Sunflower Value Chain Enhancements for the Rural Economy in Tanzania: A Village Computable General Equilibrium-CGE Approach

Poverty is prevalent and widespread in rural Tanzania, where agriculture is the main activity. The government is making significant public investments intended to speed the growth of agriculture as a means to accelerate inclusive economic growth. In line with public investments, the government is promoting public–private partnerships by encouraging the use of improved agricultural innovations and linking farmers to markets, seeking to increase their yields and income. However, there is a paucity of empirical evidence using multipliers analysis about the extent of how gains in agricultural productivity and market linkages for farmers in rural areas help improve the economy at the household level. This paper assesses the welfare effects of the sunflower value chain for a rural economy in Tanzania using a computable general equilibrium (CGE) model for the selected village, which has a high potential for sunflower. Findings highlight the use of the CGE model, first, for analyzing and understanding the economic sectors at a village level. Second, the effects of various upgrading strategies promoted for improving rural farming communities by the government and non-governmental development partners at the micro-scale are analyzed and potential agricultural commodity value chains identified. The multiplier analysis provided insights regarding the potential of sunflower crops for the village economy.


Introduction
In Sub-Saharan Africa, including Tanzania, agriculture-based economic growth in rural areas has great potential for reducing poverty compared to non-agriculture growth [1]. The agricultural sector employs about 80% of the rural population and contributes about 50% of the export earnings, while its share of Tanzania's GDP is about 25% [2,3]. However, poverty is still prevalent and widespread in rural Tanzania, with about 43.8% of residents living below the poverty line [4,5]. The agriculture sector predominantly consists of small-to medium-scale farmers and a small number of emerging large-scale farmers [5]. Moreover, these farmers have low productivity due to, among other reasons, poor agronomic practices [6]. Other hindering factors include a lack of expanding trade, costly transport to the remote areas due to poor infrastructure, and a lack of logistics, such as cooperatives that could facilitate vertical and horizontal marketing integration [7][8][9]. However, there is potential for improving and expanding crop cultivation in these areas, including simple moisture conservation techniques, like tied ridges, that can increase yields and productivity [3,6,10].
Interventions curbing these hindrances and ensuring food access, such as the establishment of income or employment generating activities for the poor, is even more crucial, as their labor is often

Study Area Selection and the Underlying Data
This study is based on primary data collected in 2014 by the Trans-SEC project baseline household survey. Thus, results can be directly linked to information from the household-level data set. The study was conducted in the Dodoma region, which is semi-arid (350-500 mm above sea level) and has a food system that is primarily based on the cultivation of millet, sorghum, and sunflower, as well as on raising livestock [28,29]. The region is particularly sensitive to food insecurity [3]. In order to identify an appropriate study site, a village scoping study was carried out. The selection of the case study site involved a range of criteria, including market access and the physical accessibility of the village, which has implications for the level of transaction costs incurred by farmers when selling their commodities [30,31]. Moreover, land availability for the production of food crops and sunflower as a potential cash crop was also taken into consideration when selecting the village [30]. Within this context, Idifu, a village located in Idifu ward, Chamwino district, was identified as an appropriate study site. The main outside markets are Mvumi (division center) and Dodoma (capital city of the region and of Tanzania), where most large-scale sunflower processors in the region are located.
Out of the 1205 households living in Idifu, 150 (12.5%) were interviewed in January and February 2014. The households were chosen randomly, based on household lists provided by the village executive officer. The respondent was the household head, who was asked the economic activities, and, at times, his wife, who was asked questions pertaining to the purchases of food items. In addition, a follow-up discussion was conducted with key informants in the village in order to cross-check and validate information. Among others, issues about nearby markets and the main commodities imported and exported from the village were discussed. The questionnaire focused on different aspects of farmers' participation in sunflower value chains (production, processing, as well as the trading of seeds, oil, and sunflower seedcake) and other agricultural commodities, including potential cash and food crops, such as sorghum, pearl millet, and maize. Input-output relationships covering all village transactions within and between households were modeled, with an emphasis placed on both subsistence consumption of agricultural food and cash crops, including all by-products and their uses (fertilizer, food, seed, seedcakes). Regarding other potential crops in the study village, small-scale farmers cultivate food crops, primarily pearl millet and sorghum. Imports mainly consist of maize, beans, and rice, while sunflower and livestock represent the major export commodities of the case study village. The information gathered built the basis for developing a village social accounting matrix (SAM).

