Вы находитесь на странице: 1из 12

Strategic Management Journal, Vol.

15, 167-178(1994)


Department of Management, lJniversity of Massachusetfs- Boston Boston, Massachusetts, U.S.A.

This paper argues that chaos theory provides a useful theorectical framework for understanding the dynamic evolution of industries and the complex interactions among industry actors. It is argued that industries can be conceptualized and modeled as complex, dynamic systems, which exhibit both unpredictability and underlying order. The relevance of chaos theory for strategy is discussed, and a number of managerial implications are suggested. To illustrate the application of chaos theory, a simulation model is presented that depicts the interactions between a manufacturer of computers, its suppliers, and its market. The resuhs of the simulation demonstrate how managers might underestimate the costs of international production. The paper concludes that, by understanding industries as complex systems, managers con improve decision making and search for innovative solutions.

INTRODUCTION One of the enduring problems facing the field of strategicmanagmentis the lack of theoretical tools available to describe and predict the behaviorof firms and industries.For example, even if we know that oligopolisticindustriesare periodsof stabilityalternating likely to experience with periods of intense competition, we do not know when they will occur or what will be the outcome. Similarly, it is almost impossibleto predict the impact of the advent of a new competitor or technology in an industry. The problem is that industries evolvein fundamental a dynamicway over time as a result of complex interactionsamong firms, government, labor, consumers, financial institutions, and other elementsof the environment. Not only does industry structure influence firm behavior, but

Key words: Chaostheory, simulation,international, supply chain

firm behavior in turn can alter the structure of an industry and the contours of competition. Existing theoretical models, however, tend to assume relatively simple linear relationships without feedback.Indeed,manystrategic theories attempt to classify firms and industries and to describe appropriate strategiesfor each class; examplesinclude the Boston ConsultingGroup matrix for resource allocation and Bartlett's of international strategies(Bartlett classification and Ghoshal,1989).Although thesemodelsare basedon recurrent patternsthat we recognizein the real world, there are usually far too many for the modelsto havemuchpredictive exceptions value. Chaostheory, which is the study of nonlinear dynamic systems, promises to be a useful conceptualframework that reconcilesthe essenwith the emerof industries tial unpredictability genceof distinctivepatterns(Cartwright, 1991). Although chaostheory was originally developed Radzicki in the contextof the physicalsciences, (1990) and Butler (1990) amongstothers have

-2095 -r2 t94tB01.67 ccc 0143 1994 by John Wiley & Ltd. Sons, @)


D. Levy

noted that social, ecological, and economic have confoundedmathematicians for years. A systems alsotend to be characterized by nonlinear similar problem afflicts someonewho is trying relationships andcomplex interactions that evolve to calculate the path of an object in the dynamically over time. This recognitionhas led gravitational pull of two or more bodies.While to a surgeof interestin applyingchaostheory to we canusesimpleNewtonian equations to predict a numberof fields,includingecology(Kauffman, the orbits of planetsaroundthe sun with a high l99l), medicine(Goldberger,Rigney and West, degreeof accuracy, the mathematics involvedin 1990) internationalrelations (Mayer-Kressand the case of two or more 'suns'become intractable. Grossman,1989),and economics (Baumol and The problem can be illustratedon a terrestrial Benhabib, 1989; Kelsey, 1988).t Despite the level by observing the motion of a simpletoy, a apparent applicability of chaostheoryto the field metal ball suspended over two or more magnets. of business strategy', there has been surprisingly The ball will tracea series of patternsthat never little work in this area. exactlyrepeatthemselves, and yet aie not totally This paperintroduces readers to chaostheory, random. and discusses its relevance to the socialsciences The paradox here is that the motion of the in general and to aspects of strategy in particular, metal ball is driven by the same Newtonian including planning and forecasting, and the equations as the well understood caseof a single impact of changeon firms and industries.The gravitational attractor.If we knew precisely the application of chaos theoryto a business situation original location, speed, and direction of the is illustrated using a simulation model of an ball, we ought to be able to predictits path with international supply chain. The model, which is a reasonabledegree of accuracy. How is it based on the author's researchinto the supply that deterministic systems can give rise to chain of a California-based computercompany, unpredictability?The explanation is that tiny depicts the complex interactionsbetween the variations in the motion of the ball are magnified firm, its suppliers, andits markets. The simulation every time it swingsby one of the magnets.It resultsillustrate the managerialimplicationsof is a combination of this divergenceand the applyingchaostheory to strategicmanagement. repeatedinteractionsthat give rise to 'chaotic' The model demonstrates how small disruptions behavior. Mathematically,chaotic systemsare to the supply chain interact to make the chain represented by differentialequations that cannot highly volatile, imposingsignificant costson the be solved,so that we are unableto calculate the organization. Although forecasting is very difficult stateof the systemat a specific future time 't'. in the supply chain, distinct patterns emerge At the limit, chaoticsystems can becometruly which are useful for managers. The simulation random. A toss of a coin or the roll of a die also shows that by understanding the supply are, in theory, deterministic systems, but yield chainas a complexdynamicsystem, it is possible more or less random outcomes.Not only is it to identify managerialapproaches that lower the impossibleto toss a coin twice in exactly the cost of operatingthe supplychain. sameway, but on each toss the coin is subject to slightly different air currents, themselves a result of turbulent air flow (Ford, 1983). AN INTRODUCTION TO CHAOS To overcomethe problem of intractabledifferTHEORY ential equations,researchers usuallymodel systems as discrete difference equations, which Chaostheory is the studyof complex,nonlinear, specifywhat the state of the systemwill be at dynamic systems.The field was pioneered by time 't* 1' given the stateof the systemat time Lorenz (1963), who was studying the dynamics 't.' Computer simulationscan then be used to of turbulent flow in fluids. Although we all seehow the systemevolvesover time. recognize the swirlsand vorticesthat characterize One of the major achievements of chaos theory turbulentflow, the complexities of turbulentflow is its ability to demonstrate how a simpleset of deterministic relationships canproducepatterned yet unpredictableoutcomes. Chaotic systems t See also special issues of. Journal of Economic Theory, never return to the same exact state, yet the 40(1), 1986, and Journal of Economic Behavior and Organization,8(3), 1987. outcomesare boundedand createpatternsthat

