Master thesis
Assessment of control systems and a consideration of
alternatives: an exploratory case study among
administrative employees and their superiors within a
lead logistics provider
P.J. Veeningen Mennestraat 21 3882 AN Putten e-mail: peterveeningen@gmail.com tel. nr.: 06 4255 2514 Student number: 1752332 Amersfoort / Putten, July 2009 University of Groningen
Faculty of Business and Economics MSc Business Administration
Specialization in Organizational and Management control HAVI Logistics B.V. (Amersfoort)
Supervisor Groningen: prof. dr. H.J. ter Bogt Co-assessor Groningen: prof. dr. G.J. van Helden
Supervisor HAVI: mr. M. Lenardt (manager finance & administration) Publication
Assessment of control systems and a consideration of
alternatives: an exploratory case study among
administrative employees and their superiors within a
lead logistics provider
“Question everything. Learn something. Answer nothing.”
(Greek playwright, c. 480-406 BC)
“Interdependence is a higher value than independence.”
(S.R. Covey, 2004, p.3)
University of Groningen comes to an end. In September 2007 I started at the intake year for students from a university of applied sciences at the faculty of economics and business, which focused on learning about scientific methods, as well as knowledge about Organizational & Management Control (which is also my specialization within MBA). This year was in between my previous study (‘BSc Business economics – controller’ at the Windesheim University of Applied Sciences in Zwolle) and MBA, which I started in September 2008.
During the previous six years I have gained, in my opinion, valuable theoretical and practical knowledge about organizations, management, (business) economics, finance, logistics, administrative organizations, bookkeeping, tax and law, and IT and professional software. During the last two years I learned on top of this how to look at, and address different kind of issues.
At the beginning of February 2009, I had a first meeting – arranged by both a friend and an employee of HAVI, Gerrit Aalten – with Marco Lenardt, who became my supervisor from HAVI until the thesis would be finished. All meetings with Marco and with other employees went very good, for which I thank them. I also thank them for the resources they made available, and for the trust they have put in me and the university from the little that they knew about it.
During that period, the supervisor from the University of Groningen, prof. dr. Henk ter Bogt, was also assigned to me, who I thank for giving necessary comments on my work and other support.
I finally thank the following persons who were helpful and/or necessary to get this thesis finished or for my motivation: my parents, my brothers, dr. Ben Crom, prof. dr. Jan van Helden, Hanneke van Dijk, Tom Hoorn, Nick Koele, Berry Ligtermoet, Albertine van der Werf, Herman Jan Wolthuis, Tsjibbe Wiersma, and all other friends, family, and acquaintances who had a directly positive influence.
regarding HAVI’s control systems effectiveness were determined. During the second stage these research topics were investigated more in-depth, comparing data from interviews and questionnaires from administrative employees and their superiors. This has led to a list of 31 considerations about HAVI’s control systems and a proposal for a strategic control system.
TABLE OF CONTENTS
INTRODUCTION ... 6
STAGE I: GENERAL CHARACTERISTICS OF HAVI’S CONTROL SYSTEMS FOR RESEARCH TOPICS ... 8
GENERAL LITERATURE ABOUT CONTROL SYSTEMS... 8
CHARACTERIZING HAVI’S CONTROL SYSTEMS... 10
IN-DEPTH RESEARCH TOPICS... 12
STAGE II: IN-DEPTH LITERATURE REVIEW... 14
METHODOLOGICAL LITERATURE: CATEGORIZATION OF SCIENTIFIC METHODS... 14
METHODOLOGICAL LITERATURE:DETAILED DISCUSSION OF SOME METHODOLOGICAL ISSUES... 15
METHODOLOGICAL LITERATURE:QUESTION DESIGN AND ITS LIMITATIONS... 16
ACADEMIC LITERATURE: INTRODUCTION... 18
ACADEMIC LITERATURE:CONTROL SYSTEMS... 19
ACADEMIC LITERATURE:FORMALIZATION OF STRATEGIC PLANNING AND GOAL CONGRUENCE... 21
ACADEMIC LITERATURE:REMAINING RESEARCH QUESTIONS... 23
CONCLUSION... 24
STAGE II: IN-DEPTH RESEARCH DESIGN AND METHOD... 26
DEFINITION OF THE CASE STUDY... 26
DATA SOURCES... 26
ASSUMPTIONS ABOUT THE DATA... 28
METHODS OF DATA ANALYSIS... 29
STAGE II: IN-DEPTH RESULTS ABOUT HAVI’S CONTROL SYSTEMS ... 30
FORMALIZATION OF THE (STRATEGIC) PLANNING AND CONTROL SYSTEMS, AND EMPLOYEES BEST ABLE FOR DETERMINING ACTIONS... 30
MOTIVATION BY CONTROL SYSTEMS (INCLUDING APPRAISAL) AND ACHIEVABILITY OF WORK... 39
ASSESSMENT OF THE INTERVIEWS AND QUESTIONNAIRES BY HAVI’S EMPLOYEES... 42
SUGGESTIONS FOR A STRATEGIC CONTROL SYSTEM... 43
CONCLUSION... 45
CONCLUSION AND LIMITATIONS ... 47
REFERENCES ... 51 APPENDIX I THESIS PLANNING HAVI
APPENDIX II RESEARCH PROTOCOL
APPENDIX III ORIENTATION INTERVIEW ABOUT HAVI’S CONTROL SYSTEM APPENDIX IV OTHER POINTS OF ATTENTION FOR SUB RESEARCH QUESTIONS 2 - 4 APPENDIX V INVESTIGATIVE QUESTIONS FOR STAGE II
INTRODUCTION
From an interview with the manager finance & administration of HAVI Logistics B.V. (HAVI) – a lead logistics provider for McDonalds and BP petrol stations – the question arose whether other control systems than the ones already used could be useful for the achievement of the organizational objectives. Different options and areas of interest were considered, like the application of a balanced scorecard, enhancement of personnel empowerment, and the improvement of accounting measures. However, it was stressed that mainly the human resource aspect is an important field for HAVI to take into account in this research. The interview concluded with the statement that improvements to HAVI’s control systems should be considered, when this leads to a better implementation of HAVI’s strategy. To achieve this goal, first it must be determined what HAVI’s current control systems are and how they are functioning. The objective of this research is formulated as follows:
Research objective: Determine how HAVI’s control systems are functioning and
whether other control systems might be necessary for the achievement of HAVI’s organizational objectives.
