Main Proceedings Document
Review and validation of early J apanese local government SD models
Saburo Kameyama, Faculty of Commerce, Chuo University
E-mail: kameyama@ tamacc.chuo-u.ac.jp
Hidenori Kobayashi, Faculty of Policy Studies, Chuo University
E-mail: kobaken0@ fps.chuo-u.ac.jp
Toru Suetake, Arthur Andersen Tokyo
E-mail: tohru.suetake@ jp.arthurandersen.com
Abstract
During 1970s, many Japanese local governments built their SD models for formulating their long
and medium term development plans. However, as we mentioned in our presentation last year ISDC
at Wellington, these models have not been used any more for their policy making.
However, sufficient time has passed to verify these Japanese local models. We have reviewed some
of these models such as Saitama Prefecture Model, Hyogo Dynamics Model and Tokyo Metropolitan
Model and found some meaningful hindsight as follows:
Conceptual validity: SD as the design methodology based on the core concept of feedback is
exclusively valid. Without the feedback concept it is impossible to design
complex social systems.
Institutional validity: Connected with institutional resources such as annual budget, SD model can
carry its validity into the practical planning and implementation processes of
local government policy.
Operational validity: Simulation results of above models were not always valid. But errors can be
accrued defendable to initial conditions, while intrinsic operational properties
of the model remains.
Key Word: local government model, macro economics
1. Japanese local government models developed during 1970s
The System Dynamics was introduced to Japan in later 1960s by Professor Watanabe, Professor
Sakakura and other pioneers. After their introduction, a kind of boom has happen and started many
introductory seminars and application trials in many areas.
As one of this big streams, in during 1970s, there were a kind of SD modeling boom in Japan for
local governments aimed to give some direction of their visioning and long term planning. These SD
simulation results expected to give some trustful perspectives on their future and also the local
government staffs believes that they can control perfectly if they know the right results or right
target.
Japanese Local Government Models
Miyagi Dynamics Model
Saitama Model
| Tokyo Metropolitan Model
Narashino City Population Model
and
Locations
Shiga Model
Hiroshima Model
Fig- 1: Map of local government model area
1
The major local government SD models developed in those days is as follows:
Table 1: Models and Major Simulation Results
‘Name of the models
Major simulation results
Information Resources
Miyagi Dynamics Model
Target years is 2025 (50 years after)
Population would be grow to 3,578 thousand (1.8
times of present population)
‘Aged person over 65 years would be 15% of the total
population (3.6 times)
Household would be increased to 1,181 thousands
(2.2 times)
GNP would be 10 trillion Yen (8.2 times) and
agriculture sector would be reduced from 11% to 3%,
industry sector would be grow from 26% to 30% and
service sector would be grow from 63% to 67% in
target year
GNP per capita would be growing from 630 thousand
Yen to 2.9 million Yen.
They expect very optimistic view on environmental
issues and NOx of air pollution would be reduced to
30% (8 thousand tons per year) and BOD of water
pollution would be reduced to 40% (9 thousand tons
per year)
Electric consumption would be increased to 27.3
billion KWHiyear (6.1 times)
* Overview of Miyagi Dynamics Model,
manuscript report by Mr. Syoichi
Toyohara, Planning Department of
Miyagi Prefecture, June 1977
* Challenge with System Model to
Socio-economic Environmental
Dynamics: Development of Miyagi
Prefecture Long-term Total Planning
Model, by T. Matuzaki et al,
Chiba Prefecture Model
Target yearis 1999 (21 years after)
Population would be increase from present 4.6 million
to 6.6 Million.
‘Aged person over 65 years old would be increased
from present 6% to 14% of total population.
GNP would be increased from present 4.6 trillion Yen
to 18.3 trillion Yen.
Service sector would be rapidly increase from present
51% to 60%.
GNP per capita would also present 1.1 million Yen to
2.7 million Yen
‘They rather have pessimistic environmental issues and
NOx of air pollution would be increased to 1.89 times
and BOD of water pollution would be increased to
1.15 times worse. Also worried about solid waste
material
+ Chiba Prefecture Vision of 21"
Century, February 1980
+ Development Study of Chiba
Prefecture System Dynamics Model,
by Japan Long Term Credit Bank,
March 1978
Narashino-shi Population
Model
Only forecast population of Narashino City.
Target yearis 2000.
‘The population would be increase from 135 thousand
in 1985 to 145 thousand in 2000.
+ A Population Forecast of Narashino-
shi by the Method of System
Dynamics, by Toshiyuki Horiuchi, et
al, Report of Chiba Institutes of
‘Technology, No.27, 1982
Saitama Model
‘Aim to check and validate effectiveness and reliability
of SD modeling.
Target year is 1972.
Calculate during 1965 to 1972 on public sector capital
stock, population, industry output, air pollution, traffic
and find that the difference is within 10% compare
with real statistic figures.
