INTERNATIONAL FOOD

POLICY RESEARCH INSTITUTE

A Standard Computable
sustainable options for ending hunger and poverty
General Equilibrium (CGE)
Model in GAMS

Hans Lofgren
Rebecca Lee Harris
Sherman Robinson
With assistance from
Marcelle Thomas
Moataz El-Said

MICROCOMPUTERS IN POLICY RESEARCH 5

The International Food Policy Research Institute ( IFPRI )
IFPRI was established in 1975 to identify and analyze national and international strategies and
policies for meeting the food needs of the developing world on a sustainable basis, with partic-
ular emphasis on low- income countries and poor people ; to make the results of its research
available to all those in a position to use them ; and to help strengthen institutions conducting
research and applying research results in developing countries .

Future HarvestTM and the Consultative Group on International

Agricultural Research (CGIAR)

IFPRI is one of 16 international food and environmental research organizations known as the
Future Harvest Centers . The Centers are principally funded by governments , private founda-
tions, and regional and international organizations , most of which are members of the CGIAR .

About this Series

Microcomputers in Policy Research represents IFPRI's ongoing collective experience in adapting

microcomputer technology for use in food policy analysis in developing countries . Designed to
provide hands-on methods and clear instruction through the extensive use of examples, the pri-
mary purpose of the volumes in this series is to share IFPRI's experience with potential develop-
ing country users , although other users may find them helpful as well .

About GAMS

The General Algebraic Modeling System ( GAMS) is a high - level modeling system for mathemati-
cal programming problems . It consists of a language compiler and a stable of integrated high-
performance solvers . GAMS is tailored for complex, large -scale modeling applications , and
allows you to build large maintainable models that can be adapted quickly to new situations .
GAMS allows the user to concentrate on the modeling problem by making the setup simple.
The system takes care of the time - consuming details of the specific machine and system soft-
ware implementation . For more information , visit www.gams.com .

Copyright 2002 International Food Policy Research Institute . All rights reserved . Sections of this report may be reproduced with-
out the express permission of but with acknowledgment to the International Food Policy Research Institute .

ISBN 0-896-29720-9
A STANDARD COMPUTABLE
GENERAL EQUILIBRIUM (CGE)
MODEL IN GAMS

HANS LOFGREN
REBECCA LEE HARRIS
SHERMAN ROBINSON

with assistance from

MARCELLE THOMAS
and
MOATAZ EL-SAID

MICROCOMPUTERS IN POLICY RESEARCH 5

INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE
Copyright © 2002 International Food Policy Research Institute

All rights reserved. Sections of this book may be reproduced without the express
permission of, but with acknowledgment to, the International Food Policy
Research Institute.

International Food Policy Research Institute

2033 K Street, N.W., Washington, D.C., 20006-1002, U.S.A.
Telephone +1-202-862-5600; Fax +1-202-467-4439; www.ifpri.org

Library of Congress Cataloging-in-Publication Data

Lofgren, Hans.
A standard computable general equilibrium (CGE) model in GAMS / Hans
Lofgren, Rebecca Lee Harris, Sherman Robinson ; with assistance from Marcelle
Thomas and Moataz El-Said.
p. cm.
Includes bibliographical references and index.
ISBN 0-89629-720-9 (alk. paper)
1. Agriculture—Economic aspects—Mathematical models. 2. Food supply—
Mathematical models. 3. Equilibrium (Economics)—Mathematical
models. I. Harris, Rebecca Lee, 1969- II. Robinson, Sherman. III.
Title.
HD1415.L64 2002
338.1'01'51—dc21 2002152627
CONTENTS
Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .iv

Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .v

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .vi

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1

2. The Social Accounting Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3

3. Overview of the Standard CGE Model . . . . . . . . . . . . . . . . . . . . . . . .8

Institutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .10

Commodity Markets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .11

Macroeconomic Balances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .14

4. Mathematical Model Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . .18

Price Block . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .18

Production and Trade Block . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .23

Institution Block . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .31

System Constraint Block . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .35

5. The Standard Model in GAMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . .42

Appendix A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .46

Appendix B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .60

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .68

iii
 TABLES
1. The Basic SAM structure used in the CGE model . . . . . . . . . . . . . . .5

2. Standard SAM for Zimbabwe, 1991 . . . . . . . . . . . . . . . . . . . . . . . . . . .6

3. Alternative closure rules for macrosystem constraints . . . . . . . . . . .13

4. Notational principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .18

5. File structure in GAMS standard CGE modeling system . . . . . . . . .43

iv
FIGURES
1. Production technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .9

2. Flows of marketed commodities . . . . . . . . . . . . . . . . . . . . . . . . . . . . .12

3. The structure of GAMS model and data files . . . . . . . . . . . . . . . . . .44

v
 PREFACE
Over the past decade, the increasing power and reliability of microcom-
puters and the development of sophisticated software designed specifically
for use with them has led to significant changes in the way quantitative
food policy analysis is conducted. These changes cover most aspects of the
analysis, ranging from the collections and analysis of socioeconomic data
to the conduct of model-based policy simulations. The venue of the com-
putations has shifted from off-site mainframes dependent on highly
trained operators and significant capital investment in supporting equip-
ment, to desktop and laptop computers dependent only on the occasional
availability of electricity. This means that it is now feasible to quickly
transfer new techniques between IFPRI and its collaborators in develop-
ing countries, that the costs of policy analysis have been substantially re-
duced, and that a new level of complexity and accuracy in policy analysis
is now possible.
As with any new technology, however, substantial costs in time and
money are involved in learning the most efficient ways of using this new
technology and then transmitting these lessons to others. This series, Mi-
crocomputers in Policy Research, represents IFPRI's ongoing collective ex-
perience in adapting microcomputer technology for use in food policy
analysis in developing countries. Publication decisions are made on the
basis of a review by an external referee. The manuals in the series are pri-
marily for the purpose of sharing these lessons with potential users in de-
veloping countries, although persons and institutions in developed coun-
tries may also find them useful. The series is designed to provide hands-
on methods for quantities food policy analysis. In our opinion, examples
provide the best and clearest form of instruction; therefore, examples—in-
cluding actual software codes wherever relevant—are used extensively
throughout this series.
Computable general equilibrium (CGE) models are used widely in pol-
icy analysis, especially in developed-country academic settings. The pur-
pose of the fifth volume in the series, A Standard Computable General
Equilibrium (CGE) Model in GAMS, by Hans Lofgren, Rebecca Lee Har-
ris, and Sherman Robinson, with assistance from Marcelle Thomas and
Moataz El-Said, is to contribute to and facilitate the use of this class of
models in developing countries. The volume includes a detailed presenta-
tion of a static "standard" CGE model and its required database. The
model is written for application at the country level; however, only mini-
mal changes are needed before it can be applied to a region within a coun-
try (such as a village) or to a farm household involved in production and
consumption activities. The model incorporates features developed over
recent years through IFPRI's research projects. These features—of partic-
ular importance in developing countries—include household consumption
of nonmarketed ("home") commodities, explicit treatment of transaction
costs for commodities that enter the market sphere, and a separation be-
tween production activities and commodities that permits any activity to
produce multiple commodities and any commodity to be produced by mul-
tiple activities. The manual discusses the implementation of the model in
GAMS (the General Algebraic Modeling System) and is accompanied by a
CD-ROM that includes the GAMS files for the model, sample databases,
simulations, solution reports, and a social accounting matrix (SAM)

vi
 vii

aggregation program. Although the volume provides a standardized

framework for analysis, the analyst is not forced to make "one-size-fits-all"
assumptions. The GAMS code is written to give the analyst considerable
flexibility in model specification.
Howarth Bouis and Hans Lofgren, Series Editors
1. INTRODUCTION
Over the past 25 years, computable general equilibrium (CGE) models
have become a standard tool of empirical economic analysis. In recent
years, improvements in model specification, data availability, and com-
puter technology have improved the payoffs and reduced the costs of pol-
icy analysis based on CGE models, paving the way for their widespread
use by policy analysts throughout the world. The purpose of this manual
is to contribute to and facilitate the use of CGE models, making them ac-
cessible to a wider group of economists. The manual includes a detailed
presentation of a static, “standard” CGE model implemented in a com-
puter modeling language called GAMS (General Algebraic Modeling Sys-
tem). It also provides a sample database in an accompanying CD-ROM.1
Although most CGE models have been developed for countries, the
basic framework applies, and has been applied, in settings ranging from
the world (divided into multiple regions) to disaggregated regions within
a country, such as villages, and even to households. In most applications,
the markets and prices in the model represent actual markets with money
used as a medium of exchange. However, especially in household models,
they may be viewed as “implicit” markets where the solution wages and
prices represent “shadow prices” or “exchange values.” Our standard
CGE model is written for application at the country level and has been im-
plemented with a number of country data sets, but only minimal changes
are needed to apply the model to a region within a country or to a pro-
ducer-consumer household.
The standard model includes a number of features designed to reflect
the characteristics of developing countries. The specification follows the
neoclassical-structuralist modeling tradition presented in Dervis et al.
(1982). It incorporates additional features developed in recent years in re-
search projects conducted at IFPRI. These features, of particular impor-
tance in developing countries, include household consumption of nonmar-
keted (or “home”) commodities, explicit treatment of transaction costs for
commodities that enter the market sphere, and a separation between pro-
duction activities and commodities that permits any activity to produce

The authors would like to thank Ed Taylor for a constructive review, and Renger van
Nieuwkoop and Jennifer Chung-I Li for useful comments.
1We assume that the reader has a basic familiarity with CGE modeling using GAMS.
Brooke et al. (1998) is the basic reference on the GAMS software; it also includes a
self-contained tutorial. The basics of GAMS-based CGE modeling are summarized in
Robinson et al. (1999). Lofgren (2000a, 2000b) presents a set of hands-on exercises
in CGE modeling with GAMS. Extensive treatments of CGE methods are found in
Dervis et al. (1982), Robinson (1989), Shoven and Whalley (1992), Dixon et al.
(1992), and Ginsburgh and Keyzer (1997). References to and examples of CGE-based
analyses of food policy in developing countries are found in the Trade and Macro-
economics Division section of the IFPRI website (www.ifpri.org).

1
2

multiple commodities and any commodity to be produced by multiple ac-

tivities.
The CD-ROM provided includes the GAMS files for the CGE model,
sample databases, simulations, solution reports, and a social accounting
matrix (SAM) aggregation program. In the GAMS code, the model is ex-
plicitly linked to a file for country data, including a “standard” SAM that
follows the format required for the standard CGE model and a set of elas-
ticities. Optionally, the user may provide quantity data for primary factors
(for example, labor types) that appear in the SAM. In the model code, this
data set is used to define model parameter values in a manner that assures
that the base solution to the model exactly reproduces the values in the
SAM. In other words, the model is “calibrated” to the SAM. It is, more-
over, straightforward for users to develop new data sets for other applica-
tions.
The CGE model and the accompanying GAMS code are written to give
analysts considerable flexibility. He or she can choose between alternative
treatments for macroeconomic balances and for factor markets. It is also
possible to exclude various features that appear in the standard model,
such as home consumption and transaction costs. The country database to
which the model should be applied can incorporate a wide range of policy
tools as well as any desired degree of disaggregation of production activi-
ties, commodities, households, and enterprises. Flexibility in terms of
model structure and the fact that model parameters are derived from an
empirical database (which may be very detailed) permit the analyst to cap-
ture country-specific aspects of economic structure and functioning.
Hence, although the manual provides a standardized framework for
analysis, the analyst is not forced to make “one-size-fits-all” assumptions.
We consider this CGE model and the accompanying computer code as
work in progress and encourage readers and users to send us their com-
ments. A number of extensions are possible. For example, users may be in-
terested in adding alternative treatments of production technology or a
more detailed treatment of policy tools. However, when new features are
added, there is a tradeoff between additional versatility and additional
complexity. Unless the new features are of general interest, they should
preferably be added in the context of specific applications that use the cur-
rent, relatively simple model as their starting point. As noted earlier, the
model can be easily adapted for application to regions within a country or
to a household that is involved in production and consumption. More fun-
damental changes would be needed to make it dynamic or to turn it into a
world model.2
The remainder of this manual is organized as follows: Chapter 2 de-
scribes the standard SAM. Chapter 3 provides an overview of the features
of the CGE model, followed by an equation-by-equation description in
Chapter 4. Chapter 5 describes the structure of the GAMS files for the
standard CGE model and its database and discusses how they may be used
for policy analysis. The appendixes include the mathematical model state-
ment in summary form and core sections of the GAMS code for the model.

2To apply the model to a region or a household, the only changes needed involve the
addition of new rules for closing the accounts for the government and the rest of the
world (now representing the economy outside the region or the household). The
database (including the SAM) should then represent a region or a farm household.
2. THE SOCIAL ACCOUNTING
MATRIX
A social accounting matrix (SAM) is a comprehensive, economywide data
framework, typically representing the economy of a nation.3 More techni-
cally, a SAM is a square matrix in which each account is represented by a
row and a column. Each cell shows the payment from the account of its
column to the account of its row. Thus, the incomes of an account appear
along its row and its expenditures along its column. The underlying prin-
ciple of double-entry accounting requires that, for each account in the
SAM, total revenue (row total) equals total expenditure (column total).4
Table 1 shows an aggregated SAM with verbal explanations in the cells
instead of numbers. With one exception, it has all of the features required
for implementation with the standard CGE model. The exception is that
in the standard SAM, taxes have to be paid to tax accounts, disaggregated
by tax type, each of which forwards its revenues to the core government
account. The tax types are divided into direct taxes (on domestic non-
government institutions and factors), commodity sales taxes, import
taxes, export taxes, activity taxes, and value-added taxes. Also note that,
in the standard SAM, payments are not permitted in the blank cells of
Table 1. Any original SAM that includes such payments should be re-
structured before being implemented with the standard CGE model.5
Table 2 shows a real-world standard SAM for Zimbabwe in which the
tax accounts are treated in the required manner.6 In addition, it has mul-
tiple accounts for activities, commodities, factors, and domestic non-

3For general discussions of SAMs, see Pyatt and Round (1985) and Reinert and
Roland-Holst (1997); for perspectives on SAM-based modeling, see Pyatt (1988) and
Robinson and Roland-Holst (1988).
4The GAMS program checks that the SAM that is entered is balanced (meaning the
row and column totals are equal for each account). If the absolute value of the sum
of account imbalances exceeds a cutoff point, an optimization program is used to es-
timate a balanced SAM. The program, which minimizes the entropy distance of the
cells of the estimated SAM from those of the initial SAM subject to the constraint
that row and column totals are equal, is primarily intended to remove rounding er-
rors. For SAM estimation in GAMS in a setting with substantial imbalances in raw
data (not only rounding errors), see Robinson and El-Said (2000) and Robinson, Cat-
taneo, and El-Said (2001).
5One common case would be payments from the government to factors (for the labor
services provided by government employees). To restructure the SAM to work with
the standard model, the preferred approach is to reallocate such payments to a
commodity for government services that pays a government service activity which,
in turn, pays the labor account.
6For other examples of SAMs that have the required structure, see the “data sets”
page on IFPRI’s website (www.ifpri.org).

3
4

government institutions. In each category, the GAMS code can handle any
desired disaggregation, including having just a single account. In any real-
world application, the preferred disaggregation of the SAM and the CGE
model depends on data availability and the purposes of the analysis. It is
typically preferable to include relatively detailed treatment in areas of in-
terest while keeping the database relatively aggregated in other areas.7
With regard to the structure of the standard SAM, a number of fea-
tures are noteworthy. First, the standard SAM distinguishes between ac-
counts for “activities” (the entities that carry out production) and “com-
modities.” The receipts are valued at producer prices in the activity ac-
counts and at market prices (including indirect commodity taxes and
transaction costs) in the commodity accounts. The commodities are activ-
ity outputs, either exported or sold domestically, and imports. This sepa-
ration of activities from commodities is preferred because it permits ac-
tivities to produce multiple commodities (for example, a dairy activity may
produce the commodities cheese and milk) while any commodity may be
produced by multiple activities (for example, activities for small-scale and
large-scale maize production may both produce the same maize commod-
ity). In the commodity columns, payments are made to domestic activities,
the rest of the world, and various tax accounts (for domestic and import
taxes). This treatment provides the data needed to model imports as per-
fect or imperfect substitutes vis-à-vis domestic production.8
Second, the matrix explicitly associates trade flows with transactions
(trade and transportation) costs, also referred to as marketing margins.
For each commodity, the SAM accounts for the costs associated with do-
mestic, import, and export marketing. For domestic marketing of domes-
tic output, the marketing margin represents the cost of moving the com-
modity from the producer to the domestic demander. For imports, it rep-
resents the cost of moving the commodity from the border (adding to the
c.i.f. price) to the domestic demander, while for exports, it shows the cost
of moving the commodity from the producer to the border (reducing the
price received by producers relative to the f.o.b. price). The Zimbabwe
SAM in Table 2 shows how these transaction costs appear in commodity
and activity accounts in the standard SAM: A services activity, in Table 2
called transportation (account 4), produces a commodity (account 8) that,
like other commodities, may be purchased for intermediate use by activi-
ties and for final use by institutions. However, the transportation
commodity also receives payments from three special accounts, represent-
ing the transaction costs associated with domestic sales, imports, and
exports (accounts 10-12).9 These special accounts are paid by the accounts

7The CD-ROM that accompanies this manual includes a program for aggregating an
existing SAM.
8In addition, our model code makes it possible to treat selected imports as separate,
“noncomparable” commodities (not produced domestically). In the commodity rows,
such import commodities receive payments from one or more domestic users. In the
columns, these payments would be passed on to the accounts for the rest of the
world, import marketing margins, and relevant taxes. The columns for this category
of imports do not have any payments to domestic activities.
 Table
The
1-
model
CGE
the
in
used
structure
SAM
Basic

Expenditures

Savings- Rest
the
of
Receipts Activities Commodities
Factors Households Government Investment
Enterprises World
)(ROW Total