This study has a two-stage design with stage I as an orientation on HAVI’s control systems, and stage II as an in-depth study on research topics determined during stage I (see also appendix II).
In stage I general statements about HAVI’s control systems are made. According to Merchant and Van der Stede (2007), four general types of control systems, applied in certain tightness, can be distinguished within each organization: action, result, personnel, and culture controls. Either one or combinations of these systems are available within HAVI. These control systems have several elements (advantages, disadvantages, and criteria), which are compared with theoretical considerations about these elements. From this comparison, a central research question and four sub research topics are described at the end of stage I which are investigated during stage II.
Stage II describes the in-depth study on the research topics. It contains the same structure as stage I, however, more specific literature and information from HAVI is used to address the research topics. At the end of stage II possible necessary considerations for HAVI’s control systems are described and a proposal for a strategic control system is made. After stage II further stages with the same approach might also be necessary, however, this is beyond the scope of this research.
systems and about advising alternatives. However, existing theory already might provide the same insights, which is assessed first.
Important in this study are methodological considerations. This research objective has a normative (and explanatory) nature, in the sense that previous literature will be used to assess HAVI’s control systems and by describing how it could possibly work more properly. However, this research is also an exploratory case study, which is less suitable for these conclusions to be drawn. Exploratory studies can, however, provide valuable new insights, and due to the fact that the central research question (with relevant constructs) is not yet known, conducting an exploration first is more feasible than immediately drawing conclusions. Furthermore, according to Eisenhardt (1989), opportunities may rise during the exploration to conduct a cross-literature or cross-case study, which may provide even a better chance of gaining new insights (e.g. from constructs not yet associated with each other).
Hopper and Powell (1985) state that each research methodology might be biased from the fact that people involved in a study might not always make the most rational (objective) choices. Furthermore, contexts (business environments) change, which could make the conclusions from this study in the future less applicable. Moreover, individual choices or behavior within HAVI or from the researcher himself can possibly bias the results. Even a more radical approach to this study might be necessary, in a way the researcher is not used to. However, the ambition of this research remains to understand how it can be determined what factors are involved in assessing the effectiveness of (a part of) HAVI’s control systems and how to address alternatives. By applying both a normative approach, and by accounting for individual choices and behavior, possible limitations to this research are tried to be reduced. Figure 1 summarizes the introduction.
Figure 1: Research objectives and possible biases from research methodology Literature study into how control
systems (CS) can be assessed
Assess HAVI’s CS Objective Incomplete description Sufficient CS: no control alternatives Insufficient CS: examine control alternatives Stage II, …, i Possible biases from
STAGE I: GENERAL CHARACTERISTICS OF HAVI’S CONTROL SYSTEMS FOR RESEARCH TOPICS
In this section, first, different points of attention are described which are used in an interview with the manager finance & administration to characterize HAVI’s control systems. This section concludes with the research topics which will be investigated during stage II, based on the comparison of the theoretical points of attention and HAVI’s control systems’ characteristics.
General literature about control systems
Merchant and Van der Stede (2007) describe three processes which lead to the development of control systems:
1. Strategy formulation: Defining how organizations should use their resources to meet objectives (e.g. by examining the internal and external environment). A clear strategy yields a larger set of feasible control alternatives (focusing only on needed behavior, i.e. critical factors), however, not even the most elaborate strategic vision is complete enough to detail every desired action and contemplate every possible contingency.
2. Objective setting: It is necessary for employees to understand what the organization is trying to accomplish (financially and/or non-financially). Many employees do not always unanimously agree on how to balance responsibilities between stakeholders (e.g. among owners, employees, debtors or the society at large). However, most organizations develop effective compromise mechanisms to resolve conflicts among stakeholders and reach some level of agreement about the objectives they will pursue.
3. Developing (management) control systems: Used to make employees behave appropriately. This can be segregated into three control problems, where choices have to be made about in whether to rely on employees or managerial interventions through different systems:
3a. Do employees understand what is expected of them? (Do employees know what to do to achieve the organizational objectives);
3b. Will employees work consistently hard and try to do what is expected of them? (Are employees motivated to focus on the organizational objectives);
3c. Are employees capable of doing a good job?
Merchant and Van der Stede (2007) describe four types of control systems which address problems 3a up to 3b at a certain level. These four types are result control, action control, personnel control, and cultural control, which are defined with some keywords in table 1. These types of control systems can have, among other, the following characteristics:
4b. Encourage autonomy/creativity. This can be seen as the counterpart of 4a, because no prescriptions are used. Autonomy can lead to better motivated employees who can solve individual problems more creative, where standardization could not achieve this;
4c. Ex-post measurement of the results/actions. Subjective measures can be useful when objective measures have many flaws; however, evaluators often have limited knowledge about evaluatees’ performance. Furthermore, information about how work must be improved is often not provided. Subjective measures are also often mistrusted and can lead to an excuse culture. 4d. Set performance targets. Targets for desired results should be challengingbut achievable. If not, internalization is less likely, due to demotivated personnel;
4e. Provide rewards for good results/actions (e.g. monetary and/or psychical); 4f. Rely on employees instead of managerial restrictions and/or monitoring;
4g. Rely on mutual monitoring between employees instead of managerial restrictions and/or monitoring. This can be encouraged by providing group rewards, where employees do not want their bonus be influenced by underperforming colleagues.
Merchant and Van der Stede (2007) also describe some requirements for the four control systems: 5a. Knowledge of desired results/actions. Employees should know what to do (with result controls) or managers should be able to determine what employees should do (with action controls). This should be congruent with the organizational goals (see 2. objective setting); 5b. Ability to influence desired results or that desired actions are taken. With many uncontrollable factors managers or employees can be held less accountable for their activities. Moreover, holding employees accountable for what they can not control can demotivate them; 5c. Ability to measure results or to track actions. Even if results/actions can not be measured effectively, the used measure still is effective if it evokes the right behavior. Furthermore, measures are more likely to work well if they are precise, objective, timely communicated (which leads to earlier interventions), well communicated and internalized, and understandable. Type of control system Topics Characteristics Requirements Examples
Result controls
Result accountability 3a, 3b 4b, 4c, 4d, 4e 5a, 5b, 5c Pay for performance
Action controls
Behavioral constraints 3b 4a 5a, 5b Separation of duties Preaction reviews 3a, 3b, 3c 4a 5a, 5b Budgeting process Action accountability 3a, 3b, 3c 4a, 4c, 4e 5a, 5b, 5c Prespecified policies Redundant capacity 3c 4a 5a, 5b Extra employees
Personnel/cultural controls
Selection and placement 3a, 3b, 3c 4b, 4f 5a, 5b Application procedure Training 3a, 3c 4b, 4f 5a, 5b On job, in classrooms Providing right resources 3c 4b, 4f 5a, 5b Providing information Strong culture 3a, 3b 4b, 4f, 4g 5b Codes of conduct
The more the mentioned requirements are met, the tighter the control systems are applied. Behavioral constraints are also applied more tightly when they are more reliable and restrictive. Preaction reviews are applied more tightly when executed more frequent, more detailed, and performed by informed person(s). Finally, result accountability, action accountability, and a strong culture are tighter when rewards are more significant to the person(s) rewarded.
