FORESIGHT AND SYSTEM DYNAMICS IN
MODELLING THE DYNAMICS OF ORGANIZATIONAL
KNOWLEDGE
Paulo Edy Nakamura
Brazilian Aeronautical Command
pnakamura@ gmail.com
Ricardo Matos Chaim
University of Brasilia, Brasilia, Brazil
mchaim@ gmail.com
Abstract
The loss of organizational knowledge due to the departure of skilled staff generates negative impacts on
both an organization's productivity and its functional capacity to understand the key strategies and actions
necessary to achieve stated goals. In order to identify, understand and model the factors likely to be
involved in the causes and impacts of the loss of organizational knowledge, this article examines the
dynamics of knowledge processes in a Project Management Office of the Brazilian Aeronautical
Command, which is ible for the impl ion of complex projects in Brazil.
Presented here is a dynamic model based on concepts from Nonaka and Takeuchi’s “Theory of
Organizational Knowledge Creation”, and on Michel Godet’s methodology for identifying key variables
within Strategic Foresight Studies. The model enables the simulation of diverse scenarios in which are
represented the consequences of changes in key variables of influence. The work concludes with a
discussion of concerns and opinions that could facilitate both the formulation of policy and active
intervention aimed at minimizing the unwanted impacts that result from the loss of an abstract entity
known as organizational knowledge.
Keywords: System Dynamics. Scenarios. Organizational Knowledge. Knowledge Management.
Dynamic Models. Complex Projects. Strategic Foresight.
1 Introduction
Organizational en are ically complex by nature, and in them individual knowledge and
experience, when properly valued, organized and controlled, are critical to success and survival. As such,
leading organizations stand out by being well-informed and having information and knowledge that allow
them to adopt the best strategies for achieving their goals (CHOO, 2006). They also have leaders,
managers and decision makers who can recognize and value personal knowledge, individual experience
and training as a key resource called organizational knowledge.
Organizational knowledge, whether explicit or implicit, is an abstract entity inherent to human activity
which neither transfers nor can be shared with ease or spontaneity (ALVARENGA NETO, 2008). As
with other types! of knowledge, the management of such an asset is no easy task. Unlike the
of physical (such as materials, personnel and finances), which can be measured
and
defined.
A good example here is Brazilian Air Force Command (COMAER), which oversees the Coordinating
is subjective and lacks indicators that are explicit or well-
Committee of the Combat Aircraft Program (COPAC), an agency responsible for coordinating activities
related to the development, modernization and acquisition of aeronautical systems for use in the Brazilian
Air Force (FAB). Organizational knowledge permeates the activities of these groups, most of which
resides in the minds of group members and leaders, both past and present. This implicit knowledge,
experience and accumulated memories are combined with explicit knowledge, policies, guidelines and
manuals to achieve the goals and objectives of each organizational unit.
As these activities are carried out in projects of medium and long-term duration, that is, from 5-15 years,
the organization inevitably passes through organizational and structural changes, especially with the
departure of highly qualified and experienced staff. This natural process results in the loss of significant
portions of acquired experiences and accumulated knowledge, important assets for the direction and
continuation of administrative proceedings within COPAC.
In previous work (BONOTTO, 2005; TAVARES, 2008), it was found that the continuous turnover of
qualified | gradually d acci experience and organizational memory. This in tum
harms all of project ially during the jation, signing and of
contract execution.
Hence arises the need for an understanding of the factors involved in the management of accumulated
knowledge, how such processes occur, and of how they are influenced by human and organizational
factors, factors that can also contribute to the formation and maintenance of knowledge organizations of
this nature.
2 Development
Understanding the mechanisms that enable the formation, flow and loss of organizational knowledge
enables the adoption of policies and strategies that can ensure the survival and competitiveness of an
organization in a turbulent environment. Thus, this study addresses the following topics:
a) Analyze the principles and concepts of Knowledge M (KM) in an
ible for ing complex ical projects (COPAC);
b) Identify and present a dynamic model to represent the turnover of specialized personnel and its
relation to the formation of organizational knowledge; and
&
Identify and i igate the di ics present in izational change in terms of personnel
policy and the efforts employed to create organizational knowledge.
+ Popular, philosophical, religious or scientific (LAKATOS and MARCONI, 1986, p.20).
2.1 Knowledge Management
Different authors define Knowledge Management in different contexts and for different purposes; they
conceptualize the term in different ways based on different epistemologies. This paper will therefore
incorporate an interdisciplinary perspective that can capture the dynamic nature of Knowledge
Management. This allows Knowledge Management to be defined as:
“The effective leaming processes associated with exploration, exploitation and sharing of human
knowledge (tacit and explicit) that uses appropriate technology and cultural environments to
enhance an organization’ s intellectual capital and performance.”
Source: Jashapara, A., (2004), Knowledge Management: An Integrated Approach, Prentice Hall, Harlow,
Essex: page 12.
According to Nonaka and Takeuchi (1995) there are two types of knowledge: tacit (implicit) and explicit.
Tacit knowledge consists of one's experiences, mental models, beliefs and opinions. Explicit knowledge
is the kind of knowledge that can be defined and shared easily through information technology.
