You are on page 1of 9

chaos.doc http://www.aom.pace.edu/bps/Papers/chaos.

html

Please Note: Those without graphical browsers (such as Mosaic or Netscape) will find it difficult to read
this paper as it contains in-line graphics for many of the formulas and charts.

Over 1000 downloads between 1995 and 1998

downloads since October 7, 1999

Go to Working Papers Index

FROM CHAOS TO COMPLEXITY IN STRATEGIC PLANNING

STEVEN E. PHELAN

School of Commerce

La Trobe University

Bundoora, VIC, 3083

AUSTRALIA

Ph: +61 3 479 1259

Fax: +61 3 479 2909

E-mail: s.phelan@latrobe.edu.au

Presented at the 55th Annual Meeting of the Academy of Management

Vancouver, British Columbia, Canada

August 6-9, 1995

ABSTRACT

Chaotic systems are known to exhibit a sensitive dependence on initial conditions that makes long-range
planning and prediction impossible. However, complex systems can exist in at least four qualitatively
different states (one of which is chaos). Strategic planning is useful in at least two of these four states.

Introduction

Chaos theory has the potential to contribute valuable insights into the nature of complex systems in the
business world. However, care must be taken when reading popular accounts of chaos in the
management literature. As is often the case with the introduction of a new management metaphor,

1 of 9 8/1/00 9:52 PM
chaos.doc http://www.aom.pace.edu/bps/Papers/chaos.html

"chaos" tends to be suddenly seen in almost all managerial systems (Levy, 1994).

Recent research on complex systems in physics and biology has revealed that chaos is only one of four
possible states into which the behaviour of complex systems may be classified. More importantly, this
research has also revealed that systems do not tend to gravitate towards chaotic behaviour but rather
towards an area of complexity between chaos and order.

This paper is organized into three sections. The first section focuses on chaos theory and its implications
for strategic planning. The second section focuses on the nature of complex systems and the four states
in which they can exist, while the final section attempts to describe the implications for strategic
planning of the new theories of complexity.

What is chaos theory?

Chaos theory can be compactly defined as "the qualitative study of unstable aperiodic behaviour in
deterministic nonlinear dynamical systems" (Kellert, 1993:2). A researcher defines a system of interest
and can represent important variables and their interrelationships by a set of equations. The system (or
more technically its equations) is dynamical because the equations are capable of describing changes in
the values of system variables from one point in time to another. Nonlinear terms involve complicated
functions of the system variables such as:

Chaos theorists have discovered that even simple nonlinear sets of equations can produce complex
aperiodic behaviour. The most familiar example being the logistic equation of the form:
where x lies between 0 and 1. This system is deterministic in the sense that no stochastic or chance
elements are involved. Figure 1 depicts the behaviour of this system for varying levels of r.

At values of r <2 iterating over the logistic equation will result in the system stabilising at x=0. Between
r=2 and r=3 the system reaches equilibrium at progressively higher and higher values of x. At around
r=3 the system is seen to bifurcate into two values. The steady state value of x alternates periodically
between two values. As r continues to increase it continues to increase in periodicity, alternating
between 2, then 4, 8 and 16 points. When r is approximately 3.7 another qualitative change occurs - the
system becomes chaotic. The output ranges over a seemingly infinite (non-repeating) range of x values.
This behaviour is represented in Figure 1(d).

2 of 9 8/1/00 9:52 PM
chaos.doc http://www.aom.pace.edu/bps/Papers/chaos.html

Chaotic systems are also unstable, exhibiting a sensitive dependence on initial conditions. This feature
was first recognised by Edward Lorenz in 1963. While running computer simulations on a set of three
nonlinear equations, Lorenz decided to take a shortcut on data entry while attempting to replicate a
previous simulation. Lorenz inadvertently rounded off the initial parameters from four decimal places to
two (ie. 0.8872 became 0.89). Lorenz was intrigued to find that the time paths of the equations diverged
exponentially over time (Gleick, 1987). This sensitivity to initial conditions can also be seen in the
logistic equation described above, where small perturbations in the value of r will lead to large
differences in the values of x as t gets larger.