In choosing between different control types, personnel/cultural controls are worthy of considering first, because they are relatively cheap to implement and have relative few side effects.
An advantage of action controls is that it is not necessary to have motivated employees to ensure that desired work is done, because it is determined for them what they must do. However, it is often still better to have motivated employees for better results, more creativity, and to prevent collusion. Furthermore, documentation of best practices can improve effectiveness. However, with action controls there can be too much focus on the actions itself instead of on the necessary ones which lead to the right results (determined by what is needed in the future for a better adaptation to the changing environment instead from documented past best practices).
An advantage of the result and personnel/cultural controls is that they, if implemented well, increase motivation and lead to the desired activities, even when desired actions are unknown. Also, by not describing how employees should do their work, they can creatively find best solutions for individual cases. However, result control and personnel/cultural controls can fail when the requirements are not met. Finally, not all employees like or need to have autonomy.
This brief theoretical introduction of (management) control and control types has led to a list of issues (1a – 5c) which might deserve attention in empirical research that intends to characterize the control systems of an organization.
Characterizing HAVI’s control systems
In appendix III interview questions, based on the previous points of attention (1a – 5c), are enclosed, which were used in an interview with the manager finance & administration to learn about HAVI’s control systems. Below, the findings from this interview are presented.
The five year strategy, set by HAVI Europe, focuses on growth and product/market development (by adding more and different customers). HAVI Europe has disaggregated this corporate strategy into key strategies focusing on periods of three years, which are disaggregated again into yearly targets* for HAVI (called tactics within HAVI). These tactics focus on growth,
* According the manager finance & administration, each target within HAVI has to be specific, measurable,
market development, product and service quality, costs, and social responsibility. During a strategic meeting, focusing on the budget year of 2009 (January to January), each department manager of HAVI presented their achievements of last year and their new budget targets (coordinated by the finance & administration department, based on historical data) and the planned activities to achieve these tactics. These activities include capacity expansion plans, (human) resource allocation plans, and other activities with respect to the tactics. In a second meeting adjustment between department budgets took place.
As a preaction review control system, planned investments have to be approved by HAVI Europe, the managing director of HAVI, and several department managers.
Another control system HAVI currently uses is a combination of action accountability, and selection and placement: operational employees† are weekly or monthly checked by (department) managers‡ with ex-post measures about quality, compared with targets (e.g. by checking the number of stock shortages and the productivity at the order-pickers). The main ex-post measure for HAVI is total delivery reliability, measured in on-time deliveries and the number of correct deliveries. Department managers are, however, free in the use of the ex-post measures with respect to setting targets and the use of ex-ante prescriptions of work (job descriptions and process protocols). The manager finance & administration (also a department manager) stated that HAVI relies on department managers to lead the employees and to motivate and treat them socially. This means that department managers determine the restrictions for employees, however, they themselves are relied on their personal qualifications (determined during selection and placement) to take responsibility for the functioning/behavior of their employees. The finance & administration department does facilitate departments with ex-post measures of actions, which is also planned to be done more frequent in the future. Furthermore, according to the manager finance & administration, the process protocols will be further developed for a more effective transfer of knowledge about working practices to new employees.
Another personnel control is the possibility for operational employees to get promotion to management functions. According to the manager finance & administration, this increases employees’ motivation and initiatives, and it functions like a bonus. HAVI also provides employees on the job training and department managers with a training on European level.
†
Where I mention operational employees I mean administrative employees and other subordinates of (department) managers.
‡ Department managers are superiors of other managers and/or team leaders/supervisors and operational
HAVI will also have some redundant capacity in the future by attracting a new management director assistant. This employee will get training on the job to take over tasks of the managing director to make HAVI ready for expected growth.
As result accountability is regarded bonuses (on top of a fixed salary) for department managers, which are made dependent on the achievement of budget and performance targets. These managers have to explain their own results for this reason to the managing director. Bonuses for administrative personnel (part of the operational employees) depend on the illness rate, performance measures, and a yearly appraisal (‘beoordelingsgesprek’) with measures of results which could be subjective. Bonuses for operational employees only depend on the illness rate and the same appraisal.
A second result control is a two-yearly evaluation of managers and the company by the operational employees.
So far a brief résumé of control types used by HAVI.
In-depth research topics
Based on the orientation interview and control systems’ requirements, the following fields of interest to investigate about HAVI’s control systems are mentioned here.
Firstly, the manager finance & administration stated that the strategic planning might possibly be optimized on resource allocation or target setting, for a better knowledge of the desired long-term results and actions (see also 1. strategy formulation). Department managers currently autonomously propose strategic plans; however, more or less formalization of this process and/or responsibilities might lead to better goal congruence between the organization and the individual departments. Kaplan and Norton (1996) described the balanced scorecard, which can be used for translating strategies into operational performance measures. Elements from this theory will be explored for possible points of interest about the control systems’ functioning.
Thirdly, according to the manager finance & administration, employees experience their work as challenging and not too boring and the illness rate and the labour turnover (compared to the industry) are seen as good measures of employees’ satisfaction with the control systems. The manager finance & administration also stated that there are not many uncontrollable factors, due to the fact that HAVI operates in a stable environment. In this perspective, it is possibly interesting to investigate whether department managers and operational employees are indeed motivated by the used control systems and whether they experience their work as achievable. This also relates to the following point of interest.
HAVI’s employees are (partly) evaluated on the achievement of results by using a – possible – subjective appraisal with related bonuses. Objective controls (4c, 5c) might be experienced as better influenceable. The functioning of this result accountability control system is another topic which will be investigated.
