PEP Standard CGE Models

Macro-micro development policy modeling group assists developing country researchers in constructing models of their national economy to simulate the impact of macroeconomic shocks/policies on various dimensions of poverty and welfare. To do so, it applies a combination of macro-micro modelling and simulation techniques including, for the macroeconomic level approach, a Computable General Equilibrium (CGE) framework. CGE models account for the structural aspects of a country’s economy, i.e. the interactions among sectors and institutions, and their links with the global economy.

Over time, the leaders of the research group came to find that there was a need for reference models that were more elaborate and closer to real-life conditions than the ones that had been used so far. Moreover, policymakers face new challenges that call for impact assessments that look both forward in time and beyond national boundaries, to the global economy. And so four lead researchers of the modeling group (below) devoted time and energy to creating a series of standard CGE models that can respond to these needs.

The growing family of PEP standard CGE models

The first of these models, PEP 1-1, designed for country-level studies, was developed as an operational tool for researchers to easily adapt a relatively standard model to their national economy.

Building on this basic model, PEP 1-t was created to include evolution in time. Technically speaking, PEP 1-t is a "recursive dynamic" version of PEP 1-1 (which is a static or single-period model): it extends the analysis to multiple periods, linking each one to the past through variables inherited from the previous period. 

Since 2011, three PEP standard CGE models have been put online:

  • PEP-w-1 is a single-period WORLD model and PEP-w-t is its recursive dynamic version.
  • The latest innovation is PEP-w-t-fin,
    • To our knowledge, this is the only recursive dynamic world model that includes international financial assets.
       

Free public access under a Creative Commons license

All of these models are truly open source: free public access is available under a Creative Commons license (see below). Each model is fully documented and all the relevant programs are freely accessible. Better still, with the kind permission of GTAP, a 14-region, 4-commodity, aggregated database is also made available to calibrate the new world models.

PEP is proud to offer the international modelling community, models that are not only fully operational for applied studies, but also perfectly suitable for training. Indeed, the team made painstaking efforts to produce documentation that is complete, including references to theoretical underpinnings and detailed mathematical developments that link model equations and calibration procedures with the theory. 
 

Models

Access each model, including detailed information, related tools and files:

PEP-1-1 (single-country, static version) v2.1

Authors: Bernard Decaluwé, André Lemelin, Véronique Robichaud, Hélène Maisonnave

The PEP-1-1 model (1 period – 1 country) is the first major output of a project that emerged spontaneously from the long-standing association between the co-authors. They propose a static computable general equilibrium (CGE) model designed for the study of a national economy and intended to be an operational tool for PEP researchers and other users. With it, researchers will be able to develop a relatively standard model, and easily apply it to their country, whatever the particular structure of their social accounting matrix (SAM). Quoting the co-authors:

PEP-1-1 is to be our basis, from which to further deepen our understanding of CGE analysis and develop modeling techniques that will tackle new problems. But, both in sharing our experience and exploring new paths, we want to remain in the realm of operational model building. So our intentions are all at once pedagogical, experimental, and practical.

The present model differs significantly from Decaluwé, Martens and Savard’s EXTER model, which has been used extensively in the past by PEP researchers who had been trained in one of the many modeling “Schools” held over the years in many parts of the world. In many respects, the PEP-1-1 model is richer than the more pedagogical EXTER.

First, the PEP-1-1 model distinguishes several categories of workers and of capital. Also, PEP-1-1 is capable of taking into account a broader set of tax instruments, and it models all possible transfers between institutions (agents). Finally, the GAMS code, presented in a separate document, has been written in a general form, thanks to the use of sets. This will facilitate the application of PEP-1-1 to variously aggregated SAMs, by means of a few simple steps to make the SAM directly readable into the GAMS program.

Readers are invited to send their comments to André Lemelin at: andre.lemelin@inrs.ca

Access PEP-1-1 GAMS code, user guide, debugator and ppt presentation

Debugator (based on PEP-1-1)

This document provides some hints to debug a Computable General Equilibrium (CGE) model written in GAMS. The aim of this document is to provide a methodology to help the readers debug their own errors. The document tries to be systematic and pedagogical.

Some errors have been introduced in the model. They are a sample of the type of mistakes that may be encountered while running the GAMS code, such as compilation errors, execution, calibration and specification errors. This document is not an exhaustive list of all the different problems that may be encountered. The intention is to guide the reader on how to debug his or her code using hypothetical examples.

For a document that deals with using GAMS in general (not specifically applied to CGE), please refer to McCarl (2009) or Rosenthal (2008).