The Village SAM Framework
The SAM provides a snapshot of the economy under investigation: the structure of production, inter-sectoral linkages, and the distribution of factor value added among different socioeconomic groups, as well as its links to the outside world [30,[32][33][34][35]. The village SAM provides a base for building and calibrating the Idifu village computable general equilibrium models ( Figure 1). 2018, 10,xFOR PEER REVIEW 4 of 21 potential cash and food crops, such as sorghum, pearl millet, and maize. Input-output relationships covering all village transactions within and between households were modeled, with an emphasis placed on both subsistence consumption of agricultural food and cash crops, including all byproducts and their uses (fertilizer, food, seed, seedcakes). Regarding other potential crops in the study village, small-scale farmers cultivate food crops, primarily pearl millet and sorghum. Imports mainly consist of maize, beans, and rice, while sunflower and livestock represent the major export commodities of the case study village. The information gathered built the basis for developing a village social accounting matrix (SAM).

The Village SAM Framework
The SAM provides a snapshot of the economy under investigation: the structure of production, inter-sectoral linkages, and the distribution of factor value added among different socioeconomic groups, as well as its links to the outside world [30,[32][33][34][35]. The village SAM provides a base for building and calibrating the Idifu village computable general equilibrium models ( Figure 1). The level of disaggregation in a village SAM depends on the analyst's preferences, research objectives, and data availability. For this study, production activities are disaggregated to include all on-farm activities, off-farm activities, and natural resource extractions. In addition to the production activities, separate commodity accounts that represent each of the major household consumption categories and the domestic product market are constructed. The exogenous accounts include the local government, enterprises/firms, and outside the village. The values entered in the village SAM are converted from the local currency, Tanzanian Shillings (TZS), into purchasing power parity US Dollars so that the welfare effects can be compared in an international context [31]. Calculation of the conversion factor used in converting all monetary values from TZS to PPP USD, the CPI for 2013, referenced 2010 [36,37].

Village CGE Model Framework
We use the integrated standard CGE models developed by IFPRI [38] to construct the Idifu CGE. Our village model exhibits the following characteristics: a "small" open economy assuming The level of disaggregation in a village SAM depends on the analyst's preferences, research objectives, and data availability. For this study, production activities are disaggregated to include all on-farm activities, off-farm activities, and natural resource extractions. In addition to the production activities, separate commodity accounts that represent each of the major household consumption categories and the domestic product market are constructed. The exogenous accounts include the local government, enterprises/firms, and outside the village. The values entered in the village SAM are converted from the local currency, Tanzanian Shillings (TZS), into purchasing power parity US Dollars so that the welfare effects can be compared in an international context [31]. Calculation of the conversion factor used in converting all monetary values from TZS to PPP USD, the CPI for 2013, referenced 2010 [36,37].