With these two concepts, Nonaka and Takeuchi have developed a dynamic model (Figure 1), known as
SECI (Socializati Externalizati Combination and Internalizati This framework was in tum
based on studies in the Theory of Organizational Knowledge Creation (NONAKA and TAKEUCHI,
1995).
Dialogue
Socialization Externalization
Linking
Bold (Tear G Explicit
ens Knowledge
Internalization Combination
Learning by doing
Figure 1. Nonaka and Takeuchi SECI model (1995)
These two authors argue that knowledge is initially created by individuals and becomes organizational
knowledge through a process represented above (see Figure 1).
"A spiral emerges when the interaction between tacit and explicit knowledge is elevated
dynamically from a lower ontological level to higher levels"
Source: Nonaka and Takeuchi (1995), The knowledge-creating company. New York: Oxford
University Press: p. 57.
This spiral is created by the four modes of knowledge conversion through which knowledge is converted
from one knowledge type to another. These modes of knowledge conversion include socialization (from
tacit to tacit knowledge), externalization (from tacit to explicit knowledge), combination (from explicit to
explicit knowledge), and internalization (from explicit to tacit knowledge).
The following literature review compares this theory with other schools of thought, particularly as
expressed in the work of Davenport and Prusak (1998). With a more pragmatic view, these authors
concentrate their studies on organizational processes that allow the generation, codification and transfer
of knowledge.
For this research, the SECI model was chosen because it views the social as dynamic by nature
(NONAKA and KONNO, 2000). Humans play an important role and organizational networks generate
knowledge through social processes of sharing, exploring and creating tacit knowledge (stories,
experiences and concepts) and explicit knowledge (raw data, bank and organized data reports).
Analysis of works by Tavares (2008) and Bonotto (2005), which describe the vulnerabilities of
duction and project in COPAC, reveals similarities between the SECI model
and the processes in which knowledge is transferred through this organization.
knowledge
Through these works, an important issue was identified: the loss of organizational knowledge due to the
departure of skilled staff. Here could be seen a physical resource (humans) influencing an intangible
entity (knowledge), and this is a common problem for many knowledge-based organizations.
Peter Massingham (2008) also explores this issue in a case study within the Australian Ministry of
Defense. In this work, the author simulates the consequences of the departure of key personnel and
discusses the results of this research in the context of existing literature. He concludes by presenting a
preliminary conceptual model able to identify and isolate the causes and nature of the risks involved.
Following suit, the dynamic model here was developed to understand the factors driving the management
of accumulated Organizational Knowledge, and how they are influenced by human and organizational
activity.
2.2 The System Dynamic Approach
Due to the dynamic nature of the SECI model, with the presence of cause and effect relationships and
feedback loops, a system dynamics approach was chosen to model such behavior.
System dynamics is an aspect of systems theory, used for understanding the dynamics of complex
systems. It is also a computer-aided approach which can be applied to dynamic problems arising in
complex social, managerial, economic and ecological systems.
The basis of the method is the recognition that the structure of any system — the many circular,
interlocking, imes time-delayed relationships among its — is often just as important in
determining its behavior as the individual components themselves (http://www.systemdynamics.org).
The approach begins with defining problems dynamically, proceeds through mapping and modeling
stages, then arrives at steps for building confidence in the model and its policy implications.
Relevant here is the work of Sterman (2000), who defines a sequence of steps that allow the dynamic
modeling of the proposed problem. This methodology can be applied in different areas of knowledge
(STERMAN, 2000), especially those in which it is difficult to predict the behavior of key variables, and
where the system is relatively complex.
The necessary steps may be laid out as:
1 Problem Articulation;
2 Formulation of Dynamics Hypotesis;
3 Structure of the simulation;
4 Tests;
5 Project and evaluation of policies.
Articulating the problem is the first and the most important step in modeling (STERMAN, 2000), and it
and the model was accordingly designed to explore policies that could mitigate the loss of organizational
I ledge in a Project M: Office due to the departure of skilled staff .
Within this first step, it is also necessary to characterize the problem dynamically, that is, as a pattern of
behavior unfolding over time, which shows how the problem arose and how it might evolve in the future.
So, a reference mode, literally a set of graphs and other descriptive data showing the development of the
problem over time, should be developed (STERMAN, 2000).
In this stage, several project managers from COPAC were involved to identify the time horizon, define
those variables and concepts considered important for understanding the problem and designing policies
to address it. A Casual Loop Diagram (CLD) was built to represent both hypothetical causes of a system's
dynamics and the mental models of individuals and teams.
2.3 The MICMAC method
To build a first Causal Loop Diagram, it is necessary to select a range of elements that the modeler
considers relevant (STERMAN, 2000). This selection process, which identifies key variables over less
important ones, involves qualitative data collection.
In 1997, James Ritchie-Dunham demonstrated a method to summarize the findings from multiple Casual
Loop Diagrams as analytical tools, and to check their rigor. Part of this methodology integrates the
MICMAC approach (in the French acronym: Matrix of Crossed Impact Multiplications Applied to a
Classification) (GODET, 2006).
