System dynamics for business strategy: a phased approach

  • Published on
    06-Jun-2016

  • View
    214

  • Download
    0

Embed Size (px)

Transcript

<ul><li><p>System dynamics for business strategy:a phased approachJames M. Lyneisa</p><p>Abstract</p><p>Detailed, calibrated system dynamics models oer an eective tool for supporting businessstrategy. They correspond to the business lines and planning approaches of theorganization, serve as an important check on the adequacy of the model as a representationof the system, provide accurate assessments of the costbenefit tradeos of alternativestrategies, and allow results to be more easily sold to others. However, detailed models alsohave disadvantages: they are complex, can be more dicult to understand, and run the riskof becoming black boxes. This paper discusses a four-phased approach to consultingthat, building from other system dynamics modeling styles (systems thinking and small,insight-based models), oers an eective means of developing detailed models whilesimultaneously educating the client. The approach is illustrated with case examples fromthe credit card and aircraft industries. Copyright *c 1999 John Wiley &amp; Sons, Ltd.</p><p>Syst. Dyn. Rev. 15, 3770, (1999)</p><p>Over the past 20 years, covering more than 50 consulting engagements as theprincipal modeler, I have worked with my colleagues at Pugh-RobertsAssociates to develop a unique modeling style and approach to businessstrategy consulting. This approach is premised on what we see as the primarypurpose of our use of system dynamics modelsto aid managers inestablishing strategy and tactics. We provide tools, analysis of strategic issues,and business advice. The education of managers in the ideas and concepts ofsystem dynamics is a secondary purpose in some, but not all, engagements.Most corporate strategic analyses are episodic, often triggered by a crisis or</p><p>the need to solve an urgent problem. The situation is analyzed, optionsevaluated, and decisions made. The result of these strategic analyses is usuallya decision to take a certain set of actions, for example, to build a new plant, toenter new markets (and/or abandon others), to introduce a new product (and/ordrop others), to engage in a price war, and so on. System dynamics models canplay an important role in understanding the problem and its causes, deter-mining the consequences of alternative courses of action, and testing alterna-tives under dierent scenarios.1</p><p>Our approach to business strategy consulting has evolved over time. In theearly years, the approach was heavy on product and light on process. Likemany in system dynamics, and in management science in general, we took theview that as experts we would solve the clients problem for him, and presenthim with the solution. We gradually recognized that elegant solutions did not</p><p>37</p><p>System Dynamics Review Vol. 15, No. 1, (Spring 1999): 3770 Received April 1998Copyright *c 1999 John Wiley &amp; Sons, Ltd. CCC 0883-7066/99/01003734 $17.50 Accepted June 1998</p><p>aPugh-Roberts Associates, 41 William Linsky Way, Cambridge MA 02142, USA</p><p>James M. Lyneis is aSenior Vice Presidentwith Pugh-RobertsAssociates, a Divisionof PA ConsultingGroup. He holds aPh.D. in BusinessAdministration fromthe University ofMichigan and was anAssistant Professor atMITs Sloan School ofManagement. He hasapplied systemdynamics to problemsof business strategy inthe telecommuni-cations, electricutility, aerospace, andfinancial servicesindustries.</p></li><li><p>necessarily lead to implementation, and our style of consulting changed toinclude increased client involvement (Roberts 1977;Weil 1980). At the same time,our product was evolving tomeet the needs of clients (and to take advantage ofthe increased power of computers). We evolved from the smaller, policy-basedmodels which characterized the M.I.T. approach to more detailed models, alongwith the use of numerical time series data to calibrate thesemodels (Lyneis 1981).During the 1980s, academic research began to focus more on process and</p><p>the use of models in support of business strategy. In his papers StrategySupport Models (1984) and The Feedback View of Business Policy andStrategy (1985), Morecroft describes his view of the use of system dynamics inthe support of business strategy. He proposes that strategy models shouldsupport the development of business strategy in the same way that decisionsupport systems support day-to-day decisionsby providing appropriateinformation and analyses. He argues that the role of the model is to extendmanagement argument and debate rather than provide answers. The model isused to enhance management intuition (mental models) through an iterativeexchange between those mental models and the computer model structure andresults, with appropriate revisions to each until a consensus about strategyemerges. And he describes a two-phase process involving first, BusinessStructure Analysis, and second, Simulation Modeling, to achieve this. Addi-tional research started in the late 1980s focused on more eective ways toinvolve the client in the actual building of the model (see for exampleRichardson and Andersen 1995; Vennix 