Activities Marketed Home- Activity

outputs consumed income
outputs output
)(gross

Commodities Intermediate Transaction Private Government Investment Exports Demand

inputs costs consumption consumption
Factors aValue
- dded Factor
income Factor
from
ROW income
Households income
Factor Interhousehol
Surplus
tod Transfers
to Transfers
to Household
to
households transfers households households households income
from
ROW
Enterprises income
Factor Transfers
to to
Transfers Enterprise
to
enterprises enterprises enterprises income
from
ROW
Government ,
taxes
Producer ,
taxes
Sales Factor
income
to
Transfers to
Surplus Transfer
to Government
value
-a dded etariffs
, xport ,government
government
to Govern
, ment Government income
tax taxes direct
taxes
factor enterpr
direct ise from
ROW
taxes
Savings- Household Enterprise Government Foreign Savings
Investment savings savings savings savings
Rest
the
of Imports income
Factor Surplus Government Foreign
World
R
)( OW ROW
to ROW
to transfers
to exchange
ROW outflow
Total Activity Supply Factor Household Enterprise Government Investment Foreign
expenditures expenditures expenditures expenditures inflow
exchange
5
 SAM
S
-
2
Table
tandard
,1991
Zimbabwe
for
6

Activities Commodities Transaction

costs Factors Households institutions
Other
Category 5
4
123 6 7 8 9 11
10 12 15
14
13 16
17 19
18 20 21 22 23 24
Total
25

Activities
Agriculture
,1.
slarge
- cale 0
0 50,250 5,250
0
Agriculture
,2.
small
-scale 0 0 0 0 6070 0 0 0 0 0 0 0 0 0 685 0 0 0 0 0 0 0 0
1,355
Industry
3. 0 0 0 0 0 1007,859 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 107,859
Transportation
4. 0 0 0 0 0 0 0
80,263 0 0 0 0 0 0 0 0 0 0 0 0 0 0 8,263
0
Other
5.
services 0 0 0 0 0 0 105,781 0 0 0 0 0 0 0 0 0 0 0 0 0 1005,781
0000

O O O O
OOOO

oooo
Commodities
Agriculture
6. 244 72,697
0 97 0 0 0 0 0 0 0 0 0 0 616 629 0 0 0 0 0 2,964 0-7,223
30
Industry
7. 3,619
3,075
5,859
165
1,145 0 0 0 0 0 0 0 0 7,256
0
5,236 0 310 0 0 02,513
32,083
-494
3,399
Transportation
8. 183
176
29
38 282 0 0 0 0 986
1,689
3,444 0 0 0605 662 0169 0 0 0 08,263
services
Other
9. 30
715
281 421
1,685 0 0 0 0 0 0 0 0 0 0 1,913 2,561 40,295 0 0 12,784
0,5986,283
costs
Transaction
10.
sales
Domestic 0 0 0 0 0 2,788 0
657 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 30,444
Import
11. s 0 0 0 0 0
9 1,680 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 10,689
Exports
12. 0 0 0 580
0 406 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 90086
Factors
13.
Labor 2,447
2,936
684
755
6,028
0 0 0 0 0 0 0 0 0 0 0 0 0 0 012,851
Capital
14. 1,950
5,386
260
1,719 3,524 0 0 0 0 0 0 0 0 0 0 0 0 0 012,839

ос
Land
15. 458
0
137 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0594
000

000
000

000

Households
16.
Rural 0 0 0 0 0 0 0 0 0 0 011,979
594
,605 0 1,113
5,526 0
259 0 0 102 0 011,179
Urban
17. 0 0 0 0 0 0 0 0 0 0 0 011,220 0
127 0
3346
,306 0 0 0 0 0 104,998

institutions
Other
Enterprise
18. 0 0 0 0 0 0 0 0 0 0 0 0 100,7330 0 0 01,209 0 0 0 0 011,942
19.
Government 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 03,727 1,861
1,478 291 0 70,357
Direct
20.
taxes 0 0 0 0 0 0 0 0 0 0 0 0 0 0 709 1,351
1,667 0 0 0 0 0 0 30,727
taxes
Indirect
21. 176 43 188
524 0546 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 10,478
tariffs
Import
22. 0 0 0 0 10 1,800 051 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1,861
23.
world
of
Rest 0 0 0 0 0 48 7,550 0 450 0 0 0 26 0 0 0 535 418 0 0 0 0 0 90,027
I24.
Savings
-nvestment 0 0 0 0 0 0 0 0 0 0 00 1,415 2,280 908
-504 0 0 10,559 0 0 5,658
25.
Stock
change 0 0 0 0 0 0 0 0 0 0 0 00 0 0 -525
0 -525

Total -525
5,658
9,027
1,861
1,478
3,727
7,357
11,942
14,998
11,179
594
12,839
12,851
986
1,689
3,444
16,283
32,083
7,223
15,781
8,263
17,859
1,355
5,250 0

:T999
Source
(1
Bautista
).and
homas
dollars
presented
Zambian
million
in
Note
. ata
:D
are
 7

for marketed agricultural and industrial commodities (accounts 6 and 7).

Thus the total value of each commodity includes these transaction costs.
The standard CGE model will also work with SAMs without this treat-
ment of (and these accounts for) transaction costs.
Third, as noted, the government is disaggregated into a core govern-
ment account and different tax accounts, one for each tax type. This dis-
aggregation is often necessary because the economic interpretation of
some payments may otherwise be ambiguous. In any given application,
the SAM may exclude any (or all) of the individual tax accounts. In the
SAM, payments between the government and other domestic institutions
are reserved for transfers.
Fourth, the domestic nongovernment institutions in the SAM consist
of households and enterprises. The enterprises earn factor incomes (re-
flecting their ownership of capital and/or land). They may also receive
transfers from other institutions. Their incomes are used for direct taxes,
savings, and transfers to other institutions. As opposed to households, en-
terprises do not consume. Assuming that the relevant data are available,
it is preferable to have one or more accounts for enterprises when these
have tax obligations and a savings behavior that are independent of the
household sector. The enterprise sector should be disaggregated in a man-
ner that captures differences across enterprises in terms of tax rates, sav-
ings rates, and the shares of retained earnings that are received by differ-
ent household types. For example, in some settings it may be appropriate
to disaggregate enterprises into the categories nonagricultural (meaning
earnings from nonagricultural capital), small-scale agricultural (earnings
from land and capital controlled by small farmers), and large-scale agri-
cultural (earnings from land and capital of large farmers). Technically, the
standard CGE model requires that the SAM have at least one household
account; enterprise accounts are not necessary.
Finally, the SAM distinguishes between home consumption, which is
activity-based, and households’ marketed consumption, which is commod-
ity-based. Home consumption, which in the SAM appears as household
payments to activities, is valued at producer prices—that is, without mar-
keting margins and the sales taxes that may be imposed on marketed com-
modities.10 Household consumption of marketed commodities appears as
payments from household accounts to commodity accounts, the values of
which include marketing margins and commodity taxes. The standard
CGE model also accepts a SAM without (explicit) home consumption.

9The distinction between intermediate use of transportation services and their use
in output marketing (giving rise to transaction costs) is that intermediate input use
is part of the production process whereas use in marketing is incurred only if the
output is actually marketed (as opposed to being home-consumed). Input-output ta-
bles typically include information on marketing margins but in a less (or differently)
disaggregated format than that proposed for the standard model SAM. Hence, addi-
tional data and analysis may be needed if the model user wishes to construct a SAM
with the proposed treatment of marketing margins.
10In the model, home consumption demand is for the commodity output(s) of the ac-
tivities that, in the SAM, receive payments from households (compare with footnote
7 and equations 18 and 34 in Chapter 4).
 3. OVERVIEW OF THE STANDARD
CGE MODEL
The standard CGE model explains all of the payments recorded in the
SAM. The model therefore follows the SAM disaggregation of factors, ac-
tivities, commodities, and institutions. It is written as a set of simultane-
ous equations, many of which are nonlinear. There is no objective func-
tion. The equations define the behavior of the different actors. In part,
this behavior follows simple rules captured by fixed coefficients (for ex-
ample, ad valorem tax rates). For production and consumption decisions,
behavior is captured by nonlinear, first-order optimality conditions—that
is, production and consumption decisions are driven by the maximization
of profits and utility, respectively. The equations also include a set of con-
straints that have to be satisfied by the system as a whole but are not nec-
essarily considered by any individual actor. These constraints cover mar-
kets (for factors and commodities) and macroeconomic aggregates (bal-
ances for Savings–Investment, the government, and the current account
of the rest of the world).
This chapter summarizes the basic characteristics of the model. Un-
like the more detailed presentation in Chapter 4, it uses no mathematical
notation.

ACTIVITIES, Each producer (represented by an activity) is assumed to maximize prof-

MARKETS
PRODUCTION, its, defined as the difference between revenue earned and the cost of fac-
tors and intermediate inputs. Profits are maximized subject to a produc-
AND FACTOR tion technology, the structure of which is shown in Figure 1. At the top
level, the technology is specified by a constant elasticity of substitution
(CES) function or, alternatively, a Leontief function of the quantities of
value-added and aggregate intermediate input. The Leontief alternative is
the default. The CES alternative may be preferable for particular sectors
if empirical evidence suggests that available techniques permit the aggre-
gate mix between value-added and intermediate inputs to vary. Value-
added is itself a CES function of primary factors whereas the aggregate in-
termediate input is a Leontief function of disaggregated intermediate
inputs.
Each activity produces one or more commodities according to fixed
yield coefficients. As noted, a commodity may be produced by more than
one activity. The revenue of the activity is defined by the level of the
activity, yields, and commodity prices at the producer level.
As part of its profit-maximizing decision, each activity uses a set of fac-
tors up to the point where the marginal revenue product of each factor is
equal to its wage (also called factor price or rent). Factor wages may differ
across activities, not only when the market is segmented but also for mo-
bile factors. In the latter case, the model incorporates discrepancies that
stem from exogenous causes (for example, wage differences across activi-
ties resulting from considerations such as status, comfort, or health risks).

8
 9

Figure 1—Production technology

Commodity outputs
(fixed yield coefficients)

Activity level
(CES/Leontief function)

Value-added Intermediate
(CES function) (Leontief function)

Primary Composite
factors commodities

Imported Domestic

The user can choose between alternative factor market closures

(mechanisms for equilibrating supplies and demands in factor markets).
According to the default closure, the quantity supplied of each factor is
fixed at the observed level. An economywide wage variable is free to vary
to assure that the sum of demands from all activities equals the quantity
supplied. Each activity pays an activity-specific wage that is the product of
the economywide wage and an activity-specific wage (distortion) term. For
the default closure, the latter terms are fixed.
Alternatively, it is possible to assume that a factor is unemployed and
the real wage is fixed. This assumption may, for example, be appropriate
in settings where there is considerable unemployment for a given labor
category. Compared with the default closure, the only change is that the
economywide wage variable is fixed (or exogenized) while the supply vari-
able is “flexed” (or endogenized). Each activity is free to hire any desired
quantity at its fixed, activity-specific wage (which, implicitly, is indexed to
the model numéraire). In this setting, the supply variable is superfluous;
it merely records the total quantity demanded.
Under a third closure, the factor market is segmented and each activ-
ity is forced to hire the observed, base-year quantity—that is, the factor is
activity-specific. This closure may be preferred in short-run analyses or
when there are significant quality differences between the units of a fac-
tor that are used in different activities—for example, units of non-
agricultural capital used in different industrial and service activities.
For this case, the quantities of activity-specific factor demands and the
10

economywide wage are fixed while the activity-specific wage terms and the
supply variables are flexible.

INSTITUTIONS In the CGE model, institutions are represented by households, enter-

prises, the government, and the rest of the world.
The households (disaggregated as in the SAM) receive income from
the factors of production (directly or indirectly via the enterprises) and
transfers from other institutions. Transfers from the rest of the world to
households are fixed in foreign currency. In fact, all transfers between the
rest of the world and domestic institutions and factors are fixed in foreign
currency. The households use their income to pay direct taxes, save, con-
sume, and make transfers to other institutions. In the basic model version,
direct taxes and transfers to other domestic institutions are defined as
fixed shares of household income whereas the savings share is flexible for
selected households. The treatment of direct tax and savings shares is re-
lated to the choice of closure rule for the government and savings–
investment balances. This topic is discussed further in the final section of
this chapter, on macroeconomic balances. The income that remains after
taxes, savings, and transfers to other institutions is spent on
consumption.
Household consumption covers marketed commodities, purchased at
market prices that include commodity taxes and transaction costs, and
home commodities, which are valued at activity-specific producer prices.11
Household consumption is allocated across different commodities (both
market and home commodities) according to linear expenditure system
(LES) demand functions, derived from maximization of a Stone–Geary
utility function (for details, see Blonigen et al. 1997, 223–225, and Dervis
et al. 1982, 482–485).
Instead of being paid directly to the households, factor incomes may be
paid to one or more enterprises. Enterprises may also receive transfers
from other institutions. Enterprise incomes are allocated to direct taxes,
savings, and transfers to other institutions. Enterprises do not consume.
Apart from this, the payments to and from enterprises are modeled in the
same way as the payments to and from households.
The government collects taxes and receives transfers from other insti-
tutions. In the basic model version, all taxes are at fixed ad valorem rates.
The government uses this income to purchase commodities for its con-
sumption and for transfers to other institutions. Government consump-
tion is fixed in real (quantity) terms whereas government transfers to do-
mestic institutions (households and enterprises) are CPI-indexed. Gov-
ernment savings (the difference between government income and spend-
ing) is a flexible residual.
The final institution is the rest of the world. As noted, transfer pay-
ments between the rest of the world and domestic institutions and factors
are all fixed in foreign currency. Foreign savings (or the current account

11In the standard SAM, home consumption is only disaggregated by activity and
household, not by commodity, activity, and household. When households consume
from activities that produce multiple outputs, extraneous, non-SAM data are needed
to allocate home consumption across the commodities produced by each relevant
multiple-output activity.
 11

deficit) is the difference between foreign currency spending and receipts.

Commodity trade with the rest of the world is discussed in the next sec-
tion. Thereafter, the final section of this chapter discusses the rules for
clearing the macroeconomic balances (the macroclosures)—that is, how
equilibrium is achieved in the balances for the government, the rest of the
world, and the Savings–Investment account (where institutional savings
are aggregated and allocated to domestic investment).

MARKETS With the exception of home-consumed output, all commodities (domestic

COMMODITY
output and imports) enter markets. Figure 2 shows the physical flows for
marketed commodities along with the associated quantity and price vari-
ables as defined in the model equations discussed in the following section.
Domestic output may be sold in the market or consumed at home. For
marketed output, the first stage in the chain consists of generating aggre-
gated domestic output from the output of different activities of a given
commodity. These outputs are imperfectly substitutable as a result of, for
example, differences in timing, quality, and distance between the locations
of activities. A CES function is used as the aggregation function. The de-
mand for the output of each activity is derived from the problem of mini-
mizing the cost of supplying a given quantity of aggregated output subject
to this CES function. Activity-specific commodity prices serve to clear the
implicit market for each disaggregated commodity.
At the next stage, aggregated domestic output is allocated between ex-
ports and domestic sales on the assumption that suppliers maximize sales
revenue for any given aggregate output level, subject to imperfect trans-
formability between exports and domestic sales, expressed by a constant
elasticity of transformation (CET) function. In the international markets,
export demands are infinitely elastic at given world prices. The price re-
ceived by domestic suppliers for exports is expressed in domestic currency
and adjusted for the transaction costs (to the border) and export taxes (if
any). The supply price for domestic sales is equal to the price paid by do-
mestic demanders minus the transaction costs of domestic marketing
(from the supplier to the demander) per unit of domestic sales. If the com-
modity is not exported, total output is passed to the domestic market.
Domestic demand is made up of the sum of demands for household
consumption, government consumption, investment (the determination of
which is discussed below), intermediate inputs, and transactions (trade
and transportation) inputs.
To the extent that a commodity is imported, all domestic market de-
mands are for a composite commodity made up of imports and domestic
output, the demands for which are derived on the assumption that do-
mestic demanders minimize cost subject to imperfect substitutability. This
is also captured by a CES aggregation function.12 Total market demand is
directed to imports for commodities that lack domestic production and to
domestic output for non-imported commodities.
The derived demands for imported commodities are met by interna-
tional supplies that are infinitely elastic at given world prices. The import

12This function is also referred to as an Armington function, named after Paul Arm-
ington who introduced imperfect substitutability between imports and domestic
commodities in economic models (Armington 1969).
 commodities
marketed
F
of
Figure
-lows
2
12

Commodity
outpu
from t Aggregate
1activity exports
QXAC
( PE
Q
(
) E
PXAC
)
Aggregate
CES CET
output
P
|) X
Q
(

Commodity Household
Domestic
outpu
from t consumption
sales QQ
(
|P
) H
activity
n
P
|QDS-
( D
QXAC
( +
PDD
) Government
PXAC
)
Composite consumption
CES )|P
(QQ G
commodity
Q
( QPQ
) +
Investment
Qdst
(
q
|+ INV
Aggregate PQ
)
imports +
Q
PM
) M
( Intermediate
use
| INT
)
Q
(
P Q

Note
: ES
;C
.substitution
transformation
of
elasticity
constant
isET
 13

prices paid by domestic demanders also include import tariffs (at fixed ad
valorem rates) and the cost of a fixed quantity of transactions services per
import unit, covering the cost of moving the commodity from the border
to the demander.13 Similarly, the derived demand for domestic output is
met by domestic suppliers . The prices paid by the demanders include the
cost of transactions services, in this case reflecting that the commodity
was moved from the domestic supplier to the domestic demander. The
prices received by domestic suppliers are net of these transaction costs .
Flexible prices equilibrate demands and supplies of domestically marketed
domestic output.
Compared with the alternative assumptions of perfect substitutability
and transformability, the assumptions of imperfect transformability (be-
tween exports and domestic sales of domestic output) and imperfect sub-
stitutability (between imports and domestically sold domestic output)

Table 3-Alternative closure rules for macrosystem constraints

Constraint

Government Rest of the World Savings-Investment

GOV- 1 : ROW- 1 : SI- 1 :

Flexible government Fixed foreign savings; Fixed capital formation ;
savings; fixed direct flexible real exchange rate uniform MPS point change
tax rates for selected institutions

GOV-2: ROW-2: SI-2 :

Fixed government savings ; Flexible foreign savings; Fixed capital formation;
uniform direct tax rate fixed real exchange rate scaled MPS for selected
point change for selected institutions
institutions

GOV-3: SI-3:
Fixed government savings; Flexible capital formation ;
scaled direct tax rates for fixed MPS for all non
selected institutions government institutions

SI-4 :
Fixed investment and gov-
ernment consumption ab-
sorption shares (flexible
quantities) ; uniform MPS
point change for selected
institutions

SI-5 :
Fixed investment and gov-
ernment consumption ab-
sorption shares (flexible
quantities) ; scaled MPS for
selected institutions

Notes: For the specified closure rules, the choice for one of the three constraints does not
constrain the choice for the other two. MPS is marginal propensity to save.

