What Should be the Focus of Agricultural Policy Reforms in Sub-Saharan Africa? A CGE Analysis

Abstract

While a lot of research has been conducted on agricultural subsidies and other forms of policy transfers in developed and developing countries alike, substantial data constraints have characterised those conducted in developing countries. For this study, we employ a novel and uniquely developed dataset on these policies in Sub-Saharan Africa (SSA), to analyse the impact of policy reforms, using the latest available GTAP 9.1 Data Base, in the widely employed GTAP framework, for the first time. We simulate the scenarios of removal of output subsidies, removal of ‘market development gaps’ within and outside the country. Our results indicate that removing market development gaps is likely to increase the agricultural output without affecting trade much, while removing the subsidies could harm output a lot by import-substitution of the costly domestic output. We conclude that governments in SSA may do well to focus on developing their markets better rather than cutting the assistance to their farmers, which could in fact be counter-productive instead of raising the efficiency of domestic farmers through competition.

JEL Classification: C68, Q11, Q13, Q17, Q18

Keywords

Africa GTAP Policy Support Market Failures,Agriculture,Sub-Sahran Africa

1. Introduction

Most countries adopt policies in support of their agricultural sectors that generate significant trade, market and price distortions. While a lot of research has been conducted on agricultural subsidies and other forms of policy transfers in developed and developing countries alike (Alston & Hurd, 1990; Alston, Sumner, & Vosti, 2008; Legg, 2003; Mohanty, Pohit, & Roy, 2002; Monke & Pearson, 1989; Sadoulet & De Janvry, 1995; Swinnen, 1994; Wise, 2004), substantial data constraints have characterised those conducted in developing countries. Following a first experiment by Krueger, Schiff, and Valdes (1988), Anderson and Valenzuela (2008) followed by Anderson and Nelgen (2010) bridged this gap in the literature by estimating these distortions to agricultural incentives in developing countries.

Concerted efforts by the Monitoring and Analyzing Food and Agricultural Policies (MAFAP) team at the Food and Agricultural Organization (FAO) have resulted in the updating of this kind of a dataset covering the period 2005–2013. The MAFAP programme is implemented by the FAO and seeks to introduce country-owned and sustainable systems to monitor, analyse and reform food and agricultural policies to set up more effective, efficient and inclusive policy frameworks in developing and emerging economies.

Exploiting this dataset from the MAFAP, we extend the standard GTAP Data Base (version 9.1, year 2011) to include domestic support, export subsidies, export restrictions as well as detailed transaction costs, in a few commodity-specific value chains or in the agricultural sector as a whole for some developing countries. The countries currently covered by the MAFAP include Bangladesh, Burkina Faso, Benin, Burundi, Ethiopia, Ghana, Kenya, Malawi, Mali, Mozambique, Nigeria, Rwanda, Senegal, South Sudan, Tanzania and Uganda. However, only Burkina Faso, Ethiopia, Ghana, Kenya, Malawi, Mali, Mozambique, Tanzania and Uganda are included in this study. We employ the Altertax tool (Malcolm, 1998) in the GTAP framework to incorporate the MAFAP indicators on policy support.

As a further attempt to fill a major gap in the policy literature, in this paper we perform several real-world policy simulations involving the agricultural sectors of these selected developing countries. Most of these MAFAP countries have adopted agricultural or trade policies or budgetary transfers to stimulate agricultural production and productivity growth in an attempt to achieve food self-sufficiency, or rather self-food reliance, in the wake of the high food price crises of 2007/08 and 2010/11. These policies include a wide range of measures from highly distortive administered producer or consumer prices, to border protection or export restrictions, to slightly less distortive inputs subsidies, and more WTO-compatible types of direct or indirect transfers to agents or groups of economic agents to support marketing, research, extension, infrastructures (feeder roads, storage facilities, etc.).

This paper is organised in the following way: Section 2 describes the dataset developed and used, Section 3 summarises the methods used to incorporate the MAFAP into the GTAP, Section 4 explains the policy simulations undertaken, Section 5 shows the results and Section 6 concludes.

2. Data Sources and Methodology

2.1 Overview of the MAFAP Data and Indicators

The MAFAP methodology includes computation of indicators for price-incentive analysis and public expenditure analysis. The data requirement and sources are described for the two methodology components.

The MAFAP price incentives analysis is commodity and country specific. The MAFAP methodology on price incentives allows generating five commodity-specific indicators: (i) price gap; (ii) nominal rate of protection (NRP); (iii) effective rate of protection (ERP); (iv) nominal rate of assistance (NRA); and (v) the market development gap (MDG). The first two are calculated at three points along the value chain: retail, wholesale and farm gate, while the other three are only calculated at the farm gate level. All indicators are calculated using two different types of data: observed and adjusted. Observed indicators include all direct taxation over the specific commodity, while the adjusted indicators account for all indirect taxation and market inefficiencies as well. The MDG captures the gap between the observed and adjusted measures. However, due to data limitations in estimating the effective rates of protection (ERP) for all commodity and country pairs, we only report the price gaps, NRP, NRA, and MDG in this study.

While the MAFAP database currently includes 15 Sub-Saharan countries, data for only 10 of them are considered for the period 2005–2013. The study does not include Nigeria due to data constraints. Similarly, the nominal rates of protection at retail level are not released because the required information is not yet available for all commodities and countries.

Indicators are computed on an annual basis from 2005 onwards and updated regularly. Commodities to be analysed are selected on the basis of their contribution to the country’s food security, import bill and export revenue. Products with a high potential in promising or emerging value chains such as onion, beans cashew nuts, coffee or tea are also taken into account. All the indicators are publicly available and the data elements are available on request (see Table 1).

