Effective Strategies For Internal Outsourcing and Offshoring of Business Services: An Empirical Investigation O. Zeynep Aksin1 and Andrea Masini23 11 December 2005 1 College of Administrative Sciences and Economics, Koc University, [email protected] 2 Corresponding author: Department of Operations and Technology Management, London Business School, [email protected] 3 Author names are listed in alphabetical order. Abstract The growing pressure to reduce costs and improve efficiency induces many organizations to undertake shared services initiatives. This consolidation and streamlining of common business functions is also known as insourcing, in-house services, business services, or staff services. While adoption of a shared service structure is viewed by many as an appropriate strategy to pursue, most companies still struggle to devise optimal strategies and to generate adequate returns on investments for their projects, because none of the approaches that are commonly adopted is recognized as universally effective. This paper builds upon the “structure-environment” perspective to uncover configurations of Shared Services organizations and to explain why and under what circumstances some of these configurations exhibit superior results. The conceptual model proposed challenges the notion of “best practice” and suggests that the effectiveness of a shared services project depends on the degree of complementarity between the “needs” arising from the environment in which a company operates and the specific capabilities developed to address these needs. The theoretical findings are validated empirically through the analysis of a large sample of European firms that recently undertook initiatives in this domain. Four dominant configurations of shared service organizations are uncovered, and their relationship to performance is explored. Keywords: shared services, offshoring, outsourcing, strategy, empirical analysis 1 INTRODUCTION Increasing competition, the progressive globalization of the economy caused by the lowering of trade barriers, and the emergence of new market players that benefit from significant labor cost advantages is pressuring western firms to devise strategies to reduce costs and improve the efficiency of operations. Among the various alternatives the development of shared services (SS) is increasingly regarded as a potential solution to this problem and is being progressively adopted by a growing number of firms. For example Citigroup has established Citigroup Business Services, a shared services organization located in three regional centers around the globe, dealing with financial reporting, payroll and benefits administration, purchasing, accounts payable, and premises management functions for the entire group. Shared Services is the strategy of standardizing, streamlining, and consolidating common business functions and processes in an organization, in order to improve efficiency and effectiveness with both cost reduction and overall profitability in mind. According to Bywater (2001) it is: “The co-location of internal services which are removed from the business to whom they provide the service-internal outsourcing”. In a report by the Economist Intelligence Unit (1998), shared services are described as: “The apportioning of standardized and consolidated business functions or processes with a service mentality to ensure effective operation”. “Shared services-also known as insourcing, in-house services, business services, or staff services-is more than centralizing routine transactions: the idea is to leverage best practices, specialized knowledge, and technology to create satisfied internal and external customers” (The Conference Board, 1998). According to Goold et al. (2001) it is one of the roles that corporate headquarters can take. In general, it is believed that moving to a shared service organization (SSO) helps companies save costs, increase available time for value-added activities in line positions, 1 improve measurement capability, and achieve better service quality due to a more focused management attention. American Express and Citigroup report cumulative cost savings three years after adoption worth more than half a billion and 350 million US dollars respectively. Average cost savings of 25-30% are not unusual after adopting a shared service structure (Quinn et al. 2000). Not surprisingly, an increasing number of companies consider adopting SSOs. According to a study by Bywater (2001) over 90% of Fortune 500 and Europe 500 companies have already or are planning to implement SSOs. Shared service organizations are seen by some as the step taken before outsourcing, and by others as the alternative to corporate outsourcing. Sometimes the shared service organization is a third party owned business unit. Thus, in the current debate on business process outsourcing practices and offshoring of services, understanding shared service structures and their performance implications will constitute an important piece. To answer questions about whether to outsource or not, companies need to understand appropriate in- house solutions better. Similarly, the question of offshoring requires a good understanding of service structures that guarantee high performance. Yet, in spite of this rising interest, there is still some uncertainty about the real benefits of shared services. For instance, of the companies that participated in the Bywater study, 3% are reported to have implemented a shared services strategy, and then rejected it. This suggests that while rarely reported, shared service implementations can also fail. According to Steven Kerr from General Electric (Quinn et al. 2000): “Shared services, like outsourcing, is not a panacea for all functions. Sometimes it works and sometimes it is not the right strategy - especially if it has been forced upon reluctant business units. It is not a hammer!” By the same token, there is also uncertainty about the most appropriate ways to conceive, implement and manage shared services organizations. A number of alternative - and often antithetical - approaches to SSOs have been proposed and implemented. As a result, 2 existing SSOs differ among each other with respect to their overall objectives, their functional scope and even with respect to the specific business model adopted, which span from purely introverted approaches to more extroverted ones that privilege customer service over cost reduction (Citigroup, 2002). Yet, none of these models has proven to be generally superior. The blurred picture on the practice side is also symptomatic of a knowledge gap at the theoretical level. To our knowledge, there is no academic study to date that examines the link between strategy, implementation and resulting performance in SSOs. Claims are mostly based on conceptual arguments, perhaps supported by anecdotal evidence that associates certain variables and certain performance measures. However, an empirical justification of a link between SSO strategies and performance is lacking. Given the prevalence of this organizational structure and the large potential benefits associated with its success, understanding how firms organize their SSOs and determining superior configurations is a critical issue for businesses at large. Researchers have also generally overlooked the question of whether different SS strategies display the same degree of effectiveness for firms that have different operational and organizational needs. This is an important gap, because recent studies that revisited the relationships between strategy, organizational structure and management processes have challenged the notion of “single best practice” (Atuahene-Gima and Ko, 2001) and suggested that similar strategies, organizational forms and – by analogy, similar SS models, may produce different results in different environmental settings (Brown, 1994; Brown and Magill, 1998). In summary, a review of the extant literature suggests that there is a need for a sounder characterization of the predominant SSO models (beyond the simple stylized classifications proposed by industry practitioners) and, also, of a better understanding of the relative effectiveness of these models. 3 The objective of this paper is to fill this gap in the literature and to examine the relationship between different SS strategies and the characteristics of environments in which SSOs operate. Recognizing the need for exploratory research that shed light on the predominant SS models and following the increasing interest received by organizational gestalts (Hambrick, 1984; Miller, 87 and 90, Meyer et al. 1993) in the field of operations strategy (Miller and Roth, 1994; Bensaou and Venkatraman, 1995; Dennis and Meredith 2000), we conduct a configurational analysis. That is, we seek to understand whether SSOs organize themselves according to “internally consistent combinations of strategy, organizational architecture and technology that provide superior performance in a given environment” (Tidd and Hull, 2002, p. 7). To this end, we adopt a deductive and theory-driven approach (Ketchen and Shook, 1996), which permits the development of testable frameworks while still maintaining enough degrees of freedom to uncover “natural” configurations. Our contribution is twofold. First, using data from the Citigroup Business Services Annual Survey of European Shared Services 2002 we identify strategic groups of companies that exhibit similarities with respect to their SS operations and their environment (i.e. configurations of SSOs). Second we use the proposed classification to explain performance differences across different companies in the sample and to shed some light on the relationship between the characteristics of SSOs and their business results. The remainder of the paper is organized as follows. Section 2 discusses some of the most common approaches to shared services and provides a theoretical background. A stylized conceptual model is presented to uncover configurations based on the hypothesis that – to improve performance – firms need to achieve a fit between their shared service needs and their SS strategy. The variables that define SS needs and SS strategy are then introduced. Section 3 describes the methodology and our overall analytical approach. Section 4 presents 4 and discusses the results. We first establish four different clusters of shared service organizations, and subsequently illustrate the link between these clusters and organizational performance. The paper ends with concluding remarks and a discussion of contributions, limitations and directions for future research. 2 BACKGROUND AND LITERATURE REVIEW 2.1 Alternative Models for Shared Services Organizations The diversity of organizational forms and management approaches observed in existing shared service organizations is symptomatic of the many decisions managers face in setting- up and running SSOs. These involve both the general objective and the scope of the SSO, the business model adopted as well as the specific strategy followed for its implementation. First and foremost, the motivations for establishing an SSO can be very different. For instance, while for some companies the primary objective is cost cutting, others consider service level improvement as the most important priority. Standardization of processes or focusing on core capabilities are other common motivations. The functional scope can vary as well. Whereas most SSOs start with the transactional and administrative side of the finance and human resource functions, the scope of activities can also extend to information technology, supply and support functions. It is possible for an SSO to encompass portions of the professional and technical aspects of these functions as well. The Amherst Group Limited and Solutia Inc. distinguish between transaction-intensive services and knowledge based professional services (The Conference Board, 1998). Labeled as centers of scale and centers of expertise respectively, these services are then managed and funded differently given their differing strategies. In addition to differences in functional scope and depth, there are also differences in the business models adopted (Bywater, 2001; Citigroup, 2002). A typical model follows an introspective approach, where the organization is treated as a cost center that serves internal 5 clients. In