13Note that these transaction costs are not ad valorem. The rates the ratio between
the margin and the price without the margin—change with changes in the prices of
transactions services and/or the commodities that are marketed.
14

permit the model to better reflect the empirical realities of most countries.
The assumptions used give the domestic price system a degree of inde-
pendence from international prices and prevent unrealistic export and im-
port responses to economic shocks. At the disaggregated commodity level,
these assumptions allow for a continuum of tradability and two-way trade,
which is commonly observed even at very fine levels of disaggregation.

MACROECONOMIC The CGE model includes three macroeconomic balances: the (current)
balance, the external balance (the current account of the bal-
BALANCES government
of
ance payments, which includes the trade balance), and the Savings–
Investment balance. In the GAMS code, the user chooses among a rela-
tively large number of pre-programmed alternative closure rules for these
balances. The choices made have no influence on the solution to the base
simulation but will typically influence the results for other simulations.
The closures are summarized in Table 3.14
For the government balance, the default closure (GOV-1) is that gov-
ernment savings (the difference between current government revenues
and current government expenditures) is a flexible residual while all tax
rates are fixed. Under the two alternative government closures, the direct
tax rates of domestic institutions (households and enterprises) are ad-
justed endogenously to generate a fixed level of government savings. For
the first of these alternative closures (GOV-2), the base-year direct tax
rates of selected domestic nongovernment institutions (households and
enterprises) are adjusted endogenously by the same number of percentage
points. For the second (GOV-3), the rates of selected institutions are mul-
tiplied by a flexible scalar.15 For these three government closures, govern-
ment consumption is fixed, either in real terms or as a share of nominal
absorption, depending on the treatment of the Savings–Investment bal-
ance, discussed below. In other words, we do not specify a closure where
government savings and direct tax rates are both fixed and government
consumption is the adjusting variable.
For the external balance, which is expressed in foreign currency, the
default closure (ROW-1) is that the real exchange rate is flexible while for-
eign savings (the current account deficit) is fixed. Given that all other
items are fixed in the external balance (transfers between the rest of the
world and domestic institutions), the trade balance is also fixed. If, ceteris
paribus, foreign savings are below the exogenous level, a depreciation of

14Macroclosure of CGE models is a contentious topic with a large literature. For sum-
maries, see Robinson (1991), Rattsø (1982), and Taylor (1990).
15The difference between these two closures in terms of simulated changes in post-
tax incomes may be substantial, as illustrated by an example with two institutions—
an enterprise and a household that each, under base conditions, have incomes of 200
and face direct tax rates of 20 percent and 10 percent, respectively. Assume that total
direct tax collection has to increase from 60 to 90 to reach a fixed level of govern-
ment savings (assuming, for simplicity, no income changes). Under the first closure,
the rates would increase by 7.5 percentage points for both entities, to 27.5 percent
for the enterprise and 17.5 percent for the household. The payments would increase
by 15 percentage points for both. Under the second closure, the new tax rates would
be 30 percent and 15 percent (multiplying both base rates by 1.5), respectively. The
tax payments increase by 20 percentage points for the enterprise and 10 percentage
points for the household.
 15

the real exchange rate would correct this situation by simultaneously

(i) reducing spending on imports (a fall in import quantities at fixed world
prices) and (ii) increasing earnings from exports (an increase in export
quantities at fixed world prices). Under an alternative closure (ROW-2),
the real exchange rate (indexed to the model numéraire) is fixed while for-
eign savings (and the trade balance) is flexible.16
For the Savings–Investment balance, closures are either investment-
driven (the value of savings adjusts) or savings-driven (the value of in-
vestment adjusts). The default closure (SI-1) is investment-driven. Real
investment quantities are fixed. In order to generate savings that equal
the cost of the investment bundle, the base-year savings rates of selected
nongovernment institutions are adjusted by the same number of percent-
age points. Implicitly, it is assumed that the government is able to imple-
ment policies that generate the necessary private savings to finance the
fixed real investment quantities.
Four additional closures are also specified. The first alternative (SI-2)
is also investment-driven. It differs from the default in that, instead of ad-
justing base-year savings rates by a fixed number of percentage points, the
rates of selected institutions are multiplied by a scalar (compare with the
above discussion of the treatment of direct tax rates under alternative
government closures). The second alternative (SI-3) is savings-driven. All
nongovernment savings rates are fixed. The quantity of each commodity
in the investment bundle is multiplied by a flexible scalar to ensure that
the investment cost equals the savings value.
The last two alternatives (SI-4 and SI-5) are “balanced” closures,
which may be viewed as variants of investment-driven closures although
they also impose an adjustment rule for government consumption. Under
these, adjustments in absorption are spread across all of its components
(household consumption, investment, and government consumption).17
The nominal absorption shares of investment and government consump-
tion are fixed at base levels, although this could be generalized. (Except
for SI-4 and SI-5, government consumption is fixed in real terms.) Given
this specification, the residual share for household consumption is also
fixed. For the first balanced closure (SI-4), the savings rates of selected in-
stitutions are adjusted by an equal number of percentage points (compare
with SI-1). For the second balanced closure (SI-5), the savings rates of
selected institutions are scaled so as to generate enough savings to finance
investment (compare with SI-2). The balanced closures are compatible
with any combination of the pre-programmed closures for the government
and the rest of the world.
The appropriate choice between the different macroclosures depends
on the context of the analysis. Given that this is a single-period model, a
closure combining fixed foreign savings, fixed real investment, and fixed
real government consumption may be preferable for simulations that

16For a discussion of the real exchange rate in neoclassical, trade-focused CGE

models, see Devarajan et al. (1993).
17Under the other investment-driven closures, the quantities of investment and gov-
ernment consumption are both fixed. Hence, household consumption is the only part
of absorption that adjusts (in response to changes in savings rates). Under the
savings-driven closure, the bulk of the adjustment is carried by investment.
16

explore the equilibrium welfare changes of alternative policies . In terms of

the rules in Table 3 , this closure combines ROW- 1 with SI- 1 or SI- 2 and
any one of the three specified government closures . In the literature on
macroclosures, this is known as "Johansen closure. "18 Such a closure
avoids the misleading welfare effects that appear when foreign savings
and real investment change in simulations with a single-period model—
ceteris paribus, for the simulated period , increases in foreign savings and
decreases in investment raise household welfare (and vice versa for de-
creases in foreign savings and increases in investment) . This result is mis-
leading because the analysis does not capture welfare losses in later
periods that arise from a larger foreign debt and a smaller capital stock .
With regard to government consumption , the model does not capture its
direct and indirect welfare contributions; to avoid misleading results , it is
also preferable in welfare analysis to keep this variable fixed .
Another macroclosure often used in applied work is the savings-driven
"neoclassical closure" in which investment is determined by the sum of
private, government, and foreign savings. It is distinguished from the Jo-
hansen closure in that it uses SI-3 instead of SI- 1 or SI-2 . Both the
savings-driven neoclassical closure and the investment-driven Johansen
closure seem extreme when looking at the historical experience of coun-
tries adjusting to macroshocks. If the analysis aims at capturing the likely
effects of an exogenous shock or policy change in a given (historical, cur-
rent, or future) setting, perhaps in order to explore the role for comple-
mentary policies, it is generally preferable to impose a closure that more
closely mimics the real world, with simultaneous adjustments in the three
components of absorption. Under these circumstances, a macroscenario
that incorporates a balanced closure (in Table 3 , SI-4 or SI-5 ) is a useful
option.
The Johansen, neoclassical, and balanced closures all assume no link
between macrovariables and aggregate employment . If full- employment is
assumed in the factor markets, these closures will yield different effects of
shocks on the composition of aggregate demand, but with little or no ef-
fect on aggregate GDP. It is also feasible in the standard model to specify
a “Keynesian " closure in which aggregate employment is linked to
macrovariables through a Keynesian multiplier process. This closure is an
example of a structuralist macromodel of the type advocated by Lance
Taylor ( 1990) . In this Keynesian closure, investment is fixed in real terms .
In the labor market (in one of the labor markets if labor is disaggregated),
it is assumed that the real wage is flexible in a setting with unemploy-
ment. Adjustment in the real wage induces firms to change their labor de-
mand and employment sufficiently to generate incomes and savings that
are needed to finance the fixed quantity of real investment. In this model ,
an increase in exogenous real investment (or in real government expendi-
ture) will generate a fall in the wage, an increase in employment, an in-
crease in income, and an increase in savings to finance the increased in-
vestment. In the context of the standard model, the easiest way to imple-
ment this closure is to (i ) introduce a modified investment-driven macro-
closure that is identical to SI- 1 except that the MPS adjustment variable

18
A closure of this type was used in the first CGE model, developed by Leif Johansen
(1960).
 17

is fixed; and (ii) for one labor type, introduce a modified version of the de-
fault factor-market closure where not only the wage variable, WF, but also
the labor supply variable, QFS, is flexible.
Finally, it is often informative to explore the impact of experiments
under a set of alternative macroclosures. The results provide important
insights into the real-world tradeoffs that are associated with alternative
macroeconomic adjustment patterns.
 4. MATHEMATICAL MODEL

STATEMENT

This chapter presents the mathematical model statement equation by

equation. In its mathematical form , the CGE model is a system of simul-
taneous , nonlinear equations. The model is square—that is, the number of
equations is equal to the number of variables . In this class of models , this
is a necessary (but not a sufficient) condition for the existence of a unique
solution. The chapter divides the equations into four blocks : prices , pro-
duction and trade, institutions , and system constraints . New items (sets ,
parameters , and variables) are defined the first time that they appear in
the equations . Table 4 summarizes the notational principles . Parameter
and variable names are chosen to facilitate interpretation ; most impor-
tantly, commodity and factor quantities start with q, commodity prices
with p , and factor prices with w .

Table 4- Notational principles

Item Notation

Endogenous variables Upper-case Latin letters without a bar

Exogenous variables Upper-case Latin letters with a bar
Parameters Lower-case Latin letters (with or without a bar) or
lower-case Greek letters (with or without super-
scripts)
Set indices Lower-case Latin letters as subscripts to variables and
parameters
Notes: Exogenous variables are fixed in the basic model ver-
sion but may be endogenous in versions with different
treatments of macro- or factor-market closures .

Notes: For the specified closure rules, the choice for one of the three constraints does not
constrain the choice for the other two . MPS is marginal propensity to save.

PRICE BLOCK The price system of the model is rich , primarily because of the assumed
quality differences among commodities of different origins and destina-
tions (exports , imports, and domestic outputs used domestically) . The
price block consists of equations in which endogenous model prices are
linked to other prices ( endogenous or exogenous) and to nonprice model
variables .

Import Price PMс = pwm ·• (1 + tm ) · EXR + PQ · icm c

c'e CT

import import tariff exchange rate cost oftrade CЄ CM (1)

= adjust- (LCUper + inputs per
price price
(LCU) (FCU) ment FCU) import unit

18
 19

where
CEC = a set of commodities (also referred to as c ' and C ') ,
CЄ CM (CC) = a set of imported commodities,
с ∈СТ (с С) = a set of domestic trade inputs (distribution commodi-
ties) ,
PM = import price in LCU (local- currency units) including
transaction costs ,
pwmc = c.i.f. import price in FCU (foreign- currency units) ,
tmc = import tariff rate ,
EXR = exchange rate (LCU per FCU) ,

||||
=
PW composite commodity price (including sales tax and
transaction costs) , and
icmec = quantity of commodity c' as trade input per imported
unit of c .

The import price in LCU (local-currency units) is the price paid by do-
mestic users for imported commodities (exclusive of the sales tax) . Equa-
tion ( 1 ) states that it is a transformation of the world price of these im-
ports , considering the exchange rate and import tariffs plus transaction
costs (the cost of trade inputs needed to move the commodity from the
border to the demander) per unit of the import. For all commodities, the
market price paid by domestic commodity demanders is the composite
price, PQ ; in this equation , PQ applies only to payments for trade inputs .
The domain of the equation is the set of imported commodities (a subset
of the commodity set) . The model includes one equation like (1 ) for every
imported commodity.
Note that the notational principles make it possible to distinguish be-
tween variables (upper-case Latin letters) and parameters (lower- case
Latin letters) . This means that the exchange rate and the domestic import
price are flexible, while the tariff rate and the world import price are fixed .
The fixedness of the world import price stems from the " small- country"
assumption . That is , for all its imports, the assumed share of world trade
for the modeled country is so small that it faces an infinitely elastic sup-
ply curve at the prevailing world price.

Export Price PE = pwe (1 - te ) EXR Σ PQoice c

c'e CT

export export tariff exchange rate costoftrade

CECE (2)
price = price adjust- (LCU per inputs per
(LCU) (FCU) ment FCU) exportunit

where
CECE (CC) = a set of exported commodities (with domestic produc-
tion) ,
PE = export price (LCU) ,
pwee = f.o.b. export price (FCU) ,
te = export tax rate,
ice c = quantity of commodity c ' as trade input per exported
unit of c .
2020

The export price in LCU is the price received by domestic producers

when they sell their output in export markets. This equation is similar in
structure to the import price definition . The main difference is that the
tax and the cost of trade inputs reduce the price received by the domestic
producers of exports (instead of adding to the price paid by domestic de-
manders of imports) . The domain of the equation is the set of exported
commodities , all of which are produced domestically.19

Demand Price of PDD = PDS

PDS PQ · icdc' c
Domestic Non c'e CT
traded Goods
domestic cost oftrade CECD (3)
domestic
demand = supply inputs per
unit of
price price domestic sales

where
CECD (CC) = a set of commodities with domestic sales of domestic

output,
PDD = demand price for commodity produced and sold domes-
tically,
PDS = supply price for commodity produced and sold domesti-
cally, and
icdec = quantity of commodity c ' as trade input per unit of c
produced and sold domestically.

The model includes distinct prices for domestic output that is used do-
mestically. In the presence of transaction costs, it is necessary to distin-
guish between prices paid by demanders and those received by suppliers.
Equation (3 ) defines the demand prices as the supply price plus the cost of
trade inputs per unit of domestic sales of the commodity in question.

Absorption PQ ·• (1 - tqc) · QQc = PDD · QD + PM QM

absorption
domestic demand price import price CE (CDCM) (4)
(at demand times + times
prices net of domestic sales quantity import quantity
sales tax)

where
QQc = quantity of goods supplied to domestic market (com-
posite supply) ,
QDc = quantity sold domestically of domestic output,
QM = quantity of imports of commodity, and
tqe = rate of sales tax (as share of composite price inclusive
of sales tax) .

19The model does not include any commodities that are imported for immediate re-
export. As long as such trade uses domestic factors (and, possibly, intermediate in-
puts) , it can be handled without any changes in model structure by including an ac-
tivity in the SAM that imports a nonproduced commodity and exports all of its
output.
 21

Absorption is total domestic spending on a commodity at domestic de-

mander prices . Equation (4 ) defines it exclusive of the sales tax. Absorp-
tion is expressed as the sum of spending on domestic output and imports
at the demand prices, PDD and PM. The prices PDD and PM include the
cost of trade inputs but exclude the commodity sales tax (compare with
equations 1 and 3) .
The equation as a whole applies to all commodities that are imported
and/or have domestic sales of domestic output (the union of the sets CD
and CM) . It does not apply to commodities for which the entire output vol-
ume is exported . Each of the two terms on the right-hand side applies only
to its relevant set (CD and CM, respectively) . In the GAMS code, PM and
QM are fixed at zero for commodities that are not elements in the set CM;
similarly PDD and QD are fixed at zero for commodities that are not ele-
ments in the set CD. This approach is followed throughout : all variables
that should be excluded from the model are fixed at zero . The equation
would be transformed into an explicit definition of absorption at market
prices or of the composite price (the price paid by domestic demanders , in-
clusive of the sales tax) if it were divided by (1-tq) or ( 1-tq) .QQ.

Marketed Output PX QX = PDS · QD + PE ·QE

Value
producer price domestic supply price export price
times marketed = times + times CECX (5)
output quantity domestic sales quantity export quantity

where
PX = aggregate producer price for commodity,
QXc = aggregate marketed quantity of domestic output of
commodity,
QEc = quantity of exports, and
CECX (CC) = a set of commodities with domestic output .

For each domestically produced commodity, the marketed output value

at producer prices is stated as the sum of the values of domestic sales and
exports.2º Domestic sales and exports are valued at the prices received by
the suppliers, PDS and PE , both of which have been adjusted downwards
to account for the cost of trade inputs (compare with equations 2 and 3) .
The domain limitation to domestically produced commodities (the ele-
ments in the set CX) has to be stated explicitly given that the model in-
cludes a category of imported commodities without domestic production .
The domestic part applies only to elements in CD whereas the export part
applies only to elements in CE. In the GAMS code, the variables PE and
QE are fixed at zero for commodities that are not elements in the set CE.
PX and QX are referred to as " aggregate" values since they may apply to
an aggregation of output from different domestic producers of the same
commodity. By dividing through by QX, this equation could be rewritten
as an explicit definition of PX.

20This value excludes the value of home-consumed output .

 22
22

Activity Price PA = ΣPXACac ac

се с

activity = producer prices αεΑ (6)

price times yields

where
αεΑ = a set of activities ,
PA = activity price (gross revenue per activity unit) ,
РХАСа с = producer price of commodity c for activity a , and
Ѳа с = yield of output c per unit of activity a.

The gross revenue per activity unit, the activity price, is the return
from selling the output or outputs of the activity, defined as yields per ac-
tivity unit multiplied by activity- specific commodity prices, summed over
all commodities. This allows for the fact that activities may produce mul-
tiple commodities .

Aggregate PINTA = PQ · ica a

Intermediate Input се с
аєА (7)
Price
aggregate intermediate input cost
intermediate per unit ofaggregate
input price intermediate input

where
PINTA = aggregate intermediate input price for activity a , and
icac a = quantity of c per unit of aggregate intermediate input
α.

