Responsibility Centers: In Search of Its Determinants
-
A Conceptual Framework
Master Thesis
Coen Zieleman
Abstract: This study has been triggered by a management problem of Stork Netherlands, concerning the question whether or not to turn its after sales service department from a cost center into a profit center. Although widely applied within organizations, in the academic world surprisingly little research has been conducted in the field of responsibility centers (RCs). This study is the first of its kind that attempts to develop a comprehensive conceptual framework that explains both RC usage in general and the use of specific RC types. The likelihood of RC usage within an organization appears to be positively related to 1) the extent of decentralization; 2) the ability to measure the outputs of its agents and 3) to local cultures of its agents that are characterized by individualism and masculinity. A negative relationship has been established between RC usage and 4) both the degree of horizontal interdependencies and need for horizontal cooperation; 5) the extent of environmental uncertainty and 6) local cultures characterized by a high power distance. Furthermore, RC usage is indirectly influenced by the corporate strategy and the nature of technologies being used. With regard to the choice for a specific RC type, an organizational unit is likely to move from a cost or revenue center to a profit or investment center if 1) its size and capital intensity increases; 2) the extent of environmental uncertainty increases and 3) its business unit strategy shifts from a defender to a prospector. Moreover, as the facilitators „ability to measure outputs‟ and „knowledge of transformation process‟ increase in magnitude, the likelihood of a profit or even an investment increases as well. If these independent variables move in opposite direction, the latter facilitators are expected to be decisive in the decision whether or not to use RCs, and if so, which RC type to use. A limited field study has been conducted within Stork Netherlands and its Belgian subsidiary. The findings of this exploratory study appeared to be largely in line with the developed propositions. Finally, it has been concluded that it is both unfeasible and undesirable to turn Stork Netherlands‟ service department from a cost into an investment center.
Responsibility Centers: In Search of Its Determinants
-
A Conceptual Framework
Master Thesis
Name: Coen Zieleman
Student number: S1722093
Master Program: Master of Science in Business Administration Specialization: Organizational and Management Control Telephone number: 0630293021
Email: coenzieleman@gmail.com
PREFACE
Proudly I present to you the result of five months of intensively conducting research and its commensurate reporting. With this study – conducted in the light of my master graduation – I attempted to combine several theories in order to develop a framework that increases our understanding of the concept of responsibility centers. The past couple of months have been an enormously challenging and valuable experience to me. Analyzing a rather practical issue from an academic point of view has proven to be a perfect blend of theory and practice, and has kept me passionate and driven until the very end of the research process.
Acknowledgements
First of all, I am extremely grateful to dr. B. Crom. With his reviews and constructive feedback, he made a significant contribution to this paper. Without our pleasant cooperation, I had never been able to arrive at the point that I have reached now. Moreover, my thanks are indebted to co-assessor dr. E.P. Jansen, for reviewing and assessing this thesis. Furthermore, I gratefully acknowledge the contributions of mr. W. Mars, who in fact triggered me to explore the concept of responsibility centers. Also, he enabled and accompanied me during a visit in Belgium. This visit abroad has been a valuable experience to me and this international aspect truly appeared to be of added value to this thesis. Moreover, my thanks go to CEOs and managers of both Stork Netherlands and Stork Belgium, who provided me with sufficient information in order to complete my field research. Finally, I would like to thank my family and friends for their support, and maybe even more important, for the essential and motivating moments of happiness during the past couple of months.
TABLE OF CONTENTS 1. INTRODUCTION ...5 2. RESEARCH DESIGN ...7 2.1 Research objectives ...7 2.2 Research questions ...7 2.3 Methodology ... 10 3. LITERATURE REVIEW ... 13
3.1 Responsibility center types ... 13
3.1.1 Revenue Centers ... 13
3.1.2 Cost Centers ... 13
3.1.3 Profit centers... 14
3.1.4 Investment centers ... 14
3.1.5 Hybrid RC types ... 15
3.2 Responsibility center consequences ... 15
3.2.1 Consequences resulting from decentralization ... 15
3.2.2 Responsibility accounting principle consequences ... 17
3.2.3 Consequences for organizational unit performance ... 20
3.2.4 Consequences for, or resulting from the form of control ... 20
3.3 Responsibility center usage determinants ... 22
3.3.1 Decentralization ... 22
3.3.2 Interdependencies ... 24
3.3.3 Corporate strategy ... 25
3.3.4 Cultural differences ... 26
3.3.5 External environment ... 30
3.4 Responsibility center type determinants ... 32
3.4.1. Size... 33
3.4.2 Business unit strategy... 34
3.4.3 External environment ... 36
3.4.4 Technology ... 37
3.4.5 Contingency factors of practical nature ... 38
4. RESPONSIBILITY CENTERS: AN EXPLORATORY FIELD STUDY ... 46
4.1 Stork Netherlands vs. Stork Belgium ... 47
4.1.1 Stork Netherlands and its principal ... 48
4.1.2 Stork Belgium and its principal ... 49
4.1.3 Comparison ... 51
4.2 Service department SNL vs. Service department SBE... 52
4.2.1 Service department SNL and its principal ... 52
4.2.2 Service department SBE and its principal ... 54
4.2.3 Comparison ... 55
4.2.4 Consequences of RC usage ... 56
4.3 Conclusion of field research ... 57
5. CONCLUSION AND DISCUSSION ... 60
5.1 Conclusion ... 60
5.2 Discussion ... 61
LIST OF REFERENCES ... 62
APPENDIX I: LIST OF PERSONS INTERVIEWED ... 66
1. INTRODUCTION
“Would it be appropriate to turn our After Sales service department from a cost center into a profit center?” This rather straightforward enquiry of the controller of my current employer served as the trigger for this study. Although this question seems to be uncomplicated at first sight, searching the academic literature on the subject of responsibility centers left me stuck with numerous unanswered questions, such as “which factors justify the choice for a certain type of responsibility center over another?” and “what are the possible consequences of the choice for a particular type of responsibility center?”. In the academic world, surprisingly little research has been conducted in the field of responsibility centers. A responsibility center has been defined by Merchant (1998, p. 302) as “the apportioning of responsibility (or accountability) for a particular set of outputs and/or inputs to an individual or a group of individuals”. This definition is very close to decentralization, since the latter has been defined as “a type of organizational structure which refers to where decisions are taken within the organization, i.e., the level of autonomy that is delegated to managers for their decision-making” (Chia, 1995, p. 813). Nonetheless, although responsibility centers and decentralization are closely related, they definitely do not constitute the same notion. The difference lays in the notion of responsibility. The performance analyses of decentralized organizational units are not necessarily based on financial results of this department, and its managers are not personally held responsible for these results. Contradictory, the performance analyses of responsibility centers definitely are based on financial results of this department, and its managers are (to a certain extent) personally held responsible for these results. An elaboration on this important difference will be found later on in this study.
