1
Reynolds, D., & Biel, D. (2007). Incorporating satisfaction measures into a restaurant productivity index. International Journal of Hospitality Management, 26, 352-361
(Awarded Best Paper of the Year 2007 by IJHM).
INCORPORATING SATISFACTION MEASURES INTO A RESTAURANT PRODUCTIVITY INDEX
Dennis Reynolds
College of Business and Economics
WashingtonStateUniversity
Todd Hall 477
Pullman, WA99164
509-335-4344
509-335-3857 (fax)
David Biel
School of Hotel Administration
CornellUniversity
119 The Knoll
Ithaca, NY14850
An earlier version of this manuscript received the “Editor’s Choice Award” at the 2005 International Conference on Services Management in Delhi, India. We thank the meeting-paper reviewers, and especially Dr. Vinnie Jauhari,Conference Program Chair, as well as the IJHM editor and his reviewers for their advice and counsel.
Dennis Reynolds, Ph.D., is the Ivar Haglund Distinguished Professor of Hospitality Management at the Washington State University School of Business and Economics in Pullman, Washington, USA.
David Biel is a research assistant and student at the Cornell University School of Hotel Administration in Ithaca, New York, USA.
INCORPORATING SATISFACTION MEASURES INTO A RESTAURANT PRODUCTIVITY INDEX
ABSTRACT
The increasing stature of the foodservice industry in the global service economy suggests that productivity analyses—similar to those performed in non-service-based settings—would benefit multiunit operators by maximizing their desirable operational outcomes while minimizing expenses and other detrimental conditions such as low job satisfaction. This paper suggests that such analyses might be possible through the application of a holistic productivity metric—one that includes traditional operational variables such as revenue, profit, food cost, and labor cost, and previously ignored variables such as guest and employee satisfaction as well as retention equity. Through data gathered from a single chain’s 36 corporate-owned, same-brand casual-theme restaurants located in metropolitan centers across the United States, we found that factors leading to maximum outputs such as controllable profit and retention equity include employee satisfaction in addition to expected variables such as cost of goods sold and number of seats. Most notably, employee satisfaction as an input proved to be the most volatile variable in maximizing operational outputs.
Keywords: Productivity, data envelopment analysis, employee satisfaction, guest satisfaction
INCORPORATING SATISFACTION MEASURES INTO A RESTAURANT PRODUCTIVITY INDEX
In today’s highly competitive global foodservice arena, multi-unit restaurant operators are embracing every opportunity to maximize their operational efficiency. The methods used to benchmark and assess productivity have been limited, however, to overly simplistic approaches and, as a result, have offered limited utility. As Sigala (2004) noted, “Although there have been attempts to identify satisfactory productivity-monitoring procedures, these have been heavily criticized and no generally accepted means of productivity measurement exists” (p. 39).
Consider also the increasing complexity of restaurant operations, which, conjoined with constantly escalating consumer expectations and demands, adds to the challenges faced by foodservice operators. Traditional partial-factor productivity statistics, such as meals-per-labor hour, simply do not reflect adequately the many factors that influence the metric. Moreover, service-process matrices, such as those detailed by Schmenner (1986), offer constructive guidelines for assessing productivity but are difficult to integrate into some distinct service-industry segments such as foodservice.
Following Reynolds’s (1999) call for a productivity index that is truly holistic and Kohli and Jaworski’s (1990) assertions regarding business analyses that include a customer-service orientation, this study explores several input and output variables while introducing a comprehensive productivity metric. In particular, we investigate the effects of the customary unit-level financial indicators (e.g., profit, revenue, cost of goods, labor cost, occupancy cost), constraining variables (e.g., number of seats, square footage), and previously unexplored variables such as employee and guest satisfaction; we also consider related factors such as retention equity. The study makes its primary contribution in developing a more holistic productivity metric than those considered previously. We also evaluate the importance of guest and employee satisfaction data in assessing unit-level restaurant productivity.