The MICMAC method is proposed by Michel Godet within “Strategic Foresight”. In effect, MICMAC
does allow participants to see the influence that one variable exercises on another through a third, a
fourth, even a fifth. Highlighted are the determining factors (or 'main determinants’) of the situation under
investigation. The input variables and results or output variables help participants understand the
organization and structuring of the system under the microscope (GODET, 2001).
Godet proposes the use of a matrix to evaluate and ranks the key factors agents in a system. This involves
a method consisting of three basic steps:
1 Defining relevant variables;
2 Specifying relationships between variables;
3 Identifying key variables in the group.
The cross-impact matrix, called Matrix of Direct Influences (MDI) (Figure 2), can be constructed as a
table, where each entry in the matrix is a variable, each variable in row i, on trade-in, moves to the
variable in column j. Direct influence I(i,j) can be rated as follows: No influence (0), Weak influence (1),
Moderate influence (2) and Strong influence (3). Once the grading is done for all entries, it is possible to
observe the sum of each row, which will indicate the influence level. The sum of the columns,
meanwhile, shows the level of dependency.
To facilitate the process, the software MICMAC version 6.1.2 - 2003/2004, developed by the French
Computer Innovation Institute 3IE, was used. This program can also perform the matrix multiplication
applied to graded direct impacts many times. This process produces a Matrix of Indirect Influences (MII)
which allows the study of the diffusion of impacts along the chain of influence and also feedback. We can
therefore rank the variables according to their influence on the system as a whole.
Oa “oe se 3
Ze
SAE AE LAE 8
Variable 1 uaa) hiaz) f(a) faa) tag)
Variable 2 (2a) ahs) fea) han
Variable 3 (3.3) 1(32) (aay 1(3,4) en}
Variable 4 (4,4) naa) aay (a (4)
Variable i 1 (n3) 1 (n2) I(n,3) (nn)
Dependence Yen Diver Pras 14) ran
Figure 2. Cross impact matrix - Matrix of Direct Influences (MDI).
As the method suggests, the first goal of such an analysis is to stimulate group thinking and to initiate
reflection on ‘counter-intuitive’ aspects of system behavior. It should be remembered that there is not
simply one "official" reading of Micmac results. Numerous meetings were necessary to obtain both an
homogeneous list of variables and also a common of respective ings. Unstructured
and non-directed interviews were conducted with members and former members of COPAC, and these
aimed to identify and understand the factors that influence the flow and knowledge management in the
organization.
At this point, a limitation of the method became clear. This concerns the subjective character of the list;
even after extensive discussions, differences in terms of the relations between variables have remained.
In our illustrative COPAC case, a first set of relevant variables were identified through interviews and
relevant literature, but these were always attached to both the SECI model and to the problem, 18
variables were identified (Table 1).
The MDI (ANNEX I - MDI) were analyzed through MICMAC software and both direct and indirect
influences can be represented in a map (Figure 3). This is divided into four quadrants representing four
types of variables: 1. Buffer Variables; 2. Dependent Variables; 3. Relay Variables and 4. Influent
Variables. The results can be observed in figures 4 and 5.
4 3
INFLUENT VARIABLES RELAY VARIABLES
z
1 2
DEPENDENCE
Figure 3. The influence x dependence chart
Table 1. Relevant Variables
N?_Long label Short label
1 Shared Knowledge Shrd_Knw
2. Individual nowledge Ind Knw
3 Non-shared Knowledge N_Shrd_Knw
4 Organizational Knowledge Org _Knw
5. Staff Training by extemals Ext Tmg
6 Staff Training by intemals Int Stf Tr
7 New PMP (Project Managent Professional) New PMP
8 New Rookie New Rk
9 Interpersonal Interactions IPers_Int
10 Research and Development Rsr_& Dev
11 Organizational Policy Org Pol
12 Human Resources Quality HR Qual
13 Rookie Quit Rk Quit
14 PMP (Project Manager Professional) Quit PMP_Quit
15 Information Tecnology Iv
16 Best Practices Exchange Best_Pret
17. Sharing Capacity Shr Cap
18 Time to be Promoted Time Prom
As an option, MICMAC software can generate a direct (from MDI) or indirect (from MII) relations
graph. A direct graph (Figure 6) helped the modelers to establish a first reference for a Casual Loop
Diagram (CLD) construction. Of course, this was a generalization needing further analysis.
al a cc
fauaera)
ia
Set)
8 (Ora_Knn} 8 ‘ER Gual}
3 i 5 *
5 (ok | =
‘4
[Shrd_Knw]
4
[Time Prom]
‘dependence
Figure 4.Direct influence/dependence map Figure 5. Indirect influence/dependence map
Best Pr 3
eee Pers nib Ny
3 1 . he a
: 2 co
[Org Po \" 2
lOrg_Knv :
2 kee PMP_Quil Ext Ten
3 . a .
