Professional Documents
Culture Documents
Kun Chang Lee
Kun Chang Lee
249
0-8186-1060-3425/93 $03.000 1993 IEEE
competitor (rival)’ s strategic capabilities to:
activity might vary with time. (1) provide insight into the
Simulation can intrinsically spectrum of SP decision problem being
accommodate the time constraint in considered,
its problem-solving mechanism because ( 2 ) suggest a time-variant strategy
many discrete-event simulation as environments change,
techniques, most popular in use ( 3 ) include only relevant data in
today, require time variable to show the SP information systems.
the changes of system states over
time. If this simulation ability is The knowledge-based SP system
combined with the conventional proposed in this paper shows the
knowledge-based SP systems, it can characteristics of intelligent SP
suggest more realistic solutions than simulator. The reasons are as
the conventional knowledge-based SP f01lows:
systems.
(1) The knowledge for monitoring
Most knowledge-based SP systems use changes in the strategic environments
the knowledge only for reasoning can be extracted from CM.
according to the input facts, not for ( 2 ) A set of environmental factors
analyzing the surrounding that seem relevant to a strategic
environments. As is well known, a goal can be identified with the aid
wide variety of environmental factors of the CM knowledge.
surrounding the organization may have (3) The strategic simulation by
a great influence on the SP incorporating those environmental
performance. If knowledge can be factors identified above can be
extracted from environments and is designed and performed to assist in
used for monitoring the change of building effective SP decision under
environments, then the SP systems competitive situations.
will adapt to environments more
effectively, improving its Section 2 addresses six major tasks
250
by section 4 in which an illustrative the formulation of a set of goals for
numerical example is introduced and the organization. Environmental
simulated by using the mechanism analysis includes identification of
proposed in this paper. Finally, environmental factors which seem
concluding remarks end this paper relevant to some target goals.
with some further research topics. Strategy formulation refers to the
25 1
goal structure
1‘ 1’ T
252
+' I-
# of people +
in a city t I migration
into city
modernization 2
amount of garbage
per area sanitation
facilities
# of diseases
per 1000 residents
per area
+
A I
253
F indirectly or directly. They (B and as a knowledge engineering tool for
E) are therefore "irrelevant" in extracting a causal knowledge from
light of outcome F. Determination environments. In line with this
like this is frequently difficult to perception, Kosko [121 has proposed a
make by direct inquiry or fuzzy cognitive map (FCM) in which a
observation, while it becomes quite measure of causation of one factor on
clear when using CM purposely. another lies between -1 and 1. The
FCM can be organized into a matrix
As for the second kind of CM usage, which is capable of not only evolving
one can address the question, "Does a
with time but also producing outputs
CM-based causal knowledge as a whole
(or consequence vectors) for some
work well as we expect ?'I To answer
specific inputs (or stimulus
this, let us assume that one finds
vectors), and therefore allowing
that the value of F is relatively
I' what- if analysis.
'I
254
to use a neural network-based Through strategic control stage,
learning [ 4 l . those results are feedbacked to
previous stages such as goal
3.3 Value in Strategic Planning
formulation, environmental analysis,
and strategy formulation. After
As discussed in section 2, six major
repeating these steps, SP decisions
tasks of SP process indicate that
building effective SP decision would become more timely, alert, and
relevant to the changes of
be much easier with viewing the
surrounding environments. In sum, CM
overall dynamics of SP decision
plays the role of providing a causal
problems through CM in which the
knowledge between relevant factors
causal relationships between various
and aiding the decision maker
kinds of factors are depicted. Since
construct effective SP decisions.