LITERATURE REVIEW
Researchers have demonstrated that superior financial performance correlates strongly with productivity. Most notably, Schmenner (2004) provided theoretical support for this relationship using the service businesses as the primary focus. Others have presented empirical support using a variety of service-industry segments including lodging (Morey and Dittman 1995; Hu and Cai 2004), midscale restaurants (Reynolds 2004), and on-site foodservice (Reynolds 1998).
Correspondingly, the use of and focus on productivity has evolved dramatically during the last 25 years. This development has been fueled largely by increasingly stringent resource constraints with a disproportionate rise in labor-related expenses. Given the labor intensiveness of hospitality-related businesses, interest in productivity analyses has focused predominately on labor and its corollaries: service outcomes per employee (e.g., rooms cleaned, meals served), labor hour, or labor-dollar value (Ball, Johnson, and Slattery 1986).
Variable Identification
Building on Reynolds’s (1998) definition of productivity as the effective use of resources to achieve operational goals, researchers and practitioners have recently acknowledged the importance of productivity analyses that are more comprehensive than any single-factor indices. Brown and Dev (1999), for example, posited the use of capital productivity output by considering three single-factor statistics, including income before fixed charges per full-time equivalent employee. Similarly, Jones and Hall (1996) used multiple single-factor statistics as a means to consider desirable service-delivery outcomes such as perceived service relative to labor expenditure.
In moving toward what Bucklin (1978) termed “multi-factor” productivity analysis, Sigala (2004) aptly explained that the largest problem is the identification of inputs and outputs. Reynolds (1998) added that any meaningful productivity statistic must not only accurately identify inputs and outputs, but must integrate all critical variables if such a measure is used to assess overall operational productivity.
So which variables are critical for a holistic productivity measure in a restaurant setting? Ball, Johnson, and Slattery (1986) suggested that three broad categories of variables are essential: financial, physical, and composite (reflecting financial and physical variables). Furthermore, many researchers (e.g., Rimmington and Clark 1996) have explained that discreet quantitative measures, such as revenue and net profit, are ideal since these encapsulate broader aspects of the operation.
Regarding outputs, then, several researchers have demonstrated the criticality of revenue (e.g., Pilling, Donthu, and Henson 1999; Thore, Phillips, Ruefli, and Yue 1996). Perhaps of greater importance to operators, profit—often calculated as controllable income to reflect profit before corporate overhead costs are allocated—has proved vital to any robust measure of unit-level productivity (Grifell-Tatjé and Lovell 1999). Moreover, consideration of both revenue and profit is considered paramount to gauging an individual operating unit’s financial viability (Ingenito and Trehan 1996).
Guest satisfaction has been the most elusive output variable, yet many consider it the most important as an indicator of long-term success. Lee-Ross (1994) and Witt and Witt (1989) underscored the importance of service quality and guest satisfaction even before complex analyses of productivity in the service industry had been developed. This output has also been identified as a consequence of service quality that explains a considerable portion of customers’ intentions to purchase (Brady, Cronin, and Brand 2002). Central to the notion that quality service leads to positive guest satisfaction, Mohr and Bitner (2000) reported that the effort of the service provider has a strong positive influence on satisfaction with the transaction as an output.
Finally, two empirical studies focusing on guest satisfaction as a viable output variable in productivity analyses offered strong evidence of its significance (Lothgren and Tambour 1999; Parasuramam, Zeithaml, and Berry 1994). Heskett et al. (1994) highlighted this finding by noting how one quick-service restaurant chain has found units in the top quadrant in guest-service ratings outperform the others by all measures. Ojasalo (2003) summarized the point well: “Due to customers’ increasing influence . . . they cannot be regarded as passive recipients of the provider’s outputs, but should be seen as an integrated part” (p. 14).
Related to guest satisfaction is retention equity. Rust, Zeithaml, and Lemon (2000) define retention equity as the strength of the relationship between the customer and the firm. Retention equity is a potentially valuable output variable, as these authors note, since high retention equity indicates “the customer’s tendency to stick with the brand, above and beyond objective and subjective assessments of the brand” (p. 95). Furthermore, retention equity is linked to emotional ties that the customer has with the brand. Typically measured through purchasing frequency, retention equity indicates customers’ perception of extraordinary benefits and relational linkages that make them very reluctant to switch to another dining option (Blattberg, Getz, and Thomas 2001).
As for inputs, financial measures that have proved important to productivity analysis include labor cost (Burritt 1967; Yoo, Donthu, and Pilling 1997), cost of goods sold (Brown and Hoover 1990; Burritt 1967), controllable fixed expenses (Sarkis 2000), and uncontrollable expenses (Reynolds and Thompson 2002). Physical inputs that have proved important include service capacity, such as square footage or number of seats (Doutt 1984), and environmental characteristics, such as proximity to shopping centers and competitive conditions (Goldman 1992; Ortiz-Buonafina 1992).
Employee satisfaction intrigues us in having been posited as critical while going largely untested in productivity-analysis studies. Koys’s (2003) comprehensive study provided evidence of the strong relationship between employee satisfaction and restaurant performance. Similarly, Rucci, Kim, and Quinn (1998) offered both theoretical and case-study evidence of the effect of employee satisfaction on sales and profit. Spinelli and Cavanos (2000) identified a significant correlation between employee and customer satisfaction in a hospitality company. As substantiation for including this variable in productivity analyses, Kennedy, Lassk, and Goolsby (2002) presented evidence of an indisputable relationship between employee satisfaction and organizational goals such as sales and profit.
Data-Envelopment Analysis
Given the large number of aforementioned output and input variables, the challenge is to integrate and analyze them simultaneously so as to identify meaningful differences among operating units. As Caplow notes (1983), “An organization is efficient if, among similar organizations, its output is relatively high in relation to its input” (pp. 80-81). A widely accepted approach is Data-Envelopment Analysis (DEA), a non-parametric method that considers both controllable and uncontrollable variables and produces a single relative-to-best productivity index corresponding to each unit under comparison. Such a metric also allows operators, as recommended decades ago by Farrell (1957), to use the best performing units as the bases for evaluation.
As fully described by Charnes, Cooper, Lewin, and Seiford (2001), DEA extends the productivity ratio analysis by integrating the weighted sum of outputs to the weighted sum of inputs. In applying DEA, the weights are estimated separately for each restaurant to maximize efficiency. Moreover, the weights estimated for restaurant i are such that when they are applied to corresponding outputs and inputs from other units in the analysis, the ratio of weighted outputs to weighted inputs is less than or equal to 1 (interpreted as a percentage). On a more general basis, assuming that the number of outputs and inputs is infinite, the maximum efficiency of restaurant o as compared with n other restaurants is calculated as follows:
Maximum Po = subject to 1 for all j = 1,…n
Ur, Vi0; r = 1,…, s; i = 1,…m
where
Yrj is the rthoutput for the jth restaurant
Xij is the ith input for the jth restaurant
Ur and Viare the variable weights estimated and used to determine the relative efficiency of o
s is the number of outputs
m is the number of inputs
As Avkiran (2002) noted, DEA benchmarks units by comparing their ratios of multiple inputs to produce corresponding outputs and plotting them on a multidimensional frontier. Such a frontier allows for units that are most similar to be assessed by comparison with the top performers in their peer groups. Wöber (2002) explained that benchmarking is useful particularly when indicators span operations that are dissimilar but compete for similar target-market constituents. Using DEA-generated productivity indices facilitates such comparison. Furthermore, the method’s ability to take into account such a wide variety of output and input variables makes it ideal for hospitality applications (Reynolds 2003).
METHODOLOGY
The sample consisted of all 36 same-brand corporate units of a casual-theme restaurant chain with stores located in major metropolitan centers across the United States. The company was selected in part for the geographic diversity of its locations; the company also allowed us access to all financial information required for this study. These financial data were gathered from month-ending financial statements. Physical characteristics (e.g., number of seats, square footage) were also provided by the privately held corporation. The chain has positioned each unit in densely populated metropolitan settings using a uniform location-selection strategy; thus, environmental characteristics are similar across units.
