 2004-2005. It shows that 22 out of 41 retail companies have progressed in terms of MPI during the period. Here, we select Netto Foodstores Ltd as a sample that has registered the highest improvement in MPI (1.712) and discuss it in more detail. The MPI for the company is 1.712, which means there is an increase in improving the performance in terms of “Turnover” and “Profit before taxation” for the given level of “Total assets”, “Shareholders funds”, and “the number of employees” in the year of 2005 compared to the year of 2000. It can be seen that the progress of MPI for Netto Foodstores Ltd was contributed by a significant increase in technical efficiency change (1.829) rather than a decrease in technology change (0.936) over the time. The change in technical efficiency is the diffusion of best-practice technology in the management of the activity and is attributed to investment planning, technical experience, and management and organization in the companies. Hence, it can be concluded that the diffusion of best-practice technology in Netto Foodstores Ltd improved in the period, and the improvement of MPI is contributed by better efficiency progress rather than technology changes. Moreover, for the period under analysis, we verify that there is an increase in the VRS technical efficiency change (1.026) and scale efficiency change (1.783). The improvement in pure technical efficiency implies that Netto Foodstores Ltd has conducted investment in organizational factors in accordance with the company management, for example better balance between inputs and outputs, and efficient quality management. Another sample of Allied Domecq Ltd also analyzed, its MPI is 1.008, indicating more outputs (turnover, and profit before taxation) was produced using a given level of inputs (total assets, shareholders funds, and the number of employees). However, the progress of MPI for Allied Domecq Ltd is contrary to Netto Foodstores Ltd, it was contributed by technology changes (1.016) rather than technical efficiency change (0.993). In addition, there are no change in VRS technical efficiency change (1.000), and a reduction in scale efficiency change (0.993) over the time.

As above-mentioned analysis, Table IV demonstrates that 22 retail firms have progressed in terms of MPI between 2000 and 2005. Among these 22 companies, Netto Foodstores Ltd was ranked first for the highest improvement in MPI (1.712), and followed by Asda Group Ltd (1.218); HMV Music Ltd (1.194); Boots Company Plc (The) (1.174); River Island Clothing Co. Ltd (1.161); and Jewson Ltd (1.131). Among those 19 retail companies who have suffered regress in MPI, IKEA Ltd has revealed the highest regress (0.859) during the period, the next one is Matalan Retail Ltd (0.873). The MPI for Farmfoods Ltd is 1.00, which indicates that no change in average productivity of the region in the period considered. Technological change is the result of innovation, i.e. the adoption of new technologies, and communication system, by the best-practice companies. Approximately, half of retail companies (about 50 percent, 20 out of 41) have introduced the advanced and efficient retailing technologies over the last six years. Let us undertake a further analysis, Jewson Ltd has revealed the highest progress (1.131) in terms of technology change, it can be argued that Jewson Ltd has obtained better achievements than other retail companies in the UK in adopting efficient technologies for retailing operations. It is followed by Asda Group Ltd (1.124), the company also seems to achieve better than others. In addition, the Table IV shows that the 41 retail companies in UK have obtained various levels of improvement or regress in the terms of VRS technical efficiency change and scale efficiency change.

Based on the above-mentioned analysis, it can be concluded that these 41 retail companies in the UK experienced progress, regress and no change in MPI over the last six years 2000-2005. 22 out of 41 (about 54 percent) retail companies have revealed progress, but only one firm, i.e. Farmfoods Ltd has expressed no change in the productivity. Also, approximately 50 percent of retail companies have shown good achievements on introducing efficient technologies for retailing operations.