Village CGE Model Framework
We use the integrated standard CGE models developed by IFPRI [38] to construct the Idifu CGE. Our village model exhibits the following characteristics: a "small" open economy assuming households maximize a nested-linear expenditure system (LES) utility function on commodities either produced domestically (in the village) or imported from outside the village. Factors of production are capital (K i ) and labor (L i ), which combine using a constant elasticity substitution (CES) aggregator function to form the value-added and are assumed to be non-tradable and exogenously fixed at the household level but mobile among sectors. On the other hand, the intermediate inputs (X ij ) with a Leontief aggregator function combine with the value added by a Leontief technology function to form the domestic produced commodities (XD i ) that are tradable. Saving, investment, and unemployment are assumed to be endogenously determined. The government sector maximizes its utility using a Cobb-Douglas aggregator function, with endogenous taxes and flexible exchange rates.
Since we assume a small open economy in this model, various commodities (indexed by a subscript i) (XD i ) produced in the village are either sold to a domestic market (XDD i ) or to an export market (E i ). This transformation of domestic production is modeled according to the Constant Elasticity of Transformation (CET) function. Conversely, the village supply of the commodity will come from commodities produced in the village (XDD i ) using the Armington assumption, combined with commodities imported from outside the village (M i ), to form the composite commodities (X i ). These composite commodities are either used as intermediate inputs (X ij ) into the production process of the domestically produced commodities or sold for final consumption as household consumption (C i ), government consumption (CG i ), and investment (I i ) ( Figure 2). either produced domestically (in the village) or imported from outside the village. Factors of production are capital (Ki) and labor (Li), which combine using a constant elasticity substitution (CES) aggregator function to form the value-added and are assumed to be non-tradable and exogenously fixed at the household level but mobile among sectors. On the other hand, the intermediate inputs (Xij) with a Leontief aggregator function combine with the value added by a Leontief technology function to form the domestic produced commodities (XDi) that are tradable. Saving, investment, and unemployment are assumed to be endogenously determined. The government sector maximizes its utility using a Cobb-Douglas aggregator function, with endogenous taxes and flexible exchange rates.
Since we assume a small open economy in this model, various commodities (indexed by a subscript i) (XDi) produced in the village are either sold to a domestic market (XDDi) or to an export market (Ei). This transformation of domestic production is modeled according to the Constant Elasticity of Transformation (CET) function. Conversely, the village supply of the commodity will come from commodities produced in the village (XDDi) using the Armington assumption, combined with commodities imported from outside the village (Mi), to form the composite commodities (Xi). These composite commodities are either used as intermediate inputs (Xij) into the production process of the domestically produced commodities or sold for final consumption as household consumption (Ci), government consumption (CGi), and investment (Ii) (Figure 2).  In this model, we assume that households face the same utility function and prices. We then assume the households maximize the LES utility function instead of the CES function because the values of elasticities are not very realistic, so there are disadvantages to using the Cobb-Douglas utility function that assumes the income elasticities are equal to one, own price elasticities equal to minus one, while cross-price elasticities are equal to zero. A simple generalization of the Cobb-Douglas function, which is more flexible with respect to values of the elasticities, is the Stone-Geary utility function that leads to the Linear Expenditure System (LES).
This is maximized subject to the budget constraint given as: Interm.  In this model, we assume that households face the same utility function and prices. We then assume the households maximize the LES utility function instead of the CES function because the values of elasticities are not very realistic, so there are disadvantages to using the Cobb-Douglas utility function that assumes the income elasticities are equal to one, own price elasticities equal to minus one, while cross-price elasticities are equal to zero. A simple generalization of the Cobb-Douglas function, which is more flexible with respect to values of the elasticities, is the Stone-Geary utility function that leads to the Linear Expenditure System (LES).

Ci
This is maximized subject to the budget constraint given as: where first order condition and rearrangement reads so as to obtain the tangency condition. In addition, we use the budget constraint equation to obtain the relation of commodities as C i . Multiplying the commodity equation with respective prices, we obtain the expenditure PD i ·C i for each commodity consumed by the households in the village. The expenditures on the commodity, the left-hand side of the expenditure equation, can be divided into two terms: the first part PD i ·µH i is the minimum expenditure on commodity i to which the consumer commits himself in order to obtain a minimum subsistence level, µH i ; this is interpreted as the minimum required quantity that the consumer purchases first. Then, PD j ·µH j is the minimum expenditure on other commodities, so that Y − PD i ·µH i − PD j ·µH j is the income that remains after the consumer has purchased the minimum required commodities quantities. This marginal income is called "supernumerary income" and is allocated over all commodities based on fixed fractions. Assuming the prices of the commodities are equal to one for the LES function, the Frisch parameter is defined as the negative inverse of the fraction of supernumerary expenditure, Y − PD i ·µH i − PD j ·µH j in the total expenditure in the benchmark equilibrium situation.
The household's income from capital, labor, unemployment benefits, and other transfers is defined as Y, while savings are a fixed fraction of net income, so that in the presence of an endogenous income tax (ty), the household's demand for consumption of commodities follows the LES utility function that is maximized in the optimization process, subject to budget constraints that also account for the taxes on the consumption commodities, C i .
In addition, when there is disequilibrium (unemployment) in the labor market, the equilibrium wage rate cannot be determined by the intersection of the labor supply and demand curves. There are many ways of modeling wage determination; in our model, we introduce a new equation, assuming that the real wage changes in response to variations in the unemployment rate following a constant elasticity (elasPL). Here also we make a distinction between the benchmark equilibrium and the proposed change.
Moreover, households are assumed to engage in both the production and consumption of agricultural commodities. As a producer, the aim is to increase profit, which is achieved either by minimizing costs on factors of production or maximizing quantities produced. For this case, we assume the households choose to minimize the costs of production, and follow a linear-homogeneous CES production functions in capital and labor (K and L), that combine with intermediate input (X ij ) using a Leontief function to obtain the domestic produced commodity (XD i ).
The linear CES production functions for K and L are specified as in Equations (3) and (4): and, The quantity of intermediate inputs (X ij ) is determined assuming an Armington aggregator function for the intermediate inputs imported (M i ) and domestically produced (XDD i ) as specified in Equation (5): Thus, these combine by the Leontief aggregator function to obtain domestically produced commodity (XD i ).
The households (firms) have the choice between selling their commodity on the domestic (village) market or foreign market (outside the village), so as to maximize revenues, subject to the transformation function, for which we select the function form that exhibits constant elasticities of transformation (CET). However, in the literature on Applied General Equilibrium models, there is a formulation wherein the exponents use the plus sign instead of the minus sign (i.e., "+ρT i " instead of "−ρT i "). The reason behind this sign change is that, for the CES function, we minimize total costs, so that the elasticity of substitution between the factors of production is positive; while for the CET function, we maximize total revenues, so that we obtain a negative elasticity of transformation. Further, we assume a degree of homogeneity h = 1; i.e., a constant return to scale and re-parametrization results in obtaining the commodities demanded for village consumption (XDD i ) and export (E i ) from the domestically produced commodities (XD i ).