1996).More recently, process seems to have gained the upper hand over</p><p>product. Following the publication of Peter Senges The Fifth Discipline(1990), the use of systems thinking, archetypes, and organizational learningby consultants, with little if any computer modeling, seems to have growndramatically.2 In some of these uses, the product seems to be an afterthought.This paper describes our approach to business strategy consulting. The</p><p>approach strikes a balance between product and process that producessuccessful change in business strategy. The approach is an extended andrefined version of that described by Morecroft (1985) (and of that of Richmond1997). It contains four phases:</p><p>1. Business Structure Analysis.3</p><p>2. Development of a Small, Insight-Based Model.3. Development of a Detailed, Calibrated Model.4. On-going Strategy Management.</p><p>The approach builds on the strengths of three styles of system dynamicspractice that have evolved over the years: (1) systems thinking and the</p><p>38 System Dynamics Review Volume 15 Number 1 Spring 1999</p></li><li><p>archetypes; (2) small, insight- or policy-based models; and (3) large, decision-oriented models.4 The successful application of this approach is illustratedwith two case studies.</p><p>A phased approach to successful implementation</p><p>I define a successful consulting engagement as one in which the client imple-ments a solution to the problem, and that solution works. To achieve suchsuccess, the consultant must achieve the right mix of product and process.Four outcomes are needed:</p><p>1. An eective understanding, structuring, and analysis of the clients problem.2. Education of the client in the dynamics of their business, so as to obtain</p><p>their active and knowledgeable participation in the structuring of theproblem and in the analysis and interpretation of results (and avoid theblack box syndrome).</p><p>3. Selling of others within the company, but not involved in the project, on therecommended course of action.</p><p>4. For long-term success, providing the means for ongoing learning andplanning with the model (strategy management).</p><p>The four-phased approach discussed below is designed to accomplish theseobjectives.In the ideal situation, we work with a small client team that owns the</p><p>problem and is responsible for making the required decisions. However, this isnot always possible and often we must instead work either with a team ofadvisors to the key decision maker(s) or with a support unit such as strategicplanning (drawing in functional expertise as needed). In these less optimalsituations, while we try to engage the real decision-maker(s) along the way withregular briefings, the need to sell the results of the analysis becomes muchmore important (see Graham andWalker 1998 for a more detailed discussion ofthe consulting process in this type of situation).At the beginning of the project, a project plan and timeline describe the</p><p>phases of work, the objectives of each phase, and planned meeting dates anddeliverables. Expectations for client participation in the project are described.We have found that setting expectations regarding participation and deliver-ables at the beginning makes for a smoother project, and reassures clientsanxious for bottom line results. They know when they will be getting insightand when they will be getting accurate assessments of expected performanceimprovement.</p><p>Lyneis: System Dynamics for Business Strategy 39</p></li><li><p>Phase 1. Business structure analysis</p><p>The purpose of the first phase of analysis is to clearly define the problem ofinterest, the likely causes of that problem, and any constraints that may arise inimplementing a solution. It identifies the performance objectives of theorganization and possible solutionsall to be rigorously tested in later phases.During this phase, the consultant team reviews company documents, the</p><p>business press, and available company data. It interviews company managers,and possibly customers and competitors. It identifies the key drivers ofbusiness performance: what compels customers to buy this product, whatcompels them to buy from one supplier rather than another, what drives theinternal acquisition and allocation of resources. It identifies where majorexternalities aect the business (e.g., the economy, regulations, etc.).Phase 1 builds heavily on the tools and techniques of what is now commonly</p><p>called systems thinking:</p><p>. behavior-over-time graphs (reference modes);</p><p>. causal-loop and mixed causal, stock-flow diagramming;</p><p>. system archetypes;</p><p>. mental simulation.