System Dynamic Modelling for a Balanced Scorecard: A Case Study

By

Steen Nielsen (E-mail: sni@asb.dk)

& Erland H. Nielsen

(E-mail: ehn@asb.dk)

January 2006 Addresses for correspondence:

Steen Nielsen, Department of Business Studies Aarhus School of Business,

Fuglesangs Allé 4, Aarhus 8200, Denmark

Tel: +45 89486688

The paper is presented at the Annual Congress of the EAA March 22-25 2006,

Dublin Ireland.

1

System Dynamic Modelling for a Balanced Scorecard: A Case Study

By Steen Nielsen*

Erland H. Nielsen†

Abstract

Purpose – The purpose of this research is to make an analytical model of the BSC foundation by using a dynamic simulation approach for a ‘hypothetical case’ model, based on only part of an actual case study of BSC.

Design/methodology/approach – The model includes five perspectives and a number of financial and non-financial measures. All indicators are defined and related by a number of assumed cause-and-effect relationships. Time lags are included to stipulate one of the most important characteristic of BSC. We use the model to simulation different strategies or scenarios and to reach a kind of optimum or ‘steady-state-condition. We, also demonstrate through three different scenarios how different variables influence the optimum time that normally also influence decision making.

Findings – The results shows how profit, and ROCE (Return-On-Capital-Employed) are influenced by three chosen variables: skills, customer base and work in process. We are able to show that only minimal changes may create large effects on profit.

Research limitations/implications – Our analytical model is based on both assumed but also on case documented assumptions of a concrete model. In every such kind of model, the sensitivity of our results with respect to the assumptions should be analysed in subsequent studies. Besides, our model should be extended to cover a whole BSC of an actual company.

Practical implications – The model may be used as the first step in putting numbers on an integrated BSC model. The model is our first attempt to study an analytical version of BSC and to make inferences concerning some optimal or rational behaviour of an assumed relationship between different indicators and perspectives. The credibility or benefit of the model is insight into the finality of a model but also as a teaching exercise of BSC. It is relativity easy to extend our model to more realistic circumstances, by including more measures or change time lags. Using the assumed cause-and-effect relationships between financial and non-financial performance measures, attention should also be paid to the definitions and the number of measures.

Originality/value – A large number of case studies and surveys are now present in the literature of BSC. However, there is a lack of more theoretical and analytical modelling of the BSC. Our paper just throws a little light to this modelling approach.

Keywords: Balanced Scorecard, model, analytical, time lag, indicators, financial measures, non-financial measures, and steady-state-condition.

1 IntroductionI The aim of the holistic model of BSC is to implement and explain the strategy of the firm

with subsequently translation into more operational terms (Drury, 2002; Garrison & Noreen, 2000).

*Steen Nielsen is Ph.D. and associate professor at the Department for Business Studies in the Aarhus School of Business in Denmark, and member of Management Accounting Research Group.

†Erland H. Nielsen is associate professor at the Department of Business Studies in the Aarhus School of Business in Denmark, and member of Logistic Research Group.

2

The emergence of information has made many of the fundamental assumptions from the industrial age competition of traditional financial measures obsolete. It is no longer enough to gain sustainable competition advantage just by rapidly deploying new technology into physical assets or by excellent management of financial assets alone. The information age for both manufacturing and service organizations requires new capabilities for success (Johnson & Kaplan, 1987). Companies in a global world economy must develop and retain loyal customers, introduce innovative products mobilize employee skills invest in technology and implement new data bases and systems. Separate and single develop measures only tell a fraction of a company’s story. Therefore, the financial accounting model must expand to incorporate the valuation of the company’s intangible and intellectual assets that satisfy customers and employees (Kaplan & Atkinson, 1998). The balanced scorecard is developed to communicate the multiple, linked objectives that modern companies must achieve to compete on the capabilities. BSC translate mission and strategy into objectives and measures in different perspectives. BSC has also developed over time, from performance measurement system (Kaplan & Norton, 1992), to the so-called strategy map approach (Kaplan & Norton, 2004) with focus on a generic business model based on causality between measures and perspective to an interactive controlling system that drives learning and improvements (Simon, 1994) to a model that also encompasses intellectual capital and finally to an integrated strategic planning process through the BSC (Horváth & Ralf, 2004).

By focusing and controlling relevant key perspectives the company achieve a ‘simple’ description of how a company can attain its strategic goals. The generic model includes four different perspectives: financial perspective, customer perspective, internal perspective, and learning and growth perspective, as stated by Kaplan & Norton (1992, 1996b). The model aims you to control a numbers of key performance areas (or perspectives), based on some form of a cause and effect logic. Subsequently, Kaplan and Norton (2001a,b,c) has developed an overall strategic management concept that consist of: i) transform strategy to operational terms, ii) get the company to focus on this strategy, iii) make strategy every person’s job to achieve, iv) make strategy a continuous process, and v) mobilize changes through a strong and effective management. However, as pointed out by Olve et al., (1997, 2000), the interest in non-financial measures is only a belated awakening to what others have been doing

The news in BSC – in contrast to other holistic models – is the balance and the comprehensiveness and future view. Also the importance of the latest version of BSC is put on cause and effect logics through the concept of ‘strategy maps’. An example of a complete and integrated BSC is, however, still hard to find in the real world, although companies are trying to use BSC as described by Kaplan and Norton.

Maisel (1992) also points to fact that financial measures need to be supplemented by non-financial measures, such as customer’s attitude, quality, innovation and flexibility. Financial measures may in fact create barriers for meaningful communication between perspectives. How many perspectives should be used depends on the company’s strategy. Contrary to Kaplan & Norton (1996a,b), Maisel (1992) discusses a specific perspective for Human Resource Measures. By Kaplan & Norton (1996b) this perspective is included in the learning and innovation perspective. The innovative and developing perspective includes the lifecycle of the products. Because of constant change, intense competition, and more responsiveness to satisfying customer demands, companies need to strike balance between efficiency, develop new products to lower costs (Maisel, 1992). Therefore, BSC needs to stress TQM and continuous improvements.

In conclusion, balance is needed between the horizontal and a vertical view, between financial and non-financial measures, between external and internal measures, between management control concepts such as costs and revenues, between short and long run and between hard and soft values.

3

Balance also implies a balance between lead and lagging indicators (Kaplan & Norton, 1996a, p. 56; 1996, p. 25).

BSC is only one out of several new holistic accounting innovations. However, in contrast to more partial and single-purpose oriented innovation, such as ABC, target costing, kaizen costing BSC was first introduced in 1992 by Kaplan & Norton in the article, ‘The Balanced Scorecard – Measures that Drive Performance’ from Harvard Business Review, as one of more holistic concepts or innovations with the focus on management the company from it’s vision and strategy (other models or company models are the EFQM model (European Foundation for Quality Management), the TQM model (Dale & Plunkett, 1995; Conti, 1999), the Performance Pyramid (McNair et al., 1990), Business Models (Eccles et al., 1991), Intellectual Capital Statements (Mouritsen et al., 2002) or classical budgeting models (Shank & Govindarajan, 1993). Kaplan and Norton (1996b, 2004) point out that it is important to be able to evaluated the economic effect or outcome in the financial perspective (e.g. by an improvement – or reduction - in ROI, ROS, ROA, RI, or ROCE). BSC is a tool to transform the company’s vision, mission and strategy into operational terms and integrated connections, or as Kaplan & Norton (1996b,p.2) write, ‘The Balanced Scorecard translates an organization’s mission and strategy into a comprehensive set of performance measures that provides the framework for a strategic measurement and management system’.

