You are on page 1of 13

The current issue and full text archive of this journal is available at www.emeraldinsight.com/1463-5771.

htm

Measuring and benchmarking the performance of coffee stores for retail operations
Seong-Jong Joo
School of Business, Clayton State University, Morrow, Georgia, USA

The performance of coffee stores

741

Philipp A. Stoeberl
Department of Management, John Cook School of Business, Saint Louis University, St Louis, Missouri, USA, and

Kristin Fitzer
College of Business, Central Washington University, Ellensburg, Washington, USA
Abstract
Purpose The purpose of this paper is to measure and benchmark the retail operations of selected coffee stores owned by a specialty coffee company. Design/methodology/approach Data envelopment analysis is used for benchmarking the performance of eight coffee stores for two years using internal annual reports. Findings Major ndings are that the inefcient stores need to improve occupancy related expenses and revenues from non-coffee items. In addition, the coffee stores locate in an afuent residential area outperform the stores in the business district. Research limitations/implications The limitation of this paper is in the use of nancial data. To overcome this issue, non-nancial data should be included. Originality/value The contributions of this paper are twofold: providing a framework for performance measures within a benchmarking perspective and evaluating the comparative efciency of coffee stores using internal data, which is not accessible to the general public. Keywords Coffee, Retailing, Performance measures, Benchmarking, Data analysis Paper type Research paper

1. Introduction Coffee has been around since the fteenth century and is the second most traded commodity next to oil (Hoover, 2007). The coffee shop industry domestically includes 20,000 stores with a combined annual revenue of about 11 billion dollars (IMCR Case Studies and Management Resources, 2007). The coffee industry is not only interesting but also offers many challenges as it has become increasingly competitive, demanding efciency in all areas of operations for success. Measuring and benchmarking the performance or efciency of coffee stores using various factors, which is an important part of the managerial process, may not be simple and consequently requires a very high level of effort. The use of conventional efciency measures and nancial ratios will provide simple and straightforward results. The advantages of these measures are that they are readily available and easy to use and understand. However, they
The authors would like to thank the reviewer(s) for their valuable comments and suggestions.

Benchmarking: An International Journal Vol. 16 No. 6, 2009 pp. 741-753 q Emerald Group Publishing Limited 1463-5771 DOI 10.1108/14635770911000088

BIJ 16,6

742

provide only a partial picture of performance and are unable to incorporate multivariate aspects of performance measures. In fact, when we measure and benchmark the performance of organizations, we need to include multiple factors such as purchasing, labor, building and markets for their individual effects and, at the same time, for their overall effects. Managers will want to know the comparative performance of the stores determining which factors are more inuential for the protability of a store and/or which factors should be improved to increase protability. We demonstrate an approach for measuring and benchmarking multivariate aspects of performance using actual data from a coffee company. Particularly, we analyze the operating efciency of the coffee stores using data envelopment analysis (DEA), which is a well-known approach used in models for the comparative efciency measures of organizations or organizational units. Measuring that part of performance is an essential process for management. Without information or feedback on the performance of organizations, it will be difcult to manage them efciently and effectively. We failed to nd a scholarly study about performance measures and benchmarking for coffee stores at the time of this study. Accordingly, this study will be a rst attempt in this area. In addition, the use of internal data, which is not accessible by the public, is unique to this study. Because internal data is used, the anonymity of the company will be maintained throughout the study. The contributions of this study include providing a practical framework for performance measures within a benchmarking perspective in general, and analyzing comparative efciency for the retail operations of selected coffee stores using pertinent internal data along with market or location information in particular. This study is applicable to similar businesses for measuring and benchmarking performance. The rest of this study consists of reviewing related studies within a benchmarking perspective, exploring DEA for a methodology, describing the choice of data and variables and presenting and discussing results followed by conclusions that include the summary of ndings and the limits and future direction of the study. 2. Related studies For benchmarking the efciency of coffee stores for retail operations, we employ DEA, which is widely used for measuring and benchmarking the relative efciency of various entities. Selected recent DEA studies for benchmarking involve measuring the performance of nancial institutes (McEachern and Paradi, 2007), healthcare organizations (Duffy et al., 2006; Ferrier and Valdmanis, 2004), railway zones (George and Rangaraj, 2008), seaports (Cullinane et al., 2005), airports (Graham, 2005), third party logistics providers (Min and Joo, 2006), schools (Johnes, 2006), a charity organization (Joo et al., 2007), telecommunication service providers (Debnath and Shankar, 2008; Kwon et al., 2008) and utility companies (Goncharuk, 2008). Because this study focuses on the comparative efciency measures of coffee stores as retail operations, we investigate the related literature but fail to nd one at the time of this study. Coffee stores increasingly carry light meals and can be part of the restaurant industry in a broad sense (the fast casual market). Accordingly, we expand our literature review to the restaurant industry and note a handful of studies measuring operating efciency using DEA.