Table 1.
Categories of MAFAP

Indicators and Variables

Publicly available Available on request

Price gaps
Nominal rates of protection
Nominal rates of assistance
Market development gap Benchmark prices
Access costs
Output prices
Input prices
Quantities

Indicators and Variables
Price gaps Nominal rates of protection Nominal rates of assistance Market development gap	Benchmark prices Access costs Output prices Input prices Quantities

Source: www.fao.org/in-action/mafap

Table 2.

MAFAP Country Coverage

Burkina Faso	Ghana	Ethiopia	Kenya	Malawi	Mali	Mozambique	Tanzania	Uganda
Live cattle	Cassava	Barley	Cassava	Cotton	Cattle	Cashew nuts (processed)	Cashew nuts	Cattle
Cotton	Maize	Coffee	Cotton	Groundnuts	Cotton	Cashew nuts (raw)	Coffee	Cassava
Maize	Palm oil	Haricot beans	Dry beans	Maize	Groundnuts	Cassava	Maize	Coffee
Onion	Rice	Lentils	French beans	Sugar	Maize	Cotton	Rice	Cotton
Rice	Yam	Live cattle	Maize	Tea	Milk	Maize		Cow milk
Sesame		Maize	Potatoes	Tobacco	Millet	Rice		Maize
Sorghum		Sorghum	Rice		Rice			Rice
		Teff	Sorghum		Sorghum			Sugar cane
		Wheat	Sugar cane					Tea
		Sesame	Tea					Wheat
			Wheat

Source: www.fao.org/in-action/mafap

Table 2 gives an overview of the countries and commodities currently covered by the MAFAP programme as used in this paper. Most of these value chains were analysed up to 2013 (Angelucci, Balié, Gourichon, Mas Aparisi, & Witwer, 2013). The price data is usually collected in the rural wholesale markets in each country by the national authorities. The MAFAP project does not typically collect primary data, but relies on national institutions to do so. However, the MAFAP project team undertakes specific data mobilisation surveys and/or missions to fill critical gaps in data. Moreover, this team promotes consultations and exchanges of experiences among participating countries to adopt common approaches for commodity and market selection as well as data management and estimation of indicators. As such, all data entered in the MAFAP database are subject to automatic and manual robustness and consistency checks before they are published. After validation, data are published on the FAO/MAFAP website (www.fao.org/in-action/mafap) with profile webpages for each MAFAP participating country that display the data in various graphs and tables for users. The data displayed are comparable and consistent across countries. The data inputs required for the computation of indicators for each commodity and potential sources are provided hereafter.¹

The benchmark prices are annual nominal prices of the commodity at the country’s border, where the commodity is imported or exported. When higher frequency data are available (such as quarterly, monthly, or daily), the annual average is computed. For a net import, the benchmark price is the cost of insurance and freight price. For a net export, the benchmark price is the free on board price. The sources usually include the UN Comtrade database, the FAOSTAT trade database or national sources (e.g., ministry of trade or statistics).

The exchange rate is the annual average of the nominal exchange rates between the local currency and USDs. The main sources are the International Monetary Fund database and the World Bank’s World Development Indicators database.

Domestic prices at the wholesale, farm gate or retail levels are annual nominal prices. When higher frequency data are available (such as quarterly, monthly or daily) the annual average is computed. The point in the value chain where the domestic product competes with the internationally traded product is called the point of competition. For a net export, this may be the price at the international auction or the price at the border. For a net import, the price at the point of competition may be the price at the main wholesale market where the product is traded. The nominal producer price is sought at the commodity’s main production area. Usually the main source of data comes from the national level (such as commodity boards, producer organisations, the ministry of agriculture, planning or trade or the statistics bureau). The MAFAP occasionally also uses the FAO GIEWS database as well as the CoutrySTAT database.

Access costs from the border to the point of competition, from the farm gate to the point of competition and from the point of competition to retail are also necessary to compute the MAFAP indicators. Marketing costs include the costs of border clearance, storage and handling, inland transport, government taxes and fees, bribes, marketing margins of traders and/or processors, etc. The main sources of data include national sources (e.g., commodity boards, producer organisations, and the ministries of agriculture, statistics, planning or trade), private companies (i.e., processors, estates, etc.), value chain and marketing studies/publications, and also the ‘Enabling the Business in Agriculture’ database for port costs. Data are usually collected with annual frequency even if in some cases this is not possible and therefore they are inferred using the available information from previous years.

Average quantity and quality conversion ratios are used to render products comparable at various stages of the value chain. The quantity conversation ratio relates to the volume of a given commodity generated from one unit of raw inputs of the same commodity. This information is only relevant for those commodities that undergo processing between the farm gate and point of competition or between the border and point of competition. For example, the conversion ratio for rice is equal to the volume (tons) of milled rice produced per one unit (ton) of paddy rice. Similarly, the quality conversion ratio accounts for quality differences between the domestic product and the internationally traded product. Indeed, in order to compare like with like between the internationally and domestically produced commodities, a conversion accounting for the quality difference needs to be made. For example, if most domestic milled rice is 30 per cent broken and milled rice imports are 100 per cent whole, the quality conversion ratio would be 0.7. Thus, the import price for milled rice would be multiplied by 0.7 to make it comparable to the price for domestic milled rice. For the quantity and quality conversion ratios, the national sources (e.g., commodity boards, producer organisations and ministries of agriculture, statistics, planning or trade) including private companies (i.e., processors, estates, etc.) are usually preferred.

Finally, the MAFAP methodology enables the incorporation of data obtained in the analysis of public expenditure into the price incentives analysis to construct an indicator that captures public spending in support of a specific commodity or group of commodities, in addition to policy and market performances already captured by the NRP. Combining price and budget information, the NRA provides a more complete picture of incentives, particularly in cases where budgetary payments may be compensating for disincentives to producers. Data on input subsidies (in nominal, unit prices) for producers of the commodities selected for analysis are obtained from the MAFAP public expenditure database and analysis also available through the MAFAP website (www.fao.org/in-action/mafap).