However, we could not find out that this model use for
planning process of Saitama Prefecture
+ Study on Urban Area Government
Planning Support System, Japan
Operation Research Society, March
1975
Toyo Metropolitan Model
Target yearis 2005.
Population of Tokyo metropolitan area would be
increase rapidly.
Shortage of water resources in Tokyo metropolitan
area obstacle economy growth.
* Study on Population dynamics in
‘Tokyo Metropolitan Area, manuscript
draft of Dr. Masahige Tyji, for Japan
Industrial Engineering Society, Fall
1976
+ Study on Water Resources in Tokyo
Metropolitan Area, manuscript draft of
Dr. Masahige Tuji, for Japan Industrial
Engineering Society, Spring 1977
Name of the models
Major simulation results
Information Resources
Kanagawa Model
Expand double of old aged population (over 65) in the
21% century
Increase nitro-oxide (NOx) and other pollution rapidly
to be double of present state
Decrease the green area percentage of urban area drop
down to half of present state
Become secure to keep water resources and only 70%
of the demand could fulfil
Industrial structure change rapidly shift to leaded by
the service sector
+ Nagakiyo Takahashi and Nagashige
Shinozaki "Control and Disclosure of
Local Government Information’,
Jichitai No Keiei To Koritsu, I: Keiei to
Gyouzaisei Unei (Management and
Efficiency of Local Government, I:
Management and Operation, ed.
Makoto Takahahi), Gakuyoshobo,
Tokyo 1982, p131-147
Shiga System Dynamics
Target yearis 2000.
Population grew to 1.6 million but not so rapidly
increase the old age percentage
Decrease 10,000 ha agriculture land
Industry and service sector increase rapidly
GDP would be 5.8 times and GDP per capita would be
3.2 times of present status
Water pollution could be reduce 30% at 1990 but
again increase and almost same level in target year
They have pessimistic opinion on air pollution and
solid waste problem
‘Also they have pessimistic opinion on local
government financial issues.
+ Regional Prediction using System
Dynamics, by Yoshio Morita and
Sadaichi Okada, Nihon Denki Giho
No.120/1977
KLUD: Kyoto area Land
Use Model
“Target year is 1995.
Land use could be rapidly change and face to over
population and rapidly decrease of green area.
‘They recommend to govemment organizational
reform for abolish sectionalism to accomplish much
effective city planing and urban re-development.
+ KLUD I: Development of KLUD
Kyoto area Land Use Model, 1974,
City Planning Development, Kyoto
City
+ KLUD Il: Kyoto City structure shift
from 971 to 1995, 1975, Ib
Appendix of KLUD No.1: KLUD
dynamic model equation, 1976, Ib
Osaka Model Target yearis 2000 + Dynamics Model for Urban
‘They have very pessimistic future Environment, Resources and
Environment protection could be progress with Economic Growth I, by Isao Maniwa,
effective energy consumption policy and improve Obtemon Economic Study Report,
efficiency and productivity. July 1974
However, social development oriented policy and low | * Dynamics Model for Urban
pollution policy may face up with difficulty to keep Environment, Resources and
balance with high economic growth Economic Growth II, by Isao Maniwa,
Ohtemon Economic Study Report,
1976
Hyogo Dynamics Model Target year is 2020. * Long-term Prospecting about
Population would be increase from present 4.8 million
to 6.3 million in 2010 but decrease after that. But
urban area population increase from present 3.7
million to 4.2 million in 1995 but decrease after that.
‘Aged population over 65 years old could be increase
from present 6% to 20% in 2010 in urban area. 20%
in 1995 and 26% in 2020 in the prefecture level.
Moving population to other prefectures and urban
areas would be increase rapidly than coming from.
Aged mortuary rate is higher than infant mortuary
tate.
They recommend to prevent good socio-economic
welfare for aged people.
They also recommend protecting to change into slum
in urban area for urban re-development with deeply
concemed environment protection.
Water resources would be shortage and recommend of
recycling sewerage water.
Population, Pollution, Resources and
Industry of Hyogo Prefecture,
manuscript report of Planning
Department, Hyogo Prefecture,
October 1973
Name of the models Major simulation results Information Resources
Hiroshima City Model | » Target yearis 1990. + A Dynamics Model to Analyze the
* Life style oriented public infrastructure investment Repercussions of Urban Policies, by
policy cause the overpopulation and this may cause Hiroyuki Kitajima and Ryoichi Sasaki,
11% increase of air pollution, heavy water pollution manuscript report at July 1980
and traffic jam,
+ Industry development oriented public policy cause
over population and heavy air and water pollution.
+ People may rather choose public transportation with
suitable policy on transportation sector.
Remark: U$-Yen conversion rate in those days was fixed at 360Yen/U$
2. Model and information resources
We already mentioned that after booming of the System Dynamics in Japan during 1970s, however,
these models are not used for long term planning of Japanese local government anymore suddenly
after 1980s. There are several reasons and we already mentioned at the ICSD last year.