determining which factors should be
included in CM depends on the
4. COGNITIVE MAP KNOWLEDGE-BASED
subjective judgment of decision SIMULATION IN STRATEGIC PLANNING
maker, it is recommendable that
multiple experts are involved in 4.1 Knowledge-Based Simulation
building various CMs for a specific
SP decision problem. After CMs are Today, simulation is one of the most
completed, then they may be merged widely used techniques for management
into one meaningful and appropriate planning [241 because the aim of
CM for the particular SP decision business simulation is to help the
problem. The merging process is well manager analyze complex problems and
described in 1231. The CM can help make decisions in a more scientific,
view the causal structure of SP objective way. Corporate planning
decision problems as a whole which models [171 is a typical example of
leads to identify relevant factors business simulation methods.
that should be more thoroughly Meanwhile, as expert systems
analyzed later. Therefore, those techniques recently emerge as one of
factors identified can be most widely used decision making
incorporated into the SP-related methods in management,
strategic simulation mechanism as knowledge-based simulation technique
exogenous variables. The simulation is now receiving attention.
results are analyzed and used as a Simulation researchers have used
basis for making more efficient SP those knowledge-based simulation
decisions against uncertain future. techniques either to provide powerful
255
decision support functions [221 or to compared to that of conventional
act as intelligent front-ends to expert systems.
provide assistance with
experimentation and output analysis 4.3. Simulations in Strategic
256
I CM Analysis I
I .............................................
1 Causal Knowledge 1- User %
:Interface/
.............................................
Identified
1'
Strategic Simulator
.......................................................
User
,C&
/ \ .......................................................
n Strategy
Implementation
.......................................................
, C & [ Interface
User
/ \ .......................................................
257
a competitor. To clearly understand Environments can be classified into
the experimental procedures, we three categories: uncontrollable,
follow the six simulation steps semi-controllable, and controllable.
mentioned in previous section. Environments such as economic
conditions regardless of domestic or
Step 1. Goal formulation
international, interest rates,
political stability, etc. are
We will use Deal’s dynamic
uncontrollable. Quality control
advertising model [61 as our major
activity, sales price, productivity,
strategic simulation model. The form
etc. are controllable because they
of its objective function is the sum
might become favorable or unfavorable
of two items: (1) profits
according to the efforts of the
accomplished during a fixed SP period
and (2) relative market share at company. Semi-controllable
environments possess two aspects of
terminal planning time which will be
both uncontrollable and controllable
described in the sequel. Accordingly,
the new-entering company is assumed environments. Examples are
competitiveness, brand image, etc.
to set a goal to maximize its profits
Competitiveness represents a general
during a certain SP period. To
competitive position of a firm in the
accomplish this goal, the company
target market, which may be
possesses a strategic goal (i.e.,
operationalized goal) which is to determined by complicated interacting
forces of various exogenous and/or
make its relative market share at the
endogenous factors described so far.
terminal planning time as large as
possible. A large relative market Environments facing the new-entering
company are assumed in this paper to
share will allow a strategically
stable position in the target market, encompass those three kinds above.
Semi-controllable factors:
Step 2. Environmental analysis Competitiveness
Market demand
258
Controllable factors: shown in Figure 5, an initial version
Quality control of CM matrix can be built as in the
Sales price
Productivity following Table 1.
Competitiveness
Sales Price
IT L Quality
I +
AI
Productivity
Conhit ions
Ecodomic +- ll
I
+
Advertisements
259
Description of Notations:
I: Market Share +1: Positive cause-and-effect
11: Competitiveness 0: No cause-and-effect
111: Quality Control -1: Negative cause-and-effect
IV: Sales Price
V: Productivity
VI : Market Demand
VII: Economic Conditions
VIII: Competitor’s Advertisements
260
Stimulus Input Changes of four factors
V: Productivity = + . 3 IV: Sales Price = -.l
VII: Economic Conditions = -.3 V: Productivity = + . 3
VIII: Competitor’s advertisements VII: Economic Conditions = - . 3
= +.5 VIII: Competitor’s Advertisements
= +.2
26 1
the new-entering company. In derivative with respect to time t
specific, strategy concerned is to (that is, dx/dt), tf and to are
appropriately allocate advertising respectively terminal time and
expenditures among the planning initial time, U and v are
period considering the competitor's respectively piecewise continuous
advertising behavior. We found that vector functions called controls or
competitive advertising mode1 strategies, U ( . ) is an algebraic
formulated in differential games [61 terminal condition. @(.I is a
is most suitable for this type of performance index at terminal time tf
strategic simulation because two representing herein a relative market
simulation entities including the share at terminal planning time. A
new-entering company and the saddle point is defined as the pair
* *
competitor must be equally considered (U ,v 1 satisfying the relation
into the simulation dynamics. The
nature of differential games depends J(u*,v) 2 J(u * ,v* J(u,v*
on the system dynamics with which
each of two simulation entities would for arbitrary U E U, v E V. If this
* *
maximize its performance value. relation is satisfied, U and v are
Typical structure of the differential called optimal pure strategies and
* *
game problem can be briefly described J(u ,v 1 is called the value of the
as follows [31: Determine a saddle game or performance index.