The MAFAP’s public expenditure analysis is conducted at the aggregated and disaggregated levels. At the aggregate level, MAFAP seeks information on overall national public expenditure, and on budgeted and actual allocations. The analysis is disaggregated at the programme and project levels, includes budgeted and actual amounts, accounts for spending at the central and decentralised levels, regardless of the implementing agency/ministry, and incorporates both donor and government outlays on-budget and off-budget. These data are obtained for the period starting with the fiscal year 2005/06.

Under the MAFAP, public expenditures for agriculture include: (i) agriculture-specific expenditure, which are expenditures on individual agricultural agents (e.g., input subsidies), or on a sector as a whole (e.g., agricultural research); and (ii) agriculture-supportive expenditures, which are expenditures in support of rural development, such as on rural infrastructure, rural education and rural health, as these also have an important role in indirectly supporting agricultural sector development. All the measures that comply with these criteria are considered. General public expenditure measures throughout the economy are not considered, even if they generate monetary transfers to the agricultural sector, and private expenditures are not considered.

The MAFAP public expenditure analysis also requires qualitative information on the budgetary process in the country: the institutional architecture together with a detailed explanation of the functioning of the budget. Moreover, it calls for a thorough description of all the policy projects and programmes that will be considered in the analysis: their objectives, activities, status of implementation, commodities targeted, and the level of government implementing the project. Each measure needs to be well documented to facilitate MAFAP classification, for seven years before the period of analysis.

The sources of information include the ministries of finance, agriculture and planning in the different countries. Moreover, when possible, the MAFAP project team also seeks to obtain data from donors for off-budget expenditures, and data from other ministries/agencies for expenditures not recorded in the ministry of finance or agriculture’s budget such as Presidential Initiatives, for instance.

2.2 The GTAP Model

In this study, we rely on the comparative static regional general equilibrium model GTAP also called the standard model, the framework of which is well documented in Hertel (1997), also illustrated in Figure 1. The standard GTAP model is a multiregion, multisector, computable general equilibrium model, assuming perfect competition and constant returns to scale. Bilateral trade is treated following the Armington assumption that products traded internationally are differentiated by country of origin. Like other CGE models, this model generates smaller and more realistic responses of trade to price changes, than usually obtained through models of homogeneous products. The GTAP model is well documented in the GTAP book and in various research and technical papers (e.g., Jensen, 2010; Narayanan, Aguiar, & McDougall, 2015). It allows for a wide range of closure options, including unemployment, tax revenue replacement and fixed trade balance closures. A few partial equilibrium (PE) closures are also available to users, which enable comparison of the results with studies based on PE assumptions more easily.

Figure 1.

Source: GTAP model, Purdue University.

The version used in this study contains innovations including (i) the treatment of private household preferences using the non-homothetic constant difference elasticity (CDE) functional form, (ii) the explicit treatment of international trade and transport margins, and (iii) a global banking sector which intermediates between global savings and consumption.

The GTAP9 database used in this study includes data on consumption, production and trade that are globally consistent for three benchmark years of 2004, 2007 and 2011. We employ the 2011 version of this database. The GTAP database is composed of input–output (I–O) statistics contributed by members of the GTAP network. These I–O tables allow us to capture the inter-sectoral linkages within each country in the GTAP model. The GTAP 9 database includes separate I–O tables for 120 individual countries representing 98 per cent of global gross domestic product (GDP) and 92 per cent of the world’s population, along with 20 composite regions which aggregate smaller economies. The list of countries included as individual regions in GTAP is provided in Figure 2. Countries in white background are captured as individual regions in the database while countries in a dark background are aggregated into ‘rest of’ the world.

In the version 9 of the GTAP database used in this study, economic activities correspond to 57 products and services following the United Nations Central Product Classification and the International Standard Industrial Classification. The data base includes 14 agricultural sectors (including forestry and fishing), four mineral extraction sectors, eight food processing sectors, 16 manufacturing sectors, three utility distribution sectors (construction, wholesale and retail trade), three transportation sectors, six service sectors and dwellings.

Figure 2.

Source: GTAP, Purdue University.

In this study, all the policy instruments are represented as ad valorem tax equivalents that create wedges between the prices that prevail in the international market and that are assumed to be undistorted and prices that include the effects of policies in domestic markets. Domestic support is usually intended as the combination of market price support and budgetary payments following the OECD PSE methodology. In this analysis, we focus on market price support while budgetary transfers are omitted unless they are clearly tight to one commodity in the form of a commodity-specific input subsidy, for example. The latter case implies that both the market price-support component and the commodity-specific budgetary transfer could be captured in the NRA.

Moreover, it should be noted that disentangling market price support from the border measures goes beyond the scope of this paper. Hence, when dealing with domestic support issues, we focus on the behavioural equations in the production technology representation of the standard GTAP model (production tree). For a specific production activity, this production tree combines intermediate inputs and primary factor inputs such as land, labour, capital and natural resources applying a nested structure. The production technology tree is shown in Figure 3. This figure shows the nested structure of the model. Intermediate inputs and value added are combined to produce one output following a Leontief function. The value added aggregates the production factors following a CES function (ESUBVA). The lower nest of the production function aggregates imported intermediate inputs from different regions following a CES function (ESUBM), while the upper nest determines the combination of domestically produced inputs and imported ones (ESUBD). While capital and labour are considered mobile, natural resources and land are treated as sluggish endowments. The primary factors of production (land, capital, labour and natural resources) are assumed to be fully employed in each region. These factors are not allowed to move across regions. It should be noted that the elasticity of substitution between factors is much smaller for the agricultural sectors than the other sectors (0.26 as opposed to more than 1.05). In other words, the model assumes an inelastic substitution of production factors in the agriculture sector. Moreover, land is a factor that is considered specific to agriculture—and which has the effect of dampening the supply response of sectors requiring land. As a result, an increase in the demand for agriculture production will result in an increase in the land price. As such, growth in agriculture production associated with constant land supply will require substituting land by other primary factors.