Anthony and Govindarajan (2007, pp. 131-132), argue that an organization is a collection of responsibility centers. In the organizational control literature, four types of responsibility centers are distinguished: 1) revenue centers, where outputs are measured in monetary terms; 2) cost centers, where inputs are measured in monetary terms; 3) profit centers, where both inputs and outputs are measured and 4) investment centers, in which the relationship between profit and investment is measured.
view their analysis as a step towards a richer theory of decentralization and responsibility centers. They also state that a next natural step would be to develop a model that addresses how organizational subunits should be grouped into responsibility centers. For instance, what are the trade-offs in keeping separate revenue and cost centers as opposed to a single profit center? (Melumad et al., 1992, p. 32).
However, although this study already dates from 1992, it appears that, after having searched for studies that have elaborated on the research of Memulad et al., surprisingly little research has been conducted on this subject subsequently. The study of Schoute and De With (2006) seems to be the only study entirely dedicated to responsibility centers that has been performed since then. However, this study only concerns some statistical analysis between a couple of contextual factors and the usage of either a profit or an investment center, without providing a fundamental, qualitative explanation for the divergence in the use of responsibility centers. Furthermore, the authors only considered characteristics of the organization as a whole, without paying attention to characteristics of individual responsibility centers. Nonetheless, the authors confirm that the number of empirical studies to date has been very limited. Given the importance of responsibility centers as a part of management control systems, this observation can be seen as a surprising establishment (Schoute and De With, 2006, p. 256). Therefore, the phenomenon of responsibility centers (to be abbreviated with RCs hereafter) would, at the least, be an original topic for further research.
2. RESEARCH DESIGN
2.1 Research objectives
The objective of this research is twofold. First of all, and partially based on Melumad‟s suggestions for future research, I have the ambition to develop a conceptual framework that explains the diversity in the usage of RCs in general. Secondly, I attempt to explain the variety in different RC types. That is, for instance, which factors cause the preference for a profit center over the choice for a cost center, or the preference for an investment center over the choice for a profit center? Obviously, there are some straightforward assumptions, such as the prevalence of a cost center for administrative functions (Merchant and Van der Stede, 2007, p. 273). However, beyond a couple of these logical examples, a large portion of RC variety seems to remain unexplained. In order to ease the search for the explaining determinants, this study also contains an identification of the consequences of RC usage (both in general and the use of specific RC types) both for the management control systems of a particular RC and the effects on the behavior of an RC manager.
By analyzing RCs – which can be conceived as a rather practical issue – from an academic point of view, the overall objective of this paper is to enhance our fundamental understanding of RC (type) usage and enable the reader to place this subject in a proper context with other management control related issues (for instance, strategy and extent of decentralization). Therefore, the core question of this research basically comes down to the following:
Which contingency factors determine the usage of RCs in general and, subsequently, the choice for a particular RC type?
2.2 Research questions
In order to answer the above stated question as accurately as possible and eventually provide a comprehensive explanatory framework, this core question will be decomposed into several research questions. First of all, it is important to determine which types of RCs are distinguished in the literature. This will enable the reader to understand the subject and to place it in the right context, and it defines the issues for the rest of the paper. Therefore, my first research question:
RQ1: Which types of responsibility centers can generally be distinguished?
resulted from isomorphism, there will always be a reason for implementing RCs. Therefore, the decision to mark an organizational unit as a specific type of RC should be based on the consequences and intentions of this particular choice. Hence, in order to be able to identify determinants, the (theoretical) consequences of the choice for a particular center should be known in front. Although it is intuitively appealing first to identify the determinants and next the consequences, this study will pursue this order the other way around. Being familiar with both implications for the control systems and behavioral implications will ease the search for the determinants of RC types. By doing so, the RC consequences will serve as an instrument in order to identify the RC (type) determinants.
According to Anthony and Govindarajan (2007, p. 133), each type of RC requires a different planning and control system. However, the authors remain rather general about what these implications are. This leads us to my next research question:
RQ2a: What are the consequences of responsibility center usage – both in general terms and in terms
of the choice for a particular responsibility center – for an organization‟s planning and control systems?
However, answers to this question will inevitably lead to some straightforward findings. For instance, a Return on Investment measure will be highly feasible in an investment center (Merchant and Van der Stede, 2007, p.271). This result cannot really be classified as a major break-through in science. In order to develop a broader understanding of the consequences of RC usage, it is useful not only to consider the technical, but also the (dysfunctional) behavioral consequences, since not the firm itself, but its employees act (Norreklit, 2000, p. 80). Therefore, the extension of research question 2:
RQ2b: What are the influences of responsibility center usage – both in general terms and in terms of
the choice for a particular responsibility center – on the behavior of responsibility center managers?
RQ3: What is the relationship between decentralization of an organization and the use of
responsibility centers?
Regardless of whether decentralization appears to be a determinant of RC usage in general or not, it is useful to be familiar with the contingency factors (other than, probably, decentralization) that determine the usage of RCs in general. First of all, this sets the boundaries and departing points for the search of factors that influence the choice for a particular type of RC. Secondly, it structures the framework to be developed and, as a final point, it provides a more comprehensive understanding of the phenomenon RCs. Therefore, my fourth research question:
RQ4: Which factors increase the likelihood of responsibility centers usage in general?
Answers on these questions enable us to have a better understanding of the phenomenon of the usage of RCs in general. However, the question of which factors in particular determine the use of a specific type of RC remains unanswered. Literature on contingency theories is expected to be highly valuable in deriving possible determinants of the use of specific RC types. According to Chenhall (2003, p. 127), researchers have attempted to explain the effectiveness of management control systems by examining designs that best suit the nature of environment, size, technology, structure, strategy and local cultures. Chenhall provides a critical review of findings of contingency-based studies over the past 20 years. This study will serve as the basis for the process of identifying factors that possibly determine the choice for a particular RC. Therefore, my fifth and final research question:
RQ5: Which factors possibly determine the choice for a particular type of responsibility center?