In this study, we find that already some of detail information are disappear. The Shimada SD
Library at Senshu University is the only SD information resources library in Japan. We use the
information in this library. However, this library has limited information because they have only
documents provided from Professor Shimada and do not try to collect other information from other
providers. We try to collect more detail information about the activities of those SD golden days in
Japan and send questionnaires to prefectures and cities of these models before start of this study. But
we could not get any replay until now. This is all information we have now about Japanese local
government SD models developed during in 1970s.
Even though there are limitations of information we have in these models and also these models are
not use for planning of local government nowadays, however, we fell some of these concepts and
ideas are still useful and we can learn things from these models and model builder’s experiences.
Also, it may be the last chance to collect information about these models because many model
builders in these days are retired and we strongly afraid that the memory of those pioneer are also
gone forever. That is the reason we start this study.
Following list is the information we collect for this study and what kind of information we can get.
It shows how limited information we have. About most models, we only know the concept and the
simulation results.
Table 2: Models and Information Resources
Models and Information Resources
Structure and concept
Information
Detail simulation
results
Equations or
program source code
Miyagi Dynamics
Overview of Miyagi Dynamics Model, manuscript | Overview description and _ | Detail table of No information
report by Mr. Syoichi Toyohara, Planning feedback loop concept simulation results and
Department of Miyagi Prefecture, June 1977 diagram but enough to population growth
know the model structure. | graph.
Challenge with System Model to Socio-economic | Overview description and | Much detail graphs | No information
Environmental Dynamics: Development of Miyagi | concept diagram, stock and | and simulation results
Prefecture Long-term Total Planning Model, by 7. | flow diagram tables
Matuzaki et al,
Chiba Prefecture SD model
Chiba Prefecture Vision of 21" Century, February] Just overview description | Comparison table of _ | No information
1980 and overview conceptual _| start year and target
diagram, year,
Development Study of Chiba Prefecture System | Detail explanation about _ | No information No information but
Dynamics Model, by Japan Long Term Credit
Bank, March 1978
the model structure with
detail model structure
diagram.
some of them can
know from the
description of
element relations.
Narashino-shi Population Model
‘A Population Forecast of Narashino-shi by the
Method of System Dynamics, by Toshiyuki
Horiuchi, et al. Report of Chiba Institutes of
Technology, No.27, 1982
Detail structure diagram
and enough description
Simulation result
tables.
DYNAMO program
source code
Saitama Model
Study on Urban Area Government Planning
Support System, Japan Operation Research Society,
March 1975
Detail explanation and
detail feedback diagram,
Detail simulation
results tables and
graphs.
Equations and
program source code
Tokyo Metropolitan Model
Study on Population dynamics in Tokyo
Metropolitan Area, manuscript draft of Dr.
Masahige Tuji, for Japan Industrial Engineering
Society, Fall 1976
Diagrams and overview
explanation
One graph
No information
Study on Water Resources in Tokyo Metropolitan
Area, manuscript draft of Dr. Masahige Tyji, for
Japan Industrial Engineering Society, Spring 1977
Relation concept chart
One graph
No information
Kanagawa Model
Nagakiyo Takahashi and Nagashige Shinozaki
“Control and Disclosure of Local Government
Information”, Jichitai No Keiei To Koritsu, I: Keiei
to Gyouzaisei Unei (Management and Efficiency of
Local Government, I: Management and Operation,
ed. Makoto Takahahi), Gakuyoshobo, Tokyo 1982,
pl3i-147
Not much information
Brief results only.
No information
Shiga System Dynamics
Regional Prediction using System Dynamics, by
Yoshio Morita and Sadaichi Okada, Nihon Denki
Giho No.120/1977
Overview explanation and
conceptual feedback
diagram
Result graph
No information
KLUD: Kyoto area Land Use Model
+ KLUD I: Development of KLUD: Kyoto area
Land Use Model, 1974, City Planning
Development, Kyoto City
+ KLUD Il: Kyoto City structure shift from 971 to
Detail explanation with
conceptual diagram and
feedback loop diagram.
Detail simulation
results tables.
Not enough
information but
many of them can
know from the
1995, 1975, Ib description of
* Appendix of KLUD No.1: KLUD dynamic model element relations.
equation, 1976, Ib No program source
code,
Osaka Model
+ Dynamics Model for Urban Environment, Detail explanation with | Detail simulation Equations and
Resources and Economic Growth I, by Isao conceptual diagram and | result graphs. program source
Maniwa, Ohtemon Economic Study Report, July | feedback loop diagram. codes.
1974
+ Dynamics Model for Urban Environment,
Resources and Economic Growth II, by Isao
Maniwa, Ohtemon Economic Study Report, 1976
Models and Information Resources Structure and concept Detail simulation Equations or
Information results program source code
Hyogo Dynamics
Long-term Prospecting about Population, Conceptual relation chart | No graph and tables] No information
Pollution, Resources and Industry of Hyogo and overview explanation.
