To Main Proceedings Document
Business M odeling Process
Shoji Hidaka
NTT DATA Corporation
Shinkasumigaseki Bldg. 18th Foor
3-3-2 Kasumigaseki, Chiyoda-ku
Tokyo 100-0013, Japan
Telephone: +81-3-3506-4631, Facsimile: +81-3-3506-4638
E-mail: hidakas@ noanet.nttdata.co jp
Abstract
This paper discusses the key issues in modeling process. In teaching business simulation,
the most frequently asked question is, “How can I get to develop such models?” It is
easier to explain the structure of existing models and managers can easily learn how to
use simulation software. Modeling is integration of several techniques; techniques for
collecting necessary information, techniques for summarizing them, techniques for
mapping them into diagrams, and computer simulation technology. In this paper, I
discuss the key points in modeling process through my experience in teaching business
modeling not in classroom but in real business fields. I also introduce the effectiveness
of the KJ Method, a useful technique in collecting and summarizing information, which
is frequently used in Japanese organizations. This paper is based on the case study in
NTT Data, applying the SD modeling to software quality management. This case study
has two main implications for effective modeling. First, it suggests that understanding
whole process of modeling is the most important to develope a good modeler. Second,
some techniques forTQM are also useful for SD modeling.
1. Business Modeling Process
Frequently Asked Question
When I teach SD modeling to managers in my company, the most frequently asked
question is, “How can I get to develop such models?” This means that managers can
understand the SD modeling but they cannot become a good ‘modeler’ even after the
SD course.
In the course, as the time is limited, I usually show several sample models and explain
the structure and formula. I also explain how to develop simulation models with special
software, such as i-think or Powersim.
Managers can understand the structure of the models and get easily boring in the third
model. They quickly learn how to use simulation software because those software have
Windows-based simple user interface. In addition, most the mangers have systems
engineering background. (NTT Data is the largest systems integrator in Japan.)
However, they often complain that they cannot develop their own simulation models by
themselves. This may be the similar situation when students in painting class saw
beautiful pictures and learned how to use painting tools but they cannot draw such
beautiful pictures.
I think that developing simulation models with software is only a part of the modeling
processes and is not the most important one. Managers need to learn total process of
modeling. From this point of view, I would like to analyze the modeling process in
detail first.
Modeling Process
Below is the standard SD modeling process which I leamed at the London Business
School as an MBA student.
PROBLEM OR ISSUE
!
BOUNDARY DEFINITION AND
PROPOSALS FOR
ACTION
f
POLICY EXPERIMENTS
CONCEPTUALIZATION
a, SIMULATION EXPERIMENTS
STRUCTURAL DEFINITION
NS DETAIL MAPPING & /
EQUATION FORMULATION
When I saw this chart for the first time, I thought that this is similar to the ‘QC story’,
which is used for the process in Japanese TQM activities. The second chart shows the
standard ‘QC story’ in Japan.
Selection of a Theme
V tL
Insights and Future Direction
Current Situation Analysis ‘
| Prevention of Backsliding
Setting the Target '
| Effect analysis
Cause Analysis /
ty Implementation
Solution selection —*
The main differences between the two chart are:
1) ‘QC story’ itself is the feedback loop process. (often called ‘PDCA’ cycle)
2)‘QC story’ emphasizes more on the early stages of the process. (Selection of the
theme, Current situation analysis)
This ‘QC story’ provides good suggestions to understand the modeling process in detail.
Before using simulation software, there are several steps to do in modeling.
In addition, when we develop models with causal loop diagrams, we usually start from
the very simple model to the detailed one. Further, we sometimes develop parts first and
combine them later.
Below is the modeling process model, I use recently in my classes. Based on the
standard model which I learned at the LBS, I add some elements from the QC story and
my own experiences.
SELECTION OF A THEME
“
SCHEDULING
ACTION
\
SIMULATION
t
SOLUTION SELECTION
INFORMATION COLLECTION
!