By adding up the topics to investigate the following central research question is formulated:
Central research question: How can the functioning of the control systems,
available to the department managers to control the operational employees, be investigated?
This question is that broad due to the methodological considerations previously mentioned in this study (see introduction). The central research question will be focused on the four already mentioned fields of interest for HAVI, rewritten as the following sub research questions:
1. Does more or less formalization of the (strategic) planning and control systems lead to better goal congruence (results)?
2. Can (department) managers or operational employees determine best which actions should be taken?
3. Are (department) managers and operational employees motivated by the used control systems and is work experienced as achievable?
STAGE II: IN-DEPTH LITERATURE REVIEW
This section first focuses more in-depth on methodological points of interest with respect to the central research question. These points of interest determine the research design used to gain insights into which method might lead to best knowledge about the sub questions. The second part of this section focuses in-depth on points of interest from academic Management Accounting and Control literature to determine the content of the further research, but it also discusses some further methodological considerations. Every italic word (except from citations) is a point of attention for this research; these points are summarized in table 2 and 3. In the next section the points of attention from this section will be combined with the previous section which results in the research design.
Methodological literature: categorization of scientific methods
When a researcher wants to investigate a certain problem, a first issue is the type and design of the research he wants to conduct.
According to Blumberg et al (2005, p. 124) there is a wide distinction in research between qualitative studies (e.g. focusing on words and sentences) and quantitative studies (e.g. focusing on numbers and figures). Investigations often start with qualitative studies exploring new or unknown phenomena (i.e. research questions) and forming theories/hypotheses, which are later tested on validity (i.e. explained or tried to be proven) by quantitative studies. However, according to Blumberg et al (2005, p. 126) the “quality of any research does no so much depend on whether it is qualitative or quantitative, but rather it depends on the quality of its design and how well it is conducted.”
The design of studies can further be classified by:
• method of data collection (monitoring or questioning);
• researcher control (describing non-influenceable real-life occurred phenomena or conducting a controlled experiment);
• purpose (describing phenomena or making conclusions about causalities between variables);
• time dimension (a cross-section of one moment in time or a longitudinal study);
• the topical scope (broadly generalizeable and supportable outcomes about few variables from statistics about a large population or emphasis on a full contextual analysis of fewer events with less generalizeable and supportable outcomes);
Blumberg et al (2005, p. 132) describe several qualitative (exploratory) techniques: secondary data analysis (e.g. a literature review), using experience from other persons, a discussion with a focus group, and a two-stage design (with the first stage as to define research questions by exploration and the second stage as to develop the more focused research design).
Quantitative (explanatory) techniques focus on statistical techniques to describe statistical evidence about single variables or relations between variables. Data from surveys (interviews, questionnaires), experiments or databases is often used for this purpose
To build (qualitative or quantitative) field research, Blumberg et al (2005, p. 430) state that before research is conducted, the following has to be considered: type of data needed to answer the research questions, communication approach, usage of structured questions or unstructured questions (or a combination of the two), and whether the questioning should be undisguised or disguised. According to Van der Stede et al (2005) the quality of research also depends on the accuracy of data entry and disclosure and reporting.
According to Blumberg et al (2005, p. 205 - 226) to make significant and relevant statements about a population (HAVI in this case) a proper sample must be taken. When a population is small and variable, and the sample will in no way represent its population, only a census of the results from the interviewees is more proper. The size of the sample should depend on the variation in the population parameters and the precision needed by the researcher.
Methodological literature: Detailed discussion of some methodological issues
A widely discussed concept in field research is the case study. For the design of case studies, Blumberg et al (2005, p. 195-196) state that the research requires a clear purpose definition, a detailed process planning, a thorough research design, and high ethical standards applied (rights of involved actors are protected). Research design can be tested on whether:
• limitations are revealed (quality of the study and whether deviation from the planning occurred);
• there is a clear analysis and weighing of information; • the findings are presented unambiguously;
• the conclusions are justified.
Yin (1989, p. 41) adds to this that research design can be tested on:
• construct validity (e.g. by citing multiple sources of evidence and by a detailed documentation of evidence);
• external validity (study is reproducible for a later broader generalization by applying a broad range of theories in the case);
• reliability (the study must be reproducible by carefully planning the data collection in a research planning/protocol).
According to Yin (1989) no single research strategy exists for analyzing information, which makes case studies more difficult to conduct. Furthermore, Grünbaum (2007) states that the definition of a unit of analysis is ambiguous. For example, according to Vaughan (1992) the unit of analysis can be everything from a construct to the case itself. This makes it even more difficult to find valid and reliable case studies. To solve this, Grünbaum (2007) states that researchers must clearly distinguish the unit(s) of analysis from the case(s). Other authors (Keating, 1995; Otley and Berry, 1998) indicate that researchers must be clearer in case studies about their theoretical position, and what contributions to theoretical development they intend to make (e.g. theory discovery, theory refinement to make theory more testable, and theory refutation from practical or theoretical evidence).
Methodologicalliterature: Question design and its limitations
Taylor (1967, p. 88) stated that questions themselves can trigger various employees into “bringing back answers, recommendations or alternatives for improvement or cost reduction” about non-essential or dysfunctional processes or procedures, however, he also supports that questions and questionnaires need to be well constructed. The following points of interest possibly support the latter.
Firstly, according to Spörrle et al (2007), open-questions often serve for exploratory purposes, and closed (multiple choice) questions to present possible alternatives, which requires prior knowledge about the unit of analysis. This raises the dilemma that a decision between alternatives might stay away when using open questions while the trigger for employees to explore and determine possible alternatives might stay away when using closed questions.
399-400; Schwarz, 1999). The instruction for questions can also lead to different answers (Vautier, 2003).
Thirdly, according to Roberson and Sundstrom (1990) a method of getting a higher response is by starting with questions which are of most concern to the respondent (e.g. their wage). According to Blumberg (2005, p. 251) a method of getting better response is by informing the respondent well and making them comfortable with the questionnaire and its (existing or non-existing) consequences. Schwarz (1999) advises to use objective and quantifiable scales as much as possible, and not estimations (like with a low to high frequency scale). For example, instead of asking: “how often do you watch television”, which can result in answers between a number of hours and statements like “frequently”, formats should be used. For example, “I watch television, ___ hours per day”, including the explanation that this has to be expressed in number of hours.