Prefecture, manuscript report of Planning
Department, Hyogo Prefecture, October 1973
Hiroshima City Model
‘A Dynamics Model to Analyze the Repercussions of | Overview and conceptual | Overview graph and | No information
Urban Policies, by Hiroyuki Kitajima and Ryoichi | diagram tables
Sasaki, manuscript report, July 1980.
3. Validation of these models
We decide to validate these models with 3 levels: conceptual level, institutional level and
operational. Of course we already know other validation methods but in this case, this three kinds of
validation are suitable for old model for we can use statistic data, historical background and results
of their policy decision making as validation criteria.
3-1) Conceptual validity
SD as the design methodology based on the core concept of feedback is exclusively valid. Without
the feedback concept it is impossible to design complex social systems. However such as KLUD,
some of these model use econometric methods and we feel somehow not fully follow the System
Dynamic’s philosophy that tacitly request to composed with relation of only with flows and stocks.
Therefore, we think considering concept of the model carefully. As concept validity, we select
following criteria as validation elements:
- Consistency: Consist feedback loop concept
- Focus: Is the model focus on purpose or objectives to the simulation? Or on the other word, multi
purpose or single purpose
- Structure: Whether model structure is adequate for purposes of the simulation objectives.
Consistency is validated with the existence of detail feedback philosophy. But some of the models
are not perfectly build by 100% pure flows and stock feedback models. We could understand that the
machine (computer) does not had enough memory and computing ability in those days. Especially
on finance sector, it is sometimes so difficult to make 100% perfect elements relationship with
linkage of macro level and micro level economic factors. There are many unknown economic factors
between national level, prefecture level and municipality level by industry sectors and household
expenditures. Also science and technology development factor is difficult to make component. On
the other hand, in such model so called hybrid which composed with econometric part and SD part
make shorten its life of usage. For the sake of in-composed with econometric part, KLUD and
Saitama Model forecast at most 10 years.
Focus, or single purpose or multi purposes affects complexity of the model. Single purpose makes
rather simple model structure and easy to trace or validate. Multi purposes makes much complex
model and sometimes difficult to trace or validate. Also we think multi purposed model is hard to
keep maintaining and keep using planning processes.
Structure is also very important element to check the completeness of concept. It also related with
difficulty to keep maintain the model and using for planning processes. We suspected that complex
model such as KLUD, though which has single purpose, makes difficult to keep maintaining.
We try to trace as much resource we can but confessed that the information resources are much rely
on academic paper which sometimes drop off many related information such as back ground
information of the model builder and their true objectives or demand. For example, the Saitama
Model developed to check the effectiveness of SD model for planning processes as first objectives.
The model builder got the conclusion that SD simulation results are within 10% error of the statistic
data. Then we know the model was presented to Saitama Prefecture but even we try to trace the
subsequence of the Saitama Model but could not get much information. Model, program source
codes and recommendations are still in mist.
Table 3-1: Conceptual Validation
Name of the models
Consistency
Focus
Structure
Miyagi Dynamics
Model
Not enough information for
validating consistency in detail.
Multi purposes but seem to be
that main concern is future of life
standard rather than future of
industry or finance.
Composed with 6 components,
population, land use,
infrastructure, industry,
environment and life style.
Chiba Prefecture We think this model have perfect | Multi purposes and main Composed with 8 components,
Model feedback loops and keep objective of this model is to population and household,
consistency in detail. provide information for planning | public infrastructure,
staff. transportation and traffic,
We think this model has rather | environment, finance, industry,
difficulty to clearly focus on labor market and land use.
specific issues.
Narashino-shi Perfect consistency of feedback | Single purpose that only for Mainly simple 9 elements loops,
Population Model loop. calculate population of the population density, population,
Narashino City. household, land use, land usage,
land price, factory area,
commerce area and population
moving rate.
Saitama Model
We think this model have perfect
feedback loops but doubt to have
consistency in detail.
Suspected that use some linear
relation for simplify such as
using attractiveness multiplier.
Main purpose is not assist to
planning but validate the
effectiveness of SD techniques
on govemment planning.
Briefly composed with 4
sectors, industry, population,
finance, pollution and public
services.
Toyo Metropolitan Not enough information Not enough information Not enough information
Model
Kanagawa Model Not enough information Not enough information Not enough information
Shiga System Seem to be good feed back loop | Multi purpose but rather focus | Mainly composed with 5
Dynamics in overview level but do not have | on environment protection components, industry,
enough information for check the
detail.
issues,
population, land use, finance
and public services and the
environment centered on Biwa
Lake.
Name of the models Consistency Focus Structure
KLUD: Kyoto area | This model have perfect Single purposed focus on land | Composed with 6 sectors,
Land Use Model feedback loops but doubt to have | use only. population, household,
consistency in detail. commerce area, industry area,
Suspected that use some linear traffic and environment.
relation for simplify in finance
sector.