Figure 3.

Source: Adapted from Hertel (1997).

2.3 Incorporating MAFAP Indicators into the GTAP Data Base

In order to incorporate the MAFAP-based taxes/subsidies data into the GTAP, we employ the widely used Altertax tool (Malcolm, 1998). This entails changing the taxes and subsidies in the dataset, without affecting the balance and other parts of the data base. The closure and elasticity parameters are designed to suit the requirement of changing nothing other than the taxes and subsidies. A virtue of this tool is that it can preserve the balance as well as the initial economic structure of the data base, with the only changes being the tax rates that are targeted.

The GTAP 9.1 data base (Narayanan et al., 2015) is an assembly of trade, protection, input-output, consumption and macro-economic datasets from various established sources across the world. We incorporate the MAFAP indicators into the GTAP database to enable us to work on simulations that reflect real world policies, using 2011 for these variables. For all the variables in the MAFAP dataset, we identify the corresponding variables in the GTAP framework, so as to modify the GTAP data base accordingly.

We specifically make the following assumptions:

If NRAs are positive, they are treated like output subsidies;

If MDGs are negative, they are treated like output taxes; and

If MDGs are positive, they are treated like import tariffs, uniform across partners as they provide additional protection to farmers.

In this study, we augment the GTAP database with data from the MAFAP.² Table 2 shows how GTAP sectors are mapped to MAFAP sectors, while Tables 3 and 4 provide an overview of countries and commodities covered by the MAFAP programme that we are focussing in this study. It is important to note that only a subset of the MAFAP dataset and resulting indicators are used in the current study.

Tables 3 and 4 provide some descriptive statistics on the average NRAs and MDGs over the period analysed in this study. In Table 3, figures in bold indicate very high positive values for NRAs suggesting a high level of support. In Table 4, figures in bold indicate very negative MDGs suggesting substantial market inefficiencies for a given commodity/country pair.

Table 3.
MAFAP Border Average NRAs (%) for 2005–2013

Burkina Faso Ethiopia Ghana Kenya Malawi Mozambique Tanzania Uganda

Rice 6.8 −14.9 27.7 −27.8 83.3 87.4

Wheat 48.3 5.1 −2.5 −19.4

Corse grains −47.0 −36.5 −39.1 −3.8 68.6 −1.8 −20.4 50.7

Vegetables and fruits 6.9 −24.6 9.7 −3.6 −3.8 −3.6 15.5

Oilseeds −39.2 −13.3 −34.3 −22.0 −12.7 8.7

Sugar 48.2 3.5 −32.9

Other crops −33.7 −6.4 −9.4 −53.8 4.5

Beef & sheep meat −22.9 −54.9 −30.8

	Burkina Faso	Ethiopia	Ghana	Kenya	Malawi	Mozambique	Tanzania	Uganda
Rice	6.8		−14.9	27.7		−27.8	83.3	87.4
Wheat	48.3	5.1		−2.5				−19.4
Corse grains	−47.0	−36.5	−39.1	−3.8	68.6	−1.8	−20.4	50.7
Vegetables and fruits	6.9	−24.6	9.7	−3.6		−3.8	−3.6	15.5
Oilseeds		−39.2	−13.3	−34.3	−22.0	−12.7		8.7
Sugar				48.2	3.5			−32.9
Other crops		−33.7		−6.4	−9.4		−53.8	4.5
Beef & sheep meat	−22.9	−54.9						−30.8

Source: www.fao.org/in-action/mafap

Table 4.

MAFAP Border Average MDGs (%) for 2005–2013

	Burkina Faso	Ethiopia	Ghana	Kenya	Malawi	Mozambique	Tanzania	Uganda
Rice	24.1		9.7	2.8		−1.5	−2.7	−10.4
Wheat	−8.9	−0.1		2.6				0
Corse grains	−29.7	−3.0	8.4	−1.3	−7.9	−0.4	−9.9	−3.2
Vegetables and fruits	−7.8	−5.9	−18.2	2.8		−5.0	−18.4	−10.7
Oilseeds		4.9	−10.9	−18.5	−13.6	−10.1		−2.4
Sugar				−65.0	−2.6			−75.9
Other crops		9.5		−3.0	−26.2		−4.6	−7.6
Beef and sheep meat	−9.6	−33.8						−26.7

Source: www.fao.org/in-action/mafap

We also compute the output shares of granular FAO sectors in more aggregated GTAP sectors, using the mappings that we developed between the FAO and GTAP sectors, while incorporating the NRAs and MDGs into the GTAP data base.

2.4 Choice of Policy Simulations

Analysis of the MAFAP dataset suggests that factors other than only trade and price policies explain the general pattern of production disincentives across commodities in Sub-Saharan Africa (SSA). We consider that most of the explanation for these frequent disincentives to production lies in the type and mix of policy measures adopted by governments. Indeed, border policies that are generally favourable to producers are often combined with excessive market access costs (transport, handling, storage, margins, etc.), that reveal important inefficiencies or even underdevelopment of the value chains (World Bank, 2008), which have tended to lower prices received by producers. However, there were offsetting mechanisms for producers resulting from other forms of support and primarily budgetary transfers, such as input subsidies. Consistent with previous findings (Bates, 1981; Demeke et al., 2009; Maetz et al., 2011; Schultz, 1964), the main issue is the lack of policy coherence and transparency to improve market price signal transmissions to farmers.

In this context, a number of specific policy issues have been examined in the literature, including agricultural inputs subsidies, border protection through tariff and non-tariff measures, and targeted investment in agricultural infrastructure in order to reduce market access and transaction costs. While the use of inputs by farmers is constrained by several factors, including the lack of access to credit, high cost of inputs (generally imported), price variability, and high market and financial risks (Demeke et al., 2014; Dorward & Chirwa, 2011), it is recognised that input subsides are costly for generally scarce national budgets, often not sustainable and typically not effective at improving productivity in the long run (Jayne, Mather, & Mghenyi, 2010).