point for
In fact, differential game is a
t
J = Q[x(tf),tfl + J f L(x,u,v,t)dt branch of optimal control theories
.I and its applications to a wide
O
variety of MS/OR problems such as
subject to the constraints finance, marketing, production, and
economics, etc. can be found in the
x = f(x,u,v,t),
book by Sethi and Thompson [211.
262
potential market size. Total 7 ) The new-entering company tries
potential market size is therefore t o take advantage of the competitor’s
assumed to be a fixed amount of non-optimal play which can be
sales, M. observed as the negative deviations
2) The advertising media is assumed in the state values.
to be relatively continuous such as Based on these seven assumptions and
radio, television, and newspapers. strategic goal, following set of
3) Each of the market competitors strategic simulation equations may be
possesses a dynamic equation for its built based on Deal’s advertising
strategical behaviors (i.e., equation model [61:
x = f(x,u,v,t) 1 which is similar to where subscript 1 and 2 respectively
each other except state and control denote the new-entering company
variables. (pursuer) and the competitor
4) The sales effectiveness of (evader), both J2 are
J1 and
succeeding advertising expenditures performance index t o be optimized
diminishes as both company’s sales under the given situations, xi(t) is
increase. the sales amount of firm i at time t,
t
M p1 J1 = wlx1(tf)/[xl(tf) + x2(tf)l + J t [clxl(t)- u;(t)l dt
0
subject to
xl(t) = -alxl(t) + blul(t)[M - xl(t) - x2(t)l/M,
x2(t) = - a x (t) + b2u2(t)[M
2 2
- xl(t) - x2(t)l/M.
263
In the performance index, U is Step 5: Strategy evaluation and
1
squared for analytical convenience to implementation
calculate the derivative of the
* *
optimal controls, U and v . It is Strategy evaluation is based on the
noteworthy that performance index of criterion: "Will the proposed
each company follows a goal hierarchy strategy be any good in the future
formulated in step 1. To solve this ?'I. For strategy evaluation, we
kind of differential game problems, obtained expected performance index
we used an adaptive algorithm to take J for each case after SP period,
1
advantage of the competitor's which is summarized in Table 4 .
non-optimal play [181. Let us suppose
Case ?erformance Index
that following sets of parameters
(after SP period)
shown in Table 1 are available to the
I 180.86
new-entering company. M is given I1 157.90
I11 164.08
500.0, xl(to) = 40.0 and x2(t0) =
IV 170.56
100.0, ul(t) = 1.00 and uz(t) = 1.00, v 152.77
and planning period is denoted as t E
VI 133.70
VI I 174.67
[0.0,5.01, being discretized into
integers for analytical simplicity. Table 4. Performance index J
1
After differential game-based
strategy simulation with these The values of performance index in
*
parameters, we obtained U for each Table 4 are obtained from summing two
1
case. terms: profits during SP periods and
-Case a a C C w w
1 2 bl bz 1 2 1 2
I .20 .25 1.10 1.10 .60 .80 15.0 8.00
I1 .25 .25 2.00 1.50 .50 .60 22.5 8.00
I11 .25 .25 2.00 1.50 .SO .60 25.0 5.00
IV .25 .25 1.10 1.10 .50 .60 25.0 8.00
V .20 .25 1.10 1.10 .50 .60 15.0 8.00
VI .15 .25 1.10 1.10 .50 .60 15.0 8.00
VI I .25 .40 1.10 1.10 .50 .60 15.0 8.00
264
relative market share at the terminal
planning time. From the results shown Time Expected Sales Amount
in Table 4, we conclude that case I 1 84.70
will result in most promising 2 69.78
strategy in the future. Then the
* 3 57.34
following strategy (i.e.,ul) related 4 47.26
to case I is implemented for the 5 38.99
new-entering company.
