ESTIMATION, VALIDATION AND STOCHASTICITY
IN A MODEL OF ORGANIZATIONAL STRATEGY
F. JAVIER DIAZ S. AND ISAAC DYNER
UNIVERSIDAD NACIONAL ‘DE COLOMBIA
A.A. 1027.MEDELLIN. COLOMBIA
ABSTRACT
This paper reports results of research on model building
and validation applied to a complex system immerse in
an environment of permanent change.
A model was built in the context of an organization
operating under a highly centralized structure but with
poor institutional integration and poor information
technology tools. The conveniently articulated strategies
chosen to dilute institutional problems are:
decentralization, leadership and implementation of
information systems.
1. INTRODUCTION
In the search of a methodology capable to represent dynamic
and adaptable systems immerse in uncertain environments, it
is necessary to incorporate dynamic parameters estimation,
prospective validation, and stochastic components.
The main interest then focuses on the internal system
structure and on the possibility of system intervention
by means of political decisions in order to introduce the
planned behavior.
The proposed parameter estimation technique must be dynamic,
prospective and flexible, characteristics which stimulate the
search an incorporation of additional information.
It is advisable to use some appropriate index to measure the
"goodness of fit" between the wanted trajectory and the
simulated one. Besides, the gap must be calculated through
time and corrective measures must be taken.
66 SYSTEM DYNAMICS '93,
If random effects take extreme unfavorable values with high
frequency, this could lead the system to a possible crisis.
Besides, even if this frequency is not high, a sequence of
extreme happenings could lead the system to:an unfortunate
low performance level making impossible its recuperation,
which would produce an imminent catastrophe.
2. THs SYSTEM
According to, Dyner and Diaz (1992) it is necessary to design
strategies on decentralization, leadership and information
systems for highly centralized organizations but with poor
institutional integration and poor information technologies.
These ideas are based on the institutional evaluation
methodology proposed in Suchman (1967).
The mechanisms of decentralization and the recognition of
leadership as the main strategies will imply structural
changes, the redefinition of authority levels, assignment of
appropriate functions to the personnel and modifying
processes and procedures.
The extensive incorporation of information technologies
supports structural changes and the acquisition and use of
information. The suitable information flow through flexible
and reliable channels facilitates the timely decision making.
Ackoff (1981) makes references to the fluctuations between
centralization and decentralization as something that a good
number of organizations suffer.
Jacobsen and House (1992) sketches the characteristics of a
social situation that leads to the charismatic leadership and
describes the cycle that reverts to the beginning situation
that supported this charismatic leadership. Such cycle seems
appropriate to our case of study.
Brooke (1992) recognizes the importance of © central
administration to coordinate local authorities and to solve
possible local disputes. The central level should assume the
task of formulating general policies, to distribute functions
and attributions and to establish agreements with other
entities.
In the following section a general model to represent a
system operating under the described environment is
presented. The. model incorporates the strategies being
examined in order to evaluate their impact in the
organization.
SYSTEM DYNAMICS '93 67
3. THE MODEL
The purposes pursued with the model developed are described
as follows:
- To understand institutional behavior.
- To examine organizational behavior under new polices and
strategies.
- To analyze the manner in which the system structure respond
to external disturbances.
- To establish the effect of unsuccessful projects and the
impact of uncertainties.
- To understand the training effects on the system behavior.
- To examine the incidence of resources on the institution
results.
Figure 1 shows structural relationships of the selected
strategies. The model represents an organizational system of
general nature, however it was conceived for a particular
institution.
As may be observed resources increase according to the level
of leadership and decline with the financing appraisal of
projects. These projects will be financed if their total
costs are feasible with the assigned budget during each
period;. otherwise they will be taken to a waiting list or
will not be considered. Some amount of resources ‘are used for
personnel. training.
Leadership is included.in a positive cycle along with
resources and training. Projects formulation generates two
cycles: A positive one reinforcing the level of leadership
through successful projects, and a negative one due to
unsuccessful projects which increases resistance to change.
The activities related to personnel training strengthen
leadership by two mechanisms, directly as well as through
diminishing resistance to change.
