Beruflich Dokumente
Kultur Dokumente
Suggestions for improvements to the T21-SF and to this Documentation are welcome and should be addressed
to Matteo Pedercini: mp@millennium-institute.org
Introduction
Purpose of the model
The Threshold21 (T21) is a System Dynamics based model that has been designed as a
support tool for national development planning. T21 is structured to analyze medium-long
term development issues at the nationwide level. The model integrates in a single framework
the economic, the social, and the environmental aspects of development planning. The level
of aggregation used make it ideally suited to look at resources allocation issues among
different ministries. T21 is conceived to complement budgetary models and other shortmedium term planning tools by providing a comprehensive and long term perspective on
development.
T21 is useful at four levels in the planning process. First, the participatory process of model
development provides insights on the coherence and consistency of objectives, hypotheses
and data used for policymaking in different sectors. Second, the base run simulation of the
model offers an outlook into the key development issues the country/region might face in the
future. Third, alternative scenarios provide an understanding of how different strategic
choices or external conditions can impact future development, and how policies
synergistically interact. Fourth, the resulting development plan provides a clear basis for
action in the various sectors, as well as for monitoring and evaluation of performance.
History of T21
Millennium Institutes plan from the beginning was to produce a computer -based national
development planning model consisting of a set of dynamically integrated sectors that
together would be adequate to represent the long term development of most countries,
industrialized and developing. As a step in this direction, we reviewed the literature on
models for planning national development and assembled a library of hundreds of such
models. Millennium Institute (MI) then studies some of the most interesting and useful of
these models and published a book cataloging about fifty of the most interesting and useful
models 2. The Threshold 21 (T21) model was based on this detailed review of the
development modeling literature. For example, MI has studied carefully the literature relating
to:
Gerald Barney, W. Brian Kreutzer, and Martha J. Garrentt, Managing a Nation: the Microcomputer Software
Catalog, Institute for 21st Century Studies and Westview Press, 1991.
National Environment models from the Dutch National Institute of Health and
Environment (RIVM) and other organizations
National Climate -Economy models from Yale University and other organizations
Agricultural production models from the UN Food and Agriculture Organization, the
US Department of Agriculture, and other organizations
Many of these models are not dynamic, integrated, or long-term, and MI specifically wanted
these characteristics in its model. It was possible to learn general concepts for earlier models,
but it was necessary to formulate T21 from scratch.
The earliest version of the T21 model was developed by Dr. Bob Eberlein, Ventana Systems,
Harvard, MA as a gift to MI. We at MI acknowledge and thank Dr. Eberlein and Ventana for
this gift.
Dr. Weishuang Qu then picked up the T21 development work and has led the work for more
than a decade. In the last few years, he has had able support from Dott. Matteo Pedercini.
For many years T21 model has been a work in progress. The current version is based on
nearly twenty years of study, research, development, application, testing, and revision for
twenty countries. The history of country applications of the model is summarized in Error!
Reference source not found..
Table 1: Chronology of T21 Country Applications and the Sectors Added with the
Applications
Date
1994-6
1996-7
1997
1997
1997
1999
1999
1999
1999
2000
2001
2001
2002
Italy
China
Malawi
Mexico (alpha)
USA (beta)
Somaliland (beta)
Ghana (alpha)
Guyana
Bhutan (beta)
2003
2004
Mozambique
Saint Lucia
2002
2003
2004
Papua, Indonesia
Cape Verde
USA
2004
Ghana
2005
2006
Mali
Jamaica
Note: Alpha models are preliminary models based on international data only. Beta models have had the benefit
of some interaction with professionals in the country, but are not fully refined applications.
The T21 model has evolved over two decades as applications to individual countries were
developed, tested, modified, and refined. The clients included ministries of finance, ministries
of planning, academic institutions, UN agencies, and the World Bank. New clients needed
new sectors added, for example health, nutrition, education, energy, water, etc. As these
sectors were developed, they were added to one of the country specific applications of the
model. Generally new country-specific applications were developed from the version of the
model most recently completed.
By the early 2000s, T21 had been customized for nearly twenty developing and industrialized
countries (see Error! Reference source not found.), and the models basic structurewith
customizationwas found to well represent countries as diverse as Malawi, Bangladesh,
China, Italy, and the United States. It was at this time that the MI Board of Trustees of
Millennium Institute (MI) asked the MI staff to implement the original vision for the model,
namely (1) that there be a T21 model (called the Starting Framework) from which all future
T21 country-specific applications would be customize d and developed; and (2) that following
the completion of a country-specific application, the T21 Starting Framework Model be
updated with any developments that are in the future to become standard in T21 models. This
change has now been made, and the resulting model is the T21-Starting Framework (T21-SF).
This document represents the official documentation of T21-SF in its generic (not countryspecific) version.
Not all country-specific applications of the T21 model will include all of the sectors in the
T21 Starting Framework (T21-SF). For example, if a country has no fossil fuels, the countryspecific T21 application will omit the fossil fuel energy sector. Generally, however, most of
the sectors in the T21-SF w ould appear in each country-specific application.
This volume provides the Explanatory documentation of the model, which highlights T21s
general structure and the reasoning behind the modeling decisions for each sector. In addition
to this documentation, the source code contains an equation-by-equation documentation for
each equation and variable. The key equations for each sector are described in the
Explanatory documentation, while all the equations are described in the equation-by-equation
documentation.
b)
Represents the important elements of complexity feedback relationships, nonlinearity and time delays that are fundamental for proper understanding of
development issues;
c)
d)
e)
f)
The model provides policymakers and other users with a general direction of the
consequences to be expected from pr esent and alternative choices. T21 includes linkages and
feedbacks among all the components in the three spheres of economy, society, and
environment. These inter-component connections are critically important in a model being
used by policy makers to guiding a country towards a sustainable future. For instance, water
limitations may affect the production of food, which in turn will affect the health of workers
and the productivity of the economy, which will further affect government revenue,
household consumption and savings, and international trade. With T21, policymakers of
different ministries of a country can work together, observe the whole picture of how a
decision in one part will affect other parts, and decide which policy will improve best the
overall health of a nation.
Structural Overview:
Model boundaries and time-horizon
The Threshold 21 Starting Framework is a generic structure that represents development
mechanisms that can be found in most developing and industrialized countries. As such, it
covers a broad range of issues that countries all over the world face: from poverty to
environmental degradation, from education to health, from economic growth to demographic
expansions. In other words, T21 SF is designed to cover the most common long-term issues
countries encounter in the development process.
As the model has very broad scope, it is particularly important to identify its boundaries, in
order for any potential user to assess its suitability for analyzing the issues of interest.
Endogenous, exogenous and excluded variables
A first level of boundaries defines what variables are considered endogenous , exogenous or
excluded from the model (see Figure 1). Variables that are considered an essential part of the
development mechanisms object of the research are endogenously calculated. For example,
the Gross Domestic Product (GDP) and its main determinants, population and its main
determinants, and the demand and supply of natural resources are endogenously determined.
Geographic boundaries
The focus of T21 SF is on the specific country being analyzed. A lthough the model can also
address the impacts of developments in the rest of the world (e.g., oil prices, changes in
commodity prices) on the country, the model is centered on the internal issues of the country.
The model can therefore address questions of what a country can do to help further the
development of its ow n progress and sustainability.
As illustrated in Figure 2, the model also determines outputs from the country to the rest of
the world, e.g. CO2 emissions. However, the small country assumption is made, and the
performance of the country is assumed to have no relevant effect on the development in the
rest of the world. For example, oil prices are generally considered exogenous, as well as other
commodities prices. This assumption is generally relaxed when the country being analyzed
has a particularly relevant role at world level with respect to one of the issues analyzed. For
example, when analyzing energy problem in U.S.A., oil prices are considered endogenous.
several thousands feedback loops. Given the size and the level of complexity of the model, its
structure has been reorganized into smaller logical units, labeled as modules.
A module is a small piece of structure which internal mechanisms can be understood in
isolation from the rest of the model3. The size of this piece of structure is chosen also
considering the amount of information that the user can take in at once, and so as to fit the
size of standard computers monitors (all modules fit in a 1024x768 screen). T21 SF is
composed of 37 modules.
The 37 modules composing T21 SF are grouped into 18 sectors: 6 social sectors, 6 economic
sectors, and 6 environmental sectors. Sectors are groups of one or more modules based on
their functional scope. For example, the water sector groups the water demand and water
supply modules; and the education sector groups the primary education and secondary
education modules.
Society, economy and environment are known as the three spheres of T21. All sectors in T21
SF belong to one of the three spheres, depending on the type of issue they are designed to
address. Modules are built to be in continuous interaction with other modules in the same
sector, across sectors, and across spheres.
Table 2 lists the modules of T21 SF and the sectors they belong to, for each of the three
spheres: society, economy and environment. The modules listed below are not normally
sufficient to carry out a detailed country-specific analysis: in most cases additional modules
have to be introduced to capture the particular reality of a country. Similarly, some modules
may need to be eliminated during the customization process because not relevant to a specific
countrys situation. T21 SF should thus be considered as a starting point for creating a fully
customized national model for development planning, and not as a rigid framework.
Examples of modules that are very often added to the initial structure are the indicators
modules. These are modules created to calculate or organize specific indicators, including but
not limited to MDGs, HDI and GDI. These and many other are ready on-the-shelf modules to
be linked to the starting framework as needed.
Table 2: Modules, Sectors and Spheres of T21 SF
SOCIET Y
ECONOMY
ENVIRONMENT
Population Sector:
1. Population
2. Fertility
3. Mortality
Education Sector:
4. Primary Education
5. Secondary Education
Health Sector:
6. Access to basic health care
7. HIV/AIDS
8. HIV children and orphans
9. Nutrition
Infrastructure Sector:
10. Roads
Labor Sector:
Production Sector:
14. Aggregate Production and Income
15. Agriculture
16. Animal husbandry-fishery-forestry
17. Industry
18. Services
Technology Sector:
19. Technology
Households Sector:
20. Households accounts
Government Sector:
21. Government revenue
22. Government expenditure
23. Public investment and consumption
24. Gov. balance and financing
Land Sector:
30. Land
Water Sector:
31. Water demand
32. Water supply
Energy Sector:
33. Energy demand
34. Energy supply
Minerals Sector:
35. Fossil Fuel production
Emissions Sector:
36. Fossil Fuel and GHG emission
Sustainability Sector:
37. Ecological footprint
As it is emphasized later on in the text, although it is possible to understand the internal mechanism of a
specific module in isolation from the rest of the model, fully understanding its functioning and relevance requires
studying its role in the whole models structure.
11. Employment
12. Labor Availability and Cost
Poverty Sector:
13. Income distribution
The scope of each module, as well its affiliation to a specific sector, can be easily deducted by
the name used to identify it. In the following chapters of this documentation, a detailed
description of the particular assumptions and formulations used in each module is also
provided.
The structure of the individual modules is based on well accepted work in the field,
translated in stock and flow language by MI modelers and integrated with ad-hoc research.
The major distinctive characteristic of T21, however, lies in the way the various modules are
linked together, forming a complex network of feedback loops which is the determinant of the
models behavior.
A comprehensive map containing all the cross-module linkages in T21 SF would be way too
extensive and complex to bring any information to the reader. Therefore, in this
documentation we decided to present those linkages in two different ways: first, in the
documentation of each module, incoming and outgoing linkages are specifically identified;
second, in the following section of this chapter the major cross-sector and cross-sphere
linkages are described.
Society
Economy
Environment
production
population
labor
investment
health
society
economy
infrastruc
ture
education
technology
government
poverty
households
row
land
energy
water
environment
emissions
minerals
sustainabi
lity
Social Sectors
The social sectors and contained modules of T21 SF are listed in Table 3. The next pages
present the structure of each module detail.
Table 3: The social sectors of T21 and corresponding modules
Society
Population Sector:
1. Population
2. Fertility
3. Mortality
Education Sector:
4. Primary Education
5. Secondary Education
Health Sector:
6. Access to Basic Health Care
7. HIV/AIDS
8. HIV Children and Orphans
9. Nutrition
Infrastructure Sector:
10. Roads
Labor Sector:
11. Employment
12. Labor Availability and Cost
Poverty Sector:
13. Income distribution
1. Population module
MIGRANTS' SEX
DISTRIBUTION
INITIAL
POPULATION
MIGRANTS' AGE
DISTRIBUTION
NET MIGRATION PER
1000 HABITANTS TIME
SERIES
net migration
total
population
total adult
deaths
<Time>
Population
<births>
secondary student
crude death rate
primary student
crude death rate
<deaths>
total deaths
under five
mortality rate
<TIME STEP>
infant mortality
<DEATH RATES
TABLE>
effective death rates
per age group
<Time>
population cohort
shift
population group
population under
five
share of population
per gender
average life
expectancy
measured le
life expectancy
The flexibility of usin g one-year age cohorts in T21 makes for easy application of the model
in any country and does not significantly complicate the programming or data input.
Explanation
Major Assumptions
All births happen at the first of the year
The aging process (the shift of individuals from one age cohort to the next) happens
only at the first of each year
Net migration is exogenously determined
Immigrants have the same fertility and mortality behavior as the rest of the population
Input Variables
Variable Name
Births
Deaths
Death rates table
Module of Origin
Fertility
Mortality
Mortality
Output Variables
Variable Name
Population
Total population
Destination Module
Same Sector
Other Social Sectors
Fertility,
Primary Education,
Mortality
Secondary
Education,
HIV/AIDS, HIV
Children and
Orphans, Labor
Mortality
Access to Basic
Health Care, Income
Distribution,
Nutrition,
Life expectancy
Adult crude death
rate
Primary student
crude death rate
Secondary student
crude death rate
Under five mortality
rate
Constants and Table Functions
Variable Name
Initial Population
Migrants sex distribution
Migrants age distribution
Net migration per 1000 habitants time series
Other Sectors
Primary Education,
Secondary Education
Industry
Primary Education
Secondary Education
Fertility
Type of Variable
Constant
Constant
Constant
Time Series
Functional Explanation
The T21 Population module has one stock variable, the population stock, which is affected by
three flows: births, deaths , and net migration .
The Population Stock
The population stock is disaggregated into 2 genders and 82 age-cohorts. Modeling the agecohort system and aging process are conceptually straightforward, as illustrated in Figure 6.
In T21, a single equation shifts the population in each age cohort (minus cohort deaths) to the
next age cohort at the end of the year, except the cohort over 80. For this cohort, deaths are
subtracted and last years age 79 cohort is added.
Births
Womb
of
Society
0-1
1-2
>80
The function INTEG performs the mathematical process of integration (accumulation), and
the subscript new born refers to the first age cohort, or the cohort of infants born within that
year (societys womb).
Since no one is born directly into any of the other age cohorts, the second equation (which
applies to all other age cohorts) is as follows:
Population [sex, age over 0] = INTEG (-deaths [sex, age over 0]+net migration[sex,
age over 0], INIT POPULATION[sex, age over 0])
in which age over 0 refers to the categories age 0 to age over 80.
The age-shifting equation is as follows:
When the condition MODULO(Time,1)<TIME STEP/2 is true (when Time reaches the end of
each year), the subscripted variable population shifts from one age cohort to the next. The
second to last parameter of the function (the value 1) causes the function to add (not replace)
the age 79 cohort (Year Ago 79) to the last cohort age over 80 (Year Ago Over 80). The last
parameter of the function (the value 0) is used to replace the first age cohort, i.e., to set the
first cohort to zero at the end of each year to store new births the next year.
The variable initial population specifies the value for the stock of population for both males and
females in each of the 82 age cohorts when the simulation begins, such as 1980 or 1990. The
first age cohort is initialized at zero at the beginning of the simulation. It is like the womb
of society and is used to acc umulate new births each year.
The flows
Details about the calculation of births are in the Fertility module section.
Details about the calculation of deaths are in the Mortality module section.
Other components
In the Population module, three fundamental variables are calculated: under-five mortality, infant
mortality, and life expectancy.
The first two variables are calculated per the World Bank definition:
Under five mortality rate = SUM (deaths [sex!,under 5!]) * 1000/SUM (births[sex!])
Infant mortality = SUM (deaths [sex!, AGE 0]) /SUM(births[sex!]) *1000
Life expectancy is calculated from effective deaths rates per age group , by applying the inverted
death rates table used to calculate deaths in the Mortality module. If the indicated life expectancy
is the only variable affecting the death rates per age, then life expectancy is equal to indicated life
expectancy. If other variables affect death rates per age, then the effect of these variables will
be reflected in the difference between life expectancy and indicated life expectancy.
In the top right corner of the sketch, crude death rates for adults, primary and secondary
students are calculated. These three variables are calculated in a similar way; the sum of
deaths for a certain age group divided by the total number of individuals in that group. For
example, the adult crude death rate is calculated as:
Adult crude death rate = total adult deaths/SUM(Population[sex!,adult age!])
These three crude death rates are an important indicator of demographic development and are
used in the Production (adult crude death rate) and Education sectors (primary and secondary
student crude death rates).
References.
Shorter, F.C., R. Sendek, and Y. Bayoumy, Computational Methods for Population
Projections, New York: The Population Council, 1995.
Hughes, B., International Futures (IFS) Documentation, Working Draft 1, 1997.
Sehgal, J., An Introduction to Techniques of Population and Labor Force Projections,
International Labor Office, Geneva, 1989.
US Immigration and Naturalization Service, 1993 Statistical Yearbook of the Immigration and
Naturalization Service, Washington: US Government Printing Office, 1993.
A. Wils, 1996, PDE-Cape Verde: A Systems Study of Population, Development, and
Environment, Laxenburg, Austria: IIASA, WP-96-009.
2. Fertility module
<average adult
literacy rate>
EFFECT OF ADULT
LITERACY ON DESIRED
NUMBER OF CHILDREN
TABLE
<Perceived Relative Pc
Disposable Income>
EFFECT OF ECONOMIC
CONDITION ON DESIRED
NUMBER OF CHILDREN
TABLE
INITIAL DESIRED
NUMBER OF CHILDREN
PER WOMAN
TIME TO PERCEIVE
CHANGE IN UDER FIVE
MORTALITY RATE
<under five
mortality rate>
<adult literacy
rate>
PROPORTION OF
WOMEN HAVING OUT OF
WEDLOCK SEX
PROPORTION OF
WOMEN MARRIED
desired number of
children per woman <Population>
consciously controlled
fertility rate
unconsciously
determined
fertility rate
contraceptive
prevalence rate
proportion of
babies per sex
sexually active
women
total fertility rate
effective contraceptive
prevalence rate
total births
births
Explanation
Major Assumptions
That Total Fertility Rate (TFR) has two components: (1) a consciously controlled part
related to education level, desired family size, and contraceptive use and availability
and (2) an unconsciously controlled part related to cultural factors;
Fertile females can be classified into two types, each with its own fertility rate: (1)
those who practice conscious birth control, and (2) those who do not;
The proportion of females who practice conscious birth control changes over time.
The major factors affecting this change are the female adult literacy rate and, if
implemented, family planning programs;
The fertility rate for females who do not use birth control is the natural fertility rate
which changes with female life expectancy;
The fertility rate for those who practice conscious birth control changes primarily in
response to the trend of societys desired family size;
The variables that affect desired family size are per capita real GDP and male and
female adult literacy rates;
The probability distribution underlining the age specific fertility rates is constant over
time;
The fraction of women using conscious birth control is considered the same across all
age cohorts.
Input Variables
Variable name
Population
Under-five mortality rate
Adult literacy rate
Average adult literacy rate
Perceived relative pc disposable income
Module of Origin
Population
Population
Primary Education
Primary Education
Households Accounts
Output Variables
Variable Name
Births
Total fertility rate
Constants and Table Functions
Variable Name
Age specific fertility distribution
table
Initial contraceptive prevalence
rate
Effect of female literacy rate on
contraceptive prevalence rate
table
Initial adult female literacy rate
Unconsciously determined
fertility rate time series
Destination Module
Same Sector
Other Social
Sectors
Population
AIDS Children
Other Sectors
Type
Table
function
Constant
Table
function
Constant
Time series
US Census Bureau
Constant
Table
function
Constant
Constant
EDSTATS
Data on total fertility rate (UN Population
Division) and contraceptive prevalence rate
(US Census Bureau)
Data on total fertility rate (UN Population
Division) and under five mortality rate (UN
Population Division)
Data on income (IMF) and total fertility rate
(UN Population Division)
Local Experts
Local Experts
Constant
Constant
Local Experts
Data on births (UN Population Division)
Table
function
Constant
Functional Explanation
The T21 model distinguishes between two types of fertile females, those who use birth control
and those who do not. The size of each group is determined by the size of female population
cohorts and by the variable proportion using conscious control . Both of these variables change
over time. Those who use birth control have a consciously controlled fertility rate , and those who
do not use birth control have an unconsciously determined fertility rate.
The literature review (Appendix A) found seven major elements affecting fertility. These
elements relate to natural factors (length of lactation, coital frequency, infant mortality) or to
direct birth control factors (contraceptive usage and contraceptive effectiveness) or to indirect
birth control factors (economic conditions and education). Some other causal factors
mentioned in the literature were not included in the model because they were either relatively
less significant in most cases (nutrition and health), or relatively non-variant (age of first
menarche, age of menopause, duration of fertile period), or non-conclusive (religious,
political, and economic values). 4
The equation for the total fertility rate (TFR) follows from the causal relationships in Figure 7.
The equation, which is the weighted sum of controlled and natural fertility rates, is as follows:
Total fertility rate= consciously controlled fertility rate*effective contraceptive
prevalence rate+ unconsciously determined fertility rate*(1-effective contraceptive
prevalence rate)
The seven elements identified in our literature review are all considered in the calculation of
the total fertility rate in T21. In particular:
Length of lactation and coital frequency are implicitly considered in the estimation of
the unconsciously determined fertility rate;
Economic conditions (relative pc disposable income), education (adult literacy rate ) and
infant mortality ( under five mortality rate) are all used to determine the consciously
controlled fertility rate;
Education also affects the proportion using conscious control. From the literature, the
decision whether to use conscious control is primarily made by the female, while
decisions regarding family size involve both sexes. Consequently, female literacy rate
is used to affect proportion using conscious control, and average literacy rate is used to
affect desired number of children per woman;
Contraceptive effectiveness is used to determine the effective contraceptive prevalence
rate;
The variable contraceptive prevalence rate has a numeric value at any point in timesuch as
0.35, which means at that time 35% of females are using conscious control. This
percentage is the overall average for all females between the ages 15 and 49. Actual values of
contraceptive prevalence rate for different age cohorts can be different for different age groups,
such as 50% for age 30 and 10% for age 15. In T21-SF, a simplifying assumption was made,
namely that the fraction of women using conscious birth control is constant over all age
cohorts. This means that a single value of contraceptive prevalence rate can be applied to all
female childbearing cohorts.
These other causal variables can be included on a case by case basis if they are needed in some country -specific
applications of the T21 -SF.
The proportion of the fertile population using conscious control (contraceptive prevalence rate)
depends on the female adult literacy. The equation is as follows:
contraceptive prevalence rate = INITIAL CONTRACEPTIVE PREVALENCE
RATE*effect of female literacy rate on contraceptive prevalence rate
The effective contraceptive prevalence rate also considers the effectiveness of birth control
methods:
Effect contraceptive prevalence rate = contraceptive prevalence rate*
EFFECTIVENESS OF CONTRACEPTION METHODS
The consciously controlled fertility rate is determined by three variables: desired number of
children per women, perceived under five mortality rate and effectiveness of birth control , as shown
in the following equation:
consciously controlled fertility rate = (desired number of children per
woman*(1+Perceived Under Five Mortality Rate/1000))/EFFECTIVENESS OF
BIRTH CONTROL
The fertile female population is made up of all women married or sexually active outside of
wedlock, as described in the following equation:
sexually active women [ag childbearing] = Population[FEMALE,ag
childbearing]*(PROPORTION OF WOMEN MARRIED+PROPORTION OF
WOMEN HAVING OUT OF WEDLOCK SEX)
Desired number of children per woman is
To convert TFR into births, we use the variable age specific fertility distribution table.5 This
distribution is actually the probability density function indicating the distribution of births
over the population of fertile women. Thus, births are equal to the total fertility rate
multiplied by the fertility distribution multiplied by the total number of females of child
bearing age, multiplied by the proportion of babies per sex, as shown in the following
equation:
births[sex] = total fertility rate*SUM(sexually active women[ag childbearing!]*age
specific fertility distribution table[ag childbearing!])*proportion of babies per
sex[sex]
References.