265
II I 1i very difficult for decision makers to
Planning identify those factors relevant to
strategic goals being considered.
Future research topic which seems
.1604 69.70
I 3 I .1918 I 103.00 1 interesting is clarification of the
4 .1578 85.40
role of cognitive map knowledge in
5 .0805 69.70
266
[61 K.R. Deal, " Optimizing strategic planning decision making",
advertising expenditures in a dynamic Proceedings of the 24th Annual Hawaii
duopoly", Operations Research, vol. International Conference on System
27, 1979, 682-692. Sciences, vol. IV, 1991, 173-180.
[71 C. Eden, S. Jones, and D. Sims, [161 A.R. Montazemi and D.W. Conrath,
Thinking in Organizations, Macmillan "The use of cognitive mapping for
Press Ltd., London, England, 1979. information requirements analysis",
MIS Quarterly, March 1986, 45-56.
[81 D.R. Ford and B.J. Schroer, "An
expert manufacturing simulation 1171 T.H. Naylor, Corporate Planning
system", Simulation, vol. 48, no. 5, Models, Addison-Wesley, 1979.
May 1987, 193-200.
[181 S.J. Park and K.C. Lee,
[91 J.H. Grant and W.R. King, "Differential game approach to
"Strategy formulation: analytical and competitive advertising model",
normative models", in D. Schendel and Journal of the Korean Institute of
C.W. Hofer (Eds.): Strategic Industrial Engineers, vo1.12, no.1,
Management: A New View of Business June 1986, 95-105.
Policy and Planning, 1979, Little,
Brown and Company (Inc.), 104-122. 1191 Y.V. Reddy, M.S. Fox, N. Husain,
and M.McRoberts, "The knowledge-based
[lo] J. Haddock, "An expert system simulation systems", IEEE Software,
framework based on a simulation vol. 3, no. 2, March 1986, 26-37.
generator", Simulation, vol. 48, no.
2, Feb. 1987, 45-53. [201 D. Schendel and C.W. Hofer
(Eds.), Strategic Management: A New
[ill N.G. Hall, "Diagnosing problems View of Business Policy and Planning,
with the user interface for a Little, Brown and Company (Inc.),
strategic planning fuzzy DSS", IEEE 1979.
Transactions on Systems, Man, and
Cybernetics, vol.18, no.4, 1988, [211 S.P. Sethi and G.L. Thompson,
638-646. Optimal Control Theory: Applications
to Management Science, Martinus
[121 B. Kosko, "Fuzzy associative Nijhoff Publishing, 1981.
memories", in: A. Kandel (Ed.1, Fuzzy
Expert 'Systems. Read ing : [22l M.L.G. Shaw and B.R. Gaines, "A
Addison-Wesley, 1987. framework for knowledge-based systems
unifying expert systems and
1131 J.C. Larreche and V. Srinivasan, simulation", Proceedings of the
"STRATPORT: A model for the Conference on Intelligent Simulation
evaluation and formulation of Environments, P.A. Luker and H.H.
business portfolio strategies", Adelsberger (Eds.), SCS Simulation
Management Science, vo1.28, no.9, Series 17, no. 1, San Diego, Calif.,
1982, 979-1001. Jan. 1986, 38-43.
261