</p><p>Graphs of problematic behavior over time, often with objectives for futureperformance highlighted, focus the model design discussion. Figure 1 illustratesthe behaviors of interest in the two case studies presented below. In the firstcase, the client, a major association of credit-card issuers, had been losingmarket share steadily over a number of years. Would this trend continue? Whatcould be done to turn the situation around? In the second case, the client wasconcerned with the future behavior of commercial jet aircraft orders. Was themarket peak at hand? How low would orders drop, and how long would thenext downturn last?With the problem behavior in mind, causal-loop diagramming or mixed</p><p>stock-flow, causal-loop diagramming is an eective means of conceptualizingthe causeeect structure of the system believed to create the behavior.5 It isalso one of the primary techniques used to facilitate group model-buildingdiscussions. Clients find this approach extremely valuable as a means fordiscussing the dynamics of the business, and for getting everything on thetable. It begins to create the shared understanding of the business problemthat will be essential to achieve agreement and buy-in for the solution. Indeed,as discussed in a later section, some clients feel that this is the most valuablephase of the whole process.</p><p>40 System Dynamics Review Volume 15 Number 1 Spring 1999</p></li><li><p>Fig. 1. Behavior modesof interest in the casestudies</p><p>Case 1 Market Share Trend for Credit Card Company</p><p>Lyneis: System Dynamics for Business Strategy 41</p></li><li><p>In some cases, model diagramming draws on the system archetypes,common generic problem behaviors and structures observed over and overagain in dierent businesses (see for example Senge 1990 and Kim and Lannon1997). We make very little explicit use of archetypes in actually structuring theproblem. The situation at our credit-card client contained several examples ofsuccess to the successful and limits to success archetypes (illustrated inFigures 2 and 3). However, these were noted after the model was developed as ameans of explaining some of the dynamics, rather than as a means ofidentifying structures creating problem behavior.Once a conceptual model is developed using these techniques, mental</p><p>simulation is used to estimate how the structure can create the problembehavior. Mental simulation is a first check on the logic of the analysisdoesthe assessment embodied in the conceptual model seem capable of explainingthe observed problem(s)? Mental simulation is also used to identify the possibleimpact of alternative courses of action.In summary, Phase 1 delivers:</p><p>. an understanding of the nature of the problem in dynamic terms, objectivesfor the business, and constraints imposed on possible solutions;</p><p>. a conceptual model showing how the pieces fit together;</p><p>. a description of how the conceptual model is believed to create the observedproblem;</p><p>. ideas for possible solutions to the problem;</p><p>. lists of issues, uncertainties, and insights or key ideas (nuggets) (If theclient is involved in the process all along, by the end the solutions to theproblem often seem obvious and in hindsight they sometimes wonder whatvalue the project has been, other than in confirming their intuition. It can beextremely important therefore to document insights along the way to serveas a reminder of the learning that has transpired);</p><p>. lists of numerical data and other information that will be needed to supportlater phases of work.</p><p>While Phase 1 and the systems thinking that is a key part of it are a necessarystart, it should not be the end point. Two problems limit its eectiveness insupporting business strategy. First, simple causal diagrams represented bysystem archetypes, while useful pedagogically, take a very narrow view of thesituation (typically, one or two feedback loops). In reality, more factors arelikely to aect performance, and it is therefore dangerous to draw policyconclusions from such a limited view of the system. A more complete repres-entation of the problem considers more feedback eects and distinguisheslevels from rates, but introduces the second problem: research has shown that</p><p>42 System Dynamics Review Volume 15 Number 1 Spring 1999</p></li><li><p>the human mind is incapable of drawing the correct dynamic insights frommental simulations on a system with more than two or three feedback loops(Sterman 1989; Paich and Sterman 1993). In fact, without the rigor and check ofa formal simulation model, a complex causal diagram might be used to argueany number of dierent conclusions. In addition to overcoming these</p><p>Fig. 2. Application ofsuccess tosuccessful archetype</p><p>Lyneis: System Dynamics for Business Strategy 43</p></li><li><p>limitations, as discussed below, formal modeling adds significant value to thedevelopment and implementation of eective business strategies.</p><p>Phase 2. Development of a small, insight-based model</p><p>The power of system dynamics comes from building and analyzing formalcomputer models. This is best done in two steps. In the first, a small, insight-based model is developed to understand the dynamics of the business so as togenerate insights into the direction of actions needed to improve behavior. Thesmall, insight-based model is also the next logical progression in the educationof the client in the methods and techniques of system dynamics modeling. Inthe second quantitative modeling step (Phase 3 below), a more detailed versionof the first model is developed, and is calibrated to historical data. Its purpose</p><p>Fig. 3. Application ofthe limits to succe...</p></li></ul>

Recommended

View more >