Download PEP-1-1 GAMS debugator

PEP-1-t (single-country, recursive dynamic version) v2.1

Authors: Bernard Decaluwé, André Lemelin, Véronique Robichaud, Hélène Maisonnave

The PEP-1-t model follows the PEP-1-1 single-country, single-period model. It is a recursive dynamic model, the second research tool of the Standard PEP Model project that emerged spontaneously from the long-standing association between the co-authors.

With this model, we pursue the goals that we had initially set ourselves: to crystallize our joint CGE modeling experience, and to share the result with the PEP MPIA Network, and with the modeling community at large. In addition, PEP-1-t is one step towards a deepening of our understanding of CGE analysis and the development modeling techniques that will tackle new problems.

We nonetheless persist in wanting to remain in the realm of operational model building. In that respect, we are fortunate that three members of the team have been involved in a Unicef-Financial and Fiscal Commission project set up to analyse the impact of the crisis on child poverty in South Africa. This was an opportunity to apply a version of PEP-1-t adapted to the South African context and the project objectives.

The adapted model was subjected to extensive maltreatment; that flushed out a few bugs which we promptly corrected. In the end, all were quite happy with the model’s performance. As usual, we have tried to eliminate as many errors and "misprints" as possible in this document. Needless to say, we welcome comments that will help us improve either the model or its presentation.

Readers are invited to send their comments to André Lemelin at: andre.lemelin@inrs.ca

Download PEP-1-t GAMS code, user guide and ppt presentation

PEP-w-1 (multi-region, single-period world model) v4.0

Authors: André Lemelin, Véronique Robichaud, Bernard Decaluwé and Hélène Maisonnave

This paper documents the PEP-w-1 model developed by four members of the PEP research network. It is the third research tool of the Standard PEP Model project that emerged spontaneously from the friendship and long-standing association between the co-authors.

The PEP-w-1 model extends the PEP-1-1 single-country, single-period model to a worldwide multiregional model. With this model, we pursue the goals that we had initially set ourselves:

  • to crystallize our joint CGE modeling experience,
  • to share the result with the PEP modeling group, and with the modeling community at large.

In addition, PEP-w-1 is another step towards a deepening of our understanding of CGE analysis and the development modeling techniques that will tackle new problems.

In a sense, however, PEP-w-1 is a transitional model, a stepping stone towards the recursive dynamic PEP-w-t, and its successor PEP-w-t-fin, with international financial assets.

Version 4.0 replaces version 3.3. Changes from the previous version are considerable:

  • PEP-w-1 version 4.0 uses the GTAP 8.1 database, whereas version 3.3 used GTAP 7
  • Changes in the model itself
  • Changes in the parametrization procedure
  • Formulations in the GAMS code were also altered to clarify, uniformize and streamline the code in several instances 
  • The interested user will find details in the “What’s new in version 4.0?” section of PEP-w-1 v4.0.pdf below.

The authors wish to thank the Global Trade Analysis Project (GTAP: www.gtap.agecon.purdue.edu) for granting us permission to distribute our 14 region, 4 industry aggregation of the GTAP 8.1 data base. As usual, we have tried to eliminate as many errors and  "misprints" as possible in this document. Needless to say, we welcome comments that will help us improve either the model or its presentation.

Readers are invited to send their comments to André Lemelin at: andre.lemelin@inrs.ca

Download PEP-w-1 GAMS code and user guide

PEP-w-t (multi-region, recursive dynamic world model) v4.1

Authors: Véronique Robichaud, André Lemelin, Bernard Decaluwé and Hélène Maisonnave

The PEP-w-t model extends the PEP-w-1 worldwide multi-regional model to a recursive dynamic model. Its dynamics is similar to the dynamics of PEP-1-t, insofar as the GTAP database permits. PEP-w-t is the fourth research tool of the Standard PEP Model project that emerged spontaneously from the friendship and long-standing association between the co-authors. With this model, we pursue the goals that we had initially set ourselves: to crystallize our joint CGE modeling experience, and to share the result with the PEP MPIA Network, and with the modeling community at large. In addition, PEP-w-t is another step towards a deepening of our understanding of CGE analysis and the development modeling techniques that will tackle new problems.

Version 4.1 replaces versions 4.0 and 1.4. PEP-w-t version 4.1 uses the GTAP 8.1 database. Changes since version 1.4 have been considerable. With respect to version 4.0, version 4.1 is an update. For that reason, the version 4.0 document remains valid and has not been replaced. But we urge users to take a look at the document entitled “The PEP standard multi-region, recursive dynamic world CGE model: Update, with a reference scenario”.     

The authors wish to thank the Global Trade Analysis Project (GTAP: www.gtap.agecon.purdue.edu) for granting us permission to distribute our 14 region, 4 industry aggregation of the GTAP 8.1 data base. As usual, we have tried to eliminate as many errors and "misprints" as possible in this document. Needless to say, we welcome comments that will help us improve either the model or its presentation.