Shorter, F.C., R. Sendek, and Y. Bayoumy, Co mputational Methods for Population
Projections, New York: The Population Council, 1995.
3. Mortality module
INITIAL MEDIUM TERM
AVERAGE PC INCOME IN
USD PPP
<real pc gnp in
usd in ppp>
<quality of
nutrition>
LOCAL CONDITION LE
ADJUSTMENT
PARAMETER
NORMAL LIFE
EXPECTANCY
TABLE
MALE FEMALE LE
DIFFERENCE
EFFECT OF NUTRITION ON
LIFE EXPECTANCY TABLE
<fraction of water
available>
indicated life
expectancy
DEATH RATES
TABLE
EFFECT OF WATER
AVAILABILITY ON LIFE
EXPECTANCY TABLE
<average access to
basic health care>
non-aids death
rates per age
<Population>
Demographers at the Population Council found that the pattern of age specific death rates vary slightly in
different regions of the world. Specifically, they found that there are four patterns: (a) West (base); (b) East
(higher infant mortality rates than in west, increasingly high rates over age 50, and lower for other ages, relative
to West); (c) North (lower infant mortality rates, low rates beyond 45 or 50, and higher for other ages, relative to
West); and (d) South (higher mortality rates for ages under 5, lower mortality between 40 -65, and higher over
65, relative to West). We have tested these four life tables in ten countries, and they have proven to be
extremely accurate.
tables. In T21, we developed a method to deal with HIV/AIDS and other specific factors
affecting mortality only for some age cohorts, as explained below.
The life tables included in T21 were developed at the Population Council. 7 The tables have
proven highly accurate for several country applications of T21 (for example, China, USA,
Mexico, Italy, and Malawi) when life-table-based projections were compared with historic
data for the period 1980 to 2000.
Explanation
Assumptions
Variations in GNP per capita can explain major changes in life expectancy at birth;
Variations in the level of health care, nutrition, availability of water and pollution also
effect life expectancy;
Cohort death rates can be determined by indicated life expectancy using a death rates
table and adding abnormal demographic conditions, such as deaths from AIDS or other
epidemics;
HIV/AIDS (which primarily kills people at young adult and working ages) is an
abnormal demographic condition, and AIDS-related deaths are modeled as a departure
from the normal.
Input Variables
Variable Name
Real pc GNP in USD in PPP
Population
Average access to basic health care
Quality of nutrition
Fraction of water available
CO2 emissions per hectare
AIDS children deaths
AIDS deaths
Module of Origin
Aggregate Production and Income
Population
Access to basic health care
Nutrition
Water demand
Fossil fuel and GHG emissions
HIV children and orphans
HIV/AIDS
Output Variables
Variable Name
Deaths
Destination Module
Same Sector
Other Social Sectors
Population
Education
Type of Variable
Table function
Table function
Table function
Other Sectors
Coale, A.J., and Demeny, P., Regional Model Life Tables and Stable Population, second edition, Academic
Press, 1983.
Division)
Time for changes in access to basic Constant
health care to affect life expectancy
Time for income changes to affect
life expectancy
Constant
Constant
Table function
Table function
Table function
Constant
Constant
Constant
Table function
Functional Explanation
Many elements play a crucial role in the determination of life expectancy and the variables
affecting life expectancy can substantially differ from one country to another. For example,
there are countries in which pollution is a major cause of death and significantly contributes
toward reducing life expectancy. In other countries, pollution is not an issue while malaria is a
major cause of low life expectancy.
In T21, we assumed that six elements most directly influence life expectancy: income,
nutrition, health care, water availability, pollution and HIV/AIDS. When customizing the
model to a specific country, additional country-specific elements can be considered when
calculating life expectancy.
The method of computing age -specific death rates endogenously proceeds in four steps. The
first step is to calculate the normal life expectancy at birth based on income only. Then, the
indicated life expectancy is calculated based on four additional factors: health care,
availability of water, nutrition, and pollution. The third s tep involves using the life tables to
calculate age-specific death rates based on the indicated life expectancy. Finally, other factors
that affect only specific age cohorts, such as HIV/AIDS, are taken into account to calculate
the final death rates.
Normal life expectancy
One way to represent the effect of income on life expectancy is through Gross National
Product (GNP). There are many published correlations between life expectancy and GNP. A
good example is the smooth plot of national average life expe ctancy against GNP at
purchasing power parity published in 1997 by the World Bank (Figure 9). The WB graph
shows a rapid rise in life expectancy with increasing GNP per capita at low GDP, a slowing of
the increase after $5,000 per capita (1991 dollars, PPP), and a saturation (flattening out)
beyond $20,000 per capita. In other words, there is little gain in life expectancy from growth
in GNP beyond $20,000 per capita. The graph also suggests a 90% confidence interval of
about ten years of life expectancy, meaning that, at each income level, 90% of the national life
expectancy values for all countries are within 10 years of each other.
A curve like the one in Figure 9 suggests that differences in per capita GNP can explain a big
part of the differences in life expectancy in different countries 8.
In T21, the variable normal life expectancy is calculated using the relationship shown in Figure
10. This relationship is related to lower confidence bound in the WDI graph. Implicit in the
curve in Figure 10 is the assumption that GDP per capita determines life expectancy at birth
up to a GNP per capita of $20,000 (1991 dollars, PPP).
While there is a correlation between GNP and life expectancy, GNP does not directly cause life expectancy to
change. GNP is an aggregate measure of goods and services produced per year, and it is the effect of the
consumption of specific goods and servicesnot aggregate GDPthat cause increases or decreases in life
expectancy.
10000
20000
30000
40000
50000
60000
Figure 10: Relationship between Life Expectancy and GNP Assumed in T21
Indicated life expectancy
As discussed before, changes in life expectancy in T21 may also come from other elements.
Starting from normal life expectancy, which is calculated based on GNP, the effects of health
care, nutrition, water availability, and pollution are considered when calculating indicated life
expectancy. In T21, these effects include assumptions (in the form of a table function) that
relate the variable (such as access to basic health care) to life expectancy. An example is
provided in Figure 11.
Effect on life
expectancy
1.2
1.15
1.1
1.05
1
0.5
0.6
0.7
0.8
0.9
1.1
The effect of water availability on life expectancy is calculated based on the table function effect of
water availability on life expectancy table, using as input fraction of water available :
Effect of water availability on life expectancy = EFFECT OF NUTRITION ON LIFE
EXPECTANCY TABLE(quality of nutrition)
The effect of health basic care on life expectancy is calculated based on the table function effect of
basic health care on life expectancy table, using as input average access to basic health care :
Effect of basic health care on life expectancy = SMOOTH N(EFFECT OF BASIC
HEALTH CARE ON LIFE EXPECTANCY TABLE(average access to basic
health care), TIME FOR CHANGES IN ACCESS TO BASIC HEALTH CARE
TO AFFECT LIFE EXPECTANCY, EFFECT OF BASIC HEALTH CARE ON
LIFE EXPECTANCY TABLE(average access to basic health care), 1)
Note that a first order delay formulation is used in this case as we assumed a consistent delay
existed between changes in access to basic health care and consequent changes in life
expectancy.
The effect of ai r pollution on mortality is calculated based on the effect of air pollution on life
expectancy table, using as input the CO2 emissions per hectare, which is assumed to be a good
proxy for PM10 emissions:
effect of air pollution on mortality= SMOOTH N(EFFECT OF AIR POLLUTION ON
LIFE EXPECTANCY TABLE(co2 emissions per hectare),TIME FOR CHANGES
IN POLLUTION TO AFFECT LIFE EXPECTANCY,EFFECT OF AIR
POLLUTION ON LIFE EXPECTANCY TABLE(co2 emissions per hectare),1)
The effect of air pollution on life expectancy table represents the relationship between fossil fuel
emissions per hectare of land and mortality. It has been estimated based on data from a study
by AEA Technology Environment, commissioned by the EU. Note that a first order delay
formulation is used in this case as we assumed a consistent delay existed between changes in
air pollution and consequent changes in life expectancy.
The indicated life expectancy is calculated separately for each sex: the difference between the
two is represented by the parameter male female le difference . The equation is as follows:
Indicated life expectancy [female]= normal life expectancy*Effect Of Basic Health
Care On Life Expectancy*effect of nutrition on life expectancy*effect of water
availability on life expectancy*Effect Of Air Pollution On Life Expectancy
+MALE FEMALE LE DIFFERENCE/2
Indicated life expectancy [male]= indicated life expectancy[FEMALE]-MALE
FEMALE LE DIFFERENCE
Death rates
In T21, the death rates table is used to link indicated life expectancy w ith age -specific death
rates. The relationship inherent in the death rates table for males is illustrated in Table 4. The
first column is age, and the first row is life expectancy at birth. All the other cells are age
specific death rates. For instance, when male life expectancy at birth is 39.695 years (about
40 years), male infant mortality rate (death rate for age 0, or the first year of life) is 0.18951,
or 189.51 per thousand live births, and the death rate of a 40-year-old male is 0. 01584.
Age
0
1
5
10
20
30
40
50
60
70
80
The death rates computed in the previous section can be regarded as the death rates under
normal conditions. They have to be adjusted for abnormal situations, such as an epidemic
affecting some specific age cohorts (e.g. HIV for the adult population and TBC for the underfive population). These additional causes of death can be introduced by modifying the age group-specific mortality (formulation used for TBC) or by adding to the total deaths the
deaths from the disease (formulation used for AIDS) that are calculated in another module.
References
World Resources Institute, World Resources 1998-99: A Guide to the Global Environment,
Oxford: Oxford University Press, 1998.
Coale, A.J., and Demeny, P., Regional Model Life Tables and Stable Population, second
edition, New Y ork: Academic Press, 1983.
Sehgal, J., An Introduction to Techniques of Population and Labor Force Projections,
International Labor Office, Geneva, 1989.
Shorter, F.C., R. Sendek, and Y. Bayoumy, Computational Methods for Population
Projections, New Yor k: The Population Council, 1995.
UN Population Division, Sex and Age Annual, 1950 -2050 (The 1994 revision) (on disks)
The World Bank, World Development Indicators 1997 , Washington D.C., 1997
SHARE OF EDUCATION
EXPENDITURE FOR PRIMARY
EDUCATION TIME SERIES
PRIMARY EDUCATION
EXPENDITURE
IMPLEMENTATION TIME
<education
expenditure>
primary
education
expenditure
INITIAL EXPENDITURE
PER PRIMARY SCHOOL
AGE CHILD
<gdp deflator>
real primary
education expenditure
primary education
expenditure
implementation
<Perceived Relative Pc
Disposable Income>
INITIAL IMPLEMENTED
PRIMARY EDUCATION
EXPENDITURE
Implemented
expenditure per
Primary
primary school
Education
age child
Expenditure
average gross
enrollment rate
<Population>
effect of income on
primary intake
EFFECT OF INCOME ON
PRIMARY ENROLLMENT
TABLE
EFFECT OF PRIMARY
EDUCATION SECTOR
PERFORMANCE ON PRIMARY
GROSS INTAKE RATE TABLE
potential primary
enrollment rate
<Population>
TIME TO ENROLL
STUDENTS
<Population>
primary net
entrance rate
<Population>
Primary
Students
primary
graduation
<primary student
crude death rate>
literate youth
females males ratio
average adult
literacy rate
<total adult
population>
adult literate
population
Literate
Population
INITIAL
LITERACY RATE
primary students
INITIAL PRIMARY
deaths
adult primary
STUDENTS
graduation
literate deaths
literate population
cohort shift
<Time>
<DEATH
RATES
TABLE>
<life expectancy>
<TIME STEP>
PROPORTION OF ADULTS
GRADUATING PER YEAR
Explanation
Major Assumptions
Entrance and dropout rates depend on income and government expenditure per school
age child;
Primary school lasts for 6 years;
Graduates from primary schools are literate;
All new entrants enter primary school at grade one;
The children in school have the same life expectancy as the children who do not go to
school;
migration of children in school is not considered;
The literate population has the same life expectancy and migration behavior as the rest
of the population.
Input Variables
Variable Name
Module of Origin
Population
Net migration
Total adult population
Death rates table
Life expectancy
GDP deflator
Education expenditure
Primary student crude death rate
Perceived relative pc disposable income
Population
Population
Population
Population
Population
Relative prices
Government expenditure
Population
Households accounts
Output Variables
Variable Name
adult literacy rate
Average adult literacy
rate
Destination Module
Same Sector
Type of Variable
Time series
Constant
Constant
Constant
Table function
Table function
Constant
Constant
Constant
Constant
Constant
Other Sectors
Industry, Water
demand
Functional Explanation
The calculation of the adult literacy rate is divided into three parts. First, the level of
education offered by the government is calculated; then the flows affecting the stock of
students are calculated (the intake, graduation and dropout and death flows); and finally the
literate population is calculated.
The level of education services
As indicators of the level of education services available, we use the implemented expenditure
per school age child. To calculate this, the implemented primary education expenditure (an
exponential average of the governments education expenditure) is divided by the number of
primary school age children:
Expenditure per primary school age child= Implemented Primary Education
Expenditure/SUM(Population[sex!, primary school age!])
Intake flow
The expenditure per school age child is compared to its initial value (determining the relative
expenditure per primary school age child ) and used to calculate effect of primary education sector
performance on primary enrollment. This effect is multiplied by the effect of income on primary
intake rate in order to calculate the potential primary enrollment rate :
Potential primary enrollment rate = effect of income on primary intake[sex]*effect of
primary education sector performance on primary enrollment[sex]
The primary net entrance rate is obtained as the difference between the potential enrollment rate
and the actual gross enrollment rate, multiplied by the school age population and divided by
the time necessary to enroll new students:
Primary net entrance rate[sex]= (potential primary enrollment rate[sex]-gross
enrollment rate[sex])*SUM(Population[sex, primary school age!])/TIME TO
ENROLL STUDENTS
Graduation flow
Once children enter school, they must complete six grades before they are considered literate.
The subscript [grades] is used to model the primary students stock on six levels: students enter
the stock at grade 1 and graduate at grade 6 (unless they drop out of school and do not become
literate). The graduation flow is calculated as following:
Primary graduation = Primary Students[sex,GRADE 6]/TIME FOR ONE GRADE
The variable time for one grade represents the time required to complete one grade: if there are
a substantial amount of students repeating one grade, this is set to a value higher than 1.
Death flow
The primary students deaths flow is calculated based on the primary students stock and on the
primary student crude death rate :
Primary students deaths [sex,grades]= Primary Students[sex,grades]*primary student
crude death rate[sex]
The literate population
As students graduate and leave primary school, they enter the literate population stock. This
stock is also influenced by migration and death flows, according to the same rules used in the
Population module. In addition, the literate population stock increases as adults become literate
(e.g. due to an adult literacy program). The flow of adult primary graduation is considered a
constant fraction of the adult population in T21, but can be modeled as dependent on the
funds made available by the government for adult literacy programs.
The adult literacy rate is then calculated by dividing the adult literate population (the part of
literate population over 15 years) by the total adult population:
Adult literacy rate [sex] = MIN(1,adult literate population[sex]/ total adult
population[sex])
References
Gerald Barney, Managing a Nation: the Microcomputer Software Catalog , pp. 182-3, (the
Peru 21st Century Model), Institute for 21st Century Studies and Westview Press,
1991.
World Bank, World Development Indicators 2000, on CD-ROM.
secondary education
expenditure
<education
expenditure>
<gdp deflator>
SECONDARY EDUCATION
EXPENDITURE
IMPLEMENTATION TIME
INITIAL EXPENDITURE PER
SECONDARY SCHOOL
AGE PERSON
<Population>
<Perceived Relative Pc
Disposable Income>
EFFECT OF INCOME
ON SECONDARY
ENROLLMENT TABLE
real secondary
education expenditure
secondary education
expenditure implementation
expenditure per
secondary school age
person
effect of income on
secondary enrollment
<Time>
SHARE OF EDUCATION
EXPENDITURE FOR SECONDARY
EDUCATION TIME SERIES
Implemented
Secondary
Education
Expenditure
INITIAL IMPLEMENTED
SECONDARY EDUCATION
EXPENDITURE
EFFECT OF SECONDARY
EDUCATION SECTOR
PERFORMANCE ON SECONDARY
ENROLLMENT TABLE
effect of secondary
education sector on
enrollment
secondary gross
enrollment rate
potential secondary
enrollment rate
<Population>
INITIAL
SECONDARY
STUDENTS
INITIAL SECONDARY
<Population> EDUCATION RATE
<DEATH RATES
TABLE>
secondary students
enrollment
Secondary
Secondary
Graduates
secondary
secondary gross
Students
intake
graduation
<net migration>
<TIME TO ENROLL
total secondary
STUDENTS>
secondary students
graduates
net secondary
<Population>
TIME FOR ONE
deaths
graduates
GRADE IN
migration
secondary
SECONDARY
<secondary student
education index
SCHOOL
crude death rate>
secondary
graduates deaths
<life expectancy>
secondary
gradutes shift
<TIME STEP>
<Time>
Explanation
Major Assumptions
Entrance and dropout rates depend on income and government expenditure per
secondary school age person;
Secondary school lasts for 6 years;
All new entrants enter secondary school at grade one;
The children in school have the same life expectancy as the children who do not go to
school;
the migration of children in school is not considered
The population of secondary school graduates have the same life expectancy and
migration behavior as the rest of the population
Input Variables
Variable Name
Population
Net migration
Death rates table
Life expectancy
Secondary student crude death rate
Literate population
Perceived education expenditure
Time to enroll students
Perceived relative pc disposable income
Module of Origin
Population
Population
Population
Population
Population
Primary education
Primary education
Primary education
Households accounts
Output Variables
Variable Name
Total secondary
graduates
Secondary students
enrollment
Destination Module
Same Sector
Other Social Sectors
Labor availability and
cost
Labor availability and
cost
Other Sectors
Type of Variable
Time series
Constant
Constant
Constant
Table function
Table function
Constant
Constant
Constant
Functional Explanation
The structure of the Secondary Education module is very similar to that of the Primary
Education module. For an explanation of Secondary Education module, refer to the functional
explanation section in the Primary Education module.
References
Gerald Barney, Managing a Nation: the Microcomputer Software Catalog , pp. 182-3, (the
Peru 21st Century Model), Institute for 21st Century Studies and Westview Press,
1991.
World Bank, World Development Indicators 2000, on CD-ROM.
HEALTH EXPENDITURE
IMPLEMENTATION
<gdp deflator>
TIME
<health
expenditure>
health expenditure
implementation
real health
expenditure
<total population>
Implemented
Health
Expenditure
HEALTH CARE
ACCESS INCOME
THRESHOLD
<per capita relative
mean income>
<REFERENCE
INCOME>
fraction of population
without prompt access
to basic health care
Explanation
Major Assumptions
Access to basic health care is related to income, income distribution, and government
expenditure on health care;
Families with income levels above a certain threshold (country specific) have access
to basic health care anyway, independent from the level of service offered by the
government;
For households with income levels below the income threshold, a fraction of them will
have access to basic health care. That fraction is related to the government per capita
expenditure in health care.
Input Variables
Variable Name
Total population
Module of Origin
Population
GDP deflator
Health expenditure
Reference income
Per capita relative mean income
Std dev for per capita income
Relative pric es
Government expenditure
Income distribution
Income distribution
Income distribution
Output Variables
Variable Name
Destination Module
Same Sector
Other Social
Sectors
Mortality
Other Sectors
Constant
Table function
Constant
Constant
Local Experts
Functional Explanation
Calculating access to basic health care is divided into three major steps. The first step is the
calculation of the per capita level of health infrastructure and personnel put in place by the
government, which is the per capita perceived health expenditure. To do this, perceived health
expenditure (an exponential average of the governments health expenditure) is divided by total
population :
Per capita perceived health expenditure = Perceived Health Expenditure/total
population
Then second step is to calculate fraction of population within 5 km from health care point, which is
determined as a function of relative per capita perceiv ed health expenditure.
Fraction of population within 5 km from health care point = MIN (1,INITIAL
FRACTION OF POPULATION WITHIN 5 KM FROM HEALTH CARE
POINT*EFFECT OF PER CAPITA HEALTH EXPENDITURE ON ACCESS TO
BASIC HEALTH CARE TABLE(relative per capita per ceived health
expenditure))
The relative per capita perceived health expenditure represents the ratio between the current and
initial per capita perceived health expenditure.
Finally, to obtain the fraction of the population with access to basic health care, we consider
the fraction of the population that, with a high income, has access to basic health care even if
they live more than 5 kilometers away from a health care point. To do so, we first calculate
the fraction of the population without prompt access to basic health care:
fraction of population without prompt access to basic health care = (1-fraction of
population within 5 km from health care point)*(1-fraction of population having
access to health care anyway)
Access to basic health care is calculated as 1 minus the fraction of the population without
prompt access to basic health care.
Note: alternative representation of access to Basic Health Care
Depending on the structure of the health sector in the country being analyzed and the
availability of data, different formulations can be used to represent the access to basic health
care (ABHC). An alternative formulation that is often used in T21is based on an incomeABHC curve that relates access to basic health care to income level. Combining this curve
with the income distribution curve, we can compute access to basic health care for each
income level. Most importantly, the ABHC-income curve shifts left or right depending on
government expenditure: the higher the government expenditure for basic health care, the
higher the ABHC. This formulation is particularly useful when data on the fraction of
population within 5km from a health care point are not available, or when ABHC is more
related to the cost of the service than to the distance from the healt h center.
References
Gerald Barney, W. Brian Kreutzer, and Martha J. Garrentt, Managing a Nation: the
Microcomputer Software Catalog, pp. 182-3, (the Peru 21st Century Model), Institute
for 21st Century Studies and Westview Press, 1991.
International Monetary Fund, IMF Country Report No. 04/182, Mali: Poverty Reduction
Strategy Paper Annual Progress Report, June 2004
World Bank, World Development Indicators 2000, on CD-ROM.
7. HIV/AIDS module
BASE SUSCEPTIBLE <Hiv Population>
adult prevalence
CONTACT RATE
EFFECT OF HIV RELATED
susceptible
contact rate
DEATHS ON SEXUAL
average adult
effect of hiv treatment
<Population>
BEHAVIOR TABLE
prevalence
on sexual behavior
FRACTION
OF
HIV
<Cumulative Aids
susceptible population's
POPULATION
RECEIVING
EFFECT OF HIV
Deaths>
generated contacts
total susceptible <Time>
CARE AND
PROPER TREATMENT TIME
proportion of adults
TREATMENT
ON
population
SERIES
INCUBATION
with hiv safe behavior
susceptible SHARE OF HAART PATIENTS
TIME TABLE
BASE HIV
hiv positive
<Population>
population WILLING TO CHANGE THEIR
POSITIVE
contact rate
SEXUAL
BEHAVIOR
CONTACT
RATE
probability of an
<non-aids death effect of hiv care and
hiv population's
infectious contact
susceptible population
rates per age> treatment on incubation
generated contacts
<Time>
age distribution
time
SEX
BASE
INFECTIVITY
natural deaths
DISTRIBUTION
INCUBATION
OF INFECTIONS
TIME
incubation time
infections divided by
total new infections
stage of generation
Cumulative
Hiv
Aids Deaths
Population
infections
aids deaths
initial hiv aids sex
distribution
INITIAL HIV AIDS
POPULATION AGE
DISTRIBUTION
<Time>
initial hiv
population
TOTAL INITIAL HIV
POPULATION
total hiv
population
HIV POPULATION
INITIAL STAGE
<TIME STEP>
DISTRIBUTION
Explanation
Major Assumptions
The transmission of the virus among adults occurs primarily through sexual contact;
The HIV-infected population is subjected to the same natural death rate (due to non
HIV-AIDS related causes) as the rest of the population;
HIV-infected people are subjected to AIDS-related death after a given incubation
time;
The awareness of the disease and of its consequences alter the sexual behavior of the
susceptible population.