Lorenz concluded that the sensitivity to initial conditions "... implies that two states differing by
imperceptible amounts may eventually evolve into two considerably different states. If, then, there is any
error whatever in observing the present state - and in any real system such errors seem inevitable - an
acceptable prediction of an instantaneous state in the distant future may well be impossible" (Lorenz,
1963:133)

This gives rise to the notion of the "butterfly effect" where the flapping of a butterfly's wings in Brazil
may make the difference between calm weather and a tornado in Texas next month (Gleick, 1987).

The Lyapunov exponent is a mathematically precise measure of the degree of sensitivity to initial
conditions. The Lyapunov exponent takes the one-dimensional form . If lambda<0 then the initial
differences will converge exponentially. If lambda="0" then the displacements will remain constant over
time, while if lambda>0 small differences will magnify over time. All chaotic systems have a lambda
value that is greater than zero.

Strategic Planning in Chaotic Systems

Attempting to perform strategic planning on a chaotic system exhibiting a sensitive dependence on

3 of 9 8/1/00 9:52 PM
chaos.doc http://www.aom.pace.edu/bps/Papers/chaos.html

initial conditions is a rather futile exercise. As Lorenz has observed, even small errors in specifying
initial conditions can be magnified exponentially as the system evolves.

The use of Lyapunov exponents enables this point to be demonstrated mathematically. In order to
forecast monthly data five years or sixty months in advance with an error of +/- 0.1 then the initial data
would need to be specified with an accuracy of assuming lambda=1. Clearly this level of
accuracy is infeasible. Of course, this also assumes that some deterministic nonlinear set of equations
describing the strategic situation is also available. As long as lambda>0 the system is guaranteed to
demand impossible levels of accuracy in the measurement of initial parameters for quite modest values
of t and future error tolerance levels.

We have discussed in some depth the reasons why scientists believe that long-term planning is very
difficult in aperiodic non-linear deterministic systems. It is not clear that such an analysis could be easily
transferred to models in management and the social sciences.

Initially, the system of interest would have to be specified in terms of nonlinear dynamical equations.
Few researchers in the social sciences have attempted to identify non-linear deterministic behaviours in
their systems of interest. Quantitative research in the social sciences has tended to be statistical and
linear in nature. Of course, it is possible that the appeal of chaos theory may excite an interest in
developing nonlinear models.

The researcher would also need to demonstrate that the system was capable of chaotic behaviour over
some valid region of its parameters. By running digital simulations of the nonlinear systems, researchers
would hope to discover regions of chaos in the models that could be linked with phenomena in the
observed world. Ideally, the Lyapunov exponent could then be calculated and found to be greater than
zero.

Levy (1994) has used a nonlinear simulation of international supply chains to demonstrate that managers
might underestimate the costs of international production; resulting in disruptions and volatility in the
production function. Levy also demonstrated that managers should be able to control the process and
shift the system back into a stable state. Levy's paper represents one of the first attempts to model chaotic
systems in a business management environment.

Notwithstanding Levy's excellent approach to researching chaos, chaos theory has been represented more
on a metaphorical or intuitive level in the recent management and strategy literature. "Intuitively, the
patterns we observe...all point to the importance of chaos in the practice of business management. The
failure to predict...provide[s] further intuitive support" (Stacey, 1991:361). Stacey (1991) then goes on to
say that he can only find two publications that adopt a chaos approach to management and organization.
It would seem that increased turbulence in the business world and the widely-reported accelerating rate
of change is sufficient to label the system as chaotic.

Given the difficulties in empirically demonstrating the applicability of chaos theory to business
management, one would expect the management literature to be relatively silent on the implications of
chaos theory to managerial practice. Interestingly, this is not the case (Cooksey & Gates, 1994; Muller &
Watts, 1993; Peters, 1988; Stacey, 1991; Vinten, 1992). Despite the limited evidence, bold claims are
being made about chaos theory being the "next major breakthrough in management" (Vinten, 1992:1).
One article has even managed to link chaos theory with libertarian and Eastern-oriented philosophies
(Cooksey & Gates, 1994) .

4 of 9 8/1/00 9:52 PM
chaos.doc http://www.aom.pace.edu/bps/Papers/chaos.html

All authors agree that long-term or strategic planning is futile in a chaotic environment: "...if one accepts
the premise that the dynamic of success is chaotic... all forms of long-term planning are completely
ineffective" (Stacey, 1991:188). This is, of course, one of the central features of chaos theory and we
have discussed earlier why this conclusion is perfectly defensible.