Donthu and Yoo (1998) argue that there is no uniform, broadly accepted denition of productivity measurement methodology for retailing and summarize four major problems related to retail productivity measures. First, salesperson productivity has been used comparably for retail productivity due to the labor-intensive nature of retailing. Second, existing studies do not measure productivity among individual stores. Instead, they focus on two extreme levels such as salespersons (an excessive micro level) and retail industries or aggregation of stores (a broad macro level). Third, most existing productivity measures are simple and absolute ones such as conventional productivity measures and ratios. Comparative productivity measures among peer retail stores are necessary for benchmarking and managerial processes. Last, the problems of existing productivity measures such as productivity indexes and regression analyses sometimes hinder the understanding of retail productivity. To overcome these problems, Donthu and Yoo (1998) suggest DEA for retail productivity measures and demonstrate the application of DEA for measuring the relative efciency of 24 fast-food chain restaurants in a major metropolitan city. They include two output variables: sales in dollars and the results of customer satisfaction surveys and four input variables: store sizes in serving area, the experience of managers in years, promotion/give-away expenses in dollars, and location such as in a shopping mall or free-standing, in a basic DEA model. However, they do not provide sources of efciency or inefciency, which are identied by employing different types of returns-to-scale, such as constant returns-to-scale and variable returns-to-scale. Lan et al. (2006) use DEA for developing an efciency-based recruitment plan for fast food chain restaurants in Taiwan. They select 27 fast food chain restaurants in South Taiwan and include three output variables: prot after control, cash ow and total number of customers served and four input variables: salary for employees, salary for managers, benets and utility expenses. Except for one output variable, total number of customers served, all variables are measured nancially. They identify sources of efciency (or inefciency) such as pure technical or managerial efciency and scale efciency (SE) by employing different types of returns-to-scale. Pure technical efciency (PTE) represents efciency due to management. Meanwhile, SE denotes efciency because of the different scales used for computing efciency in DEA models. Accordingly, PTE is useful for managers to improve the efciency of organizations. After computing efciency scores, they perform a series of what-if analyses to determine recruitment requirements. Unlike Donthu and Yoo (1998) and Lan et al. (2006) do not include a location variable. Reynolds (2003) points out the problems of methodologies for evaluating productivity for restaurant operations and suggests the use of DEA to include a wide range of factors, which inuence restaurants sustainability. He lists three popular productivity measures with their shortcomings: partial factor productivity measures that lack comprehensive measures of operating efciency; total factor productivity measures that are unable to provide comparative efciency for multiple operations and regression analysis that fails to generate benchmark information. To illustrate the benets of DEA, he explains the results of two studies: one that analyzes 83 midscale, corporate owned and operated restaurants (Reynolds and Gu, 2004) and the other that evaluates a chain of 62 full service restaurants (Reynolds and Thomson, 2007). For measuring the operating efciency of the restaurants, both

The performance of coffee stores

743

BIJ 16,6

studies include controllable and uncontrollable variables in their models. In addition, Reynolds and Thomson (2007) incorporate the effect of location into their model. 3. Methodology DEA is a useful approach for measuring relative efciency among similar organizations or objects, which are called decision-making units (DMUs) in DEA analyses. Because DEA can identify relatively efcient organizations or DMUs among a group of given DMUs, it is a promising tool for benchmarking. DEA is a special application of linear programming (LP) based on the frontier methodology of Farrell (1957), advanced by Charnes et al. (1978) and further advanced by Banker et al. (1984). There are numerous variations of DEA models for different analyses since Charnes et al. (1978). To explore the mathematical property of a basic DEA model, let E0 be an efciency score for the base DMU 0 then: nP o R ur0 yr0 r1 o Maximize E 0 nP 1 I vi0 xi0 i1 subject to: nP
R r1 ur0 yrk i1 vi0 xik

744

o for all k 2

nP I

o #1

ur0 ; vi0 d where:

for all r; i

yrk is the observed quantity of output r generated by unit k 1,2, . . . ,N. xik is the observed quantity of input i consumed by unit k 1,2, . . . ,N. ur0 is the weight to be computed given to output r by the base unit 0. vi0 is the weight to be computed given to input i by the base unit 0.

d is a very small positive number.