2.3 Methodology
The proposed study will consist of two subsequently related parts: a theoretical section and an empirical part. The first part, which predominates this study, comprises an investigation of previous literature. By relating findings from earlier studies, this section seeks to provide answers on each of the five research questions developed above. By integrating these answers, the ultimate objective of this part is to develop a theoretical contingency framework that describes the possible determinants of both RC usage in general and the choice for a particular type of RC.
Answers on the first research question will be provided by drawing on management control textbooks, since this is a rather straightforward and introductory question. Management control and organizational behavior literature will represent the foundation of answers to the second research question, while the third research question will primarily be derived from management control and organizational design literature. Numerous studies on the topic of decentralization appear to have been conducted, so no problems with regard to information ability are expected. This is in contrast to the final couple of research questions. Since approximately zero studies on RC determinants have been performed so far, there is a high need for formulating an explicit starting point. This starting point is represented by the study of Chenhall (2003), which identifies contingency factors that explain management control system variety in more general terms. By doing so, I expect to create a so-called „snowball effect‟ that ultimately leads to a set of contingency factors that probably influence the choice for a particular RC. By exploring additional literature that deals with these contingency factors specifically, I intend to judge whether a certain contingency factor might actually influence the choice for a certain center. Nonetheless, it will be unavoidable to draw some inferences in this part. With regard to the final research question, the determinants of RC type will be marked as the independent variables, whereas the actual type of RC (cost, revenue, profit or investment center) serves as the dependent variable.
The ultimate objective of the literature review section is the development of propositions (by the ends of a theoretical framework) regarding a set of determinants of RC types. The empirical part will build from these propositions.
study that – in combination with the extensive literature study – this paper would become too voluminous for a master thesis. By doing so, the objective of this limited field research has become to serve as a stepping stone for future, larger-scale research in the field of RCs.
Eisenhardt (1989a, p. 534) argues that case studies can be used to accomplish various aims: to provide description, to test theory or to generate theory. Given the time constraint and the fact that relatively little is known about this subject, this qualitative study uses a relatively limited sample. If the case study purposes defined by Eisenhardt would be reflected on a continuum ranging from providing description to generating theory, the purpose of this case study is located in between „providing description‟ and „testing theory‟. By doing so, the purpose of the empirical part is to gain more in-depth insight into the phenomenon of RCs and to make a first attempt to relate the suitability of the developed propositions to the day-to-day practice. However, the aim of this empirical section is explicitly not to make generalizable assertions about the population, since the sample size is far too small for this purpose.
2.3.1 Sampling design
This empirical study will use a non-probability sample. That is, the RCs to be investigated have not been selected randomly. Instead, the centers have been selected deliberately in order to obtain a sample with a certain variety in RC types (or dependent variables). This is the reason that Stork Belgium has been incorporated in the field research. Considering the Belgian entity in the sample is interesting in particular, since this company has a service department that has been marked as a profit center. Stork Netherlands also recognizes a service department, but this organizational unit has been designed as a cost center.
2.3.2 Data collection
Given the relatively little number of studies dedicated to RCs, the aim of this study is to explore a phenomenon about which little is know. By doing so, I wish to gain in-depth insights into the phenomenon of RCs. With regard to the field research part, it focuses on collecting data on a limited number of subjects. These are all characteristics of an exploratory study, and this study can therefore be marked as such.
3. LITERATURE REVIEW
Chapter 3 is dedicated to an extensive literature study. First, section 3.1 will elaborate on the different RC types to be used, followed by section 3.2 that identifies the possible consequences of RC usage. Having determined these consequences, the chapter continues with section 3.3, which identifies contingency factors that influence the decision to implement RCs in general and, hereafter, factors that influence the choice for a particular RC type in section 3.4. Altogether, this will result in a conceptual framework of RC (type) usage determinants in section 3.5, which constitutes the conclusion of this chapter.
3.1 Responsibility center types
The management control literature generally distinguishes four basic types of RCs. According to Anthony and Govindarajan (2007, p. 131), RCs are classified according to the nature of the monetary inputs and/or outputs that are measured for control purposes:
3.1.1 Revenue Centers
Revenue centers are characterized by outputs that are measured in monetary terms, but no formal attempt is made to relate input to output. A typical example of this type includes marketing and/or sales units that do not have authority to set selling prices. Neither are they charged for the costs of products sold (Anthony and Govindarajan, 2007, p. 133).
3.1.2 Cost Centers
According to Zimmerman (1997, p. 182) cost centers are established whenever a subunit is assigned the decision rights to produce some output and the unit‟s efficiency in achieving this objective is to be measured and rewarded. Its managers are evaluated based on their efficiency in applying inputs to produce outputs. They are not judged on revenues or profits, since managers are not held responsible for selling the final services or products. Two subtypes of this RC type can be distinguished: engineered and discretionary cost centers:
be made for every cost item (Anthony and Govindarajan, 2007, p. 134). Furthermore, Zimmerman (1997, p. 183) argues that managers of this RC type are primarily concerned with the task of minimizing costs: to choose the most efficient mix of labor and materials in order to produce a given level of outputs.
Contradictory to that of engineered cost centers, the outputs of discretionary cost centers cannot be measured in monetary terms. The term „discretionary‟ reflects management‟s discretion in decisions regarding certain business activities, for instance the appropriate amounts to spend on R&D practices, marketing efforts or public relations. Whereas the difference between budget and actual expense is a measure of efficiency within engineered cost centers, deviations from budget within discretionary cost centers simply imply differences between budgeted and actual input, without incorporating any value of output (Anthony and Govindarajan, 2007, pp. 134-136). In addition, Merchant and Van der Stede (2007, pp. 274-275) state that control in this RC type is usually exercised by ensuring that the discretionary cost center adheres to a budgeted level of expenditures while successfully accomplishing the tasks assigned to it.
3.1.3 Profit centers
Managers of profit centers are held accountable for the organizational unit‟s profit, relying on the internal accounting systems to measure performance. These are often composed of several cost and revenue centers that are being integrated for management control purposes. Profit centers are set up when the knowledge required to make the product mix, quantity, pricing and quality decisions is specific to the division and costly to transfer (Zimmerman, 1997, pp. 184-185).