Guest satisfaction data were obtained through a random sample of guests visiting each store during the period corresponding to the aforementioned financial information. The voluntary survey, which was distributed randomly to guests (one per table) and did not include any incentive except for the opportunity to provide feedback, included questions corresponding to the firm’s primary objectives: food quality, service quality, ambiance, value, and overall dining experience (α. = .91). We also asked questions regarding patronage frequency as a measure of retention equity, as discussed by Blattberg et al. (2001). On average, 32 surveys were completed per store with an average response rate of 72%. (One unit within the chain failed to complete this aspect of the study and was therefore dropped from the analyses.)
Employee satisfaction was assessed though a confidential, anonymous survey. The survey included questions corresponding to the job-descriptive index (see Kinicki, et al. 2002) and featured a summated five-point Likert scale (α. = .93). Owing to differing staffing levels, all front-of-the house employees (n = 37.8 per unit) were surveyed with a 100% response rate (average surveys per unit = 33.7).
In addition to traditional financial indicators such as revenue and cost of goods, we included rent and taxes and insurance as a method for adjusting for regional economies. Similarly, constraining variables such as number of seats and square footage were considered since larger restaurants should be expected to produce larger sales, which are proportional to greater expenses. Table 1 presents the complete list of factors for which we obtained data.
Table 1: Inputs and Outputs Considered
Variable / Input/Output / Measured As / Controllable/UncontrollableRevenue / Output / $ for the period / N/A
Controllable Income (profit) / Output / $ for the period / N/A
Guest Satisfaction / Output / Summated scale / N/A
Retention Equity / Output / Average purchase frequency / N/A
Cost of goods sold / Input / $ for the period / Controllable
Labor cost / Input / $ for the period / Controllable
Employee satisfaction / Input / Summated scale / Controllable
Rent / Input / $ for the period / Uncontrollable
Taxes and Insurance / Input / $ for the period / Uncontrollable
Square Footage / Input / Ft2 / Uncontrollable
Number of Seats / Input / Number / Uncontrollable
Prior to applying DEA, we ensured that each input was related to at least one output, that the inputs are independent, and that the outputs are independent. This analysis indicated a very strong relationship between total income and controllable income. While most firms pursue maximization of both sales and profit (e.g., Ingenito and Trehan 1996), the ultimate goal of this (and other) firms is improving the bottom line. Moreover, since the controllable input variables provide adequate scaling indices for subsequent steps in the analyses, we removed total sales as an output variable prior to proceeding.
Our next step, then, was to perform stepwise multiple regressions, using the output variables as dependent variables and the input variables as independent variables. As demonstrated in Table 2, while guest satisfaction did not appear to play a significant role, retention equity did contribute to the model. Similarly, rent and square footage proved unimportant for the subsequent DEA application.
As a confirmatory step, the data were then fit with a semiparametric regression model. As a variation to the more traditional linear model, this approach allows us to evaluate the empirical evidence with a broader application of the causal linkages previously suggested. Unlike classic regression methods, semiparametric methods automatically fit linear and nonlinear functional relationships (Ruppert, Wand, & Carroll, 2003). Thus, semiparametric regression models allow for better estimation since they can be formulated to include fewer unjustified assumptions than traditional regression models. These models supported the relationships identified in Table 2.
Furthermore, there were no substantial violations of the regression assumptions underlying this analysis. Additionally, no individual units were found to be overly influential in the analysis. We performed a Bootstrap model validation, with 5000 trials, to check the robustness of the model fit (Harrell 2001). The bias-corrected estimate of the coefficient of determination and the estimated global shrinkage parameters for the intercept and slope are - suggest a robust model fit with little evidence of being overfit.
Table 2: Relationships between Inputs and Outputs
Variables / Controllable Income / Guest Satisfaction / Retention EquityCost of Goods Sold / 1.36*** / N/A / .019**
Labor Cost / 1.71*** / N/A / N/A
Employee Satisfaction / N/A / N/A / 3.455**
Rent / N/A / N/A / N/A
Taxes and Insurance / 9.08** / N/A / N/A
Square Footage / N/A / N/A / N/A
Number of Seats / 1,134.84*** / N/A / N/A
*p<0.05, **p<0.01, ***p<0.001