Apart from this type of budgetary transfer from tax payers to farmers, governments often also use changes in tariffs, tax increases or exemptions to affect price levels for farmers. These policies have been typically analysed using CGE models in order to account for cross-sector linkages and distributional effects. Taking into account market-specific transaction costs as well as missing markets, we examine whether output taxation combined with fertiliser subsidies, as is usually the case in most MAFAP countries, is welfare-enhancing compared to a policy scenario where subsidies have been discontinued. Furthermore, considering that farmers in MAFAP countries have tended to receive lower prices than those prevailing in international markets, we test the potential welfare-enhancing effects for both net food buyers and net sellers of better international price transmission as a result of further trade liberalisation. Finally, in reference to the commitments made by African government in the Maputo (2003) and the Malabo (2014) declarations to allocate at least 10 per cent of the national budget to agriculture, we investigate the impact and fiscal implications of an increase in agricultural spending to 10 per cent of the national budget with an exclusive focus on indirect support to the agricultural sector (e.g., on public goods such as agricultural research, feeder and rural roads, extension and training, etc.). These features are captured as a reduction in transaction costs and/or a rise in productivity in agricultural sectors owing to such developments.

The aim of this paper is to shed some light on the likely impacts of a few policy reform scenarios through policy simulations. Primarily, we consider three alternative scenarios:

Removal of NRAs provided to the farmers;

Removal of MDGs within the country, which hinder output expansion; and

Removal of MDGs along with the border protection, which impede imports.

For all our scenarios, we employ the standard GTAP model closure, which assumes perfect competition in all markets, full employment of all factors and zero profits.

3. Results

Based on our simulation design and the data we have collected on distortions and subsidies in the framework of the MAFAP project, a reduction in the subsidy in each sector in each country separately, would theoretically cause a fall in output and exports due to an increase in prices, and a resultant increase in imports, in a PE framework. This may also happen in a CGE framework if each sector in each country is shocked separately. However, the combination of all these countries reducing their subsidies simultaneously results in multiple possible outcomes. For example, when subsidies are higher in one country (A) than the other (B), it may be possible that country B may gain in comparison to country A in sectors where it has lower subsidies, since the removal of subsides means a lower increase in prices in country B and a higher increase in prices in country A.

We examine the results of the three simulated scenarios on sectoral output, imports, exports and prices for all the SSA countries in our dataset. Table 5 shows the results of removing NRAs on output. The effects are largely negative and huge. However, we observe a moderate increase in output for some sectors particularly in Mozambique, Malawi and Kenya. This is because these countries have little or no NRAs to begin with, and hence, with hardly any price rise (Table A3), they emerge as more competitive than others when the other countries remove their NRAs.

Table 5.
Removal of NRAs: Percentage Change in Output, GDP and Welfare

Region Burkina Faso Ethiopia Ghana Kenya Malawi Mozambique Tanzania Uganda

GDP (Real) −0.99 0.01 −28.21 −0.12 0.03 −1.65 −0.56 −0.58

Welfare (equivalent variations as a % of initial income) 5.16 0.03 −32.92 −0.10 0.02 −2.09 −1.43 −0.78