The Government
The government levies taxes on the consumption of commodities and on capital and labor used by firms, as well as on household income. Consequently, total tax revenues (TAXR) for the government is given by: The government pays unemployment benefits to the household at the replacement rate, denoted by "trep", and it transfers money for other purposes, such as pensions and disaster relief (through the TASAF program). These other transfers are made nominal using the Laspeyres consumer price index, in the presence of endogenous taxes (PCINDEX), resulting in the total government transfers (TRF) being obtained. Government expenditures on capital (KG), labor (LG), and consumption commodities (CG i ) are obtained by subtracting the total transfers and the endogenous taxes from the total tax revenues and the remains are the government's (real) saving (SG). Assuming that the government maximizes the Cobb-Douglas utility function: with: Maximization of the utility function subject to government expenditure equation results in CG i , KG, and LG.

Investment Demand
We assume that village households and firms use financial services from formal and informal financial agents in the village or nearby major market centers; in this case the Mvumi ward center, which is 5 kilometers from the village. The financial agent maximizes a Cobb/Douglas utility function: subject to: S = ∑ n i=1 PD i ·I i , where: S = total saving, PD i = commodity price, and I i = commodity investment demand. Therefore, the total saving equals the household savings (SH): S = SH.

Modelling the Foreign Sector
It is assumed that our economy is that of a small country and that the village does not exert an influence on either the world price of exports (PWEZ i ) or on that of imports (PWMZ i ), and thus the village economy is not affected by the impact of policy changes on world prices. The import price in local currency (including tariffs) is given by: while the export price in local currency is: finally, the balance of payment in (USD), is given by:

