Model Selection Tool

Sample policy questions to guide model selection from the AGRODEP Model Library


What is the impact of export, import or production taxes/subsidies imposed at various stages of a specific value chain?

Trade restrictions or domestic taxes/subsidies can be imposed along various stages of a value chain. For instance, a tax or a subsidy can be applied at the raw material stage and another tax or subsidy at the processing stage. To examine the impact of such policies, one needs a well-designed multimarket model able to represent a multiproduct-value chain (crushed seeds, feed, oil and biodiesel for instance for an oilseed value chain).

The Export Restrictions And import Tariffs Overall impacts (ERATO) model was developed to examine the impact of trade restrictions imposed along three stages of a value chain. It can be extended to include more stages in the analysis. The model is in its structure a static multimarket partial equilibrium model, which does not require huge data for calibration. It is worth noting that in addition to trade restrictions, the model can assess the impact of productivity shocks as well.


What is the (ex-post) impact on trade flows of a Regional Trade agreement or a monetary union?

To study the impact on trade flows of a trade agreement or the formation of a monetary union, the gravity model is an interesting tool. First it allows the user to work at the disaggregated level (HS6 for instance). Second it requires a limited amount of data (trade flows, GDP, distance, tariffs mainly) compared to other types of models.  Third the trade agreement or the monetary union can be easily modeled through dummy variables (although some endogeneity issues may arise). While gravity models present a lot of advantages compared to PE or CGE models, they cannot capture all the effects of trade agreements such as trade deviations and terms of trade effects.


What is the impact of a trade agreement on imports and exports at the (disaggregated) tariff line level?

Trade agreements are often negotiated at a very detailed level and one is interested in the impacts at those levels of sectorial disaggregation. CGE models are calibrated to social accounting matrices which are in turn built on national accounts. Thus, SAMs barely cover more than 50 sectors. To study the impacts of trade liberalization at the detailed product level, one needs a partial equilibrium model based mainly on trade data which are available at a high level of disaggregation (5,000 products at the HS6 level).

The Partial Equilibrium Trade Simulation (PETS) model is the type of model designed to study the impact of trade agreements on import and export flows at the finest level of both sectorial and geographical disaggregation. Compared to gravity models, it can capture trade deviation and terms of trade effects. Its recursive dynamic structure also allows the user to simulate trade reforms that should be implement gradually.

Spatial equilibrium models also constitute an efficient analysis tool to evaluate at the disaggregated level the potential effects of a change in trade and transportation costs on market, trade and welfare variables (supply and demand, producer and consumer prices, volume and direction of trade, consumer and producer surpluses). Trade and transportation costs are often important in Africa and constitute variables policy makers usually try to reduce through trade liberalization, simplification in administrative or regulatory practices or improving infrastructure. The Spatial equilibrium model developed in AGRODEP library is a special class of partial equilibrium model allowing the user to work at the subnational level (if necessary) with disaggregated data on transportation costs. They are particularly interesting when working with neighboring countries for which good estimates of transportation costs are available.

Although, not a pure spatial model, the Economy-Wide Multi-market (EMM) model offers a good alternative to the user interested in studying the impact of reducing trade costs, with more detailed representation of the supply side of the economy.


What is the economywide wide impact of changing a particular domestic policy instrument (tax reform, input subsidy…) or implementing an investment plan such as the National Agricultural Investment Plans (NAIPs) ?

There are a series of domestic policies that are important with general equilibrium effects but for which the impact on third countries can be summarized (aggregated). Those policies include tax reforms, input or output subsidies and taxes, productivity shocks and public investment plans. For these type of policies with a focus on the national economy (impact on various households’ categories for instance), single country computable general equilibrium models are appropriate.  They can be static or dynamic depending on the type of policy under consideration. For instance, the introduction of a tax in one particular year would require a static model while the gradual implementation of a productivity increase to catch up with a technology frontier would require a dynamic version.

In the library, the IFPRI Standard model, the PEP 1-1 and PEP 1-T models are interesting single country CGE models available to the user for domestic policy analysis. One could also envisage the use of the MIRAGRODEP model  if policy changes have regional implications or shift comparative advantages.


What is the impact on poverty and inequality of changing a domestic policy instrument (tax reform, unilateral trade liberalization) or simulating an agricultural productivity shock?

The use of CGE models to study policy shocks on national economies has become a standard approach. Yet CGE models calibrated to aggregate Social Accounting Matrices (SAMs) are insufficient to account for impacts at the household level and to analyze the effects on income distribution.  Indeed, even highly disaggregated SAMs contains hardly 10 categories of households.  To address this gap, micro models, which incorporate detailed information at the household level (household surveys data), should be linked to the macro model.

The Micro-macro distributional analysis toolbox developed by AGRODEP contains the codes used to run microsimulations linked to CGE models in a top down approach. It allows the users to study the impact on poverty and inequality of various policy scenarios simulated with CGE models.


What is the impact of climate change (Shared Socioeconomic Pathways) on global food production and consumption?

Climate change is one of the most important emerging issues in the 21st century. Agriculture is going to be affected mainly through changes in yields due to changing rainfall conditions. Adaptation and mitigations strategies will also play an important role facing climate change and various narratives (scenarios) have been developed to describe how humanity would deal with it (Shared Socioeconomic Pathways).

The IMPACT suite of Models built around a core global partial equilibrium multimarket model of agricultural production, demand, trade, and prices is well designed to study scenarios involving climate change and mitigation measures. The suite includes an explicit water module to investigate the long-term dynamics of how water demand and availability would affect future food production and a land use model to handle competing demands for land and changes in greenhouse gas emissions due to land-use change in future analysis.


What is the impact of a currency devaluation on core macroeconomic variables (GDP, investment, inflation and unemployment)?

Many interesting topics in economics (monetary policy, exchange rate policy…) require the use of a macroeconomic model to be thoroughly covered. Here CGE models which are generally focused on the real sector of the economy cannot help. One rather needs well designed macro-financial models able to tackle issues related to the nominal part of the economy.

AGRODEP Macroeconometric modelling toolbox allows the user to study various macro policy shocks  using the Eviews software. It combines macroeconomic theory and time series techniques in a coherent framework and does not require a significant amount of data. In most of the cases the WDI database of the World Bank is enough to calibrate the model.