The Balanced Scorecard developed by Kaplan and Norton (1992, 1996a,b,c, 2001a,b,c, and 2004) has become a very popular concept and an important innovation in almost all countries around the world. A large number of books, articles, and papers have been published that examines the concept or part of it in theory writings (Maisel, 1992; Nørreklit 2000). Besides, Kaplan has developed his own problem based research methodology, ‘Innovation Action Research: Creating New Management Theory and Practice’ (JMAR, 1998) as a template for solving practical problems. One direction began appearing in the mid of the 90’s and dedicated to case-studies (Ax & Bjørnenak, 2005; Malina & Selto, 2001, Olve, et al., 1997; Kraus & Lind, 2002/03). Another direction began from end 90’s and beginning of 2000 and was conducted to a number of surveys (Hoque & James, 2000; Töpfer et al. 2002) and in the Nordic countries (Bukh et al., 2000; Malmi et al., 2001; Nielsen Sørensen, 2004). Now a quantitative or model direction is beginning to show, dedicated to analytical oriented studies and frameworks concerning the problem of transforming BSC into a dynamic model (Laitinen, 2004, 2005; Schöneborn, 2003). Often company invests and implement in more than one accounting innovation at a time (Joshi, 2001, Nielsen et al., 2002) which also indicate a great interest in new accounting innovations.

Also a large number of cases which also have emphasized the opportunities for a success-oriented enterprise control have been published in recent years. The reason for this fast spreading of the BSC in all kind of businesses around the world is probably the first real attempt to integrate the strategic dimension and objectives of the company and to formulate specific measures that should be used to increase profit. However, there also seems to be heavy difficulties in formulating these strategic objectives or goal into the company’s operational processes and finally to locate the benefit in the P & L statements. What has to be done in the operational and concrete area is to make sure that the objectives or goal will be achieved. How are these operational measures linked with the strategic objectives, and how can financial and non-financial measures be linked together? I.e. how can we translate the different perspectives and different kinds of measures into a common languish that can be used to evaluate and control the company? The traditional four perspectives -financial, customer, internal business processes, and learning and growth - only provide a generic model without any real description of how the measures should be linked and used as drivers for future performance. The core elements of BSC are strategic success factors, which should guarantee the survival of the company as well as its financial success in the long run. It is, however, stressed

4

that the BSC does not only represent a mathematical model or system. However, the core idea of a BSC as pointed out several times is that the financial success of an enterprise can be measured or determined in the long run. However, an isolated view and maximization of financial results may also lead to a neglect of other measures, to a misuse of the financial element or to conflicting signals about the organization values (Kaplan & Norton, 1996a). This will not create a long-term consensus for the member of the company. Therefore writers on BSC often use the metaphor of entering a cockpit of a jet plain. On the other hand, only a mixture of financial and non-financial measures represents a limited application of the full power of BSC because of the missing strategy indicators (Kaplan & Norton, 1996a).

Kaplan and Norton point out these relations or hypotheses relations between perspectives as an important facet of BSC (Kaplan & Norton, 1996b). These perspectives should be connected to the financial and monetary success of the company. Therefore, the recognition of single elements or measures and their influence on other measures including the financial area are important for a success oriented company’s control view. To bring strategies in action, measures have to be implemented which guarantee the achievement of the strategic goals and an increase in profitability in the long run. However, trying to connect measures to one single outcome measure has been tried before e.g., in the DuPont Powder Company, formed in 1903, where the financial outcome was on the Return of Investment (ROI). One problem is related to the high complexity of incorporating several measures to a single financial measure including both different time delays and stochasticity in measures. These problems are not yet part of the practical problems for BSC, probably because companies have not yet reached this analytical step. Therefore, the purpose of our paper is to start and develop this step and to illustrate a network dynamic model which integrates different time lags situations with the help of causal loop diagrams in a system-dynamic research. Even though our model may labelled as a naive estimation of the real world it has some creditability and gives some insight of the extreme difficulties that are to be expected in the further if this analytical perspective is going to be explored. However, by using a dynamic model approach it becomes possible to conduct and test different scenarios by the help of simulation.

There has been some discussion concerning how the cause-and-effect relations should be interpreted. Nørreklit (2000, 2003) and Nørreklit et al. (2004) has taken the point that BSC should be based on a generic cause-and-effect relationships mostly takes for the natural science or on logic relationships. Others, (e.g. Bukh & Malmi 2005; Ittner et al., 2000; Kaplan & Norton, 1996b, 2001a,b,c, 2004) has taken the point that the cause-and-effect should be based on a more pragmatic, practical contingency or assumed approach. Even though we mostly agree with the last mentioned approach which also is in line with what Mattessich (1995a,b) calls a finality approach, we also think that these relationships must be conducted by statistical hypothesis testing methods, like all other kinds of relationships in economics. The important point is that the assumed relationship between different kinds of indicators, actual works, which may first be evaluated after some time. Besides, statistical test and estimation have a long tradition in management accounting (see Kaplan, 1982; Hirsch, 2002). However, the word ‘causal’ has many meanings (James et al., 1987). As we see it, the cause-and-effect relationship in a social science cannot be driven by a stringent cause-and-effect relationship known from the natural science. We see management accounting basically as an applied research approach in line with Ittner et al., (2002), Chenhall, (2003), Samuelson (1990), Horngren, (1977), and Kaplan (1998). Besides, ‘the assumed cause-and-effect approach’ or hypnotises is in no respect different from other approaches within management accounting, e.g. the cost driver in ABC, relationships in ABM between customers and profit, in forecast techniques for revenue and costs, or in developing strategies in association to the economic results of the company. Beside, this assumed approach is also used within econometrics and statistical analysis (correlation, regression, covariance structures, structural equation modelling etc. (Johnson, 1986; Kemnta, 1986). Kaplan & Norton (1996a, p. 67) also point to this possibility, by saying; ‘managers can test the

5

business theory’s hypothesised causal chain of performance drivers and outcomes. If we didn’t use such an approach, we would not be able to come up with any model at all. However, this cause-and-effect may of cause be verified in practical situations (Bukh & Malmi, 2005).

The purpose of this paper is to

1. developed an analytical base model based on a theoretical foundation of BSC with the aid of the simulation language Vensim;

2. apply the model for different scenarios concerning a change in strategy combined with problems searching for steady-state-conditions.

There have only been a few writings concerning the problem of converting BSC into an analytical model. Laitinen (2004) derive a theoretical foundation for BSC, by the use of shadow prices and different strategies. In Laitinen (2005) a microeconomic analysis of BSC on practical company is used, here Nokia Corporation. The model includes demand, production, and objective mathematical functions. Strategy is described by the objective function to be maximized. However, an experimental simulation strategy has been mentioned by Brinberg & McGrath (1988). In building a BSC-model is no different from building models in other areas e.g. management science or logistics.

This has also been pointed out by Kaplan & Norton (1996a, p. 67) who says: ‘at this point, the Balanced Scorecard can be captured in a systems dynamics model that provides a comprehensive, quantified model of a business value creation process.’

The paper is organized as follows. In the next section we present a short introduction to financial and non-financial measures on the background of BSC. In section three, we introduce the concept of model in a classical sense and try to convert the specific assumption on the model of a BSC. In section four, we introduce our framework for our BSC model, its assumptions, containing elements of skills, customer base and work in process. In section five, we introduce three numerical simulation scenarios; one for the given basic ideas, one for changing the number of trainings, and one for changing the number of staff. Finally, the sixth section we summarize the main results of the study and discuss the limitations and making suggestions for further research on the topic. .

2 Financial and non-financial measures Several economic theories discus the choice of performance measures and indicate that

performance measures and reward systems should incorporate measures that provide incremental information on managerial effort (Ittner & Larcker, 1998). It is important to define measures that also improve competitiveness (Ghalayini et al., 1997; Neely et al.,1996).

Many measures used in BSC have already been used and defined in earlier literature for performance measures, both in Anglo-Saxon literature (e.g. Solomon, 1965), in German literature (Hahn, 1974) and in the Nordic literature, e.g. Johansson & Őstman (1995). Several of these may be called ‘generic measure’, e.g. market share; customer satisfaction; customer retention; profitability; EVA (Economic Value Added); because they are already accepted and have been used for decades.

One of the main problems in BSC is the combination and integration of financial and non-financial measures. Non-financial measures are normally considered short termed or operational measures, whereas financial measures are long termed and strategic measures, Atkinson & Kaplan (1999). In BSC these two types of measures must work together. At the same time assumed relations and measures between different perspectives e.g. the outcome profitability measure from a customer perspective should be related to the sale in the financial perspective. The construction of BSC also allows other accounting innovations to be included, e.g. TCO (Total Cost of Ownership)

6

for suppliers, TCS (Total Cost to Serve) in the customer perspective, or the theory of ABC/M for the internal and external perspective. In contrast to ABC, where the separability theorem of activities, cost pools and cost driver must exits (Christensen & Demski, 1995), indicators in BSC should correlate. The correlation between measures and perspectives, also gives the opportunity for simulation, to search for optimization/minimization stages, to change substitutions between different input factors or to choose between different outcome measures.