It may be noted that the company with the highest progress (Netto Foodstores Ltd) is a Danish company. Other foreign companies operation in the UK, namely Asda Group Ltd and Wal-Mart Stores (UK) Ltd have also shown progress in MPI during the period of study. Hence, there seems to be a linkage between the ownership (foreign or UK) with the efficiencies. Moreover, there seems to be a reasonable linkage between the legal form and efficiencies. There are many private companies (e.g. Allied Domecq Ltd and HMV Music Ltd) that have registered progress while several public companies (e.g. JJB sports) do not seem to have registered good progress. The next section deals with these linkages quantitatively by employing statistical concepts. We use DEA efficiencies as indicators of performance instead of MPI in the next section.

Drivers of efficiency using bootstrapped Tobit regression

We conducted both Tobit regression and bootstrapped Tobit regression analysis. Recent DEA literature supports using bootstrapped Tobit regression in order to overcome the problem of inherent dependency of efficiency scores when used in regression analysis (Xue and Harker, 1999; Simar and Wilson, 2007). Hence, we report the results of bootstrapped Tobit regression in this paper. The Stata software program (version 10.0) was used to carry out the bootstrapped Tobit analysis. In order to be more objective and reliable, we undertook bootstrapped Tobit analysis by considering the impact of one variable at a time instead of using a multiple regression equation. The result of analysis is presented in the Table V. The observed coefficients, t-ratios, Wald χ ², Prob > χ ², Pseudo R ², Sigma, and Log likelihood are reported in this table. It reveals that efficiency scores are positively and statistically significant with three variables (type of ownership, legal form and retail characteristic). Thus, the bootstrapped Tobit regression analysis shows that ownership, legal form and retail sector can be considered as the driving forces influencing efficiency of retailers. The other two variables, namely head office location and years of incorporation, do not have significant relationship with efficiency.

Results of bootstrapped Tobit regression analysis in Table V show that type of ownership has a significant positive coefficient. This means that the VRS efficiency of foreign retailers is on an average 11.31 percent higher than that of local retail companies. This result agrees with the observation on the impact of foreign and local retailers using the MPI analysis in the previous section. In addition, the legal form (private limited, public quoted, and public not quoted) of a company seems to have significant relationship with its VRS efficiency. This suggests that private retail companies seemed to be more efficient than the other two legal forms. Based on the regression results shown in Table V, it may be safe to think that the efficiencies of private companies could be, on an average, 8.8 percent more than their public counterparts. Table V also shows that retail characteristic of companies is also statistically significant with their efficiencies. Looking at the results of Table V, food retail companies seem to be more efficient than other retail sectors in the UK. The efficiency, on an average, increases by about 3.4 percent when we move from fashion retailing, to DIY and home improvement retailing, to home appliances retailing, and, to food retailing. The DEA analysis also obtained the same results (Table III), four out of ten (40 percent) CRS efficient companies are food retailer, i.e. Asda Group Ltd, Farmfoods Ltd, Lidl Ltd, and Netto Foodstores Ltd The next section provides a discussion on this and additional results.

Discussion and managerial implications

In this section, we critically analyze the DEA, MPI and bootstrapped Tobit regression results described in the previous sections, and discuss their managerial implications as well.

The results derived from DEA analysis indicate that only ten retail companies have the highest CRS efficiency of 1.00. And among these ten CRS efficient companies, there are four food retailers, namely, Asda Group Ltd, Farmfoods Ltd, Lidl Ltd, and Netto Foodstores Ltd Three efficient companies, i.e. Asda Group Ltd (USA), Lidl Ltd (Germay), and Netto Foodstores Ltd (Denmark), are foreign-owned companies in UK. This presents that these three firms have generated the maximum possible outputs (turnover, profit before the taxation) for the given level of inputs (total assets, shareholders funds, and the number of employees). According to Retail Knowledge Bank (2001), there were totally four foreign-controlled food retailers in 2001 in the UK, namely, Asda Group Ltd (USA), Lidl Ltd (Germany), Netto Foodstores Ltd (Denmark), and Aldi Ltd (Germany). To our knowledge, the number has not changed until today. Food retailing is the largest sector within the UK's retail market. There are 102,511 grocery stores in 2006 (IGD Research, 2007). Tesco and Sainsbury are the two major UK based firms in the food sector, but CRS efficiency of Tesco Plc was 0.734 that is close to the average CRS efficiency of 0.73 in 2005, while that of Sainsbury Plc (0.693) was below the average level. Therefore, it can be argued that the efficiency of foreign-owned food retail companies seems to be higher than that of local retailers. Under the VRS assumption, there are 16 retail companies that have achieved efficiencies. But Argos Ltd, JJB Sports Plc, New Look Retailers Ltd, and Whitbread Group Ltd seem to be performing much worse VRS efficiency. Inefficiency reflects the failure of these firms to obtain the maximum feasible output given the amount of inputs used.