The fractional programming model can be converted to a common LP model without much difculty. Although the LP model can be solved with ordinary LP software, the use of various DEA solvers can save time and effort for solving the LP model. Sherman and Ladino (1995) summarize the capability of DEA in the following manner: . identies the best practice DMU that uses the least resources to provide its products or services at or above the quality standard of other DMUs; . compares the less efcient DMUs to the best practice DMU; . identies the amount of excess resources used by each of the less efcient DMUs; and . identies the amount of excess capacity or ability to increase outputs for less efcient DMUs, without requiring added resources.

Major DEA studies have utilized selected organizations and departments or branches in organizations as DMUs for measuring their efciency. Moreover, some studies have demonstrated that DEA can be used for evaluating personal efciency by choosing, for example, physicians as DMUs (Chilingerian, 1995; Chilingerian and Sherman, 1994; Chu et al., 2003). Thus, DEA can be employed for measuring the efciency of any entity, which has inputs and outputs and is homogeneous with peer entities in an analysis. According to a recent DEA study, there is a remedy for a group of entities that are not homogeneous (Haas, 2003). Thus, DEA can be applied to the wide variety of DMUs without much restriction as long as DMUs satisfy the basic requirements of inputs and outputs. In addition, DEA is applicable to DMUs with categorical and uncontrollable (or environmental) input data (Athanassopoulos and Thananssoulis, 1995; Mahajan, 1991). Because DEA is solved with LP, it inherits the same limitations as found in using LP. Especially, degeneracy with LP can be a problem for benchmarking studies by neglecting an alternative optimal solution (Fumero, 2004). In this case, a two stage LP method is suggested for detecting degeneracy. In this study, involving benchmarking operations of coffee stores, Charnes-Cooper-Rhodes (CCR), Banker-Charnes-Cooper (BCC), and bilateral CCR and BCC models are employed. A CCR model utilizes constant returns to scale so that all observed production combinations can be scaled up or down proportionally. A BCC model, on the other hand, uses variable returns to scale and is solved by a piecewise linear convex frontier. In addition, both input oriented and output oriented models are used depending on the selection of variables. To get pure technical efciencies, BCC efciency scores are compared to the results of the CCR model. For understanding location differences, bilateral CCR and BCC models are applied. 4. Data and variables We have data for eight coffee retail stores, associated with a premier coffee retailer, for two years, current and past year, and maintain 16 DMUs per model. They are located in two different districts: four stores are located in an afuent residential area and the remaining four stores are found in a typical business area in Seattle, Washington. We will test for a location difference in operating efciencies of coffee stores along with input and output variables. We have two different sets of variables for two models. Model 1 includes one output variable: total sales and four input variables: cost of sales, wages and benets, other expenses and occupancy expenses. Total sales consist of total restaurant and retail sales. This variable shows the total sales being taken in by each retail location. The rst input variable used in this model, cost of sales includes purchasing and other costs that are associated with the goods for sale. Wages and benets, the second input variable, gives total costs of labor associated to the sales of each store. Some items included are vacation, bonuses, payroll taxes and salaries. Other expenses, the third input variable, encompass items such as telephone, building and equipment, maintenance, security, education and training. The fourth variable used in this model is occupancy expenses, and it incorporates the cost of rent, utilities, insurance and other costs associated with the occupancy of the retail location. Model 1 will show the overall efciency of retail stores major expenses. Occupancy expenses are considered as uncontrollable. Model 2 contains two output variables: sales-restaurant and sales-retail and three input variables: cost of sales, wages, and other expenses. The two output variables are