Osaka Model This model have perfect Multi purposed but focus on ‘Composed with 8 sectors, local
feedback loops both conceptual | providing information for policy | government policy, social
level and detail level. decision support. capital, investment plan, energy
and resources, industries,
population, economic growth
and environment,
Hyogo Dynamics Do not have enough information | Multi purpose Briefly composed with 4
Model sectors, population, industries,
resources and environment
pollution.
Hiroshima Model
Do not have enough information
Multi Purpose but rather focus
on transportation sector policy
and environment
Briefly composed with 2 level,
macro level and micro level.
Marco level handle industry and
population. Micro level
composed with three sectors,
traffic and communications,
Environment and life style. But
we feel this macro micro layer
structure may not accomplished
in this model because there are
no description about time delay
linkage between two layers.
We can conceptual validated only 5 models out of 11, Chiba Prefecture Model, Narashino-shi
Population Model, Saitama Model, KLUD and Osaka Model. Some of them are not completely
followed SD’s flows and stock loop relation concept. This is because they want to introduce the
simplification for avoiding the arguments on especially economic sector. However excluding that
point, these models are conceptually acceptable.
Also many of these models are multi purpose and aimed to provide the information for policy
decision-makers of planning division. But this multi-purpose does not mean to give some standard
form to the model. Every model has their own background needs and they change the structure of the
models for this background needs. For example, Shiga Model is multi purposes model but rather
focus on environment issues stand for this prefecture’s unique position. This prefecture has the
largest lake in Japan and the lake area shared nearly 50% of the area. Also this lake is main water
resource for Kyoto and Osaka and this prefecture has very heavy duty to protect from water
pollution.
All models’ structures are typical sector related model. Sector relations could be validated whether
loop is closed within the system. In this term, all models have closed loop with in the system as
sector relation and concept level. However, as we mentioned early, some of these models use
econometric techniques and this make linear relationship between parameters. We have many
experiences to know that the economic and financial relation sometimes has non-linear relation.
Also, the relation between the sector is so called non-hierarchy structure. Hiroshima City Model is
only model, which clearly try to identify macro level and micro level with hierarchy structure.
However, we feel they may not accomplished this macro-micro layer structure. As our experiences,
model builder must consider time delay of information between two layers, macro level and micro
level. It is sometimes very difficult to keep consistency and feel to insert some functionality to give
judgement after get some interim result on the simulation. For this purpose, we once make interim
result tables between the simulation for linkage of macro and micro level. Therefor, we think this
Hiroshima City Model may have non-hierarchy structure (flat structure) and may not consider time
delay mechanism. However, we think that, for basic purpose of the model, this model may no
necessary to have hierarchy structure.
We could not validate other models because shortage of detail information with in the model’
sectors.
3-2) Institutional validity
Connected with institutional resources such as annual budget, SD model can carry its validity into
the practical planning and implementation processes of local government policy.
As institutional validity, we select following criteria as validation elements:
- Initiatives: Who is model builder in term of concept development or concept determination
- Background demands: What kind of information they truly need?
- Recommendation: What kind of recommendations they provide and how it was affected to their
policy.
Initiatives, is very important even all models are actually developed by computer scientist or system
engineer. Some of these models are developed only for study purpose. For example, main purpose of
Saitama Model is for check whether SD is effective and could simulate real world. In this term,
initiatives are conduced by OR scientist but not local government staff. On the other hand, Miyagi
Dynamics Model and Hyogo Dynamics are good example that the government staff takes initiatives
for their long term planning.
Background demands are also so important for try to make in such models. Generally in those days
in Japan, local governments have mainly three big issues. One is population movement and
economic structure transform. Economic structure shift from agriculture based to industry based.
That makes people move from rural area to urban area and the government staff feel the needs of re-
develop of urban area. Second issue is environment protection. They afraid the air pollution and
water pollution. It is also connect with over population of urban area, they feel the need to build
more water supply and sewerage system, control the number of car and try to shift more public
transportation including bus and subway. Third issue is financial problem. The government staff was
afraided that they could not get enough tax income for conducting many infrastructure developments
and social welfare needs for growth of aged population. This treats are much strong especially
surrounded prefectures of Tokyo and Osaka such as Chiba Prefecture, Saitama Prefecture, Kanagawa
Prefecture which are neighbor prefectures of Tokyo, Hyogo Prefecture, Shiga Prefecture and even
Kyoto City which are neighbor prefectures of Osaka. They've afraid to keep balance of huge public
infrastructure development investment. However, Miyagi Prefecture does not have much in such fear
compare with other surrounded prefectures.
Strong recommendations also related with whether government staff take initiatives or not. If the
government staff takes initiatives of model building and simulation, the recommendations also built
into their planning processes including to keep sufficient budget for implementation. We validated
the results of recommendations with their new laws, regulations, policy statements and new projects.