2 A Time Prom] is
d 2S ¢ 1
Set [Shrd_Knvy ind Knvl 7
7 3 Rsr_& Dev
IN_Shrd_Kmi es
Rk_Quit New OME 3
3 R_Qual is
i
Weakest influences
— Weak influences
— Moderate influences
— Relatively strong influences
— Strongestinfluences
Figure 6. Direct influence graph
In a strict formalistic application, this method is heavily reliant on participant choices. The results can be
strongly biased by dominating competencies within the group. Therefore, it is necessary to set up as
multidisciplinary a team as possible to constantly check the understanding and meaning of the variables.
Obtaining a consensus does not mean no errors. However, collective, participatory method greatly limits
the risks of incoherence and, at the same time, offers the opportunity to build up together a common
experience, a common knowledge (GODET, 2006).
In this work, this method proposed by Godet, despite not following a strict formalistic application, proved
to be useful in the identification and organization of variables to be considered. Moreover, it allowed the
initial draft of the model to be realized.
3 Dynamic Model
To build the model, a framework to capture the dynamics of staff tunover was required. This had to
reflect both possible improvements and deteriorations in organizational knowledge. The framework,
known as a “coflow”, is used to keep track of the attributes of various items as they travel through the
stock and flow structure of a system (STERMAN, 2000).
A mental model (Figure 7) was created to represent the COPAC coflow. In this structure each NEW
ROOKIE flowing into the PMP TOTAL adds the marginal attribute (knowledge) to the KNOWLEDGE
TOTAL. Similarly, for each outflow from PMP TOTAL there is a corresponding drain from the
KNOWLEDGE TOTAL.
PMPTOTAL,
Figure 7. COPAC’s Knowledge coflow structure. Adapted from STERMAN, 2000.
Individual knowledge increases with training and declines as professionals forget or as changes in the
process make existing experience obsolete. The KNOWLEDGE TOTAL associated with a PMP is not
conserved and the coflow structure should include additional flows into or out of the KNOWLEDGE
TOTAL stock. This is a “nonconserved coflow” (STERMAN, 2000).
The COPAC model (ANNEX I) can be divided into two distinct connected sub models, as observed in
figure 7, the first (AGING CHAIN SUBMODEL) reflects staff tumover, while the second (SECI
SUBMODEL) simulates the relationships that determine the formation of organizational knowledge. The
model was created in Powersim Studio 10 Professional (10.00.5486.6) software.
3.1 The Aging Chain sub model
The turmover model uses a structure called "Chain of Aging" (STERMAN, 2000) (Figure 8) and it
assumes a staff policy based on the inputs and outputs rates of ROOKIE, staff with low level business
knowledge, and Project Manager Professionals (PMP). The Project Management Professional (PMP) ® is
an industry-recognized certification from the Project Management Institute (PMI) for project managers.
The PMP® demonstrates that you have the experience, education and competency to lead and direct
projects (www.pmi.org).
In COPAC is not necessary to have a PMP certificate to be considered an experienced manager, but the
requirements to get this certificate were used to define the transition from a rookie to an experienced
manager. So, the variable “time to be promoted” was defined: 2 years as a default period for promotion in
the COPAC model. This variable also depends on other factors (Figure 10) which have a negative
(Balancing) loop, for example, increasing staff training by externals will accelerate the prospects of
promotion.
In this model, a COPAC policy of tumover control was established. The “rookie acquisition rate” depends
on the “PMP quit rate” (assumed to be exogenous) by a fraction of 40%, defined as the “increase margin
factor’. This is enough to guarantee a total staff increase rate of 5% per year to support future incoming
projects.
sl ead
ft
srowth rate
INCREASE MARGIN é)
enon
i +
+ ‘4
new PHP|
staff ota
/
+ +
[Ce +
) on
rookie accustion promotion Rate PMP quit rate
ae Rookie = MP S
PMP QUIT FRACTION
ome tobe
promoted
Figure 8. Model "Chain of aging". Adapted from STERMAN, 2000.
This is a two-level promotion chain, adapted from Sterman (2000), for rookie and PMP, which provides a
way to model the learning curve for new staff members. It should be embedded in a full model of the
organization so as to also characterize the dynamic nature of knowledge flow.
3.2 The SECI sub model
To represent the SECI sub model (Figure 1) and, after that, link it to the Aging Chain model (Figure 8) it
was necessary to consider knowledge, both tacit and explicit, as resources (stocks), where they
accumulate and either decay or get utilized.
The knowledge, generally speaking, passes through all steps of the SECI model (Socialization,
Externalization, Comt and Internali ) as a knowledge sharing process. This process by which
knowledge held by an individual is converted into a form that can be useful by other individuals. It is also
important because it provides a link between the individual and the organization by moving knowledge
that resides within individuals to the organizational level, where it is converted into economic and
competitive value (HENDRIKS, 1999).
This capacity to share knowledge is also a necessary condition for the creation of new knowledge
(NONAKA and TAKEUCHI, 1995). As knowledge sharing refers to the sharing of not just codified
knowledge but also beliefs, experiences, and contextualized practices (DAVENPORT and PRUSAK,
1998), it is only through such sharing that a base of jointly held knowledge, necessary for mutual
understanding, can be created (NONAKA and TAKEUCHI, 1995).
So, the model (figure 9) basically has three stocks: Rookie Non-Shared Knowledge, PMP Non-Shared
Knowledge and Shared Knowledge.