There are two types of performance measures; performance drivers also called lead indicators and output measures or lag indicators. Performance drivers also named as metrics are the ones that tend to be unique for a particular business unit and reflect the uniqueness of the business unit’s strategy (Kaplan & Norton, 1996a). If e.g. the company want to reduce lead time in the internal perspective as part of its strategy, the company should chose those underlying lead indicators that support this part of the strategy e.g., product design, speed of processing, DFM (Design for Manufacturability). The indicators are not a goal in itself but must support outcome drivers and the strategy. BSC will normally also create new processes and new indicators that should be used in combination with existing measures in the strategic planning (Kaplan & Norton (1996, p. 11). Normally, BSC also clarify the company’s strategic goals and at the same time identify critical drivers that should ensure the strategy. However, as known from practice, the learning process from BSC is a major factor. BSC also gives the possibility to make feedback from the strategic level and the strategic learning. At the same time goal congruence, incentive and autonomy must all be considered simultaneously to secure consistency and transparency and prevent unknowingly sub-optimization. Typically, maximum 25 measures are prescribed as sufficient in the generic model which are both consistent and mutually reinforcing (Kaplan & Atkinson, 1998).

3 A-priori Model-considerations The most important objective for BSC is the feedback for learning and improvements for

employees, the communication and information, and learning for executives (Kaplan & Norton, 1996a, b p. 15). This presumes a stringent model-approach build on a-priory knowledge and tests of how different performance indicators are connected, both within a single perspective but also between different perspectives. The final goal is to be able to control the operational performance of the company, and therefore be able to secure the competitions of the company and to increase the value of the company in the long run.

Looking in classical literature for model building, e.g. in operation research accounting (Churchman et al, 1957; Hillier & Lieberman, 1989; Sterman, 2000; Sterman et al., 1997) (or in micro economics e.g. Varian, 1997), accountants may get help to build and secure the validity of the BSC model (here validity as used by Brinberg & McGrath (1988, p. 13) i.e. ‘integrity, character, or quality to be assessed relation to purposes and circumstances’). But also in the area of analytical accounting research (Christensen & Demski, 1995; Magee, 1986; Mattessich, 1984) and in a system oriented approach (Bjørnenak & Olson, 1999), different model approaches can be found.

Using the assumptions above implies that the BSC model should provide a means to explore the consequences of alternative policies and environmental circumstance in such a setting (Sterman et al., 1997). Translating the thoughts of model builders such as Sterman (2000) and Churchman et al., (1957) into the BSC means that developers of BSC must consider the following ingredients:

Problem Articulation/Boundary Selection: I.e. the general problem formulation of the BSC model, i.e. the definition of and operational definition of strategy and vision. This also includes the number and choice of perspectives, the choice of including both performance-drivers (lead indicators) and output measures (lag indicators), and the feedback loops both forward and backwards. It also includes

7

the determination of key variables (depending and independent) and the defintions of each indicator. An important factor is the time horizon or time lag i.e. how far in the future but also how far back in the past lies the roots of the problem. Finally, the dynamic problem (known as the reference modes i.e. literally a set of graphs and other descriptive data showing the development of the problem over time) i.e. what is the historical behaviour of the key concepts and variables.

Formulation of the Dynamic Elements: What are current theories of the problematic behaviour? All areas of BSC are covered of different theories (Chenhall, 2003). The endogenous focus implies the formulation of the dynamic hypothesis that explains the dynamics as endogenous consequences of the feedback structure (loops). Using and developing maps of causal structure based on initial hypothesis, key variables, reference modes and other available data should be used. Besides, strategy maps and other tools such as, model boundary diagrams, subsystem diagrams, stock and flow maps (e.g. employees’ knowledge vs. cost flows), could be used.

Formulation of a Simulation Model: This process should be ‘easy’ when the above mentioned steps are conducted. This step implies the specification of structure, decision rules, estimation of parameters, knowledge of behavioural relationships and setup initial conditions. Determination of time lags or the dynamic aspect between relation and algorithm functions between different perspectives, i.e. the time lag from input effort and to effect in the next perspective (static vs. stochastic). Varian (1997) also points to the determination of constraints. E.g. who are the people that make the choices, what are the constraints they face, how do they interact, what adjusts if the choices aren’t mutually consistant? Kaplan & Norton (2004, p. 366) also argues ‘The collection of linked objectives, targets, initiatives, and budgets for a strategy theme, …, provides a complete stand-alone business case for the piece of the strategy’. However, Kaplan & Norton (2004) also use the term ‘theme’ as the building blocks of the strategy, meaning that each thema of the strategy, should be considered as having ‘its own microeconomic action plan’. Accumulting the microecnomic plans, provide the macroeconomic model of the enterprise (Kaplan & Norton, 2004, p. 369).

Testing: I.e. comparison to the reference modes above and by asking ‘does the model reproduce the problem behavior adequately for the intended purpose’? Does the model behave realistically when stressed by extreme conditions? How does the model behave given uncertainty in parameters, initial conditions, model boundary, and aggregartion? This has also been pointed out by Kaplan & Norton (1996, p. 258) in what they call ‘Management game simulation’, or ‘without quantifying, a strategy objective is simply a passive statement of intent’ (Kaplan & Norton, 2004, p. 365).

There is no cookbook recipe for successful modelling, no procedure you can follow to guarantee a useful model. Modelling is inherently creative, because individual modellers have different styles and approaches (Sterman, 2000, p. 87). It is important to be aware of the structure of BSC model and how the model behave e.g., exponential growth, goal seeking, oscillations, steady-state-solution etc. These behave characteristics depends on feedback loops, stocks and flows, linearity and non-linearities created by the interaction of the physical and institution structure of the model. Besides, traditional concepts such as cost vs. spending, capacity utilization etc. have to be determined (Cooper & Kaplan, 1999).

8

For each relation, the company has to select and determine all variables, i.e. those performance measures that measure the underlying indicators. E.g. lead time may be described as

where represents processing time, inspection time, transportation time, and waiting time. How these variables relate to each other may be a specific problem of its own. The variables may be linear or non-linear and the function may be separable or non-separable. Technically, the BSC-model has many similarities with statistical models, such as simultaneous equation systems (Kmenta, 1986) or the batteries of techniques within the term of ‘Structural Equation Models’ (SEM) (Johnson & Wichern, 1998). Simulation techniques may be used to search for different kinds of solutions (Law & Kelton, 1998).

1 2, 3 4LT = f(x ,x x ,x ) 1x 2x 3x

4x

The time lag or time delay problem must be considered if BSC should be used as a model predicting future financial results. The time delay will be presented both between performance measures and between perspectives. The time delay is defined as a process whose output lags behind its input in some fashion (Sterman, 2000, p. 411). There might be material delay and information delay. Both types assume that there is at least one stock within every delay. In some cases the time delay will be short, but in cases there will be longer time periods, e.g. customer satisfaction and the customer profitability or between a customer outcome and company net profit. The dynamic assumption and the feedback loops will imply that is will be possible to reach a steady-state-solution before an evaluation of the model can be conducted. E.g., if time is expressed by where t represents starting point and t represents ending point between setup and finishing a product, then by setting time to 50 time units, we get a total delay value of 86 time units. For the BSC it is important that we can go both ways, i.e. – top-down from the financial perspective to a lower perspective or from bottom-up to the financial perspective. A top-down example could be by using the BSC in the strategy-focused business planning process, where marked target may be expressed in a specific profitability return (e.g. through ROCE or EVA). From these targets the company can use the BSC-model to simulate the demands for all performance measures down to lower level perspectives.

10.152( ) 1.2 tf t e= 1 2

If this is the case the company must decide whether to use nominal or relative estimates and how the results will be represented by columns, bars, pies, radars, surfaces, histograms or frequency polygons. Finally, the manual of the model should include plan of actions for what should be done if bottleneck problems turns up or if research and development is not able to make the suggested improvements. All variables and performance measures should be consistent, depending and transparent. When the model building step is finished it will be relatively easy to implement or change to a new strategy by changing variables, implement new parameters or changing the time delay dimension.