Agriculture Rice −100 −0.19 −35.34 11.11 3.42 6.12 −61.95 −98.96

Wheat −54.78 −0.2 38.3 −4.71 2.16 7.24 7.38 −100

Veg/fruits 328.77 1.23 −92.87 −11.45 0.87 −34.81 1.91 −1.47

Sugar −2.52 0 −16.96 0.52 −0.62 0 −9.01 −1.46

Cotton −75.68 0 16.35 2.02 1.72 15.41 13.68 52.96

Live cattle −3.66 0.25 −65.49 0.36 0.09 −0.99 0.35 8.23

Other grains −2.78 0.02 −64.08 −0.51 1.45 −2.48 −3.93 −12.95

Oil seeds −43.17 −0.47 −46.76 0.1 −1.04 11.84 1.52 24.95

Other crops −14.58 −2.14 195.43 1.53 −2.04 21.28 17.36 −52.45

Milk −3.35 0 −22.58 0.51 0.21 3.4 1.18 −3.71

Meat livestock −3.05 −0.02 −52.69 0.27 0.1 0.78 2.13 7.44

Non-agriculture Extraction −5.9 −0.01 26.2 0.22 −0.04 1.32 5.27 4.02

Processed food −4.87 0 −99.06 0.54 0.3 0.64 −15.68 −3.79

Textwapp −3.44 −0.09 51.26 0.09 −1.06 1.42 12.85 11.76

Light manufacturing −10.3 −0.11 81.56 0.28 0.39 5.07 10.08 12.88

Heavy manufacturing −17.86 −0.17 114.95 0.1 −0.01 8.3 13.51 12.64

Utilities consumption −2.51 0.02 −18.73 0.28 −0.01 3.36 −0.94 0.67

Transport Communication −1.78 −0.1 18.09 0.12 0.11 1.68 1.06 2.26

Other services 5.58 −0.02 0.49 1.13 −0.01 2.84 2.79 3.21

	Region	Burkina Faso	Ethiopia	Ghana	Kenya	Malawi	Mozambique	Tanzania	Uganda
GDP (Real)		−0.99	0.01	−28.21	−0.12	0.03	−1.65	−0.56	−0.58
Welfare (equivalent variations as a % of initial income)	5.16	0.03	−32.92	−0.10	0.02	−2.09	−1.43	−0.78
Agriculture	Rice	−100	−0.19	−35.34	11.11	3.42	6.12	−61.95	−98.96
	Wheat	−54.78	−0.2	38.3	−4.71	2.16	7.24	7.38	−100
	Veg/fruits	328.77	1.23	−92.87	−11.45	0.87	−34.81	1.91	−1.47
	Sugar	−2.52	0	−16.96	0.52	−0.62	0	−9.01	−1.46
	Cotton	−75.68	0	16.35	2.02	1.72	15.41	13.68	52.96
	Live cattle	−3.66	0.25	−65.49	0.36	0.09	−0.99	0.35	8.23
	Other grains	−2.78	0.02	−64.08	−0.51	1.45	−2.48	−3.93	−12.95
	Oil seeds	−43.17	−0.47	−46.76	0.1	−1.04	11.84	1.52	24.95
	Other crops	−14.58	−2.14	195.43	1.53	−2.04	21.28	17.36	−52.45
	Milk	−3.35	0	−22.58	0.51	0.21	3.4	1.18	−3.71
	Meat livestock	−3.05	−0.02	−52.69	0.27	0.1	0.78	2.13	7.44
Non-agriculture	Extraction	−5.9	−0.01	26.2	0.22	−0.04	1.32	5.27	4.02
Processed food	−4.87	0	−99.06	0.54	0.3	0.64	−15.68	−3.79
	Textwapp	−3.44	−0.09	51.26	0.09	−1.06	1.42	12.85	11.76
	Light manufacturing	−10.3	−0.11	81.56	0.28	0.39	5.07	10.08	12.88
	Heavy manufacturing	−17.86	−0.17	114.95	0.1	−0.01	8.3	13.51	12.64
	Utilities consumption	−2.51	0.02	−18.73	0.28	−0.01	3.36	−0.94	0.67
	Transport Communication	−1.78	−0.1	18.09	0.12	0.11	1.68	1.06	2.26
	Other services	5.58	−0.02	0.49	1.13	−0.01	2.84	2.79	3.21

Source: Authors’ calculations from GTAP model analysis

The GDP and welfare effects of the NRA removal are significant and negative for most countries. Burkina Faso loses about 1 per cent of its GDP but gains 5 per cent in terms of welfare, because of the consumer side benefits of removing the NRAs which operate as a tax on consumption. Ethiopia and Malawi register small positive effects while in Kenya the effects from removing NRAs are small and negative. The most notable loss is registered by Ghana that may lose about one-third of its GDP and welfare by removing NRAs, since many of its agricultural sectors could decline significantly as a result. This result shows that the economy of Ghana appears to be still quite dependent on the agriculture sector.

The reason why output generally declines is evident from Tables A1 and A4 of the annexes as well as Figures 6 and 7, which document a widespread decline in exports due to a considerable rise in prices (shown in Figure 4), attributable to the removal of subsidies (Table A2). We indeed observe that the price surges are more important in those countries where NRAs were high (see Table 6 on descriptive statistics). The rise in prices is especially impressive for rice in Burkina Faso (above 1,000%), but also for Uganda and to a lesser extent, Tanzania.

Table 6.

Sectors of Modelling Focus

Sectors Modelled	Corresponding GTAP Sectors^a	Aggregated Sectors for Reporting Purposes
Paddy and processed rice	pdr pcr	Grains
Wheat	wht	Grains
Coarse grains	gro	Grains
Vegetables and fruits	v_f	Veg and fruit
Oilseeds	osd	Oilseeds and vegetable oil
Vegetable oils	vol	Oilseeds and vegetable oil
Raw and processed sugar	c_b sgr	Other crops
Other crops	ocr	Other crops
Cattle, sheep and beef	ctl cmt	Beef and sheep/lamb meat
Milk and dairy	rmk mil	Dairy

Source: Authors’ aggregation of the GTAP database.

Note: ^a www.gtap.agecon.purdue.edu/databases/v9/v9_sectors.asp

Figure 4.

Source: Authors based on results from GTAP model analysis

Figure 5.

Source: Authors based on results from GTAP model analysis

Besides rice, the price increase is generalised across commodities in Burkina Faso at levels well above 45 per cent for most commodities. However, we also observe a slight decline in prices across the board in Kenya, Malawi, Mozambique and Tanzania and more pronounced declines in Ghana. We find tiny sectors with almost no production are wiped out in some cases, like wheat in Ghana.

In the GTAP model, changes in ad valorem taxes or subsidies are fully passed on to prices, and hence we see the complete effect of these changes on prices. Further, because of the rise in prices, which happens in many sectors and countries especially Burkina Faso and Uganda, we also observe an increase in imports; with some exceptions (see 5 below and Table A6 in the annexes).

Most of the large increases seen in the charts correspond to very small initial values, meaning that these changes do not refer to large changes in absolute terms. For example, as seen in Figure 6, sugar exports from Ghana increases by 571 per cent from a small initial level of exports, which was almost zero (US$0.00009 million or US$90!). Similarly, Ghana’s NRA is rather high in the vegetables and fruits sector, but non-existent in other sectors. Therefore, while the vegetables and fruits sector of Ghana is practically eliminated by the price rise induced by removing NRAs, other sectors remain competitive and hence expand so much that some of the decline in aggregate output due to the catastrophic fall in vegetables and fruits sector is offset. For this type of effect to happen, a large expansion in other sectors, which are small to begin with, is necessary. Such substantial effects can be seen in the tables.

Figure 6.

Source: Authors based on results from GTAP model analysis

Table A1 of the annexes and Figure 9 below indicate the impact on output of a reduction in the MDG. Such a reduction would basically consist in a reduction of market inefficiencies resulting from excessive transportations costs ultimately borne by farmers, rents captured by middlemen depressing producer prices, or illicit tax collection practices imposed on farmers when accessing markets. The results obtained are mixed. We note that in general, changes resulting from a removal of a MDG are less important in magnitude but also much more positive than the results obtained from removing NRAs.