Chaos Theory and Strategy embody mathematicalconstants(Feigenbaum, 1983).It is the promiseof findinga fundamental order and structurebehind complexeventsthat probablyexplains the greatinterestchaostheory in so many fields. has generated


relationships that reflectvery complexunderlying include the interaction of several potentially chaotic systems;crop prices, for example, are influenced by the interaction of economic and The searchfor a simpleset of weathersystems. equationsto explain complex phenomenamay be a futile attempt to construct grand 'metaChaostheory and the social sciences theory,' a projectthat is rejectedin the postmodsee ern paradigm. The applicationpresentedhere of chaostheory enthusiastically Proponents is pointing to the ubiquity usesa different approach;field study research signsof it everywhere, in the socialworld used to derive a set of relationships among of complex,dynamicsystems and the resemblance betweenpatternsgenerated variables and the influence of external systems a methodsuggested by simulatednonlinear systemsand real time is modeledprobabilistically, series of stock exchangeor commodity prices. by Kelsey(1988). also differ in the chaostheory is Social and physicalsystems From a theoreticalperspective, with the postmodernparadigm,which source of unpredictability. In the physical congruous positivismas it acknowl- world, unpredictability arises due to many deterministic questions and diversityof experience. iterations, nonlinearity, and our inability to edges the complexity has had a profound influ- definestartingconditionswith infinite precision. While postmodernism ence on many areas of social scienceand the In the social world, far less accuracy is and the in definingstartingconditions, it hasbeenneglected by organization possible humanities, and Parker, specificationof the system structure itself is until very recently(Hassard theorists much lessprecise. 1ee3). A final difference is that physical systems of chaos Despiteits attractions, the application naturallaws,whereas by unchanging is still in its infancy, are shaped theory to the socialsciences and there are thosewho think that expectations social systemsare subject to intervention by of Investigations are too high (Baumol and Benhabib, 1989). individualsand organizations. may resemble the economictime seriesby chaos theoristshave Although real life phenomena that relationshipsamong ecopatternsgenerated by simplenonlinearsystems, usually assumed that does not mean that we can easily model nomic actors are fixed over time. In reality, and forecast these phenomena; it is almost methods of stabilizing the economy have impossible to take a set of data and determine changed from the use of the gold standard the system of relationshipsthat generatesit and balanced budgets to Keynesian demand (Butler, 1990). In fact, there is considerable management and, later, to monetaristcontols. debate in the economicsand finance literature Human agency can alter the parametersand about how one tests a data seriesto determine very structures of social systems, and it is ambitiousto think that if it is chaotic or simply subject to random perhapsunrealistically (Brock and Malliaris, 1989; Hsieh, the effects of such intervention can be endoinfluences 1991). Moreover, it is important to recognize genized in chaotic models.2 Nevertheless, that many systemsare not chaotic, and that chaotic models can be used to suggestways systems can transition between chaotic and that people might interveneto achievecertain better goals. The application presented here, for nonchaotic states. Chaostheory is perhaps can reduce seen as an extension of systemstheory (Katz example,showshow management to improve of the supply chain into the realm the volatility and Kahn, 1966; Thompson,1967) of nonlinear dynamics rather than as a total performance. paradigmshift. It is possible that the applicationof chaos has been constrained by 2 To some extent, the distinction between endogenous and theory to socialscience the fact that it has developed in relation to exogenous variables in a model is one of convenience; a physical systems,without taking into account factor that is exogenous in a simple model might become endogenous in a more complex and comprehensive one. fundamentaldifferencesbetween physical and Exogenous factors can be included as random variables in often chaotic systems for modeling purposes (Kelsey. 1988). In the socialworld, outcomes socialscience.