This may be the best institutional validation if based on the previous two information, who take
initiatives and their background demands.
Table 3-2: Institutional validity
Name of the models Initiatives Background Needs Recommendation
Miyagi Dynamics Govemment staff: Planning Underpopulation problem that | Need policy considering rapidly
Model Department of Miyagi Prefecture | people move to Tokyo. age population growth, and
Govemment.
They make project team and
build the model.
‘Also search to give satisfaction
to the people keep living in the
prefecture.
balanced socio-economic
development of urban and rural
areas.
Environment protection policy
is very important to keep
maintain the level of life
standard and must accelerating
public infrastructure investment
including water supply an
sewerage system, public park,
traffic safety, safety consumer
products, etc.
Focus on social services to keep
sustain and up the life standards
for give more satisfaction to the
people living in the prefecture.
Chiba Prefecture Govemment staff: Planning Over population problem in bed | Not clear from the information
Model Department of Chiba Prefecture | town area and under population | resources.
Govemment problem in rural area.
Rapidly growth of industry
sector and pollution problem.
Narashino-shi ‘Academic side: Chiba Institutes | Over population problem and | Not clear from the information
Population Model of Technology necessity of urban development. | resources.
Saitama Model ‘Academic side: Japan Operations | Over population problem in bed | Not clear from the information
Research Academy town area and under population | resources.
problem in rural area.
Rapidly growth of industry
sector and pollution problem.
Toyo Metropolitan ‘Academic side: (Over population problem, Not clearly mentioned but
Model shortage of house, pollution, suggest the relation between the
shortage of water resource, and | number of university, GNP
so many typically on big cities. | growth water resources and
population.
Kanagawa Model Not enough information Not enough information Not enough information
Shiga System Govemment staff: Project team — | Pollution of water resources Need to take protection of water
Dynamics of Shiga Prefecture pollution.
Must consider financial
problem,
10
Name of the models
Initiatives
Background Needs
Recommendation
KLUD: Kyoto area
Govemment staff: Study team in
Needs of urban re-development
Need to develop public transport
Land Use Model Kyoto City Government to keep balance of historical system including subway.
environment. Need to control strict control of
Jand use under urban
development plan.
Osaka Model ‘Academic side: study team ‘Over population and heavy Environment protection could
pollution. be progress with effective
First oil shock makes economic | energy consumption policy and
recession. improve efficiency and
productivity.
However, social development
oriented policy and low
pollution policy may face up
with difficulty to keep balance
with high economic growth
Hyogo Dynamics Govemment _ Staff: Planing | Over population problem inbed | They recommend to prevent
Model Department of Hyogo Prefecture | town area and under population | good socio-economic welfare
problem in rural area. for aged people.
Rapidly growth of industry They also recommend
sector and pollution problem. protecting to change into slum
in urban area for urban re-
development with deeply
concemed environment
protection.
Water resources would be
shortage and recommend of
recycling sewerage water.
Hiroshima CITY Not clear Rather unique position in They do not recommend
Model industry area of the Seto Inland | anything, Just describe the
Sea.
Industrial pollution problem.
forecast based on the simulation
results.
On 5 models, the local government staff takes initiatives. In such model and simulation results gives
very clear recommendations. However, models take initiatives by academic side are not give so
much clear recommendations. Also these models seem to be much focus on providing the
information rather than recommendations.
Based on the models’ background, these models could categorized briefly into three types, urban
model such as Osaka model and Tokyo model, suburb model such as Chiba Prefecture Model,
Hyogo Dynamics and rural model such as Miyagi Dynamics. The local governments of these urban
models have every trouble the big cities have such as over population, urban re-development,
pollution and shortage of resources. The local governments of these suburb models have complicated
features. They have urban problems that big cities have such as over population, shortage of public
infrastructure but also have under population in their rural area. The local governments of these rural
models have under population problem and feels some necessity to give other satisfaction or value to
keep maintain the population or stop the under population.
Recommendations also clearly show the importance of government staff initiatives. They take much
clear recommendations and could take incorporate into their planning processes. We can know from
their new regulation such as environment protection laws, policy statement and new projects such as
rural area infrastructure development. For example, Shiga Prefecture takes strict water-pollution-
11
control law for protecting the pollution of the Shiga Lake. This law request to stop usage of synthetic
detergent. We believe that their recommendations makes the government's action to take in such
pioneering anti-water-pollution activities in Japan. Also, we believe that the recommendations of
Miyagi Dynamics Model keep behavior of the governments’ people's life style oriented policy in this
prefecture. In Hyogo Prefecture’s case, they also succeed to keep urban re-development with many
parks and museum, and keep balance of industry development in the prefecture. Kyoto City still
keeps balancing policy of socio-economic development and historical landscape. They also succeed
to build the subway. All of these cases could have sufficient budget to keep their policy
implementation.
Therefore, government staff's initiatives on model building and strong recommendation are key
factor of institutional validity.