Input Variables
Variable Name
Population
Death rates per age
Module of Origin
Population
Morta lity
Output Variables
Variable Name
Aids deaths
Adult prevalence
Destination Module
Same Sector
HIV Children and
Orphans
HIV Children and
Orphans
Other Social
Sectors
Mortality
Other Sectors
Services
Infectivity
Initial HIV AIDS population age
distribution
Total initial HIV population
Base contact rate HIV
Constant
Constant
Constant
Constant
Constant
WHO/UNAIDS
Data on HIV population and
infection rate (WHO/UNAIDS)
WHO/UNAIDS
Data on length of stages from
Dangerfield and Roberts (1999)
Data on HIV population and
infection rate (WHO/UNAIDS)
WHO/UNAIDS
Constant
Local experts
Table function
Constant
Constant
Constant
Constant
Functional Explanation
The module can be divided into three major structures: the calculation of the HIV-infection
rate within adult groups (15-49 years old); the internal structure of the HIV population stock;
the calculation of the adult-prevalence rate.
Infection rate
The infection rate is determined based on the SIR model infection rate:
SIR Infection Rate= (c*i*I)(S/N),
Figure 16 illustrates the mechanisms that determine the growth (reinforcing loop R1) and the
decline (balancing loop B1) of the epidemic. As individuals in the uninfected population are
infected, they become part of the HIV population. Each infected individual generates a certain
number of infectious contacts, depending on the stage of advancement of the disease. The
infectivity of contacts also changes as the disease evolves. Infectious contacts lead to more
infections and to an even higher HIV population. This mechanism determines the exponential
growth in the HIV population observed in many countries (see the reinforcing loop R1). On
the other hand, not all infectious contacts happen between an infected and an uninfected
individual: it is also possible that an infected person is in contact only with other infected
people. In particular, the higher the proportion of infected individuals with respect to the total
susceptible population, the lower the probability that an infected person will contact an
uninfected one. This mechanism slows down and eventually stops the epidemic of HIV (see
the balancing loop B1).
Figure 16: Sketch of the Mechanisms of Growth and Decline of the HIV/AIDS Epidemic
HIV Population
As illustrated in Figure 16, infected individuals (the HIV population stock) go through five
stages with the development of the disease, each with different characteristics in terms of
infectivity and symptoms. Individuals in the HIV population stock are divided by gender and
age in order to keep track of age and gender -related aspects of the epidemic. Notice that in
case accurate data is not available this module can be simplified by reducing the number of
stages to two, without dramatically modifying the models behavior.
During the five-stage pe riod, infected individuals infect new ones and boost the diffusion of
the disease. The shift of individuals from one stage of the disease to the following one is
regulated by stage specific incubation times. These incubation times can change as a
consequence of particular treatments provided to the patients to diminish their symptoms and
increase their life expectancy.
As part of the HIV population, infected individuals are still subjected to the natural death rate
that affects the rest of the population. It is only in the last stages that one dies from AIDSrelated causes. The equations for the stages in the HIV population stock are reported below:
HIV Population[sex, age, HIV Stage 1]= INTEG (infections[sex, age]-natural
deaths[sex, age, HIV Stage 1]-HIV Population[sex, age, HIV Stage 1]/incubation
time[HIV Stage 1], initial HIV population[sex, age, HIV Stage 1])
HIV Population[sex, , HIV Stage 2]= INTEG (HIV Population[sex, age, HIV Stage
1]/incubation time[HIV Stage 1] -natural deaths[sex, age, HIV Stage 2] -HIV
Population[sex, age, HIV Stage 2]/incubation time[HIV Stage 2], initial HIV
population[sex, age, HIV Stage 2])
HIV Population[sex, age ,HIV Stage 3]= INTEG (HIV Population[sex, age ,HIV Stage
2]/ incubation time[HIV Stage 2]-natural deaths [sex, age, HIV Stage 3]-HIV
Population[sex, age ,HIV Stage 3]/incubation time[HIV Stage 3],initial HIV
population[sex, age, HIV Stage 3])
HIV Population[sex, age, Early AIDS]= INTEG (HIV Population[sex, age, HIV Stage
3]/incubation time[HIV Stage 3]-natural deaths[sex, age, Early AIDS] -HIV
Population[sex, age, Early AIDS]/incubation time[Early AIDS], initial HIV
population[sex, age, Early AIDS])
HIV Population[sex, age, Late AIDS]= INTEG (HIV Population[sex, age ,Early
AIDS]/ incubation time[Early AIDS] -AIDS deaths[sex, age]-natural deaths[sex,
age, Late AIDS], initial HIV population[sex, age, Late AIDS])
Adult prevalence
To calculate adult prevalence, the stock of the HIV population is divided by the adult
population:
Adult prevalence [sex] = SUM (Hiv Populat ion[sex!, susceptible!, hiv!])/SUM
(Population [sex!, susceptible!])
Note that adult prevalence is calculated for women and men separately, to keep support the
analysis of gender-related aspects of the epidemic.
References
Kermack , W. O., McKendrick, A.G. (1927) Contributions to the mathematical theory of
epidemics (Part I) Proc. Roy. Soc., A , 115: 700-21. 1927.
Dangerfield, B. and Roberts, C., 1999, Optimisation as a Statistical Estimation Tool: An
Example in Estimating the AIDS Treatment -Free Incubation Period Distribution,
System Dynamic Review, vol. 15, no. 3, pp. 273-291
B. C. Dangerfield, Y. Fang and C. A. Roberts (2001) "Model-Based Scenarios for the
Epidemiology of HIV/AIDS: the Consequences of Highly Active Antiretroviral
Therapy", System Dynamics Review, Vol. 17, No. 2, pp119-150.
BASE PERINATAL
TRANSMISSION
RATE
perinatal
transmission rate
effect of proper
treatment on perinatal
transmission rate
Aids
Children
<Time>
aids children
natural deaths
<births>
<adult
prevalence>
<deaths>
aids children
deaths INITIAL AIDS
CHILDREN
POTENTIAL REDUCTION IN
PERINATAL TRANSMISSION
RATE DUE TO PROPER
TREATMENT
<non-aids death
rates per age>
aids children
cohort shift
total aids children
<Time>
AVERAGE LIFE
EXPECTANCY FOR
AIDS CHILDREN
INITIAL AIDS
ORPHANS
TIME TO
BECOME ADULT
<aids deaths>
new aids orphan
under 18 crude
death rate
Aids
Orphans
orphans becoming
aids orphans adults
natural deaths
<Population>
Explanation
Major Assumptions
The infection of children with HIV occurs primarily through mother-to-child
transmission
The HIV-infected children are subjected to the same natural death rate (due to non
HIV-AIDS related causes) as the rest of the population
The HIV-infected children are subjected to AIDS related death after a given
incubation time
The children orphaned by AIDS are subjected to the same natural death rate (due to
non HIV-AIDS related causes) as the rest of the population
Input Variables
Variable Name
Module of Origin
Population
Births
Total fertility rate
Deaths
Death rates per age
Adult prevalence
Aids deaths
Population
Fertility
Fertility
Mortality
Mortality
HIV/AIDS
HIV/AIDS
Output Variables
Variable Name
Destination Module
Same Sector
Other Social
Sectors
Mortality
Other Sectors
Type of Variable
Constant
Constant
Constant
Constant
Constant
Constant
UNAIDS/WHO
Local experts
Functional Explanation
The AIDS Children and Orphan module consists of two separate stock and flow structures.
The first structure represents the mechanism of infection and the development of the disease
within children. Children born to an infected mother have a certain probability (the perinatal
transmission rate) of contracting AIDS in the early stages of their lives. The calculation of the
number of children born every year with AIDS is as follows:
Aids children births = births[sex]*adult prevalence[FEMALE]*perinatal transmission
rate
Children born with AIDS accumulate in the AIDS children stock and they are subjected to both
natural and AIDS-related death. Natural deaths are calculated according to the same rules as
uninfected children, while AIDS-related deaths are calculated as follows:
Aids children deaths = Aids Children[sex, age]/AVERAGE LIFE EXPECTANCY
FOR AIDS CHILDREN
The second structure represents how a parent dying from AIDS leaves orphaned children and
how these children eventually become adults or die in their younger years. The yearly
number of new AIDS orphans is calculated as follows:
New aids orphans = SUM(aids deaths[sex!,susceptible!])*(1-PROPORTION OF
CHILDREN WHO LOST ONE PARENT FOR AIDS THAT HAD ALREADY
LOST THE OTHER)* total fertility rate/3
The total fertility rate is used in this equation to calculate how many children become orphans
when one parent dies, and it is divided by three because it is assumed that the parent dying of
AIDS will die in the first period of their reproductive life.
AIDS orphans can either die from natural causes or grow up and become a dults. Natural
deaths for AIDS orphans are calculated based on the under-18 crude death rate, while the flow
of orphans becoming adults is calculated by dividing the stock of orphans by the average time
it takes for a child to become an adult:
Aids orphans natural deaths = Aids Orphans*under 18 crude death rate
Orphans becoming adult = Aids Orphans/TIME TO BECOME ADULT
References
www.unaids.org
www.niaid.nih.gov
9. Nutrition module
<crops production
in tons>
<crops production
in tons>
<net export>
<Producer Prices>
total fat
consumed
STOCKING CAPACITY
BUFFERING TIME
net export
production ratio
Calories Per
Person Per Day
total calories
consumed
PROTEINS PER
TON OF FOOD
VEGETABLE
WHO STANDARD
CALORIES
Fat Per Person
Per Day
quality of nutrition
total proteins
consumed
Proteins Per
Person Per Day
total micronutrients
consumed
Micronutrients Per
Person Per Day
PROTEINS PER
TON OF MEAT
<real production
by sector>
WHO STANDARD
MICRONUTRIENTS
WHO STANDARD
PROTEINS
<total population>
MICRONUTRIENTS PER
TON OF FOOD
MICRONUTRIENTS
VEGETABLES
PER TON OF MEAT
DAYS IN ONE
YEAR
Explanation
Major Assumptions
Food produced and not exported is locally consumed;
The mix of the types of vegetable produced is constant;
The mix of the types of meat produced is constant.
Input Variables
Variable Name
Total population
Crops production in tons
Meat production for domestic market
Fish production in tons
Net export
Producer prices
Real production by sector
Module of Origin
Population
Agriculture
Animal husbandry, fishery and forestry
Animal husbandry, fishery and forestry
International trade
Relative prices
Aggregate production and income
Output Variables
Variable Name
Destination Module
Same Sector
Quality of nutrition
Other Social
Sectors
Mortality
Other Sector s
Type of Variable
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
N.A.
Gregor ian calendar
WHO
WHO
WHO
Local experts
Functional Explanation
T21 uses the domestic supply of animals, vegetables and total population to calculate per
capita consumption of proteins, fat calories and micronutrients. Proteins, calories and
micronutrients are compared to international standards from WHO (World Health
Organization) to derive an aggregate measure of the quality of nutrition.
Domestic animal and vegetable consumption
The domestic consumption of animal food is calculated as the sum of the production of meat
for domestic consumption and the production of fish (assumed entirely domestically
consumed):
Domestic animal consumption in tons = meat production for domestic market + fish
production in tons/2
Note that fish production is divided by two to account for the lower contents of nutrients per
kg of fish with respect to meat.
The domestic consumption of vegetables is calculated as the difference between the domestic
production of vegetables and the net export of agricultural non-cotton goods:
Domestic food vegetables consumption in tons = SUM(crops production in
tons[crops!])-food vegetables net export
Domestic consumption of vegetables and animal food are then used to calculate the total
consumption in terms of calories, proteins, fat and micronutrients. As an example, the total
consumption of calories is calculated as follows:
Total calories consumed = domestic animal consumption in tons*CALORIES PER
TON OF MEAT + domestic food vegetables consumption in tons*CALORIES
PER TON OF FOOD VEGETABLES
Consumption of fat, proteins and micronutrients are calculated accordingly to the same logic.
The total amount of nutrients consumed is divided by the total population and by the number
of days in one year to obtain the daily per capita consumption of fat, calories, proteins and
micronutrients. These last three variables are then weighted to derive a quality of nutrition
index:
Quality of nutrition = 0.625* Calories Per Person Per Day/WHO STANDARD
CALORIES + 0.375*Proteins Per Person Per Day/WHO STANDARD
PROTEINS + 0*Micronutrients Per Person Per Day/WHO STANDARD
MICRONUTRIENTS
The weight given to the different nutrients is arbitrarily chosen. Note that in this case a weight
of 0 is given to the consumption of micronutrients, because data on micronutrients was not
available.
References
Hunter Colby, Mark Giordano, and Kim Hjort, The ERS China CPPA Model:
Documentation, 1997
US Department of Agriculture, International Agricultural Baseline Projections to 2005, AER750, USDA/ERS.
US Department of Agriculture, World Agriculture, Trends and Indicators, 1970-91, USDA,
1993
World Bank, World Data 1995, on CD-ROM.
<Agricultural Land
In Use>
proportion of agricultural
land over total land
<TOTAL LAND
AREA>
PUBLIC SERVICES
TARGET ROAD
DENSITY
<Time>
<total population>
relative road
density
Roads Under
Construction
roads completion
AVERAGE ROADS
COMPLETION TIME
budget for new
transportation
infrastructure
indicated roads
maintenance cost
roads upgrade
cost
<gdp deflator>
feasible upgrade
roads upgrade
implemented
fraction of
necessary
maintenance
real roads
expenditure
<roads
expenditure>
Functioning
Roads
AVERAGE ROADS
MAINTENANCE
COST PER KM
effect of road
proximity on yield
INITIAL EFFECT OF
ROAD PROXIMITY ON
YIELD
roads disruption
total roads
roads construction
starts
AVERAGE ROADS
CONSTRUCTION
UNIT COST
total roads in
agriculture area
desired road
upgrade
AVERAGE ROADS
UPGRADE COST
PER KM
INITIAL EXISTING
ROADS
<Time>
DESIRED ROAD
UPGRADE TIME
SERIES
MINIMUM TIME TO
UPGRADE ROADS
Explanation
Major Assumptions
Roads are of two types: paved and unpaved;
Unpaved roads can be upgraded to paved roads;
Road mainte nance has priority over road construction when budgets are limited;
The density of rural roads has an important effect on agricultural productivity.
Input Variables
Variable Name
Total population
Total land area
Agricultural land in use
Roads expenditure
Module of Origin
Population
Land
Land
Government expenditure
GDP deflator
Relative prices
Output Variables
Variable Name
Destination Module
Same Sector
Other Social
Sectors
Other Sectors
Agriculture
Type of Variable
Constant
Constant
Constant
Arbitrarily chosen
Local experts
Constant
Time series
Local experts
Local experts
Constant
Local experts
Table function
Constant
Constant
Time series
Constant
Constant
Constant
Constant
Functional Explanation
The structure of the Roads module can be divided into three sub-structures: (1) the budget for
road construction, maintenance and upgrade; (2) the physic al flows of road construction,
maintenance and upgrade; and (3) the effect of road proximity on agriculture yield.
Budgets for road construction, maintenance and upgrades
The government real expenditure for roads is determined by roads expenditure deflated by the
GDP deflator:
Real roads expenditure = roads expenditure/gdp deflator
This budget is allocated with priority to the maintenance and upgrade of the existing
structures, with respect to building new infrastructures. The budget for the construction of
new roads is calculated as:
The budget for the upgrade of existing structures is exogenous ly determined, while the
indicated budget for road maintenance (the expenditure that the government should sustain to
keep the existing roads in good order) is calculated as the stock of existing roads times the
average maintenance cost per km:
Indicated roads maintenance cost = SUM(AVERAGE ROADS MAINTENANCE
COST PER KM[roads type!]*Functioning Roads[roads type!])
Roads construction, upgrade and maintenance
The budget for the construction of new roads is allocated with the construction of paved or
unpaved roads, depending on the parameter fraction of funds for road construction for paved roads.
The funds allocated to the construction of each type of road are then divided by the average
cost per km to determine how many km of road the government starts building every year:
Roads construction starts [unpaved] = budget for new transportation infrastructure*(1FRACTION OF FUNDS FOR ROAD CONSTRUCTION FOR PAVED ROADS
TIME SERIES)/AVERAGE ROADS CONSTRUCTION UNIT COST
[UNPAVED]
Roads construction starts [paved] = budget for new transportation
infrastructure*(FRACTION OF FUNDS FOR ROAD CONSTRUCTION FOR
PAVED ROADS TIME SERIES)/AVERAGE ROADS CONSTRUCTION UNIT
COST [PAVED]
The road construction starts accumulates in the stock of roads under construction , before
becoming functioning roads after a certain average construction time. Functioning roads are
disrupted when proper maintenance is not implemented:
Roads disruption [road type] = Functioning Roads[roads type]*(1-implemented
fraction of necessary maintenance)/AVERAGE ROADS LIFE WITHOUT
MAINTENANCE[roads type]
Functioning unpaved roads can be upgraded to paved roads, depending on the road upgrade
sought by the government and available funds:
Roads upgrade [unpaved] = MIN(feasible upgrade, MIN(desired road upgrade,
Func tioning Roads[UNPAVED]/MINIMUM TIME TO UPGRADE ROADS))
Effect of roads proximity on yield
Assuming a target for rural road density that guarantees full productivity of agriculture land,
relative roads density in agricultural areas is calculated as the current density (the roads per
hectare of agriculture land) over target density. The effect o f road proximity on yield is then
calculated through the table function illustrated below in Figure 20:
Effect of road proximity on yield = EFFECT OF ROAD PROXIMITY ON YIELD
TABLE(relative rural roads density level)
The effect of road proximity on yield is then normalized (divided by the initial effect of road
proximity on yield) to determine the relative effect of road proximity on yield, which affects
productivity in the Agriculture module.
References
Stifel D, B. Minten, and P. Dorosh. 2003. Transaction Costs and Agricultural Productivity:
Implications of Isolation for Rural Poverty in Madagascar. IFPRI MSSD Discussion
Paper No. 56. IFPRI, Washington D.C.
<INITIAL CAPITAL
AGRICULTURE>
<relative labor
technology>
agriculture desired
capital labor ratio
<Capital Agriculture>
<INITIAL CAPITAL
INDUSTRY>
initial agriculture
capital labor ratio
<labor force
availability>
Agriculture
Employment
agriculture labor
demand
indicated agriculture
employment level
INITIAL INDUSTRY
EMPLOYMENT
TIME TO HIRE IN
AGRICULTURE
<total labor
supply>
unemployment rate
Industry
Employment
TIME TO HIRE IN
net industry hiring
INDUSTRY
indicated industry
employment level
total employment
INITIAL SERVICES
<Capital Industry>
industry labor
EMPLOYMENT
demand
<INITIAL CAPITAL
<labor force
SERVICES>
availability>
<relative labor
initial services capital
technology>
Services
services desired
labor ratio
Employment
net services hiring
capital labor ratio
indicated services
employment level
services
labor
<Capital Services>
TIME TO HIRE IN
demand
SERVICES
<relative labor
technology>
industry desired
capital labor ratio
Explanation
Major Assumptions
Employment levels depend on the amount of productive capital and current levels of
technology;
Employment levels cannot be higher than the available labor force.
Input Variables
Variable Name
Initial capital agriculture
Capital agriculture
Capital industry
Initial capital industry
Module of Origin
Agriculture
Agriculture
Industry
Industry
Services
Services
Technology
Labor availability and cost
Labor availability and cost
Output Variables
Variable Name
Industry employment
Services employment
Agriculture labor demand
Industry labor demand
Services labor demand
Destination Module
Same Sector
Other Social
Sectors
Other Sectors
Industry
Services
Labor availability
and cost
Labor availability
and cost
Labor availability
and cost
Type of Variable
Constant
Constant
Constant
Constant
Constant
Constant
Functional Explanation
The calculation of employment levels for the three production activities (agriculture, industry
and services) is very similar, so only the description of the calculation for industry
employment appears here. Users can understand the formulations used to calculate
employment for the other two types of production activity by analogy.
To determine employment levels, labor demand from the Industry module is calculated first,
and then it is assumed employment is adjusted to labor demand, unless labor supply is
insufficient to cover demand.
Labor demand
To determine labor demand, the desired capital labor ratio is determined first, that is the
optimal capital labor ratio sought by producers. The desired capital labor ratio increases as
technology increases, as production processes bec ome more automated and require less labor
per unit of capital:
Labor demand is calculated as the capital currently in place in the Industry module, divided
by the desired capital labor ratio:
Industry labor demand = (Capital Industry/industry desired capital labor ratio)
Employment
Employment is represented as a stock variable in T21, and it adjusts toward industry labor
demand, unless the labor supply is insufficient to meet the demand. The only flow
accumulating in the stock of employment is net hiring:
Industry net hiring = (indicated industry employment level-Industry
Employment)/TIME TO HIRE IN INDUSTRY
Indicated employment levels represent feasible employment levels, when considering the
demand of producers and the available labor force:
Indicated industry employment level = industry labor demand * labor force availability
Labor force availability represents the fraction of labor demand that can be satisfied by the
current labor supply, and is calculated in the Labor Availability and Cost module.
Note: agriculture employment
References
<agriculture
labor demand>
<industry labor
demand>
<services labor
demand>
labor force
availability
labor demand
supply balance
PERCENTAGE OF LABOR
DEMAND FOR SKILLED
LABOR
<total secondary
graduates>
skilled labor
demand
<Population>
total labor supply
LABOR FORCE
AVAILABILITY
FRACTION
INITIAL LABOR
DEMAND SUPPLY
BALANCE
relative labor demand
supply balance
FRACTION OF
SECONDARY STUDENTS
WORKING PART TIME
ELASTICITY OF LABOR
COST TO DEMAND
SUPPLY BALANCE
skilled labor
availability
<Time>
relative labor cost
EFFECT OF UNIONS
ACTIVITY ON LABOR
COST TIME SERIES
Explanation
Major Assumptions
Changes in labor cost are mainly caused by changes in the labor supply/demand
balance;
Skilled labor supply corresponds to the stock of graduates from secondary school;
Skilled labor demand is a fixed fraction of labor demand.
Input Variables
Variable Name
Population
Agriculture labor demand
Industry labor demand
Services labor demand
Total secondary graduates
Secondary students enrollment
Module of Origin
Population
Employment
Employment
Employment
Secondary education
Secondary education
Output Variables
Destination Module
Variable Name
Same Sector
Employment
Other Social
Sectors
Other Sectors
Technology
Agriculture, industry
and services
Type of Variable
Constant
Initial
Constant
Local experts
Internally determined by Vensim as
the first value simulated for labor
demand supply balance
Local experts
Time series
Local experts
Functional Explanation
The Labor Availability and Cost module can be divided into two major structures: one (right
part of the sketch in FIGURE 17) describes the mechanisms that determine the availability
and cost of general labor; the other (left part of the sketch in FIGURE 17) is used to calculate
the availability of skilled labor.
Availability and cost of labor
The availability of the labor force is calculated as the ratio of labor supply and demand.
Labor supply is calculated as a fraction of the population in the productive age, minus those
enrolled in secondary school and do not work:
Total labor supply = SUM(Population[sex!,labor age!])*labor force availability
fraction-SUM(secondary students enrollment[sex!])*(1-fraction of secondary
students working part time)
Total labor demand is calculated as the sum of labor demand for each type of production
activity, calculated in the Employment module:
Total labor demand = agriculture labor demand+industry labor demand+services labor
demand
Note that the formulation if then else is used to make sure that when labor supply exceeds
labor demand, labor availability is equal to 100%.