a different and more extroverted model, the shared service organization is run as a customer-centric separate business entity. In some cases the SSO also serves external clients, as in the well known example of Shell Services International. The Bywater (2001) study labels these two business models as the incremental approach and the aggressive approach respectively. Quinn et al. (2000) describe an evolution from a basic model to one of a separate business entity. A similar spectrum of conservative, moderate, advanced SS strategies are suggested by Connell (1996). Similarly, there is no general agreement about the most effective approaches to implement an SSO, as one finds differences in implementation strategy as well as performance. Existing SSOs exhibit differences with respect to their location (greenfield, i.e. a new site vs. brownfield, i.e. an existing building), size, technological infrastructure, and human resource practices. For instance, while most claim that business potential is maximized with an aggressive approach that designs the SSO as a separate business entity (The Conference Board 1998, Quinn et al. 2000, Bywater 2001, Citigroup 2002), others argue that success requires a fit between strategy and implementation (The Conference Board 1998, Economist Intelligence Unit 1998). This diversity is clearly symptomatic of the fact that no superior SS model has yet been identified. Indeed, a preliminary analysis conducted on the data set collected for this research indicates a very weak direct relationship between specific managerial choices and performance. For instance we find very limited support for the common claim that the adoption of a customer-centric extroverted model leads to success in SSOs. Sometimes a more introspective approach can result in better outcomes. However, the fact that similar models have been applied in different contexts with very different results suggests that the proposed dichotomy between introverted and customer- centric approaches is not exhaustive. Combined with the observation that some choices are 6 clearly restricted by parameters outside the control of SSO management, this also suggests that rather than looking for best models as such, managers should consider alternative SS configurations in relation to the specific needs and constraints of their firm. Based on these preliminary observations and supported by the contingency view in strategy we argue that different types of environments call for different types of SSO structures and practices. Testing this view requires an elaboration of common environments and common SSO models, or in other words the identification of dominant SSO configurations. 2.2 A configurational approach to study SSOs The notion of organizational configurations or gestalts (Miller and Friesen, 1977) is one of the most well-established concepts in strategy research. According to Miller (1981, p5): “Instead of looking at a few variables or at linear associations among such variables we should be trying to find frequently recurring clusters of attributes or gestalts”. Conceptual arguments for configurations and an overview of the early literature can be found in Miller (1986) and Miller (1987). While the notion of gestalts has gained increased recognition from a conceptual standpoint, researchers have followed different approaches to study configurations, and they have traditionally distinguished between conceptually established typologies (Hambrick, 1984) and empirically derived taxonomies (Hatten and Hatten, 1985). Clustering approaches have been the primary method of investigation to establish taxonomies and have also gained acceptance outside the strategy literature. They have for example been used in operations management (Miller and Roth, 1994; Boyer et al.,1996; Boyer et al., 2000; Dennis and Meredith, 2000; Verma and Young, 2000; Duray et al., 2000; Kathuria, 2000; Jonsson, 2000; Diaz et al., 2000; Yeung et al., 2003) and information systems (Segars and Grover, 1999; Lee et. al., 2004) settings. 7 In this paper, we also establish configurations empirically using cluster analysis, as in Hambrick (1984) and Bensaou and Venkatraman (1995). However, while most applications of cluster analysis are exploratory in nature, and do not test a conceptual model – our work has the distinguishing feature of combining the two approaches. The point of departure of our analysis is the well-known theoretical perspective that, for purposes of effectiveness, firms should deploy strategies or develop capabilities in accordance with the requirements of the environment in which they operate. While it enjoys a long history in management literature (Lawrence and Lorsch, 1967; Thompson, 1967), this perspective has taken different conceptualizations in the various disciplines that it has - directly or indirectly - influenced: in economics with the notion of complementarity (Milgrom and Roberts, 1990), or in operation management, where both production systems (Keller et al., 1974) and supply chains (Fisher, 1997) and supplier-buyer relationships (Bensaou and Venkatraman, 1995) have been analyzed in relation to different environmental contingencies. Building upon this perspective, we propose a conceptual model of fit that describes the main decision variables in setting up a shared service organization vis à vis the most important requirements that this organization must address (Figure 1). We suggest the general hypothesis that in shared service organizations performance is given by the presence of fit between environmental needs (i.e. factors that influence performance of an SSO not controlled by the manager of an SSO) and the combination of specific SS strategic, tactical and operational decisions made by the SSO manager. For sake of simplicity we refer to this set of variables as ‘SS capabilities’. Accordingly, to establish configurations we follow the two-stage procedure described in Hambrick (1984) and Bensaou and Venkatraman (1995). In the first stage we cluster organizations based on their shared service needs, and in the second stage, within each SS needs group we cluster organizations based on their shared service capabilities. 8