Business units are often created as profit centers. However, these terms are not synonymous. First of all, a business unit manager‟s authority may be constrained in various ways (Anthony and Govindarajan, 2007, p. 189), implying that business units are not by definition profit centers. Furthermore, these authors state that functionally organized organizations can also be comprised of profit centers (p. 191). Hence, just as business units do not necessarily have to be profit centers, profit centers neither are business units by definition.
3.1.4 Investment centers
3.1.5 Hybrid RC types
Merchant and Van der Stede (2007, p. 274) argue that there can be considerable variation within each RC type, such that RCs can be arrayed on an almost seamless continuum from cost or revenue center on the one extent to investment centers on the other extent. For instance, a profit center can be analyzed on contribution margin instead of net profits. This continuum will be further elaborated in section 3.4. However, in order to develop an unambiguous framework, combinations of pure RC types will not explicitly be dealt with in this study. Having arrived at the ultimate theoretical framework, the reader can fairly well depict hybrid RC types on the RC type continuum. Tracing the determinants and consequences of a particular hybrid RC type then becomes a rather straightforward task.
3.2 Responsibility center consequences
As has been emphasized in the previous chapter, using RCs should not be an objective by itself, but rather should be means in order to achieve higher overall performance. Therefore, the decision to mark an organizational unit as a specific type of RC should be based on the consequences and intentions of this particular choice. Hence, in order to be able to identify determinants, the (theoretical) consequences of the choice for a particular center should be known in front. Being familiar with both implications for the control systems and behavioral implications will ease the search for the determinants of RC usage in general and RC types in particular. Therefore, this section attempts to identify consequences of RC (type) usage, both for the control systems and managerial behavior. During the literature reviews on this topic, it appeared that the RC usage has consequences that result from several concepts or has consequences for organizational characteristics: 1) consequences that result from the concept of decentralization; 2) consequences that result from the concept of responsibility accounting; 3) consequences for organizational performance and 4) consequences from or for the concept of different control forms.
3.2.1 Consequences resulting from decentralization
Merchant and Van der Stede (2007, p. 275) argue that firms‟ financial RC structures are, to a large extent, coincident with the managers‟ areas of authority. Although a more precise relationship between decentralization and RC usage will be defined in the next section, it is obvious that decentralization is inevitably related to the use of RCs and will therefore have considerable consequences for an organization. Leaving the question whether decentralization is either a determinant or consequence of RC usage to be unanswered until the next section, the consequences of decentralization are assumed to be (indirect) consequences of RC usage.
consequences of decentralization that are directly related to the use of RCs and might either influence RC usage in general or the choice of a particular RC type.
Most studies to decentralization choose the agency theory as a departing point, originally founded by Jensen and Meckling in1976. These authors define an agency relationship as a contract under which one or more persons (the principal) appoint another person (the agent) to perform some service on their behalf, which involves delegating some decision making authority to the agent (Jensen and Meckling, 1976, p. 308). In the light of RCs, the RC manager can be considered as an agent who has been delegated certain tasks (minimizing manufacturing costs, maximizing ROI) by the principal. This principal is represented by the manager of a higher-level organizational unit or by the owners of the firm in case that an entire organization is considered as an RC.
Jensen and Meckling state that, in case that both the agent and principal maximize their utility, the agent presumably will not always act in the best interest of the principal. Generally spoken, it is impossible for the principal to ensure that the agent will make optimal decisions from the principal‟s viewpoint at zero cost. Hence, the use of RCs is inherently related to the incurrence of so-called agency costs. These agency costs consist of 1) the expenditures incurred by the principal in order to monitor the agent; 2) bonding costs, or the costs involved in a principal‟s attempt of expending resources to guarantee that the agent would limit his activities, such as contractual guarantees to have the financial accounts audited by a public account and 3) a residual loss, which constitutes the monetary equivalent of welfare reduction due to the divergence between the agent‟s actual decisions and those decisions that would maximize the welfare of the principal. Hence, decentralization comes at a cost. This is supported by Mookherjee (2006, p. 387), who argues that delegation can cause significant control losses due to managerial risk aversion or due to the principal‟s limited capacity to monitor local conditions and the agent‟s decisions.
allows a firm to respond more quickly to (changed) local conditions and is therefore more appropriate in volatile environments.
Hence, the use of RCs (albeit via decentralization) can yield several advantages but also incur additional costs. It cannot be determined beforehand, but rather it will be context-specific whether the use of RCs leads to a net benefit or net loss. Although Dalton et al. (1980, p. 59) argue that the limited evidence to date tends to support a negative relationship between centralization and performance for managers and professionals on average, it would be far too straightforward to assume that decentralization leads to superior performance in every situation. Or, as Eisenhardt (1989b, p. 61) states: the heart of the principal-agent theory is the trade-off between the cost of measuring behavior on the one hand and the cost of measuring outcomes and transferring risk to the agent on the other hand. Put in context with this research, it will be seen that the contingency factors to be determined in section 3.3 and 3.4 will determine whether or not the benefits of a large extent of decentralization are outweighed by its additional costs and, consequently, whether or not RCs will be implemented.
3.2.2 Responsibility accounting principle consequences
Responsibility accounting is a reporting system that classifies accounting (and often other) information about an organization‟s activities according to the managers who are responsible for them (Merchant and Van der Stede, 2007, p. 271). Responsibility accounting is traditionally based on the assumption that RC managers are individually accountable for an organizational subunit such as a department or division. A fundamental objective of responsibility accounting is the separate measurement of each RC manager‟s performance as a means of motivating self-interested RC managers to show goal-congruent behavior (Rowe, 2008, pp. 165-168). Therefore, responsibility accounting and RC usage are closely related, and I define RCs here as “the crystallization and formalization of the responsibility accounting principles”. By pursuing this definition, responsibility accounting consequences can be conceived as consequences of RC usage.
to align the objectives of RC managers with those of the company‟s owners, or, in other words, in order to economize on the agency costs that have been mentioned in the previous subsection. Accounting measures are assumed to be relatively congruent with the organization‟s objectives of profit maximization. Taken this into account, implementation of the responsibility accounting concept is supposed to result in better goal alignment between ownership and control, which eventually results in higher overall performance.