The ratio of labor supply and demand is a lso used to determine the relative cost of labor. The
effect of the demand supply balance on labor cost is calculated as following:
Effect of demand supply balance on labor cost = relative labor demand supply
balance^ elasticity of labor cost to demand supply balance
The relative labor cost is then calculated as the initial labor cost (not explicitly in the equation,
since it is assumed equal to 1) times the effect of the demand supply balance, times the effect
of union activity on labor cost:
Relative labor cost = effect of demand supply balance on labor cost +effect of unions
activity on labor cost (Time)
Availability of skilled labor
To calculate the availability of skilled labor, the demand for skilled labor is compared to
supply. The demand for skilled labor is calculated as a fixed share of the demand for labor
from each type of production activity:
Skilled labor demand [agri] = agriculture labor demand*PERCENTAGE OF LABOR
DEMAND FOR SKILLED LABOR[agri]
Skilled labor demand [ind] = industry labor demand*PERCENTAGE OF LABOR
DEMAND FOR SKILLED LABOR[ind]
Skilled labor demand [serv] = services labor demand*PERCENTAGE OF LABOR
DEMAND FOR SKILLED LABOR[serv]
Skilled labor supply is assumed to correspond to the sum of people with a secondary school
degree:
Total skilled labor supply = SUM(total secondary graduates[sex!])
Finally, the availability of skilled labor is calculated as the ratio between total skilled labor
demand (the sum of skilled labor demand for the agriculture, industry and services) and total
skilled labor supply:
Skilled labor availability = total skilled labor supply/total skilled labor demand
References
<disposable income>
<total population>
mean per
capita income
INCOME
CLASS SIZE
income level
demo income
mean
income distribution
<gdp deflator>
STD DEV OVER
MEAN RATIO
TABLE
GINI COEFFICIENT
TIME SERIES
<Time>
<disposable
income>
share of income to
lowest quintile
gini coefficient
standard
deviation
POVERTY LINE
IN CFA87
share of population
below twice the poverty
line
relative poverty
line
<total population>
total population
below poverty line
average income of
population below
poverty line
share of population
below poverty line
average relative pc
poverty line
<REFERENCE
INCOME>
<gdp deflator>
income gap ratio
Explanation
Major Assumptions
Income is log normally distributed
Gini coefficient is exogenously determined
Input Variables
Variable Name
Total population
Disposable income
GDP deflator
Module of Origin
Population
Households account
Relative prices
Output Variables
Destination Module
Same Sector
Variable Name
Per capita relative mean
income
Std dev of per capita income
Reference income
Other Social
Sectors
Access to basic
health care
Access to basic
health care
Access to basic
health care
Type of Variable
Time series
Table function
Constant
Constant
Constant
Other Sectors
Functional Explanation
From a mathematical perspective, the Income Distribution module is probably the most
complex part of the T21 model. Assuming a log normal distribution of income, two
parameters must be determined to calculate income distribution: the mean and standard
deviation of income. The Households module of T21 generates disposable income, from
which we can compute mean income. The standard deviation of income can be computed
from the Gini coefficient and the mean of income, as seen in the next section. The next
section describes the general characteristics of a log normal distribution, while the subsequent
sections describe the method used to calculate income distribution and other indicators.
Characteristics of the log normal distribution
If X is log normally distributed as LN(, 2), and Y = log X, then Y is normally distributed as
N(, 2), i.e., the log of a log normal variable becomes a normal variable. (In other words, if
Y is normal, then X=EXP(Y) is log normal)
The relationship between the parameters , and , are:
= LN(2) - 0.5 * LN(2 + 2)
(1)
2 = LN(1 + 2/2)
(2)
= EXP( + /2)
(3)
2 = EXP(2* + 2) * (EXP(2) - 1)
(4)
The log normal distribution has the following density function, in which x is the income level:
f(x) = 1/(SQRT(2*)**x) * EXP(-((ln x - )^2)/(2*^2)
(5)
From the density function we see that, once we know the mean () and standard deviation ()
of the log normal (or income) distribution, we need to compute the mean () and the standard
deviation ( ) of the corresponding normal distribution before we arrive at the density
function.
The cumulative function of the log normal distribution cannot be written in an analytically
closed form.
The mean income
The mean pc income is calculated as the household income in real terms, divided by the
population:
Mean per capita income = (disposable income/gdp deflator)/ total population
The standard deviation of income
In order to compute the standard deviation of income from the Gini coefficient, we developed
a specific table function (see Figure 24). The values used for this table function were tested in
several countries and proven sufficiently accurate for the type of application required for T21
(Qu 2002).
In order to write the density function of income distribution, we need to first compute the
mean and standard deviation of the corresponding Normal Distribution, which are (refer to
Equations 1 and 2):
The density function of income distribution can now be computed as (see Equa tion 5):
income distribution[IncClass] = 1/(SQRT(2*3.14159)*standard deviation*income
level [incclass]) * EXP((-(LN(income level[incclass])-mean)^2)/(2*standard
deviation^2))
Notice that [IncClass] is an array, so that the density function values at all the income levels
can be computed with a single equation.
Indicators
With income distribution and its mean and standard deviations, a variety of indicators are
calculated. Two common indicators are the share of population living below the poverty line and
the poverty gap ratio . The share of the population living below the poverty line is calculated using
the mean and standard deviation of income, and LNNORMAL function:
Share of population below poverty line = LNNORMAL(1,relative poverty line,per
capita relative mean income,std dev of per capita income)
The LNNORMAL function accepts four variables as input: lower bound income level, upper
bound income level, mean household income, and standard deviation of household income.
Its output is the percentage of the population between the bounds. When the lower bound is
zero, 1 should be used instead, as zero will cause a divided by zero error in the function.
The poverty gap ratio is calculated from the UN definition based on the income gap ratio and the
average i ncome of population below poverty line. This last variable is calculated using an external
function QUINWLTH. This function takes three input variables, a fraction of the poorest
population (such as 0.2 or a variable), mean household income, and standard deviation of
household income. The output of the function is the fraction of total income made by the first
input variable: fraction of the poorest population.
Income gap ratio = (average relative pc poverty line- average income of population
below poverty line/REFERENCE INCOME)/average relative pc poverty line
Average income of population below poverty line = (QUINWLTH(share of population
below poverty line, per capita relative mean income, std dev of per capita
income)*disposable income/gdp deflator)/total population below poverty line
Poverty gap ratio = income gap ratio* share of population below poverty line
References
William H. Greene, Econometric Analysis, 3rd Edition, Prentice Hall 1993 (p.71)
David W. Pearce, The MIT Dictionary of Modern Economics, 4th Edition, the MIT Press
1994 (p. 254)
Camilla Toulmin et al., Mali Poverty Profile , SIDA, IIED Drylands Programme, May 2000
Weishuang Qu et al., A Model for Evaluating the Policy Impact on Poverty, Proceedings of
the 19th International Systems Dynamics Conference, Palermo, Sicily, Italy, August
2002.
Economic Sectors
The economic sectors of the T21 are listed in Table 3. The next pages present the structure of
each module detail.
Table 5: The economic sectors of T21 and corresponding modules
Economy
Production Sector:
14. Aggregate Production and Income
15. Agriculture
16. Animal husbandry-Fishery-Forestry
17. Industry
18. Services
Technology Sector:
19. Technology
Households Sector:
20. Households Accounts
Government Sector:
21. Government Revenue
22. Government Expenditure
23. Public Investment and Consumption
24. Government Balance and Financing
25. Government Debt
ROW Sector:
26. International Trade
27. Balance of Payments
Investment Sector:
28. Relative Prices
29. Investment
nominal
production by
sector
real production
by sector
<services
production>
<net savings>
<gdp deflator>
<financing>
<adjustment>
<errors and
omissions>
FRACTIONAL
GROWTH OF ROW
GDP TIME SERIES
<agriculture
production>
<arrears>
resources
check 2
real gdp at
factor cost
nominal gdp at
factor cost
<Time>
real gdp at
market prices
Relative
Row Gdp
<consumption>
<investment>
<total population>
real pc gdp
nominal gdp at
market prices
<Import Taxes By
Sector>
<public
consumption>
<private
consumption>
<net factor
income>
import taxes
consumption
domestic
savings
total import
<total export>
real pc gross
national product
<total
population>
PPP PARAMETER
TIME SERIES
1987 US GDP
DEFLATOR
BASE 1990
resources
check 3
real pc national
income
<Time>
<gdp deflator>
<investment>
<current account>
<resources
balance>
resources check 1
EXCHANGE
RATE 1987
ppp parameter
real pc gnp in
usd in ppp
Explanation
Major Assumptions
Real GDP at factor cost is the sum of real sectors production
Real GDP at market prices also includes import taxes
Gross national product is equal to nominal GDP at market prices plus net factor
income
Gross national income is equal to gross national product plus total net transfers
Input Variables
Variable Name
Agriculture production
Industry production
Services production
Total population
Import taxes by sector
Net factor income
Total net transfers
Current account
Module of Origin
Agriculture
Industry
Services
Population
Government Revenue
Balance of Payments
Balance of Payments
Balance of Payments
Investment
Private consumption
Public consumption
GDP deflator
Total export
Resources balance
Financing
Errors and omissions
Arrears
Adjustment
Net savings
Investment
Households Accounts
Public Investment and Consumption
Relative Prices
Balance of Payments
Balance of Payments
Balance of Payments
Balance of Payments
Government Balance and Financing
Government Balance and Financing
Households Accounts
Output Variables
Variable Name
Real production by sector
Sector GDP ratio
Nominal production by sector
Nominal GDP at factor cost
Real GDP at market prices
Module of Destination
Same Sector
Other Economic Sectors
Relative Prices, International
Trade
Investment
Government Revenue,
Relative Prices
Government Revenue
Other Sectors
Energy Demand
Functional Explanation
This module contains several important economic indicators, among which the key income
and production measures. These indicators are calculated according to international standard
definitions, as described below:
Real GDP at factor cost = agriculture production +industry production +services
production
Nominal GDP at factor cost = real GDP at factor cost*GDP deflator
Real GDP at market prices = real GDP at factor cost +import taxes/GDP deflator
Nominal GDP at market prices = real GDP at market prices*GDP deflator
Gross national product = nominal GDP at market prices +net factor income
Gross national income = gross national product +total net transfers
Consumption = private consumption +public consumption
Total import = consumption + investment+ total export - nominal GDP at market
prices
Domestic savings = nominal GDP at market prices -consumption
National savings = investment +current account
Real pc GDP = real GDP at market prices/total population
Real pc gross national product = (gross national product/gdp deflator)/total population
Real pc GDP in USD = (real pc gross national product/EXCHANGE RATE
1987)/"1987 US GDP DEFLATOR BASE 1990"
Real pc GDP in USD in PPP = real pc GNP in USD/PPP parameter
Note in particular that the total level of imports is determined as the residual in the GDP
equation in the aggregate production and income sector. Using the classical economic
nomenclature, given the level of production (Y), investment (I), consumption (C) and export
(X), import (M) is calculated as:
M=+I+C+X Y
This type of closure ensures the consistency of economic flows in T21. Other variables can be
chosen as residual in the GDP identity, depending on the particular structure of the economy
studied, the particular research questions to be addressed, and the available data.
The module also contains three check variables (resources check 1, resources check 2, resources
check 3 ) which serve as indicators the internal consistency of the system of national accounts
in T21. These variables should always be equal to zero. Their equations are:
Resources check 1 = investment +consumption nominal GDP at market prices
+resources balance
Resources check 2 = net savings +financing +errors and omissions arrears
adjustment
Resources check 3 = gross national income consumption national savings
Two other key variables are calculated in this sector: sector GDP ratio and the relative ROW
GDP . The sector GDP ratio (sector is intended in this case as type of economic activity:
agriculture, industry and services) represents the relative contribution of each type of
production activity to the GDP and is calculated as the production of each type of goods and
services divided by the total production:
Sector GDP ratio [sectors] = real production by sector [sectors]/real GDP at factor cost
The relative ROW GDP represents the relative size of the rest of the worlds GDP compared to
its initial value. It is calculated based on an exogenously determined growth rate.
References
Dornbusch, R. and Fischer, S., Macroeconomics, 6th edition. McGraw-Hill, Inc. 1994
AVERAGE LIFE OF
AGRICULTURE
CAPITAL
INITIAL CAPITAL
AGRICULTURE
<investment
crops>
Capital
Agriculture
<Time>
depreciation
agriculture
INITIAL CROPS
PRODUCTION IN
TONS
AGRICULTURE
CAPITAL
relative agriculture ELASTICITY
capital
total factor
productivity
agriculture
AGRICULTURE
PRODUCTS VALUE
ADDED TIME SERIES
yield
<Effect Of Health On
<effect of education <relative effect of road
Productivity>
proximity on yield>
on productivity>
INITIAL
HARVESTED
AREA
agriculture
production
crops production
average
yield
<Agricultural Land
In Use>
<labor force
availability>
<livestock
production>
crops production
in tons
relative
harvested area
<fish
production>
<forestry
production>
main crop
area fraction
harvested area
crop intensity
index
<Time>
CROP INTENSITY
INDEX TIME SERIES
Explanation
Major Assumptions
Input Variables
Variable Name
Investment crops
Effect of energy price on productivity
Relative labor technology
Effect of water stress on agriculture
productivity
Relative effect of road proximity on yield
Effect of health on productivity
Effect of education on productivity
Labor force availability
Agricultural land in use
Forestry production
Fishing production
Livestock production
Module of Origin
Investment
Energy Supply
Technology
Water demand
Roads
Industry
Industry
Employment
Land
Animal husbandry-fishery- forestry
Animal husbandry-fishery- forestry
Animal husbandry-fishery-forestry
Output Variables
Variable Name
Crops production in tons
Agriculture production
Crops production
Capital agriculture
Average yield
Crop intensity index
Module of Destination
Same Sector
Other Economic Sectors
Investment
Investment
Technology
Type of Variable
Constant
Constant
Constant
Time series
Time series
Time series
Constant
Constant
Other Sectors
Nutrition
Employment
Land
Land
Functional Explanation
The major output of the Agriculture module (see Figure 26) is agriculture production. Crops
production is modeled using a Cobb-Douglas (CD) production function with main inputs
land, capital, and technology. Productivity is also determined by the availability of water,
roads, oil and healthy labor force (basic and educated). Crops production is calculated for two
types of crops separately: the main crop and the rest. The type of crops considered in the
agriculture production function should be determined based on the characteristics of
agriculture production in the country being modeled.
In the top right of the sketch, crops production is added to fishery, forestry and livestock
productio ns to determine total agriculture production. The following paragraphs describe how
the Cobb-Douglas production function is applied to the agriculture production in T21, and
how eventually total agriculture production is determined.
Note that values for agriculture production are measured in physical quantities or at constant
prices.
The Cobb-Douglas production function in T21
Where A represents the total factor productivity (TFP), K represents the stock of capital, and
L represents labor or, as in the case of the agriculture production in T21, the stock of land
used for agriculture. The constant a represents the elasticity of output to capital: the ratio
between the percentage change of output and the percentage change of an input. The elasticity
of output to labor is set to 1-a, assuming that there are constant returns to scale (the
production function is thus first order, homogeneous).
In T21 the standard CD production function is presented in unit-consistent form, and TFP is
expanded to include several different elements. To explain the final form of production
function used in T21, we have to proceed in steps.
First, in T21 the CD production function is implemented using the values for capital and land
relative to their initial values, or normalized. In the original CD function, capital and land are
used directly to determine production: this leads to inconsistency in the units of measure used
(capital is measured in currency units, land in hectares and production in currency per year).
Normalizing K and L (dividing them by their initial value) avoids unit errors, does not modify
the results of the production function and conforms to good System Dynamics practice 9 (see
Sterman, 2000). In the following paragraphs we will consider the terms Y, K and L as
normalized.
Second, the TFP includes several different elements affecting the productivity of factors. In
the T21 agriculture production function we explicitly considered technology, a healthy labor
force (uneducated and educated) and the availability of water, roads and oil. The T21
formulation for TFP for agriculture in T21 can be written as:
From a System Dynamics perspective, K ^a and L^(1-a) represent the effects of capital and land on production,
and are used in a multiplicative form in the production function
Note that since all the elements composing the TFP are used in a multiplicative form, it is
algebraically indifferent whether a specific effect is assumed to affect the productivity of K or
L.
The actual equations used in T21 to calculate the agriculture production reflect exactly the
logic described above, but use more transparent variable names to identify the various factors
and effects. The relative (normalized) levels of capital and land are calculated as:
Relative agriculture capital = Capital Agriculture/INITIAL CAPITAL
AGRICULTURE
Relative harvested area [crops] = harvested area[crops]/INITIAL HARVESTED
AREA[crops]
Note that the subscript [crops] is used to distinguish among the land used for the different
crops.
The total factor productivity is calc ulated as following:
Total factor productivity agriculture = effect of education on productivity*effect of
energy price on productivity *effect of water stress on agriculture
productivity*labor force availability*relative effect of road proximity on yield
*relative labor technology[AGRI]*Effect Of Health On Productivity
Note that the effects of the various elements on TFP are calculated in other modules, and are
explained in other sections of this documentation.
is calculated by multiplying the effects of change in capital, land and
TFP by the initial level of production:
Crops production in tons
To obtain crops production in monetary values, production in tons is multiplied by the value
added per ton of crop, and production from all crops is summed up:
Crops production = SUM(crops production in tons[crops!]*agriculture crops value
added per ton[crops!])
Finally, total agriculture production is calculated as the sum of crops production, forestry, fishery
and livestock productions:
Agriculture production = crops production +livestock production+ fishing production
+forestry production
Cobb-Douglas alternative transformations
Y = A * (K/L)^a * L
In this case, production is obtained by multiplying the land L times the average productivity
of land (K/L)^ a, and by the total factor productivity A. Note that the average productivity of
land is a function of the capital intensity per unit of land (K/L), and it increases at decreas ing
rate as K/L increases.
Agriculture capital
Initial capital can be estimated in several ways, depending on the data available. When direct
estimation of capital stock does not exist, a commonly used approach is the perpetual
inventory method 10.
Other components
The relative technology level for agriculture is calculated in the Technology module and
documented in the homonymous section.
is calculated as the ratio between production and land area used for agriculture
production:
Yield
The harvested area for each crop is calculated as the land used for agriculture multiplied by the
crop intensity index, (an aggregated measure of the number of crops obtained by the same
piece of land every year), multiplied by the fraction of land allocated to each crop:
Harvested area [main crop] = Agricultural Land In Use*crop intensity index*main crop
area fraction
Harvested area [rest] = Agricultural Land In Use*crop intensity index*(1-main crop
area fraction)
The multi -cropping index is related primarily to natural conditions, including temperature and
humidity. If these conditions do not change, the multi-cropping index remains constant 11.
References
US Department of Agriculture, World Agriculture, Trends and Indicators, 1970-91, USDA,
1993
US Department of Agriculture, International Agricultural Baseline Projections to 2005 ,
AER-750, USDA/ERS
Donald O. Mitchell, and Merlinda D. Ingco, The World Food Outlook, The World Bank,
November 1993
10
The perpetual inventory method (PIM) is a method of constructing estimates of capital stock and consumption
of fixed capital from time series of gross fixed capital formation.
11
The multi-cropping index can also be related to technology and to income per capita in some countries. In
some rural parts of China, for example, the multi- cropping index dropped after per capita incomes increased. So
this index may be modeled as a function of rural income.
<total population>
AVERAGE MEAT
EXPORT DEMAND IN
TONS
meat production
in tons
<Time>
<labor force
availability>
<Time>
<total population>
fish production in
tons
FORESTRY PRODUCTION
AVERAGE FISH VALUE
IN CUBIC METERS TIME
ADDED PER TON TIME
SERIES
SERIES
forestry
production
<Time>
VALUE ADDED PER
in cubic meters
CUBIC METER OF WOOD
TIME SERIES
feasible forestry
<Forest Land>
<Time>
production
AVERAGE
FOREST DENSITY
livestock
production
fish production
forestry production
<TIME TO CONVERT
FOREST LAND>
Figure 27: Sketch of the Animal Husbandry, Fishing, and Forestry module
Explanation
Major Assumptions
Per capita meat and fish demand are a function of income alone;
Supply meets demand for livestock and fish;
Forestry production is limited by the existing amount of forest land;
Increase in the labor supply does not increase production, but the lack of it can limit
production.
Input Variables
Variable Name
Module of Origin
Households accounts
Population
Population
Labor availability and cost
Energy Supply
Land
Land
Output Variables
Variable Name
Livestock production
Fishing production
Forestry production
Meat production for domestic
market
Fish production in tons
Forest cut for wood
production
Module of Destination
Same Sector
Other Economic
Sectors
Agriculture
Agriculture
Agriculture
Other Sectors
Nutrition
Nutrition
Land
Type of
Variable
Table function
Constant
Constant
Constant
Constant
Time series
Time series
Time series
Constant
Time series
Constant
Functional explanation
The sketch in Figure 27 can be divided in three parts: the top represents meat demand and
production; the central represents fish demand and production; and the bottom represents
forestry production.
Note that values for production in this module are measured in physical quantities or at
constant prices.
Meat demand and production
Agricultural production also includes meat production. Meat includes seven categories: pork,
beef, lamb, poultry, egg, fish, and milk. T21 SF uses a general meat to represent the sum of
demand of the seven meat categories.
To calculate meat production, first per capita meat demand is determined. Per capita meat
demand is assumed to adjust over time as per capita income changes. The non-linear
relationship binding income and per capita meat demand is described in Figure 28 12. More
specifically, per capita meat demand is calculated as following:
Per capita meat demand = PC MEAT DEMAND TABLE(Medium Term Average Real
Pc Income In Usd In Ppp)*LOCAL CONDITIONS MEAT ADJUSTMENT
FACTOR/TON TO KILO CONVERSION FACTOR
Note that the local conditions meat adjustment factor is used to consider aspects of the dietary
habits in a country that are not related to income (e.g. cultural aspects).
per capita yearly meat demand in kg
80.00
70.00
60.00
50.00
40.00
30.00
20.00
10.00
0.00
0
2000
4000
6000
8000
10000
12000
Figure 28: Assumed Relationship between per Capita Income and Meat Demand
Meat production for domestic markets is
population:
Meat production for domestic market = total population*Per Capita Meat Demand
Domestic meat production in tons is the sum of meat production for domestic market and the net
export of meat:
Meat production in tons = meat production for domestic market +AVERAGE MEAT
EXPORT DEMAND IN TONS
Note that the average meat export demand in tons is considered constant.
12
This assumption is based on "International Agricultural Baseline Projections to 2005", AER-750, Washington,
DC: US Department of Agriculture, Economic Research Service.
The value added from livestock production is calculated as the meat production in tons, times
the meat value added per ton, times the effects of oil price and labor force availability on
production.
Livestock production = meat production in tons*MEAT VALUE ADDED PER TON
TIME SERIES(Time)*effect of energy price on productivity*labor force
availability
Fish demand and production
As in meat production, in T21 we only consider one generic type of fish. To calculate fish
production, first per capita fish demand is determined. The per capita fish demand is assumed to
adjust over time as per capita income changes, based on a certain elasticity of fi sh demand to
income:
Per capita fish demand = INITIAL PER CAPITA FISH DEMAND*Perceived Relative
Pc Disposable Income^ELASTICITY OF FISH DEMAND TO INCOME
The per capita fish demand is then multiplied by total population, to obtain fish production in tons :
Fish production in tons = total population*Per Capita Fish Demand
Finally, the value added from fish production is obtained by multiplying fish production in
tons by the value added per ton of fish and by the effect of energy price and labor force
availability on production:
Fish production = fish production in tons*AVERAGE FISH VALUE ADDED PER
TON TIME SERIES (Time)*effect of energy price on productivity*labor force
availability
Forestry production
Forestry production is determined as forestry production in cubic meters , multiplied by the value
added per cubic meter of wood time series , times the effect of energy price and labor availability
on production:
Forestry production = forestry production in cubic meters*VALUE ADDED PER
CUBIC METER OF WOOD TIME SERIES (Time)*effect of energy price on
productivity*labor force availability
Forestry production in cubic meters is based on the exogenous forestry production in cubic meters
time series , and on the actual availability of forest land:
References
Economic Research Service, US Department of A griculture, "International Agricultural
Baseline Projections to 2005", AER-750, Washington: USDA/ERS.