The activity-specific aggregate intermediate input price shows the cost

of disaggregated intermediate inputs per unit of aggregate intermediate
input . It depends on composite commodity prices and intermediate input
coefficients , which show the quantity of input commodity c per unit of ag-
gregate intermediate input (not per unit of output).

Activity Revenue PA (1 - ta ) · QAa = PVA QVA + PINTA。 · QINTA

and Costs
value-added aggregate
activity price intermediate
= price times αεΑ (8)
(net oftaxes) input price times
times activity level quantity
quantity

where
taa = tax rate for activity,
QA = quantity (level) of activity,
QVAa = quantity of (aggregate) value-added ,
||||||

QINTA = quantity of aggregate intermediate input , and

PVA = price of (aggregate) value-added .
 23

For each activity, total revenue net of taxes is fully exhausted by pay-
ments for value-added and intermediate inputs. Given the above defini-
tions of PA and PINTA, equation ( 8 ) implicitly defines the value-added
price, PVA.

Consumer Price CPI = PQ cwtsc

Index сес (9)
consumer = prices times
price index weights

where
cwts = weight of commodity c in the consumer price index, and
CPI consumer price index (exogenous variable) .

Producer Price DPI = PDS - dwts

Index for Nontraded CEC
(10)
Market Output producerprice index = prices times
fornon-traded outputs weights

where
dwtse = weight of commodity c in the producer price index, and
DPI = producer price index for domestically marketed output .

Equations (9) and ( 10) define the consumer price index and the pro-
ducer price index for domestically marketed output. The CPI is fixed and
functions as the numéraire in the basic model version ; alternatively, the
DPI may be fixed . A numéraire is required since the model is homoge-
neous of degree zero in prices-a doubling of the value of the numéraire
would double all prices but leave all real quantities unchanged . All simu-
lated price and income changes should be interpreted as changes vis -à-vis
the numéraire price index.

The production and trade block covers four categories: domestic produc-
PRODUCTION AND tion and input use; the allocation of domestic output to home consump-
TRADE BLOCK tion , the domestic market , and exports ; the aggregation of supply to the
domestic market (from imports and domestic output sold domestically ) ;
and the definition of the demand for trade inputs that is generated by the
distribution process .

Production is carried out by activities that are assumed to maximize

profits subject to their technology, taking prices (for their outputs, inter-
mediate inputs, and factors) as given . In other words, it acts in a perfectly
competitive setting. The CGE model includes the first-order conditions for
profit-maximization by producers . As noted in the preceding section (see
Figure 1 ) , two alternative specifications are permitted at the top level of
the technology nest : the activity level is either a CES or a Leontief
24

function of the quantities of value-added and aggregate intermediate

21
input use.

1
α
-Pa pa
CES Technology : · ·
Q4, = œ“ · (8 a%; · QVA¸¯
а "² + (1–82 ) · QINTA
,~ P " ) »
Activity Production a = ACES ( 11 )
Fucntion activity quantity ofaggregate value added,
level
my] = CES[ quantity of aggregate intermediate input

1
QVAa PINTAa δα
a 1+pa
CES Technology:
Value-Added- QINTA PVA 1-8a
a
Intermediate - Input
Ratio value-added
intermediate-input : a = ACES ( 12)
intermediate- =
f value-added
input quantity price ratio
ratio

where
a ЄACES(CA) = a set of activities with a CES function at the top of the
technology nest ,
aa = efficiency parameter in the CES activity function,
δα = CES activity function share parameter, and
pa
α = CES activity function exponent .

The user specifies the activities , if any, that belong to the set ACES. If
not in ACES, an activity belongs to the set ALEO, which is introduced
below. Activities in ACES have CES technology at the top level of the
technology nest . In other words, the activity level is a CES function of
value- added and aggregate intermediate input use (equation 11) . The op-
timal mix of intermediate inputs and value-added is a function of the rel-
ative prices of value -added and the aggregate intermediate input
(equation 12 ) .22 Below, in equation ( 18) , the activity level determines the
quantity of commodity outputs produced by each activity. The exponent ,

21For the alternative with CES technology at the top of the technology nest, the
profit-maximization problem, which applies to each relevant activity, a, is as follows :

Σ PO · OINTас - WF, WFDIST . OF,f a subject to equations

maximize PAa · (1 –-– ta¸ ) · QA - сес feF

( 11) , (15) , and (17); equations ( 7) , (8) , ( 11 ) , ( 12) , ( 15 ) , ( 16) , and (17) are the first-
order conditions. For the alternative with Leontief technology at the top , the profit-
maximization problem includes equations ( 13) and (14) among its constraints but
excludes equation 11 ; the Leontief first-order conditions include equations ( 13) and
(14) instead of equations (11) and (12). In both optimization problems, all quantities
are variables whereas the decisionmakers view all other items as parameters or ex-
ogenous variables (including all prices and wages).
221
In general, when writing nonlinear equations to be solved numerically, it is good
practice to avoid division by a variable that the solver treats as possibly going to zero.
Accordingly, in the GAMS version of equation ( 12) , QINTA was moved to the right-
hand side . Parallel adjustments were made in equations (22) and (25) .
 25

pa, is a transformation of the elasticity of substitution between value-

added and the aggregate intermediate input : the higher this elasticity, the
smaller the value of på and the larger the optimal change in the ratios be-
tween the quantities of value-added and the intermediate input aggregate
in response to changes in their relative prices.23

Leontief QVA₁ = ivaa QAa

Technology: a Є ALEO ( 13 )
Demand for demandfor = activity
value added f level
Aggregate Value-
Added

Leontief
QINTA = intaa QAa a = ALEO ( 14 )
Technology:
Demand for demandfor aggregate activity
Aggregate
intermediate input ·] = [ level
Intermediate Input
where
a ЄALEO(CA) = a set of activities with a Leontief function at the top of
the technology nest,
ivaa = quantity of value- added per activity unit, and
intaa = quantity of aggregate intermediate input per activity
unit.

For the alternative model version with a Leontief function at the top
of the technology nest , equations ( 11) and ( 12) are replaced by equations
( 13) and ( 14 ) where the demands for value- added and the aggregate inter-
mediate inputs are defined as Leontief functions of the activity level. Each
activity is an element in either ACES or ALEO, but not both .

1
va
να
-Pa a ЄA (15)
Value-Added and QVA = avaΣ8ya
a 8fa QFfa J
Factor Demands feF

quantity of aggregate = CES factor

value added inputs

va va
Factor Demand WF, WFDIST
f a = PVAа ( 1 − tva ) · QVAΣ8f a · QF Pa
ƒЄF'
va αε Α
ναa ·QFƒ a"Pa"-1 (16 )
•Έ
87fa fa fe F

marginal cost of = marginal revenueproduct

factorfin activity a offactorf in activity a

1
23For CES functions, σ = 2 where is the elasticity of substitution and the ex-
1 +ρ
ponent.
26

where
f =F(= F') = a set of factors,
tvaa = rate of value-added tax for activity a,
ava = efficiency parameter in the CES value-added function ,
Seca
α = CES value-added function
share parameter for factor f
in activity a,
QFfa = quantity demanded of factor f from activity a,
να = CES value-added function exponent,
pva
WFf = average price of factor, and
WFDIS fa = wage distortion factor for factor f in activity a (exoge-
T
nous variable) .
Equation ( 15) states that , for each activity, the quantity of value-added
is a CES function of disaggregated factor quantities . According to equa-
tion (16) , activities demand factors at the point where the marginal cost of
each factor (defined on the left-hand side as the activity- specific factor
price) is equal to the marginal revenue product (net of intermediate input
costs) of the factor. In the GAMS code , the domain of equation (16) is lim-
ited to the factor-activity combinations that appear in the base-year SAM .
Similar domain restrictions apply to other equations that are defined over
mappings between multiple indices (for example, equation 17) . The expo-
να
nent, pa, is a transformation of the elasticity of factor substitution : the
να
higher this elasticity, the smaller the value of pa and the larger the opti-
mal change in the ratios between different factor quantities in response to
changes in relative factor prices (compare with footnote 16) .
The fact that the average factor price is an endogenous variable while
the activity-specific “wage-distortion " factor is exogenous reflects the
treatment of factor markets in the basic model version (see equation 39
below).

Disaggregated QINT a = icae a · QINTA αεΑ

(17)
Intermediate Input се с
Demand intermediate demand aggregate intermediate
forcommodity c input quantity
from activitya for activity a

where
QINTca = quantity of commodity c as intermediate input to activ-
ity a.

For each activity, the demand for disaggregated intermediate inputs is

determined via a standard Leontief formulation as the level of aggregate
intermediate input use times a fixed intermediate input coefficient .

Commodity QXACac
ас + QHAach = 0acс Q¹a
Production and hЄH
Allocation household home αεΑ
marketed quantity production (18)
consumption = of commodityc се СХ
of commodity c
of commodity c
from activity a from activitya
from activitya
 27

where
QXACac = marketed output quantity of commodity c from activity
a, and
QHAach = quantity of household home consumption of commodity
c from activity a for household h .

On the right-hand side, production quantities , disaggregated by activ-

ity, are defined as yields times activity levels . On the left- hand side , these
quantities are allocated to market sales and home consumption. Note that
this equation permits (i) any commodity to be produced by one or more ac-
24
tivities and (ii) any activity to produce one or more commodities.²

1
ас paасc-1
-pac
Output Aggregation QX =0ac8ac
Σδα · QXACac
Function аєА

CE CX ( 19)
aggregate activity-specific
marketed == CES marketed
production of production of
commodity c commodity c

where
aac = shift parameter for domestic commodity aggregation
function ,
бас
ac = share parameter for domestic commodity aggregation
function, and
ac
ρε = domestic commodity aggregation function exponent .

First-Order PXAC₁a c = PXQX ас QXACP

8c ac • Sac · QXAC-ас
P
Condition for Output аЄA'
Aggregation αεΑ
(20)
Function marginal cost ofcom- = marginal revenueproduct of се СХ
moditycfrom activity a commodity cfrom activity a

Aggregate marketed production of any commodity is defined as a CES

aggregate of the marketed output levels of the different activities produc-
ing the commodity (equation 19) . The optimal quantity of the commodity
from each activity source is inversely related to the activity- specific price
(equation 20) . QX appears as the output, sold at the price , PX, and pro-
duced with the inputs, QXAC, that are purchased at the prices , PXAC.
More specifically, the choice between commodities from different
sources is cast as an optimization problem . Equations ( 19) and (20) are the

24In the SAM , home consumption is represented by payments from households to ac-
tivities. For the case of home consumption out of activities with multiple outputs, it
is necessary to complement the information in the SAM with data on the allocation
of consumption across the different activity outputs.
28

first-order conditions for maximizing profits from selling the aggregate

output, QX, at the price, PX, subject to the aggregation function and the
disaggregated commodity prices, PXAC. A decline in the price, PXAC, of
one activity relative to others would shift demand in its favor without to-
tally eliminating demand for other, higher-price sources . The degree of
substitutability between different producers depends on the value of på ,
which is a transformation of the elasticity of substitution (compare with
footnote 16) . Its values, and those of the elasticity, are restricted to assure
that the corresponding isoquant is convex to the origin . In terms of pro-
duction economics, this is equivalent to a diminishing technical rate of
substitution .
It should be noted that, for the case where there is a single producer
ас
of a given commodity, the value of the share parameter, ασ would be
Sac
unity and, as a result, QXAC = QX and PXAC = PX, irrespective of the
values for the elasticity and the exponent .

1
—
Output QX = α · (8 · QEº¹ + (1 − 8′) · QDº¹ ) ž
Transformation
(CET) Function CE (CE CD) (21)
aggregate marketed == CET export quantity, domestic
domestic output sales of domestic output

where
at = a CET function shift parameter,
St = a CET function share parameter, and
= a CET function exponent .
ρ

Equations (21 ) and (22) address the allocation of marketed domestic

output, defined in equation ( 19) , to two alternative destinations : domestic
sales and exports . Equation (21 ) reflects the assumption of imperfect
transformability between these two destinations . The CET function ,
which applies to commodities that are both exported and sold domesti-
cally, is identical to a CES function except for negative elasticities of sub-
stitution. The elasticity of transformation between the two destinations is
a transformation of på , for which the lower limit is one . The values are re-
stricted to assure that the isoquant corresponding to the output transfor-
25
mation function is concave to the origin.²

1
ОЕ PE 1-8- 1
C
Export- Domestic
Supply Ratio QD PDS c St
CE (CE CD) (22)
export-domestic = export-domestic
supply ratio f price ratio

1
25For CET functions, N = where is the elasticity of transformation and p the
1 + p
exponent. As Q varies from zero to infinity, the value of pέс varies from infinity to one .
In equation (22) , as ptc approaches one from above, the elasticity of the QE-QD ratio
with respect to changes in the PE-PDD ratio increases.
 29

Equation (22 ) defines the optimal mix between exports and domestic
sales . Equations ( 5) , ( 21 ) , and (22) constitute the first-order conditions for
maximization of producer revenues given the two prices and subject to the
CET function and a fixed quantity of domestic output . Note that equation
(22) assures that an increase in the export- domestic price ratio generates
an increase in the export-domestic supply ratio (that is, a shift toward the
destination that offers the higher return).

Output QX = QD + QE
Transformation for cЄ (CD CEN) ~ (CE ~ CDN) (23)
Domestically Sold
domestic market
Outputs Without aggregate
marketed = sales of domestic + exports [for
Exports and for domestic output output [for CE (CE CDN)]
Exports Without cЄ (CD CEN)]

Domestic Sales

where
CЄCEN (CC) = non-exported commodities (complement of CE) , and
CE CDN (CC) commodities without domestic market sales of domes-
tic output (complement of CD ) .

This equation replaces the CET function for domestically produced

commodities that do not have both exports and domestic sales . It allocates
the entire output volume to one of these two destinations .

1
-pa
Composite Supply 20¸ = œ¶ · (8ª · QM²²² + (1−8ª ) · QD´? ² ) På
CE (CM CD) (24)
(Armington)
Function composite import quantity, domestic
-
f use of domestic output
supply

where
а = an Armington function shift parameter,
Sa = an Armington function share parameter, and
Pe = an Armington function exponent .

Imperfect substitutability between imports and domestic output sold

domestically is captured by a CES aggregation function in which the com-
posite commodity that is supplied domestically is "produced" by domestic
and imported commodities entering this function as "inputs . " When the
domain of this function is limited to commodities that are both imported
and produced domestically, it is often called an "Armington" function,
named after the originator of the idea of using a CES function for this pur-
pose. The elasticity of substitution between commodities from these two
sources is a transformation of for which the lower limit is minus one
(compare with footnote 16) .

1
Import-Domestic QM PDD ба 1+pa
CE (CM CD) (25)
Demand Ratio QDc PMc 1-89

import-domestic domestic-import
demand ratio price ratio
] =/ [
30

Equation (25) defines the optimal mix between imports and domestic
output. Its domain is thus limited to imports with domestic production .
Note that the equation assures that an increase in the domestic-import
price ratio generates an increase in the import- domestic demand ratio
(that is, a shift away from the source that becomes more expensive) .26 To-
gether, equations ( 4 ) , ( 24) , and (25) constitute the first- order conditions
for cost-minimization given the two prices and subject to the Armington
function and a fixed quantity of the composite commodity.

Composite Supply QQ = QD + QM
for Non-imported
CE (CD CMN) (CM CDN) (26)
Outputs and Non-
produced Imports domestic use of
composite = marketed domestic imports [for
supply output [for CE (CMCDN)]
CE (CD CMN)]

where
CE CMN (CC) = a set of non-imported commodities.

The Armington function is replaced by equation (26) for the union of

commodities that have either imports or domestic sales of domestic output
but not both . For any commodity in this category, it imposes equality be-
tween "composite supply" and one of the variables on the right-hand side.

=
Demand for QT Σ (icm QM + ice c' · QE + icd c' · QDc ')
Transactions c'e C'
Services demandfor sum ofdemands с∈СТ (27)
transactions for imports, exports,
services and domestic sales

where
QTc = quantity of commodity demanded as transactions serv-
ice input .

Total demand for trade inputs is the sum of the demands for these in-
puts that are generated by imports (from moving commodities from the
border to domestic demanders), exports (from moving commodities from
domestic producers to the border) , and domestic market sales (from mov-
ing commodities from domestic producers to domestic demanders) . In all
three cases, fixed quantities of one or more transactions service inputs are
required per unit of the traded commodity.

26See footnote 16 for the definition of the elasticity of substitution . In equation (25) ,
as the value of på approaches minus one from above, the elasticity of the import-
domestic demand ratio with respect to changes in the PDD-PM ratio increases .
 31

INSTITUTION
BLOCK

Factor Income YF, = WF, WFDIST

f a · QFƒ a
аєА
fЄF (28)
sum ofactivitypayments
income of
(activity-specific wages
factorf
times employment levels)

where
YFf = income of factor f.

Institutional Factor if = shif¡ · · [ (1 - tf ) . YF, – trnsfrowf · EXR

YIF,,
Incomes
income of share ofincome income offactorf i Є INSD
institution i = offactorfto (29)
(net oftax and feF
fromfactorf institution i transferto RoW)

where
i Є INS = a set of institutions (domestic and rest of the world) ,
i =INSD(ĊINS) = a set of domestic institutions,
YIF i f = income to domestic institution i from factor f,
shifif = share of domestic institution i in income of factor f,
tff = direct tax rate for factor f, and
trnsfri f = transfer from factor f to institution i .

Equation ( 28 ) defines the total income of each factor. In equation (29) ,

this income is split among domestic institutions in fixed shares after pay-
ment of direct factor taxes and transfers to the rest of the world.27 The lat-
ter are fixed in foreign currency and transformed into domestic currency
by multiplying by the exchange rate. This equation makes reference to the
set of domestic institutions (households , enterprises, and the govern-
ment) , a subset of the set of institutions, which also includes the rest of
world.

Income of domestic, YIi CPI + trnsfr; row · EXR

f+
YI₁ = Σ YIF;ƒ Σ TRII + trnsfr; gov
Nongovernment fe F i'e INSDNG'
Institutions transfers
transfers transfers
income of factor from other domestic i Є INSDNG (30)
institution i income from from
non-government ROW
institutions government

where
i ЄINSDNG(=INSDNG' - INSD)
= a set of domestic nongovernment institutions,
YI = income of institution i (in the set INSDNG ) , and
TRII = transfers from institution i' to i (both in the set INS-
DNG).