However, the assumption that financial measures reflect economic activities appears not always to be valid. Merchant and Van der Stede (2007, p. 441) argue that these accounting measures are far from perfect indicators of firm value and value changes. Research has shown that the correlation between annual accounting profits and stock price changes are positive, but usually fall within in the range of 0.20 and 0.30. This indicates that accounting-based financial measures are only imperfect proxies of economic activities. If organizations rely too heavily on RC accounting-based financial measures in evaluating RC performance and rewarding its managers, it faces the risk that RC managers focus too heavily on these „noisy measures‟. Since these measures are only partially correlated with actual economic activity, responsibility accounting and RC usage might lead to suboptimal organizational performance due to managers focusing on the „wrong‟ performance indicators.
Motivating managers to maximize accounting measures can also create behavioral displacement problems, with managers who become to act myopically as the most potential damage (Merchant and Van der Stede, 2007, p. 443). Managers are said to be myopic when they become excessively concerned with short-term profits or returns, which comes at the expense of long-term value creation. This problem arises when managers focus on financial measures in short periods.
One form of this is known as operating myopia. This problem can occur in every RC type where revenues are part of the evaluation measures (to be precise: in revenue, profit and investment centers). One example of this is known as channel stuffing, which involves boosting sales in the near term by offering lower prices to customers. This may lead to a disproportionate decrease of sales in the longer run. Another form of operating myopia is boosting profits at the expense of the firm‟s stakeholders‟ goodwill. If RC managers force employees to work excessive overtime at the end of a period in order to finish production and report sales within the current period, the quality of these products may come into question. This usually leads to a disproportionate decrease of customer satisfaction and, consequently, to a loss of sales in future periods. Shipping lower-quality products may also lead to excessive cost of field repairs or product returns in a subsequent period.
investment policy in favor of projects that realize cash flows early and away from those projects that realize their cash flows later or, stated slightly different, selecting projects which pay off quickly over those projects that maximize firm value. In which ways can investment myopia hurt the firm‟s overall performance? Performance of investment centers is often (in 80 to 93% of investment centers according to surveys) analyzed in terms of Return on Investment (a measure of profit divided by an amount of capital invested), which creates several problems (Merchant and Van der Stede, 2007, pp. 443-450). The fist problem is one of suboptimization. ROI measures can encourage managers to make investments that boost the ROI of the RC, but are not in the best interest of the company as a whole. Assume an RC ROI of 15%, whereas the overall company ROI is 20%. An investment opportunity with a project ROI of 17% is likely to be undertaken by the investment center manager, since it increases his ROI. However, this is clearly not in the best interest of the firm, since it has a decreasing effect on overall ROI. Obviously, this line of reasoning can also be pursued the other way around: investment center managers who reject certain projects since it decreases RC ROI, although this project would have increased overall ROI (in case that RC ROI is higher than overall firm ROI). A second problem stems from difficulties in the denominator of the ROI measure. The asset values that are taken from the balance sheet are not always fair representatives of real economic values. ROI is usually overstated if net book values are used, since older assets are far below their replacement values due to purchase at lower prices and they have longer been depreciated. Furthermore, ROI can be artificially increased by accelerated depreciation. This causes the denominator to decrease and, consequently, the ROI to rise. Since a survey of Reece and Cool (1978, p. 33) shows that 85% of firms use net book values in ROI computations, this problem caused by using net book values is quite severe in most organizations.
When do managers behave myopically with regard to their investment decisions? According to Narayanan (1985, p. 1473), a manager will invest myopically in an attempt to increase the market‟s perception of his value if investment opportunities are not readily observable and the risk averse manager is not tied to the firm for more than a single year. He does so in order to maximize his wage.
feasible, organizations still rely heavily on these financial measures in their application of the responsibility accounting principle (Merchant and Van der Stede, 2007, p. 443).
To summarize, the responsibility accounting principles that stem from RC usage can lead to increased goal congruence. If managers are held responsible for certain measures, they are likely to focus on these measures that are in the best interest of the company as a whole, eventually resulting in higher performance. However, since firms heavily rely on accounting-based financial measures in performance evaluation, several dysfunctional problems may occur. First, these measures are not always as highly correlated with real economic values as assumed. Second, it can lead RC managers to behave myopically in operating terms (in case of revenue, profit and investment centers) and, in addition, it encourages these managers to engage in investment myopia if their RC has been marked as an investment center. As also has been hypothesized with the decentralization consequences, the contingency factors to be determined in the following couple of sections are likely to determine whether or not the advantages of responsibility accounting are outweighed by its disadvantages and, consequently, whether or not RCs will be implemented. This will be shown in section 3.3 and 3.4.
3.2.3 Consequences for organizational unit performance
Chu et al. (2002) conducted research to RCs in 160 Taiwanese hospitals during the period from 1994 to 1996. These researchers related RC usage to hospital (in)efficiency, where inefficiencies are defined as differences between actual output and the maximally feasible output. On its turn, hospital outputs are defined in terms of number of surgeries, number of emergency visits, total intensive care client days etcetera. Chu et al. (2002, p. 1235) found that hospitals that implemented the RC system perform better in terms of efficiency than hospitals that did not implement RCs. However, this finding only holds when the RC system was integrated with formal incentive schemes such as rewards that are, at least partially, dependent on departmental profits. This also confirms the in management control literature commonly cited statement “no results without rewards”. It should be remarked that these findings are applicable to hospitals and cannot directly be generalized to other businesses. However, since business culture in the for-profit sector can be considered as at the least as competitive as that of in the non-profit sector, it would be more reasonable to assume that the findings of Chu et al. also do hold in the for-profit sector than to assume that these do not hold. Although it would be far too straightforward to assume that RC usage leads to higher performance always and in every case, the findings of Chu et al. indicate that, ceteris paribus, the use of RCs can definitely result in higher levels of efficiency if this is combined with formal incentive schemes where RC managers are subject to.