<investment
industry>
<average adult
literacy rate>
INDUSTRY
CAPITAL
ELASTICITY
INITIAL CAPITAL
INDUSTRY
<relative labor
technology>
relative capital
industry
effect of education
on productivity
Effect Of Health
On Productivity
<relative skilled labor
availability>
industry
production
total factor
productivity industry
INITIAL ADULT
LITERACY RATE
AVERAGE LIFE
CAPITAL
INDUSTRY
relative industry
employment
INITIAL INDUSTRY
PRODUCTION
<Industry
Employment>
<INITIAL INDUSTRY
EMPLOYMENT>
EFFECT HEALTH ON
PRODUCTIVITY
TABLE
TIME FOR CHANGES IN LIFE
EXPECTANCY TO AFFECT
PRODUCTIVITY
INITIAL AVERAGE
LIFE EXPECTANCY
Explanation
Major Assumptions
Industrial production follows a Cobb-Douglas form, with main production factors
including labor and capital;
Technology, health, education and the availability of energy also affect productivity.
Input Variables
Variable Name
Investment industry
Effect of energy price on productivity
Rela tive labor technology
Average adult literacy rate
Average life expectancy
Relative skilled labor availability
Industry employment
Initial industry employment
Module of Origin
Investment
Energy supply
Technology
Primary education
population
Labor availability and cost
Employment
Employment
Output Variables
Variable Name
Industry production
Capital industry
Effect of education on
productivity
Effect of health on labor
productivity
Module of Destination
Same Sector
Other Economic
Sectors
Investment
Technology
Other Sectors
Water demand
Employment
Agriculture, services
Services
Type of
Variable
Constant
Constant
Table function
Constant
Constant
Table function
Table function
Constant
Constant
Constant
Functional Explanation
The industry production function
is calculated as:
Capital industry is a stock that accumulates via investment and reduces by depreciation:
Capital industry = INTEG (investment industry - depreciation industry, INITIAL
CAPITAL INDUSTRY)
Initial capital can be estimated in several ways, depending on available data. If direct
estimations of capital stock do not exist, a commonly used technique is the perpetual
inventory method.
Effect of education, health and technology on productivity
The effects of basic education and skilled labor availability on productivity are calculated
through two distinct table functions (see Figure 30 and Figure 31). Respectively, the table
functions use as input the relative (normalized) average adult literacy rate and the relative
(normalized) skilled labor availability. The two table functions are estimated based on
productivity and primary and secondary education data.
Figure 30: the assumed relationship between adult literacy rate and productivity
Note that the relationship between adult literacy rate and agriculture productivity is assumed
non-linear. As literacy rates increase, productivity increases at a decreasing rate, and
eventually saturates at a level 20% higher than the initial one. Similarly, as the availability of
skilled labor increases productivity increases at a decreasing rate, and saturates at a level 10%
higher than the initial one. These assumptions are consistent with the law of decreasing
marginal returns: resources are first allocated to the use that has the highest possible return,
and therefore resources that are available in a second moment will be used for less productive
activities.
Figure 31: the assumed relationship between availability of skilled labor and
productivity
The effect of health on labor productivity is calculated using the table function in Figure 32.
The table function uses the relative (normalized) life expectancy (a proxy for health) as input
and produces as output a multiplicative effect that is then applied to labor productivity. We
assume that health affects productivity in two ways. First, as health decreases (increases), the
strength and dynamism of the labor force decreases (increases). Second, as health decreases
(increases), mortality increases (decreases) and consequently workers have a higher (lower)
turnover rate, which implies for the whole labor force a slower (faster) accumulation of
knowledge and skills.
Figure 32: assumed relationship between life expectancy and labor productivity
Note that the relationship between life expectancy and productivity is assumed non-linear:
productivity decreases fast as life expectancy decreases below the initial value; however the
effect saturates in the case of extremely high life expectancy. The underlying assumption is
that beyond certain levels, a further increase in the life expectancy leads only to a limited
increase in productivity.
The relative labor technology for industry is calculated in the Technology module (documented
in the homonymous section of the documentation) and is directly used in a multiplicative
form in the TFP equation.
References
Meadows, D. and et al, Dynamics of Growth in a Finite World , Wright-Allen Press 1974.
Dornbusch, R. and Fischer, S., Macroeconomics, 6th edition. McGraw-Hill, Inc. 1994
World Bank, World Development Indicators 2000, on CD-ROM.
Amor Tahari et al, Sources of growth in Sub -Saharan Africa, IMF working paper, September
2004
Barry Bosworth et al, Accounting for differences in economic growth , Conference on
"Structural Adjustment Policies in the 1990s: Experience and Prospects" October 5-6,
1995, organized by the Institute of Developing Economies, Tokyo, Japan.
<investment
services>
depreciation services
AVERAGE LIFE
CAPITAL SERVICE
SERVICES
CAPITAL
ELASTICITY
INITIAL SERVICES
PRODUCTION
services
production
<Services
Employment>
<relative labor
technology>
total factor
productivity
services
<Effect Of Health On
Productivity>
relative services
employment
<INITIAL SERVICES
EMPLOYMENT>
<effect of education
on productivity>
Explanation
Major Assumptions
Services production follows a Cobb-Douglas form, with main production factors
including labor and capital;
Technology, health, education and the availability of energy also affect productivity.
Input Variables
Variable Name
Investment services
Effect of energy price on productivity
Relative labor technology
Effect of education on labor productivity
Effect of health on labor productivity
Services employment
Initial services employment
Module of Origin
Investment
Energy Supply
Technology
Primary education
Industry
Employment
Employment
Output Variables
Variable Name
Module of Destination
Same Sector
Services production
Capital services
Constants and Table functions
Variable Name
Initial capital services
Services capital elasticity
Type of
Variable
Constant
Constant
Constant
Constant
Other Economic
Sectors
Investment
Technology
Other Sectors
Employment
Functional Explanation
The structure of the Services module is very similar to that of the Industry module. For a
functional explanation of the Services module, refer to the functional explanation section in
the Industry module.
References
Dornbusch, R. and Fischer, S., Macroeconomics, 6th edition. McGraw-Hill, Inc. 1994
World Bank, World Development Indicators 2000, on CD-ROM.
Lynn R. Brown, The potential impact of aids on population and economic growth rates,
June 1997, 2020 Brief No. 43, IFPRI
Amor Tahari et al, Sources of growth in Sub -Saharan Africa, IMF working paper, September
2004
Barry Bosworth et al, Accounting for differences in economic growth , Conference on
"Structural Adjustment Policies in the 1990s: Experience and Prospects" October 5-6,
1995, organized by the Institute of Developing Economies, Tokyo, Japan.
<Capital
Agriculture>
<INITIAL CAPITAL
INDUSTRY>
<INITIAL CAPITAL
SERVICES>
<Capital Services>
<INITIAL CAPITAL
AGRICULTURE>
<investment non
mining industry>
average
technology level
relative resources
technology
Technology
<Relative
Labor Cost>
technology discard
technology investment
<depreciation
agriculture>
desired level of
technology for the
new capital
RELATIVE
TECHNOLOGY
COST TABLE
<depreciation
industry>
<depreciation
services>
INITIAL AVERAGE
LEVEL OF
TECHNOLOGY
relative labor
technology
<Time>
<relative average
energy price>
The creation and discarding of capital technology (for simplicity technology in the rest of
the text) in T21 is represented as a co-flow of investment and depreciation in the Production
sector. Based on this structure, two aggregate indicators are derived, which represent the state
of labor -related and resources-related technology for agriculture, industry and services capital.
To calibrate technological growth in T21, we use information from available growth
accounting studies carried out in the country object of the analysis or in countries with similar
characteristics.
Explanation
Major Assumptions
Input Variables
Variable Name
Investment crops
Investment industry
Investment services
Initial capital agriculture
Initial capital industry
Initial capital services
Capital agriculture
Capital industry
Capital services
Depreciation agriculture
Depreciation industry
Depreciation services
Relative labor cost
Relative average energy price
Module of Origin
Investment
Investment
Investment
Agriculture
Industry
Services
Agriculture
Industry
Services
Agriculture
Industry
Services
Labor availability and cost
Energy demand
Output Variables
Variable Name
Module of Destination
In the Same Sector
In Other Economic
Sectors
Agriculture,
industry, services
Type of
Variable
Time series
In Other Sectors
Energy demand,
water demand,
fossil fuel
production
Employment
Constant
experts opinion
For definition equal to 1
Functional Explanation
To explain how relative levels of labor-related and resources-related technology are calculated
in T21, we have to proceed in steps. First we describe how we measure technology; second
how technology can increase or decrease over time; then explain the meaning and use of
technology indicators.
Technology
Technology is
Note from the equation above that subscripts are used to differentiate between labor -related
and resource -related technology (subscript [tech], containing the elements [labor, resources]);
and among the different types of production activities (subscript [sectors], containing the
elements [agri, ind, serv]).
The desired level of technology for new capital is calculated as:
Desired level of technology for the new capital [sectors, labor] = INITIAL AVERAGE
LEVEL OF TECHNOLOGY [sectors, LABOR]*Relative Labor Cost/RELATIVE
TECHNOLOGY COST TIME SERIES [sectors, LABOR] (Time)
Desired level of technology for the new capital [sectors, resources] = INITIAL
AVERAGE LEVEL OF TECHNOLOGY [sectors, RESOURCES]*relative
average energy price/RELATIVE TECHNOLOGY COST TIME SERIES [sectors,
RESOURCES] (Time)
In the first of the two equations above, decisions about levels of labor -related technology for
new capital are made based on the ratio between the cost of labor (from the Labor Availability
and Cost module) and the cost of technology (which is exogenously determined). It is
assumed that as labor costs increase (decreases), the labor -related technology of new capital
proportionally increases (decreases), to reduce (increase) the amount of labor employed per
unit of capital. In other words, as labor becomes more expensive, producers seek a more
capital intensive type of production process; if labor becomes cheaper, producers will opt for
labor intensive production.
The same logic and assumptions are used to determine resource-related technology for new
capital. Note that in the second equation, the average energy price is used as a proxy for the
cost of energy and water.
Technology Discard
Technology discard is
Based on the stock of technology and the stock of capital in the Production sector, the average
level of technology is determined:
Average technology level [AGRI, tech] = Technology [AGRI, tech]/Capital Services
Average technology level [IND, tech] = Technology [IND, tech]/Capital Services
Average technology level [SERV, tech] = Technology [SERV, tech]/Capital Services
The average technology level represents the amount of technology (measured in technology
units) that is embedded in each unit of capital in agriculture, industry and services.
This variable is then normalized, to obtain the relative labor technology and the relative resources
technology :
Relative labor technology [sectors] = average technology level [sectors,
LABOR]/INITIAL AVERAGE LEVEL OF TECHNOLOGY [sectors, LABOR]
Relative resources technology [sectors] = average technology level [sectors,
RESOURCES] /INITIAL AVERAGE LEVEL OF TECHNOLOGY [sectors,
RESOURCES]
These two variables are aggregate measures of the technological level of the capital over time,
and are used in several other sectors of the model.
Relative labor technology is
technology as
proxy for the efficacy of the equipment used for discovery and extraction of
fossil fuels.
References
World Bank, World Development Indicators 2000, on CD-ROM.
John D. Sterman, Business Dynamics, New York: Mc Graw -Hill, 2000
INITIAL REAL PC
DISPOSABLE INCOME
<gdp deflator>
<nominal gdp at
market prices>
private domestic
savings
<budgetary
revenue>
<interest on
domestic debt>
<private transfers>
disposable income
<domestic
financing>
<arrears>
TIME TO PERCEIVE
CHANGES IN TOTAL
IMPORT
propensity to
consume
private savings
ELASTICITY OF
PROPENSITY TO SAVE
TO INCOME
net savings
TIME TO ADJUST
RESERVES
net change in
reserves
propensity to
save
private
consumption
<adjustment>
<errors and
omissions>
Gross
International
Reserves
desired
reserves
desired import
coverage
private domestic
investment
<Time>
private investment
IMPORT
COVERAGE
TIME SERIES
<total import>
Perceived Total
Import
INITIAL
PROPENSITY
TO SAVE
Perceived Relative Pc
Disposable Income
TIME TO PERCEIVE
CHANGES IN PC
INCOME
total households
revenue
<private factor
income>
real pc disposable
income
relative pc
disposable
income
foreign direct
investment
Explanation
Major Assumptions
All value added created in the economy is transferred to households;
All domestic taxes and duties are paid by households;
The propensity to consume is determined based on per capita income level (Engels
Law);
Input Variables
Variable Name
Nominal GDP at market prices
GDP deflator
Interest on domestic debt
Private transfers
Private factor income
Budgetary revenue
Total population
Financing
Errors and omissions
Arrears
Adjustment
Domestic financing
Resources balance
Private capital and financial transfers
Total import
Public investment
Module of Origin
Aggregate Production and Income
Relative prices
Government debt
Balance of payments
Balance of payments
Government revenue
Population
Balance of payments
Balance of payments
Government balance and financing
Government balance and financing
Government balance and financing
Balance of payments
Balance of payments
Aggregate Production and Income
Government expenditure
Output Variables
Variable Name
Module of Destination
In the Same Sector
In Other Economic
Sectors
Disposable income
Perceived relative pc
disposable income
Private consumption
Private investment
Constants and Table functions
Variable Name
In Other Sectors
Income distribution
Water demand,
fertility, primary
education, secondary
education, animal
husbandry-fisheryforestry
Aggregate Production
and Income
Investment
Type of
Variable
Constant
Constant
Constant
Constant
Constant
Functional Explanation
The Household Account module contains several important elements that are functional in the
determination of consumption, savings and investment. These elements are described in detail
below. Note that economic flows in this module are considered at current prices, with the
exception of real pc disposable income.
Disposable income
Household disposable income is determined in the upper left part of the sketch in Figure 35. To
calculate disposable income, first total households revenue is determined as the sum of value
added in the economy, net payment for factors of production lent abroad, net remittances
received, and interests paid by the government to households:
Total households revenue = nominal GDP at market prices +private factor income
+private transfers +interest on domestic debt
Disposable income is
that the formulation MIN is used to make sure that the propensity to save can never exceed 1.
Private Consumption and savings
Private consumption is
Gross international reserves are calculated based on imports. It is assumed that the banking
system will manage the reserves so as to always guarantee a certain number of months of
coverage of imports. The level of reserves desired by the banking system is thus:
Desired reserves = Perceived Total Import*desired import coverage
Gross international reserves are then represented as a stock, which adjusts towards the desired
level of reserves over time:
Net change in reserves = (desired reserves-Gross International Reserves)/TIME TO
ADJUST RESERVES
The net savings in the banking system is the net difference between domestic savings and
investment. In other words, net savings represents all the money saved by households and the
government plus that which is not invested. Net savings are assumed equal to net change in
reserves, plus adjustment and arrears from the government budget, and minus errors and
omissions from the balance of payments:
Net savings = net change in reserves -errors and omissions +adjustment +arrears
Note that variable errors and omissions, and adjustment and arrears are set to historical values
for the past and zero for the future.
Private Investment
Private domestic investment is
financing, minus net savings :
Note that under foreign direct investment in T21, we include all types of private capital and
financial transfers from the balance of payments.
References
World Bank, Model building RMSM-X reference guide, July 1995.
World Bank, World Development Indicators 2000 on CD-ROM.
Taxes On Income
And Profits
direct taxes
<Time>
TAXES ON GOODS
AND SERVICES TIME
SERIES
<nominal production
by sector>
<import>
IMPORT TAXES
TIME SERIES
tax revenue
TIME TO
COLLECT TAXES
Import Taxes
By Sector
INITIAL IMPORT
TAXES
<Time>
<nominal gdp at
factor cost>
OTHER TAX
REVENUE TIME
SERIES
taxes on goods
and services
Taxes On Goods
And Services By
Sector
indirect
taxes
taxes on
international
trade
budgetary revenue
overall fiscal
pressure on gdp
Other Tax
Revenue
<nominal gdp at
factor cost>
NONTAX
REVENUE TIME
SERIES
<total households
revenue>
nontax
revenue
total revenue
<Time>
SPECIAL FUNDS
ETC TIME SERIES
grants
<Time>
Explanation
Major Assumptions
Input Variables
Variable Name
Nominal GDP at factor cost
Nominal production by sector
Module of Origin
Aggregate Production and Income
Aggregate Production and Income
Import
Total households revenue
International trade
Households accounts
Output Variables
Variable Name
Module of Destination
In the Same Sector
In Other Economic
Sectors
Relative prices
In Other
Sectors
Investment, relative
prices
Balance of payments
Government balance and
financing
Government expenditure
Government balance and
financing, government
debt
Type of
Variable
Time series
Time series
Time series
Time series
Constant
Time series
Time series
Time series
Households accounts
Functional Explanation
The Government Revenue module has a tree -shaped structure where the various revenue
items are summed together to determine the total revenue and grants. The following
paragraphs describe how the different sources of revenue are determined. Note that economic
flows in this module are considered at current prices.
Taxes
Taxes on income and profits and other tax revenue
Note that the values used to initialize the delay functions above correspond to the values
estimated by the IMF for those types of taxes for 1990.
The total taxes on goods and services are the sum of the taxes calculated for the different types
of products:
Taxes on goods and service = SUM (Taxes on goods and services by sector[sectors!])
Import taxes are also differentiated depending on the type of product (agricultural, industrial,
services) they are applied to. They are determined as a share of the value of imports:
Import taxes by sector [sectors] = DELAY N( import[sectors]*IMPORT TAXES
TIME SERIES[sectors](Time) , TIME TO COLLECT TAXES, INITIAL
IMPORT TAXES[sectors], 1)
Import taxes by sector are eventually summed up to obtain taxes on international trade . Taxes on
income and profits are equivalent to direct taxes, and the sum of taxes on goods and services,
taxes on international trade and other taxes determines indirect taxes. The sum of direct and
indirect taxes constitutes the tax revenue :
Nontax revenue = nominal gdp at factor cost *NONTAX REVENUE TIME SERIES
(Time)
Nontax revenue
Budgetary revenue = MIN (total households revenue, tax revenue +nontax revenue)
Note that in the equation above, a MIN formulation is used to ensure that budgetary revenue,
withdrawn from households, never gets bigger than total households revenue.
Special funds and grants
Special funds etc are
total revenue :
Finally grants, which are also exogenously determined, are added to total revenue to obtain
revenue and grants:
Reven ue and grants = total revenue +grants
References
IMF, Mali: Statistical Annex, IMF Staff Country Report No. 98/14, 1998 International
Monetary Fund
IMF, Mali: Selected Issues and Statistical Annex, IMF Staff Country Report No. 00/128, 2000
International Monetary Fund
IMF, Mali: Selected Issues and Statistical Annex, IMF Staff Country Report No. 04/10, 2004
International Monetary Fund
<Time>
general public
services expenditure
defense
expenditure
expenditure and
net lending
<gdp deflator>
social services
expenditure
<Time>
ECONOMIC SERVICES
EXPENDITURE AS SHARE OF
BUDGET TIME SERIES
OTHER EXPENDITURE AS
SHARE OF BUDGET TIME
SERIES
<Time>
EDUCATION
EXPENDITURE AS SHARE
OF BUDGET TIME SERIES
HEALTH EXPENDITURE
AS SHARE OF BUDGET
TIME SERIES
ROADS EXPENDITURE AS
A SHARE OF BUDGET
TIME SERIES
extraordinary
expenditure and lending
economic services
expenditure
real budgetary
expenditure
other
expenditure
education
expenditure
health
expenditure
<Time>
budgetary
expenditure
<interest on public
debt>
roads expenditure
<revenue and
grants>
Explanation
Major Assumptions
General public services expenditure is related to the population size;
Extraordinary expenditure and lending is exogenous;
The size of the other items of expenditure calculated in this module is related to the
size of the government budget.
Input Variables
Variable Name
Revenue and grants
Special funds etc
Interest on public debt
GDP deflator
Total population
Module of Origin
Government revenue
Government revenue
Government debt
Relative prices
Population
Output Variables
Variable Name
General public expenditure
Defense expenditure
Social services expenditure
Economic services
expenditure
Other expenditure
Education expenditure
Health expenditure
Roads expenditure
Extraordinary expenditure and
lending
Expenditure and net lending
Budgetary expenditure
Module of Destination
In the Same Sector
In Other Economic
Sectors
In Other Sectors
Primary
education
Access to basic
health care
Roads
Type of
Variable
Time series
Time series
IMF data
Time series
Time series
IMF data
IMF data
Time series
IMF data
Time series
IMF data
Time series
IMF data
IMF data
Time series
IMF data
Time series
Functional Explanation
In this module, expenditures are organized using a functional classification, i.e. according to
the destination of the funds allocated by the government. The following paragraphs describe
how the various expenditures are calculated and tot al expenditures are determined. Note that
all the economic flows in this module, except real budgetary expenditure, are expressed in
nominal terms (current prices).
General public services expenditure
It is assumed that expenditures for general public services are strongly related to population
levels. In particular, we assumed that in order to guarantee a proper level of general public
services, the government spends a certain amount of funds per capita every year. General
public services expenditure is therefore calculated:
General public services expenditure = total population*REAL PC GENERAL
PUBLIC SERVICES EXPENDITURE TIME SERIES (Time)*gdp deflator
The real pc general public services expenditure time series is a time series containing past
government expenditures per capita and an estimate of per capita expenditure for the next
decades.
Extraordinary expenditure and lending, interest on public debt and special funds etc
The other items of expenditure are calculated as fractions of the total revenue and grants. For
example, education expenditure is calculated as:
Education expenditure = EDUCATION EXPENDITURE AS SHARE OF BUDGET
TIME SERIES (Time)*revenue and grants
of expenditure, excluding
References
IMF, Mali: Statistical Annex, IMF Staff Country Report No. 98/14, 1998 International
Monetary Fund
IMF, Mali: Selected Issues and Statistical Annex, IMF Staff Country Report No. 00/128, 2000
International Monetary Fund
IMF, Mali: Selected Issues and Statistical Annex, IMF Staff Country Report No. 04/10, 2004
International Monetary Fund
public
consumption
current
expenditure
CAPITAL SHARE OF
ECONOMIC SERVICES
EXPENDITURE TABLE
<budgetary
expenditure>
<extraordinary
expenditure and
lending>
public investment
<economic services
expenditure>
<Time>
CAPITAL SHARE OF
GENERAL SERVICES
EXPENDITURE TABLE
<general public
services expenditure>
<roads
expenditure>
CAPITAL SHARE OF
ROADS EXPENDITURE
TABLE
<health
expenditure>
capital
expenditure
CAPITAL SHARE OF
DEFENSE EXPENDITURE
TABLE
CAPITAL SHARE OF
HEALTH EXPENDITURE
TABLE
<other
expenditure>
<defense
expenditure>
CAPITAL SHARE OF <social services CAPITAL SHARE OF
<education
EDUCATION
SOCIAL SERVICES
expenditure> EXPENDITURE TABLE expenditure>
EXPENDITURE TABLE
CAPITAL SHARE OF
OTHER EXPENDITURE
TABLE
Explanation
Major Assumptions
The capital share of various items of expenditure is exogenously determined;
Extraordinary expenditures and lending is considered consumption expenditure;
Interest on public debt is not considered as consumption or investment, but as transfers
to domestic and foreign households.