Σ shift.f = 1 .
27To assure that the total factor income is distributed, it is necessary thatieINSD
32

Domestic nongovernment institutions form a subset of the set of do-

mestic institutions . The total income of any domestic nongovernment in-
stitution is the sum of factor incomes ( defined in equation 29) , transfers
from other domestic nongovernment institutions (defined below in equa-
tion 31 ) , transfers from the government (indexed to the CPI) , and trans-
fers from the rest of the world.28

Infra-Institutional TRIIshii ¿' (1 - MPS ) (1- TINS ) . YI

Transfers i Є INSDNG
share ofnet income income ofinstitution (31)
transferfrom = i'ЄINSDNG'
institution i'to i ofinstitution i' i', net ofsavings and
transfered to i directtaxes

where
shiiii = share of net income of i ' to i (i’ĒINSDNG'; iĒINS-
DNG),
MPS; = marginal propensity to save for domestic nongovern-
ment institution (exogenous variable), and
TINS; = direct tax rate for institution i (i ЄINSDNG) .

Transfers between domestic nongovernment institutions are paid as

fixed shares of the total institutional incomes net of direct taxes and sav-
ings. The values of MPS and TINS are defined in separate equations, dis-
cussed below.

Household EH shii¡ n
Σ -MPS ) (1 - TINS ) ·YI
Consumption iЄ INSDNG
hЄH (32)
Expenditures household income household income, net ofdirect
disposablefor taxes, savings, and transfers to
consumption other non-government institutions

where
i e H(CINSDNG)
= a set of households , and
EH = household consumption expenditures .

Among the domestic nongovernment institutions, only households de-

mand commodities . In equation (32) , the total value of consumption
spending is defined as the income that remains after direct taxes, savings,
and transfers to other domestic nongovernment institutions.

28The fact that government transfers are indexed to the CPI makes the model homo-
geneous of degree zero in prices. This indexing influences the results when the DPI is
the model numéraire. If the CPI is the numéraire, it has no effect.
 33

Household m
PQc · · QHch = PQc •· Y" h + Beth EH - PQ c'h - ΣΣ
Yon ΣPXACac Yach
Consumption c'e C аЄAc'εC
Spending on household consumption total household consumption CEC
Marketed spending on market f spending, marketprice ofc, and other (33)
hЄH
Commodities commodity c commodityprices (market and home)

where
QHch = quantity of consumption of marketed commodity c for
household h ,
m
Y'ch = subsistence consumption of marketed commodity c for
household h ,
yhac
h = subsistence consumption of home commodity c from ac-
tivity a for household h , and
m
Bc = marginal share of consumption spending on marketed
h
commodity c for household h.

Household PXACac · QHAach = PXACac Yach + Bach

Consumption
Spending on EH - PQ Yoh - ΣPXACacc' Yach αεΑ
Home Commodities c'e C aЄ Аc'ε C
CEC (34)
household consumption total household consumption spending, hЄH
spending on home commodity f producerprice, and other
cfrom activity a commodityprices (market and home)

where
Bhch = marginal share of consumption spending on home com-
modity c from activity a for household h.

It is assumed that each household maximizes a “Stone-Geary” utility

function subject to a consumption expenditure constraint. The resulting
first-order conditions , equations (33) and ( 34) , are referred to as LES (lin-
ear expenditure system) functions since spending on individual commodi-
ties is a linear function of total consumption spending, EH. Two functions
are needed since household consumption is for two types of commodities:
(i) consumption of marketed commodities (purchased at market prices;
equation 33 ) and (ii) consumption of home production (valued at their op-
portunity cost, the activity- specific producer price not including market-
ing costs; equation 34) . Explicit demand functions may be derived by di-
viding both sides of each equation by the relevant price.

Investment Demand QINV = IADJ · qinv

39
CEC (35)
fixed investment adjustment factor
times
demand for
commodity c base-year fixed
investment
34

where
QINV = quantity of fixed investment demand for commodity,

IADJ = investment adjustment factor (exogenous variable) ,

and
qinve = base-year quantity of fixed investment demand.

Fixed investment demand is defined as the base-year quantity multi-

plied by an adjustment factor. For the basic model version , the adjustment
factor is exogenous, in effect also making the investment quantity
exogenous. Inventory investment is also included in the model, but is
treated as an exogenous demand ( see equation 40 below) .

Government QG = GADJ ·• 98c

Consumption
government adjustment factor
Demand consumption times
demand for base-year government CEC (36)
commodity c consumption

where
QGc = government consumption demand for
commodity,
GADJ = government consumption adjustment factor
(exogenous variable) , and

98c = base-year quantity of government demand .

Similarly, government consumption demand, in which the main com-

ponent tends to be the services provided by the government labor force, is
also defined as the base-year quantity multiplied by an adjustment factor.
This factor is also exogenous and, hence , the quantity of government con-
sumption is fixed .

Government YG = · ·
ΣTINS · YI + Σff YF + Σ ινα PVA · QVA
Σtva
Revenue iЄ INSDNG fe F αε Α
.
+ Σtaa · PAa ' QA + Σtm pwmс QM с EXR + Σte pwe · QE · EXR
αε Α се СМ CE CE

+ Σ14с PO 00, + Σ YIFgovf + trnsfrgov row · EXR

CE C feF

direct taxes directtaxes value-

government = + added
revenue from + from
institutions
-CCHCHO factors ] tax
(37)
activity import export
tax tariffs taxes

sales transfers
+ factor + from
tax income
ROW

where
YG = government revenue.
 35

Total government revenue is the sum of revenues from taxes , factors ,

and transfers from the rest of the world.

Government EG = Σ PQ • QG + Σ trnsfr gov CPI

се с iЄ INSDNG
Expenditure
transfers to domestic (38)
government government
non-government
spending"]= [consumption institutions

where
EG = government expenditures.

Total government spending is the sum of government spending on

consumption and transfers.

SYSTEM
CONSTRAINT
BLOCK

Factor Markets QFa = QFSƒ

Σ fa
ає А fЄF (39)

demand for supply of

factor f factor f

where

QFS = quantity supplied of factor (exogenous variable) .

Equation (39) imposes equality between the total quantity demanded

and the total quantity supplied for each factor. In the basic model version ,
all demand variables are flexible while the supply variable is fixed . The
factor wage, WF , is the equilibrating variable that assures that this equa-
tion is satisfied : an increase in WF, raises the wage paid by each activity,
WF WFDISTfa which is inversely related to the quantities of factor de-
mand, QF . All factors are mobile between the demanding activities .
Two other factor-market closures are programmed in the GAMS ver-
sion . To specify the case with unemployment at a given wage for a factor,
the supply variable for the factor is unfixed (QFS,) while its economywide
wage is fixed (WF ) . The model remains square (one endogenous variable
is added but another is removed) . Each activity is free to employ the quan-
tity it desires (QFfa) at a fixed wage (WF · WFDISTƒ «) . The free supply
variable, QFS,, records the total employment level .
Alternatively, to specify the case of a fully segmented factor market
with fixed factor demands (for example, short-run fixity of nonagricultural
capital use) , the variables for factor demand and the economywide wage
are fixed (written QFra
fa and WF ) while the variables for supply and wage
distortions are unfixed (written QFS, and WFDIST ) . The model again
remains square—that is, the economywide wage variable and a set of ac-
tivity-specific factor-demand variables are fixed while the supply variable
and a set of activity- specific wage-distortion variables are unfixed .
36

Activity-specific wages, WF, WFDISTfa , vary to assure that the fixed ac-
tivity-specific employment level, QFfa , is consistent with profit-maxi-
mization (compare with equation 16) . Also for this formulation, the en-
dogenous aggregate factor supply variable merely records the total em-
ployment level .

Composite QQ = ΣQINTea + ΣQHch + QG + QINV¸ + qdst¸ + QT¸

Commodity Markets αε Α he H

CEC (40)
composite = intermediate household government
+
supply use consumption consumption

+ fixed stock trade

investment]+[change
•]+ [inputuse

where

qdste = quantity of stock change.

Equation (40 ) imposes equality between quantities supplied (from

equations 24 , 25 , and 26) and demanded of the composite commodity. The
demand side includes endogenous terms (from equations 17 , 27, 33, 35,
and 36) and a new exogenous term for stock change. Among the endoge-
nous terms, QG and QINV are fixed in the basic model version (compare
with equations 35 and 36 ) . The composite commodity supply, QQ, drives
demands for domestic marketed output, QD, and imports, QM. The
market-clearing variables are the quantities of import supply, for the im-
port side, and the two interrelated domestic prices, PDD and PDS, for do-
mestic market output .

Current-Account Σ pwm QM + Σtrnsfrrowf = pwe, QE + trnsfr row + FSAV

Balance for the се СМ fe F CECE iЄ INSD

Rest of the World, in factor institutional (41)

import export foreign
+ transfers revenue transfers
Foreign Currency spending to RoW savings
from RoW

where
FSAV = foreign savings (FCU) (exogenous variable) .

The current-account balance, which is expressed in foreign currency,

imposes equality between the country's spending and its earning of for-
eign exchange. For the basic model version, foreign savings is fixed ; the
(real) exchange rate (EXR) serves the role of equilibrating variable to the
current-account balance. The fact that all items except imports and ex-
ports are fixed means that, in effect , the trade deficit also is fixed . Alter-
natively, the exchange rate may be fixed and foreign savings unfixed . In
this case, the trade deficit is free to vary.
 37

Government YG = EG + GSAV
Balance
government government government
(42)
revenue ] =[expenditures savings

where
GSAV = government savings .

The government balance imposes equality between current govern-

ment revenue and the sum of current government expenditures (not in-
cluding government investment) and savings. Savings may be negative.
The alternative mechanisms for maintaining this balance are associated
with equation (43) .

Direct Institutional
TINS₁ = tins; · ( 1 + TINSADJ · tins01 , ) + DTINS · t.
Tax Rates
direct tax base rate adjusted point change i = INSDNG (43)
=
ratefor for scalingfor + for selected
institution i selected institutions institutions

where

TINSi = rate of direct tax on domestic institutions i ,

tinsi = exogenous direct tax rate for domestic institution i ,
TINSADJ = direct tax scaling factor ( = 0 for base; exogenous vari-
able) ,
tins01; = 0-1 parameter with 1 for institutions with potentially
flexed direct tax rates, and
DTINS; = change in domestic institution tax share ( = 0 for base ;
exogenous variable).

Equation (43 ) defines the direct tax rates of domestic nongovernment

institutions. For the basic model version, all variables on the right-hand
side are fixed, in effect fixing the values for the direct tax rate variable for
all institutions . In this setting, government savings is the endogenous
variable that clears the government balance.
In the GAMS implementation of the standard model, two alternative
closure rules are coded for the government balance (see Macroeconomic
Balances in Chapter 3 ) . For both alternatives, government savings is
fixed . In the first case, DTINS is the flexible variable that clears the gov-
ernment balance by scaling the base-year tax rates of each tax-paying in-
stitution. In this setting, the rates will change by a uniform number of
(percentage) points for all institutions with a value of 1 for the parameter
tins01 (that is , for all institutions with potentially flexed direct tax rates) .
Hence, the initial tax rate has no impact on the rate change . In the second
case, TINSADJ becomes a variable while DTINS is fixed . For this closure,
the changes in TINS are relatively large for institutions with relatively
large base-year rates (if they have a value of 1 for tins01 ) .
38

Notice that when GSAV is fixed for the two alternative closure rules ,
another variable is made endogenous, thus maintaining a model with an
equal number of variables and equations . The choice between alternative
closure rules should depend on the empirical context . For example, if the
government pursues a policy of raising effective direct tax rates to main-
tain fixed savings in a setting with reduced other revenues and/or in-
creased government spending, will it raise rates for all or only a subset of
the nongovernment institutions? For the targeted institutions , will the
government aim at uniform point increases or will it raise rates in pro-
portion to current rates?

.
Institutional MPS₁ = mps , · ( 1 + MPSADJ · mps01, ) + DMPS · mps01 ,
Savings Rates
savings base rate adjusted pointchange i ЄINSDNG (44)
ratefor = forscalingfor + for selected
institution i selected institutions institutions

where

mpsi base savings rate for domestic institution i ,

MPSADJ = savings rate scaling factor ( = 0 for base) ,

MPS01; = 0-1 parameter with 1 for institutions with potentially

flexed direct tax rates, and
DMPS = change in domestic institution savings rates ( = 0 for
base; exogenous variable) .
Equation (44) defines the savings rates of domestic nongovernment
institutions . Its structure is the same as that of equation (43) . Whether
one or none of the variables MPSADJ and DMPS is flexible depends on
the closure rule for the Savings-Investment balance . For the basic model
version, DMPS is flexible , permitting MPS to be adjusted by a uniform
rate for selected (one or more) nongovernment institutions .

Savings-Investment ΣMPS (1 - TINS, ) · YI, + GSAV + EXR

ie INSDNG
Balance
· FSAV = PQ QINV + PQ qdst (45)
сес се с

non-govern- + government + foreign = fixed stock

mentsavings savings savings investment change
„ ] +[~

Equation (45 ) states that total savings and total investment have to be
equal . Total savings is the sum of savings from domestic nongovernment
institutions, the government, and the rest of the world, with the last item
converted into domestic currency. Total investment is the sum of the val-
ues of fixed investment (gross fixed capital formation) and stock changes.
 39

In the basic model version, the flexible variable, DMPS, performs the
task of clearing this balance (compare with equation 44) . None of the
other items in the Savings-Investment balance is free to vary to assure
that the balance holds. Given that the balancing role is performed by the
savings side , this closure represents a case of "investment-driven" sav-
ings . In the GAMS code, additional Savings-Investment closures have also
been programmed . Under closure 2 (see Table 3) , DMPS is fixed and
MPSADJ is flexible. For closure 3, in which investment is savings- driven,
IADJ is flexible whereas both MPSADJ and DMPS are fixed .
Up to this point, the model as stated is not square ; the number of
equations exceeds the number of variables by one. However, the model sat-
isfies Walras' law: one equation is functionally dependent on the others
and can be dropped . The Savings-Investment balance or the current-
account balance is commonly eliminated . ) After eliminating one equation ,
the model is square and , in the absence of errors in formulation , a unique
solution typically exists . Instead of dropping one equation , it is also possi-
ble to add one variable to the macroeconomic balance equations . The so-
lution value of this variable should be zero. If not, one or more equations
are not satisfied and a general equilibrium solution has not been found .
This is the approach followed in the GAMS version of the model . A vari-
able called WALRAS is added to the Savings-Investment balance . No
equation is dropped.
After this adjustment, the model presented is complete and self-
contained. In the basic model version, three more equations (and three
new variables that appear in them) are added . The reason for including
these is that they permit the formulation of the “balanced " Savings-In-
vestment closures 4 and 5. We will return to the closure issue later, after
presenting the new equations and their notation .

Total Absorption QHch + ΣΣΣ PXACac QHAach

TABS = Σ Σ ΡΩ, • OH
he H cЄ C aЄA CEC heH

+ Σ PQ • QG + ΣPQ QINV + PQ · qdst

се с CE C се с
(46)

household household
total = government + fixed stock
market home
absorption consumption investment change
consumption consumption

where
TABS = total nominal absorption .

Total absorption is measured as the total value of domestic final de-

mands , which equals GDP at market prices plus imports minus exports .
The new variable , TABS, records this value.
40

Ratio of Investment INVSHR · TABS = Σ PQ QINV + PQ · qdst

се с се С
to Absorption
(47)
investment-
total = fixed stock
absorption +
ratio absorption investment change

where
INVSHR = investment share in nominal absorption .

The right-hand side of this equation defines the total investment value
(compare with equation 45) . On the left-hand side, total absorption is mul-
tiplied by a new free variable, INVSHR. At equilibrium, this variable
measures the ratio between investment and absorption.

Ratio of GOVSHR · TABS = Σ PQ ·QG

Government CEC

Consumption to government (48)

consumption- total government
Absorption absorption
absorption consumption
ratio

where
GOVSHR = government consumption share in nominal absorption
.

This final equation is similar to equation (47) except that investment

is replaced by government consumption . The right-hand side defines the
value of government consumption (compare with equation 38 ) . On the
left-hand side , total absorption is multiplied by a new free variable,
GOVSHR , which measures the ratio between government consumption
and absorption .
The presence of equations (46) , (47) , and (48) and the three new vari-
ables makes it possible to specify Savings-Investment closures 4 and 5 ,
which represent versions of "balanced" macroeconomic adjustment that
may be preferable for model simulations aimed at generating plausible ,
real-world responses to shocks (compare with discussion of Macroeco-
nomic Balances in Chapter 3) . For the investment-driven , Savings—
investment closures 1 and 2, the burden of adjusting to absorption shocks
is assumed in full by household consumption.29 Under closure 3, with sav-
ings-driven investment, the adjustment burden falls on investment.
Savings-Investment closures 4 and 5 in Table 3, which are also pro-
grammed in the GAMS version of the model, impose a balanced adjust-
ment in the aggregate components of absorption . Under both, the shares
of nominal absorption for investment and government consumption
(INVSHR and GOVSHR) are fixed at base levels while the quantity ad-
justment factors for fixed investment demand and government consump-
tion (IADJ and GADJ) are endogenized . The two closures differ as to
whether DMPS or MPSADJ is the flexible variable that generates the
Savings-Investment equilibrium .
 41

Under these two closures, any change in total absorption would, in

nominal terms, be spread evenly across all three components of absorp-
tion; given the shares for investment and government consumption, the
share for household consumption is implicitly defined. Adjustments in the
nongovernment savings value clear the savings- investment balance. The
magnitude of the savings adjustment, which is influenced by changes in
investment and government consumption (for the latter via changes in
government savings), determines the availability of resources for house-
hold consumption.