Scenario and Simulations Description
The Innovating Strategies to Safeguard Food Security Using Technology and Knowledge Transfer: A People Centered Approach (Trans-SEC) Project introduced various Upgrading Strategies (UPSs) aimed at improving peoples' livelihood in the region under study [39,40]. Among others, a market linkage and an in-situ rainwater harvesting technology UPSs were introduced in the Idifu village. Therefore, our study predicts the welfare effects of the introduced UPSs using scenarios and simulations. The scenarios are developed based on the existing evidence on the effects of these UPSs in Sub-Saharan Africa, Tanzania, and the study village in particular.
Improving the rural poor necessitates increasing productivity in traditional food value chains while also entering more integrated high-value markets where, instead of selling raw materials, farmers can participate in value addition and new product development value chains. These are hindered in rural settings due to a lack of, or low investment in, input supply, storage, handling, and processing [41]. It is often argued that developing horizontal and vertical linkages are an effective organizational innovation in food value chains that helps small-scale producers and traders overcome obstacles that would otherwise discourage participation in high-value markets [42,43]. In high-value sunflower markets, farmers would sell semi-processed sunflower oil instead of seeds. Consequently, it is important to consider the establishment of farmer-processor linkages that would allow farmers organized in groups in Idifu to participate in the value addition and high-value sunflower markets, where the prices offered to farmers are higher than when middlemen come to the village. Evidence from the Trans-SEC project report indicates that farmers in the villages receive about half of the prices of the crops offered at the Dodoma city market when selling to middlemen [39,44]. In 2012-2014, findings show that sunflower prices offered to farmers linked to village markets are between Tanzanian Shilling (TZS) 35,000-42,000 per 65 Kg, while in villages where farmers are not linked to markets, reported sunflower pricing of TZS 35,000 per 100 Kg instead of 65 Kg [45]. Such price differences, in most cases, are aggravated by information asymmetry for the farmers operating individually, so that these farmers are misinformed about the weight of the bag [39,44,45]. In addition, a study conducted in 2015-2016 by Balchin et al. [46] reports an average sunflower farm gate price of about TZS 300, while market price was 660 per Kg of sunflower seeds. Therefore, market linkages are assumed to reduce market information asymmetry with regard to the prices of agricultural commodities in other markets outside the village. In term of yields, a study conducted by Njeru et al. [47] in Embu County, Kenya, reports an average maize grain yield of 0.7 t/ha for farmers using flat cultivation, while farmers using tied ridges have an average yield of 2.2 t/ha. Nyamagara and Nyagumbo [48], conducting a study of crops cultivation, such as maize, sorghum, and sunflower, in a semi-arid smallholder farming environment in central Zimbabwe, report an increase of yield of 100-450% for farmers who use tied ridges. In addition, Mudatenguha et al. [49], conducting a study of semi-arid regions in Rwanda, find that maize yields doubled when using tied ridges cultivation techniques compared to flat cultivation. Moreover, a field study of soil moisture management practices by Kabanza and Rwehumbiza [50] finds that tied ridges increase sorghum yields from 0.4 to 2 t/ha in Dodoma. A study conducted by Germer et al. [51] at Ilolo, a village near Idifu, finds that sunflower yields averaged 1.4 t/ha for farmers using tied ridges, but an average yield of 0.6 t/ha for farmers using flat cultivation. In addition, using a simulated economic risk analysis of tied ridges at Idifu village, Mwinuka et al. [52] report that about 95% of farmers would use tied ridges in their farming practices.
In lieu of the above, first, we consider the market linkage scenario (PrES1), where we simulate the agricultural sector by assuming commodities' prices increase twofold when farmers are linked to processors or traders from outside the village. Second, optimistically, we take an average of each of the reported yields to create our second scenario (QuIS2), where we consider a twofold increase the crop yields when farmers use the tied ridges cultivation technique. Moreover, considering the diversity of the farmer's factors of production endowment [39,[53][54][55], we predict that the potential crop yields for farmers could increase by about 50-80% of the reported crop yields on research farms and under closely supervised on-farm experiments. Therefore, the study uses a comparative static CGE model for the Idifu village economy to simulate the effects of the two UPS scenarios; these provide contrasts to the existing practices. Finally, we compare the results for the increase in output and prices for the agricultural sector in general, later separating the sunflower sub-sector so as to address the objective of this paper and provide insights about the effects of the upgrading strategies on the village economy.