Input Variables
Variable Name
Budgetary expenditure
Economic services expenditure
General public services expenditure
Defense expenditure
Module of Origin
Government expenditure
Government expenditure
Government expenditure
Government expenditure
Government expenditure
Government expenditure
Government expenditure
Government expenditure
Government expenditure
Government expenditure
Government expenditure
Output Variable s
Variable Name
Module of Destination
In the Same Sector
Public investment
Public consumption
Constants and Table functions
Variable Name
Capital share of economic services
expenditure time series
Capital share of general services expenditure
time series
Capital share of defense expenditure time
series
Capital share of social services expenditure
time series
Capital share of education expenditure time
series
Capital share of other expenditure time series
Capital share of health expenditure time
series
Capital share of roads expenditure time series
In Other Economic
Sectors
Households accounts
Households accounts
Type of
Variable
Time series
Time series
IMF data
Time series
IMF data
Time series
IMF data
Time series
IMF data
Time series
Time series
IMF data
IMF data
Time series
IMF data
In Other
Sectors
IMF data
Functional Explanation
The calculation of capital and current expenditures in the public investment and consumption
module is straightforward. First, total capital expenditures is calculated by multiplying each
item of expenditure by its capital share:
Capital expenditure = general public services expenditure*CAPITAL SHARE OF
GENERAL SERVICES EXPENDITURE TIME SERIES(Time)+
defense expenditure*CAPITAL SHARE OF DEFENSE EXPENDITURE TIME
SERIES(Time)+
education expenditure*CAPITAL SHARE OF EDUCATION EXPENDITURE TIME
SERIES(Time)+
health expenditure*CAPITAL SHARE OF HEALTH EXPENDITURE TIME
SERIES(Time)+
economic services expenditure*CAPITAL SHARE OF ECONOMIC SERVICES
EXPENDITURE TIME SERIES(Time)+
social services expenditure*CAPITAL SHARE OF SOCIAL SERVICES
EXPENDITURE TIME SERIES(Time)+
Note that capital shares of each item of expenditure are an exogenously determined time
series. Also note that capital expenditures, as well as the other economic flows in this module,
are considered at current prices.
Public investment is set equal to capital expenditure, and current expenditure is determined as
the difference between public investment and budgetary expenditure:
Budgetary expenditure = budgetary expenditure-public investment
References
IMF, Mali: Statistical Annex, IMF Staff Country Report No. 98/14, 1998 International
Monetary Fund
IMF, Mali: Selected Issues and Statistical Annex, IMF Staff Country Report No. 00/128, 2000
International Monetary Fund
IMF, Mali: Selected Issues and Statistical Annex, IMF Staff Country Report No. 04/10, 2004
International Monetary Fund
adjustment
<Time>
ARREARS TIME
SERIES
arrears
overall fiscal balance
cash basis
<expenditure and
net lending>
<revenue and
grants>
government
financing
overall government
budget cash basis
<adjustment>
external financing
<Time>
SHARE OF FINANCING
FROM FOREIGN SOURCES
overall government
TIME SERIES
budget order basis
<arrears>
<interest on
foreign debt>
<grants>
<public
investment>
<public
consumption>
<budgetary
revenue>
public resources
from abroad
public resources
gap
public domestic
savings
domestic financing
public resources
balance
<arrears>
<interest on
domestic debt>
<adjustment>
Explanation
Major Assumptions
Government financing is the residual in the government accounts;
Domestic and foreign sources are always available to finance the government deficit.
Input Variables
Variable Name
Expenditure and net lending
Revenue and grants
Interest on foreign debt
Grants
Public investment
Public consumption
Budgetary revenue
Interest on domestic debt
Module of Origin
Government expenditure
Government revenue
Government debt
Government revenue
Public investment and consumption
Public investment and consumption
Government revenue
Government debt
Output Variables
Variable Name
Module of Destination
In the Same Sector
Domestic financing
External financing
Government debt
Government debt
Type of
Variable
Time series
Time series
Time series
In Other Economic
Sectors
Households accounts
Balance of payments
In Other
Sectors
Functional Explanation
The following paragraphs describe how government financing and public resources balance are
determined. Note that all the economic flows in this module are expressed in nominal terms
(current prices).
Government financing
To determine the required government financing, first overall fiscal balance order basis is
determined:
Overall fiscal balance order basis = revenue and grants -expenditure and net lending
The overall fiscal balance cash basis is then calculated as the balance order basis plus adjustment
and arrears:
Overall fiscal balance cash basis = overall fiscal balance order basis +adjustment
+arrears
Note that adjustment usually represents statistical errors in the measurement of public financial
flows. Arrears are corrections to accounting that are used to keep track of expenditures
occurring in a certain fiscal period in the order basis system, but that are accounted for in a
different period in the cash basis system. Both arrears and adjustment are set to zero after for
the future.
Government financing is
imbalance:
Government financing = -overall fiscal balance cash basis
Government financing consists
The following paragraphs describe how the variables in the right hand side are determined.
The public resources gap in the equation above is defined as the public domestic savings minus
the public investment:
Public resources gap = public domestic savings -public investment
The public domestic savings , in turn, are given by the budgetary revenue minus the public
consumption.
The public resources from abroad are given by the sum of external financing and grants, minus
interest paid to foreign entities:
Public resource from abroad = external financing +grants-interest on foreign debt
Finally, the transfer from public to private includes interest on domestic debt and domestic financing
(with a negative sign, since it represents a net flow from households to the government):
Transfer from public to private = -domestic financing +interest on domestic debt
References
IMF, Mali: Statistical Annex, IMF Staff C ountry Report No. 98/14, 1998 International
Monetary Fund
IMF, Mali: Selected Issues and Statistical Annex, IMF Staff Country Report No. 00/128, 2000
International Monetary Fund
IMF, Mali: Selected Issues and Statistical Annex, IMF Staff Country Report No. 04/10, 2004
International Monetary Fund
<Time>
changes in domestic
debt from other sources
interest on
domestic debt
Government
Domestic Debt
net change in
domestic debt
government
debt over gdp
CHANGES IN DOMESTIC
DEBT FROM OTHER
SOURCES TIME SERIES
<Time>
FRACTIONAL INTEREST
RATE ON DOMESTIC DEBT
TIME SERIES
<domestic
financing>
total
government
debt
<nominal gdp at
market prices>
<Time>
other foreign debt
changes
<Time>
external financing
in usd
<external
financing>
interest on public
debt
<nominal exchange
rate with us>
Government
Foreign Debt
interest on foreign
debt
interest on foreign
debt in usd
FRACTIONAL INTEREST
RATE ON FOREIGN DEBT
TIME SERIES
<revenue and
grants>
Explanation
Major Assumptions
Changes in debt can happen through financing or other sources;
Other financing sources are exogenous;
The same interest rate is applied to all domestic debts;
The same interest rate is applied to all foreign debt;
Interest rates are exogenous.
Input Variables
Variable Name
Domestic financing
Module of Origin
Government balance and financing
External financing
Nominal GDP at market prices
Nominal exchange rate with US
Total export
Revenue and grants
Output Variables
Variable Name
Interest on domestic debt
Interest on public debt
Module of Destination
In the Same Sector
Government balance and
financing
Government expenditure,
Public investment and
consumption
Government balance and
financing
In Other Economic
Sectors
Households accounts
In Other
Sectors
Balance of payments
Type of
Variable
Time series
Time series
Time series
IMF data
IMF data
Time series
IMF data
IMF data
Functional Explanation
The following paragraphs describe how the stocks of domestic and foreign debt are calculated
and how interest paid by the government is derived. Note that all the economic variables in
this module are expressed in nominal terms (current prices).
Domestic debt
Domestic debt is of one type only and includes both domestic debt with the central bank and
domestic debt with commercial banks. Domestic debt is represented as stock variable and
accumulates a net flow of debt change:
Net change in domestic debt = domestic financing +changes in domestic debt from
other sources
Domestic financing is calculated in the Government Balance and Financing
changes in domestic debt from other sources is exogenously determined.
module, while
Foreign debt
Foreign debt is represented as a stock variable, and its value is expressed in U.S. dollars
(USD). The flow that accumulates in the stock of foreign debt (external financing in USD )
includes external financing and changes in debt of another nature ( other foreign debt changes):
External financing in USD = external financing/nominal exchange rate with us +other
foreign debt cha nges
Interests
The interest the government pays on debts is calculated applying an exogenous marginal
interest rate to the stock of domestic and foreign debt. The interest on foreign debt in USD is
thus calculated as:
Interest on foreign debt in USD = Government Foreign Debt*FRACTIONAL
INTEREST RATE ON FOREIGN DEBT TIME SERIES(Time)
Then, by applying the nominal excha nge rate, interests on foreign debt in USD are converted to
local currency:
Interest on foreign debt = Interest on foreign debt in USD *nominal exchange rate with
us
The formulation used to determine interest on domestic debt follows the same logic:
Interest on domestic debt= Government Domestic Debt*FRACTIONAL INTEREST
RATE ON DOMESTIC DEBT TIME SERIES(Time)
Note that when aggregating different types of debt into one variable, it might result that the
overall interest rate is negative. This can happen, for example, in case the government has a
large credit with the commercial banks but he receives little or no interest on it, and at the
same time it has a smaller debt with the central bank, on which it pays substantial interests. In
this case the net position of the government would be positive, but still it would have to pay
interests, resulting in a negative interest rate. In such situation the debt should be split into
debt with the central bank and debt with commercial banks, and interest rates properly
represented.
Finally, the total interest paid by the government is calculated as the sum of interest on
domestic and foreign debt:
Interest on public debt = interest on foreign debt + interest on domestic debt
Indicators
Some important indicators are also calculated in this module, including: interest on foreign debt
over export, interest on foreign debt over government revenue , total government debt and
government debt over GDP . These indicators are fundamental in monitoring the level and the
sustainability of public debt, and are calculated as follows:
Interest on foreign debt over export = interest on foreign debt/total export
Interest on foreign debt over government revenue= interest on foreign debt/revenue
and grants
Total government debt = government domestic debt+Government Foreign
Debt*nominal exchange rate with us
Government debt over GDP = total government debt/nominal gdp at market prices
References
IMF, Mali: Statistical Annex, IMF Staff Country Report No. 98/14, 1998 International
Monetary Fund
IMF, Mali: Selected Issues and Statistical Annex, IMF Staff Country Report No. 00/128, 2000
International Monetary Fund
IMF, Mali: Selected Issues and Statistical Annex, IMF Staff Country Report No. 04/10, 2004
International Monetary Fund
relative domestic
consumer price
<nominal gdp at
market prices>
fraction of import demand
satisfied by import
effect of convenience of
domestic goods on import
NOMINAL
EXCHANGE RATE
ELASTICITY OF
total import
WITH US TIME SERIES
IMPORTS TO
effect of gdp
demand
import
IMPORT PRICES INITIAL NOMINAL
on
import
total imports in usd
GDP MP
import demand
<nominal gdp at
INITIAL
<real production
market prices>
IMPORTS
nominal exchange
by sector>
ELASTICITY OF
total import of
rate with us
ELASTICITY OF
IMPORT TO
import
production
goods
balance of trade
effect of relative
IMPORT TO GDP
RELATIVE PRICES
ratio
in usd
prices on imports <nominal gdp at
<Time>
<relative prices>
market prices>
<import prices>
export
ELASTICITY OF EXPORT
total export of
production ratio
TO RELATIVE PRICES
total exports in usd
goods
export gdp ratio
net export
effect of relative
<INFLATION
ELASTICITY OF
prices on export
EU INFLATION RATE AFTER
EXPORT TO ROW GDP
RATE
AFTER
2002
effect of row gdp
2002>
export
<Producer Prices>
on export
<Relative Row
effect of exchange
EXCHANGE RATE
fractional growth of
Gdp>
<Time>
rate on export
WITH EU IN 2002
exchange rate with eu
nominal exchange
INITIAL
after 2002
rate with eu
SHARE OF EXPORTS TO
EXPORTS
<Time>
relative exchange
INDUSTRIALIZED
rate with eu
COUNTRIES TIME SERIES
NOMINAL
Exchange Rate
INITIAL NOMINAL EXCHANGE
RELATIVE EXCHANGE RATE WITH
With Eu After 2002 growth of exchange
EXCHANGE RATE RATE WITH EU
DEVELOPING COUNTRIES
WITH EU
TIME SERIES
rate with eu
Explanation
Major Assumptions
Export is a function of ROW GDP growth, relative prices, and exchange rate;
Total imports are the residual in the GDP identity;
Sector distribution of imports depends on domestic GDP, consumer prices and import
prices.
Input Variables
Variable Name
Import prices
Domestic consumer price
Nominal GDP at market prices
Relative prices
Producer prices
Real production by sector
Inflation rate after 2002
Total import
Relative ROW GDP
Module of Origin
Relative prices
Relative prices
Aggregate Production and Income
Relative prices
Relative prices
Aggregate Production and Income
Relative prices
Aggregate Production and Income
Aggregate Production and Income
Output Variables
Variable Name
Relative exchange rate with
EU
Nominal exchange rate with
US
Import
Export
Net export
Module of Destination
In the Same Sector
In Other Economic
Sectors
Relative prices
In Other
Sectors
Government debt
Balance of payments
Balance of payments
Government revenue,
relative prices
relative prices
Nutrition
Type of
Variable
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Time series
Functional Explanation
The following paragraphs describe how exports and imports are calculated. Note that all the
economic variables in this module are expressed in nominal terms (current prices).
Export
Exports for each type of goods and service is determined based on the initial level of export,
the ROW GDP growth, relative prices, and the exchange rate.
The effect of ROW GDP on export is determined by applying to the relative (normalized) ROW
GDP the elasticity of export to ROW GDP:
Effect of ROW GDP on export [sectors] = Relative ROW GDP^ ELASTICITY OF
EXPORT TO ROW GDP[sectors]
Note that the subscript [sectors] is used to keep track separately of the intensity of the effect
on the various types of export products (agricultural, industrial and services).
Similarly, the effect of relative prices on export is calculated as:
Effect of relative prices on export [sectors] = (Relative Prices[sectors])^ELASTICITY
OF EXPORT TO RELATIVE PRICES[sectors]
The formulation used to determine effect of exchange rate on export is more elaborate, as it
considers the influence of changes in the exchange rate with EU (the major industrialized
trade partner for the country under analysis) and with the average for developing countries:
Effect of exchange rate on export = (relative exchange rate with EU)*SHARE OF
EXPORTS TO INDUSTRIALIZED COUNTRIES (Time) +RELATIVE
EXCHANGE RATE WITH DEVELOPING COUNTRIES *(1-SHARE OF
EXPORTS TO INDUSTRIALIZED COUNTRIES (Time))
Note that the effect of exchange rate on export is assumed linear, i.e. with an elasticity value of
1.
Finally, export is calculated as:
Export [sectors] = MIN(real production by sector[sectors]*Producer Prices[sectors],
INITIAL EXPORTS[sectors]* effect of row gdp on export[sectors]* effect of
relative prices on export[sectors]* effect of exchange rate on export)
Note that the MIN formulation is used to guarantee that under no circumstances can export
exceed production.
Import
The total level of imports is determined as the residual in the GDP equation in the aggregate
production and income sector. The distribution of total import among the types of goods and
services is calculated based on initial import and the effect of convenience of domestic goods
(with respect to imported goods), the effect of relative prices, and the effect of domestic GDP.
The effect of convenience of domestic goods on imports is calculated elevating the relative
(normalized) convenience of domestic goods to the power of the elasticity of im ports to such
variable:
Effect of convenience of domestic goods on imports [sectors] = relative convenience of
domestic goods[sectors]^ELASTICITY OF IMPORTS TO IMPORT
PRICES[sectors]
which is only used to determine the fraction of imports per type of goods and
services and not the total level of import, is subsequently calculated as:
Import demand [sectors] = INITIAL IMPORTS [sectors]* Effect Of Convenience Of
Domestic Goods On Import[sectors]* Effect Of Gdp On Import[sectors]* effect of
relative prices on imports[sectors]
Finally, the import for each type of goods and services are calculated as:
Import [sectors] = import demand[sectors]*fraction of import demand satisfied by
import
The fraction of import demand satisfied by import is determined as total imports calculated as the
residual of the GDP identity, divided by total import demand.
References
Devarajan, Lewis and Robinson, From stylized to applied models: Building multi-sector
models for policy analysis, Working paper No. 616, UC Berkeley, September 1991.
World Bank, World Development Indicators 2000 on CD-ROM.
total export
<import>
trade balance
resources balance
net services
current account
private factor
income
errors and
omissions
<Time>
overall balance
debt relief
<nominal gdp at
market prices>
PRIVATE CAPITAL
TRANSFERS OVER GDP
TIME SERIES
PRIVATE FINANCIAL
TRANSFERS OVER GDP
TIME SERIES
<external
financing>
private capital and
financial transfers
financing
Explanation
Major Assumptions
Private factor income and private transfers are exogenous;
Private capital and financial transfers are affected by the level of fiscal pressure.
Input Variables
Variable Name
Export
Import
Interest on foreign debt
Grants
Nominal GDP at market prices
External financing
Module of Origin
International trade
International trade
Government debt
Government revenue
Aggregate Production and Income
Government balance and financing
Output Variables
Module of Destination
Variable Name
In Other
Sectors
Type of
Variable
Private factor income time series
Time series
Private transfers time series
Time series
Private capital transfers over GDP time series Time series
Private financial transfers over GDP time
Time series
series
Functional Explanation
Accounting in the Balance of Payments module has a tree-shaped structure (see Figure 42).
On the left part of the sketch, basic cross-border economic flows are represented. These flows
are gradually added together to calculate more aggregate indicators, until eventually the
overall balance of payments (the sum of all flows) is determined. The upper part of the sketch
contains the flows belonging to the current account, while the lower part of the sketch
contains those belonging to the capital and financial account.
Current account
In the top left corner of the sketch, import and export from the International Trade module are
used to define the import and export of goods and separately of non-factor services. The trade
balance is then defined as:
Trade balance = export of goods +import of goods
Similarly, net services (the balance of non factor services) is calculated as:
Net services = export of non factor services -import of non factor services
The sum of trade balance and net services gives the resources balance :
Resources balance = trade balance +net services
Just below in the sketch, the net factor income is determined. The net factor income includes the
public factor income (assumed to be equal to the interest on foreign debt, taken with a negative
sign) and the private factor income (exogenous):
Net factor income = private factor income +public factor income
Similarly, net transfers are determined as the sum of private transfers and official transfers . Private
transfers are exogenously determined, while official transfers are assumed equal to net grants
received by the government:
Net transfers = official transfers +private transfers
Finally, resources balance , net factor income and net transfers are added up to obtain current
account balance :
Current account = net factor income +resources balance +total net transfers
Capital and financial account
Capital and financial account consists
financial transfers:
Capital and financial account = private capital and financial transfers +public capital
and financial transfers
Public capital and financial transfers are
debt relief:
References
International Monetary Fund, Balance of Payments Manual, fifth edition, available online at
http://www.imf.org/external/np/sta/bop/BOPman.pdf
Barth, Ric hard, and William, Hemphill, Financial Programming and Policy, IMF Institute,
2000.
<relative consumer
prices>
ELASTICITY OF
DEMAND TO
RELATIVE PRICES
<supply>
INITIAL SECTOR
DEMAND SUPPLY
DISEQUILIBRIUM
INITIAL PC
<total population>
DEMAND
feasible share of pc
<real
pc national
<import>
demand
indicated
income>
demand supply
<export>
pc demand
ELASTICITY OF
<import prices>
balance
PRICE TO DEMAND
INITIAL SECTOR
INITIAL REAL PC
SUPPLY BALANCE <gdp deflator>
indicated producer
PRODUCER PRICES
INCOME
TIME TO
prices
ELASTICITY OF
ADJUST PRICE
<real production
INITIAL GDP
DEMAND TO
inflation rate
Producer
by sector>
INCOME
DEFLATOR
Prices
relative
gdp
producer price
INITIAL
REFERENCE TIME
deflator
change
HORIZON TO CALCULATE INFLATION RATE
pp index
domestic produced
cp index
INFLATION TREND
<RELATIVE
domestic share domestic marketed
EXCHANGE RATE
consumer prices
<real production
goods and services
WITH
DEVELOPING
by sector>
relative prices relative consumer
COUNTRIES>
import share
<export>
real import
prices
SHARE OF IMPORTS FROM
import prices
<Taxes On Goods And
INDUSTRIALIZED
before tariff
Services By Sector>
domestic
import prices
COUNTRIES TIME SERIES
effective
consumer prices
indirect tax rate
<INITIAL SECTOR
<Time>
<nominal production
<import> <relative exchange
PRODUCER PRICES>
by sector>
<Time>
rate with eu>
initial row
gdp deflator
<IMPORT TAXES
ROW INFLATION
price
GDP DEFLATOR
TIME SERIES>
RATE
<Time>
TIME SERIES
Row Prices
General Level Of
PRICES IN 2002
INFLATION RATE
Prices After 2002
general level of
row prices growth
AFTER 2002
prices growth
demand
supply
Consumer prices have a balancing effect: as the demand for one good increases, consumer
prices tend to increase. The higher the consumer price for one good, the lower the demand
will be for that good next time around.
Explanation
Major Assumptions
Producer prices depend on the supply/demand balance for goods and services;
Consumer prices depend on producer prices, import prices and indirect taxation;
Demand for goods and services is affected by relative consumer prices and real
income (Engel Curves);
Cross-product price elasticity values are all zero, which means that when, e.g.,
industry prices go up, it will affect the industry market, but it will not affect the price
of other products (agriculture and services).
Input Variables
Variable Name
Real production by sector
Import
Export
Taxes on goods and services by sector
Nominal production by sector
Total population
Relative exchange rate with EU
Import taxes by sector
Real pc national income
Relative exchange rate with developing
countries
Module of Origin
Aggregate Production and Income
International trade
International trade
Government revenue
Aggregate Production and Income
Population
International trade
Government revenue
Aggregate Production and Income
International trade
Output Variables
Variable Name
Sector producer price
Relative prices
Domestic consumer price
GDP deflator
Module of Destination
In the Same
In Other Economic
Sector
Sectors
International trade
Investment
International trade
Investment
Agriculture, government
expenditure, households
accounts
In Other Sectors
Income distribution,
primary education,
access to basic health
care, mortality, roads
International trade
Type of
Variable
Constant
Time series
Constant
Constant
Constant
Constant
Constant
Constant
Constant
IMF data
Equal to 0 for default
Constant
Constant
Constant
Time series
Constant
IMF data
Functional Explanation
The Relative Prices module can be divided into four functional parts: the calculation of
demand (top right of the sketch in Figure 43); the calculation of producer prices (middle of
the sketch); the calculation of relative prices (middle left of the sketch); and the calculation of
consumer prices (bottom left of the sketch). The following paragraphs describe these four
parts.
Demand for goods and services
Relative prices reflects
supply and demand situations for agricultural and industrial goods, and
for services. Demand for goods and services responds to changes in prices and income per
capita.
As per capita income increases, consumer demand does not increase for all types of goods and
services in the same way. In general, demand for agricultural goods tends to be less elastic
with income changes than the demand for industrial goods, which in turn is less elastic than
the demand for services (a relationship know as Engels Law 13). This means that as income
increases, consumers tend to spend a higher share of their income on services and a lower
share on agricultural and industrial goods. Consequently, as a country develops demand
patterns change and production tends to adapt to the new demand.
At the per capita income levels of the United States, for example, most additional income
becomes a demand for services rather than agricultural or industrial goods. In other words,
income elasticity of services demand at incomes per capita typical of the US is la rger than that
of industry demand or agriculture demand. Similarly, when people with US-income levels
suffer a reduction in income, they cut their demand for services more than for industrial and
agricultural goods.
13
This relationship is associated with Ernst Engel, a 19th century German statistician
In T21, indicated per capita demand (per capita demand before considering price levels, based
on income only) of agricultural and industrial goods and services is first computed, using
specific income elasticity for each type of goods and services:
Indicated per capita demand [sectors] = (INITIAL PC DEMAND [sectors])*(real pc
national income/INITIAL REAL PC INCOME)^ ELASTICITY OF DEMAND
TO INCOME [sectors]
Note that the subscript [sectors] is used to differentiate among different types of products
(agricultural and industrial goods, and services).