The output results shown here vary between largely positive at best and slightly negative at worst. On average across countries, it appears that removing the gaps in market development is far more effective in maintaining a similar sectoral mix than removing NRAs. Table A1 of the annexes shows that the GDP and welfare effects of the MDG removal are negligible in comparison to the NRA removal. This suggests that the MDG removal could benefit sectors without any cost to the overall economy, while the costs of the NRA removal are substantially higher, as we discussed before.

Rice in Burkina Faso seems to be an outlier, particularly sensitive to the removal of market inefficiencies. The strong effects on rice are expected for a landlocked country where distance to the port operates like a natural protection (raising the cost of imports) and rice production is encouraged, as a result. Farmers have experienced positive nominal rates of protection (NRPs) between 2005 and 2013, although these have been substantially less important than those received by wholesalers. This is indicative of the role of the market inefficiencies in hampering price transmission along the value chain such as for example an increase in output price in world or regional markets. Such an effect on price signals is captured by the MDG.

Figure 8 indicates that there is not a very large flooding of imports (See Table A5). This can be seen as positive from the perspective of import protection without a need of raising tariffs. The figure also shows a drastic reduction in imports of rice in Burkina Faso which is consistent with the massive output expansion. It is also interesting to note the sharp decline in imports of vegetable and fruits in Kenya, as domestic production increases to meet the demand.

Figure 7.

Source: Authors based on results from GTAP model analysis

Figure 8.

Source: Authors based on results from GTAP model analysis

Figure 7 shows that exports do not change dramatically either. Kenya and Burkina Faso are major exceptions to this general observation. Burkina Faso witnesses a massive expansion in exports of rice while this increase is more modest for fruits and vegetable in Kenya.

Table A7 (in the Annexes) shows, again, a mixed picture for price. We observe a low increase in prices to a major fall in prices, in most countries except Ethiopia, Ghana, Malawi and Tanzania, which experience small price reductions. Overall, we can say that domestic MDG removal is a far more positive scenario compared to the removal of NRAs. We however note one exception with exports which surge in general, much like in the NRAs removal scenario. Therefore, important MDGs appear to have stronger and positive export-enhancing effects, due to the changes in relative prices in different countries.

Figure 9.

Source: Authors based on results from GTAP model analysis

4. Conclusions

Given the role of the agricultural sector in the economy of most developing countries, particularly in SSA, it is important to evaluate the effects of agricultural policy reforms. However, owing to the complexities of these policies and lack of data, there have not been many comprehensive economy-wide analyses of these reforms especially in SSA. In this paper, we explain the methodology used to construct a novel and unique dataset that includes agricultural policies, MDGs and distortions to agricultural production incentives. For the first time, we employ the MAFAP dataset developed by the FAO, in a global CGE framework to analyse the impact of policy reforms in SSA countries.

We examine a couple of scenarios and we find that removing NRAs has quite visible and somewhat catastrophic output-reducing effects in many countries. This is due to price rises and import surges. However, exports increase in some countries and offset the negative effects to some extent. Moreover, we also find that removing the MDGs is less disruptive than removing the NRAs. We observe largely small effects of such a policy. Furthermore, the removal of NRAs leads to huge macro-level impacts on GDP and welfare, while the cost of removing MDGs appears negligible at the macro level. We conclude that, for most of the countries studied, removing the MDG is a far more positive scenario compared to the removal of NRAs.

Therefore, our results suggest that, for government interested in supporting production and market participation by farmers, it would be more effective to focus on eliminating or at least reducing market inefficiencies that translate into very high costs for producers, rents captures by intermediaries and illicit practices, eventually lowering farmers’ margins, rather than to target reforms of explicit trade and price policies. However, removing the MDGs is not a straightforward policy. It would imply substantial investments in roads and other public goods to reduce the huge market inefficiencies primary related to the substantially higher transport costs and marketing margins observed in SSA compared to other economies. Achieving such massive investments appears to be a daunting task for most governments with notorious limited budget capacities. Future research could explore the net effects on the economy of the MDG removal and the public investment required to eliminate the MDG.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Appendix

Table A7.

Removal of Domestic MDGs: Percentage Change in Prices

Region	Burkina	Ethiopia	Ghana	Kenya	Malawi	Mozambique	Tanzania	Uganda
Region	Faso	Ethiopia	Ghana	Kenya	Malawi	Mozambique	Tanzania	Uganda
Rice	−27.24	−0.02	−0.01	−2.47	−0.04	0.06	−0.01	0.69
Wheat	0.99	−0.03	−0.01	−2.98	−0.01	−0.06	−0.01	0
VegFruits	0.92	−0.04	−0.01	−13.75	−0.05	0.09	−0.03	0.85
Sugar	1.1	−0.03	−0.01	1.7	−0.03	0.11	−0.01	0.83
Cotton	0.67	−0.03	0	0.03	−0.02	0.06	0	0.64
LiveCattle	1.11	−0.03	−0.01	1.55	−0.05	0.09	−0.02	0.7
OthGrains	1.06	−0.03	−0.01	−1.94	−0.06	−3.19	−0.01	0.85
OilSeeds	0.91	−0.03	−0.01	1.48	−0.04	0.1	−0.01	0.79
OthCrops	−3.84	−0.03	−0.02	−0.38	−0.04	0.05	−0.02	−0.97
Milk	1.1	−0.03	0	2.03	−0.05	0.08	−0.02	0.77
MeatLstk	0.9	−0.03	−0.01	0.61	−0.04	0	−0.01	0.57
Extraction	0	−0.02	0	0.32	0	0	0	0.03
ProcFood	0.07	−0.02	−0.01	0.64	−0.04	−0.12	−0.01	0.43
TextWapp	0.06	−0.02	0	0.47	−0.03	0.03	0	0.29
LightMnfc	0.06	−0.02	0	0.5	−0.02	0.03	0	0.24
HeavyMnfc	0.04	−0.01	0	0.33	−0.02	0.02	0	0.2
Util_Cons	0.09	−0.02	−0.01	0.58	−0.02	0.02	0	0.21
TransComm	0.07	−0.02	−0.01	0.57	−0.02	0.02	0	0.28
OthServices	0.15	−0.02	−0.01	0.64	−0.03	0.02	0	0.31

Source: Authors’ calculations from GTAP model analysis

Notes

References

Alston

J. M.

, & Hurd

B. H.

(1990). Some neglected social costs of government spending in farm programmes. American Journal of Agricultural Economics, 72(1), 149–156.