In the absence of any ability to plan or control the future, managers are urged to develop an adaptive
stance and a preparedness to react to unexpected and unanticipated events. The term, organizational
learning, is often used to describe the process whereby groups and individuals within the organization
challenge existing mental models of behaviour and learn to rapidly and creatively adapt to a

changing environment (Muller & Watts, 1993; Senge, 1990; Stacey, 1991; Stacey, 1993). In the absence
of foresight, it is clear that a competitive advantage may be gained by effectively adapting to novel and
unpredictable situations faster than the competition.

Is strategic planning then defunct? Should management academics remove the word planning from their
syllabi, and substitute learning or adaptation instead? The success of this argument lies in proving that
business is in fact a complex system in a state of chaos. As we have seen, this proposition is difficult to
test empirically. However, other work in the physical sciences has investigated the properties of complex
systems. The next section argues that while business is a complex system, current research would
indicate that chaos is but one of four qualitative states in which a system may exist and that such systems
would appear not to gravitate towards chaos.

Research on Complex Systems

In a recent working paper, Vriend described several features of complex (adaptive) systems:

"A 'complex system' is a system consisting of a large number of agents that interact with each other in
various ways. Such a system is 'adaptive' if these agents change their actions as a result of the events in
the process of interaction." (Vriend, 1994:1)

Complex adaptive systems have been recognised and studied in biology, physics and economics
(Anderson, Arrow, & Pines, 1988; Kauffman, 1992; Waldrop, 1992). A collection of firms each striving
to achieve a competitive advantage over one another, and adjusting their strategies accordingly, clearly
meets with Vriend's definition. Profit outcomes are jointly determined by the interactions of the firms
(and their strategies), while the ability of firms to alter their actions over time is evidence of adaptive
behaviour. Even a struggle of political coalitions within a firm to determine strategic agendas can be
viewed as a complex adaptive system.

Much of the research on complex systems in biology has focused on the use of cellular automata
(Langton, Taylor, Farmer, & Rassmussen, 1992; Wolfram, 1986) . In these models the state of a single
cell is determined, via a set of rules, by the state(s) of its immediate neighbours (often in 1-, 2- or
3-dimensional space). States are designated as 'on' or 'off' or perhaps 'living' or 'dead'. The Game of Life,
a simple computer program popular among micro-computer buffs in the early to mid 1980s, is a
well-known form of cellular automata. Researchers have been concerned with the emergent behaviour of
the system; the consistent patterns that emerge from the interaction of many elements under different
rule sets.

Researchers have identified four classes of behaviour in cellular automata (Kauffman, 1991; Langton, et
al., 1992; Wolfram, 1986). In Class I models, any combination of live and dead cells quickly approaches

5 of 9 8/1/00 9:52 PM
chaos.doc http://www.aom.pace.edu/bps/Papers/chaos.html

a steady equilibrium state where all cells are 'dead'. Thus, life is extinguished. In Class II models, the
cells develop into static groupings or patterns of live cells, or perhaps groups of cells that oscillate
between fixed states.

Class III models are the opposite of Class I and II models. Class III models are chaotic - the cells
alternate wildly between 'on' and 'off' positions and there are no predictable patterns or stability. The
final set of models, Class IV, are a combination of Class II and III "...coherent structures that propagated,
grew, split apart, and recombined in a wonderfully complex way" (Waldrop, 1992:226). Class IV models
are capable of producing "extended transients" such as the Game of Life's gliders: structures that can
survive and propagate for an arbitrarily long time. To a greater or lesser extent the behaviour of these
extended transients is stable and predictable. However, there is also a degree of uncertainty in Class IV
systems, and extended transients may be destroyed by an interaction with another "entity" or completely
random "mutation".

There is a growing body of evidence in physics and biology, that complex systems tend to evolve to a
state of complexity at the edge of chaos (ie. into Class IV systems). Studies on phenomena as disparate as
sandpiles, earthquakes and artificial life have found that systems move towards complexity or Class IV
behaviour (Bak & Chen, 1991; Kauffman, 1991; Langton, et al., 1992). Class IV behaviour enables
entities in the system to maximise the benefits of stability while retaining a capacity to change.