New Rookies bring a certain amount of experience (knowledge) with them, but the model treats only part
of this as useful for project management. The knowledge, at this stage, is individual and non-shared with
others because there haven’t been any kinds of interaction yet. The same occurs for new PMP entering the
organization.
For example:
new rookie individual knowledge = ‘growth rate** rookie factor’
bw sacron Steere 0 fk ls oem cur STO
we as ON
“nowtodae me clade
cas fu ,
ae,
oe
Traependent
é | Bek oo =
dhouts . pores ned ,
wea se Se oil co
Saf non-anared
owiadge aut ate
Figure 9. Model for knowledge sharing in a SECI process.
All the SECI modes (Socialization, Externalization, Combination and Internalization) were modeled
through rates that increase the shared and non-shared stocks: new individual knowledge acquisition rate,
PMP acquisition knowledge rate, rookie shared acquisition rate and PMP shared acquisition rate.
Socialization (implicit-to-implicit) consists of sharing knowledge through social interactions (tacit to
tacit) or better, sharing experience, know-how and secrets directly at work through a tutor and apprentice
liaison.
The process of externalization (tacit-to-explicit) gives a visible form of tacit knowledge and converts it to
explicit knowledge. Tacit knowledge may take the form of metaphors, analogies, concepts, hypotheses, or
models" (NONAKA and TAKEUCHI, 1995). In externalization, individuals are able to articulate their
knowledge and know-how and, in some cases, the know-why and the care-why.
Combination (explicit-to-explicit) is the process of recombining discrete pieces of explicit knowledge into
a new form. No new knowledge is created at this step. It is rather to improve what we have gathered so
far, to make a summary or review report, a brief analysis or a new database. The content has been
organized logically to extract more sense.
The last conversion process, internalization, occurs through diffusing and embedding newly acquired,
consolidated knowledge. In some way, internalization is strongly linked to "learning by doing"
(NONAKA and TAKEUCHI, 1995), it converts or integrates shared and/or individual experiences and
knowledge into individual mental models.
Once internalized, new knowledge is then used by members who broaden it, extend it, and reframe it
within their own existing tacit knowledge.
The rates that increase the shared and non-shared stocks depend on factors and variables (figure 10)
identified during the MICMAC process (and subsequent discussions). This helped the modelers to define
the weights or coefficients (X 1, X2, ..., Xn) by which they multiply the variable. In such cases case, the
sum of the coefficients should be 1.0 (X1+X2+...+Xn=1.0)
For example:
‘rookie knowledge increase factor’ = O *(1+{ X1 *'STAFF TRAINING BY INTERNALS'+ X2
*'STAFF TRAINING BY EXTERNALS'+X3 *'RESEARCH AND DEVELOPMENT'+ X4
*'INFORMATION TECHNOLOGY ’))
In this example, © is the initial value of this parameter (Q=0.5). It is an individual factor that will raise
d
(or a Rookie's k ledge stock, di ding on the impact of each variable.
At this point, note that the model runs in reference mode, i.e., an initial condition is assumed (the zero
referential) and all the variables (under study) are considered as variations around the zero. For example,
if there are 10 (ten) external training sessions per year and they increase to 12 (twelve) per year, this
represents an effort 20% above the initial condition. The coefficient X 2 should thus be equal to 0.2.
One limitation of this approach is that the ref dition or level for of i
resources may alter over time (WARREN, 2008).
HUMAN RESOURCES ORGANIZATIONAL RESEARCH AND -—=INTERPERSONAL INFORMATIONAL «STAFF TRAININGEY STAFF TRAININGEY EST PRACTICES
‘QUALITY POLICY DEVELOPMENT INTERACTIONS TECHNOLOGY INTERNALS EXTERNALS EXCHANGE
&
doe
rookie knowledge = tims tobe PMP knowledae seuaion laa
ineeage Factor promoted incense Factor nee
PMP TIME
REQUIREMENT
Figure 10. Factors affecting shared and non-shared knowledge acquisition rates.
The loss of organizational knowledge is also proportional to the obsolescence of knowledge. In this sense,
the model assumes a lifespan for useful knowledge (shared and non-shared), represented as the
‘SHARED KNOWLEDGE DECAY FACTOR’ and the ‘NON-SHARED KNOWLEDGE DECAY
FACTOR’ respectively. These factors have a negative polarity, i.e., the longer the factor is (i.e knowledge
with a longer ‘life’), the slower is the decay rate.
We can also observe that non-shared knowledge depends directly on individual knowledge; it declines
over time and is forgotten as members no longer use it in the everyday ways that normally reinforce it. By
contrast, shared knowledge is collective and so takes more time to be forgotten. So the SHARED
KNOWLEDGE DECAY FACTOR is generally greater than the NON-SHARED KNOWLEDGE
DECAY FACTOR.
The dynamic behavior of organizational knowledge is directly influenced by the gains and losses of
shared and non-shared knowledge. So, the model represents Organizational Knowledge as a sum of
shared and non-shared knowledge (Figure 11).