D. Levy
errors would lead us to think that better models and a more accurate specificationof starting conditionswould yield better forecasts,useful for perhapsmonths if not years into the future. Chaos theory suggests otherwise; the payoff in terms of better forecasts of building more complexand more accurate modelsmay be small. Similarly, we cannot learn too much about the future by studyingthe past: if history is the sum of complex and nonlinear interactionsamong peopleand nations,then history doesnot repeat itself. Concerningurban planning, Cartwright (1991)has noted that we have to acknowledge that 'a completeunderstanding of some of the thingswe plan may be beyondall possibility.' The notion that long-termplanningfor chaotic systems is not only difficult but essentially impossible has profound implications for organizations trying to set strategy based on their anticipationof the future. Rather than expend large amountsof resources on forecasting, strategic planning needs to take into account a number of possible scenarios.Moreover, too narrow a focus on a firm's core products and markets might reduce the ability of the organization to adapt and be flexible in the face of change. The proliferation of joint-ventures and the acquisition by large firms of stakes in entrepreneurialenterprisescan perhaps be understoodas attempts to keep a foothold in a number of potential scenariosin the face of uncertaintyand accelerating change. Industries do not reach a stable equilibrium The traditional approachto understanding the influenceof industry structure on firm behavior and competitive outcomes has been derived from microeconomics, with its emphasis on comparative statics and equilibrium.More recent applicationsof game theory have attempted to accountfor interactionsamongsmall numbersof firms (usuallytwo), yielding predictionsabout, for example, investmentsin R&D or plant capacity to seize first-mover advantages.Even the most complexgame theoreticmodels,however, are only considered useful if they predict an equilibrium outcome. By contrast, chaotic systems do not reacha stable equilibrium;indeed, they can never passthrough the sameexactstate more than once. If they did, they would cycle endlesslythrough the same path becausethey


To understandthe relevanceof chaostheory to strategy,we need to conceptualize industriesas complex, dynamic, nonlinear systems. Firms interact with each other and with other actorsin their environment, suchas consumers, labor, the government, and financial institutions. These interactions are strategic in the sense that decisionsby one actor take into accountanticipated reactions by others, and thus reflect a recognition of interdependence.Although interfirm behavior has been modeledformally in economicsand businessstrategy using game theory (Camerer, I99L), these models tend to presumethe emergence of equilibrium and do not adequately reflect industry dynamics. As Porter (1990) emphasizes,the evolution of industries is dynamicand path dependent: corporate (and country-level) capabilities acquired during previous competitive episodesshape the contextfor future competitivebattles.Moreover, the accumulationof competitive advantagecan be self-reinforcing,suggesting at least one way in which industriesare nonlinear. If industries do behave as chaotic systems,a number of implicationsfor strategycan be drawn. Long-term planning is very difficult In chaotic systems, small disturbances multiply over time because of nonlinearrelationships and the dynamic, repetitivenatureof chaoticsystems. As a result,suchsystems are extremelysensitive to initial conditions, which makes forecasting very difficult. This is a problem that has confronted meteorologiststrying to model the weather: the fundamentalproblem is trying to use finite measurementsin an infinite world. A related problem is that as systemsevolve dynamically,they are subject to myriad small random(or perhas chaotic)influences that cannot be incorporated into the model. Formulatinga long-termplan is clearly a key strategictask facing any organization.People involved in planning,whether in business, economics,or someother area,have alwaysknown just models,that forecasts that modelsare always are uncertain,and that uncertaintygrows over time. Nevertheless. our conventional understanding of linearmodelsand the influence of random

Chaos Theory and Strategy are driven by deterministic relationships.The implicationis that industriesdo not 'settle down' and any apparentstability, for examplein pricing patterns,is likely to be shortlived. or investment Chaos theory also suggeststhat changesin industrystructures canbe endogenous. Corporate decisionsto enter or exit the market, or to developnew technologies, alter the very structure of the industry, which in turn influencesfuture firm behavior. One of the most provocativeand controversialelements of chaos theory is that chaotic systemscan spontaneously self-organize into more complexstructures (Allen, 1988).The notion has been applied to biological evolution (Laszlo, 1987) as well as to economicsystems (Mosekilde and Rasmussen, 1986). In the context of business strategy,the conceptcould potentially be applied to the evolution of complexorganizational relationshipssuch as long-term contracts and technical cooperation with suppliers, and hybrid forms of organizationalcontrol such as joint ventures.Chaostheory suggests that new, more complex organizationalforms will appear more frequently than if they were simply the result of random mutations. Dramatic changecan occur unexpectedly Traditional paradigms of economics and strategy, which are generally based upon assumptions of linear relationships and the use of comparative static analysis, lead to the conclusion that small changesin parametersshould lead to correspondingly small changes in the equilibrium outcome.Chaostheoryforcesusto reconsider this conclusion.Large fluctuationscan be generated internally by deterministicchaoticsystems. Models of population growth based on the logistic differenceequation illustrate how sudden,large changesin population levels can arise from the dynamics of the system rather than from the influenceof external shocks(Radzicki, 1990).3 Similarly, if economic systemsare chaotic then we do not need to search for wars or natural