Fourthly, when asking about questions, relating to satisfaction, Schwarz (1999) states that the memory (temporary or long-term) on which respondents base their answer, can also determine the answers. Respondents can even answer questions, where they in fact have no knowledge about. Involving these factors in the questions may lead to different cognitive processes, which can differ per respondent. McFadden et al (2005) conclude that it may be impossible to cover the “correct” responses, and that future research needs to define how potentially difficult questions can be asked, and how researchers can deal with cognitive processes.
Classification of research *
Qualitative research Quantitative research
Secondary data analysis Questionnaires Experience from persons Experiments
Research design:
Method of data
collection Discussion within a focus groups Databases Two-stage design
Research design:
Researcher control
Describing uninfluenceable real-life occurred phenomena or conducting a controlled experiment
Research design:
Purpose
Exploring (form theories/hypotheses about single variables or relations between variables)
Explaining (describe prove about single variables or relations between variables), often hypothesized during qualitative research.
Research design:
Time dimension
Cross-sectional or longitudinal study
Research design:
Topical scope
Emphasis on a full contextual analysis of fewer events with less generalizeable and supportable outcomes
Broadly generalizeable and supportable outcomes about few variables from statistics about a large population * NB. Variables under qualitative research and quantitative research apply to both columns. Variables only under qualitative or quantitative research only apply to these columns.
Classification of research*
Qualitative techniques Quantitative techniques
Communication approach
Structured questions or unstructured questions (or a combination of the two),
Research design:
Further
considerations Undisguised or disguised questioning. Accuracy of data entry
Disclosure and reporting.
Sample if population is steady and big. Census if population is small and variable Determining the size of a sample and population: depends on population
parameters and precision needed by researcher Detailed process planning
High ethical standards
Clearly distinction between unit(s) of analysis and the case(s).
Research design:
Further considerations
for (case) studies Theoretical position (theory discovery, refinement, refutation) Limitations are revealed
Clear analysis and weighing of information;
Ex-post test of (case study)
research design Findings are presented unambiguously; Conclusions are justified.
Construct validity Internal validity External validity
Reliability
Open-questions often serve for exploratory purposes
Closed (multiple choice) questions requires prior knowledge about the unit of analysis
Participants perceptions and cognitive factors
Multiple choice questions can lead to different conclusions than open-questions. Different meanings of words and numbers
Type and order of a scale (objective and quantifiable scales preferred) Instruction influencing answers
More response by starting with questions which are of most concern of the respondent
Better response by informing the respondent well and making them comfortable with the questionnaire
(Short-term and long-term) memory influencing answers
* NB. Variables under qualitative research and quantitative research apply to both columns. Variables only under qualitative or quantitative research only apply to these columns.
Table 2: Research design selection; methodological variables (2/2)
This methodological literature has led to a list of issues (table 2) which might deserve attention when designing or assessing a study. Next an introduction to the discussion of academic literature is given. From this literature elements of control systems will be considered which might also deserve attention during the empirical research.
Academic literature: introduction
control systems, and effects of result accountability control systems (e.g. appraisals). These points of attention are more in-depth derivatives of the points of attention mentioned in the previous section, and are used here to learn how these constructs can be used in control systems’ research.
Academic literature: Control systems
Control systems are also defined as ‘management control systems’ (Merchant and Van der Stede, 2007), ‘management information systems’ (Beckett, 1965), and ‘management accounting systems’ (Reid and Smith, 2000). Depending on how exactly they are defined, all these concepts can be more or less synonyms; here, they are regarded as such. Research is often conducted on individual parts of control systems; however, some researchers suggest a more comprehensive analysis. For example, Malmi and Brown (2008) state that many researchers consider single themes or practices which are seemingly unconnected, while they in fact invariably are dependent on a broader control system. Prince (1964) and Mauldin and Ruchala (1999) also suggest that, when doing research, interdependencies between different factors, and the change of the factors themselves, should be taken into account. Nowadays, more complex decisions are able to be made due to developments in Information Technology (IT), which results in a growing need for integration of different sub-disciplines in accounting research, Therefore, task-orientation and system design and implementation is needed that recognizes interrelations among major categories of contextual factors (contingencies): technological (IT, production systems, etc.), organizational, and cognitive.
Many other researchers (e.g. Chenhall, 2003; Reid and Smith, 2000; Hopwood, 1983) support that control systems vary within organizational contexts, e.g. on the basis of such factors such as the nature of the environment, size, structure, strategy, and national culture. This also possibly indicates that strategic control systems and management control systems should not be studied separately (Merchant and Van der Stede, 2007). Rather, objectives for control systems, derived from organizational and environmental circumstances, specific to the studied case, should be taken into consideration (Otley, 1980). Furthermore, depending on consensus or conflict about these objectives, or the degree to which they are imposed to (or implemented by) different stakeholders, can result in a different nature of these objectives (Otley and Berry, 1980). Some of the previously mentioned factors will be further discussed here.
documents, etc.) often dominate formal information, also leading to proper organizational results in the different cases. Winn and Keller (2001) attempted to describe a model to make a hierarchy between different objectives for different stakeholders from these many sources of information (from newspaper items to phone calls with customers); however, this model still seems quite intuitive and too unstructured to serve as a basis to determine the content of control systems (in determining which actions should be taken, etc.).
Secondly, according to Luft and Shields (2003), the amount of formalization (i.e. decide to arrange something according to a fixed structure) depends on the operational and financial separation in the organizational structure and the dominance of financial reality, while these factors also can influence each other. Macintosh and Daft (1987) found that when interdependencies between departments increase, they at first use more operating procedures, then more complex budget and statistical reports, and at its highest dependency, usage of these three control systems declines again. Bouwens and Van Lent (2006) add to this that when interdependencies between departments become increasingly important, disaggregated information also becomes more important, while aggregated information also contains more noise (uncontrollable factors of other departments). So more complex interdependencies call for more complex control systems, however, interdependencies at their highest complexity require less. This also relates to Ouchi’s (1979) model, which will be discussed below.
Thirdly, a more covering (classical) approach, to determine whether control systems can be more or less formal, depends on contingencies such as knowledge of the transformation processes and measurability of output (Ouchi, 1979). Formal rules and measures become more feasible when these factors are at a high level. However, although this reasoning seems to make sense, from the previous explanation it can be concluded that formal control systems may also become more or less feasible due to many other factors and dependencies. To add, Rankin et al (2008) showed from an experiment that superior authority alone improves budget proposals, even when the same information is used.
to determine whether they are enabling, the following four issues should be regarded: are employees able to ‘repair’ (i.e. solve) problems in control processes, is there enough internal transparency to understand the working of these processes, is there enough global transparency to determine whether these processes fit in the organization as a whole, and are processes flexible enough to be influenced.