Decentralization directly forms a negative cycle which
tenders to its stabilization. Additionally, this cycle is
connected with the ones corresponding to leadership. Besides,
its growth is reinforced with the increase of training which
reduces resistance.
68 SYSTEM DYNAMICS '93.
a“ RESOURSES
[TRAINING
a
PROJECTS. SUCCESSFUL
FORMULATION 3
ae)
execuT on
_« FAILED
systens woh
Res! wa By bh z
TO CHANGES
DECENTRALIZATION
a =
DUPLICITY FossieiL ity oF
DECENTRALIZATION
puNeT Ce
\ INSTITUTIONAL,
INTEGRATION
FIGURE 1. A Support Model for institutional strategies.
The variable resources form a negative cycle: They make
possible projects formulation and evaluation but reduces
resources availability.
Some susceptible points to produce structural sensibility on
the system will be introduced next.
The positive cycle conformed by leadership - resources -
training. Such. levels and particularly the variable
leadership can strengthen the structural behavior.
Training leads to reduce the resistance levels which at the
same time reinforces leadership and information systems
(increased as well by successful projects). And therefore
strengthen the decentralization cycle.
By suitable decisions making it would possible to reverse the
system declining. tendency. Some of the recommended measures
are listed next.
- To increase the proportion of successful projects. Its
adequate control can be ‘conceived as an institutional
commitment to establish some mechanisms that facilitate
SYSTEM DYNAMICS '93 69
success and promotion of such projects.
- To establish a campaign to increase decentralization and
leadership appraisals. Strengthening institutional presence
on the lower units it is possible to increase its attainments
and reduce resistances.
- Training activities must-have a high priority in terms of
budget assigned.
4. ESTIMATION, VALIDATION AND RESULTS
Parameters are -in a certain way- reduced structures;
modifying them may produce system dynamics changes. A
redefinition of the model could implicate the elimination of
some elementary levels which need to be replace for more
complicate cycles and structures.
If the simulated time horizon is very extensive a lot of
quantities.tend to be variable more than constant values.
Through the modeling process some parameters can become
flexible in order to reproduce the system behavior.
Considering that past does not condition future on a
significant. way the tendential stages will be less probable
every time and the model validation will have to incorporate
the idea of model utility in order to plan and control the
system.
4.1 BASIC SCENARIO (Figure 2)
Supposition: Low decentralization levels, leadership to train
personnel and to develop information systems. Considerable
level of resistance and acceptable probability to obtain
successful projects.
Figure 2 shows the behavior followed by the system.
Leadership starts with a slow declination. It increases as a
result of the impact of the successful medium and long tern
projects. Its appraisal falls with the apparition of
unsuccessful projects and the increment to the resistance to
change. At the end it shows a softly declining tendency with
a small peak during the period in which it stabilizes the
change resistance.
Decentralization increases on an accelerated way during the
time in which there is a wide gap between management
concentration and decentralization. Then its increment
appraisal starts diminishing because the gap begin to close.
70 SYSTEM DYNAMICS '93
The levels of decentralization and leadership show a strong
positive correlation.
FIGURE 2. Basic Scenario
100
90
80
70
60
507
40
30
20
107
0 P
OQ 50 100 150 200-250 300 350 400 450 500 550 600 650 700
TIME,
—™ Leadership —+— Decentralization —*— Training
=> Systems... >= Resistance “ai Resources
The resistance’ to~ change increases with the unsuccessful
projects and declines with the successful ones. Information
systems decline at the beginning becatise of their own
obsolescence. They grow with successful projects and their
increment are soften with unsuccessful projects and with the
increase of the resistance to change.
The training process presents a growing behavior during all
the. simulation horizon, however its growing appraisal is
increasingly reduced and latter presents a.tendency toward
its stabilization.
The level of resources. presents during a-_short. time a rough
fall because of the money needed to finance the new projects
and the low leadership -level.. With the increase of
leadership, this fall presents a slow period of decay and
latter it comes down more rapidly. The level of resources is
not a planning object itself but a consequence from: other
variables. onthe system and its environment.