disasters to accountfor economicdepressions or a crashin the stock market. The size of fluctuations from one period to the next in chaotic systems have a characteristic probability distribution (Bak and Chen, 1991). Under this distribution, large fluctuationsoccur more frequently than under the normal distribution, suggesting that managers might underestimate the potential for large changes in industry conditions or competitors' behavior. Small exogenousdisturbances to chaotic systemscan alsocauseunexpectedly largechanges. The implication for business strategyis that the entry of one new competitoror the development of a seeminglyminor technologycan have a substantial impacton competition in an industry. An examplethat comes to mind is the way Dell's mail order strategy in the personal computer industry forced other companiesto reduce their prices and reexaminetheir traditional high-cost salesand servicechannels. Short-term forecastsand predictions of patterns can be made Although the unpredictability and instabilityof chaotic systemshas been emphasized, there is also a surprising degree of order in chaotic systems. Short-term forecasting is possible becausein a deterministicsystem, given the conditions at time 't,' we can calculate the conditions at time 't+1.' A carefullyconstructed simulation model of a complex system with accuratelyspecifiedstarting conditionscan yield useful forecasts at least for severaltime periods. Weather forecastsbased on sophisticated computer modelsusingmeasurements from thousands of points around the globe do provide useful forecasts for a few days,which is usuallysufficient for purposessuch as hurricane warnings. If we imaginethat strategicdecisions in companies are made on a monthly or even annual cycle, then industry simulation models might be able to make useful predictionsover a time horizon of severalmonths or possiblyyears. Another feature of chaotic systems that lends them a degreeof order is that they are bounded; outcomevariablessuchpricing or investments in new capacityfluctuatewithin certain boundsthat are determined by the structure of the system and its parameters but not its initial conditions. In the context of businesstrategythese bounds

3 The logistic difference equation has the form: P,*t : P, * R* (1-P,) P, a fraction between 0 and 1, representsthe population level as a proportion of the maximum carrying capacityof the environment.R is the growth rate from one cycle to the growth is constrained next. Population by the factor 1-P,, which can be understoodas a resourceconstraint.


D. Levy

might be set by feedback loops such as the betweencountries, betweenfirms in an industry, entrance of new firms or antitrust action by or even betweendepartments in a firm. the government in response to monopolistic conditions. Although we cannot forecastthe precisestate Guidelinesare neededto cope with complexity of a chaotic systemin the longer term, chaotic and uncertainty systems trace repetitive patterns which often 'Strategy'can refer to a set of guidelinesthat provide useful information. According to Rad- influence decisions and behavior. It is the zicki (1990),deterministicchaos'is characterized complexityof strategicinteractions, whether in by self-sustained oscillationswhose period and chess, soccer, or in business,that makes it amplitude are nonrepetitive and unpredictable, essentialto adopt simplifying strategies to guide yet generated by a system devoidof randomness.' decisions; even the most powerful computersare For example, while we do not know exactly unable to track all possiblemovesand counterwhere or when tornadoes and hurricanes will movesin a chessgame. General Electric'swell strike, we do know what conditions lead to their known strategyof being numberone or number occurrence, when and where they are most two in every industry in which it participatesis frequent,and their likely paths.In a similarway, a simple example of a guideline which may be we know that oligopolistic industriestend to generally useful but is not always optimal in alternatebetweenperiodsof intensecompetition every situation. We need general guidelines andperiods of morecooperative behavior, though becauseit is impossibleto specify the optimal we do not know when an industry will make the courseof action for every possiblescenario. transitionfrom one state to another.To give a It is important to distinguish the guidelines third example,we know that the economycycles and patternsof behavior that constitutestrategy through recessions and booms,though we cannot from the underlying rules of the game. In a predict very well the depth or duration of a game of chess,for example, knowing the rules particular recession (Butler, 199). Observing for playing the game does not necessarily give patterns is especiallyuseful if we can associate one insights into strategiesfor successful play. different phases of the systemwith other charac- One can only learn thesestrategies after experiteristics; for example,thereis a strongrelationship encing the complexitiesof interactions on the betweenbusiness cyclesand other variablessuch chessboard. Indeed, because of the complexity as demand, interest rates, the availability of of strategicinteractions,one does not always credit, vendor lead times, and the tightnessof know why a particular strategyis successful. the labor market. While the complexity of industry systems An intriguing aspectof the patternstraced by dictates the needfor broadstrategies, the dynamic chaotic systemsis that they are independentof natureof chaoticsystems mandates that strategies scale;in other words, similarpatternsare traced adapt.As industrystructures evolveand competiby a system whateverhorizon is usedto view it. tors changetheir strategies, a firm clearly needs Economic time series often appear to display to changeits own guidelinesand decisionrules. this property.Stockprices,for example, displaya The problem here is that there is no simple way remarkably similarpatternwhetherone observes of deriving optimal strategies for a given system. daily changes over 1 year or minute-by-minute Indeed, in a complexsystemthe best strategies changesover a day. These imagesof patterns might achievegoals indirectly and even appear within patterns are termed fractals when they counter-intuitive. The bestwayto improvequality are generatedby chaotic systems. In the natural is not necessarily to check every product several world, fractals canbe found in manyphenomena, times: it may be to improve labor relationsand from the shapeof coastlines to ice crystals.The thus gain labor's cooperation in finding waysto implications for business strategy are not entirely reduce defects. IBM's decision in 1981 to clear. One interpretation is that previousexperi- let other 'clone' manufacturers use the DOS encesin an industry are likely to recur on a operatingsystemfor personalcomputers helped much larger scale. A second interpretationis competitors but also indirectly helped IBM to that similar patterns of behavior might be build market share by creating the industry expected whether one examines competition standard.