The performance of coffee stores

745

BIJ 16,6

746

encompassed in total sales used in Model 1, but it should be noted that in Model 2, total sales have been separated into restaurant sales and retail sales. Some items included in restaurant sales are espresso drinks, pastries, and drip coffee. Items in retail sales include packaged food, brewing equipment, general merchandise, and whole bean sales. All of the input variables are the same as used in Model 1 with the exception of occupancy expenses. We exclude this uncontrollable variable and try to measure the efciency on the major operating characteristics of retail stores. Table I shows variables and their descriptive statistics for the two models analyzed in this study. 5. Results and discussion First, we apply CCR and BCC input-oriented models to Model 1 that includes total sales for an output variable and cost of sales, wages and benets, other expenses, and occupancy for the four input variables. CCR models employ constant returns-to-scale and provide technical efciency (TE), which is global optimal. Meanwhile, BCC models use variable returns-to-scale and measure PTE or operational efciency, which is local optimal. Accordingly, employing both CCR and BCC models together will provide meaningful managerial insights for understanding pure operational efciency along with TE. Because there are one output and four input variables, CCR and BCC input-oriented models are selected for analyses. Table II exhibits the results of CCR-I and BCC-I analyses for Model 1. In the CCR-I analysis, only three DMUs are 100 per cent efcient. Store 2 demonstrates strong efciency such as 100 per cent efciency in both the past and current year. Stores 1, 3, 4, 5 and 6 tend to maintain decreasing efciency scores. Stores 7 and 8 show a positive trend for their efciency scores. For pure operational efciency, nine DMUs are 100 per cent efcient in the BCC-I analysis. As shown in Table II, the efciency scores generated by BCC models are greater than or equal to those of CCR models. Stores 6, 7 and 8 along with Store 3 in the current year need to improve their pure operating efciency. The last column in Table II displays ranks for stores nancial contribution to the companys prots. It is reasonable to expect that the ranks roughly coincide with the ranks of the CCR-I model. However, the ranks of the BCC-I model look different from those two ranks. This is because the BCC model focuses on pure operational efciency without considering market conditions and other external factors that inuence DMUs efciency, which are measured with SE. SE is computed
Model Model 1 Variable Total sales Cost of sales Wages and benets Other expenses Occupancy Sales-restaurant Sales-retail Cost of sales Wages and benets Other expenses Minimum 316,612 114,095 131,850 16,360 59,269 286,060 22,107 114,095 131,850 16,360 Maximum 3,117,709 1,161,470 1,526,475 217,328 1,147,130 2,833,148 326,610 1,161,470 1,526,475 217,328 Average 1,278,542.81 419,118.94 521,115.88 80,184.56 247,954.81 1,153,197.56 125,345.13 419,118.94 521,115.88 80,184.56 SD 843,549.48 292,557.21 380,045.33 58,627.72 266,946.51 777,374.03 100,684.39 292,557.21 380,045.33 58,627.72 Variable type Output Input Input Input Input Output Output Input Input Input

Model 2

Table I. Descriptive statistics for variables

Note: In US dollars

CCR-I Store 1_Current 1_Past 2_Current 2_Past 3_Current 3_Past 4_Current 4_Past 5_Current 5_Past 6_Current 6_Past 7_Current 7_Past 8_Current 8_Past TE 89.50 93.58 100.00 100.00 86.07 100.00 70.27 71.38 64.30 65.71 68.99 75.59 67.98 66.48 64.25 58.19 Rank 5 4 1 1 6 1 9 8 14 13 10 7 11 12 15 16 PTE

BCC-I Rank 1 1 1 1 10 1 1 1 1 1 13 12 14 15 16 11 SE 89.50 93.58 100.00 100.00 88.51 100.00 70.27 71.38 64.30 65.71 98.25 98.89 98.65 96.90 95.16 63.17 Rank by total contribution 3 4 2 1 5 6 8 7 16 15 9 10 11 14 13 12 100.00 100.00 100.00 100.00 97.24 100.00 100.00 100.00 100.00 100.00 70.22 76.44 68.91 68.61 67.52 92.12