29However, for simulations with single-period models (like the current model) aimed at
exploring welfare impacts of exogenous shocks, the Savings–Investment closure 1 or 2
is often preferable since the model is unable to capture future welfare changes associ-
ated with current changes in investment (compare with the Macroeconomic Balances
discussion in Chapter 3).
 5. THE STANDARD MODEL IN GAMS
The GAMS input files contained in the CD-ROM that accompanies this
manual include country data files that enable the user to conduct simula-
tions with the standard model using data for a selected country. It is also
straightforward to apply this modeling system to alternative country data
sets generated by the user. This chapter provides a brief guide to the
GAMS files and suggestions on how to use this modeling system. The files
themselves include additional explanatory comments.
Table 5 summarizes the contents of the different files and Figure 3
provides a schematic representation of the structure of the GAMS model
and data files.30 The modeling system is segmented into two main files,
mod.gms and sim.gms. This segmentation corresponds to the two main
steps in a typical CGE modeling project. In the first main file, mod.gms,
the model, which is identical to that detailed in Chapter 4, is set up and
calibrated to a country data set that is read in the form of an “include” file
(<name>.dat). The sample data sets illustrate how data sets should be de-
fined.31 The SAMs may be included directly in the <name>.dat file or be
read into this file using GAMS GDX file command that comes with recent
GAMS release (or via a link to a spreadsheet using the program XLLINK,
which has to be installed separately for older releases of GAMS. The for-
mer approach—direct inclusion of the SAM and the rest of the data—is
often preferable because it is less error-prone, and it facilitates model doc-
umentation and transportability between different users and computers.
If the account imbalances in the SAM exceed a low cutoff point, a simple
SAM balancing program in the file sambal.inc is activated. The file
varinit.inc is used to initialize all variables at base levels. In the optional

30The CD-ROM also includes an example file for a SAM aggregation program (sam-
agg.gms). It may be used independently of the other GAMS files, in which case the ap-
propriately aggregated SAM should be inserted in the country data file. Alternatively,
after some adjustments, samagg.gms may be used as an include file in the country data
set, immediately before the inclusion of sambal.inc. If so, the set AC in the country
data file should be expanded to include all SAM accounts (both of the initial SAM and
the aggregated SAM). The rest of the sets should be defined on the basis of the ac-
counts in the aggregated SAM.
31Three sample data sets are included: test.dat, which is based on data from Mozam-
bique, and is designed to test many of the model features; swazilan.dat, which includes
macrodata for Swaziland that has only one element in each account set; and zim-
babwe.dat, a Zimbabwe data set. When applied to test.dat or swazilan.dat, the stan-
dard model can be solved using the student version of GAMS; the Zimbabwe data set
is larger and requires a full version of GAMS (including solvers for NLP and MCP
problems). Data sets for a number of other countries are also available.

42
 43

Table 5–File structure in GAMS standard CGE modeling system

File name Description
mod.gms All items (sets, parameters, variables) that appear in the stan-
dard model equations as well as the equations themselves and
the CGE model are declared. Except for the sets, these items are
also defined. The model is solved for the base.
<name>.dat Include files for mod.gms with country-specific data sets
(named after the country they represent), one of which should
be included. The data consists of set elements (used to define
model sets), a SAM, elasticities, selected physical factor quanti-
ties, commodity value shares for home consumption (if needed),
and a parameter transforming SAM tax data.
sambal.inc Include file for <name>.dat. A simple program that balances
the SAM if its account imbalances exceed a cut-off point.
varinit.inc Include file for mod.gms (and, optionally, for sim.gms). All
model variables are initialized.
varlow.inc Optional include file for mod.gms. Imposes lower limits on se-
lected model variables.
repbase.inc Include file for mod.gms. Using data from the base solution, de-
fines an economic structure table, a GDP table, and a macro-
SAM.
sim.gms Restarted from mod.gms. The file includes
(a) declarations and definitions of sets for simulations, experi-
ment parameters, closures for macrosystem constraints, and
closures for factor markets;
(b) a loop over the set of current simulations that contains def-
initions of simulation-specific parameters and variables, a
solve statement, and an include file defining report parame-
ters;
(c) preparation and processing of report parameters (in include
files), checks for errors in report parameters, and a display
of report parameters.
repsetup.inc Include file for sim.gms that includes
(a) declarations and definitions for sets used in reports; and
(b) declarations of report parameters.
reploop.inc Include file for sim.gms. For each simulation, the file defines re-
port parameters for
(a) the levels of each model variable;a
(b) the value of parameters that are subject to change in simu-
lations;
(c) the incomes and expenditures of each SAM account;
(d) national accounts data;
(e) macro- and factor-market closure;
(f) consistency checks for data in (c) and (d).
repperc.inc Include file for sim.gms. For all relevant parameters under (a)
through (d) in reploop.inc, computation of percentage change
from base for nonbase simulations.b
repsum.inc Include file for sim.gms. Summary results tables based on
report parameters defined in reploop.inc and repperc.inc.

aThese parameters have the same name as the corresponding variable with X
added at the end.
bThese parameters have the same name as the corresponding parameter in re-
ploop.inc with P added at the end.
44

file varlow.inc, lower limits close to zero are imposed for selected variables
as this may improve solver performance.
Two models are defined inside mod.gms, one for MCP (mixed-
complementarity programming) and one for NLP (nonlinear program-
ming) solvers.32 The MCP model is identical to the model presented above.
The NLP model differs in that it also includes an objective function. The
objective function is needed given that this is an optimization problem,
but it has no influence on the solution since there is only one feasible so-
lution that satisfies all constraints. After having solved the model for the
base, the program calls up the file repbase.inc, which generates a report on
the base solution.
In sim.gms, which restarts from the save files of mod.gms, simulations
are defined and carried out.33 A note at the beginning of the file specifies
the steps required when additional simulations are introduced. For each
simulation, the user can choose between alternative closures for macro-
economic constraints (compare with Table 3) and factor markets (three al-
ternatives for each factor and simulation; see summary in Chapter 3). The
user has the option of selecting the base levels of the model variables as
the solver’s starting point for selected simulations (by including the file
varinit.inc); this may facilitate the solver’s task of finding a solution rela-
tive to the default, according to which it uses the variable levels from the
preceding model solution. Report parameters are declared in the include
file repsetup.inc and defined in the include files reploop.inc, repperc.inc,
and repsum.inc. The parameters are designed to contain most of the in-
formation that an analyst may be interested in; Table 5 provides details.
Repsum.inc may be used as a starting point for user-defined reports that
highlight information of interest in a specific application.

Figure 3—The structure of GAMS model and data files

mod.gms sim.gms

<name>.dat varinit.inc repsetup.inc

varlow.inc varinit.inc
repbase.inc reploop.inc
repperc.inc
repsum.inc
sambal.inc

32For information on solvers, visit the GAMS Development Corporation website

(www.gams.com).
33For save and restart facilities in GAMS, see Brooke et al. (1998, 199).
 45

The modeling system presented can be used in a variety of ways. The

first and most straightforward approach is to carry out simulations with
one of the existing data sets without making any changes in the modeling
structure. Here the user is required only to define new simulations. The
file sim.gms includes a note that summarizes the core steps to take when
carrying out additional simulations.
In a second approach, users may wish to take the additional step of ap-
plying the model to their own data set. If so, it is preferable to structure
the data set in the same way as the sample data files. The most critical ad-
ditional step is to generate a properly formatted SAM. If an available SAM
has a different format (for example, exports from activity accounts instead
of commodity accounts or a different treatment of taxes), we strongly rec-
ommend that the user reformat the SAM (a task that can be done inside
the GAMS include file). The alternative of adjusting the model code to a
differently formatted SAM is likely to be more time-consuming and error-
prone. Once the model properly calibrates to the new data set, the user
can proceed with simulations.
The third approach is also the most involved. Here, in combination
with 1 and 2, more advanced users may wish to change the model, a step
that involves changing the files mod.gms and, quite likely, <name>.dat,
as existing model elements (sets, parameters, variables, and equations)
are modified or new ones are declared and defined. If the user is also ap-
plying the model to a new data set (as in the second approach above), it is
probably easier to divide the process into two steps, first generating a data
set to which the original model calibrates and second modifying the model.
Changes in the model structure will also require the user to modify and/or
add to the report system, for example, adding new parameters to account
for new model variables and modifying the parameters that define the in-
comes and expenditures of SAM accounts.34
After having read this manual, we recommend that users familiarize
themselves with the contents of the different files. For users who limit
themselves to the first approach, the most important task is to become fa-
miliar with the file sim.gms and its include files. For users who also add
their own database, as in the second approach described above, it is also
crucial to be aware of the detailed structure of the standard SAM (de-
scribed in Chapter 2 and exemplified in the country data files) and how it
may differ from the original format of any new SAM that the user wants
to apply. A thorough study of the modeling system is required for users
who, in addition, wish to modify the model, using it as a tool to develop
further in different directions.

34The modeling system includes consistency checks on the report parameters that will
generate error messages if, for example, the reports show imbalances between the in-
come and spending of SAM accounts.
APPENDIX A: MATHEMATICAL SUMMARY

STATEMENT FOR THE STANDARD

CGE MODEL

αεΑ activities
SETS
α ЄACES(CA) activities with a CES function at the top of the
technology nest
α ЄALEO(CA) activities with a Leontief function at the top of the
technology nest
CEC commodities

CE CD(CC) commodities with domestic sales of domestic output

C = CDN(≤ C) commodities not in CD

CECE(CC) exported commodities

CE CEN(CC) commodities not in CE

CECM(CC) imported commodities

CЄ CMN(CC) commodities not in CM

CECT( C) transactions service commodities

C ECX(CC) commodities with domestic production

feF factors
i ЄINS institutions (domestic and rest of the world)
i Є INSD( INS) domestic institutions
i Є INSDNG

(< INSD) domestic nongovernment institutions

heH(CINSDNG) households

PARAMETERS
Latin Letters
cwtsc weight of commodity c in the CPI
dwtsc weight of commodity c in the producer price index
icaea quantity of c as intermediate input per unit of
activity a
icdec'
quantity of commodity c as trade input per unit of c '
produced and sold domestically
icecc' quantity of commodity c as trade input per exported
unit of c '
icme c'
quantity of commodity c as trade input per imported
unit of c '
intaa quantity of aggregate intermediate input per activity
unit

46
 47

ivaɑ
quantity of value-added per activity unit
mps ; base savings rate for domestic institution i
mps01 0-1 parameter with 1 for institutions with potentially
flexed direct tax rates

pwe export price (foreign currency)

pwmc import price (foreign currency)
qdste quantity of stock change
98c base-year quantity of government demand
qinve base-year quantity of private investment demand
shifiif share for domestic institution i in income of factor f
shii ii' share of net income of i' to i (i' INSDNG';
i ЄINSDNG)
tα a tax rate for activity a
tec export tax rate

tff direct tax rate for factor f

tins exogenous direct tax rate for domestic institution i
tins01; 0-1 parameter with 1 for institutions with potentially
flexed direct tax rates
tmc import tariff rate
tqc rate of sales tax
trnsfri f transfer from factor f to institution i
tvaa rate of value-added tax for activity a

Greek Letters da efficiency parameter in the CES activity function

va
αα efficiency parameter in the CES value-added function
ac
da
shift parameter for domestic commodity aggregation
function
а Armington function shift parameter
α CET function shift parameter
Bach marginal share of consumption spending on home
commodity c from activity a for household h
Bh marginal share of consumption spending on marketed
commodity c for household h
δα CES activity function share parameter
Sa
acc
share parameter for domestic commodity aggregation
function
89 Armington function share parameter
8t CET function share parameter
va
διfa CES value-added function share parameter for factor f
in activity a
m
ch subsistence consumption of marketed commodity c for
household h
yhac
h subsistence consumption of home commodity c from
activity a for household h
48

I
O ac yield of output c per unit of activity a
α

a a a
pa CES production function exponent
va CES value-added function exponent
Ρα
ac
ρα domestic commodity aggregation function exponent
pa Armington function exponent
pć CET function exponent

EXOGENOUS CPI consumer price index

VARIABLES DTINS change in domestic institution tax share ( = 0 for
base; exogenous variable)
FSAV foreign savings (FCU)
GADJ government consumption adjustment factor
IADJ investment adjustment factor
MPSADJ savings rate scaling factor (= 0 for base)
QFS f quantity supplied of factor
TINSADJ direct tax scaling factor ( = 0 for base ; exogenous vari-
able)
WFDIST ƒ a
wage distortion factor for factor f in activity a

ENDOGENOUS DMPS change in domestic institution savings rates ( = 0 for

base; exogenous variable)
VARIABLES
DPI producer price index for domestically marketed output
EG government expenditures
EH consumption spending for household
EXR exchange rate (LCU per unit of FCU)
GOVSHR government consumption share in nominal absorption
GSAV government savings
INVSHR investment share in nominal absorption
MPS marginal propensity to save for domestic non-
government institution (exogenous variable)
PA activity price (unit gross revenue)
PDDc demand price for commodity produced and sold
domestically
PDS supply price for commodity produced and sold
domestically
PEC export price (domestic currency)
PINT aggregate intermediate input price for activity a
A
PM import price (domestic currency)
PQ composite commodity price
PVA value-added price (factor income per unit of activity)
PXc
aggregate producer price for commodity
РХАСас producer price of commodity c for activity a
QA quantity (level) of activity
QDc quantity sold domestically of domestic output
 49

QEc quantity of exports

QFfa quantity demanded of factor f from activity a
QG government consumption demand for commodity
QHeh quantity consumed of commodity c by household h
QHAach quantity of household home consumption of
commodity c from activity a for household h
QINTA& quantity of aggregate intermediate input
QINTca quantity of commodity c as intermediate input to
activity a
QINV quantity of investment demand for commodity
QM quantity of imports of commodity
QQc quantity of goods supplied to domestic market
(composite supply)
QTc quantity of commodity demanded as trade input
QVAa quantity of (aggregate ) value-added
QXc aggregated marketed quantity of domestic output of
commodity
QXACac quantity of marketed output of commodity c from
activity a
TABS total nominal absorption
TINS direct tax rate for institution i (i e INSDNG)
TRII transfers from institution i' to i (both in the set
INSDNG)
WFf average price of factor f
YFf
income of factor f
YG government revenue
YI income of domestic nongovernment institution
YIFif income to domestic institution i from factor f
EQUATIONS
50
50

Block
Price
price
Import pmXR
=
PM
1t
·(
E
)
P
c+
icm
wm
Q CM
CЄ )
(1
с •
c'e
CT

import import tariff rate

exchange trade
of
cost
=
price adjust- (LCU +
per inputs
per
)(LCU F
() CU ment )
FCU unit
import

price
Export .E
t
p)
1-
PE XR-
(=ewe Σ
,Ρ
i
c Ω
ce CECE )
(2
с
CT
c'e

H
export tariff
export rate
exchange trade
of
cost
adjust-
=
price L CU
(per inputs
per
)(L CU
CU
Fment FCU
) unit
export

Demand
price
of PDS
=
PDD
P
Σ
+ Q
cicd CECD )
(3
domest
nontra
goods ic
ded CT
c'e
HE E
domestic EH domestic trade
of
cost
inputs
per
demand supply
unit
of
price price
sales
domestic

Absorption PQc
tqc
(1―
)·Q
P
=
·Q
P
+
Q
· Qc
DD
D
MM CE
CD
)
CM
( )
(4

absorption domestic
demand
price price
import
demand
a(t +
times times
of
net
prices domestic
sales
quantity import
quantity
sales
)
tax

Marketed
output
value · P
·Q
PX
=
QEX
+ DS
QDE CX
CЄ )
(5

producer
-| price domestic
Hsupply
price price
export
marketed
times
= times+ times
quantity
output domestic
sales
quantity quantity
export
price
Activity ·0
ac
P
=aXAC
PA₁Ѳа
с αεΑ )6
(
ac
C
CE

activity prices
producer
= times
price yields

interm ate
Aggregediate PINTA
i
·P
a Q
=cac αΕΑ )7
(
price
input C
CE

aggregate input
intermediate
cost
intermediate aggregate
of
unit
per
price
input intermedia
input te

Activity
revenue taa
(1-
PA₁
)Q
=
P
+
QVA
·QA₁
INTA
INTA
VA αεΑ )8
(
and
costs
price
activity value
-added aggregate
intermediate
)nof
taxes
(et = times
price times
price
input
level
activity
times quantity
quantity

index
price
Consumer )
(9
Dd3 = 1 CPI
P
cwts
=Q c

consumer times
prices
=
indeex
pric weights

Producer
for
index
price DPI
d PDS
Σwts
·= )(10
nontraded
market
output
CEC
] =[ ~ ;
index
price
producer timess
price
for
non
-traded
outputs weights
51
Production
and
52

Trade
Block
1
œINTA¸¯På
=
QA₁
8
1
(
+
)|·Q VA¸¯På
ª
−8ª Pa
α
technology
A:CES
ctivity A
Є
a CES )(11
function
production act added
value
aggregate
,of
CEiviSty quantity
level input
intermediate
aggregate
of
quantity

1
QVA 8apa
a1+
PINTA
a
a
-:V dded
alue
technology
CES QINTA
a PVA 1-8a Є
a ACES 1
() 2
interm
-input ediate a
quanti
ratio ty
value
--added
intermedia
i- nput te
intermediate-
f
= -avalue
dded
input
quantity
ratio
price
ratio

QVA
ivaa
=
QAa
Leontief
:technology Є
a ALEO )(13
Demand
for
aggregate for
demand
activity
f
value
-a dded =
value
added level

QINTA
Q
·i
=nta
Aa
Leonti
techno
: ef
logy Є
a ALEO )(14
for
aggreg d
Demanate
interm deman
aggreg
for date activity
input ediate
intermediate
input level
]=ƒ [

3. O va
δ va25 QVA₁
% Fa
=
QFƒ
a 8a
-Pa
and
-a
Value
dded fa αεΑ )(15
factor
demands feF
CES
=
]
aggregate
of [
quantity factor
value
added inputs
 · · · Q
Pa
QFfa
t
1-
)(
WFWva
,
P
a=
ƒ FDIST
VA
VA8 a· ·8
Q
%F 1- vaa
-p
Factor
demand f f αΕΑ )(16
feF

marginal
of
cost = product
revenue
marginal
factor
a
activity
fin activity
f
factor
in
aof

QINTca
Q
·i
=caca
INTA
intermediate
Disaggregated αεΑ )(17
demand
input CEC
intermediate
demand diate
intermeate
aggreg
commodity
cfor input
quantity
a
activity
from a
activity
for

QXACac
Q0 HAach
+ac
'= Aa
Commodity
production hЄ
H αεΑ )(18
and
allocation household
home С
є
аХ
marketed
quantit y production
commod
of consumption commodit
oc=f
c +ity y
c
commodity
of
afrom
activity activity
afrom
activity
afrom