Organizational Knowledge
Rookie Non-Shared PMP Non-Shared Shared Knowledge
‘no wiedge Knowledge ——
Tr
ae) u
ri +
Non-Sh
Knowledge fetal
+ +
Organizationgl,
knowledge TOT!
Figure 11. Organizational Knowledge as a sum of shared and non-shared knowledge.
In order to integrate the two sub models (the Aging Chain and SECI), the work of Kim Warren (2008)
provides useful leads and insights. He has developed strategy dynamics frameworks and discusses key
issues, such as integrating intangibles into the strategic architecture, and resources “attributes” - the
qualities that change as resources are won or lost.
The resulting model can be observed in Annex I.
4 Scenarios
The understanding of how certain variables influence a complex system can be achieved through the
study of a set of future occurrences likely to result from existing conditions. As time passes, it can be
observed how these variables give rise to future situations.
Various definitions of such scenarios can be identified in the literature, especially that of Michel Godet
(2001), who claims that scenarios are "a description of a future situation and the course of events that
allows us to move from the original situation to the future situation". This theme is explored in several
publications by Godet, under the theme “Strategic Foresight which refers to an effective methodology
for understanding the future and defining directions and ional ies (Godet, 2001, 2000).
Two types of scenarios are distinguished:
- Exploratory: setting off from past and present trends, and leading to a credible future;
- Normative, or anticipatory: constructed from alternative images of the future, these can be unwanted or
even feared; they are also conceived retrospectively.
Scenarios and comparisons between present and possible future situations inform the model. The term
"scenario" was used to describe states or changes in key variables and future changes in organizational
environment from these variations.
In this study, exploratory research scenarios were simulated in the dynamic model. They encompass the
intervention of various key events or circumstances that occur between the original situation and a
possible future. This allowed the model to be effectively evaluated.
The following scenarios were established:
Scenario 1: Variations of 20%, 30%, 40% and 50% occurred among departing PMP (managers), which
kept within the policy of acquiring new members (Rookies) with an “increase margin factor” by 40%;
Scenario 2: Variations of 20%, 30% and 40% increases in “staff training by extemnals”, while keeping
constant a PMP departure rate (35%).
The first scenario represents a policy that guarantees an increase in staff of 5% per year. But, the question
to be answered here is: What is the effect on organizational knowledge of the loss of PMP managers with
experience?
Organizational Knowledge f(PMP quit rate)
tlzlale[sle lr fe ls hrohssh2|
13
1 lela i ls le > le ls hchskos els |e [s le IpIsl> hokshe| 1 le [3 |e ls lel> ls |s hohe
14 15 16
Time (months)
— PMP quit rate 20% ~~ PMP quit rate 30% -- PMP quit rate 40% -- PMP quit rate 50%
Figure 12. Graph resulting from POWERSIM software on variations in organizational knowledge related
to the ratio of output managers.
Analyzing the behavior of Organizational Knowledge variable in the COPAC model, in which the
number of trained managers increases by 5% per year, it suggests that the policy initially adopted
(scenario 1 - figure 12) will lead to a gradual loss of organizational knowledge.
The second scenario takes one variable (STAFF TRAINING BY EXTERNALS) which can be
manipulated and which should impact most on the whole system. It was chosen with the aid of
MICMAC, especially through the analysis of the Matrix sum of Direct Influences (MDI) (Table 2). This
sum ranks variables according to their influence or dependence. The objective here is to analyze the
impacts caused by the increase of one variable on the system, but a combination of several factors could
also be analyzed.
Table 2. Directly influences/dependencies rating of variables according to the MICMAC method
Ne VARIABLE INFLUENCE FACTOR DEPENDENCE FACTOR
1 Shared Knowledge
2 Individual Knowledge 16 38
3 Non-shared Knowledge 6 29
4 Organizational Knowledge 12 34
5 Staff Training by extemals 23 1
6 Staff Training by intemals 18 1
7 New PMP 12 3
8 New Rookie 2 3
9 Interpersonal Interactions 15 ul
10 Research and Development 7 5
11 Organizational Policy 9 9
12 Human Resources Quality 14 9
13. Rookie Quit u 4
14 PMP Quit 2 5
15 Information Tecnology 8 2
16 Best Practices Exchange 10 4
17 Sharing Capacity 12 8
18 Time to be Promoted 4 18
Totals 21 BI
This influence, of the “STAFF TRAINING BY EXTERNALS”, can be seen in Figure 13:
Organizational Knowledge f(tranning by externals)
Indknw.
? [eleteTeTotel Tefal sa
13
1[e[s[sls [er lelshohsha
14
sTels[e[s Ts fr Tels hobska| [2s [ss e[2Tels hokihd
15 16
Time (months)
— PMP quit rate 35% trainning effort actual ~~ PMP quit rate 35% trainning effort 30%
~~ PMP quit rate 35% trainning effort 60% — -- PMP quit rate 35% trainning effort 90%
Figure 13. Graph resulting from POWERSIM software on variation in organizational knowledge
resulting from increased training efforts.
In this scenario, increased training can reverse organizational knowledge loss. For example, an initial
condition of 10 (ten) extemal training sessions per year, could be increased to 13 (thirteen) per year, this
represents an effort 30% above the initial training condition and can increase the organizational
knowledge level, but not enough to reverse the overall tendency of loss. So, training sessions could be
raised to 60% and 90% to reverse this trend.