The performance of coffee stores

747

Notes: TE: technical efciency; PTE: pure technical efciency

Table II. Efciency scores in percentages and ranks for Model 1

as follows: SE uCCR =uBCC , where uCCR and uBCC represent the CCR and BCC scores of a DMU, respectively. Thus, we have TE PTE SE. Using this decomposition of efciency, we can identify the sources of inefciency. SE reveals unfavorable conditions for DMUs while PTE shows pure operational efciency, as explained before. For example, Stores 6 and 7 maintain high SE scores close to 100 per cent. Accordingly, their low BCC scores are due to inefcient operations. This nding indicates that the managers of Stores 6 and 7 must put forth additional effort to improve the efciency of operations. Meanwhile, Stores 4 and 5 are 100 per cent efcient in the BCC model but they are considerably inefcient for SE. Thus, their low efciency scores (TE) in the CCR model are due to unfavorable or market conditions. Because BCC models compute pure operational efciency, it is desirable to review potential improvements indicated by the BCC models; specically, what inputs to improve and by how much. Table III displays information for potential improvements suggested by the BCC-I model. Among the input variables, Occupancy, which includes expenses related to stores occupancy, requires higher levels of improvement than any other variable in the model. Potential improvements for Stores 3, 6, 7 and 8 in the current year provide store managers with managerial insights to increase efciency for their store operations. Next, we split total sales into two output variables: sales from restaurant and sales from retail operations. As a part of diversication efforts, coffee retailers tend to increase their retail operations by adding items other than coffee drinks. Considerable revenues are generated from non-coffee drinks; for example, music CDs, general merchandise, coffee utensils, cookies/cakes and non-coffee drinks. Thus, it is important to divide total sales into two sales categories. For input variables, we drop Occupancy, which is xed and uncontrollable in the short-term, such as within one year. Accordingly, we have sales and its directly related expenses in Model 2, which are controllable within the short-term. For Model 2, we apply CCR and BCC output-oriented

BIJ 16,6

Store 1_Current 1_Past 2_Current 2_Past 3_Current 3_Past 4_Current 4_Past 5_Current 5_Past 6_Current 6_Past 7_Current 7_Past 8_Current 8_Past

Total sales 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 10.17

Cost of sales 0.00 0.00 0.00 0.00 24.99 0.00 0.00 0.00 0.00 0.00 229.78 226.05 234.40 231.39 232.49 27.88

Wages and benets 0.00 0.00 0.00 0.00 22.76 0.00 0.00 0.00 0.00 0.00 243.53 235.72 238.18 231.87 242.58 242.58

Other expenses 0.00 0.00 0.00 0.00 221.54 0.00 0.00 0.00 0.00 0.00 238.34 223.56 231.09 254.58 238.11 27.88

Occupancy 0.00 0.00 0.00 0.00 2 20.91 0.00 0.00 0.00 0.00 0.00 2 61.03 2 63.76 2 73.82 2 73.53 2 62.46 2 7.88

748

Table III. Potential improvement in percentages for Model 1 by BCC-I

Note: Positive number indicates increase and negative number means reduction

models and try to determine efciency for the two types of revenues. Table IV shows the results of CCR-O and BCC-O analyses for Model 2. Stores 6, 7 and 8 have low PTE scores while keeping high SE scores. This indicates that their inefciency is due to operations, not external factors. However, Store 5 retains low SE scores meaning that their source of inefciency is due to unfavorable conditions. For only operations or PTE, Store 5 is 100 per cent efcient compared to the other stores in the model. Because the selection of the variables for the results in Table IV is limited to the ones directly related to operations, the inefciency identied

CCR-O Store 1_Current 1_Past 2_Current 2_Past 3_Current 3_Past 4_Current 4_Past 5_Current 5_Past 6_Current 6_Past 7_Current 7_Past 8_Current 8_Past TE 95.98 95.69 100.00 100.00 100.00 100.00 78.20 81.49 66.60 72.67 71.27 84.33 76.95 75.82 68.45 61.43 Rank 5 6 1 1 1 1 9 8 15 12 13 7 10 11 14 16 PTE 100.00 100.00 100.00 100.00 100.00 100.00 100.00 87.06 100.00 100.00 75.42 87.13 76.98 79.04 69.83 82.24

BCC-O Rank 1 1 1 1 1 1 1 11 1 1 15 10 14 13 16 12 SE 95.98 95.69 100.00 100.00 100.00 100.00 78.20 93.60 66.60 72.67 94.50 96.79 99.96 95.93 98.02 74.70