The advances in the information, communication, and technology coupled with global and regional competitions have changed the landscape of various service providers. Technological change is one of the most important elements of future competition in retailing, and correct adaptation of new technologies gives significant competitive advantages to firms achieving the innovation process. The results of MPI considered in our study verify that about 50 percent (22 out of 41) of retail companies have expressed progress in terms of MPI during the period 2000 and 2005. Especially, Netto Foodstores Ltd achieved the highest improvement in MPI (1.712), which was contributed by a significant increase in technical efficiency change rather than a technology change. And, approximately half of retail companies (20 out of 41) have introduced the advanced and efficient retailing technologies during the period considered. However, there were 19 firms have suffered regression in MPI, for example, IKEA Ltd and Matalan Ltd and, about 50 percent of companies have registered regress in terms of technology change in the period, which means that these 21 retail companies have not adopted or implemented effectively new technologies in their operations procedures. Information technology is a tool to enhance the overall strategy of the organization as well as being used to promote competitive advantage in the market, consequently to improve organizational performance. Technical efficiency is a consequence of various factors such as managerial policies, company financial condition, and scale. Therefore, it can be suggested that theses companies should review the operational procedures that would improve the efficiency of operations, and adaptation of new technology in retailing operations.

The results of DEA reveal that the primary cause of efficiency is the scale economies, but it does not identify the other more driving factors influencing efficiency. A bootstrapped Tobit regression allows us to investigate other efficiency drivers beyond the scale economies. The bootstrapped Tobit analysis in our study presents that legal form is one of the determinants to retailers' efficiency. As far as we are not aware of any DEA studies in retail sector that has used legal form in regression analysis. Our study shows that private retail companies in the UK seem to be more efficient than public operators. The private limited company is the most common business structure in the UK retail sector, in our study 28 out of 41 retailers are private limited companies. The regression result also indicates that retail characteristic is another driver of efficiency, and food retail companies are more efficient than non-food retailers in the UK. According to IGD Research (2007), food retailing is one of the most dynamic sectors in the UK's retail market economy, and the grocery market was worth £123.9bn in 2006, an increase of 3.4 percent on 2005. It will continue to grow at an average rate of 2.9 percent over the next five years. Moreover, over the last few years supermarkets have expanded into the less-traditional non-food categories which now account for more than 10 percent of sales through supermarkets (IGD Research, 2007). In addition, the study identifies that the national or foreign ownership is another driving force influencing efficiency in the British retail sector. Foreign retail companies in the UK seem to be more efficient local retailers. Although cultural and businesses difference might create obstacles to successful market entry and adaptation, theses obstacles can be overcome over time through learning. Moreover, the internationalization literature ( Johanson and Vahlne, 1990,1992) has implied that psychically close countries are more similar and, because similarities are easier to manage than differences, it is expected that businesses will achieve greater success in similar markets. There is, to some extent, fewer and lower psychic distance, such as cultural, regulatory, legal, and financial, for foreign retailers from EU and USA (e.g. Asda Group Ltd (USA), Lidl Ltd (Germay), and Netto Foodstores Ltd (Denmark)) to operate in the UK market. This result is contrary to Barros's (2006) study, the author has found that national retailers in Portuguese market operated better than foreign companies. On the other hand, the factor of head office location does not have significant relationship with retail efficiency, 20 out of 41 retail companies' head offices are located in London. London remains a powerful magnet for business, while other cities such as Glasgow, Edinburgh, Manchester, and Leeds have become increasingly attractive for subsequent roll-outs. Moreover, our study verifies that years of incorporation is not an efficiency driver. The reasons for this result may be that the new retail companies could be easier to adapt the new technology and business management which normally generate improved performance. These results contrast with the significant positive relationship found by Dobson and Gerrard (1989) and have suggested that older companies have not derived advantages from reputation effects or accumulated experience. Glancey (1998) has also conclude that younger companies display significantly higher growth rates than older companies.