3-3) Operational validity
Simulation results of above models were not always valid. But errors can be accrued defendable to
initial conditions, while intrinsic operational properties of the model remains.
As operational validity, we select following three criteria:
- Modification: Easiness to modify the structure of models
- Sensitiveness of initial conditions
- Accessibility to model or simulation environment
Modification or easiness to modify the structure of the model is very important element to keep
using the model on their planning processes. Log term planning basically not routine work but pile
of ad-hoc type works for planner. They need to re-calculate for specific issues suddenly as ad-hoc
work. Many times, they feel to modify the model structure of model components. Especially the
model partly using simplification needs to modify or update much frequently. Easiness of
modification is better for operational purpose. If the model does not allow the modification, in some
days, it may not used anymore.
Sensitiveness of initial conditions mainly come from the model's complexity. More complicate, the
bigger chance to falls into sensitiveness of initial condition. Sometimes, operator could not get
suitable data but sensitiveness of initial condition may cause the difficulty to make try and error
initial condition settlement. It makes also barrier of modification and accessibility.
Accessibility is also sometimes makes barrier to keep using the model for planning processes. Many
cases, planner do not have much time for arrange the simulation. If they could not get the simulation
12
results immediately, they rather use statistical data and make simple estimation. Linear statistical
estimating data can easily get using simple calculator or spread sheet. Accessibility depended on two
elements, whether the initiatives of the model building carried by the planning section of the
government staff or academic side. If the government staff take initiatives, they try to keep using for
their planing and therefore try to keep accessibility to the model. Second is computer language and
availability of the source code. If they keep having source code and compiler, it also increasing the
accessibility.
Table 3-3: Operational validity
Name of the models Modification Sensitiveness Accessibility
Miyagi Dynamics Not enough information but Not enough information Not enough information but
Model model was composed with initiatives conducted by planing
matrix described by APL
computer language and other
part describes by GSMP III. We
feel in such model may not
easily make modification in
necessary of planning.
However, they disclose stock and
flow diagram and it may possible
to modify on this diagram level.
section of the govemment.
We feel that the staff may need
official procedure to use and
feel not so easy to access,
Chiba Prefecture Easy to modify for disclose of _ | No information but the model is _| No information
Model detail model. But we feel the | so complicated and we fee! this
government staff may not do for | model may have sensitiveness to
complexity of the model, initial condition.
Narashino-shi Easy to modify because source | Original structure of the model is | Easy as same reason mentioned
Population Model codes and model structures are | simple and may not have so at modification section,
open and model is simple and
clear to understand. Even though
the model can also possible to
expand and include other sector
but we feel the government staff
do not do that for the academy
take initiatives on development
of the model,
much sensitiveness to initial
condition,
However, initiatives lead by
academic and we feel that the
government staff does not use
this model so much on their
planning,
Saitama Model
Easy fo modify for disclose of
detail model. But we feel the
government staff may not do for
academy take initiatives on
development of the model.
Not enough information
Not enough information on
accessibility but source codes
are disclosed.
Toyo Metropolitan Not enough information Not enough information Not enough information
Model
Kanagawa Model Not enough information Not enough information Not enough information
Shiga System Not enough information Not enough information Not enough information but the
Dynamics model describes with
DYNAMO.
KLUD: Kyoto area
Possible but we feel difficult to
Temay face up the sensitiveness
Not enough information but it
Land Use Model modify. Also hard task to modify | of initial condition. may be possible for many
financial sector for adapt models information are
industrial structure disclosed.
transformation.
Osaka Model Possible Original structure of the model is | Possible and the model
simple and may not have described by with FORTRAN
sensitiveness to initial condition. | language
Hyogo Dynamics Not enough information Not enough information Not enough information
Model
Hiroshima City Model
Not enough information
Not enough information
Not enough information
We could not get enough information for validate modification criteria but 6 models, Miyagi
13
Dynamics Model, Chiba Prefecture Model, Narashino-shi Population Model, Saitama Model, Osaka
Model and KLUD are seem to be easy to modify. They disclose the detail relation loop models and
some of them are even disclosed the stock and flow diagram too. If accessibility is quite good
condition, the government planning staff may keep using and maintaining as the part of planing
processes. However some of them have complicated structure for modification such as Miyagi
Dynamics Model. It may also comes from the limitation of computer abilities in those days and
availability of DYNAMO, many model describes with other computer language including
FORTRAN, APL and GSMP III. FORTRAN is popular in those days and could find many system
engineer who know well about FORTRAN language but we feel it may be not so easy to find the
system engineer who knows other computer languages such as GSMP III.
Also we have very limited information about the sensitiveness of initial condition and we have no
way but validate from the complexity of the model’s structure. If we could have chance to get
complete equations and source codes of the models, we hope to validate by our own computer
someday. Anyway, the simple and small model such as Narashino-shi Population Model and Osaka
Model obviously seem to have not in-composed with any trouble part which may fall into chaos.