Table IV. Efciency scores in percentages and ranks for Model 2

by the BCC-O model is primarily due to operations. Table V shows potential improvement for the operations of coffee stores in the BCC-O model. Inefcient stores show the high levels of improvement needed with respect to their retail sales, such as non-coffee items. For diversication and efciency, they need to increase their retail sales in addition to restaurant sales. Among the resources, wages and benets are major areas for improvement or reduction. Last, we include a Location variable in Model 1, using one for the stores in an afuent residential area (Group 1) and two for the stores in a mostly business district nearby residential areas (Group 2). The two areas are geographically separated and demographically different. We have an even number of DMUs in the two areas and test a difference in efciency with bilateral DEA models. Table VI shows the result by the bilateral CCR-I and BCC-I models. The efciency scores in Table VI are between zero and one. For computing the efciency scores of bilateral models, assume that there are two groups of DMUs such as G1 and G2. We compute efciency scores of DMUs in G1 against DMUs in group G2 or vice versa. The DMU n [ G1 is enveloped by DMUs in G2 and then the optimal efciency score is given by u * OA=OB , 1, where OA is the distance between the origin and the line connecting frontiers in Group G1 or G2, and OB is the distance between the origin and n. If the DMU n cannot be enveloped by DMUs in G2 then n should be expanded radially to B to be enveloped. The optimal efciency score is given by u * OA=OB . 1. Thus, the optimal efciency score is not bounded by one and can take a value larger than one. This property, which provides non-overlapping ranks for DMUs, is useful when we compare rank sums for two groups with a non-parametric statistical test such as the Wilcoxon-Mann-Whitney test. The rank sum statistics for both bilateral CCR-I and BCC-I models are identical. Rank sums are 36 for Group 1 and 100 for Group 2. The test statistic is 2 3.36067 with p-value, 0.00078. The difference between the rank sums is statistically signicant. Thus, we conclude that the DMUs in
Store 1_Current 1_Past 2_Current 2_Past 3_Current 3_Past 4_Current 4_Past 5_Current 5_Past 6_Current 6_Past 7_Current 7_Past 8_Current 8_Past Sales-restaurant 0.00 0.00 0.00 0.00 0.00 0.00 0.00 14.87 0.00 0.00 32.59 14.77 29.91 26.52 43.21 21.60 Sales-retail 0.00 0.00 0.00 0.00 0.00 0.00 0.01 20.86 0.00 0.00 93.59 287.29 342.92 999.90 43.21 229.83 Cost of sales 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 27.98 216.99 20.06 0.00 0.00 Wages and benets 0.00 0.00 0.00 0.00 0.00 0.00 0.00 2 4.87 0.00 0.00 2 19.18 2 15.90 2 16.31 0.00 2 14.65 2 42.88 Other expenses 0.00 0.00 0.00 0.00 0.00 0.00 20.01 220.86 0.00 0.00 29.45 0.00 0.00 230.81 21.58 224.60

The performance of coffee stores

749

Note: Positive number indicates increase and negative number means reduction

Table V. Potential improvement in percentages for Model 2 by BCC-O

BIJ 16,6

Store 1_Current 1_Past 2_Current 2_Past 3_Current 3_Past 4_Current 4_Past 5_Current 5_Past 6_Current 6_Past 7_Current 7_Past 8_Current 8_Past

Bilateral CCR-I Score Rank 1.5640 1.5482 2.9625 2.9023 1.3390 1.3710 1.0421 1.0659 0.6430 0.6571 0.6899 0.7559 0.6798 0.6648 0.6425 0.5819 3 4 1 2 6 5 8 7 14 13 10 9 11 12 15 16

Bilateral BCC-I Score Rank 1.9786 1.8996 3.1244 2.9747 1.7322 2.0437 1.7760 1.8000 1.0000 1.0000 0.7022 0.7644 0.6891 0.6861 0.6752 0.9212 4 5 1 2 8 3 7 6 9 9 13 12 14 15 16 11