The constructs mentioned above also relate to the next part of this section, while these are considered as being a part of the same control systems.
Academic literature: Formalization of strategic planning and goal congruence
Some management control systems pay explicit attention to and intend to be explicitly related to the strategy of the organization, which is the case with e.g. the balanced scorecard. However, with other systems such relationships are much more implicit.
As Merchant and Van der Stede (2007) stated, a more clear strategy yields a larger set of feasible control alternatives. Many authors have indeed found (direct and indirect) variables which have effect on control systems. However, Langfield-Smith (1997) stated that knowledge about the relation between management control systems and strategy is still limited. For example, Bouwens and Abernethy (2000) found that when organizations pursue a customer-responsive strategy (such as customization) this does not directly translate into a change in management control systems. Rather, the primary effect of this type of strategy is via greater interdependencies between departments and the necessity for more sophisticated information. Daniel and Reitsperger (1991) also found that when organizations pursue a strategy focusing on quality it is more likely that goal-setting and providing feedback relating to quality take place more frequent. Slagmulder (1997) supports that when strategic investment decisions change; organizations also adapt their control systems by processing changes in the environment with operational adaptations of control systems. However, none of the mentioned authors provide a structured guideline on how to decide when to use which control system with which strategy; rather they give an enumeration of incoherent decisions which can be made concerning strategies.
with its rate of success determined with outcome measures. The ambition of the balanced scorecard (Kaplan and Norton, 1996, p. 56, italic phrasing not in original paper) is “to articulate the strategy of the business, to communicate the strategy of the business, and to help align individual, organizational, and cross-departmental initiatives to achieve a common goal”. Furthermore, executives can use the balanced scorecard as a communication, information, and learning system. Merchant and Van der Stede (2007, p. 473) state that non-financial performance measures indeed seem to create a more strategic (long-term) focus, because, accounting measures alone ignore, for example, more future oriented intangible resources (such as research in progress and human resources). If investments are detrimental for short-term returns, managers may be reluctant to invest in the assets, necessary for long-term survival.
Other researchers have a more skeptical view on the balanced scorecard (e.g. Otley, 1999). Nørreklit (2000) brings forward the following points of critique on Kaplan and Norton’s balanced scorecard, which are also partly covered by Otley (1999).
Firstly, the cause-and-effect relation between action and result measures is problematic, because this would need two points in time to determine whether actions (means) indeed lead to certain results (ends), while the balanced scorecard only shows both measures at the same time. Linkages between means and ends can be bidirectional instead of the suggested unidirectional (which is the case with cause and effect). For example, in order to achieve good financial results, prices can be held low (from which customers become more satisfied) to attract more customers. Later on, prices can be raised for better financial results but at the cost of less satisfied customers. Thus, satisfied customers can lead to better financial results (as stated by Kaplan and Norton, 1996), but better financial results can also lead to less satisfied customers.
Secondly, no guidelines are given for how to involve other stakeholders, or how to adapt the balanced scorecard on environmental developments (e.g. with benchmarks from competitors) in the balanced scorecard.
Thirdly, the balanced scorecards pretends that it gets rooted into organizations, however, it is not taken into account that implementation might not be easy and that this process itself may also change the content and effectiveness of the system.
Previous parts, focusing on the academic literature, have dealt with (management) control systems in general and the first sub research question (formalization of strategic planning). The last part of this section deals with academic literature about sub research 2 up to 4.
Academic literature: Remaining research questions
Whether employees are performing well by or are satisfied with the controls depends on many factors (Luft and Shields, 2003). These factors can be influenced by other factors and the cause-and-effect relations are still unclear. Even worse, a(n) (approximate) replication of prior studies about causalities may even lead to opposite conclusions (Otley and Pollanen, 2000). For example, job performance is often seen as depending on factors such as budget participation, job satisfaction, and role ambiguity, however, the opposite is in an experiment also shown to be true (Nouri and Kyj, 2007). The same can probably be said about the previously presented research questions.
This ambiguous and seemingly inexhaustible amount of factors make it difficult to decide on the basis of literature with which factors it is possible to determine whether operational employees (subordinates) and/or department managers (superiors) are motivated/satisfied by the control systems. The same counts for the question whether subordinates or superiors should determine which actions should be taken and how the effectiveness of appraisals can be determined. In appendix IV points of attention are mentioned which are possibly useful to address these questions. Table 3 summarizes all academic literature discussed in this section.
Variable category
Control systems’ variables for the sub research questions to address Merchant and Van der Stede’s (2007) general control problems (Knowledge of desired
results/actions, the ability to influence desired results or that desired actions are taken, and the Ability to measure results or to track actions)
Research approach
1. Comprehensive analysis of factors of control systems, interdependencies between factors and the change of factors themselves (e.g. from discussion about objectives) is needed.
2. Contingencies: technological, organizational, and cognitive.
3. Due to the comprehensive analysis of the organization and its context, contingencies (can) apply to all four sub research questions.
1. Informal and formal.
Research variables
Information types
2. Make a hierarchy between objectives for different stakeholders from different sources of information.
1. Financial and operational separation within the organization makes formalization more or less necessary.
2. Complex interdependencies between departments require more complex control systems.
Necessity and feasibility of formalization
3. Interdependencies at its highest complexity require less complex control systems. 4. Aggregated information contains more noise.
Variable category
Control systems’ variables for the sub research questions to address Merchant and Van der Stede’s (2007) general control problems (Knowledge of desired
results/actions, the ability to influence desired results or that desired actions are taken, and the Ability to measure results or to track actions)
5. Better knowledge of the transformation processes makes formalization more feasible.
6. Better measurable output makes formalization more feasible.
Research variables
Necessity and feasibility of formalization
7. Superior authority (power) improves budget proposals (and control systems?) 1. Are employees able to solve problems in processes?
2. Are employees able to understand the processes where problems need to be solved?
Enabling or coercive control
systems? 3. Do the processes fit in the entire organization?