In competitive situations, such as those found in the biological world, predators and prey must
continuously co-evolve by adjusting to the adaptation of their opponents. It is possible for groups of
species to reach a local optimum in a fitness landscape (much like an extended transient in the Game of
Life may survive for an indefinite number of periods) but each species must be able to adapt very rapidly
if the environment changes in such a way as to make their behaviour sub-optimal. Thus a change in the
environment may cause an 'arms race' or evolutionary spurt until a local optimum is again reached.

In the Game of Life, one mutation in the wrong place can destroy an extended transient almost instantly.
Scientists have explored the probability of change occurring in complex systems at the edge of chaos. A
large number of phenomena seem to follow a power-law distribution (Bak & Chen, 1991). In this
distribution, the number of events of size N occurring is proportional to . Thus for every 1
event of size 1000 (in any scale), 1000 events of size 1 will occur. This has led to the theory of
punctuated equilibrium. Change in any complex system will tend to be small most of the time, but
occasionally a large scale change will occur (see Figure 2.)

6 of 9 8/1/00 9:52 PM
chaos.doc http://www.aom.pace.edu/bps/Papers/chaos.html

Implications for Strategic Planning

Strategic planning in a Class I system is a trivial case as the behaviour tends to 'death' or zero. The
output in Figure 1(a) could be classified as a Class I system. In Class II system such as those depicted in
Figure 1(b) or 1(c), the behaviour of the system is both stable and predictable. The 1950s and 1960s have
been characterised as a period of stable and continuous economic growth and the business world at this
time could be viewed as approximating a Class II system. Strategic planning is also a relatively trivial
exercise in Class II systems. Planning is based on identifying repetitive historical patterns and projecting
them into the future. Such planning may of course be complicated by the presence of measurement error
or complex business cycles that may revolve around eight or more points.

As we have discussed above, Class III or chaotic systems display a sensitivity dependence on initial
conditions that makes accurate predictions of future conditions virtually impossible. Strategic planning is
not possible in Class III systems beyond specifying the broad limits of the behaviour of interest (eg.
snow is unlikely to fall in Texas in the summer, small companies are unlikely to takeover large
companies).

Class IV systems are an interesting case. To the extent that extended transients in Class IV systems
display regular (ie. predictable) behaviour for prolonged periods of time then it would seem that strategic
planning is indeed possible in Class IV systems. However, planners in a Class IV world should attempt a
Stoic outlook on life. Although it can be assumed that extended transients will follow a fixed trajectory,
it should be recognised that at any time an unforeseen interaction with a chaotic element or other
transient in the system has the potential to divert or destroy the elements of the extended transient.

As the level of turbulence in the system increases, the life expectancy of an extended transient
(company/firm/organization) will decline. Thus with increasing turbulence, Class IV "complexity"
systems will move away from a resemblance to Class II "stable" systems and towards the characteristics
of a Class III "chaos" system. Organisations should then direct resources away from trying to predict the
future state of the system and towards learning new adaptive behaviours. However, to the extent that
there are islands of stability in the system, strategic planning will still be a worthwhile exercise.

Is the business environment more like a Class III or IV system? There would appear to be more evidence
to support the view that the business world is a complex system poised on the edge of chaos, rather than
a system in a state of perpetual chaos. Industries appear to go through long periods of incremental
change with the occasional discontinuous change or punctuated equilibrium (Moore, 1993; Tushman &
Anderson, 1986). This would suggest that business systems follow a power-law distribution indicative of
a Class IV system. Indeed research on technological discontinuities has revealed a whole typology of
change ranging from minor process improvements to completely new product classes. Smaller
improvements would seem to be much more numerous than large scale improvements.

On a more anecdotal level, leaders in a wide range of industries tend to hold their position for relatively
long periods of time (often for decades), suggesting that their source of competitive advantage may be
somewhat sustainable, or at least renewable (Chandler, 1990) . Examples of companies that have held a
dominant position in their industry for many years include IBM, Ford, Citibank, Exxon, AT&T,
McDonalds and ICI. This is hardly a picture of unrestrained turbulence and unpredictability that chaos
theorists have claimed exists in today's business world.

There are many factors causing environmental turbulence today including: the globalization of markets;

7 of 9 8/1/00 9:52 PM
chaos.doc http://www.aom.pace.edu/bps/Papers/chaos.html

economic deregulation; the computing and communication revolution; and the increasing importance of
green environmental issues (Blanchard, 1992) . However, it is important to note that not all industries
are affected equally by these changes and different firms will go through these changes at different times
(Mintzberg, 1993).