3.2.4 Consequences for, or resulting from the form of control
based on the periodical results of entities, corporate management can review and judge the effectiveness of the various organizational entities while leaving the actual execution of operations to people responsible for the performance of those entities. Action controls involve ensuring that employees perform certain actions that are most likely to be beneficial, and ensuring them explicitly not to perform actions that are harmful to the organization. Personnel controls are designed in order to increase the probability that employees will perform the desired tasks satisfactorily on their own (for instance, providing effective training), while cultural controls are used to shape organizational behavioral norms and to encourage employees to monitor and influence each other‟s behaviors, such as a corporate code of conduct. These authors also argue that result controls are consistent with, and even necessary for, the implementation of decentralized forms of organization with largely autonomous entities or RCs. However, there might be some causality problems here. On the one hand, it can be assumed that organizations (have to) pursue a form of results controls once their desire is to implement the concept of RCs. On the other hand, it is intuitively appealing to assume RCs to be a mechanism in order to exercise results controls. It goes beyond the scope of this study to establish this cause-and-effect relationship. Nonetheless, although this cause-and-effect relationship is not always clear-cut, from the literature it appears that the concept of RCs and the concept of results controls move hand in hand. Hence, the usage of RCs is inevitable related to pursuing a form of results control.
To summarize this overall section and to provide the final answer to research questions 2a and 2b, to the possible consequences of RC usage belong the following: derived from the concept of decentralization, RC usage utilizes benefits of local information, enables organizations to respond more quickly to (changing) local conditions and enables analyzing organizational units on an aggregated base. However, agency costs are being incurred by using RCs. Derived from the concept of responsibility accounting, RC usage results in increased goal alignment between agent and principal. Nonetheless, it might work out adversely, resulting in operating myopia and, specifically in case of investment centers, in investment myopia. Moreover, although a uniform cause-and-effect relationship cannot be reliably estimated, RC usage goes hand in hand with a form of results controls. Finally, a study in hospitals proved that RC usage definitively can lead to higher performance if combined with formal incentive schemes. Nonetheless, altogether, it appears that the results from section 3.2 remain partially inconclusive. Although some preliminary consequences have been inferred, no concrete and firm conclusions can be derived from the literature research. Therefore, no concrete propositions come forth from this section.
manager be held responsible, also considering the consequences of this RC (type) usage? Therefore, although no firm conclusions came forth, the inferred consequences from this section will serve as a device in order to identify the explaining variables, as will be seen in section 3.3 and 3.4. For instance, the possible RC consequence „myopia‟ will appear to most likely lead to the decision not to use RCs if an organization‟s departments need to cooperate intensively. Section 3.3 identifies determinants of RC usage in general, whereas section 3.4 investigates the determinants of the use of specific RC types.
3.3 Responsibility center usage determinants
Before deciding which type of RC to be used, an organization should already have made the decision to use RCs in general. Which factors might influence this choice? Given the limited research on this subject to date, no specific starting points are readily available that enable to draw on studies from other researchers. Therefore, the more universal contingency theory will serve as the main device in order to identify the determinants of RC usage. According to Fisher (1998, p. 47), contingency theory states that the design and use of control systems is dependent upon the context of the organizational setting. Hence, from an extensive study on the general management control literature, a set of factors has been selected that is frequently being used in management control studies and that is likely to influence the usage of RCs. The factors to be considered in this subsection are respectively decentralization, interdependencies, corporate strategy, cultural differences and environmental uncertainty.
3.3.1 Decentralization
As has been mentioned before, firms‟ financial RC structures are, to a large extent, coincident with the managers‟ areas of authority. Furthermore, the authors state that decisions about an organization‟s structure do not necessarily precede decisions about the type of RCs that should be used; the responsibility structure decision may come first. However, it has not been argued what the exact relationship is between decentralization and delegation of authority on the one hand and the use of RCs on the other hand. Hence, in order to develop a more comprehensive framework, it has to be determined whether the decision to decentralize is more likely to be a determinant or a consequence of the use of RCs.
Furthermore, the direction of causality can be estimated by logical reasoning: once an organization becomes decentralized, the conditions for RC usage have been created but these RCs do not necessarily have to be designed. Several factors might corporate management make to decide not to hold managers personally responsible for results (for instance, a high extent of environmental uncertainty, as will be seen later on in this section), but decentralization is still essential since corporate management cannot perfectly coordinate the local business due to this same environmental uncertainty. For instance, an organization might decide to establish an autonomous subsidiary in a part of the world in which the way of doing local business is unknown to the organization. The subsidiary then serves as test case, and the manager receives full freedom in order to explore the new geographic region. If successful, the parent company decides to elaborate the business. If not, it eliminates the test case subsidiary, and by doing so, keeping its „exploring costs‟ (which are sunk costs) relatively low. Since the parent company cannot reliably estimate whether the test case subsidiary will be viable, the manager of the latter cannot be personally held responsible for the generated results. Hence, although the decision-making processes have been decentralized to a large extent, it is not being marked as an RC.
However, this line of reasoning does not hold the other way around. A characteristic of RCs is that certain decisions are delegated to the center‟s management with instructions to optimize some financial performance variable (Memulad et al., 1992, p. 446) or, in other words, the delegation of authority, which has been defined to be equivalent to decentralization by Merchant and Van der Stede (2007, p. 27)1. Hence, whereas organizations can decentralize without using RCs, it is impossible to use RCs without having decentralized the organization to a certain extent, indicating that decentralization partially determines (or is a condition for) RC usage.
This leads to preliminary indications that the decision to decentralize an organization is a determinant rather than a consequence of RC usage, thereby providing the solution to the third research question. This finding is in line with my previous statement that RC usage should be methods in order to achieve higher organizational performance instead of an objective by itself. Therefore, the first proposition:
Proposition 1: As the extent of decentralization increases, the likelihood of RC usage increases as well.
1
3.3.2 Interdependencies
The extent of need for cooperation between organizational units might influence the decision whether or not to transform these units into RCs. Given the mentioned dysfunctional behavior that stems from the responsibility accounting principle, RC usage can lead to suboptimal performance. In case of a high need for cooperation combined with the use of RCs, operating and/or investment myopia can lead RC managers to make decisions that optimize the RC accounting measures, but are suboptimal to the firm as a whole. This statement is supported by Zimmerman (1991, p. 185), who states that motivating individual profit centers to maximize profits will not generally maximize profits for the firm as a whole when there are interdependencies among business units. Furthermore, Rowe (2008, p. 172) provides experimental evidence that when RC measurability is inseparable, competitive compared to cooperative RC boundaries leads to significantly lower group-level performance due to free riding and distrust. Thus, when RC measurability is inseparable, competitive managerial behavior can be dysfunctional.