4. Are employees flexible enough to influence the processes?
Unguided adaptations of control systems:
1. Quality strategy causes greater interdependencies between departments and the requirement of more complex information. More frequent goal setting and feedback is also more likely.
Strategy and a set of feasible control alternatives?
2. Changes in the environment cause operational adaptations of control systems.
Claims about the balanced scorecard as a structured guideline and its limitations:
1a. Translates a company's vision into a coherent and linked set of operational measures to control four different perspectives.
1b. The claimed coherence is doubtful and should be analyzed.
2a. Balanced scorecard useful to articulate and communicate the strategy, and to align individual, organizational, and cross-departmental initiatives.
2b. Implementation might not be easy and this process may also change the content and effectiveness of the balanced scorecard, which is not taken into account with the claims.
3a. The balanced scorecard contains both measures of actions and their causal results.
3b. Causal linkages are not proven, because means and ends can be also bidirectional instead of the claimed unidirectional.
3a. The balanced scorecard is a (dynamic) learning system.
3b. This claim doubtful, because no guidelines are given how to involve other stakeholders or how to adapt the balanced scorecard dynamically on environmental developments.
4a. A strategic dialogue between employees is needed to undertake a continuous search process to uncover threats and opportunities.
Other Appendix IV
Table 3: Design, role and effects of control systems’ variables (2/2)
Conclusion
for by managers. However, whether positivistic approaches should be banned could also be questionable, because, for example, managers and their control systems can be effective due to their presence alone (Rankin et al, 2008; Bruns and McKinnon 1993).
This study is also influenced by methodological issues, which can limit the effectiveness of the research design and results from this study. For such reasons, the decision about the type of research might have to be taken in part on the basis of a (subjective) interpretation of the research situation which exists in practice. On the highest level, a choice has to be made between qualitative (exploratory) and quantitative (explanatory) methods. Possibly, this choice can not be based on sound arguments, because for quantitative studies closed questions must be asked, which requires prior knowledge about the research topics of interest. However, this knowledge can be limited when prior studies (e.g. stage I in this research) are in conflict about terminology about, for example, the case and the unit of analysis, theoretical positions of studies, quality of research design, cognitive factors, and other, possibly, almost inexhaustible number of variables.
STAGE II: IN-DEPTH RESEARCH DESIGN AND METHOD
In this section points of attention from the previous sections are used for the development of the in-depth research design and method. Investigative questions for interviews and questionnaires, as well as the type of questions and the type of data collection and processing are discussed here. In the next section results and considerations about HAVI’s control systems will be discussed, which are based on the questions and research method from this section.
Definition of the case study
This study draws on data from a single case study. Conform Grünbaum (2007), the case is distinguished from the unit(s) of analysis: the case is determined as ‘administrative (operational) employees and their superiors’ within a lead logistical provider. The units of analysis are the methodological and academic variables from the previous section, rewritten used as a basis for and into investigative questions which are asked to the involved employees of HAVI.
Data sources
Two types of data were gathered in three days over a period of almost two weeks.
Firstly, interviews with five different managers and one administrative employee were held (table 4). These interviewees were chosen by the finance & administration manager, judged on their capability to answer the questions. On average, these interviews lasted about 65 minutes. Written notes were made during the interviews; however, they were not tape-recorded. Two interviewees were asked to judge reports of their interviews, based on the notes. They confirmed that the report was in line with their opinions.
In the previous section it was concluded that the choice between qualitative (exploratory) and quantitative (explanatory) studies can be difficult to be made. As can be seen in table 2, qualitative (open question) studies can create new insights and makes multiple choice questions better possible to be determined due to prior knowledge about job-specific topics. Explanatory (multiple choice question) studies, however, can create stronger and more generalizeable conclusions about qualitative studies.
versions 1 and 2). Not all questions are in the open form in version 1 and in the multiple choice form in version 2 because of the limited time available at HAVI. The questions that are asked in the open form are about strategic topics and the balanced scorecard (because of special interest shown in these topics by HAVI). Also the enabling/coercive principles are of special interest, because they can possibly provide a basis for comprehensive conclusions about control systems.
Secondly, data was collected using a questionnaire (see questions 12 up to 43 for version 3 in appendix V) for the six interviewees and twenty-five administrative employees within several departments (table 4). These were chosen to make a comparison between the interviewees and their respective subordinate departments. The questions in the questionnaires were all asked in the multiple choice form. The varying open and closed questions and question number 12 (appendix V), designed for the interviewees, were also asked to the administrative employees, however, only in multiple choice form and only focusing on their own situation. So, for example, for question 8, “I and the employees have enough possibilities to...” is changed into “I have enough possibilities to...”. All questions were asked in Dutch, except for the department manager of finance & administration. Table 5 summarizes the different data sources.
Functional breakdown of data sources
Abbre-viations
Number of persons
Interviews and questionnaires for superiors (i.e. interviewees)
(1) Department manager – Finance & administration FAM 1 (2) Replacing department manager of (1) – Finance & administration RFAM 1 (3) Supervising manager of (5) and (6) – Operations management OPM 1 (4) Administrative employee – Human resources HRAE 1 (5) Supervising department manager of (8) and (9) – Supply chain management SCM 1 (6) Department manager – Distribution center DCM 1
Questionnaires for administrative employees
(7) Finance & administration department FAD 5 (8) Customer service department CSD 9
(9) Procurement department PCD 7
(10) Human resources department HRD 1 (11) Transport planning department TPD 2 (12) More than one department More 1
Total 31
The total number of budgeted fulltime employees (FTEs) within HAVI in (1) up to (12) and other departments per 31 December 2009 is 193.97.
The total number of budgeted FTEs in the transport department (to which (11) belongs) is 40.07.
Table 4: Functional breakdown of data sources and number of samples
example, attitudes and whether organization’s efforts have had the desired effect. Likert scales can be used to compare scores from persons or groups with another distribution of scores.
Version Interviewees Administrative
employees Question # and type
of question (MC =
multiple choice) FAM RFAM OPM HRAE SCM DCM
1+2 1 – 7 open X X X X X X 1 8 – 11 open X X X 2+3 8 – 11 MC X X X X 1+2 12 open X X X X X X 3 12 – MC X 2 13 – 16 open X X X 1+3 13 – 16 MC X X X X 1+2+3 17 – 43 MC X X X X X X X
Table 5: Investigative questions and data sources
To prevent misunderstanding of words, definitions of ‘process’ and ‘information’ from the Oxford dictionary were given. Both before the interview questions were asked and before the questionnaires were have to be filled in, the employees were asked to read these definitions. The concluding part of appendix V contains the definitions, as given in the instruction for the questionnaires. Furthermore, I was present when the employees filled in the questionnaires. No questions about the questionnaire were asked during these moments by the employees.