Availability may key issues to keep using the model for planing processes. This matter comes from
three elements, easiness of modification with simple enough and still have necessary functionaries,
initiatives of the governments staff for the model building, and ideally initial data re-setting and
source code compiling can do on their personal computer. Unfortunately there are no model meet in
this condition. Osaka Model may be said only one if insisted to select because FORTRAN may
rather easy to translate into BASIC. The reason that suddenly these models do not use any more in
1980s may comes from such reasons (difficulties of modification and compiling). Models that
include the diagrams with flows and stock may also rather easy to translate into other SD simulation
software such as STELLA. Miyagi Dynamics Model, Saitama Model, Kyoto area Land Use Model,
Osaka Model and Narashino-shi Population Model disclose in such type of diagram. Chiba
Prefecture models also disclosed relation diagram and may also possible to translate into other SD
Models.
3-4) Other validation
We also try to validate available statistic data with their forecast. However, the difference does not
mean straightly existed some errors conceptually, institutionally and operationally.
14
3-4-1) Miyagi Dynamics Model (target in 1995)
Table 3-4: Miyagi Dynamics Model and Statistical Data
Simulation Standard Simulation Case T Statistics
Population (thousand) 2,783 2,786 2,338")
Population age 0-14 21.9% 21.9% 16.5%
Population age 15-64 67.2% 67.2% 68.5%
Population age 65- 10.9% 10.9% 15.1%
Household (thousand) 876 876 787*2)
Population Density*3) 381.7 382.2 321
Labor population *4) 1,285 1,287 T164
Agriculture 10.9% 10.8% 8.2%
Industry 29.6% 29.8% 27.5%
Services 59.5% 59.4% 63.9%
GNP (100 million Yen) 2 Ad6 $434 64,000*5)
Agriculture 2,025 2,021 2,699
Industry 13,420 13,446 36,923
Services 27,000 26,968 26,815*5)
*1) population in 1996
*2) household in 1997
+3) person per square kilometers
*4) thousand
+5) GNP in 1994
*) GNP of simulation use price index in 1970
*) Standard case consider the limitation of green area and try to keep more than 75% but CASE I do not consider any conditionality
of green rate.
*) Consumer index price in 1975 is 55.3 as 100.0 in 1997 seem to almost double.
Miyagi Dynamics Model seems to have under estimate on growth of aged population. This growth
of aged population (over 65 years old) goes so rapidly especially in rural area in Japan. Also they
under estimated the GNP growth. (In this case actually mean gross prefecture product. Same concept
of GNP which much less than GDP.) It may affected the economic recession since middle of 1990s
but SD model builder thought the economic growth would be goes as linear relation.
3-4-2) Chiba Prefecture Model (target in 2000)
Table 3-5: Chiba Prefecture Model and Statistical Data
‘Simulation Statistics
Population 6640 5,824*1) | Thousand
Population age 0-14 19% 15.4%
Population age 15-64 67% 72.8%
Population age 65- 14% 11.7%
Household 2,270 2,092*2) | thousand
Labor population 1,988 2,992 | thousand
Agriculture 8% 4.6%
Industry 26% 27.5%
Services 66% 67%
GNP 129,000 181,662*3) | 100 million Yen
Agriculture 3,627 4,850"4)
Industry 143 116,655*4)
Services 81,230 60,355°3)
*1) in 1996
15
*2) in 1997
*3) in 1994
*4) in 1995,
+) GNP based on price in 1975
Chiba Prefecture Model shows very good forecast except economic growth. They also over
estimated the economic growth.
Unfortunately, we could not have enough forecast information compared with statistical data on
other models. Saitama Model is for validation purpose with SD simulation and statistic data and we
decide not validate this time.
4, Conclusion
We are happy to introduce the Japanese local government models developed during 1970s and could
have chance to validate. As lessons from these models and the effort of the pioneers, following
maxims may conclude though someone may said already.
- Focus on single purpose and/or make simple model
- Make simple structure model but avoid to making hybrid structure
- Take initiatives of the government staff for model building
- Need to support their demands and recommendations
- To disclose all information for maintaining the model or/and running simulation on other computer
We though that we may find some general or standard model form through comparison and
validation of the old Japanese local government models. Every model has similar structure of several
sectors linkage block diagram. Selected sectors and elements are also very similar. However, small
change comes affected their background demands even though all of them start similar concept.
Validation results show much strong recommendation comes from the model and simulation results
the government staff takes initiatives of model building. Also validation results show that in such SD
model and simulation results support their background needs and recommendations.
5. Bibliography
- A Computer Assisted Long Range Comprehensive Planning System for a Regional Project Impact
Assessment in Land Use Planning, draft by T. Shiina et al. for IFAC Sixth Triennial Congress,
Boston/Cambridge, Aug. 1975.
16
CC BY-NC-SA 4.0