Subsequently, per capita demand is determined based on indicated per capita demand and the
consumer prices:
PC demand [sectors] = indicated pc demand [sectors]*relative consumer
price[sectors]^ELASTICITY OF DEMAND TO RELATIVE PRICES sectors]
Demand
Note that pc demand is also multiplied by the feasible share of pc demand , to ensure that the
sum of pc demand for the various goods and services does not exceed the overall per capita
income.
Producer prices
Producer prices are mainly a function of supply and demand balance for goods and services.
The paragraphs above describe how demand is determined in T21. Supply is equal to
production plus imports minus exports:
Supply = real production by sector [sectors] + import [sectors]/import price [sectors]
export [sectors]/Sector Producer Price [sectors]
The intensity of the change in prices that follow the change in the demand supply balance
varies from country to country and among the different types of goods and services. In
general, prices of agricultural goods tend to be more sensitive to supply and demand changes
than industrial goods, which in turn are more sensitive than services. This is because
agriculture production tends to be less flexible and adapts less rapidly to changes in demand
than industry or services production. The intensity of price change from changes in the supply
demand balance is represented by elasticity of price to demand supply balance . The values used
in T21 for this parameter are estimated for each country depending on its historic experience.
Indicated producer prices (producer prices before considering the time delays involved in price
changes) are thus calculated as:
Indicated producer prices = INITIAL SECTOR PRODUCER
PRICES[sectors]*demand supply balance[sectors]^ELASTICITY OF PRICE TO
DEMAND SUPPLY BALANCE[sectors]
Producer prices
prices:
Note from the equation above that producer prices are set to adjust to indicated producer prices
times the relative (normalized) GDP deflator. This captures the overall inflationary tendency
in the economy, which is not captured by the simple balance between supply and demand of
goods and services.
Relative prices
Relative prices
The PP index represents the sum of producer prices weighted by the shares of GDP derived
from the production of each different type of good and service. More precisely, the PP index is
calculated as:
PP index = SUM (Producer Prices [sectors!]*real production by sector [sectors!]) /
SUM(real production by sector [sectors!])
The following example illustrates how relative prices are calculated in general and in T21.
Producer prices indices for the United States are provided in Table 6 for 1980 and 1998. Note
that producer prices are normalized with base year 1980.
Table 6: Producer Price for USA, 1980 , 1998
1980
1998
Agriculture
1.00
1.86
Industry
1.00
1.52
Services
1.00
2.36
To calculate relative prices, we need to calculate the sum of producer prices weighted by GDP
shares per type of goods/services (the PP index). GDP shares for 1998 are: 2% for agricultural
goods, 26% for industrial goods, and 72% for services. Thus the weighted producer price
index is:
1.86*0.02 + 1.52*0.26 + 2.36*0.72 = 2.13
for 1998 are now obtained by dividing producer prices by the weighted producer
price index of 2.13. Relative prices for 1998 and 1980 (all 1.0 for the base) are shown in Table
7.
Table 7: Relative Sector Price Indices for USA, 1980, 1998
1980
1998
Agriculture
1.00
0.87
Industry
1.00
0.71
Services
1.00
1.11
From the above table, we can see that from 1980 to 1998, relative prices for agriculture and
industry fell 13% and 29%, while services rose 11%.
Relative prices
The weighted sum of relative prices of the three types of goods and services is always 1:
0.87*0.02 + 0.71*0.26 + 1.11*0.72 = 1
Consumer Prices
Consumer prices are
are defined as the import prices before tariff plus the effective tariff rate :
Finally, import prices before tariff are calculated using the overall level of prices in the rest of
the world (ROW) and the development of the exchange rates with developing countries and
industrialized countries:
Import prices before tariff [sectors] = Row Prices [sectors]*(relative exchange rate
with EU*SHARE OF IMPORTS FROM INDUSTRIALIZED COUNTRIES
TIME SERIES (Time) +RELATIVE EXCHANGE RATE WITH DEVELOPING
COUNTRIES*(1-SHARE OF IMPORTS FROM INDUSTRIALIZED
COUNTRIES TIME SERIES (Time)))
Using this definition for import prices and consumer prices , as the government increases
taxation on certain types of goods or services, demand for that good or service tends to
decrease. Consequently, producer prices decrease and investment in production capacity for
that good or service also decreases.
Other components
The GDP deflator (bottom left part of the sketch in Figure 43) and average price levels in the
rest of the world (bottom right part of the sketch) are also calculated in this module.
The GDP deflator is exogenously determined using an exogenous time series for the past and a
constant inflation rate for the future. More precisely, the GDP deflator is calculated as:
GDP deflator = IF THEN ELSE (Time<2002, GDP DEFLATOR TIME SERIES
(Time), General Level Of Prices After 2002)
General level of prices after 2002 is
level of prices growth :
The average level of prices in the rest of the world ( ROW prices) is also represented as a stock
variable, which accumulates the ROW prices growth flow:
ROW prices growth [sectors] = Row Prices [sectors]*ROW INFLATION RATE
References
Devarajan, Lewis and Robinson, From stylized to applied models: Building multisector
models for policy analysis, Working paper No. 616, UC Berkeley, September 1991.
Sterman, The energy transition and the economy, A system dynamics approach, Ph.D. Thesis,
MIT, December 1981.
<public
investment>
share of public
investment over total
EFFECTIVENESS OF
PUBLIC INVESTMENT
TIME SERIES
investment
agriculture
<crops
production>
<agriculture
production>
crops share of
agriculture production
investment crops
investment
<private
investment>
investment industry
real
investment
<gdp deflator>
<Time>
ELASTICITY OF
INVESTMENT TO
RELATIVE PRICES
<relative prices>
total investment
shares
investment services
INITIAL INVESTMENT
SHARE
effect of relative prices
on investment shares
investment shares
adjustment
INITIAL SECTOR
GDP RATIO
Investment
Shares
indicated
investment shares
relative sector
gdp ratio
INVESTMENT SHARE
ADJUSTMENT TIME
Explanation
Major Assumptions
The larger the contribution of one category of production (agriculture, industry,
services) to GDP, the larger the investment for that type of production;
The higher the relative producer prices for one type of goods or services, the larger the
investment for the production of that good or services;
The effectiveness of public investment is lower than the effectiveness of private
investment.
Input Variables
Variable Name
Public investment
Private investment
GDP deflator
Relative prices
Agriculture production
Module of Origin
Public Investment and Consumption
Households accounts
Relative prices
Relative prices
Agriculture
Output Variables
Variable Name
Investment crops
Investment industry
Investment services
Module of Destination
In the Same
In Other Economic
Sector
Sectors
Technology, agriculture
Technology, industry
Technology, services
Type of
Variable
Constant
Time series
Constant
Constant
Constant
In Other Sectors
Functional Explanation
The major function of this module is to calculate investment flow for each type of production
activity: agriculture, industry and services. To do so, first real investment is calculated (top left
of the sketch in Figure 44), and then real investment is shared based on the relative
contribution of each production activity to the GDP and the effect of relative prices.
Real investment
the gross capital formation, i.e. the effective inflow of capital for
the various production activities. Real investment differs from investment in two aspects: first it
is expressed in real terms (constant prices) while investment is expressed in nominal terms;
second, it considers that part of the public investment does not turn in capital formation. The
reason of this phenomenon is that the definition of capital investment used in government
accounts often includes items that cannot be considered physical capital formation (e.g.
professional training). Also, it is possible that inefficiencies exist in the public investment
process, and that the capital put in place by the government is not as productive as private
capital. Real investment is thus calculated by dividing investment by GDP deflator , and then
subtracting the fraction of public investment that is lost due to inefficiencies:
Real investment = (investment/GDP deflator)*(1-share of public investment over total
+share of public investment over total*EFFECTIVENESS OF PUBLIC
INVESTMENT TIME SERIES (Time))
Sector GDP ratio
The relative contribution of each type of production activity to GDP is an important element
for the allocation of investment among agriculture, industry and services. In general, for
example, the higher the share of GDP from industry, the higher the investment in industry.
This tends to be true because activities that account for bigger shares of GDP normally have a
bigger production capacity, which often coincides with a bigger capital stock. The substitution
of old capital requires a bigger investment flow, and if production is to expand 10% for all
activities, the activity that has a bigger capital stock will need to receive a bigger investment
flow.
The variable that represents the relative contribution of each type of production activity to the
GDP is called sector GDP ratio (sector is intended in this case as type of economic activity:
agriculture, industry and services) and is calculated in the Aggregate Production and Income
module.
Effect of relative prices on investment shares
Another important element for the allocation of investment among the various types of
production activities is relative prices. A high relative price for one type of good or service
indicates a strong demand for that good or service, and a higher than average profitability for
those who produce that good or service. High profitability has a significant influence on
investors who tend to direct their capital towards building capacity for the production of that
good or service. This mechanism is represented in T21 through the effect of relative prices on
investment shares . This variable is calculated by applying elasticity of investment to relative prices
to relative prices:
Effect of relative prices on investment shares [sectors] = relative prices[sectors]
^ELASTICITY OF INVESTMENT TO RELATIVE PRICES[sectors]
Investment shares
To calculate investment shares , first indicated investment shares is determined. This variable
represents the investment share that one would observe if there was no time lag between the
moment market conditions change and the moment investment decisions are actually
influenced. Indicated investment shares are calculated by combining together the effect of the
size of the various production activities with respect to the GDP and the effect of relative
prices:
Indicated investment shares [sectors] = INITIAL INVESTMENT SHARE [sectors]
*effect of relative prices on investment shares [sectors]*relative sector GDP ratio
[sectors]
Investment
Note that investment shares in the equations above are also divided by total investment shares,
or normalized, to make sure that the sum of investment shares does not exceed 1.
Investment agriculture, Investment industry and Investment services are calculated multiplying real
investment by the respective shares:
Investment agriculture = real investment*Investment Shares [AGRI]/total investment
shares
Investment industry = real investment* Investment Shares [IND]/total investment
shares
Investment services = real investment*Investment Shares [SERV]/total investment
shares
Investment industry and Investment services are
References
Devarajan, Lewis and Robinson, From stylized to applied models: Building multi-sector
models for policy analysis, Working paper No. 616, UC Berkeley, September 1991.
World Bank, World Development Indicators 2000 on CD-ROM.
Environmental Sectors
The environmental sectors of T21 are listed in Table 3. The next pages present the structure of
each module detail.
Table 8: The environmental sectors of T21 and corresponding modules
Environment
Land Sector:
30. La nd
Water Sector:
31. Water Demand
32. Water Supply
Energy Sector:
33. Energy Demand
34. Energy Supply
Minerals Sector:
35. Fossil Fuel Production
Emission Sector:
36. Fossil Fuel and GHG Emissions
Sustainability Sector:
37. Ecological Footprint
<total population>
desired urban land
INITIAL FOREST
LAND
Forest Land
PC SPACE
INITIAL
URBAN
LAND
desired change in
urban land
TIME TO CONVERT
FOREST LAND
forest loss for
agriculture
Urban Land
FALLOW LAND
CONVERSION TIME
DEGRADED LAND
RECOVERY TIME
forest to
fallow
fallow to forest
INITIAL
FALLOW
LAND
agriculture to
fallow
Desert Land
fallow to desert
forest cover
relative
forest cover
Fallow Land
maximum conversion
of fallow to agriculture
INITIAL
FOREST
COVER
TIME TO REGENERATE
FOREST LAND
fallow to urban
TIME TO ADJUST
URBAN LAND
INITIAL DESERT
LAND
INITIAL DESERTIFICATION
TIME FOR DEGRADED
AGRICULTURE LAND
relative
desired change in
<average yield>
average yield
DESIRED TIME TO ADJUST
agricultural land
AGRICULTURE LAND
Agricultural Land
INITIAL
<crop intensity
In Use
EFFECT OF AGRICULTURE
AVERAGE YIELD
index>
desired
INITIAL
INTENSITY ON LAND
arable land AGRICULTURAL
INITIAL EXHAUSTION
<average yield>
EXHAUSTION TABLE
LAND IN USE
TIME FOR AGRICULTURE
<total population>
LAND
DESIRED PER CAPITA
desired agriculture
AGRICULTURE PRODUCTION
production in tons
Explanation
Major Assumptions
Fallow land is brought into agriculture production by the increase in food demand;
Agricultural land degradation time depends on the intensity of farming;
Fallow land is lost to urban land through increases in population;
When fallow land is not sufficient to satisfy the demand for agriculture ad urban land,
forest land is cut;
Forest land is also cut in order to produce wood;
Fallow land is lost to desert land through the natural process of degradation;
Forest cover is also a determinant of the speed of desertification process.
Input Variables
Variable Name
Total Population
Crop Intensity Index
Module of Origin
Pop ulation
Agriculture
Average Yield
Forest cut for wood production
Agriculture
Animal husbandry-fishery-forestry
Output Variables
Variable Name
Agricultural Land in Use
Module of Destination
Same Sector
Other Environmental Sectors
Type of Variable
Constant
Constant
Other Sectors
Agriculture,
Water Demand,
Roads
Table Function
Table Function
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Functional Explanation
The Land module represents land use for different purpose: urban, forest, fallow, agricultural
and desert land. In addition, the stock of fallow l and is also disaggregated into cultivable fallow
land and degraded fallow land. In the following paragraphs, the flows that move land from
one use to another are described.
Fallow to Agriculture
The flow of fallow land that is transformed into agriculture land is calculated as the minimum
between desired change in agriculture land and the maximum possible conversion from fallow to
agric ulture. The maximum flow of fallow land to feasible agriculture is calculated as the stock
of fallow land [cultivable], divided by a conversion time:
maximum conversion of fallow to agriculture= Fallow
Land[CULTIVABLE]/FALLOW LAND CONVERSION TIME
The desired arable land , which is the amount of agriculture land wanted by agriculture
producers to meet the demand for agriculture goods, is calculated as desired agriculture
production divided by the average yield and crop intensity index , plus the additional demand for
arable land deriving f rom changes in the cotton price:
desired arable land= (desired agriculture production in tons/average yield)/crop
intensity index+ADDITIONAL DEMAND OF ARABLE LAND DUE TO
COTTON PRICE CHANGES(Time)
Agriculture to Fallow
The flow of agricultural land that becomes fallow land is calculated as the stock of agricultural
land divided by the time it takes to exhaust the land's nutrients, times the effect that the
average yield (representing the intensity of the culture) has on the exhaustion time:
agriculture to fallow = (Agricultural Land In Use/INITIAL EXHAUSTION TIME FOR
AGRICULTURE LAND)*EFFECT OF AGRICULTURE INTENSITY ON
LAND EXHAUSTION TABLE(relative average yield)
Figure 46: Assumed relationship between the intensity of agriculture activity (Relative
Average Yield) and the agricultural land degradation process.
The table function shown in Figure 46 represents the relationship between the intensity of
agricultural activities (yield) and the intensity of the agricultural land degradation process. It
is assumed that the higher the yield, the more intense the land degradation.
Forest to Fallow
The flow of forestland converted into fallow land is calculated as the minimum between the
sum of forest loss for agriculture, urbanization and forestry, and existing forestland divided by
the time necessary to convert forestland into useful land.
Forest to fallow = MIN(forest loss for agriculture +forest loss for urbanization +forest
cut for wood production, Forest Land/TIME TO CONVERT FOREST LAND)
Forest loss for urbanization , the
Note that if the difference between the desired change in urban land and total fallow land
difference is negative, this variable is equal to 0.
represents the change in urban land that is necessary in order to
satisfy the needs of the population. It is calculated as the difference between desired (total
population times per capita urban land needed for living, working, and entertainment) and
Desired change in urban land
actual levels of urban land, if this difference is bigger than 0. If not, this variable is equal to
0.
Forest loss for agriculture,
Fallow to Forest
The flow of fallow land that is re-transformed into forestland is calculated as the stock of
fallow land , divided by the average time necessary to regenerate the forest:
fallow to forest[degradation level]= Fallow Land[degradation level]/TIME TO
REGENERATE FOREST LAND
Fallow to Urban
The flow of fallow land that is transformed into urban land is calculated using two equations,
one for [cultivable] land and one for [degraded] land. It is assumed that for urbanization
purposes, degraded land is preferred over cultivable land, and that cultivable land will be used
only in case the existing degraded land is not enough.
The flow of degraded land is therefore calculated as the minimum between desired change in
urban land and the existing fallow land [degraded] divided by the time required to adjust urban
land:
fallow to urban[DEGRADED]= MAX(MIN( desired change in urban land , (Fallow
Land[DEGRADED]))/TIME TO ADJUST URBAN LAND,0)
The flow of cultivable land is calculated as the minimum between desired change in urban land
and fallow land [cultivable] divided by the adjustment time minus the flow of fallow to urban land
[degraded]. The MAX formulation is used in both equations to ensure that this flow is always
positive:
fallow to urban[CULTIVABLE]= MAX(MIN( desired change in urban land , (Fallow
Land[CULTIVABLE]))/TIME TO ADJUST URBAN LAND-fallow to
urban[DEGRADED],0)
Fallow to Desert
The flow of fallow land that is transformed into desert land is calculated as the stock of
[degraded] fallow land divided by the time it takes for degraded land to become desert land.
The effect of forest cover on the speed of the desertification process is also taken into
account:
fallow to desert= (Fallow Land[DEGRADED]/INITIAL DESERTIFICATION TIME
FOR DEGRADED AGRICULTURE LAND)*EFFECT OF FOREST COVER ON
DESERTIFICATION TABLE (relative forest cover)
Figure 47: Assumed relationship between forest cover and the intensity of the
desertification process.
The table function in Figure 47 represents the assumed relationship between forest cover and
the intensity of the desertification process. It is assumed that the lower the forest cover, the
faster the desertification.
References
World Bank, World Data 1995, on CD-ROM.
World Resources Institute, World Resources 1994 -95. New York: Oxford University Press,
1994.
FAO, FAOSTAT 1997 , on CD-ROM.
<Perceived Relative Pc
Disposable Income>
domestic and
municipal water
demand
PC WATER
DEMAND
<relative resources
technology>
<Agricultural Land
In Use>
<relative resources
technology>
agriculture water
demand
industry water
demand
<total population>
Explanation
Major Assumptions
Domestic and municipal water demands are influenced by total population, income,
and technology;
Industry water demand is determined by production (proportionally) and technology;
Agriculture water demand is determined by the size of the harvested area and
technology.
Input Variables
Variable Name
Agricultural Land in Use
Industry Production
Perceived relative PC Disposable Income
Relative Resources Technology
Total population
Module of Origin
Land
Industry
Households Accounts
Technology
Population
Output Variables
Variable Name
Total Water Demand
Constants and Table functions
Variable Name
Module of Destination
Same Sector
Other Environmental
Sectors
Water Supply
Type of
Other Sectors
Variable
Effect of Education on Water Use Efficiency
table
Table function
Table function
PC Water Demand
Relative Perceived PC Disposable Income in
2000
Water Demand per Ha of Arable Land
Water Demand per Unit Produced in Industry
Constant
Constant
Constant
Constant
Aquastat (FAO)
Data on income and water demand
from Aquastat (FAO)
Data on water from Aquastat (FAO)
IMF (International Monetary Fund)
data on Income
Aquastat (FAO)
Aquastat (FAO)
Functional explanation
Total Water Demand
Total water demand is
The amount of water demanded each year for industry is calculated as industry production times
water demand per unit produced, divided by relative resources technology :
industry water demand= industry production*WATER DEMAND PER UNIT
PRODUCED IN INDUSTRY/relative resources technology[IND]
Agriculture water demand is
The table in Figure 49 represents the relationship between income and per capita water
demand. It is assumed that the higher the per capita income, the higher is per capita water
demand.
Figure 49: Assumed relationship between income and per capita water demand.
Effect of Water Stress on Productivity
The effect of water stress on agriculture productivity is calculated based on the table function
EFFECT OF WATER STRESS ON AGRICULTURE PRODUCTIVITY TABLE (using as input water
stress index). Water stress index is the proportion of demanded water that is satisfied.
Figure 50: Assumed relationship between water stress and agriculture productivity.
The table function in Figure 50 represents the non-linear relationship between water stress and
agriculture productivity. It assumes that there is no effect until the stress index is lower than
0.5. Then it assumes that if the stress index goes to 1 there is a reduction in agriculture
productivity of about 5%, because part of irrigated agriculture would be cut. The maximum
effect is for a stress index of 3, assuming that agriculture is cut by 12.5%.
References
Hoekstra, A., AQUA, a Framework for Integrated Water Policy Analysis, National Institute of
Public Health and Environment (RIVM), July 1995.
Miller, G. T., Living in the Environment, seventh edition, Wadsworth, 1992
CROSS BORDER
INFLOW
<total water
demand>
total water use
fraction of water
available
<Time>
precipitation
KILOGRAMS PER
CUBIC METER OF
WATER
<TOTAL LAND
AREA>
water resources
internally produced
FRACTION OF RAIN
EVAPORATING
IMMEDIATELY
total renewable
water resources
<total water
demand>
<total population>
Explanation
Major Assumptions
Input Variables
Variable Name
Total Land Area
Total water demand
Total population
Module of Origin
Land
Water demand
Population
Output Variables
Variable Name
Module of Destination
Same Sector
Other
Environmental
Sectors
Other Sectors
Agriculture
Mortality
Type of
Variable
Time Series
Constant
FAO (Aquastat)
FAO (Aquastat)
Constant
FAO (Aquastat)
Constant
Table
func tion
Constant
FAO (Aquastat)
Data for agriculture production and
water availability (FAO)
FAO (Aquastat)
Functional Explanation
As illustrated in Error! Reference source not found., available water comes from two
sources, cross border inflow and precipitation .
Precipitation is
The table function effect of water stress on agriculture productivity table is estimated based on
local experts knowledge and on data for agriculture production and water availability, and it
is likely to differ substantially from country to country.
References
Hoekst ra, A., AQUA, a Framework for Integrated Water Policy Analysis, National Institute of
Public Health and Environment (RIVM), July 1995.
Miller, G. T., Living in the Environment, seventh edition, Wadsworth, 1992
<relative resources
technology>
<sector gdp ratio>
initial energy
demand per unit
of production
INITIAL INTENSITY
OF NON
ELECTRICITY
BTU PER BARREL OF
OIL TIME SERIES
energy intensity
of gdp
<Time>
RELATIVE WEIGHTED
AVERAGE ENERGY PRICE
NON-ELECTRICTY TIME
SERIES
RELATIVE COAL
PRICE TIME SERIES
RELATIVE NATURAL
GAS PRICE TIME
RELATIVE OIL
SERIES
PRICE TIME SERIES
TIME TO ADAPT
DEMAND TO PRICE
CHANGES
<Time>
Effect Of Price On
Energy Demand
relative energy
prices
<real gdp at market
relative average
prices>
energy price
energy demand
per unit of
production
ELASTICITY OF FOSSIL
FUELS PENETRATION TO
PRICE
<Time>
relative fossil
fuels prices
<RELATIVE OIL
PRICE TIME
SERIES>
indicated energy
demand by source
INITIAL FOSSIL
FUELS PENETRATION
RATE
ELASTICITY OF
ENERGY DEMAND
TO ENERGY PRICE
share of energy
demand by source
total energy
demand
<Time>
<Time>
electricity generation
from fossil fuel in kwh
energy demand
ELECTRICITY LOSS
FACTOR TIME SERIES
total electricity
demand in kwh
<renewable resources
electricity production>
<nuclear
production>
Explanation
Major Assumptions
Demand for energy depends on GDP, energy prices, and technology;
Electricity demand that is not produced by non-fossil fuels energy sources, is produced
through fossil fuel combustion;
Demand for fossil fuels for the generation of electricity depends on fossil fuel-specific
technology and prices.