Pre-structured formats were not used for the open questions, because this could limit the explorative nature. However, this also increases the risk that irrelevant answers will be given (questions which do not reveal what is necessary). Multiple choice questions might solve this problem. However, if I do not have enough prior knowledge about the topics involved, these questions may also be limited in value. The investigative questions are not adapted to the job-specific characteristics per se, so this still possibly biases the results. So, there are various reasons why it could be relevant to use both open (more exploratory) and multiple choice (more explanatory) questions during the interviews.
Cognitive variables (table 2), possibly biasing the results (such as the influence on the answers from the instruction, the Likert scale, etc.), were also not taken into account. Finally, other types of studies (such as a discussion within a focus group, an experiment or a statistical analysis) are not used. However, using these methods could possibly also have lead to new insights.
Assumptions about the data
administrative employee. Third, it is assumed that the questions are understood and interpreted well by each person involved in this study (including the researcher).
Methods of data analysis
In table 6 (in the next section, ‘Stage II: In-depth Results About HAVI’s Control Systems’) answers from the interviews are presented. The abbreviations of the interviewees are used to show who made which statement. In table 7 the answers from the questionnaires are presented, given by the six interviewees and the twenty-five administrative employees (from all departments). Answers from the interviewees for question 8 up to 11 and 13 up to 16 are both in table 6 and table 7.
Data from the multiple choice questions are from a cross-section study and therefore relevant statements can only be drawn about single variables (e.g. question 6 and 12 are proper for this purpose, however, question 13 and 14 are not). Furthermore, the sample itself may not represent a larger population (such as administrative personnel in general) when there are many differences in the control systems between departments and between the hierarchical levels. Finally, it is unknown whether the sample is stable (i.e. whether the turnover and number of employees is relative constant through time) or of relevant size to be used for explanations or statistical tests.
According to Blumberg et al (2005), between 20 and 25 properly structured questions are needed about a topic (in the case of this study about each of the sub research questions) when using the Likert scale for reliable conclusions. However, in the case of this research, each answer from the questionnaire (including open questions) could possibly be used, due to possible interdependencies between many variables. The following processing of the data will be applied.
To deal with the sub research questions, only answers from the interviewees and administrative employees from HAVI in total are necessary. However, in appendix VI, answers per department are also shown. In the next section each sub research questions is separately discussed. As a guide, based on the previous section, in appendix V the sub research questions are coupled to the different investigative questions, according to an expected relation between these topics. However, other relations and interdependencies between variables may also be possible.
In the discussion of the sub research questions answers from different type of questions about the same topics will be compared with each other. Both the results from table 6 and 7 and this comparison are expected to give insights which are relevant for the central research question.
Finally, a suggestion for a strategic control system (an ‘adjusted balanced scorecard’, based on the literature about control systems in the previous section) will be given.
STAGE II: IN-DEPTH RESULTS ABOUT HAVI’S CONTROL SYSTEMS In this section the results from the interviews and the questionnaires are presented in table 6 and 7. This data is used to discuss each research question, consider HAVI’s control systems (table 8) and make suggestions for a strategic control system (an ‘adjusted balanced scorecard’; table 9). Sub research question 1 and 2 are discussed together, because these are possibly invariably dependent on HAVI’s broader control system. This is in accordance with Malmi and Brown (2008), as stated in the in-depth literature section. Sub research question 3 and 4 are also discussed together, because it is expected that motivation by control systems, appraisals (which can be considered as a part of the control systems), and achievability of work are related.
Formalization of the (strategic) planning and control systems, and employees best able for determining actions
The following discussion is built up from question 1 up to 16 from the interviews (table 6). It is widely stated that information about understanding the processes of the customers and their needs, and for managing the supply chain – separated into several criteria/parameters – are most important for HAVI. Expected sales or the customer orders are according to two interviewees the basic most important parameters. Moreover, two interviewees stated that the more HAVI knows about customer needs and processes (except when there is an information overflow), the better both parties can function.
In general, for HAVI to understand the customer needs, could be achieved by finding a balance between controlling costs and to maintain a certain level of quality (in service as well as products), or by finding a balance between service and efficiency; for both with the yearly strategic budgeting process as a basis. The balance between efficiency and service (question 10.3) seems to depend on the flexibility in which choices can be made to work according to appointments with, e.g., the Mc Donalds Head Quarter on the one side, and whether tailor-made solutions for individual customers – deviating from these appointments – have to be made on the other side. With respect to the Mc Donalds example, one interviewee stated that a system is needed where HAVI, Mc Donalds Head Quarter, and the customers can all benefit from.
departments, and newsletters for HAVI’s employees and customers might be necessary to increase commitment, resulting in increased productivity and more passion for the job. This might indicate that other more or less informal information is also important. Finally, continuously benchmarking developments in the industry might be necessary to keep thinking about how things can be done better.
No comments for improvements of the yearly strategic budgeting process are made, neither on the clearness of the strategy. However, it is stated that the process of implementation and the translation of strategies into rules and procedures have to be improved. Furthermore, although one interviewee stated that each employee could tell what their operational parameters – derived from the strategies – are, it is widely acknowledged that the knowledge about the processes of the customers needs to improve for all employees.
It is widely stated that employees have enough possibilities to propose improvements and to tell mistakes, possible threats, and opportunities. However, it is also stated twice that involvement among employees and the process for proposing improvements are not functioning well. A more formal structure to direct necessary improvements into operational processes, possibly imposed by the management, might be necessary instead of the possible more ad hoc way this is functioning currently. According to one interviewee, in this structure all possible improvements should be collected and prioritized and the employee who detects opportunities for improvements should enable the necessary actions. According to another interviewee the long-term interests of HAVI and the customers run analogously, however, the short-term interests – for example between HAVI and the employees – can differ. Also a less strong hierarchy (i.e. formal organizational structure) might be necessary according to an interviewee. One interviewee stated that when decisions have to be made, too much discussion afterwards must be avoided to make the organization function better. It is also stated that motivating targets are necessary, which are realistic and, more important, realizable. Measurement of these targets can also show where improvements are necessary and for whom.