In the BSC terms like strategy, mission and vision are used. The mission provides the starting point. It defines why the organization exits or how a business unit fits within a broader corporate architecture (Kaplan & Norton, 2001, p. 72).

The vision is a short and inspired syntax sentence that all members of the company should understand. E.g., we want to be the most successful company within our field, or, we want to satisfied customers and make profitability above average, or as mentioned at Rockwater (see Kaplan & Norton 1993,p 135), ‘as our customers’ preferred provider, we shall be the industry leader’.

Although the definition of strategy differs there is general agreement that a strategy describes the general direction in which a company plans to move to attain its goal (Anthony & Govindarajan, 2004, p. 54). Strategy is defined as set of hypotheses about cause and effect, and expressed in a sequence of if-then statements (Kaplan & Norton, 1996b), in relation to what the company want to

9

compete to reach its mission and vision. E.g. if the company wants to improve the relation between the sales personals and an improvement in profit, this may take the following sequences: if we improve the employees education, then the employees have a better understanding of the knowledge of the products, and if the employees have more knowledge about the products, then they are able to target a better sales campaign, and if they can make a better sales campaign, then they are able to improve the profitability of the company. However the word strategy, may on one side have many meanings, and on the other side a company may have many different types of strategies, relating to all kinds of directions e.g. where to place a new factory or which type of supplier the company should choose for its new products. Communicating the last mentioned strategy into a BSC could be a serious mistake for the company (Anthony & Govindarajan, 2004; Porter, 1985). Kaplan & Norton (1996a,b) use and describe different strategies or themes that could be used. E.g. in the financial perspective a revenue growth and mix, improvement and asset utilization could be combined to the classical build, hold, harvest, and divest developed by Boston Consulting Group (1990) or Porters’ (1985) Relative Differentiation Position matrix. Each of these generic strategies demands different financial and non-financial measures in BSC. E.g. if the company chose a build strategy, the company should emphasize new investments, R & D, increase in market shares etc., or if the company implement a hold strategy emphasise should be put on re-investments and maintenance. However, companies often transform one state to another, e.g. from a mature to a rapid growth theme. Using BSC as a model makes this dynamic process easier. Similar templates of performance measures may be developed for other strategies (see e.g. Simons, 1994). However, several case studies have documented that it is possible to combine different measures for different strategies in a single financial metrics in a BSC (e.g. Metro Bank Strategy, see Kaplan & Atkinson, 1999).

It is also assumed that the strategy can and will be communicated to the employees using a BSC and that the measures used in BSC are the measures that the employees can influence.One of the most important aspect of BSC model is to be able to make feedback for learning and improvement, e.g. increase yield, reduce defects, improve cycle time or throughput, or reduce resource consumption (Cooper & Kaplan, 1999).

Figure 1 Relations between mission, vision and strategy

Mission(Why we exist)

Core Values(What we believe in)

Vision(What we want to be)

Strategy(Our game plan)

Inspired by: Kaplan & Norton (2004, p. 33).

10

The company can use either historical data or ex-ante oriented data. The data or the model may be either deterministic or stochastic and data may also be based on time series development. Historical data generated from the company’s MAS (Management Accounting System), can be used to make statistic tests or to find an appropriate probability distribution that matches the distribution of the data. Some relations may be weak while other is strong, some are based on belief while other are based on actual statistical tests. Testing the correlation between financial and non-financial data is important and can be used for better predictions. Software systems like SAP/R3, Oracle, Axapta, and Peoplesoft, normally has a facility for extracting data to be used in separate statistical packages like Excel, SAS, SPSS. In both cases the BSC model should be evaluated on its ability to be able to make prediction and improvements.

4 The ’Case Company’ In our generic System Dynamics model presented in this paper we have used only part of the

case company’s balanced scorecard. Our model is centred on three basic state variables:

Skills

Customer Base

Work in Process

It’s a generic setup that is used by numerous authors (e.g. Sterman 1997; Schöneborn, 2003) and as a consequence of obeying to the System Dynamics model building logic and laws the model structure is almost given in a unique fashion.

The dynamic behaviour of the proposed model is partly determined by the integration process (implicit delays) on these three state variables as well as the various assumptions explicitly defining behavioural reaction and the delays put down in smooth- as well as delay-functions.

The BSC structure that has been reasoned upon in the previous part of this paper is shown in figure 2 below.

Figure 2. BSC model structure

11

Translated into System Dynamics terms combined with the choice of state variables “Skills”, “Customer Base” and “Orders in Process” a corresponding generic structure might look like the one depicted in figure 3 as follows

Figure 4. The documentation of the dynamic model

The dynamics for the three state variables are given by their respective rate-inflow and rate-outflow functions as listed below.

Dynamics of the state variable “Skills”: Skills = INTEG (Skills Learned - Skills Lost) Skills Learned = TE*Staff*NOT Skills Lost = Skills*LR

The “Skills Learned” rate of change variable is modelled as the product of the variable “Staff” with a “training efficiency” coefficient (TE) and a coefficient indicating “the number of trainings” per time unit.

The “Skills Lost” rate of change variable is modelled simply as a exponential decay mechanism with the decay rate (LR) per time unit.

Dynamics of the state variable “Customer Base”: Customer Base = INTEG(Customer Won - Customer Lost) Customer Won = IF THEN ELSE(R+2>0,Customer Base*(R+2),0) Customer Lost = IF THEN ELSE(CL<0,-0.1*Customer Base*CL,0)

The “Customer Won” rate of change variable is modelled as a function of some index measuring (“Recommendations”+2) (R) per time unit so that if R>0 then “Customer Won” = “Customer Base” multiplied by “Recommendations” (R) and else zero. The constant value of 2

12

added to the “Recommendations” accounts for the fact that if “Recommendations” equal zero there are still a flow of customers into the firm’s base of customers from other firms.

The “Customer Lost” rate of change variable is modelled in a similar fashion as a function of some index variable “Customer Loyalty” (CL) per time unit and the mechanism is only active if CL<0 where it is given by “-0.1*Customer Base*CL” that is in fact a positive loss and zero else.

Dynamics of the state variable “Orders in Process”: WIP = INTEG (Orders-Shipments) Orders = OF*Customer Base Shipments = MIN(WIP,AP*Staff)

The “Orders” rate of change variable is modelled as the product between the “Order Frequency” (OF) per time unit and the “Customer Base”.

The “Shipments” rate of change variable is modelled as a minimum function between the variables “Work in Process” (WIP) and a handlings capacity constraint given by the product between the “Actual Productivity” (AP) per time unit and the number of employees (Staff).

As for now the model is expressed only in time units, but if coefficients as well as functional relations are put in proper sizes relative to a desired time frame the model can be assigned a calendar time dimension, which we will do as follows.

The coefficients which must have the time dimension given by definition are essentially, “Number of Trainings”

NOT =1 per month (per Staff) “Loss Rate”

LR = 0.05 per month “Order Frequency”

OF = 2 per month (per customer)

This model is hereby cast in time units equal to months. The actual settings for the variables NOT, LR and OF are clearly in this paper entirely arbitrary and primarily for illustrative purposes, as is the model structure in general. The various functional relations are primarily chosen as simple as possible, just to make a generic point of argument.

Now let us deal with the variables TE, R, CL and AP and the functional assumptions that determine their behaviour

“Training Efficiency” TE = 10/R&D activity

“Recommendations” R= CL + R&D activity

“Customer Loyalty” CL = -ADT/2

“Actual productivity” AP = ASP*Skills/20000

"R&D activity" "R&D activity" = SQRT("R&D budget p.a."/Production Cost)

The “R&D activity” is defined as the square root of the ratio between the per year budget for R&D purposes and the companies running total operative cost.