Of course, there could be an operational limit to increase the training efforts and combined scenarios
could then be used to reverse the organizational knowledge loss, for example: increasing efforts in both
STAFF TRAINING BY EXTERNALS and INTERPERSONAL INTERACTIONS.
5 Conclusions
The COPAC knowledge model represents the tumover of specialized personnel and its relation to the
formation of organizational knowledge and can be divided into two distinct connected sub models, as
observed in figure 7, the first (AGING CHAIN SUBMODEL) reflects staff tumover, while the second
(SECI SUBMODEL) simulates the relationships that determine the formation of organizational
knowledge.
The modeling process was facilitated with the use of concepts from Godet’s Strategic Foresight,
especially by the use of the MICMAC method, which provided an initial approach for the COPAC
members to understand and determine the relative importance of each variable in the system being
modeled.
After performing the simulations, the results suggest that the increased PMP quit rate (after 20%) will
lead to a gradual loss of organizational knowledge (Figure 12).
The organizational factors related to key variables, as suggested by MICMAC analysis and COPAC
members, such as increased STAFF TRAINING BY EXTERNALS have a strong influence on the system
(scenario 2). This factor can variations in ige caused by
policies (scenario 1).
A combination of changes in the influence variables, for example: INTERPERSONAL INTERACTIONS
and ORGANIZATION POLICY, can be simulated and may also be alternative solutions to reverse the
loss of organizational knowledge. Testing these before final implementation can help reveal the
capabilities and the complex combinations of factors that are involved in various feedback relationships
or in complex feedback.
The architecture of this model is generic by nature, and can be applied to other organizations by replacing
initial values of factors affecting shared and non-shared knowledge acquisition rates.
The model is easy to expand and modify in both structure and parameter values, as new data becomes
available. This can then be rerun, the results analyzed and improvements made as necessary.
In sum, the understanding of key factors of influence in Knowledge Management through a dynamic
model can provide clear guidelines for the adoption of policies that encourage all staff to share important
knowledge and experience leading to organizational success in complex projects of this nature.
Finally, our main proposal for future work is the creation and analysis of dynamic models able to evaluate
the impacts of knowledge loss on productivity within complex organizational structures.
6 References
[1] ALVARENGA NETO, R.C.D. Gestéo do conhecimento em organizagdes: proposta de mapeamento
conceitual integrativo. Sao Paulo: Saraiva, 2008.
[2] BONOTTO, M. B. Gerenciamento de projetos no ambito do subdepartamento de Desenvolvimento e
Programas - Possiveis Vulnerabilidades. Monografia - Escola de Comando e Estado-Maior da
Aeronautica, Rio de Janeiro: Universidade da Forca A érea, 2005.
[3] CHOO,C.W. A Organizacao do conhecimento: como as organizagdes usam a informacao para criar
significado, construir conhecimento e tomar decisdes. 2 ed. Sao Paulo: SENAC, 2006.
[4] COYLE, R.G. System Dynamics Modeling: a Practical Approach. London: Chapman and Hall, 1996.
[5] DAVEMPORT, T. H.; PRUSAK, L. Conhecimento empresarial: como as empresas gerenciam 0 seu
capital intelectual. 4. ed. Rio de Janeiro: Campus, 1998.
[6] DURANCE, P; GODET, M. Scenario building: Uses and abuses, Disponivel em
http://en.1 ive.fr/dy icl ‘io-building-tfsc-2010.pdf ,Acesso em 20 mar
2013.
[7] FORRESTER, J.W. Industrial Dynamics. MIT Press, Cambridge, MA : 1961.
[8] FORRESTER, J.W., LUX, N., STUNTZ, L. Road Maps: a Guide to Learning System Dynamics.
MIT Sloan School of Management, MIT Press, Cambridge, MA: 1994.
[9] GODET, M. etal. A “Caixa de ferrramentas da Prospectiva Estratégica”, Cademo CEPES, 2000.
[10]GODET, M., Creating futures - Scenario Planning as a Strategic Management Tool,
Paris:Economica, 2001.
[11]GODET, M. "L’art et la méthode" Tome II - Editions Dunod (“The art and the method”, Volume II,
Dunod Edition), 2001.
[12] HENDRICKS, P. Why share knowledge? The influence of ICT on the motivation of knowledge
sharing. Knowledge and Process Management, 1999. HISLOP, D. Knowledge Management in
Organi: A Critical Introd: . New York: Oxford University Press, 2005.
[13]JASHAPARA, A. Knowledge management: An Integrated Approach. Essex: Ashford Colour Press,
2004.
(14] LAKATOS, E.M.; MARCONI, M. A. Metodologia Cientifica.Sao Paulo: Atlas, 1986.
[15] LESSER, E.; PRUSAK, L. Communities of practice, social capital and organizational
knowledge.Whitepaper, IBM Institute for Knowledge Management, Cambridge, 1999.