1
1-
pac
ас α
QX
= ac8ac
QXACac
-Pc
function
aggregation
Output с CECX )(19
А
ає

aggregate -specific
activity
marketed CES marketed
of
production production
of
c
commodity c
commodity

PXACa
P
c·Q X
= Xc c -P
Q· XAC
Σ8ac ас P
-
Q
8
· XAC
· ac
condition
-o
First
for
rder ас c ac c Ε
αΑ )(20
'Α
αε
function
aggregation
output CECX
marginal
com-
of
cost of
product
revenue
marginal
activity
cfrom
amodity cfactivity
acommodity
rom
53
 and
Production
54

Trade
Block

)(continued P· EDº¹
QX₁
·'Q
'
α
=
′)
(1-8
²+ EP
· · − ·
transformation
Output CE
(
Є
)
OCD
c )(21
f
)C ET
( unction omesti
quantit
market
export
, aggrega
dC
= ET ycedte
outpu
domesttic sales
domestic
of
output

1
QE 1- 1-
pc
P E
-domestic
Export
ratio
supply QD PDS.c (E
cЄ
C
(
) D )(22

-domestic
export
f - omestic
dexport
=
ratio
supply ratio
price

QDc
=
QX
Q
+ E
for
transformati
Output on CE
domestic
market
commodities
e
- xported
non aggregate domestic
sales f[ or C
)( EN
( D )(23
marketed
of + exports
domestic [ or
foutput (CE
CE )]
CDN
output
)](CD
ce CEN (CE
)UCDN

I
Q
)
8
1−)
·²²
%Q
QQ₁(
+
αD´¹²
= ¶
ºM² På
—
Composite
supply CD
)(
CM
CE )(24
)f
A rmington
( unction
composite
f , omestic
dquantity
import
=
supply domestic
of
use
output

QM 89
PDDc p1
+a
ratio
demand
- omestic
d
Import QDc 1
c -89
PM CD
)(
CM
CE )(25

d- omestic
import domestic
-import
demand =
ratio ratio
price
 Q
+=
QDc
cQQc
M
for
Composite
supply CE
outputs
-i
non
mported C
( DCMN
) )(26
domestic
use
of
nonproduced
and
imports composite domestic
marketed imports
[for
+
supply = f[or
output CE
)](CM
CDN C
)( MCDN
(CD
CE )]
CMN

QDc
)
i
+
QE
QT
(
QMce
icdee
cmc
=
Demand
for CECT 2
() 7
'C
C'E
services
transactions
demand
for deman
of ds
sum
transactions imports
xports
e,for
services and
sales
domestic

Block
Institution YF₁
=
W W
·
a
Q FDIST
ƒ Fƒ
,
· F
f • feF
income
Factor )(28
А
ає
activity
of
sum
payments
income
of pecific
wages
sa-(ctivity
factor
f
)
levels
employment
times

shif¡
=
YIF₁
,'[(1−
t)·Y
,—rnsfr
fF ]
EXR
Institutional
factor
incomes f ·
row INSD
iЄ )(29
income
of share
income
of income
f
factor
of fЄF
institu
i= tion of
to
f
factor and
tax
nof
(et
f
factor
from i
institution )
RoW
to
transfer

Y
+Σ
=
YI₁
;ƒ IF Σ TRII TRII¿¿
t+rnsfr CP
+
tro Iri E
rnwsf · XR
Income
domestic
,of 'i gov INSDNG
iЄ )(30
F
fe i'e
'
INSDNG
nongovernme
institutions nt
transfers
transfers transfers
income
of factor
+ domestic
other
from +
= from from
institution
i income non
-government
government ROW
institutions
55
 Block
Institution
56

)(continued
shii¡
¡=
TRII
M
¡—
T
)(1-
,·Y INS
PS
I
Intra
-institutional INSDNG
iЄ
H income
net
of
share income
institution
of
transfers '
INSDNG
i'Є )(31
transfer
from i'
institution
of and
savings
of
i',net
i'to
institution i
to direct
transfered taxes

EH = Σ
shi
, -1-
shia) M-
1-
)·(1
·Y
)
TINS
I
PS₁
consum old
Househption hЄH 3
() 2
IЄi NSDNG
expenditure
household
income , et
income
household
nof
direct
disposable
for taxes
,asavings
to
transfers
nd
consumption institutions
- overnment
non
gother

m PQ
P
=
h
QH
|E
”
B
+
Y
-
EHHQ Thôn
P
)
-
1"-
YQ
Σ
- EEPXAC
Vach
PXAC₂e
oh
ch ch c'h
Household
consumption C'E
C c'ЄС
A
аЄ CEC
demand
marketed
for consumpti
household on total
household
consumpti hЄH )(33
on
commodities market
on
spending price
of
other
nd
spending
c,amarket
commodit
c y home
and
)(market
prices
commodity

PXACa
QHA c PX
Ya AC
ac chB
+h EH₁
PQ
Σ
-
Yach
PXACac
-ΣΣ
Yen
. ach
Household
consumption C
c'e C
cA'Є
аЄ αεΑ
demand
home
for consumption
household
total CEC )(34
consumpti
householdon ,spending
commodities commodity
home
on
spending nd
price
a,other
producer hЄH
afrom
cactivity prices
(mhome
and
)arket
commodity

QINV
I
=
q ADJ
· inv
demand
Investment CE
CINV )(35
factor
adjustment
fixed
investment
demand
for times
base
-year
fixed
c
commodity
investment
 QGGADJ
98c
Government CEC )(36
deman ption
consumd adjustment
factor
government
consumption times
demand
for government
ybase
- ear
c
commodity consumption

YG = f
+
YI
ΣTINS
tva
+PVA
,Q YF
VAa
·
Government
revenue IЄi NSDNG fe
F A
E 3
() 7
Σ
:p
Q
;E
p
+
EXR
QM
ΣAa
'·P
Σ we
XR
E
te
wm
taa
A
tm

Α
αε СМ
СЕ CE
E
Σ
+
•Q
P YIF
Q
· tq t+rnsfr ·EXR
f
gov row
gov
C
CE F
fe

]
-338
(
direct
taxes direct
taxes value-
government
= from from added
revenue institutions factors tax
======
transfers
import
activity export sales factor
f+rom
tax tariffs taxes tax income ROW

EG
•= P
t
+
QG
Σ rnsfr
Q C
· PI
Government
expenditures gov 3
)( 8
C
CE IЄi NSDNG
domestic
to
transfers
government government
spending g- overnment
non
consumption
institutions

System
Const
Blockraint ΣQFƒ
OF a=
Qf FS
market
Factor fЄF )(39
A
ає
dema
for nd supply
of
facto
f r= factor
f
57
System
58

Block
Constraint
)(continued
QQc
=
Σ
a QINTc +
Q
Q
+
Q INV
QHch
G
Σdst
q T
Composite
commodity Аає hЄ
H
markets +[
] CEC )(40
composite
• intermediate household governme
+ nt fixed + stock trade
supply use consumption consumption investment change input
use

Σ
+
QM
p
Σ trnsf
wm r ;+
Σ
·Q
pow
E
we
=rnsfr
rtrowƒ F
+ SAV
Current
balance
account CECM F
fe CE
CE INSD
iЄ )(41
world
for
the
of
rest
factor institutional
(in
currency
)foreign import t+
ransfers export foreign
= transfers
spending revenue savings
RoW
to RoW
from

YG
EGGSAV
=
balance
Government )(42
government +government government
revenue expenditures savings

=ins
TINS
T
;(1+
D
)+
,·t ins01
INSADJ
TINS
institutional
Direct
rates
tax INSDNG
iЄ 4
() 3
direct
tax adjusted
rate
base change
point
rate
for scaling+
for selected
for
iinstitution selected
institutions institutions

MPS₁
=ps01
,(1
MPSADJ
D
,)+
;·m ps
+
MPS
Institutional
savings
rates INSDNG
iЄ )(44
savings adjusted
rate
base change
point
rate
for = for
scaling+ selecte
for d
i
institution institutions
selected institutions
 TI
1-
(
ΣMPS
)·Y
,G
E
=
FSAV
P
+
QINV
qdst
INS
SAV
XR
Q
Balance
Investment
-
Savings INSDNG
iЄ C 4
)( 5
CEC CE

-gnon
overn- government
+ foreign fixed + stock
savings
ment savings savings investment change

=
TABS ΣΣ
,Η
ΡΩ QHch
P
=Q Σ
Ρ
Σ
+
PXACaOHAach
QGΩΣΣ
absorp
Total tion hЄ
cC
HЄ He
cChe
A
aЄ с
се )(46
Ω
OINV
Ρ
Σ
,+ + ΡΩ
Σ
qdste
C
CE C
CE
household household
total +
+ government
+ fixed + stock
market
home
absorption consumption investment change
consumption

INVSHR
T
qΣ
= ABS
PQ
QINV
+
·P
· dst
Q
Ratio
investment
of 4
)( 7
C
CE C
CE
absorption
to
FOOD
investment-
total fixed + stock
absorption =
absorption investment change
ratio

GOVSHR
TABS
Σ
Q PQ
•=Gc
Ratio
government
of )(48
C
CE
consumption
absorption
to
government
consumption- total government
=
absorption absorption consumption
ratio
59
APPENDIX B: CORE GAMS CODE FOR
STANDARD CGE MODEL

*SETS = == == == == == == == == == = == == == = = == == == == == == == == == == == == == == =
AC global set for model accounts-aggregated microsam accounts
A(AC) activities
ACES(A) activities with CES fn at top of technology nest
ALEO(A) activities with Leontief fn at top of technology nest
C(AC) commodities
CD(C) commodities with domestic sales of output
CDN(C) commodities without domestic sales of output
CE(C) exported commodities
CEN(C) non-exported commodities
CM(C) imported commodities
CMN(C) non-imported commodities
CX(C) commodities with output
F(AC) factors
INS(AC) institutions
INSD(INS) domestic institutions
INSDNG(INSD) domestic non-government institutions
H(INSDNG) households

*PARAMETERS == == == == == == == == == == == == == == == == == == == == == == == == == ==

Parameters other than tax rates

alphaa(A) shift parameter for top level CES function
alphaac(C) shift parameter for domestic commodity aggregation fn
alphaq(C) shift parameter for Armington function
alphat(C) shift parameter for CET function
alphava(A) shift parameter for CES activity production function
betah(A,C,H) marg shr of hhd cons on home com c from act a
betam(C,H) marg share of hhd cons on marketed commodity c
cwts(C) consumer price index weights
deltaa(A) share parameter for top level CES function
deltaac(A,C) share parameter for domestic commodity aggregation fn
deltaq(C) share parameter for Armington function
deltat(C) share parameter for CET function
dwts(C)
deltava(F,A) share parameter for CES activity production function
domestic sales price weights
gammah(A,C,H) per-cap subsist cons for hhd h on home com c fr act a
gammam(C,H) per-cap subsist cons of market com c for hhd h
ica(C,A) intermediate input c per unit of aggregate intermediate

60
 61

inta (A) aggregate intermediate input coefficient

iva (A) aggregate value added coefficient
icd (C , CP ) trade input of c per unit of com cp produced & sold dom ' ly
ice (C , CP ) trade input of c per unit of com cp exported
icm ( C , CP ) trade input of c per unit of com cp imported
mps01 ( INS ) 0-1 par for potential flexing of savings rates
mpsbar ( INS ) marg prop to save for dom non-gov inst ins (exog part )
qdst (C) inventory investment by sector of origin
qbarg (C) exogenous (unscaled ) government demand
qbarinv ( C ) exogenous (unscaled ) investment demand
rhoa (A) CES top level function exponent
rhoac (C) domestic commodity aggregation function exponent
rhoq ( C ) Armington function exponent
rhot (C ) CET function exponent
rhova (A) CES activity production function exponent
shif ( INS , F ) share of dom . inst i in income of factor f
shii ( INS , INSP ) share of inst i in post - tax post - sav income of inst ip
supernum (H ) LES supernumerary income
theta ( A , C ) yield of commodity c per unit of activity a
tins01 ( INS ) 0-1 par for potential flexing of dir tax rates
trnsfr ( INS , AC ) transfers fr inst . or factor ac to institution ins
* Tax rates
ta (A) rate of tax on producer gross output value
te (C ) rate of tax on exports
tf ( F ) rate of direct tax on factors ( soc sec tax )
tinsbar ( INS) rate of ( exog part of ) direct tax on dom inst ins
tm (C ) rate of import tariff
tq (C ) rate of sales tax
tva (A) rate of value -added tax

* VARIABLES

CPI consumer price index ( PQ-based )

DPI index for domestic producer prices ( PDS - based)
DMPS change in marginal propensity to save for selected inst
DTINS change in domestic institution tax share
EG total current government expenditure
EH ( H) household consumption expenditure
EXR exchange rate
FSAV foreign savings
GADJ government demand scaling factor
GOVSHR govt consumption share of absorption
GSAV government savings
IADJ investment scaling factor ( for fixed capital formation )
INVSHR investment share of absorption
MPS ( INS ) marginal propensity to save for dom non - gov inst ins
MPSADJ savings rate scaling factor
PA (A) output price of activity a
62
PDD(C) demand price for com c produced & sold domestically
PDS(C) supply price for com c produced & sold domestically
PE(C) price of exports
PINTA(A) price of intermediate aggregate
PM(C) price of imports
PQ(C) price of composite good c
PVA(A) value added price
PWE(C) world price of exports
PWM(C) world price of imports
PX(C) average output price
PXAC(A,C) price of commodity c from activity a
QA(A) level of domestic activity
QD(C) quantity of domestic sales
QE(C) quantity of exports
QF(F,A) quantity demanded of factor f from activity a
QFS(F) quantity of factor supply
QG(C) quantity of government consumption
QH(C,H) quantity consumed of marketed commodity c by household h
QHA(A,C,H) quantity consumed of home commodity c fr act a by hhd h
QINT(C,A) quantity of intermediate demand for c from activity a
QINTA(A) quantity of aggregate intermediate input
QINV(C) quantity of fixed investment demand
QM(C) quantity of imports
QQ(C) quantity of composite goods supply
QT(C) quantity of trade and transport demand for commodity c
QVA(A) quantity of aggregate value added
QX(C) quantity of aggregate marketed commodity output
QXAC(A,C) quantity of ouput of commodity c from activity a
TABS total absorption
TINS(INS) rate of direct tax on domestic institutions ins
TINSADJ direct tax scaling factor
TRII(INS,INSP) transfers to dom inst insdng from insdngp
WALRAS Savings–Investment imbalance (should be zero)
WF(F) economy-wide wage (rent) for factor f
YG
WFDIST(F,A) factor wage distortion variable
YF(F) factor income
total current government income
YIF(INS,F) income of institution ins from factor f
YI(INS) income of (domestic non-governmental) institution ins
=

*EQUATIONS == == == == == == == == == == == = = == == == == == == == == == == == == = == =

*Price block == == == == == == == == == == = = == == == == == == == == == == == == == == =
PMDEF(C) domestic import price
PEDEF(C) domestic export price
PDDDEF(C) demand price for com c produced and sold domestically
PQDEF(C) value of sales in domestic market
PXDEF(C) value of marketed domestic output
 63

CPIDEF
PVADEF(A)
PINTADEF(A)
PADEF(A) output price for activity a
price of aggregate intermediate input
value-added price
consumer price index
DPIDEF domestic producer price index

*Production and trade block = == == == == = = == == == == == == == == == == == == == == =

CESAGGPRD(A) CES aggregate prod fn (if CES top nest)
CESAGGFOC(A) CES aggregate first-order condition (if CES top nest)
LEOAGGINT(A) Leontief aggreg intermed demand (if Leontief top nest)
LEOAGGVA(A) Leontief aggreg value-added demand (if Leontief top nest)
CESVAPRD(A) CES value-added production function
COMPRDFN(A,C)
INTDEM(C,A)
CESVAFOC(F,A) CES value-added first-order condition
intermediate demand for commodity c from activity a
production function for commodity c and activity a
OUTAGGFN(C) output aggregation function
OUTAGGFOC(A,C) first-order condition for output aggregation function
CET(C) CET function
CET2(C) domestic sales and exports for outputs without both
ESUPPLY(C) export supply
ARMINGTON(C) composite commodity aggregation function
COSTMIN(C) first-order condition for composite commodity cost min
QTDEM(C)
ARMINGTON2(C) comp supply for com without both dom sales and imports
demand for transactions (trade and transport) services

*Institution block == == == == == == == == = = == == == == == == == == == == == == == = ==
YFDEF(F) factor incomes
YIDEF(INS)
YIFDEF(INS,F) factor incomes to domestic institutions
total incomes of domest non-gov’t institutions
EHDEF(H) household consumption expenditures
TRIIDEF(INS,INSP) transfers to inst ins from inst insp
HMDEM(C,H) LES cons demand by hhd h for marketed commodity c
HADEM(A,C,H) LES cons demand by hhd h for home commodity c fr act a
INVDEM(C) fixed investment demand
GOVDEM(C) government consumption demand
EGDEF total government expenditures
YGDEF total government income

*System constraint block = == == == == == = = == == == == == == == == == == == == == = = =

COMEQUIL(C) composite commodity market equilibrium
FACEQUIL(F) factor market equilibrium
CURACCBAL current-account balance (of RoW)
GOVBAL government balance
TINSDEF(INS) direct tax rate for inst ins
MPSDEF(INS) marg prop to save for inst ins
SAVINVBAL Savings–Investment balance
TABSEQ total absorption
INVABEQ investment share in absorption
GDABEQ government consumption share in absorption
64

* Notational convention inside equations :

* Parameters and “ invariably” fixed variables are in lower case .
* Potentially “variable ” variables are in upper case .

||
||
* Price block =

||
PMDEF ( C ) $CM ( C ) ..
PM ( C ) =E= pwm ( C ) * ( 1 + tm ( C ) ) * EXR + SUM ( CT , PQ ( CT ) * icm ( CT , C ) ) ;

PEDEF ( C ) $ CE ( C ) ..
PE ( C ) = E= pwe (C ) * ( 1 te (C ) ) * EXR - SUM ( CT , PQ (CT ) * ice ( CT , C) ) ;

PDDDEF ( C ) $ CD ( C ) .. PDD ( C ) =E = PDS ( C ) + SUM ( CT , PQ ( CT ) * icd ( CT , C ) ) ;

PODEF (C ) $ (CD ( C ) OR CM (C ) ) ..
PQ ( C ) * ( 1 tq ( c ) ) * QQ ( C ) =E = PDD ( C ) * QD ( C ) + PM ( C ) * QM ( C ) ;

PXDEF ( C ) $ CX ( C ) .. PX (C ) *QX (C ) = E= PDS ( C ) * QD ( C ) + PE ( C ) * QE ( C ) ;

PADEF (A) .. PA (A) =E = SUM ( C , PXAC (A, C ) * theta (A , C ) ) ;

PINTADEF ( A) .. PINTA (A) =E= SUM ( C , PQ ( C ) * ica ( C , A) ) ;

PVADEF (A) .. *
PA (A) ( 1 - ta (A) ) * QA (A) = E= PVA (A) * QVA (A) + PINTA (A) * QINTA (A) ;

CPIDEF .. CPI E SUM ( C , cwts ( C ) * PQ ( C ) ) ;

DPIDEF .. DPI =E = SUM ( CD , dwts ( CD ) * PDS ( CD ) ) ;

||
||
||
* Production and trade block

* CESAGGPRD and CESAGGFOC apply to activities with CES function at

* top of technology nest .