The degree of turbulence may also follow a power-law distribution with some periods in history having a
much higher level of turbulence than others. Class IV systems can experience a sharp increase in
turbulence during rare periods. The system might even become chaotic for a short period of time before
being pulled back to the edge of chaos. We tend to see the 1950s and 1960s as relatively calm periods of
economic growth, so it is clear that the business system is also capable of becoming less turbulent and
more like a Class II system.

Conclusion

Evidence has been presented that suggests that the business world may be a complex system poised at
the edge of chaos. In this state, strategic planning is still effective because stable conditions may prevail
for protracted periods of time. However, instability is also present in the system and sudden unexpected
changes are always possible. In these situations, organizations should abandon strategic planning in
favour of organizational learning where the firm must learn to rapidly adapt to its changing environment.

Disturbances in the system are likely to follow a power-law distribution in which small shocks to the
system are much more numerous than large shocks. These shocks are just as likely to make the system
less turbulent as more turbulent. Recent management writers on chaos theory are quick to see chaos in
every system and to reject the strategic planning model. Research in the field of complexity has
suggested that this may be premature and that a role for strategic planning still exists. Strategic planning
will continue to have a positive payoff as long as regions of stability remain in the system.

REFERENCES

Anderson, P. W., Arrow, K. J., & Pines, D. (Eds.). 1988. The economy as an evolving complex system.
Redwood City, CA: Addison-Wesley.

Bak, P., & Chen, K. 1991. Self-organized criticality. Scientific American, 264(1): 26-33.

Blanchard, G. 1992. Business strategy for the new Europe. European Trends, 1: 59-64.

Chandler, A. D. 1990. The enduring logic of industrial success. Harvard Business Review(March-Apr):
130-140.

Cooksey, R. W., & Gates, G. R. (1994). HRM: A management science in need of discipline. Paper
presented at the Australian & New Zealand Academy of Management Conference. Wellington, NZ.

Gleick, J. 1987. Chaos: Making a new science. London: Heinemann.

Kauffman, S. A. 1991. Antichaos and adaptation. Scientific American, 265(2): 64-70.

Kauffman, S. A. 1992. Origins of order: Self-organization and selection in evolution. Oxford: Oxford
University Press.

8 of 9 8/1/00 9:52 PM
chaos.doc http://www.aom.pace.edu/bps/Papers/chaos.html

Kellert, S. H. 1993. In the wake of chaos: unpredictable order in dynamical systems. Chicago:
University of Chicago Press.

Langton, C. G., Taylor, C., Farmer, J. D., & Rassmussen, S. (Eds.). 1992.Artificial life II. Redwood
City, CA: Addison-Wesley.

Levy, D. 1994. Chaos theory and strategy: Theory, application and managerial implications. Strategic
Management Journal, 15(Summer): 167-178.

Lorenz, E. 1963. Deterministic nonperiodic flow. Journal of the Atmospheric Sciences, 20: 130-41.

Mintzberg, H. 1993. The pitfalls of strategic planning. California Management Review 36: 32-47.

Moore, J. F. 1993. Predators and prey: A new ecology of competition. Harvard Business
Review(May-June): 75-86.

Muller, J., & Watts, D. 1993. Modelling and muddling: The long route to new organisations. European
Management Journal, 11: 361-366.

Peters, T. 1988. Thriving on chaos. New York: Macmillan.

Senge, P. 1990. The fifth discipline: The art and practice of the learning organization.
Sydney:Random House.

Stacey, R. D. 1991. The chaos frontier. Oxford: Butterworth-Heinemann.

Stacey, R. D. 1993. Strategic management and organisational dynamics. London: Pitman Publishing.

Tushman, M. L., & Anderson, P. 1986. Technological discontinuities and organizational environments.
Administrative Science Quarterly, 31: 439-465.

Vinten, G. 1992. Thriving on chaos: The route to management survival. Management Decision, 30:
22-28.

Vriend, N. J. 1994. Self-organized markets in a decentralized economy (Working Paper No.


94-03-013). Santa Fe Institute.

Waldrop, M. M. 1992. Complexity: The emerging science at the edge of order and chaos. London:
Penguin Books.

Wolfram, S. (Ed.). 1986. Theory and applications of cellular automata. Singapore: World Scientific.

9 of 9 8/1/00 9:52 PM

You might also like