Update on the NAR Model …
… Several Weeks before the First Results
Since the beginning of 2006, WoAgri has been working to develop a new, international economic model devoted to agriculture – the NAR model – through its team of economists. Previous articles1 have discussed its overall architecture and the main principles behind it. The NAR model will be released in three successive versions, the first of which, almost completely finished at this point, is described below.
Indeed, the development of the NAR model is progressing at a satisfactory rate and, as forecast in 2006, we will be able to provide the first simulations in early 2008.
Now, we can begin gradually to lift the precautionary measures of providing few details on the model's status, which we chose to implement in the interest of scientific rigor. These precautions were established to prevent us from falling into the same traps as the standard models we had studied, which produce completely unfounded results.
This was our reasoning for going to the United States in search of think tanks and scientists last June, and in providing an update on the NAR model through this document, several weeks before the first results become available.
I. Summary of factors guiding the development of the NAR model and its objectives
A. Why develop an international economic model devoted to agriculture – the NAR model?
International negotiations rely on simulations from standard economic models. These models, including the World Bank's widely used Linkage model, are "mock-ups" of reality. They are true advertising tools, used to shape and guide opinion, and they have therefore overstepped the boundaries of their initial role as a decision aid tool.
The problem is that they are biased, inappropriate and inexact reference tools:
- Biased …
Most economic models used in international negotiations (World Bank, FAPRI, OECD, etc.) are developed on the basis of the same principles of pure and perfect competition. This means that optimal conditions are achieved when all obstacles to the free circulation of agricultural products (market access, internal support and export subsidies) are removed. The WTO therefore relies heavily on the quantitative results obtained from these models to justify its strategy.
- Inappropriate …
One of the main arguments put forward by the WTO in favor of liberalizing the international agricultural markets is that it would help to improve conditions in poor countries. These models, however, are incapable of demonstrating that result, since they rely on the hypothetical "representative consumer" (meaning that all consumers are identical in terms of income and preferences). Their very construction prevents the models from studying the effect of liberalization on poverty!
International institutions, and particularly the WTO, make these models produce conclusions that they are incapable of producing. The latest audit (December 2006) performed at the request of François Bourguignon, World Bank Chief Economist, supports this analysis.
"The World Bank too often uses the results of its research to proselytize on behalf of its own policy, often without taking a balanced view of the evidence, and without expressing appropriate skepticism" (Angus Deaton, chair of the audit team and Professor of Economics at Princeton, January 2007).
- Inaccurate …
The difficulties with the Doha Round are presented as mainly affecting poor countries, since, according to the results of the World Bank's Linkage model, these countries are supposed to be the main beneficiaries of liberalizing agricultural trade.
However, these results are:
> Unpredictable: The successive estimates and reassessments provided by the World Bank Model quoted 850 billion in 2004, 350 in 2005, and were eventually revised down to 25 billion in 2006 (in other words, only $1 per inhabitant per year over 10 years!).
> Too low and too far below their margin of error (between five and ten percent of global GDP) to be able to confirm that liberalization would help these countries to rise out of poverty. Indeed, the potential gains put forward represent less than one percent of global GDP.
It was therefore essential to develop a reliable agricultural model that would place agriculture at the center of an international cooperation meeting the challenge of the 21st century.
That is the purpose of the NAR model.
B. Objectives of the NAR model
The NAR model, which has been in development since early 2006, aims to achieve six objectives:
> Account for the strategic importance of agriculture while ensuring more harmonious development of all of the world's agriculture systems.
> Account for the close interaction between agriculture and related issues, such as food security, the environment, poverty and innovation.
> Create a more reliable model of real agricultural conditions around the world (price volatility, farmer behavior in the face of risk, coexistence of different types of agriculture, etc.).
> Serve as a support tool for making policy decisions.
> Optimize each region’s degree of independence from the rest of the world in terms of agricultural potential, innovation and economic balance.
> Provide various countries, particularly France and the European Union, with a work and negotiation tool to promote international governance of agriculture.
2. An ambitious development schedule in view of obtaining initial results in early 2008
A. Architecture of the NAR model
To meet these objectives, we chose a comprehensive, balanced, calculable model based on a modular, interactive and controlled open source approach.