SYSTEM DYNAMICS '93 nn
4.2 OPTIMIST SCENARIO (Figure 3)
The techniques of prospective and dynamic estimation
considers parameters as functions of time or other variables.
The flexible estimation uses the objectives as internal
functions of the systems results. The controllable
estimation interpret these parameters as planning and control
functions trying to adjust them on a convenient way for the
accomplishment of the proposed goals.
FIGURE. 3. Optimistic Scenario
°
Qo 50 100. 150 200 250 300 350° 400 450 500 550 600 650 700
Time
—™— Leadership —+— Decentralizatio —>*~ Training
—S- Systems —*&= Resistance 7a Resources
It is important to notice how resources decrease gradually,
specially very slowly during ‘the time intervals when
leadership: shows a growing tendency.
The "Goodness of fit" between the values produced by the
model and the objectives of the system is measured by the
index of tendency.’ That: is the percentage of data variation
(the objectives) reproduced by the corresponding variable of
the model (the simulation results).
This index tries.to measure the utility of the model with
respect to knowledge if the established objectives are
feasible or not. The calculation for this exercise give the
72 ; SYSTEM DYNAMICS '93
following results: Leadership, 94.1%; decentralization,
87.6%; information systems, 97.2%; training, 97.0% and
resistance, 81.9%,
4.3 UNCERTAINTIES (Figures 4 and 5).
The basic scenario, Figure 2,°shows a similar behavior to a
deterministic one where random variables. have been removed.
It is supposed that the last one presents an average behavior
of the system under normal conditions of operation.
To illustrate the system behavior under stochastic variations
alternative scenarios were considered with different random
generator seeds. Results show a specially critical behavior
(between the weeks 200 and 400 which approximately correspond
to the years 5 and 9).
Figure 4 shows the presence of a sequence .of unfortunate
consecutive events, which lead to catastrophe in the system.
Figure 5 shows the effects of several consecutive
unsuccessful events. However such behavior is more gradual.
FIGURE 4. Unfortunate Scenario
100
904
80
70
60
50
40-
30
20
104
oO
0 50 100 150 200 250 300 350 400 450 500 550 600 650 700
Time
== Leadership —— Decentralization —*— Training
—=- Systems —>< Resistance i Resources
SYSTEM DYNAMICS '93, 73
FIGURE 5. Unsuccessful Scenario
10 i
0 50 100 150 200 250 300 350 400 450 500 550 600 650 700
Time
=™™— Leadership —+— Decentralization —*— Training
+ Systems >< Resistonce ae Resources
5. CONCLUSIONS
Simulation results show the convenience of planning and
system control, ‘the convenience of making prospective
parameters estimation and permanent model. validation, as well
as to include stochastic components.
Deterministic scenarios. and those “which incorporate
stochasticity® show. satisfactory average results. However,
when a detail study is considered critical behaviors could be
observed. :
The optimist scenario represents the trajectory towards the
system objectives according to the efforts needed to be made
and the efforts needed to control model parameters.
In this work as in system dynamics in general, validation is
focused in the context of scientific research. It is an
inherent relative concept. In relation to its use, the debate
still continues; there are available tools but the problem
depends on the modeler.
74 SYSTEM -DYNAMICS '93
REFERENCES «
ACKOFF, R. 1981. Creating the corporate future. -- New York:
John Wiley.
BROOKE, R. The relationship between central and local
government in the united kingdom, with particular
reference to fiscal relationships.’ Seminario:
competencias y recursos. D.N.P. Bogota, Colombia.
DYNER, I. and F.J. DIAZ S. 1992. A model for management
modernization and institutional improvement. Proceedings
system dynamics 1992
FORRESTER, J.W. 1981. Dindmica industrial. -- 2. ed. Buenos
Aires : El Ateneo.
JACOBSEN, Ch. & R.J HOUSE. 1992. The dynamics of charismatic
+ leadership in organization. Proceedings system dynamics
1992
POPPER, K. 1985. La légica de la investigacidén cientifica.
Madrid : Tecnos.
SYSTEM DYNAMICS '93 75
CC BY-NC-SA 4.0