A framework for benchmarking service process using data envelopment analysis and decision tree

Abstract

With the increase in the influence of service industry on the economy, more and more firms are becoming concern about service and service management. Unlike tangible products, since service is a delivery system, designing and controlling delivery process are important in management of services. One of the most important issues in managing service delivery process is how to find out and improve inefficient process in an organization. Once inefficient processes are found, it should be determined which process is to be improved first, since it is impossible for any firm to improve all of their inefficient processes simultaneously. Then, in order to get some ideas about improvement, selecting a benchmark process is required. The objective of this paper is to provide a systematic approach for benchmarking service process. We used the integrated form of data envelopment analysis (DEA) and decision tree (DT) as a benchmarking method instead of using DEA alone. Although DEA is a useful technique for benchmarking, it fails to give any clear directions as to which process should be improved first. The proposed approach enables firm’s manager to find inefficient service units in a firm-level and inefficient processes in a service unit-level. The proposed approach also allows the manager to select which process to improve first and which process a benchmark for referencing. As a result, it can be helpful for operating and managing firms’ processes. To show how useful the suggested service process benchmarking framework is, an illustrative example is addressed.

Introduction

Since the 1990s, the service sector has emerged as the fastest growing sector compared to the most economies. In addition, the share of services in production and employment has been increasing more and more (Banga, 2005). In fact, more than half of GDP is created by the service sector in developed countries, and it is expected that job markets and economies will depend on the service sector in the 21st century (Pilat, 2000). With the increase in the influence of service sector, the importance of service management is also increasing. Because service is a delivery system, unlike tangible products, designing and controlling delivery process are crucial in service management. With respect to the level of importance of the service industry and the unique characteristics of service, process design and implementation are important elements in determining the overall competitiveness of a modern organization (Fitzsimmons & Fizsimmons, 2001).

There have been some researches showing the importance of processes in assessing the performance of a firm (Chase, 1981, Chase and Tansik, 1983, Roth and Jackson, 1995, Roth and vand der Velde, 1991, Shostack, 1987). Particularly, Roth and Jackson (1995) showed empirically that process capability and execution is one of the key drivers of performance because it influences customer satisfaction and service quality. In addition, they showed that an improper process design and the poor performance of a proper process can cause process inefficiency. Hence, it is important for a firm wanting to improve its performance to identify and improve inefficient processes, especially in the service sector because of the uniqueness of the service characteristics. However, it is impossible for any firm to improve all their processes simultaneously, it is necessary to prioritize the processes that need to be changed (Hammer & Champy, 1993). In addition, it is important to find a benchmark process in order to get some ideas about how the firm could improve its processes. In fact, this is regarded as one of the most important factors for process reengineering (Davenport & Short, 1990).

Frei and Harker (1996) suggested a useful approach that can be used to find out a target process for process improvement. They evaluated the relative efficiency of a specific process over several organizations (firms or business units of firm) and decided which process to be improved and benchmarked by using data envelopment analysis (DEA). As organizations are generally composed of multiple processes, Frei and Harker (1999) suggested a method to assess overall efficiency of organization by aggregating process efficiency. Despite the contributions of their studies (Frei and Harker, 1996, Frei and Harker, 1999), there exists some limitations as follows:

• First, although their research is meaningful in that they evaluate the relative efficiency of a single process, it fails to determine which process should be improved first among the various processes in that organization.

• Second, they did not consider the fact that the processes are also managed in the organizational-level, when evaluating overall efficiency of one organization versus other using process-level efficiency. Although a process is composed of same activities and activity flows, it can be managed differently in each organization. Therefore, the same process may impact to the organization’s efficiency differently.

This paper is overcoming above-mentioned limitations of the studies by Frei and Harker. We propose a systematic framework, which can reflect the relative impact of processes to the organization’s efficiency, in order to evaluate the overall efficiency of an organization based on process-level efficiency. Also, this framework provides which process that should be considered first to improve the efficiency of organization in its own view. By using the suggested framework, the manager of the firm can identify inefficient service units (e.g., in the case of a bank, the branch office of the bank; the same example will be used hereunder) in a firm-level and inefficient processes (e.g., the process of opening a checking account, small business loan process, home equity loan process, etc.) in a service unit-level. S/he can also select which process that she deems as inefficient service units to be improved first and which process s/he should used as a benchmark.

In our study, we used DEA and decision tree (DT) as our main methodologies. DEA is a method for measuring the relative efficiency of decision making units (DMUs) with the number of inputs and outputs. It is a non-parametric approach that does not require any assumptions about a functional form of a production function (Charnes, Cooper, & Rhodes, 1978). DT is a method usually used to find out meaningful relationships and rules by systemically breaking down and subdividing the information in the data (Chen, Hsu, & Chou, 2003). There are some researches (Lee and Park, 2005, Sohn and Moon, 2004) that used these methodologies together. Lee and Park (2005) employed these methods for providing profitable customers segmentation system and Sohn and Moon (2004) made use of them for forecasting the degree of new technology commercialization. However, both of them not fully used the advantage of DEA. They only focused on evaluating efficiency of DMUs and did not use the advantage of DEA for benchmarking.

The processes of the proposed systematic approach for benchmarking service process are as follows. First, efficiency of each process in the service units is evaluated on the basis of inputs and outputs of each process by DEA (with CCR model). In the evaluation of efficiency of each process, the process is the DMU. Next, the overall efficiency of a service unit is assessed based on the efficiency of each process. To do so, the pure output (without input) DEA model suggested by Lovell and Pastor (1999) is used. This is because while the efficiency of each process is used as output factor for DEA analysis, there is no input factor. Since the overall efficiency of service unit is assessed by DEA, we can reflect relative impact of a process to the service unit’s efficiency. At this stage, each service unit is regarded as the DMU. Then, a rule to select inefficient process for improving service unit’s overall efficiency is created through a DT, which is applied to investigate the influences and roles of each process to overall efficiency of service unit.

The remainder of this paper is organized as follows. Section 2 gives an overview of DEA and DT. Section 3 explains and illustrates the service process benchmarking framework which is suggested in this research. Section 4 presents and explains the example to demonstrate the usefulness of the suggested framework. Finally, Section 5 presents the conclusions drawn from this study along with some implications and a summary of further research initiatives.