Readers are invited to send their comments to André Lemelin at: andre.lemelin@inrs.ca

Download PEP-w-t GAMS code and user guide

PEP-w-t-fin (recursive dynamic world model with international financial assets) v4.1

Authors: André Lemelin, Véronique Robichaud and Bernard Decaluwé

The PEP-w-t-fin model extends the PEP-w-t recursive dynamic world model to take into account international financial assets. PEP-w-t-fin is the fifth research tool of the Standard PEP Model project that emerged spontaneously from the friendship and long-standing association between the co-authors. With this model, we pursue the goals that we had initially set ourselves: to crystallize our joint CGE modeling experience, and to share the result with the PEP MPIA Network, and with the modeling community at large. In addition, PEP-w-t-fin is another step towards a deepening of our understanding of CGE analysis and the development modeling techniques that will tackle new problems.    

PEP-w-t-fin has already been applied to examine some implications of the 2008-2011 financial and economic crisis. Preliminary results have been presented at the Thirteenth Annual Conference on Global Economic Analysis, Penang, Malaysia, 9-11 juin 2010 (Lemelin et al., 2010). The model has also been the object of an article in Economic Modelling (Lemelin et al., 2013).

Version 4.1 replaces all previous versions of our model with financial assets, and it introduces considerable changes, in addition to be based on the GTAP 8.1 database. Like PEP-w-t-fin2, version 4.1 of PEP-w-t-fin (we decided to drop the “2”) takes into account the bilateral structure of international financial assets, on the basis of constructed data.    

The authors wish to thank the Global Trade Analysis Project (GTAP: www.gtap.agecon.purdue.edu) for granting us permission to distribute our 14 region, 4 industry aggregation of the GTAP 8.1 data base. As usual, we have tried to eliminate as many errors and "misprints" as possible in this document. Needless to say, we welcome comments that will help us improve either the model or its presentation.

Readers are invited to send their comments to André Lemelin at: andre.lemelin@inrs.ca

Download PEP-w-fin GAMS code and user guide

PEP SAMBAL (simple computer program to balance a SAM) v1.0

Authors: André Lemelin, Ismaël Fofana and John Cockburn

The Modelling and Policy Impact Analysis (MPIA) program assists developing country researchers in constructing models of their national economy to simulate the impact of macroeconomic shocks/policies on various dimensions of poverty and welfare. To do so, it applies a combination of macro-micro modelling and simulation techniques including, for the macroeconomic level approach, a Computable General Equilibrium (CGE) framework. CGE models are usually built on the basis of data assembled in a Social Accounting Matrix (SAM), which needs to be balanced. On this page, modelers will find SAM-balancing tools which have been developed or adapted by PEP researchers.

SAMBAL is a simple computer program to balance a SAM. Computable general equilibrium modeling requires a consistent and coherent benchmark data set, usually organized in the form of a Social Accounting Matrix (SAM). These data generally come from quite diverse sources and correspond to different periods of time. As a result, they often present inconsistencies. SAMBAL makes it possible to reconcile this information in order to balance a SAM. The program minimizes the changes to the base data using one of two optimizing techniques, cross-entropy (the default) or least squares. The program is an attractive and easy alternative to the arbitrary and time-consuming manual and other haphazard methods sometimes used to balance SAMs.

Download SAMBAL

GPCEMA (General Purpose Cross-Entropy Matrix Adjustment) v1.0

Author: André Lemelin

GPCEMA stands for "General Purpose Cross-Entropy Matrix Adjustment". The matrix adjustment program is based on the minimum information-gain principle (also known as minimum cross-entropy), as it has been extended to deal with negative entries by Junius and Oosterhaven (2003). In addition to handling standard matrix adjustment problems, the program automatically deals with situations where row sums, column sums, or both are constrained to zero. It is based on 
LEMELIN, André (2009), "A GRAS variant solving for minimum information loss", Economic Systems Research, 21(4), p. 399-408.

Related resources include :

Download GPCEMA

Authors

Bernard Decaluwe

Bernard Decaluwe

Former Resource Person - Macro/Modeling Group

Professor, Université Laval
Canada

Andre Lemelin

André Lemelin

Former Resource Person - Macro/Modeling Group

Associate Professor (titulaire retraité)
Institut National de la Recherche Scientifique (INRS) - Urbanisation Culture Société
Canada

FUNDED BY

Logo global affairs canada
Logo Hewlett Foundation
Logo Co-Impact
Logo BMGF
Logo IDRC - CRDI Canada
Logo Mastercard Foundation