Chaos Theory and Strategy In order to understandindirect or counterintuitive meansto an end, a systemneedsto be I understood as a whole. If systems are very
complex, then simulation models might pronl helpful in finding the most effective way to achieve a goal. The example discussed later in


J cor{poNENT Il{\tBNtclRY

I | t

! l

i vs. TARGETINv.

I i I

this paper illustrates how the best way to cut inventory in a supplychainmight be to reduce
disruptions to the chain rather than shorten lead


I nnooucrroN l<-----




pRoDucrron <-----!

I i, I v ! Thesupplychainasacomp|exdynamicsystem_qIqI!{- >

i I !


The simulation of an international supply chain how chaos theory can be applied demonstrates to the understanding of a real managerialissue. The exampleis drawn from the author'sresearch into the costs of coordinating the international supply chain of a personal computer company called California Computer Technology(CCT).4 Following Eisenhardt (1989), a case approach was usedto build and test theory in an iterative manner. of the The researchled to a conceptualization supply chain as a complex, dynamic, nonlinear system.The systemis subject to externaldisruptions, and the stagesof the chain are linked by flows of goods and information, with time lags and feedback mechanisms.The complexity of interactionsalong the supply chain is such that one cannot easily predict how the system will operateunder variousconditions,but a computer model of these processes can simulate the outcome (Lant and Mezias, 1990; Morecroft 1984).Figure 1 is a simplified representation of CCT's supply chain, showingin schematicform the flows of goods and information within the model. Solid lines represent flows of goods, dotted lines flows of information. In reality, supply chainsare often much more complex than this. CCT, for example, has hundredsof vendors,three productionsites,and in many countries. distributors and warehouses Corporate headquarter functions interact with vendors, the field sales organization, and the productionsites.Nevertheless, the diagramdoes
a The name of the company has been disguised proprietary information.


a l


l l i

chain Figure1. A Modelof CCT'ssupply

of the supplychain. Materials capturethe essence move along the chain in one direction, gaining in value at eachstage.Information is exchanged both directions among the organizationsalong the chain. This simple representationis very useful in analyzing the potential sources of coordination costs in a supply chain and the stages of the separating impact of geographically chain. It is also valuable as a tool for designing strategiesthat improve the performance of a supplychain. There are two important dimensionsto this system, uncertainty and time relationships. Rather than performing as a stable, ready state system, each stage of the chain is potentially subject to disruptions, or 'shocks.' Demand fluctuatesin an unpredictableway, production problemscan affect output, and suppliersdo not always deliver on time. When demand and production are rising, delivery and production problems are more likely. As a result of the uncertainty at each stage, flows of materials and finished systemsfluctuate in volume, and inventoriesneed to be adjustedto cope with the

to prorecr uncertainty, The linkages themselves are also subject to disruption. Shipments and communi-