Modular: Unlike the standard models, which are built in a single block, the NAR model is made up of several modules, thus avoiding a system that is at once labyrinthine and a "black box," like most of the existing models. Furthermore, users will have more flexibility in interpreting the results because the links between the modules are clearly explained.
Interactive: One of the basic premises of the NAR model is that agriculture interacts directly with the environment, development, innovation and food sovereignty. It is therefore essential for the different modules in the NAR model to be interconnected (the "Game Theory" use philosophy).
Controlled open source: Like Linux in the computing field, the principles behind the model will be known, and experts from around the world, in developed and developing countries alike, will be able to participate in improving it under the supervision of an international scientific committee.
The NAR model comprises a balanced, comprehensive and calculable central module, linking six other modules that represent the impact and effects of three strategic aspects:
> The environment
> Economic growth.
A seventh module – the "Risk" module – is directly connected to the relevant agricultural sectors.
None of the models used today takes into account all of these criteria with this type of architecture.
B. Initial results expected in early 2008
The NAR model as a whole (the Central module and the seven peripheral modules) will be completed during the first half of 2008. It will be released in three successive versions (V1 to V3), enabling us to access the first simulations starting in early 2008.
The first version of the NAR model (V1) will be completed by January 2008. NAR V1 will include the Central, Risk and Independence modules (the latter is a decisive element in the results of each simulation) along with political, economic and financial variables.
The NAR V2 model will be finished by March 2008, and include the NAR V1 model in addition to the Poverty, Innovation and Environment modules. These modules, which are almost finished for the most part, will be covered more in depth in a later document.
The NAR V3 model, which is the final version, will be finished by the end of the first half of 2008, and include the NAR V2 model as well as the Intergeneration and Growth and Global Changes modules.
This means that we will be able to provide the first simulations based on the comprehensive NAR model by the end of the first semester of 2008.
3. In anticipation of the first results, a few aspects of the NAR V1 model …
A. Central module
The Central module is the heart of the NAR model. It is a balanced, comprehensive and calculable model that accounts for demographic growth.
All of the equations that make up the Central module have been written. They total between 3,500 and 23,000 depending on the number of geographical zones and economic sectors under consideration.
Initially, the NAR model will include three regions, but we will soon be able to expand our analysis to a dozen geographical zones. The geographical division we intend to use for the first version of the NAR model differs from the division usually considered by international models, based on GTAP2 data. It will use a division that reflects the groups that are active at the WTO on the agricultural issue and distinguish between the major developed countries (North America, EU, Japan), emerging countries of the G20 (Brazil, China, India, etc.), and least developed countries. Other innovations have also been included, related to formalizing and allocating investments (distinction between gross fixed capital formation and inventory change) and the function of household consumption.
B. Risk module
In the NAR model, the Risk module serves to model the high volatility of prices for agricultural products, taking into account the specific characteristics of agriculture:
> Prices are much more volatile than in industry or services, making them very difficult to forecast.
> Decisions on current production (and not only investment decisions) made in year "T-1" are almost completely irreversible until year T.
> The coexistence of the production of physical goods and forward and futures markets (derivative products) theoretically permits complete or partial coverage of risks, but mainly promotes speculation.
By formalizing these three factors, an approach never before adopted, the NAR V1 model will be able to demonstrate that a large portion of risk is endogenous, meaning that price volatility for agricultural raw materials is due to the specific characteristics of the market (anticipation, distribution structures, speculation, etc.) and causes its own volatility.
Almost all of the equations that make up the Risk module have been written, and WoAgri economists are currently in the computer programming and robustness testing phase for the NAR V1 model. Our initial results are very encouraging, and will be described in working papers and several scientific publications.
We look forward to presenting these initial results in early 2008, along with their political and strategic implications for agricultural policy at the national, European and international levels.
|WOAgri Editorial Department |
1See articles published on our site in the Proposals and Missions section entitled "Principles of construction and global architecture of the future NAR model" (March 15, 2006); The NAR model (October 2, 2006); Risk management in the NAR model: A decisive step in agricultural modeling (December 4, 2006).
2GTAP (Global Trade Assistance and Production) is the reference database for trade modeling. Created in 1993 on the initiative of Thomas Hertel (University of Purdue, Indiana), it is improved regularly (6 versions since 1993) through a network of more than 3,700 experts located in more than 100 countries.