Input Variables
Variable Name
Nuclear production
Real GDP at market prices
Module of Origin
Energy Supply
Aggregate Production and Income
Technology
Energy Supply
Aggregate Production and Income
Energy Supply
Output Variables
Variable Name
Module of Destination
Same Sector
Other
Environmental
Sectors
Energy Supply
Energy Supply
Energy Supply
Other Sectors
Energy Supply
Type of
Variable
Time Series
Time Series
Time Series
Constant
Constant
Time Series
Time Series
Time Series
Time Series
Time Series
Constant
Time Series
Constant
Constant
Constant
Functional Explanation
The Energy Demand module is divided into electricity and non-electricity energy demand,
which includes fossil fuels (oil, natural gas and coal) and nuclear energy production. Fossil
fuel demand for electricity production is also calc ulated by the model.
Energy Demand
The energy demand per unit of production (energy intensity of GDP) is calculated as the ratio of
the initial energy intensity of GDP to the relative energy technology levels averaged for the
three production sectors:
energy demand per unit of production[energy demand type] = init ial energy demand
per unit of production[energy demand type]/SUM(relative resources
technology[sectors!]*sector gdp ratio[sectors!])
The indicated energy demand by source (both electricity and non-electric ity demand) is
calculated as the product between GDP and GDP energy intensity, and it includes the effect of
technology on demand:
indicated energy demand by source[energy demand type] = energy demand per unit of
production[energy demand type]*real gdp at market prices
The sum of both indicated electricity and non-electricity energy demand gives the total energy
demand .
The effect of price on demand is assumed to impact the choice between electricity and nonelectricity demand, but not the total amount of energy demanded. In order to do so, the model
calculates the share of energy demand by source, which includes the effect of price. The share of
energy demand by source is calculated as the indicated energy demand (electric and nonelectric) divided by total energy demand, multiplied by the effect of price on energy demand:
share of energy demand by source[energy demand type]= indicated energy demand by
source[energy demand type]*Effect Of Price On Energy Demand[energy demand
type]/SUM(indicated energy demand by source[energy demand type!]*Effect Of
Price On Energy Demand[energy demand type!])
The effect of price on energy demand is obtained by using two equations: one for electricity and
one for non electricity. The effect of energy price on electricity demand is calculated using a
first order delay of the ratio of relative electricity to relative oil price to the power of elasticity
of energy demand to energy price. The first order delay represents the time lag between changes
in prices and changes in energy consumers' behavior. The effect of energy price on non
electric ity demand is calculated in a similar way:
Effect Of Price On Energy Demand[NON ELECTRICITY]= SMOOTH
N((RELATIVE OIL PRICE TIME SERIES(Time)/RELATIVE ELECTRICITY
PRICE TIME SERIES(Time))^ELASTICITY OF ENERGY DEMAND TO
ENERGY PRICE[NON ELECTRICITY], TIME TO ADAPT DEMAND TO
PRICE CHANGES,1,1)
The energy demand, which includes both electricity and non electricity sources, for both direct
consumption and electricity production, is calculated as the total energy demand multiplied by
the share of energy demand by energy source:
energy demand[energy demand type]= share of energy demand by source[energy
demand type]*total energy demand
Fossil fuel demand for electricity production
The demand for fossil fuels for electricity generation in BTU is obtained by multiplying the
required fossil fuels production for electricity need (in KWh) by the fossil fuel efficiency in
electricity generation and by the conversion factor BTU PER KWH OF ELECTRICITY TIME
SERIES :
demand for fossil fuels for electricity generation in btu[fossil fuel] = electricity
generation by type of fossil fuel in kwh[fossil fuel]/(FOSSIL FUELS
EFFICIENCY IN ELECTRICITY GENERATION TIME SERIES[fossil
fuel](Time) /BTU PER KWH OF ELECTRICITY TIME SERIES(Time))
The electricity generation by type o f fossil fuel in KWh is obtained by multiplying the fossil fuel
share of electricity production by electricity generation from fossil fuel in kwh:
electricity generation by type of fossil fuel in kwh[fossil fuel]= normalized fossil fuels
penetration rate for electricity production[fossil fuel]*(electricity generation from
fossil fuel in kwh)
The electricity generation from fossil fuel in kwh is calculated as total electricity demand minus
renewable and nuclear electricity production (which is equal to the electricity generation by type
of fossil fuel in kwh [fossil fuel]):
electricity generation from fossil fuel in kwh= MAX(0,total electricity demand in kwh((renewable resources electricity production+nuclear production)/BTU PER KWH
OF ELECTRICITY TIME SERIES(Time)))
The normalized share of electricity produced from fossil fuels (normalized fossil fuels
penetration rate for electricity production ) is calculated by the ratio between indicated fossil fuel
penetration rate and the total adjusted penetration rate:
normalized fossil fuels penetration rate for electricity production[fossil fuel]= fossil
fuels indicated penetration rate for electricity production[fossil fuel]/SUM(fossil
fuels indicated penetration rate for electricity production[fossil fuel!])
The fossil fuels indicated penetration rate for electricity production, which represents the indicated
share of electricity production from fossil fuel, accounts for both the effect of energy prices
and technology on fossil fuel electricity penetration rate. It is calculated per each fossil fuel
separately. The shares are calculated by multiplying the initial penetration rate by the inverse
of the relative fossil fuel prices, all to the power of the elasticity of fossil fuel penetration to price,
and then multiplied by the efficiency of electrical generation from fossil fuels (effect of
technology):
fossil fuels indicated penetration rate for electricity production[fossil fuel]= (INITIAL
FOSSIL FUELS PENETRATION RATE[fossil fuel]*(1/Relative Fossil Fuels
Prices[fossil fuel])^ELASTICITY OF FOSSIL FUELS PENETRATION TO
PRICE[fossil fuel])*FOSSIL FUELS EFFICIENCY IN ELECTRICITY
GENERATION TIME SERIES [fossil fuel](Time)
References
Sterman, John D., The energy transition and the economy: A system dynamics approach,
Ph.D . Thesis, MIT, 1981;
OPEC, World Energy Model (OWEM), Oil Outlook to 2025, OPEC Review, September
2004;
Naill, R. F., Managing the Energy Transition, Ballinger, Cambridge, MA, 1977;
International Energy Agency (IEA), World Energy Outlook 2004, Annex C, WEM (World
Energy Model), 2004;
Energy Information Administration (EIA), Department of Energy, Integrating Module of the
National Energy Modeling System: Model Documentation 2004, 2004;
Energy Information Administration (EIA), Annual Energy Review 2005, 2005;
Davidsen, P. I., J. D. Sterman, G. P. Richardson, A Petroleum Life Cycle Model for the
United States with Endogenous Technology, Exploration, Recovery, and Demand,
System Dynamics Review 6(1), 1990;
Backus, G., et al. FOSSIL 79: Documentation, Resource Policy Center, Dartmouth College,
Hanover NH, 1979;
AES Corporation, An Overview Of The IDEAS MODEL: A Dynamic Long-Term Policy
Simulation Model Of U.S. Energy Supply And Demand, Prepared For The U.S.
Department Of Energy Office Of Policy, Planning, And Evaluation, Arlington, VA,
October 1993.
fossil fuel
consumption
<energy demand>
NUCLEAR
PRODUCTION
CAPACITY
nuclear production
<relative energy
prices>
INITIAL REAL
ENERGY PRICE
<energy demand>
total energy
production
REFERENCE
ELASTICITY OF
GDP TO ENERGY
PRICE
fossil fuel
dependency ratio
renewable resources
electricity production
<energy demand>
total energy
expenditure
over gdp
REFERENCE
ENERGY
EXPENDITURE
OVER GDP
TIME TO PERCEIVE
CHANGES IN
ENERGY PRICE
Elasticity Of Gdp To
Energy Prices
<relative average
energy price>
Perceived Relative
Average Energy Price
Explanation
Major Assumptions
Nuclear production capacity, hydropower and wind power generation depend on
investments and on land characteristics, therefore are represented exogenous time
series inputs;
Energy demand that is not satisfied by domestic sources is imported;
Energy prices impact economic productivity;
The intensity of the impa ct of energy prices on economic productivity is defined by
the level of energy dependency and overall energy expenditure.
Input Variables
Variable Name
BTU per KWH of Electricity Time series
Demand for Fossil Fuels for Electricity
Generation in BTU
Energy Demand
Module of Origin
Energy Demand
Energy Demand
Energy Demand
Output Variables
Variable Name
Module of Destination
In the Same Sector
In Other
Environmental
Sectors
Services, Industry,
Animal Husbandry
- Fishery
Forestry,
agriculture
In Other Sectors
Type of
Variable
Constant
Time series
Constant
Constant
Constant
Constant
Constant
Functional Explanation
Total Energy Production
The total annual energy production is calculated as the sum of fossil fuels, nuclear and
renewable energy production:
total energy production= nuclear production+renewable resources electricity
production+total fossil fuels production in btu
The total annual required energy production is calculated as the sum of demands for fossil
fuels (both for electricity and non-electricity production use), nuclear energy and renewable
energy, (which is assumed equal to production). Total required energy production differs by total
energy demand because it accounts for electricity transmission loss and fossil fuel efficiency in
electricity generation.
Effect of Energy Price on Productivity
In order to calculate the effect of energy price on productivity, T21 uses the following
formulation:
effect of energy price on productivity= Perceived Relative Average Energy
Price^Elasticity Of Gdp To Energy Prices
The perceived relative average energy price is calculated as a first order delay of the relative
average price of energy. The delay takes into account the time between the change in energy
price on the international market and the time when that change is reflected on the domestic
market.
The elasticity of GDP to energy prices is calculated as the first order delay of the indicated
elasticity of GDP to energy prices, which is equal to:
indicated elasticity of gdp to energy prices= REFERENCE ELASTICITY OF GDP TO
ENERGY PRICE*energy dependency ratio*total energy expenditure over
gdp/REFERENCE ENERGY EXPENDITURE OVER GDP
This variable is calculated by multiplying reference elasticity of GDP to energy price by energy
dependency ratio and then by total energy expenditure over GDP divided by the reference energy
expenditure over GDP. Note that both the effect of energy prices on the national accounts and
the effect of energy dependency on exposition to price volatility are included in the
formulation.
References
Sterman, John D., The energy transition and the economy: A system dynamics approach,
Ph.D. Thesis, MIT, 1981;
OPEC, World Energy Model (OWEM), Oil Outlook to 2025, OPEC Review, September
2004;
Naill, R. F., Managing the Energy Transition, Ballinger, Cambridge, MA, 1977;
International Energy Agency (IEA), World Energy Outlook 2004, Annex C, WEM (World
Energy Model) , 2004;
Energy Information Administration (EIA), Department of Energy, Integrating Module of the
National Energy Modeling System: Model Documentation 2004, 2004;
Energy Information Administration (EIA), Annual Energy Review 2005, 2005;
Davidsen, P. I., J. D. Sterman, G. P. Richardson, A Petroleum Life Cycle Model for the
United States with Endogenous Technology, Exploration, Recovery, and Demand,
System Dynamics Review 6(1), 1990;
Backus, G., et al. FOSSIL 79: Documentation, Resource Policy Center, Dartmouth College,
Hanover NH, 1979;
AES Corporation, An Overview Of The IDEAS MODEL: A Dynamic Long-Term Policy
Simulation Model Of U.S. Energy Supply And Demand, Prepared For The U.S.
Department Of Energy Office Of Policy, Planning, And Evaluation, Arlington, VA,
October 1993.
btu to mb
Identified Oil
Reserve
oil production
oil discovery
INITIAL
DISCOVERY
FRACTION OIL
INITIAL
IDENTIFIED OIL
RESERVE
<relative resources
INITIAL
INITIAL IDENTIFIED
technology>
UNDISCOVERED
COAL RESERVE
COAL RESOURCES
Undiscovered Coal
Resources
oil production in
btu
btu to mst
Identified Coal
Reserve
INITIAL
PRODUCTION
FRACTION COAL
INITIAL DISCOVERY
FRACTION COAL
INITIAL
UNDISCOVERED GAS <relative resources
technology>
RESOURCES
Undiscovered Gas
Resources
Identified Gas
Reserve
gas discovery
INITIAL DISCOVERY
FRACTION GAS
coal production
coal discovery
INITIAL
PRODUCTION
FRACTION GAS
coal production
in btu
INITIAL
IDENTIFIED GAS
RESERVE
gas production in
btu
gas production
btu to bcf
Explanation
Major Assumptions
Fossil fuel resource is finite;
Exploration and production, separately, determine the availability of recoverable
resources for production;
Technology affects the effectiveness of exploration and recovery activities 14.
Input Variables
Variable Name
Relative Resources Technology
14
Module of Origin
Technology
Prices do not affect domestic fossil fuels production for the following reason: production of a net importer
country is usually not sensitive to international energy prices because it is primarily limited by reserves
availability. Therefore, even if international prices would theoretically justify investments in production their
productivity would be too low. This is particularly true when energy prices are highly volatile.
Output Variables
Variable Name
Total Fossil Fuels Production
in BTU
Module of Destination
In the Same Sector
In Other Environmental In Other Sectors
Sectors
Energy Supply
Type of
Variable
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Functional Explanation
The recovery (production) of fossil fuels (oil, gas, and coal) is modeled us ing the structure
illustrated in Figure 55. Since the production of fossil fuels is identical for oil, gas, and coal,
only oil production is explained in the following paragraphs.
INITIAL
INITIAL
PRODUCTION
IDENTIFIED OIL
FRACTION OIL
RESERVE
btu to mb
Identified Oil
Reserve
oil production
oil production in
btu
<relative resources
technology>
INITIAL
UNDISCOVERED OIL
RESOURCES
Undiscovered Oil
Resources
oil discovery
INITIAL
DISCOVERY
FRACTION OIL
Figure 55: Structure of the Energy Supply Model for Fossil Fuels
IDENTIFIED
UNDISCOVERED
HYPOTHETICAL
SPECULATIVE
Paramarginal
RESERVES
RESOURCES
Submarginal
SUBECONOMIC
Indicated
ECONOMIC
Inferred
Measured
Exploration leads to discoveries, which gradually reduce the stock of undiscovered resources.
The identified reserve is increased through discovery and decreased by production. The
discovery rate is equal to u ndiscovered resource multiplied by initial discovery fraction divided by
the relative technology level:
oil discovery= MIN(1,INITIAL DISCOVERY FRACTION OIL*Undiscovered Oil
Resources*relative resources technology[IND])
Similarly, the production rate is equal to identified reserve multiplied by initial production fraction
divided by relative resources technology:
oil production=MIN(1,Identified Oil Reserve*INITIAL PRODUCTION FRACTION
OIL*relative resources technology[IND])
The quantity of both undiscovered resource and identified reserve is related to technology. As a
consequence of improvements in technology, resources and reserves may become recoverable
or economically extractable, thus increasing the quantity available.
Total Fossil Fuels Production
Total fossil fuels production in BTU
References
Sterman, John D., The energy transition and the economy: A system dynamics approach,
Ph.D. Thesis, MIT, 1981;
OPEC, World Energy Model (OWEM), Oil Outlook to 2025, OPEC Review, September
2004;
Naill, R. F., Managing the Energy Transition, Ballinger, Cambridge, MA, 1977;
International Energy Agency (IEA), World Energy Outlook 2004, Annex C, WEM (World
Energy Model), 2004;
Energy Information Administration (EIA), Department of Energy, Integrating Module of the
National Energy Modeling System: Model Documentation 2004, 2004;
Energy Information Administration (EIA), Annual Energy Review 2005, 2005;
Davidsen, P. I., J. D. Sterman, G. P. Richardson, A Petroleum Life Cycle Model for the
United States with Endogenous Technology, Exploration, Recovery, and Demand,
System Dynamics Review 6(1), 1990;
Backus, G., et al. FOSSIL 79: Documentation, Resource Policy Center, Dartmouth College,
Hanover NH, 1979;
AES Corporation, An Overview Of The IDEAS MODEL: A Dynamic Long-Term Policy
Simulation Model Of U.S. Energy Supply And Demand, Prepared For The U.S.
Department Of Energy O ffice Of Policy, Planning, And Evaluation, Arlington, VA,
October 1993.
WEIGHT OF N2O
WITH RESPECT TO N
<KILOGRAMS IN
ONE TON>
EQUIVALENT FOR
GREENHOUSE EFFECT
OF N2O IN CO2
TJ PER BTU
fossil fuel
consumption in tj
fossil fuel c
emission
<energy demand>
NON ENERGY FUEL
CONSUMPTION
total consumption of
energy from fossil fuel
KG OF FOSSIL
FUEL CH4
EMISSION PER TJ
EQUIVALENT FOR
GREENHOUSE EFFECT
OF CH4 IN CO2
KG OF FOSSIL FUEL
SOX EMISSION PER
BTU
MOLECULAR
WEIGHT OF C
fossil fuel
consumption
<TOTAL LAND
AREA>
KILOGRAMS
IN ONE TON
co2 emission in kg
<real gdp at
market prices>
co2 intensity
of gdp
per capita co2
emissions
co2 emissions
per hectare
Figure 57: Sketch of the Fossil Fuel and GHG Emissions module
Explanation
Major Assumptions
CO2, N 2O, and CH4 are the chief determinants of greenhouse gas generation;
CO2 emissions per hectare are a good proxy for PM10, the main determinants of
mortality related to fossil fuels emissions;
Conversion factors used to calculate emissions out of fossil fuel consumption are
constant.
Input Variables
Variable Name
Fossil Fuel Consumption
Real GDP at Market Prices
Total Land Area
Module of Origin
Energy Supply
Aggregate Production and Income
Land
Output Variables
Variable Name
Module of Destination
In the Same Sector
In Other
Environmental
Sectors
In Other Sectors
Ecological
Footprint
Mortality
Type of
Variable
Constant
Constant
ANPA data
Constant
Constant
Constant
Constant
Constant
Constant
Fossil 2 model
IPCC data
Constant
Constant
Constant
Constant
Constant
IPCC data
IPCC data
IPCC data
IPCC (Intergovernmental Panel on
Climate Change) data
Data on energy consumption (EIA)
Data on energy consumption (EIA)
IPCC data
IPCC data
Functional Explanation
In this module , fossil fuel emissions are calculated by converting consumption of oil, coal and
gas into CO2 , N2O, SO X and CH4 emissions equivalent.
Fossil Fuel SOX Emission
The emission of SOX from burning fossil fuels is calculated as the total consumption of
energy from fossil fuels times the SOX emissions per BTU of fossil fuels burned:
fossil fuel sox emission=SUM(total consumption of energy from fossil fuel[fossil
fuel!]*KG OF FOSSIL FUEL SOX EMISSION PER BTU[fossil fuel!])
Fossil fuel CH 4 emission in CO2 equivalent
Fossil fuel CH4 emissions are
Fossil fuel CH4 emissions in CO2 equivalent represent the emissions of CH4 expressed as the
CO2 emissions that would produce the same greenhouse effect. CH4 emissions in CO2 equivalent
are calculated as fossil fuel CH4 emissions multiplied by the constant to convert the CH 4
emissions into the greenhouse-effect equivalent CO2 emissions:
ch4 emission in co2 equivalent= fossil fuel ch4 emission * EQUIVALENT FOR
GREENHOUSE EFFECT OF CH4 IN CO2
As for the CH4 emissions in CO2 equivalent, N2O emissions in CO2 equivalent are obtained as
the fossil fuel N 2O emission multiplied by a conversion factor that represents the greenhouseeffect equivalent of one unit of N2O in CO2.
Fossil fuel CO2 emission
The total emission of CO 2 from the burning of fossil fuels is calculated as fossil fuel C emission
multiplied by the molecular weight of CO2 , divided by the molecular weight of C. It is assumed
that all C becomes CO 2, even though a small percentage becomes CO and CH4 :
fossil fuel co2 emission= fossil fuel c emission*MOLECULAR WEIGHT OF
CO2/MOLECULAR WEIGHT OF C
Fossil fuel greenhouse gases emissions in CO2 equivalent
References
AES Corporation, An Overview Of The IDEAS MODEL: A Dynamic Long-Term Policy
Simulation Model Of U.S. Energy Supply And Demand, Prepared For The U.S.
Department Of Energy Office Of Policy, Planning, And Evaluation, Arlington, VA,
October 1993.
Intergovernmental Panel on Climate Change (IPCC), IPCC Guidelines for National
Greenhouse Gas Inventories: Greenhouse Gas Inventory Workbook. Greenhouse Gas
Emissions.
REFERENCE PER
CAPITA CO2
EMISSIONS LEVEL
BUILT UP LAND
PC FOOTPRINT
FUELWOOD PC
FOOTPRINT
relative co2
emission
FOREST PC
FOOTPRINT
CROPLAND PC
FOOTPRINT
FISHING GROUND
PC FOOTPRINT
per capita
ecological footprint
PC BIOCAPACITY
AVAILABLE WORLDWIDE
TIME SERIES
national footprint
<total population>
<Time>
AVAILABLE
BIOCAPACITY
national footprint
relative to biocapacity
Explanation
Major Assumptions
Input Variables
Variable Name
Per capita co2 emissions
Total Population
Module of Origin
Fossil Fuel and GHG Emissions
Population
Output Variables
Module of Destination
Variable Name
In the Same
Sector
In Other
Environmental Sectors
In Other Sectors
None
Constants and Table functions
Variable Name
Available Biocapacity
Built Up Land PC Footprint
Cropla nd PC Footprint
Fishing Ground PC Footprint
Forest PC Footprint
Fuelwood PC Footprint
Grazin g Land PC Footprint
PC Biocapacity Available Worldwide time
series
PC Footprint from Reference CO2 from
Fossil Fuel
Reference per Capita CO2 Emission Level
Type of
Variable
Constant
Constant
Constant
Constant
Constant
Constant
Constant
Time series
Constant
Constant
Functional Explanation
Per Capita Footprint from CO 2 from Fossil Fuels
The per capita ecological footprint in hectares, due to emissions from the burning of fossil
fuels (per capita footprint from co2 from fossil fuels), is calculated as the reference per capita
footprint from fossil fuels multiplied by the relative level of CO2 emissions:
per capita footprint from co2 from fossil fuels= PC FOOTPRINT FROM
REFERENCE CO2 FROM FOSSIL FUEL*relative co2 emission
The CO2 footprint is the only component of the per capita ecological footprint that is assumed to
change substantially in the time horizon of the simulation.
Per Capita Ecological Footprint
The per capita ecological footprint represents the productive land and water one person requires
to produce the sustainable resources he or she consumes and to absorb his/her sustainable
wastes, all using prevailing technology. It is calculated as the sum of per capita footprint
from various sources, assuming that only the CO2 from fossil fuels per capita footprint varies
over time:
per capita ecological footprint= BUILT UP LAND PC FOOTPRINT+per capita
footprint from co2 from fossil fuels+CROPLAND PC FOOTPRINT+FISHING
GROUND PC FOOTPRINT+FOREST PC FOOTPRINT+FUELWOOD PC
FOOTPRINT+GRAZING LAND PC FOOTPRINT
National Footprint
The national footprint represents the amount of productive land and water the country requir es
to produce the sustainable resources it consumes and to absorb the waste it generates, using
currently available technology. It is calculated as the total population times the per capita
ecological footprint:
national footprint= total population*per capita ecological footprint
The national footprint relative to biocapacity is an indicator of the long-term sustainability, given
nature's biologically productive capacity. It is calculated as the national footprint divided by
available biocapacity.
National Footprint Relative to World Sustainable Footprint
This variable in an indicator of the proportion of the world's per capita biocapacity, used per
person in the country. It is calculated as the ratio of the per capita ecological footprint and PC
BIOCAPACITY AVAILABLE WORLDWIDE TIME SERIES (using as input Time):
national footprint relative to world sustainable footprint= per capita ecological
footprint/PC BIOCAPACITY AVAILABLE WORLDWIDE TIME
SERIES (Time)
References
World Wildlife Fund, Living Planet Report, Washington: World Wildlife Fund, 2002
Mathis Wackernagel, Chad Monfreda, Dan Moran, Paul Wermer, Steve Goldfinger, Diana
Deumling, Michael Murray, National Footprint and Biocapacity Accounts 2005: The
underlying calculation method, Global Footprint Network, May 25, 2005