Alston

J. M.

, Sumner

D. A.

, & Vosti

S. A.

(2008). Farm subsidies and obesity in the United States: National evidence and international comparisons. Food Policy, 33(6), 470–479.

Anderson

, & Valenzuela

(2008). Estimates of global distortions to agricultural incentives, 1955 to 2007. Washington, DC: World Bank.

Anderson

, & Nelgen

(2013). Updated national and global estimates of distortions to agricultural incentives, 1955 to 2011. Washington, DC: World Bank.

Angelucci

, Balié

, Gourichon

, Mas Aparisi

, & Witwer

(2013). Monitoring and analysing food and agricultural policies in Africa. MAFAP synthesis report 2013. Rome, Italy: MAFAP Synthesis Report Series, FAO.

Bates

R. H.

(1981). Markets and states in tropical Africa: The political basis of agricultural policies. Berkeley, CA: University of California Press.

Demeke

, Pangrazio

, & Maetz

(2009). Country responses to the food security crisis: Nature and preliminary implications of the policies pursued. Rome: FAO.

Demeke

, Spinelli

, Croce

, Pernechele

, Stefanelli

, Jafari

, & Roux

(2014). Food and agriculture policy decisions: Trends, emerging issues and policy alignments since the 2007/08 food security crisis. Food and Agriculture Organization of the United Nations (FAO).

Dorward

, & Chirwa

(2011). The Malawi agricultural input subsidy programme: 2005/06 to 2008/09. International journal of Agricultural Sustainability, 9(1), 232–247.

10.

Hertel

T. W.

(Ed.). (1997). Global trade analysis: Modeling and applications. Cambridge University Press.

11.

Jayne

T. S.

, Mather

, & Mghenyi

(2010). Principal challenges confronting smallholder agriculture in sub-Saharan Africa. World Development, 38(10), 1384–1398.

12.

Jensen

H. G.

(2010). Chapter 10G: EU domestic support data for GTAP 7.1 database. In Narayanan

G. B.

& Walmsley

T. L.

(Eds.), Global trade, assistance and production: The GTAP 7 database. Center for Global Trade Analysis, Purdue University.

13.

Krueger

A. O.

, Schiff

, & Valdez

(1988). Agricultural incentives in developing countries: Measuring the effect of sectoral and economy wide policies. The World Bank Economic Review, 2(3), 255–272.

14.

(Eds.). (1991). The political economy of agricultural pricing policy (Volume 3: Africa and the Mediterranean). Washington: The World Bank.

15.

Legg

(2003). Presidential address agricultural subsidies: Measurement and use in policy evaluation. Journal of Agricultural Economics, 54(2), 175–201.

16.

Maetz

, Aguirre

, Kim

, Matinroshan

, Pangrazio

, & Pernechele

(2011). Food and agricultural policy trends after the 2008 food security crisis. Renewed attention to agricultural development. Rome: FAO.

17.

Malcolm

(1998). Adjusting tax rates in GTAP data base. GTAP Technical Paper No. 15.

18.

Monke

E. A.

, & Pearson

S. R.

(1989). The policy analysis matrix for agricultural development (No. 04; HD1415, M5). Ithaca: Cornell University Press.

19.

Narayanan

G. B.

, Aguiar

, & McDougall

(Eds.). (2015). Global trade, assistance, and production: The GTAP 9 data base. Center for Global Trade Analysis. Lafayette: , Purdue University.

20.

Sadoulet

, & De Janvry

(1995). Quantitative development policy analysis (Vol. 1). Baltimore: Johns Hopkins University Press.

21.

Mohanty

S. K.

, Pohit

, & Roy

S. S.

(2002). Implications of Agricultural Trade Liberalisation in Selected Industrialised Countries on Trade Prospects of South Asian Countries: A CGE Modelling Analysis. In Regional Conference on Globalisation and Agriculture: Challenges for South Asia (pp. 4–5).

22.

Schultz

T. W.

(1964). Transforming traditional agriculture. Transforming traditional agriculture. New Haven and London: Yale University Press

23.

Swinnen

J. F.

(1994). A positive theory of agricultural protection. American Journal of Agricultural Economics, 76(1), 1–14.

24.

Wise

T. A.

(2004). The paradox of agricultural subsidies: Measurement issues, agricultural dumping, and policy reform. Medford: Tufts University.

25.

World Bank (2008). World development report 2008: Agriculture for development. Washington DC: The World Bank.

Indicators and Variables
Publicly available	Available on request
Price gaps Nominal rates of protection Nominal rates of assistance Market development gap	Benchmark prices Access costs Output prices Input prices Quantities

What Should be the Focus of Agricultural Policy Reforms in Sub-Saharan Africa? A CGE Analysis

Abstract

Keywords

1. Introduction

2. Data Sources and Methodology

2.1 Overview of the MAFAP Data and Indicators

Table 1. Categories of MAFAP Indicators and Variables Publicly available Available on request Price gapsNominal rates of protectionNominal rates of assistanceMarket development gap Benchmark pricesAccess costsOutput pricesInput pricesQuantities

3. Results

Footnotes

Declaration of Conflicting Interests

Funding

Appendix

Notes

References

Table 1.
Categories of MAFAP

Indicators and Variables

Publicly available Available on request

Price gaps
Nominal rates of protection
Nominal rates of assistance
Market development gap Benchmark prices
Access costs
Output prices
Input prices
Quantities