Location 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2

750

Table VI. Comparisons of efciency by location

Group 1 or in the afuent residential area maintain higher operating efciency than the DMUs in the business area. 6. Conclusion Curious and challenged by the coffee industry and specically gourmet coffee shops, and by the brewing of coffee drinks and serving of light meals, we treat coffee stores as restaurants and attempt to measure their operating efciency using internal data along with an external factor, namely, the location variable that reects market characteristics. We have data for eight coffee retail stores for two years, current and past year, and maintain 16 DMUs per DEA model. We have two sets of variables with different combinations of input and output variables and then test the two sets with various DEA models such as input-oriented CCR and BCC models, output-oriented CCR and BCC models, and input-oriented bilateral CCR and BCC models. The rst set of variables includes one output variable: total sales and four input variables: cost of sales, wages and benets, other expenses and occupancy expenses. The second set of variables has two output variables: sales-restaurant and sales-retail and three input variables: cost of sales, wages, and other expenses. By applying CCR and BCC models together to the two sets of variables, we have identied the sources of inefciency along with the comparative operating efciency of the selected coffee stores. For the rst set of variables, we nd that the relatively inefcient stores in the input-oriented BCC model show higher inefciency with respect to occupancy related expenses than on other expenses. Regarding the second set of variables, the stores in the output-oriented BCC model lack a signicant level of efciency for sales from non-coffee items. An analysis of the make-up of said items is suggested resulting in additions/reductions/or a different mix of these non-coffee items. In the bilateral models, the coffee stores located in the afuent residential area outperform the stores in the business district. Overall, this approach is useful for measuring the performance of coffee retail stores and provides managerial insights into the

company. This study is also useful for benchmarking purposes as it identies successful operations, which could be modeled by management. The contributions of this study are twofold: providing a framework for performance measures within a benchmarking perspective and evaluating the comparative operating efciency of coffee stores using real data. Finally, the limitation of this study resides within the use of nancial data. To overcome this limitation, the company needs to collect and maintain non-nancial data based on various perspectives. For example, the balanced scorecard approach is an excellent candidate for selecting variables and evaluating the performance of stores using DEA.
References Athanassopoulos, A.D. and Thananssoulis, E. (1995), Separating market efciency from protability and its implications for planning, The Journal of Operational Research Society, Vol. 46 No. 1, pp. 20-34. Banker, R.D., Charnes, A. and Cooper, W.W. (1984), Some models for estimating technical and scale inefciencies in data envelopment analysis, Management Science, Vol. 30, pp. 1078-92. Charnes, A., Cooper, W.W. and Rhodes, E. (1978), Measuring the efciency of decision making units, European Journal of Operational Research, Vol. 2 No. 6, pp. 429-44. Chilingerian, J.A. (1995), Evaluating physician efciency in hospitals: a multivariate analysis of best practice, European Journal of Operational Research, Vol. 80 No. 3, pp. 548-74. Chilingerian, J.A. and Sherman, H.D. (1994), Evaluating and marketing efcient physicians toward competitive advantage, Health Care Strategic Management, Vol. 12 No. 5, pp. 16-19. Chu, H., Liu, S., Romeis, J.C. and Yaung, C. (2003), The initial effects of physician compensation programs in Taiwan hospitals: implications for staff model HMOs, Health Care Management Science, Vol. 6 No. 1, pp. 17-26. Cullinane, K., Ji, P. and Wang, T. (2005), The relationship between privatization and DEA estimates of efciency in the container port industry, Journal of Economics and Business, Vol. 57 No. 5, pp. 433-62. Debnath, R.M. and Shankar, R. (2008), Benchmarking telecommunication service in India: an application of data envelopment analysis, Benchmarking: An International Journal, Vol. 15 No. 5, pp. 584-98. Donthu, N. and Yoo, B. (1998), Retail productivity assessment using data envelopment analysis, Journal of Retailing, Vol. 74 No. 1, pp. 89-105. Duffy, J., Fitzsimmons, J. and Jain, N. (2006), Identifying and studying best-performing services; an application of DEA to long-term care, Benchmarking: An International Journal, Vol. 13 No. 3, pp. 232-51. Farrell, M.J. (1957), The measurement of productive efciency, Journal of the Royal Statistical Society, Series A, Vol. 120, III, pp. 253-90. Ferrier, G. and Valdmanis, V. (2004), Do mergers improve hospital productivity?, The Journal of Operational Research Society, Vol. 55 No. 10, pp. 1071-80. Fumero, F. (2004), Multiple solutions identication in data envelopment analysis, Central European Journal of Operations Research, Vol. 12 No. 3, pp. 307-22.