{16] MARCONI, M. A.; LAKATOS, E. M. Técnicas de pesquisa:planejamento e execucao de pesquisas,
amostragens e técnicas de pesquisa, elaboracao, analise e interpretacao de dados. 7. ed. Sao Paulo:
Atlas, 2008.
[17] MASSINGHAM, P. Measuring the impact of knowledge loss: More than ripples on a pond ?. Sage
Publications: 2008.
[18] NONAKA, I; KONNO, N. The Concept of “Ba”: Building a Foundation for Knowledge Creation.
California Management Review 40: 4054, 1998.
[19] NONAKA, I. ;TAKEUCHI, H. The knowledge-creating company. New York: Oxford University
Press 1995.
[20]PARDUE, J. H.;CLARK, T.D.;WINCH, G.W. Modeling short-and long-term dynamics in the
lization of technical ad in IT prod industries, System Dynamics Review
Vol.15, 1999.
(21]PIDD, Michael. Modelagem empresarial: ferramentas para tomada de deciséo. Bookman: Porto
Alegre, 1998.
[22]SILVA, A. A Viabilidade de implantagao de um projeto piloto de gestéo do conhecimento no
Subdepartamento de Desenvolvimento e Programas da Aeronautica. Brasilia: UNB, 2004.
[23] STERMAN, J. Business dynamics: systems thinking and modeling for a complex world. USA: Irwin
McGraw-Hill, 2000.
(24]STERMAN, J. System Dynamics Modeling for Project Management. Sloan School of
Management, Massachusetts Institute of Technology. Cambridge, 1992.
[25] RITCHIE-DUNHAM, J. Initiating management dialog using a summary presentation that integrates
the findings from multiple SD analytical tools. Proceedings of the 15th Intemational System
Dynamics Conference, Istanbul, Turkey. August 1997.
(26]SVEIBY, K.; LINARD, K.; DVORSKY, L.Building a Knowledge-Based Strategy: A System
Dynamics Model for Allocating Value Adding Capacity. Artigo disponivel em:
http://www.sveiby.com/articles/sdmodelkstrategy.pdf . Acesso em 12 fevereiro de 2013.
[27] TAVARES, J.C.C. Produgaéo de conhecimento no ambito dos projetos da Subdiretoria de
Desenvolvimento e Programas. Monografia - Escola de Comando e Estado- Maior da
Aeronautica, Rio de Janeiro: Universidade da Forca A érea, 2008.
[28] UNB. Curso de Model de Sistemas Dinamicos C: yt apostilas de textos.Universidade de
Brasilia, Brasilia: 2012.
[29] WALTZ, E. Knowledge management in the intelligence enterprise. Boston: Artech House, 2003.
[30] WARREN, K. Strategic Management Dynamics, Wiley: New York,2008.
[31] WENGER, E.; MCDERMOTT, R.; SNYDER, W. M. Cultivating communities of practice:A Guide
to Managing Knowledge. Harvard: Harvard BusinessSchool Press, 2002.
ANNEX I
COPAC’‘s Matrix of Direct Influences (MDI)
©UPSOR-EPITA-MICMAC
18: Time_Prom
al
io
io
io
io
IN
17: Shr_Cap
16: Best_Pret
15:1T
14: PMP_Quit
13: Rk_Quit
12: HR_Qual
11:Org_Pol
10: Rsr_& Dev
:1Pers_Int
> New_Rk
> New_PMP
: Int_Stf_Tr
: Org_Knw
: N_Shrd_Knw
:Ind_Knw
9
8
7
6
5:Ext_Tmg
4
3
2
L
: Shrd_Knw
T{3 (3 [2]0 {1 [10 ]2]1/2/3]0[0]1 [10/2
3/3 ]1][2 00/0/03 /1]/0]0]/0/0[0]0]2 1/3
PRES ERED ESPACE ee
3/2]1/2 0 [1/00 lO/1]}2]1/0/o{o0}o0]0}2
S
lo
lo
lo
lo
lo
lo
lo
lo
lo
lo
lo
s
lo
lw
=
en
a
2/3 /0/0]/0 [0/0/00 /0{0/0/3/3/0]0/0
lo
lo
lo
lo
lo
lo
lo
lo
lo
lo
lo
lo
lo
lw
en
len
en
lw
lo
lo
lo
lo
lo
lo
a
lo
en
ln
en
lo
lw
en
en
en
lw
lo
lo
lo
lo
lo
lo
lo
lo
lo
lo
lo
lo
lw
*
i
ln
3/2/2/3 [0/0/00 /0/0{0]0[0}0/0]0}0
lo
lo
lo
lo
lo
lo
lo
lo
lo
lo
lo
lo
lo
len
lw
ln
en
Ol2fofofsolosolofolofojofoj2f{ojojojo
1: Shared Knowledge
2: Individual Knowledge
3: Non-shared Knowledge
5: Staff Training by externals
6: Staff Training by internals
7: New PMP
8 : New Rookie
9: Interpersonal Interactions
11: Organizational Policy
12: Human Resources Quality
13 : Rookie Quit
15 : Information Technology
16 : BestPractices Exchange
17: Sharing Capacity
18 : Time to be Promoted
COPAC Organizational Knowledge Model