INFORMATION ANALY SIS
SIMULATION
MAPPING & “a
FORMULATION
BASIC MODEL
‘
DETAILED MODEL
2. Example
A proper example often helps to understand concepts effectively. In this section, I would
like to show the processes which I used in developing the Quality Control Model last
year.
1) Selection of the theme and planning
Firstly, we develop a team for the TQM activities and discussed the theme of the year.
We usually use the ‘brain-storming’ technique for creative thinking with a team. We
chose the quality improvement of a computer system as that year’s theme.
2) Collection of the information
The data about the system’s quality was extracted from the database and analyzed by
using statistical techniques. We also interviewed some experts to understand the causes
of the poor quality of the system. Interviews are necessary because there are lots of
fuzzy or non-linear relationships which cannot be easily explained by Statistics.
3) Analysis of the information
In the next step, we wrote down what we collected into small pieces of paper and
summarized them by the ‘KJ Method’.
TIME MONEY
BUDGET
cost
MAN
SKILLS
? SOFTWARE
MANPOWER aon
4) Modeling
We developed the causal loop diagram of the quality process. We started from a very
simple model to a detailed model. This process was done by the team again, which
allowed us to share the concepts of the model. Modeling in the business fields requires
consensus of a team to develop effective solutions.
+ ‘ oe) Trouble
Quality Total ay ‘Rap
_ z
/ “Ay fo. oualty Troubles «7
/ \ 1 \ )
| + f yr) ® |
B Tubes | \ 1 ag
| \ ate
Canal Upide x Sze
correction } auay Quay Required
ng eent
be Bee ; L ye sen i Programmers
— ‘ a NN /
\ _ — sk
Total Update a Level
Budget ——* Budget ——
5) Development of the simulation model
We developed the simulation model with some simulation software, i-think and
Powersim, ran the simulation, and checked the results.
6) Reflection and Feedback
We discussed the solution to improve the quality of the system and changed the
simulation model and checked the simulation results again.
3. Techniques in Modeling
In order to develop the modeling skills, students need to learn all techniques use in the
whole process of modeling. In this section, I would like extract some of those
techniques from the previous example. Some of them tend to be overlooked in teaching
SD modeling.
1) Techniques for collecting information
Database Manipulation
Interview technique
Brain-storming
2) Techniques for analyzing information
‘KJ Method’
‘Hexagon’
3) Techniques for developing models
Causal Loop Diagram
Stock and Flow Diagram
4) Techniques for developing simulation models
Simulation Software
*KJ Method
KJ Method is a technique for summarizing information. It was developed by Jiro
Kawakita in 1970s. Most of Japanese managers can use this technique today. The chart
in the previous section was also made with this technique. The main processes of the KJ
Method are as follows.
1) Write down the information onto cards. (We often use Post-It™)
2) Collect similar cards to some groups.
3) Name each group.
4) Make a story
=
>
In the SD modeling courses, we tend to explain only diagrams and simulation tools,
which cannot develop good modelers. We need to emphasize more on the early stages of
the whole modeling process, such as collection and analysis of information. I started a
new SD modeling course this year in my corporation. In this course, I am trying to
4. Conclusion
explain the whole process of modeling and techniques used in each step. I found that
Japanese traditional techniques used in TQM activities, such as KJ Method, are often
useful in SD modeling.
Further studies
As the next step of my study about the modeling process, I would like to introduce some
techniques used in the software development into SD modeling, such as Structured
Analysis, DOA (Data Oriented A pproach), Object-Oriented A pproach.
References
Hidaka, S. (1999), System Dynamics: a New Tool for TQM, 17" International System
Dynamics Conference Proceedings
Hodgson, A.M. (1992), Hexagons for systems thinking, European Journal of
Operational Research 59, 220-230
Kawakita, J. (1967), Hassoho, Chuo-koron (in Japanese)
Senge, P.M. (1990), The Fifth Discipline: The art and practice of the learning
organization, Century Business
Vennix, J.A.M. et al. (1992), Model-building for group decision support: Issues and
alternatives in knowledge elicitation, European Journal of Operational Research 59,
18-41