The logic of the TE function is that when “R&D activity” variable is small (close to zero) it should mean that the relative weight of “R&D” in the overall business activities is small and the “Training Efficiency” should be higher due to the assumption that training on a more static product portfolio is more efficient. If “R&D activity” variable is high it means that the relative weight of “R&D” in the overall business activities is more dominant and the “Training Efficiency” is lowered

13

somewhat due to training in relation to a more varying product portfolio. The “Recommendations” variable is also a function of the “R&D activity” variable which is added linearly to a “Customer Loyalty” variable, so both an increased activity in “R&D activity” as well as an increased “Customer Loyalty” will imply an increase in “Recommendations” and vice versa.

“Actual Productivity” (AP) is defined as “Actual Standard Productivity” (ASP) scaled by the expression “Skills/20000”.

“Actual Standard Productivity” (ASP) is defined as “Standard Productivity” (SP) scaled by the expression 1-(“Time Away”*NOT/”Working Days) per Month”.

Besides the financial variables which are straightforward in their expressions and interpretations, the last vital model component is the “Actual Delivery Time” (ADT). ADT = (Orders + WIP)/(Shipments+1)

It’s a rough indicator of the actual delivery time and should be interpreted more as a psychological reaction measure to delivery problems in this model in connection with the “Customer Loyalty” variable.

Finally some constants of the model will be as follows Table 1

Constants of the Model

CONSTANTS Value Capital Employed 1e+007 Cost per Head 2 Cost per Training 10 Material Cost per Unit 10 Working days per month 20 Unit Price 100 Staff 100 Std. Productivity 100 Time Away 1 Working Days per Month 20

The model operates a couple of performance indicators besides of course “Skills”, “Customer Base” and “Work in Process”

RoCE p.a. = Profit p.a./Capital Employed Staff Utilization = Shipments/(Actual productivity*staff) Profit p.a. = SMOOTH(profit, 12) R&D activity = SQRT(R&D budget p.a./Production Cost)

These indicators obviously cover both the financial as well as the non-financial aspects of the overall systems performance as should be the case if we actually want to take a BSC-view on system performance and control.

Before running a couple of illustrative scenarios to unfold some aspects of the actual dynamics of the given BSC model proposed, it is advisable to get the expectations, as to the dynamic capabilities of the model, set right.

14

The System Dynamics tool for this operation is the so called closed loop analysis that aims at the entanglement of the signs between different elements in the model as well as for complete closed loops embedded in the model. Further analysis can be performed to establish the dominant loop structure, which essentially tells us what kind of dynamics the model will produce.

If we analyse the closed loop structure of the model from the respective stand point of the three state variables in turn we get the following statistics

• Skills - 10 closed loops

• Customer Base - 19 closed loops

• Work in Process - 10 closed loops

This immediately unveils a complicated dynamic structure of the model that would, if the model was to be used for more than illustrative purposes, require a most complicated dominant closed loop analysis. However we will settle for somewhat lesser namely the loop signs embedded in the model extracted to the BSC figure 2 and shown in figure 4.

Figure 4. The extracted model of BSC

Let us isolate a couple of important closed loops in the structure depicted in figure 4. In determining the sign that will characterize the full closed loop, the number of negative partial signs has to be summed. If the sum is even, the loop is positive or self - reinforcing, if the sum is odd then the loop is negative or stabilizing.

(1) Negative (stabilizing) loop structure:

Orders -> Actual Delivery Time -> Customer Loyalty -> Customer Base -> Orders

(2) Positive (reinforcing) loop structure:

Orders -> Production Cost -> R&D Activity -> Training Efficiency -> Skills -> Shipment -> Actual Delivery Time -> Customer Loyalty ->

Customer Base -> Orders

15

(3) Positive (reinforcing) loop structure:

Orders -> Production Cost -> Profit p.a. -> R&D Budget -> R&D Activity -> Training Efficiency -> Skills -> Shipment -> Actual Delivery Time ->

Customer Loyalty -> Customer Base -> Orders (4) Negative (stabilizing) loop structure:

Orders -> Production Cost -> R&D Activity -> Recommendations -> Customer Base -> Orders

(5) Negative (stabilizing) loop structure:

Orders -> Production Cost -> Profit p.a. -> R&D Budget -> R&D Activity -> Recommendations -> Customer Base -> Orders

(6) Negative (stabilizing) loop structure:

Shipment -> Revenue -> Profit p.a. -> R&D Budget -> R&D Activity -> Training Efficiency -> Skills -> Shipment

(7) Negative (stabilizing) loop structure:

Shipment -> Actual Delivery Time -> Customer Loyalty -> Customer Base -> Orders -> Production Cost -> Profit p.a. -> R&D Budget -> R&D Activity ->

Training Efficiency -> Skills -> Shipment (8) Positive (reinforcing) loop structure:

Shipment -> Actual Delivery Time -> Customer Loyalty -> Customer Base ->

Orders -> Production Cost -> R&D Activity -> Training Efficiency -> Skills -> Shipment

In this accounting of loops we have 3 positive self-reinforcing loops and 5 negative stabilizing loops. Now care must be observed in seeking to understand such loop information. Positive self-reinforcing loops are not necessarily destabilizing and negative stabilizing loops are not always stabilizing. How can this make sense?

If a positive self-reinforcing loop is all by itself it will indeed go crazy and result in an exploding time pattern where the variables are approaching plus or minus infinity. But if a positive self-reinforcing loop is embedded in a system of other positive- and/or negative loop, there may be certain interaction in the system that reverses an otherwise unstable development all in all resulting in a controlled outcome. With respect to negative stabilizing loops the dynamic movement towards some fix-point value can be so over exaggerated that the dynamic path results in oscillations around the fix-point but with ever increasing amplitude.

What to expect from actual simulations with the BSC model is consequently somewhat inconclusive given the loop information obtained in loop 1-8. However, the information in loop 1-8 is extremely important when certain behaviour is sought during a simulation session with a System Dynamics model. Anybody with experience in manipulating System Dynamics models in general will know that the abundance of parameters and even more numerous combinations of parameter values almost make a targeted search impossible if no structural knowledge is available – and such information constitutes the information given in loop 1-8.

16

5 Model Simulation Scenarios In the following we analyse three simple scenarios or strategies on profit according to table 2.

Table 2 Strategies and scenarios

Strategy Relevant variables We want to maximize profit within

5 years Skills = 10.000

Work in process = 1.000 Customer base = 2.000

We want to maximize profit within 5 years

Number of training is increased from 1 to 2 (≈an increase in staff of 35)

We want to maximize profit within 5 years

Order frequency increased for 2 to 3

The base model will be parameterized as described in the documentation given above and the initialization values for the state variables “Skills”, “Work in Process” and “Customer Base” will respectively be set to 10000, 1000 and 2000.

Figure 5. The parameterized BSC model

A closer view into the time plot of “Profit p.a.” and “Number of Customer outside the Customer Base” alias potential customers will be as follows, shown in figure 6.

17

Figure 6. First scenario: present situation

2 M

10,000

1 M5,000

00

0 6 12 18 24 30 36 42 48 54 60Time (Month)

"profit p.a." : CurrentNumber of Customer outside the customer base : Current

A quick inspection reveals that the company is in trouble. The company looses its customer base already within half a year and profit starts to decline around year 3. This is serious. By the way, the “Staff Utilization”=100% almost the entire period, and “RoCE p.a.”=8.4% at the end of the simulation run with a peak value of 13.19% one year and four months into the run.

To remedy on this unfortunate development we have several options. We can in this model essentially increase the “Number of Trainings” variable or we can hire more “Staff” or we can of course introduce actions with the intention to change some of the relations in the model, however this last option is of a more subtle kind and will not be considered in this work. The variable “Number of Trainings” will be raised from 1 to 2 in the following experiment.

Figure 7. Second scenario: Number of trainings increased from 1 to 2

2 M

10,000

1 M5,000

00

0 6 12 18 24 30 36 42 48 54 60Time (Month)

"profit p.a." : CurrentNumber of Customer outside the customer base : Current

18

The effect is quite dramatic in that “Number of Customer outside the Customer Base” flattens almost out totally and the “Profit p.a.” reaches a level almost 1.5 the previous level from the base run with “Number of Trainings”=1. Also “RoCE p.a.” shows great improvement in that it settles at a value of close to 16% at the end of the simulation period.