Interdependencies can appear both vertically and horizontally within organizations. First, consider the case of vertical integration. Porter (1980, p. 300) defines vertical integration as the combination of technologically distinct production, distribution, selling, and/or other economic processes within the confines of a single firm. Suboptimal decision making is likely in the situation in which supplies to an external party are favored above supplies to a subsequent internal party in the value chain, because supply to an external party might yield a higher contribution margin to the RC in case. However, if the subsequent party is dependent on these supplies and is now unable to sell to its own customers, this can lead to forgone contribution margin in terms of overall firm performance (since the subsequent party now has no semimanufactured products to process and sell, which leads to forgone contribution margin at this point in the value chain). Nonetheless, this problem can be mitigated by introducing a transfer pricing policy. In his book about transfer pricing, Eccles (1986, p. 91) states that transfers should be mandated in case of a vertical integration strategy. If there are sales opportunities to external parties, transfers should be mandated and these are based on market prices. If these external sales opportunities do not exist, transfers should also be mandated, but the price is based on full costs. This still enables the organization to evaluate the performance of organizational units independently. Hence, by setting an effective transfer pricing policy in case of a vertical integration strategy, interdependencies among organizational units do not necessarily restrict the organization from using RCs.
a new product), the units can only be analyzed by joint performance. In this case, it comes down to the sharing of intangible knowledge. Economic theory recognizes that factors such as intangible knowledge can cause the cost of separate measures of individual RC performance to increase, such that they are no longer cost effective and thus are replaced by inseparable RC measures (Rowe et al., 2008, p. 168). This is strengthened by Chenhall (2003, p. 141) who states that the more technologies are characterized by high levels of interdependence, the more informal the controls. These informal controls include fewer statistical operation procedures, less emphasis on budgets, more frequent interactions between subordinates and superiors and a greater usefulness of aggregated and integrated management control systems. Finally, there will also be practical measurability constraints in case of horizontal interdependencies, since it involves transfers of knowledge. This transfer of knowledge is far more complicated to express in financial terms than the transfer of goods as is mostly the case in a vertical integration strategy. Hence, my second statement:
Proposition 2: As the horizontal interdependencies and need for horizontal cooperation between organizational units decrease, the likelihood of RC usage increases
3.3.3 Corporate strategy
According to Anthony and Govindarajan (2007, p. 59), corporatewide strategic analysis results in decisions involving businesses to add, businesses to retain, businesses to emphasize, businesses to deemphasize and businesses to divest. According to this analysis, companies can be classified into one (or along a continuum) of the three categories: single industry, where firms operate in one line of business; related diversified, where firms operate in several industries and the respective business units benefit from a common set of core competencies or unrelated diversified, where firms operate in unrelated businesses and the connection between business units is primarily financial.
the explanatory factors can reinforce each other to a large extent). In addition, Hill (1992, p. 519) finds evidence for the fact that firms pursuing related diversification in order to realize economies of scope also need to adopt cooperative organizational arrangements. On the contrary, an unrelated diversified firm, characterized by a high degree of decision-making authority and autonomous business units, is far more likely to be organized into RCs, since the need for cooperation and coordination has mostly been disappeared. Again, this is confirmed by Hill, who proves that those organizations pursuing unrelated diversification to realize governance economies need to adopt competitive arrangements.
However, although it is therefore intuitively appealing to assume that the likelihood of RC usage increases as corporate strategy moves from single business to unrelated diversified, a critical note has to be made. As an organization moves to an unrelated diversified strategy, corporate management can only implement this diversification strategy if a certain extent of decentralization has been carried trough, since corporate management is unable to centrally coordinate the entire span of varying activities. On its turn, the extent of decentralization has been identified as a determinant of RC usage. Therefore, corporate strategy is more likely to be a direct determinant of the extent of decentralization, making it an indirect determinant of the RC usage likelihood, resulting in the following statement:
Proposition 3: As the corporate strategy moves from single business to unrelated diversified along the „corporate strategy continuum‟, the likelihood of decentralization, and, indirectly, the likelihood of RC usage increases
3.3.4 Cultural differences
Differences in local cultures is one out of the six contingency factors distinguished by Chenhall (2003), which, as has been mentioned before, are about to serve as explaining variables for the variety in RC types. However, a deeper analysis of cultural differences yielded the insight that this contingency factor does not so much explain the usage of a particular RC type, but rather does help explain the usage of RCs in general.
Generally stated, having branches in different nations is likely to commensurate with a certain extent of decentralization. As has been argued before, decentralization utilizes the benefits of local information. Therefore, it would be unfeasible to centrally coordinate the entire spam of activities abroad, and a certain amount of authority has to be delegated to local managers. Hence, differences in local cultures (or simply having branches in multiple countries) can be seen as a determinant of the extent of decentralization. However, this delegation of authority does not necessarily have to correspond with holding managers of local branches personally responsible for the realized results. Or, in other words, having geographically dispersed organizational units does not automatically have to result in RC usage, as will be seen in this subsection. The possible relationships between RC usage and cultural differences will now be analyzed by pursuing one by one the four cultural dimensions of Hofstede.
Individualism vs. collectivism
interpersonal rivalry. Taken together, this provides preliminary evidence for the proposition that a higher degree of collectivism relieves the need for responsibility accounting and the forthcoming RC usage. Or, stated the other way around:
Proposition 4a: As the degree of individualism increases (the degree of collectivism decreases), the likelihood of RC usage increases.
Masculinity versus femininity
According to Hofstede (1980, p. 46), the dominant values in masculine societies are assertiveness, the acquisition of money and not caring for other people. Feminine cultures logically represent the opposite pole. Van der Stede (2003, p. 267) predicts that organizational members‟ desire for achievement and competition in masculine countries permits a stronger focus on performance, meeting budget targets and relative performance evaluations. In feminine countries, emphasizing bottom-line performance without much consideration for the well-being of members is likely to be less accepted or even to work out counterproductive. Therefore, holding managers personally responsible for results is likely to enhance performance in masculine countries, while yielding a dysfunctional effect on performance in feminine countries. The next proposition, therefore, goes as follows:
Proposition 4b: As the degree of masculinity increases (the degree of femininity decreases), the likelihood of RC usage increases.