CESAGGPRD ( A) $ACES (A) ..

QA (A) =E = alphaa ( A) * ( deltaa ( A ) * QVA (A ) ** ( −rhoa (A) )
+ (1 -deltaa (A) ) * QINTA (A) ** ( −rhoa ( A) ) ) ** ( −1 /rhoa (A ) ) ;

CESAGGFOC (A) $ACES ( A) ..

QVA (A) = E= QINTA ( A) * ( ( PINTA ( A) / PVA (A) ) * ( deltaa (A) /
(1 - deltaa ( A) ) ) ) ** ( 1 / ( 1 +rhoa (A) ) ) ;

* LEOAGGINT and LEOAGGVA apply to activities with Leontief function at

* top of technology nest .

LEOAGGINT ( A) $ALEO ( A) .. QINTA ( A) = E = inta ( A) * QA (A) ;

LEOAGGVA (A) $ALEO (A) .. QVA (A) =E= iva (A) * QA (A) ;
 65

* CESVAPRD , CESVAFOC , INTDEM apply at the bottom of the technology nest

* ( for all activities ) .

CESVAPRD (A) ..
QVA (A) =E = alphava (A) * ( SUM ( F ,
deltava ( F , A) * QF ( F , A) ** ( −rhova (A) ) ) ) ** ( - 1 /rhova (A) ) ;

CESVAFOC ( F , A) $ deltava ( F , A) ..
WF ( F ) *wfdist ( F , A) =E=
PVA (A) * ( 1 - tva (A) )
* QVA (A) * SUM ( FP , deltava ( FP , A) * QF ( FP , A) ** ( −rhova ( A) ) ) ** ( -1 )
*deltava ( F , A) * QF ( F , A) ** ( -rhova (A) −1 ) ;

INTDEM (C , A) $ ica ( C , A) .. QINT ( C , A) = E= ica ( C , A) *QINTA (A) ;

COMPRDFN (A , C ) $ theta ( A, C ) ..
QXAC (A , C ) + SUM ( H , QHA ( A , C , H ) ) = E = theta (A , C ) * QA (A) ;

OUTAGGFN ( C ) $CX ( C ) ..
QX ( C ) =E= alphaac ( C ) * SUM (A , deltaac ( A , C ) *QXAC (A, C )
** ( - rhoac (C ) ) ) ** ( -1 / rhoac ( C ) ) ;

OUTAGGFOC ( A , C ) $deltaac (A , C ) ..
PXAC (A , C ) =E =
PX (C )
* QX (C ) * SUM (AP , deltaac (AP , C ) * QXAC (AP , C ) ** ( -rhoac (C ) ) ) ** ( - 1 )
*deltaac (A, C ) *QXAC (A, C ) ** ( -rhoac ( C ) -1 ) ;

CET ( C ) $ ( CE ( C ) AND CD ( C ) ) ..
QX (C ) = E = alphat ( C ) * ( deltat ( C ) * QE ( C ) ** rhot ( C ) +
( 1 - deltat ( C ) ) * QD ( C ) ** rhot ( C ) ) ** ( 1 / rhot ( C ) ) ;

ESUPPLY (C ) $ ( CE ( C ) AND CD ( C ) ) ..
QE (C ) = E= QD (C ) * ( ( PE ( C ) / PDS ( C ) ) *
( (1 deltat (C ) ) /deltat (C ) ) ) ** ( 1 / (rhot ( C ) -1 ) ) ;

CET2 ( C ) $ ( ( CD ( C ) AND CEN ( C ) ) OR ( CE ( C ) AND CDN ( C ) ) ) ..

QX (C) = E= QD (C ) + QE (C ) ;

ARMINGTON ( C ) $ ( CM ( C ) AND CD ( C ) ) ..
QQ (C) = E= alphaq (C ) * ( deltaq ( C ) * QM ( C ) ** ( -rhoq (C ) ) +
(1 -deltaq ( C ) ) * QD ( C ) ** ( -rhoq (C ) ) ) ** ( -1 /rhoq (C) ) ;

COSTMIN (C ) $ ( CM ( C ) AND CD ( C ) ) ..
QM (C ) = E= QD ( C ) * ( ( PDD (C ) / PM (C ) ) * (deltaq ( C ) / (1 - deltaq (C) ) ) )
** (1 / (1 + rhoq (C ) ) ) ;

ARMINGTON2 ( C ) $ ( ( CD ( C ) AND CMN ( C ) ) OR ( CM ( C ) AND CDN ( C ) ) ) ..

QQ (C ) =E = QD ( C ) + QM ( C ) ;
66

QTDEM ( C ) $ CT ( C ) ..
QT (C ) = E= SUM ( CP , icm ( C , CP ) * QM (CP) + ice ( C , CP ) * QE ( CP ) + icd (C , CP ) * QD ( CP ) ) ;

|| ||
* Institution block

||
YFDEF ( F ) .. YF ( F ) = E= SUM (A , WF ( F ) *wfdist ( F , A) * QF ( F , A) ) ;

YIFDEF ( INSD , F ) $ shif ( INSD , F ) ..

YIF ( INSD , F ) = E= shif ( INSD , F ) * ( ( 1 - tf ( f ) ) * YF ( F ) — trnsfr ( ' ROW ' , F ) *
EXR ) ;

YIDEF ( INSDNG) ..
YI ( INSDNG ) =E=
SUM (F , YIF ( INSDNG , F ) ) + SUM ( INSDNGP , TRII ( INSDNG , INSDNGP ) )
+ trnsfr ( INSDNG , ' GOV ' ) * CPI + trnsfr ( INSDNG , ' ROW ' ) * EXR ;

TRIIDEF ( INSDNG , INSDNGP ) $ ( shii ( INSDNG , INSDNGP ) ) ..

TRII ( INSDNG , INSDNGP ) =E= shii ( INSDNG , INSDNGP )
* (1 MPS ( INSDNGP ) ) * (1 - TINS ( INSDNGP ) ) * YI ( INSDNGP ) ;

EHDEF ( H ) ..
EH ( H ) =E= ( 1 — SUM ( INSDNG , shii ( INSDNG , H ) ) ) * (1 ― MPS ( H ) )
(1 TINS (H ) ) * YI (H) ;

HMDEM ( C , H ) $betam ( C , H ) ..
PQ ( C ) *QH (C , H ) = E=
PQ (C ) * gammam ( C , H )
-
+ betam (C , H ) * ( EH ( H ) SUM (CP , PQ (CP ) *gammam ( CP , H ) )
— SUM ( (A , CP ) , PXAC (A, CP ) * gammah (A , CP , H ) ) ) ;

HADEM ( A , C , H ) $betah ( A , C , H ) ..
PXAC (A, C ) *QHA (A , C , H ) =E=
PXAC (A , C) *gammah (A, C , H )
+ betah (A , C , H ) * ( EH ( H ) SUM ( CP , PQ ( CP ) * gammam ( CP , H) )
— SUM ( (AP , CP ) ,
PXAC (AP , CP) *gammah (AP , CP , H) ) ) ;

INVDEM ( C ) .. QINV ( C ) =E = IADJ* qbarinv ( C ) ;

GOVDEM ( C ) .. QG ( C ) =E = GADJ* qbarg ( C ) ;

YGDEF ..
YG E SUM ( INSDNG , TINS ( INSDNG ) *YI ( INSDNG) )
+ SUM ( f , tf ( F ) *YF ( F ) )
+ SUM ( A, tva (A) * PVA (A) * QVA (A) )
+ SUM ( A , ta ( A) * PA (A ) * QA (A) )
+ SUM ( C , tm ( C ) *pwm ( C ) * QM ( C ) ) * EXR
+ SUM ( C , te ( C ) *pwe ( C ) * QE ( C ) ) * EXR
+ SUM (C , tq (C ) * PQ (C ) * QQ (C ) )
+ SUM ( F , YIF ( ' GOV ' , F ) )
+ trnsfr ( ' GOV ' , ' ROW ' ) * EXR ;
 67

EGDEF .. EG =E= SUM ( C , PQ ( C ) * QG ( C ) ) + SUM ( INSD , trnsfr ( INSD , ' GOV ' ) ) * CPI ;

||
*System constraint block =

FACEQUIL ( F ) .. SUM ( A, QF ( F , A) ) =E = QFS ( F ) ;

COMEQUIL (C ) ..
QQ ( C ) =E = SUM (A , QINT ( C , A) ) + SUM ( H , QH ( C , H ) ) + QG ( C )
+ QINV (C ) + qdst ( C ) + QT ( C ) ;

CURACCBAL ..
SUM ( C , pwm ( C ) * QM ( C ) ) + SUM ( F , trnsfr ( ' ROW ' , F ) ) =E=
SUM ( C , pwe ( C ) * QE ( C ) ) + SUM ( INSD , trnsfr ( INSD , ' ROW ' ) ) + FSAV ;

GOVBAL .. YG =E = EG + GSAV ;

TINSDEF ( INSDNG) .
TINS ( INSDNG ) = E = tinsbar ( INSDNG ) * ( 1 + TINSADJ * tins01 ( INSDNG ) ) +
DTINS *tins01 ( INSDNG) ;

MPSDEF ( INSDNG) ..
MPS ( INSDNG ) =E = mpsbar ( INSDNG) * ( 1 + MPSADJ *mps01 ( INSDNG ) ) + DMPS *mps01 ( INSDNG ) ;

SAVINVBAL ..
SUM ( INSDNG , MPS ( INSDNG) * (1 TINS ( INSDNG ) ) * YI ( INSDNG) )
+ GSAV + FSAV* EXR = E =
SUM (C , PQ ( C ) *QINV ( C ) ) + SUM ( C , PQ (C ) *qdst (C) ) + WALRAS ;

TABSEQ ..
TABS = E =
SUM ( (C , H ) , PQ ( C ) * QH ( C , H ) ) + SUM ( ( A , C , H ) , PXAC (A , C ) * QHA ( A , C , H) )
+ SUM ( C , PQ ( C ) * QG ( C ) ) + SUM (C , PQ ( C ) * QINV ( C ) ) + SUM ( C , PQ ( C ) * qdst ( C ) ) ;

INVABEQ .. INVSHR* TABS =E= SUM ( C , PQ ( C ) * QINV ( C ) ) + SUM ( C , PQ ( C ) * qdst (C ) ) ;

GDABEQ .. GOVSHR*TABS =E= SUM (C , PQ ( C) *QG (C ) ) ;

 REFERENCES
Armington, P. A. 1969. A theory of demand for products distinguished by
place of production. IMF Staff Papers 16 (1): 159–178.

Blonigen, Bruce A., Joseph E. Flynn, and Kenneth A. Reinert. 1997.

Sector-focused general equilibrium modeling. In Applied methods for
trade policy analysis: A handbook, ed. J. F. Francois and K. A. Reinert.
New York: Cambridge University Press.

Brooke, A., D. Kendrick, A. Meeraus, and R. Raman. 1998. GAMS: A

user’s guide. Washington, D.C.: GAMS Development Corporation.

Dervis, K., J. de Melo, and S. Robinson. 1982. General equilibrium models

for development policy. New York: Cambridge University Press.

Devarajan, S., J. D. Lewis, and S. Robinson. 1993. External shocks, pur-

chasing power parity, and the equilibrium real exchange rate. World
Bank Economic Review 7 (1): 45–63.

Dixon, P. B., B. R. Parmenter, A. A. Powell, and P. J. Wilcoxen. 1992. Notes

and problems in applied general equilibrium economics. New York:
North-Holland.

Ginsburgh, V., and M. Keyzer. 1997. The structure of applied general equi-
librium models. Cambridge, Mass., U.S.A.: MIT Press.

Johansen, Leif. 1960. A multi-sectoral study of economic growth. Amster-

dam: North-Holland.

Lofgren, Hans. 2000a. Exercises in general equilibrium modeling using

GAMS. Microcomputers in Policy Research, vol. 4a. Washington, D.C.:
International Food Policy Research Institute.

Lofgren, Hans. 2000b. Key to exercises in general equilibrium modeling

using GAMS. Microcomputers in Policy Research, vol. 4b. Washing-
ton, D.C.: International Food Policy Research Institute.

Pyatt, G. 1988. A SAM approach to modeling. Journal of Policy Modeling

10: 327–352.

Pyatt, G., and J. Round. 1985. Social accounting matrices: A basis for
planning. Washington, D.C.: World Bank.

Rattsø, J. 1982. Different macroclosures of the original Johansen model

and their impact on policy evaluation. Journal of Policy Modeling 4 (1):
85–97.

Reinert, K. A., and D. W. Roland-Holst. 1997. Social accounting matrices.

In Applied methods for trade policy analysis: A handbook, ed. J. F.
Francois and K. A. Reinert. New York: Cambridge University Press.

68
 69

Robinson, S. 1989. Multisectoral models. In Handbook of development eco-

nomics, vol. II, ed. H. Chenery and T. N. Srinivasan. Amsterdam: El-
sevier Science Publishers.

Robinson, S. 1991. Macroeconomics, financial variables, and computable

general equilibrium models. World Development 19: 1509–1525.

Robinson, S., and D. W. Roland-Holst. 1988. Macroeconomic structure and

computable general equilibrium models. Journal of Policy Modeling
10: 353–375.

Robinson, S., A. Yúnez-Naude, R. Hinojosa-Ojeda, J. D. Lewis, and S. De-

varajan. 1999. From stylized to applied models: Building multisector
CGE models for policy analysis. North American Journal of Econom-
ics and Finance 10: 5–38.

Robinson, S., and M. El-Said. 2000. GAMS code for estimating a social ac-
counting matrix (SAM) using cross entropy methods (CE). Trade and
Macroeconomics Division Discussion Paper No. 64. Washington, D.C.:
International Food Policy Research Institute.

Robinson, S., A. Cattaneo, and M. El-Said. 2001. Updating and estimating

a social accounting matrix using cross entropy methods. Economic
Systems Research 13: 47–64.

Shoven, J. B., and J. Whalley. 1992. Applying general equilibrium. New

York: Cambridge University Press.

Taylor, L. 1990. Structuralist CGE models. In Socially relevant policy

analysis, ed. L. Taylor. Cambridge, Mass., U.S.A.: MIT Press.

Thomas, M., and R. Bautista. 1999. A 1991 social accounting matrix

(SAM) for Zimbabwe. Trade and Macroeconomics Division, Discussion
Paper No. 36. Washington, D.C.: International Food Policy Research
Institute.
About this Manual

The purpose of this manual is to contribute to and facilitate the use of computable general equilibrium (CGE)
models in the analysis of issues related to food policy in developing countries . The volume includes a detailed
presentation of a static " standard " CGE model and its required database and incorporates features of partic-
ular importance in developing countries . The manual discusses the implementation of the model in GAMS
and is accompanied by a CD- ROM that includes the GAMS files for the model, sample databases, simulations ,
solution reports, and a social accounting matrix (SAM ) aggregation program . Although the volume provides a
standardized framework for analysis, the analyst is not forced to make " one -size-fits-all " assumptions . The
GAMS code is written to give the analyst considerable flexibility in model specification .

About the Authors

Hans Lofgren is a senior research fellow in the Trade and Macroeconomics Division of IFPRI , where he leads
IFPRI's multicountry program , " Macroeconomic Policies, Growth , and Poverty Reduction . " Hans ' research
focuses on the analysis of macro- and micro-effects of food , agriculture, and trade policy, as well as on the
development of methods for economywide policy modeling.
Rebecca Lee Harris is the research director at the Globalization Research Center at the University of South
Florida in Tampa, Florida . She worked as a research analyst and postdoctoral fellow in the Trade and
Macroeconomics Division of IFPRI from 1997 to 2002. Rebecca's research focuses on linkages between macro-
economic policies and shocks and income distribution , concentrating on Latin America and the Caribbean .
Sherman Robinson is director of the Trade and Macroeconomics Division of IFPRI . He is an international
authority in the area of policy-oriented general equilibrium modeling . At IFPRI , he has applied these and other
tools to policy issues related to agricultural development, income distribution , poverty, intersectoral linkages,
macroeconomic policy, and international trade.
Marcelle Thomas is a research analyst in the Trade and Macroeconomics Division of IFPRI , where she works
on a variety of macroeconomic and trade issues . In recent years she has worked on building and maintaining
social accounting matrices and computable general equilibrium models for use in macroeconomic analysis for
the Philippines and Zimbabwe . She is currently working on WTO issues and their impact on the food security
of developing countries .
Moataz El - Said is a research analyst in the Trade and Macroeconomics Division of IFPRI . He is currently
working on his Ph.D. dissertation in economics from the George Washington University, and at IFPRI works on
numerous research projects related to agricultural development, income distribution , poverty, and internation-
al trade .

Also Available in this Series

Exercises in General Equilibrium Modeling Using GAMS, by Hans Lofgren , Microcomputers in Policy Research 4a , 2000.

Key to Exercises in General Equilibrium Modeling Using GAMS, by Hans Lofgren, Microcomputers in Policy Research 4b, 2000.

INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE ISBN 0-896-29720-9

2033 K Street NW, Washington , DC 20006-1002 USA
TEL + 1-202-862-5600 • FAX + 1-202-467-4439 • EMAIL [email protected]
www.ifpri.org
9780896297203 51500

Design by: Diana Flores