The “Staff” variable has also the potential to remedy on the base run situation and given the improvements in performance by increasing “Number of Trainings” from 1 to 2 it seems sensible to ask for – how big a “Staff” increase will be needed to obtain the same effect as increasing “Number of Trainings” from 1 to 2 ?

The result turns out to be that “Staff” must be increased from 100 to 135 to off-set an increase of “Number of Trainings” from 1 to 2. These two solutions to the base run problem are identical not only with respect to “Number of Customer outside the Customer Base” but also with respect to the “Profit p.a.” as well as the “RoCE p.a.” time patterns.

The “Order Frequency” in the base run was set to a value of 2. If we increase this value to 3 we get the following time plot in figure 8.

Figure 8. Third scenario: Order frequency increased from 2 to 3

2 M

10,000

1 M5,000

00

0 6 12 18 24 30 36 42 48 54 60Time (Month)

"profit p.a." : CurrentNumber of Customer outside the customer base : Current

In this case we see that the “Profit p.a.” is almost unaltered compared to the base run. But the “Number of Customer outside the Customer Base” is showing a development that is worsening compared to the base run and already there we had a problem. We are loosing potential customers faster than is comfortable but it seems not to affect “Profit p.a.” – how can this be?

Because we have increased the “Order Frequency”, fewer customers take the same amount of business to our company per time unit. This also means that we experience an increase in our “Actual Delivery Time” sooner than was the case in the base run, resulting in an earlier loss of customers due to promoted decrease in “Customer Loyalty”.

Clearly the medicine for this problem is essential the same as above.

19

6 Summary and Conclusion One basic points for BSC is to be aware of the purpose. We have formulated a simple strategy

and transformed that into an analytical BSC model base on ‘simple’ assumed cause and effect relationships between various performance measures. Even with such a simple strategy, the BSC model already becomes relatively complicated. As already shown, the most discussed cause-and-effect relationships and the time delay or time lag perspective are two important explicitly assumptions that have to be considered and evaluated in detail. The first scenario shows that with the base mode, that the company is in trouble. The company will loos its customers within a year and profit will decline around year 3. Beside the staff utilization is 100 % almost the entire period. The second scenario illustrates that the company can chose different possibilities. One such possibility is to increase the number of periods of training for the employees from one to two. This will make quit an effect on the possible new customers (will be able to take in new customers) and beside profit will increase. Finally, in the third scenario, we tested how big a staff will be need to obtain the same effect as in scenario two? The result turns out to be that staff must be increased from 100 to 135 of off-set an increase of number of trainings for one to two. Compared to the base scenario, profit and ROCE will be the same over time.

Several new references can be considered, e.g.:

How many perspectives is needed and in what priority?

How do we test different indicators and how do they work together?

How do we define a single strategy that actual works in a BSC model?

How long time before we reach a steady-state-condition in the financial perspective, if we change strategy?

How do we integrate other accounting innovations e.g. ABC, ABM, TQM in the analytical BSC?

Or how do we combine the budget into BSC, and make sure that we can simulate cash flow?

In this paper we have left out a number of more organizational problems e.g. who is responsible, who defines what, how do we involve people, and consideration about the costs of implementing. When the company has developed a BSC-model, it is relatively easy to keep measures up to date or change them or to make new cause-and-effect relationships. This updating is necessary to maintain a model that can be used forward and backwards economic evaluation.

20

References Anthony N. R. & Govindarajan V. (2004). Management Control Systems, USA: Irwin McGraw-Hill. Ax, C. and Bjørnenak, T. (2005). Bundling and Diffusion of Management Accounting Innovations – the Case of the

Balanced Scorecard in Sweden, Management Accounting Research, 16, 1-20. Bjørnenak T. & Olson O., (1999). Unbundling Management Accounting Innovations, Management Accounting

Research, 10: 325-338. Brinberg, D., McGrath J. E. (1988). Vaility and the Research Process, SAGE Publications, USA. Bukh, P.N., Frederiksen, J. and Hegaard, M.W. (2000), ‘Balanced Scorecard på dansk – Ti virksomheders erfaringer’,

Børsens Forlag, Copenhagen. Bukh, P.N., Malmi T. (2005). Re-Examining the Cause- and Effect Principle of the Balanced Scorecard, in ‘Accounting

in Scandinavia – The Northern Light’, ed. S. Jönsson & J. Mourtisen, Liber & Copenhagen Business School Press.

Chenhall, R. H., (2003). Management Control Systems Design Within its Organizational Context: Findings from Contingency-Based Research and Directions for the Future, Accounting, Organizations, and Society, 28: 127–168.

Christensen J. & Demski J. S. (1995). The classical foundations of “modern” costing, Management Accounting Research, 6, pp. 13-32. Christenson, C. (1983). The methodology of positive accounting. The Accounting Review, 58(1), 1–22. Conti, T. (1999). Organizational Self –Assessment, Kluwer Academic Publicshers, London. Cooper, R. & Kaplan, R. S. (1999), ‘The Design of Cost Management Systems’, Prentice Hall. Dale B.G. & J.J. Plunkett (1995). Quality Costing, Chapman & Hall. Drury C., 2002. Management & Cost Accounting, 5th edn. London: Thomson Eccles, R. G. (1991). The Performance Measurement Manifesto, Harvard Business Review, January-February, pp. 131-

137. Epstein, M. and Manzoni, J.F. (1997), ‘The Balanced Scorecard and Tableau de Bord: Translating Strategy into

Action’, Management Accounting, Vol. 79, No. 2, pp. 28-36. Epstein, M. and Manzoni, J.F. (1998), ‘Implementing Corporate Strategy: From Tableaux de Bord to Balanced

Scorecards’, European Management Journal, Vol. 16, No. 2, pp. 190-203. Ernst & Young (1997), ‘Measures that Matter’, pp. 1-16. Garrison R. H. & Noreen E. W, Managerial Accounting (9 ed.), McGraw Hill USA, 2000. Ghalayini A. M., J. S. Noble, T. J. Crowe, (1997). An Integrated Dynamic Performance Measurement System for

Improving Manufacturing Competitiveness, Int. J. Production Economics, Vol. 48, pp 207-225. Hahn, D. (1974). Planungs- und Kontrollrechnung als Führungsinstrument, Betriebswirtschaftlicher Verlag Dr. Th.

Gabler, Wiesbaden, Germany. Heskett J.L., Loverman, T.O., Sasser, G.W., Jr., W.E. and Schlesinger, L.A., (1997). ‘Putting the Service-Profit Chain

to Work’, Harvard Business Review, March-April, pp. 164-174. Hirsch, M. L. Jr., (2002). Advanced Management Accounting, 2.ed. Thomson, UK Hoque, Z. and James, W. (2000), ‘Linking Balanced Scorecard Measures to Size and Market Factors: Impact on

Organizational Performance’, Journal of Management Accounting Research, Vol. 12, pp. 1-17. Horngren, C. T. (1977). Cost Accounting A Managerial Emphasis, 4th ed., Prentice Hall International Ed. London. Horváth, P., Ralf, S. (2004). Why Budgeting Fails: One Management System is not Enough, Harvard Business Report,

September- October. Ittner, C. D., Larcker, D. F. (1998). Innovation in Performance Measurement: Trends and Research Implications,

Journal of Management Accounting Research, 10, pp. 205 – 238. Ittner, C. D., Larcker, D. F. 2001. Assessing Empirical Research in Managerial Accounting: A Value-Based

Management Perspective, Journal of Accounting and Economics, 32: 349 – 410. James, L. R., Mulaik, S. A., Brett, J. M. (1987). Causal Analysis, Assumptions, Modes, and Data, SAGE Publications

UK. Johansson, S.-E., Őstman, L. (1995). Accounting Theory, Integrating Behaviour and Measurement, Pitman Publishing,

UK. Johnson, J. (1986). Econometric Methods, McGraw-Hill Int. Editions, USA Johnson, H. T., Kaplan, R.S. (1987). Relevant Lost: The Rise and Fal of Management Accounting, Cambridge, MA,

Harvard Business School Press. Johnson, R. A., & Wichern D. W. (1998). Applied Multivariate Statistical Analysis, 4th edition, Prentice Hall USA. Joshi, P.L. (2001), ‘The International Diffusion of New Management Accounting Practices: The Case of India’, Journal

of International Accounting, Auditing & Taxation, Vol. 10, pp. 85-109. Kaplan R. S. and Atkinson A. A., 1998. Advanced Management Accounting, USA: Prentice Hall International. Kaplan R. S. D. P. Norton (2004). Strategy Maps. Boston, Harvard Business School Press.