The performance of coffee stores

751

BIJ 16,6

George, S.A. and Rangaraj, N. (2008), A performance benchmarking study of Indian railway zones, Benchmarking: An International Journal, Vol. 15 No. 5, pp. 599-617. Goncharuk, A.G. (2008), Performance benchmarking in gas distribution industry, Benchmarking: An International Journal, Vol. 15 No. 5, pp. 548-59. Graham, A. (2005), Airport benchmarking: a review of the current situation, Benchmarking: An International Journal, Vol. 12 No. 2, pp. 99-111. Haas, D.A. (2003), Compensating for non-homogeneity in decision-making units in data envelopment analysis, European Journal of Operational Research, Vol. 144 No. 3, pp. 530-44. Hoover (2007), Industry overview: coffee shops, available at: www.hoovers.com/coffee-shops/ ID_264/free-ind-fr-prole-basic.xhtmL (accessed 18 November 2007). IMCR Case Studies and Management Resources (2007), Starbucks: evolution of a global brand, available at: www.icmrindia.org/casestudies/catalogue/Business%20Strategy3/ BSTA064.htm (accessed 27 November 2007). Johnes, J. (2006), Measuring efciency: a comparison of multilevel modeling and data envelopment analysis in the context of higher education, Bulletin of Economic Research, Vol. 58 No. 2, pp. 75-104. Joo, S., Stoeberl, P. and Kwon, I. (2007), Benchmarking efciencies and strategies for resale operations of a charity organization, Benchmarking: An International Journal, Vol. 14 No. 4, pp. 455-64. Kwon, H., Stoeberl, P.A. and Joo, S. (2008), Measuring comparative efciencies and merger impacts of wireless communication companies, Benchmarking: An International Journal, Vol. 15 No. 3, pp. 212-24. Lan, C., Lan, Y., Chang, C. and Chuang, L. (2006), Efciency-based recruitment plan for chained quick-service enterprise, Human Systems Management, Vol. 25 No. 3, pp. 197-209. McEachern, D. and Paradi, J.C. (2007), Intra- and inter-country bank branch assessment using DEA, Journal of Productivity Analysis, Vol. 27 No. 2, pp. 123-36. Mahajan, J. (1991), A data envelopment analytic model for assessing the relative efciency of the selling function, European Journal of Operational Research, Vol. 53 No. 2, pp. 189-205. Min, H. and Joo, S. (2006), Benchmarking the operational efciency of third party logistics providers using data envelopment analysis, Supply Chain Management: An International Journal, Vol. 11 No. 3, pp. 259-65. Reynolds, D. (2003), Hospitality-productivity assessment using data envelopment analysis, Cornell Hotel & Restaurant Administration Quarterly, Vol. 44 No. 2, pp. 130-7. Reynolds, D. and Gu, Z. (2004), An exploratory investigation of multiunit restaurant productivity assessment using data envelopment analysis, Management Science Applications in Tourism & Hospitality, Vol. 16 Nos 2/3, pp. 19-26. Reynolds, D. and Thomson, G.M. (2007), Multiunit restaurant productivity assessment using three-phase data envelopment analysis, International Journal of Hospitality Management, Vol. 26 No. 1, pp. 20-32. Sherman, H. and Ladino, G. (1995), Managing bank productivity using data envelopment analysis, Interfaces, Vol. 25 No. 2, pp. 60-73.

752

About the authors Seong-Jong Joo is an Associate Professor of Supply Chain Management at the School of Business, Clayton State University in Atlanta, Georgia. He teaches graduate and undergraduate courses in Supply Chain Management. His research interests include sourcing/purchasing, supply chain collaboration, inventory management and performance measures/benchmarking. Seong-Jong Joo is the corresponding author and can be contacted at: seong-jongjoo@clayton.edu Philipp A. Stoeberl is the Mary Louis Murray Professor of Management and serves as the Interim Chair of the Department of Management at the John Cook School of Business, Saint Louis University. His teaches both graduate and undergraduate courses in Strategy and Current Issues in Management. His current research interests include performance measures. Kristin Fitzer is a recent graduate of Central Washington University and works in Seattle, Washington.

The performance of coffee stores

753

To purchase reprints of this article please e-mail: reprints@emeraldinsight.com Or visit our web site for further details: www.emeraldinsight.com/reprints

You might also like