D. Levy

cation can be delayed, and information can be INVENTORY TO DEMANDRATIOFOR lOO MONTHS misunderstood. A secondimportant dimensionof the supply o chain system is the time relationship among Z . = the stages. As a result of the time lags in u, o communication, production, and distribution,a F disruptionto one elementgenerates a sequence z O 2 of changesin other parts of the system. For = example, demand fluctuationscausechangesin F F ' I salesforecasts, productionschedules, and order z to vendors. Disruptionsoriginatingin any one z part of the system,in effect, propagateforwards and backwardsalong the chain. Disruptionscan too interact; for example, a production problem t ,,"r,"iir"r, could occur in a month when demand was 2. Simulated inventory unexpectedlyhigh, causingsome demand to go Figure levels, product shipped from Singapore to U.S. unmet. A number of researchers investigatingaspects of the supply chain have recently turned to simulation models,most of which attemptto find outcomes.CCT's managers did not expect this cost-minimizing solutions usinglinearor nonlinear volatility, because the strategicdecisionto source programming(e.g., Breitman and Lucas, 1987; from Singaporewas taken using cost estimates Cohenand Lee, 1989;Hodder and Jucker,1985; that assumed a stablesupplychain. In fact, the Hodder and Dincer, 1986). These models do instability of the chain imposed substantial not, however,deal adequately with uncertainty unexpectedcosts on CCT, primarily related to in a dynamic,multiperiodsetting. the expense of using air-freight to expedite The simulationmodel developedfor this study shipments, the opportunitycostof lost sales, and is describedin more detail in the Appendix and the costof holdingexcess inventories. In addition, in Levy (1992). The model assumesa set of CCT incurred expenses relating to the communidecisionrules and linkagesamong the stagesof cation and managerialtime neededto manage the supply chain, which are used to determine the unstable supply chain. These costs were the production plan and other variables each all underestimated becausemanagersdid not month. Each stageof the supplychainis subject appreciate the impact of complex interactions to random fluctuations,and the chain evolvesin alongthe supplychain, and tendedto treat each a dynamic fashion from month to month. disruptionas a one-timeevent. The simulation does reveal some patterns within the fluctuating inventory levels. Peak Resultsand implications inventory levelsare reached,on average,every Figure 2 shows simulated inventory levels over 5 months,thoughthe numberof monthsbetween a period of lffi months based on a version of peaks varies from 21 months; the system is the model representingproduction in Singapore clearly aperiodic. Moreover, there is a relationfor the U.S. market, which entails 30 days ship between the average time between peak shippingtime. Inventory levels are expressed as inventory levels and shippingtime: when proa proportion of monthly demand,and negative duction is available for sale the same month valuesindicatethat demandcannotbe met from (representing productionin the U.S. for the U.S. inventorythat month. market), averagetime betweenpeak inventory The most obvious feature of the graph is levels falls to around 4 months. Note also that the volatility of inventory levels. These large inventorylevelsare lessvolatile and that peaks fluctuationsillustrate well how relatively small are lower, as would be expectedwhen delivery disruptionsto the supply chain can interact with times are shorter. organizational decisionprocesses and lead times As well as illustrating the volatility of the in the systemto producelarge and unpredictable supplychain and its associated costs,the model
UJ o

Chaos Theory and Strategy

2.e 2.6 2.4 o.1t


tr o

1.9 ..8

= uJ


o.o. ! lr,
=l o.07 u0.08 j

F 1.2 6 1
= o'o

t O z

o.B o..r 0.2

-o.z -0.4 -o.3 -o.8



$ f

o.or o.o,


f, o.o"



Figure 3.

Simulated inventory levels, product made and sold in U.S.

Figure 4.

Impact of demand stability on unfulfilled demands

can be used to guide decisions concerning production location, sourcing, and optimum inventory levels. Used for this purpose, the simulation model demonstrateshow complex systems need to be understoodas a whole, and how goals can be achievedthrough indirect and nonobvious means.For example,the simulation model enablesthe cost of offshore sourcing to be estimated in termsof the incremental inventory neededto maintain demandfulfillment at some specified level. It was estimated that in order to maintainan average level of 95 percentdemand fulfillment when sourcingfrom Singaporerather than California for the U.S. market, average systeminventory levels would have to increase by more than 2 months of sales. The underlying order in the supply chain systemcan be glimpsedin Figure 4. The X-axis showsthe value of a parameterrepresenting the standard deviation of the monthly percentage change in demand, a measure of demand instability. The range of values was chosen to reflect the instability observed for CCT's products. The Y-axis showsthe averageproportion of demand that could not be fulfilled over 100 iterationsof a 36-monthperiod. There appears to be a thresholdbeneathwhich demand instability does not have a significant effect;in this region,the system doesnot exhibit chaos. Oncethe instabilityparameter approaches 0.1, the proportionof demandunfulfilledbegins to rise rapidly but smoothly and exceeds10

percent of demand for products with the most unstabledemand. While the simulation modelillustrates the costs and difficultiesof an unstablesupply chain, it also suggests approaches to solving these problems. The simulation model could be used to determineoptimal inventorylevelsfor different productsand components, and to identify those which need to be manufactured locally, based on the level of volatility associated with them. Although CCT's managers had always been products awarethat unstable shouldbe produced locally,they tendedto underestimate thesecosts. The simulationprovided a tool to analyzemore preciselywhich productswere stable enoughfor offshore manufacture. Another approach,using the insight gained from Figure 4 above, would be for managers to attempt to improve the accuracy of sales forecasting in order to reducethe costof offshore manufacture.Similarly, managerscould try to reducedisruptionsto the supplychainfrom other sources,by working with suppliers to improve quality and reducelead times, and by reducing the occurrence of internal productionproblems. Volatility can also be reducedby interveningat the boundaries of the system to change its structure.CCT, for example,has participated in the widely observed trend toward fewer suppliers. management could simUsing thesetechniques, plify and stabilize possiblymakingit the system, nonchaotic. It should be noted that the approachesto