21

Kaplan R. S. D. P. Norton, 2001. Transforming the Balanced Scorecard from Performance Measurement to Strategic Management: Part 1, Accounting Horizons, 15, March 2001: 87-104.

Kaplan R. S. & D.P. Norton, 2001. Transforming the Balanced Scorecard from Performance Measurement to Strategic Management: Part 2, Accounting Horizons, 15, June 2001: 147-160.

Kaplan, R. S., Norton, D. P. (2001), ‘The Strategy-Focused Organization – How Balanced Scorecard Companies Thrive in the New Business Environment’, Harvard Business School Press.

Kaplan, R.S. (1998), ‘Innovation Action Research: Creating New Management Theory and Practice’, Journal of Management Accounting Research, Vol. 10, No. 1, pp. 89-118.

Kaplan, R. S., Norton D. P. (1996a). Linking the Balanced Scorecard to Strategy, California Management Review, Vol 39. No. 1, pp. 53-77.

Kaplan R.S. and Norton D.P. (1996b), ‘The Balanced Scorecard – Translating Strategy into Action’, Harvard Business School Press, Boston.

Kaplan, R.S. and Norton, D.P. (1997), ‘Why does Business Need a Balanced Scorecard?’, Journal of Cost Management, Vol. 11, No. 3, pp. 5-10.

Kaplan, R.S. and Norton, D.P. (1992), ‘The Balanced Scorecard – Measures that Drive Performance’, Harvard Business Review, Vol. 70, No. 1, pp. 71-79.

Kaplan Robert S. (1982). Advanced Management Accounting, Prentice-Hall, USA. Kmenta J. (1986). ‘Elements of Econometrics, 2. ed., Macmillan Publishing Company, New York, USA. Kraus, K., Lind, J. (2002/03). Det Balanserade Styrkortets Roll I Företagens Koncernstyrning, Økonomistyring og

Informatik, Vol. 18., pp. 663-686. Laitinen, E. K., (2004). Towards a Micro Economic Approach of the Balanced Scorecard, Managerial Finance, Vol.

30, No. 6, pp. 1-27. Laitinen, E. K., (2005). Micro Economic Analysis of the Balanced Scorecard: A Case of Nokia Corporation,

International Journal of Productivity and Performance Management, Vol. 54, No. 5/6, pp. 325-339. Law, A.M., Kelton W.D., (1998). Simulation Modelling & Analysis, McGraw-Hill, Second Edition. Lyneis, J. M. (1982). Corporate Planning and Policy Design: A System Dynamic Approach, The MIT Press, London. Maisel, L.S. (1992), ‘Performance measurement: The Balanced Scorecard Approach’, Journal of Cost Management,

Vol. 6, pp. 47-52. Malina, M.A. and Selto, F.H. (2001), ‘Communicating and Controlling Strategy: An Empirical Study of the

Effectiveness of the Balanced Scorecard’, Journal of Management Accounting Research, Vol. 13, pp. 47-90. Magee R. P. (1986). Advanced Managerial Accounting, Harper & Row, New York. Malmi, T. (2001). Balanced Scorecards in Finish Companies: A Research Note. Management Accounting Research,

vol. 12, pp. 207-220. Mattessich R. (1995). Critique of Accounting: Examination of the Foundations and Normative Structure of an Applied

Dicipline, Quorum Books, USA. Mattessich R. (1984). Modern Accounting: Research: History, Survey, and Guide, The Canadian Certified General

Accountants’ Research Foundation. Mattessich R. (1995). Conditional-Normative Accounting Methodology: Incorpating Value Judgment and Means-End

Relations of an Applied Science, Accounting, Organizations and Society, 20, No. 4, 259-284. Mourtisen, J., Bukh, P. N., Larsen, H. T., Johansen, M. R. (2002). Developing and Managing Knowledge through

Intellectual Capital Statements, Journal of Intellectual Capital, Vol. 3(1), pp. 10-29. McNair, C.J., Lynch, R.L. and Cross K.F. (1990). Do Financial and Non-financial Measures Have to Agree?

Management Accounting, November, pp. 28-35. Neely, A., Mills, J, Platts, K., Gregory, M. and Richards, H. (1996), ‘Performance Measurement System Design: Should

Process Based Approaches be Adopted’, International Journal of Production Economics, Vol. 46/47, No. 1, pp. 423-431.

Nielsen, S., Melander, P. and Jakobsen, M. (2002), ‘A Survey of Modern Management Models in Danish Privately-Owned Companies’, Working Paper 02-11, Department of Accounting, Faculty of Business Administration, Aarhus School of Business.

Nielsen S., Sørensen R. (2004). Motives, Diffusion and Utilisation of the Balanced Scorecard in Denmark”, International Journal of Accounting, Auditing and Performance Evaluation, 1.103-124.

Nørreklit, H. (2000), ‘The Balance on The Balanced Scorecard – A Critical Analysis of some of its Assumptions’, Management Accounting Research, Vol. 11, No. 1, pp. 65-88.

Nørreklit, H. (2003), ‘The Balanced Scorecard: What is the Score? A Rhetorical Analysis of the Balanced Scorecard’, Accounting, Organizations and Society, Vol. 28, pp. 591-619.

Bourguignon, A., Malleret, V., Nørreklit, H. (2004). The American Balanced Scorecard versus the French Tableaus de Bord: The Ideological Dimension, Management Accounting Research, 15, pp. 107-134.

22

Olve, N.G., Roy, J. and Wetter, M. (2000), Performance Drivers: A Practical Guide to Using the Balanced Scorecard, Chichester, Wiley.

Olve, N.G., Roy, J. and Wetter, M. (1997), ‘Balanced Scorecard I Svensk Praktik’, Liber Ekonomi, Sweden. Samuelson L. A., (1990). Models of Accounting Information Systems. Stockholm, Studentlitteratur, Sweden. Schöneborn, F. (2003). Linking the Balanced Scorecard to System Danumics, Economic System Reports, No. 1. (ISSN

1431-4126). Shank, J. K., Govindarajan, V. (1993). Strategic Cost Management, New York, The Free Press. Simons R, (1994): How top managers use control systems as levers of strategic renewal, Strategic Management Journal, vol.15,169-189. Solomon, D. (1965). Divisional Performance: Measurement and Control, Homewood, H IL., Dow Jones-Irwin, USA. Sterman, J.D. (2002). Business Dynamics. System Thinking and Modelling for a Complex World. Boston 2000. Sterman, J. D., Repenning, N. P., Kofman F. (1997). Unanticipated Side Effects of Successful Quality Programs:

Exploring a Paradox of Organizational Improvement, Management Science, Vol. 43, April, pp. 503-521. Töpfer, A., Lindstädt, Föster, K. (2002), ‘Balanced ScoreCard: Hoher Nutzen zu langer Einführungszeit’, Controlling,

Heft 2, pp. 79-83. Varian H. R. (1997). How to Build an Economic Model in Your Spare Time, The American Economist, Vol. 41, No. 2,

1997, pp. 3-10. Yin R.K., (1989). Case Study Research Design and Methods, London: Sage Publications. I The authors also wish to thank MBA-students Nicholas Dalgas Fritsche, and Jesper Johnsen Steen, The Aarhus School

of Business, for their help in developing an earlier model in excel of the model.