Conclusion

This study aims to investigate economic efficiency of 41 retail companies working in the UK between 2000 and 2005. To our knowledge, this seems to be the first study to analyze the performance efficiency of retail companies in the UK, and consequently the whole UK's retail sector. The measurements have been undertaken using a popular benchmarking tool of the DEA, as well as MPI and a bootstrapped Tobit regression model.

We first use the DEA methodology to estimate efficiency on a sample of 41 retail firms during the period considered. The results using data for the year 2005 have shown that only ten retail companies are considered as efficient under CRS assumption, and 16 firms under VRS assumption. The general conclusion is that the average efficiency of retail companies in the UK was less than 75 percent over the time. Benchmarks are provided for improving the operations of poorly performing retailers. Then, the MPI has been computed to estimate productivity change. The results have shown that about 50 percent (22 out of 41) of retail companies have expressed progress in terms of MPI during 2000 and 2005, especially Netto Foodstores Ltd has achieved the highest improvement in MPI, while IKEA Ltd has suffer the highest regress in productivity during the period considered. And there are 20 out of 41 retail companies have adopted the advanced and efficient retailing technologies during this period. The analysis also has shown that there has been no change in productivity of the region during the six years of 2000 to 2005, such as Farmfood Ltd Finally, we carried out a bootstrapped Tobit analysis, under this the determinants of economic efficiency are investigated. The analysis has verified that legal form and ownership of company are the possible driving forces influencing efficiency. It can be argued that private retail companies seemed to be more efficient than the other two legal forms, and foreign retailers in the UK seem to operate better than local companies. In addition, it is found that retail characteristic is another driver of efficiency. The other two factors, i.e. head office location and years of incorporation are not the efficiency drivers.

This study has some limitations. DEA is a model that evaluates the relative efficiency of different homogeneous DMUs, based on linear programming techniques. However, although all the included firms are retail companies, the future study should take into account that the market, services and business strategy are very different among food and non-food retailing (e.g. Tesco vs IKEA). Moreover, although we tried to collect data through a number of ways, some information about retail companies in the special years is still not available. For example, during the bootstrapped Tobit regression analysis, just several environmental factors such as head office location, types of ownership, years of incorporation, legal form, and retail characteristic are within FAME. Therefore, the data set is short, and some factors have their own difficulties in the statistics analysis. The conclusions are limited. Reducing the number of observations in the DEA variables may increase the possibility that a given observation will be considered relatively efficient. Therefore, further research is needed to do in this area and to also confirm the results obtained in our study.

Equation 1

Figure 1Conceptual framework

Figure 2The pattern of CRS efficiency scores of ten efficient companies during 2000-2005

Table IPrevious research into retail efficiency using DEA

Table IIInputs, outputs, and environmental variables (data in 2005)

Table IIIDEA scores for 41 retail companies in 2005

Table IVMPI analysis of 41 retail companies during 2000-2005

Table VResults of bootstrapped Tobit regression with C = 500 samples of size 41

References

Athanassopoulos, A. (1995), "Performance-improvement decision aid systems (PIDAS) in retailing organizations using data envelopment analysis", The Journal of Productivity Analysis, No.6, pp.153-70.