D. Levy successivemonths. In order to model the stochastic natureof the supplychain,a simulation package called 'RISK'was used,which allowsa variety of probability distributionsto be assigned to each cell of the spreadsheet.Actual data and decision criteria from CCT were used to determine the structure of the model and the rangeof valuesto be usedfor variousparameters. For example, monthly bookings data revealed that the level of bookingseach month could be modeledby taking the previousmonth'sdemand plus a percentagechange that was a normally distributedrandom variablewith mean zero. Three sets of input parameterswere used for the model. The first set representsthe level of disruptionsaffectingdemand,supplierdeliveries, and production. The second set representsthe target levels of system and component inventories. The standardvaluesfor thesewere set at half a month's salesfor systems, excludingany production in transit, and 1 week's for inventory parameters group The third of input components. representsthe impact of distance on shipping time for finishedgoods, and of different vendor lead times. The main output variables of the systemwere the levelsof systemand component inventoriesand the level of demand fulfillment. the model simulatesa Using these parameters, 36-monthtime period. The model beginsat time zero with a nominal level of demand and production of 100 units, but thesevaluesevolve over time. The simulation packageenablesthe to be recalculated a specified number spreadsheet of times. On eachiteration,the entire 36-month spreadsheetis recalculatedwith a new set of random numbers. A large number of iterations can thus be used to build up a probability distribution for these output variables and to calculate a mean (expected)value. The results presented here were obtainedusingone hundred iterations for each simulation. Each simulation of lffi iterationswas run using a different set of A list of variablesrecalculated input parameters. on a monthly basisis given in the Appendix. were capThree main performancemeasures tured as output variables.Average system and average component inventory levels over 100 iterationsof the 36-monthperiod were expressed in terms of months' sales.Demand fulfillment was measuredby summingthe total number of units of demandwhich could not be met due to inventory,and dividing this total by inadequate

managingcomplexsystems describedabove con(Womack, stitutekey elements of leanproduction Jones, and Roos, 1990). Lean production can thus be conceptualized as a way to simplify and reduce the varianceof complex dynamic supply chain systems, making their behavior more predictable.Indeed, this researchsuggests that, contrary to the prevailing notion that lean production methodsconstraininternational production (Hoffman and Kaplinsky, 1988; Jones and Womack, 1985), lean production could actually facilitate international operations by reducingvolatility along the supply chain.

CONCLUSIONS Chaos theory is a promising framework that for the dynamic evolution of industries accounts and the complex interactions among industry actors. By conceptualizing industries as chaotic systems,a number of managerial implications can be developed. Long-term forecasting is almost impossible for chaotic systems, and dramatic change can occur unexpectedly;as a result, flexibility and adaptiveness are essential for organizations to survive.Nevertheless, chaotic systems exhibit a degreeof order, enablingshortterm forecasting to be undertakenand underlying patterns can be discerned. Chaos theory also points to the importanceof developingguidelines and decisionrules to cope with complexity, and of searchingfor nonobviousand indirect means to achievinggoals. The simulation model presentedhere demonstratesthat chaostheory haspracticalapplication to issuesof businessstrategy. The simulation illustrates how managementcan underestimate the impact of disruptions to an international supplychain,generating substantial unanticipated costs. It also demonstrateshow management might intervene to reduce the volatility of the supply chain and improve its performance, by reducingthe extent of disruptionsand changing the structureof the supply chain system.

APPENDIX The supply chain was simulatedusing a spreadsheet model, with columns representing the variables in the system, and rows representing

Chaos Theory and Strategy

total demandto give a ratio indicatingunfulfilled demand. The effect of distanceon shipping times and ENDING of different vendor lead times was modeled by SYSTEM INV: using different versionsof the basic model. The simulationsusedfor this paper used a version in which production in Singaporeis available for sale in the U.S. the following month (i.e., 30 daysto ship and clear customs)and vendor lead TARGET times are 60 days. SYSTEM INV:



ACTUAL DEMAND: Demand for systems each month was equal to the previous month's demand plus a random percentage change. eachmonth Salesof systems wereequalto demandunless by lack of invenconstrained tory. The best sales forecast for the next month was the previous month's demand, as no trend was built into demand fluctuations.
The level of component (or material) inventory each month was the level the previous month plus deliveries from vendors less whatduction.






depending on the version, lessa random percentage. Systeminventory at the end of each month was equal to the inventory the previous month less sales plus production the same or the previous month, depending on the model version. The target level of system inventorywas adjustedeach month to equal a proportion of the currentsales forecast. The production plan for the following month was based on the sales forecast, adjusted for the difference betweenactual and planned systeminventory. System production each month was equal to the production plan of the previous month, lessa random percentage, and constrained by the availabilityof material inventory. Unfulfilled demandequalled less demand monthly monthlv sales.