23

MODEL APPENDIX ADT = Actual Delivery Time WIP = Work in Process AP = Actual productivity ASP = Actual Std. Productivity SP = Std. Productivity TA = Time Away NOT = Number of Trainings WDPM = Working Days per Month CE = Capital Employed CPH = Cost per Head CPT = Cost per Training CL = Customer Loyalty R = Recommendations LR = Loss Rate MC = Material Cost MCPP = Material Cost Per Product OF = Order Freq. SC = Staff Cost TC = Training Cost TE = Training Efficiency PC = Production Cost NCOCB = Number of Customer outside the customer base (01) ADT = (Orders+WIP)/(Shipments+1) (02) AP = ASP*Skills/20000 (03) ASP = SP*(1-(TA*NOT/WDPM)) (04) CE = 1e+007 (05) CPH = 2 (06) CPT = 10 (07) Customer Base = INTEG(Customer Won-Customer Lost,2000) (08) Customer Lost = IF THEN ELSE(CL<0, -0.1*Customer Base*CL,0) (09) CL = -ADT (10) Customer Won = IF THEN ELSE(R+2>0,NCOCB*(R+2),0) (11) FINAL TIME = 60 (12) INITIAL TIME = 0 (13) Loss Rate = 0.05 (14) MC = MCPP*Orders (15) MCPP = 10 (16) NCOCB = MAX(10000-Customer Base,0) (17) NOT = 1 (18) OF = 2 (19) Orders = OF*Customer Base (20) WIP = INTEG(DELAY1(Orders,2,1000)-Shipments,1000) (21) Product Price = 100 (22) Production Cost = MC+SC+TC (23) Profit = Revenue-MC-SC-TC (24) Profit p.a. = SMOOTHI(profit,12,1000) (25) R&D activity = R&D budget p.a./PC (26) R&D budget p.a. = Profit p.a./5 (27) R = CL + R&D activity (28) Revenue = (Product Price)*Shipments (29) Revenue p.a. = SMOOTH(Revenue,12) (30) RoCE = Profit/CE (31) RoCE p.a. = Profit p.a./CE (32) SAVEPER = 1 (33) Shipments = MIN(WIP,AP*staff)

24

(34) Skills = INTEG(skills learned-skills lost,10000) (35) Skills Learned = TE*Staff*NOT (36) Skills Lost = Skills*LR (37) Staff = 100 (38) SC = CPH*Staff (39) Staff Utilization = Shipments/(AP*Staff) (40) SP = 100 (41) TA = 1 (42) TIME STEP = 0.0078125 (43) TC = CPT*NOT*Staff (44) TE = 10/"R&D activity" (45) WDPM = 20 Full list of loops emanating from the “Skills” variable. Loop Number 1 of length 1 Skills skills lost Loop Number 2 of length 9 Skills Actual productivity shipments revenue profit profit p.a. R&D budget p.a. R&D activity training efficiency skills learned Loop Number 3 of length 12 Skills Actual productivity shipments actual delivery time customer loyalty customer lost Customer Base orders material cost Production Cost R&D activity training efficiency skills learned Loop Number 4 of length 13 Skills Actual productivity shipments WIP actual delivery time customer loyalty customer lost Customer Base orders material cost Production Cost R&D activity training efficiency skills learned Loop Number 5 of length 13

25

Skills Actual productivity shipments actual delivery time customer loyalty recommendations customer won Customer Base orders material cost Production Cost R&D activity training efficiency skills learned Loop Number 6 of length 14 Skills Actual productivity shipments WIP actual delivery time customer loyalty recommendations customer won Customer Base orders material cost Production Cost R&D activity training efficiency skills learned Loop Number 7 of length 15 Skills Actual productivity shipments actual delivery time customer loyalty customer lost Customer Base orders material cost Production Cost profit profit p.a. R&D budget p.a. R&D activity training efficiency skills learned Loop Number 8 of length 16 Skills Actual productivity shipments WIP actual delivery time customer loyalty customer lost Customer Base orders material cost Production Cost

26

profit profit p.a. R&D budget p.a. R&D activity training efficiency skills learned Loop Number 9 of length 16 Skills Actual productivity shipments actual delivery time customer loyalty recommendations customer won Customer Base orders material cost Production Cost profit profit p.a. R&D budget p.a. R&D activity training efficiency skills learned Loop Number 10 of length 17 Skills Actual productivity shipments WIP actual delivery time customer loyalty recommendations customer won Customer Base orders material cost Production Cost profit profit p.a. R&D budget p.a. R&D activity training efficiency skills learned

Full list of loops emanating from the “Customer Base” variable. Loop Number 1 of length 1 Customer Base customer lost Loop Number 2 of length 2 Customer Base Number of Customer outside the customer base customer won Loop Number 3 of length 4 Customer Base orders actual delivery time customer loyalty

27

customer lost Loop Number 4 of length 5 Customer Base orders actual delivery time customer loyalty recommendations customer won Loop Number 5 of length 5 Customer Base orders WIP actual delivery time customer loyalty customer lost Loop Number 6 of length 6 Customer Base orders WIP shipments actual delivery time customer loyalty customer lost Loop Number 7 of length 6 Customer Base orders WIP actual delivery time customer loyalty recommendations customer won Loop Number 8 of length 6 Customer Base orders material cost Production Cost R&D activity recommendations customer won Loop Number 9 of length 7 Customer Base orders WIP shipments actual delivery time customer loyalty recommendations customer won Loop Number 10 of length 9 Customer Base orders material cost Production Cost profit profit p.a. R&D budget p.a. R&D activity recommendations customer won

28

Loop Number 11 of length 10 Customer Base orders WIP shipments revenue profit profit p.a. R&D budget p.a. R&D activity recommendations customer won Loop Number 12 of length 12 Customer Base orders material cost Production Cost R&D activity training efficiency skills learned Skills Actual productivity shipments actual delivery time customer loyalty customer lost Loop Number 13 of length 13 Customer Base orders material cost Production Cost R&D activity training efficiency skills learned Skills Actual productivity shipments WIP actual delivery time customer loyalty customer lost Loop Number 14 of length 13 Customer Base orders material cost Production Cost R&D activity training efficiency skills learned Skills Actual productivity shipments actual delivery time customer loyalty recommendations customer won Loop Number 15 of length 14 Customer Base orders

29

material cost Production Cost R&D activity training efficiency skills learned Skills Actual productivity shipments WIP actual delivery time customer loyalty recommendations customer won Loop Number 16 of length 15 Customer Base orders material cost Production Cost profit profit p.a. R&D budget p.a. R&D activity training efficiency skills learned Skills Actual productivity shipments actual delivery time customer loyalty customer lost Loop Number 17 of length 16 Customer Base orders material cost Production Cost profit profit p.a. R&D budget p.a. R&D activity training efficiency skills learned Skills Actual productivity shipments actual delivery time customer loyalty recommendations customer won Loop Number 18 of length 16 Customer Base orders material cost Production Cost profit profit p.a. R&D budget p.a. R&D activity training efficiency skills learned

30

Customer Base

Skills Actual productivity shipments WIP actual delivery time customer loyalty customer lost Loop Number 19 of length 17 Customer Base orders material cost Production Cost profit profit p.a. R&D budget p.a. R&D activity training efficiency skills learned Skills Actual productivity shipments WIP actual delivery time customer loyalty recommendations customer won

Full list of loops emanating from the “Work in Process” variable. Loop Number 1 of length 1 WIP shipments Loop Number 2 of length 5 WIP actual delivery time customer loyalty customer lost Customer Base orders Loop Number 3 of length 6 WIP actual delivery time customer loyalty recommendations customer won Customer Base orders Loop Number 4 of length 6 WIP shipments actual delivery time customer loyalty customer lost

orders Loop Number 5 of length 7 WIP shipments actual delivery time

31

customer loyalty recommendations customer won Customer Base orders Loop Number 6 of length 10 WIP shipments revenue profit profit p.a. R&D budget p.a. R&D activity recommendations customer won Customer Base orders Loop Number 7 of length 13 WIP actual delivery time customer loyalty customer lost Customer Base orders material cost Production Cost R&D activity training efficiency skills learned Skills Actual productivity shipments Loop Number 8 of length 14 WIP actual delivery time customer loyalty recommendations customer won Customer Base orders material cost Production Cost R&D activity training efficiency skills learned Skills Actual productivity shipments Loop Number 9 of length 16 WIP actual delivery time customer loyalty customer lost Customer Base orders material cost Production Cost profit profit p.a.

32

R&D budget p.a. R&D activity training efficiency skills learned Skills Actual productivity shipments Loop Number 10 of length 17 WIP actual delivery time customer loyalty recommendations customer won Customer Base orders material cost Production Cost profit profit p.a. R&D budget p.a. R&D activity training efficiency skills learned Skills Actual productivity shipments