Power distance
does not necessarily lead to RC usage, the likelihood of RC usage increases as the extent of decentralization increases (as has been argued in subsection 3.3.1). Hence, it is expected that there is both a direct and, via decentralization, an indirect relationship between the level of power distance and RC usage, leading to the following proposition:
Proposition 4c: As the level of power distance decreases, the likelihood of (decentralization, and thus) RC usage increases
Uncertainty avoidance
This final dimension indicates the extent to which people feel threatened by uncertain and ambiguous situations and try to avoid these situations by providing greater career stability, establishing more formal rules and do not tolerate unusual ideas and behavior. Employees (or, in the context of this study: RC managers) high in uncertainty avoidance prefer to avoid, reduce, or deny risk and ambiguity (Merchant and Van der Stede, 2007, p. 730). It is intuitively appealing to assume that higher uncertainty avoidance decreases the likelihood of RC usage, since the latter involves the RC managers to bear a certain amount of risk (that is, their evaluation and probably their compensation is partially based on uncertain factors such as an RC‟s annual profits). However, this relationship is not as clear-cut as one would expect at first sight.
Harrison (1993, p. 324) states that this cultural dimension has two theoretical attributes: tolerance for ambiguity and uncertainty on the one hand and attitude toward risk on the other hand. With regard to the latter attribute, it might indeed be expected that RC usage is less likely in cultures characterized by risk averseness. If managers are personally held responsible for a unit‟s results – as is the case in RC usage – their compensation might be partially based on the (financial) results of this RC. Compared to a fixed salary, this development causes the manager to face higher risks, since his annual compensation becomes more uncertain. Therefore, individuals in cultures characterized by high uncertainty avoidance may not react favorably to RC usage with the forthcoming performance-dependent compensation, since it causes them to bear more business risks, as also has been argued by Van der Stede (2003, p. 266).
level of certainty, since it becomes more transparent to managers on which aspects their evaluation and rewards are based, hereby removing the ambiguity in the evaluation process.
To summarize the domain of uncertainty avoidance: whereas the uncertainty avoidance‟s dimension attitude toward risk decreases the likelihood of RC usage, the uncertainty avoidance‟s dimension tolerance for ambiguity and uncertainty has a neutral effect on, or even increases the likelihood of RC usage. Hence, no consistent and conclusive prediction can be done with regard to the relationship between RC usage and the cultural dimension „uncertainty avoidance‟.
To summarize the overall subsection, it appears that the likelihood of RC usage increases as a local culture moves further to the extent of individualism, moves further to the extent of masculinity and is characterized by a low power distance. The three propositions developed in this subsection are expected to hold twofold. First, a multinational organization might incorporate this contingency factor in its decision to mark one particular organizational unit as an RC, while explicitly not treating another unit as RC. For instance, a division in the United States (a highly individualistic country) can be marked as a certain type of RC, while a Chinese (a highly collectivistic country) division is not marked as an RC (in this context: holding United States managers personally accountable for the organizational unit‟s results, while holding Chinese managers not). Second, this proposition is also expected to hold for organizations that operate in only one country. That is, the RC usage is expected to be higher within United States corporations than in Chinese corporations with respect to their treatment of domestic business units.
3.3.5 External environment
Just like the cultural differences, the external environment belongs to the contingency factors that are dealt with by Chenhall (2003) as well, which were initially to be introduced in section 3.4. However, from an analysis of the relevant literature it appears that this contingency factor not only helps explain the choice for a particular type of RC, but also might influence the decision whether or not to use RCs in general.
Uncertainty makes the use of results controls more difficult (Merchant and Van der Stede, 2007, pp. 724-726). There are several reasons why environmental uncertainty limits the effectiveness of results controls. First of all, if managers do not understand how to generate the desired results, it would not be reasonable to hold them responsible for these results. Furthermore, corporate management will face the problem of target setting: in unpredictable business environments, it is almost inevitable that performance targets are either too easy or hard to attain, and that these targets contain numerous uncontrollable factors, making it difficult to hold managers personally responsible for the results of an organizational department. Third, in case that this uncertainty is combined with the use of result controls, this causes employees to bear business risk. In order to limit these risks, organizations facing an uncertain environment might choose not to regard managers‟ budget targets as firm commitments to the organization and, thus, choose not to held managers personally responsible for the (financial) results. At first sight, this might be somewhat confusing with the statements made in the previous subsection, where it has been argued that RC usage can also diminish the extent of uncertainty, since it then becomes more transparent to managers on which aspects their evaluation will be based. However, latter refers to „internal‟ uncertainty (mental matters), while the uncertainty in this subsection relates to external uncertainty. Applying RCs in external uncertainty is not likely to assure internal certainty, because for managers it is still unclear how to generate the desired results.
Since RC usage is a form of results controls (as has been argued in subsection 3.2.4) and environmental uncertainty decreases the feasibility of the use of results controls, it is reasonable to assume that the likelihood of RC usage decreases as the degree of external uncertainty increases. Or, stated the other way around:
Proposition 5: As the extent of environmental uncertainty decreases, the likelihood of RC usage increases
Once the contingency factors have ensured RC usage in general, an organization faces the question which RC types to implement. Therefore, section 3.4 will consider the box „RC usage‟ as a starting point from where variables will be identified that explain the variety in RC types. In other words, the decision to implement RCs in general will be considered as a given in the next section.
3.4 Responsibility center type determinants
On this continuum, cost and revenue centers are depicted on approximately the same location, since most of their determinants will appear to be similar. However, as will be seen in this section, a couple of subtle differences exists in these determinants. This explains why revenue centers are depicted somewhat more to the right extent, but still show a significant overlap with cost centers.
Since very little research to RCs has been conducted to date, it is inevitable having to discretionally choose a starting point that might be conceived as somewhat arbitrary. This starting point is represented by the study of Chenhall (2003), which provides a critical review of findings from contingency-based studies over the past 20 years. From this, Chenhall derives a series of propositions relating management control systems to organizational context. The contingency factors which are dealt with, and which will consequently be considered in this section are 1) size; 2) business unit strategy; 3) the external environment and 4) technology. The other two factors identified by Chenhall will not be treated in this section, known as 5) differences in local culture, since in the previous section this factor appeared to be a determinant of RC usage in general, and 6) organizational structure. In a sense, the latter factor is the dependent variable to be investigated, since RC (type) usage partially determines the design of an organization. The four remaining factors will be analyzed one by one and the most likely relationships, if any, are being presented in what follows in this section.
3.4.1. Size