REFERENCES Allen, P. M. (1988). 'Dynamicsmodels of evolving systeryl" SystemDynamics Review' 4' Summer' American, 2&(l), pp. 4G53. Scientific Bartlett, C. A. and S. Ghoshal (1989). Managing : The Transnationat Solution.Harvard across Borders SchoolPress,Boston,MA. Business Baumol, W. and J. Benhabib(1989).'Chaos:Significance, mechanism, and economic applications'. 3, pp. 77-105. Journal of Economic Perspectives, 'PLANETS:A Breitman,R. L. andJ. M. Lucas(1987). planning', Interfaces, modelingsystemfor business r7(1), pp. 94-106. Brock, W. and A. Malliaris (1989). Dffirential Equations,Stabilityand Chaosin Dynamic Systems. North-Holland, New York. approachto Butler, A. (1990). 'A methodological missingthe point?', Federal chaos:Are economists Reserve Bank of St. Louis, 72(13),pp. 3G48. need Camerer,C. F. (1991).'Does strategyresearch game theory?', Strategic Management Journal, 12, Issue,pp. 137-153. Winter Special

,self-organized in pro- ""flt.t"tiJtilbn"r (1991). ever was consumed criticality',

The target level of component inventorywas adjusted each month to equal a proportion of the current sales forecast. Orders to vendors were basedon the salesforecast, the production schedulefor the following month, and a comparison of actual with target component inventory levels.
Deliveries each month equalled orders placed 1. or previously, 2 months





D. Levy
Kelsey, D. (1988). 'The economics of chaosor the chaosof economics',Oxford Economic Papers,4{1, pp. 1-31. Lant, T. K. and S. J. Mezias (1990). 'Managing discontinuous change:A simulationstudyof organizational learning and entrepreneurship',Strategic Management Journal, Summer Special Issue, ll, pp. 147-179. Laszlo, E. (1987). Evolution: The Grand Synthesis. Shambhala, Boston,MA. Levy, D. (1992). 'The costs of coordinating international production'. DBA Dissertation,Harvard University Graduate School of BusinessAdministration. Lorenz, E. N. (March 1993). 'Deterministicnonperiodic flow', Journal of the AtmosphericSciences, 20, pp. 13f141. (1989).'Chaosin Mayer-Kress, G. and S. Grossman the international arms race', Nature, 337, pp. 70r-704. Morecroft, J. D. (1984). 'Strategysupport models', StrategicManagement Journal, 5(3), pp. 215-229. Mosekilde,E. and S. Rasmussen (1986). 'Technical economicsuccession and the economiclong wave', EuropeanJournal of OperationalResearch, 25, pp. 27-38. Porter, M. E. (1990). The Competitive Advantageof Nations. Free Press,New York. Radzicki,M. J. (1990).'Institutional dynamics, deterministicchaos,and self-organizing systems', Journal of Economic Issues, 24(l), pp. 57-102. Sterman,J. D. (1989). 'Deterministicchaos in an experimental economic system', Journal of Economic Behaviorand Organization,12(29), pp. 1-28. Thompson, J. D. (1967). Organizationsin Action. McGraw-Hill, New York. Womack, J. P., D. T. Jones and D. Roos (1990). The Machine that Changed the World. Rawson Macmillan.New York.

Cartwright,T. J. (1991).Planningand chaostheory', Journal of the American Planning Association, 57(l), pp. 44-56. Cohen, M. A. and H. L. Lee (1989). 'Resource deployment analysisof global manufacturingand distribution networks', Journal of Manufacturing and OperationsManagement, 2, pp.81-104. Eisenhardt,K. M. (1989). 'Building theories from case study research', Academy of Management Review,l4(4), pp. 532-550. Feigenbaum, M. J. (May 1983).Universalbehavior in nonlinearsystems',Physica,7, pp. IG39. Ford, J. (April 1983).'How random is a coin toss?' Physics Today,36, pp. 4M7. Goldberger,A. L., D. R. Rigney and B. J. West (February1990).'Chaosand fractalsin physiology', ScientificAmerican, 263, pp. 4A9. Hassard,J. and M. Parker (eds.)(1993). Postmodernism and Organizations. Sage,ThousandOaks, CA. Hodder, J. E. and J. V. Jucker (1985).'International plant location under price and exchange rate uncertainty', Engineering Costs and Production Economics,9, pp. 225129. 'A multifactor Hodder,J. E. and M. C. Dincer (1986). model for internationalfacility location and financing under uncertainty', Computersand Operations Research, f3(5), pp. 601-609. Hoffman, K. and R. Kaplinsky (1988).Driving Force. WestviewPress,Boulder, CO. Hsieh, D. A. (1991).'Chaosand nonlineardynamics: Application to financial markets', Journal of Finance,46(5), pp. 1839-1877. Jones,D. T. and J. P. Womack (1985).'Developing countries and the future of the automobileindustry', World Development, l3(3), pp. 393407. Kauffman,S. A. (1991).'Antichaosand adaptation', Scientific American, 265(2),pp. 78-84. Katz, D. and R. L. Kahn (1966).TheSocialPsychology of Organizations. John Wiley and Sons,Chichester.