The Idifu Village Economic Structure, 2013
Results from the Idifu economic structure (Table 1) indicate that the value added (VA) share (labor, capital, and land) is about 25.6% for sorghum and millet crops, while the production (PRD) share for all commodities in the village is 24.2%. This implies that more household resources are allocated to the production of sorghum and millet than other crops, as they are more drought tolerant. Sorghum and millet are sold at the weekly markets in Idifu and other nearby villages, with their market share equaling about 19.6%. Moreover, out of the total sorghum and millet output, only 17.1% is sold outside the village; the remainder is consumed at home or recycled as inputs (seeds) for the next season. On the other hand, only 1.0% of sorghum and millet is imported from outside, which accounts for 15.5% of the demand for sorghum and millet of the village. These findings concur with those found by Eriksen et al. [56] and Mmbando et al. [57] that farmers will choose activities that make them less vulnerable to food security. Consequently, they will allocate much of their resource endowments, in this case, labor, land, and capital, to activities that are Pareto-optimal for their needs.
Sunflower is considered to be a cash crop, with a PRD share of about 4.1% and a VA share of about 4.4%. These low shares of the sunflower crop are aggravated by its production system, with most farmers intercropping sunflower with other crops, like sorghum, maize, groundnuts, or millet. About 45.6% of the produced sunflower is sold outside the village and its share of the exported commodities is about 5.8%. The village also imports about 14.2% of sunflower product from outside.
Farmers also grow maize in Idifu, but agro-ecological conditions mean it does not grow well. The production share is only 6.5%, which means that the demand for maize at the village is complemented through importation from outside the village, which accounts for about 68.5% of the village demand. Other crops grown in the village in small proportions include vegetables, groundnuts, tomatoes, and Bambara nuts. Combined, these account for about 19.3% of the PRD share, while their import share is only 0.6%, with 17% of the output sold to nearby villages.
Additionally, the livestock sub-sector, which comprises cattle, goat, sheep, and chicken, plays a great role in the village economy, as about 25.2% of the villagers are employed by the sub-sector. Its share in the exported commodities from the village is about 46%, which is 72.8% of the livestock output from the village. In term of VA and PRD, the sector shares are about 14.1% and 13.1%, respectively. Similar findings are observed in studies [58,59] showing that the livestock sub-sector supports a large proportion of the population of rural households in Tanzania by providing food, income, and animal draught power.
Conversely, about 89.7% of the non-agricultural commodities are imported from outside the village, which accounts for about 97.3% of the non-agricultural commodity demand. Its contribution in terms of employment is about 29.7%, which is the highest of all economic activities in the village. Natural resource extraction, which comprises fishing, logging, and hunting, contributes about 24.9% and 3.5% of the employment and export shares, respectively. Other studies, including Davis et al. [60], observe a high share of employment, about 44%, of the non-farm economy for households in rural Sub-Saharan Africa. This implies that a high share of non-agricultural employment depends on the availability of non-farm activities in the village, which are limited in the case of Idifu village.
Generally, agricultural activities have high shares of VA 69.8%, compared to 30.2% of the non-agricultural activities. This indicates that households depend on labor, land, and capital for income generation. Yet, in terms of employment, non-agricultural activities provide a majority of the employment: about 54.6%. Similarly, the non-farm employment share of about 57.2% for rural households in Tanzania is observed by Nagler et al. [60], while References [61][62][63] observe an increasing trend of non-farm employment in rural households that could have a positive contribution to the income of rural households. Moreover, it is worth noting that nearly every household has at least one member involved in non-agricultural activities, especially during the dry season. This is taken as an income source diversification or coping strategy whenever crop yields are not reliable. This is also noted by Wan et al. [64]: income source diversification is used by rural households to manage drought risks in semi-arid and arid regions. Conversely, although non-farm activities seem to be the best income source alternative, in most cases these are in the informal sector, requiring only low or unskilled labour. However, in the long run, these alternatives are not sustainable unless coupled with the knowledge and skills capacity to unlock barriers of entry to semi-skilled and skilled non-farm employment opportunities.

Main Household Income Sources
Findings from this study (Figure 3) indicate that the main household income source is public transfers (47%), which could be from the Tanzania Social Action Fund (TASAF) that started in the Chamwino district in 2013 due to its high level of poverty. The TASAF program is designed to help the vulnerable. While the program is good in the short run whenever there are food shortages and other livelihood needs, in the long run, government interventions should be geared toward improving productivity through improvement of availability of farm inputs and simple moisture conserving technologies. This is also suggested by Conceição et al. [65] and Elikaeli [66], who note that for Sub-Saharan countries to realize sustainable human development and food security, government intervention should be geared toward increasing yields and not toward the provision of relief funds to affected communities.
conserving technologies. This is also suggested by Conceição et al. [65] and Elikaeli [66], who note that for Sub-Saharan countries to realize sustainable human development and food security, government intervention should be geared toward increasing yields and not toward the provision of relief funds to affected communities.
The second source of household income is labor (30%), where household members work for wages in agricultural or non-agricultural activities, either within or outside the village. As pointed out earlier, household members are mostly engaged in low or unskilled labor, as there are fewer opportunities for skilled jobs. In addition, Tiffen, [12] also notes that the rural economy in Sub-Saharan Africa has ceased growing because the rural labor force is poorly educated. This implies that approaches focusing on promoting rural labor force knowledge and skills development would be helpful at expanding opportunities for the rural labor in developing countries like Tanzania.
Other sources include land rent (10%), remittances (6%), capital (5%), and enterprises (2%). One would expect a high share of income from land rent, however, it is worth noting that persistent droughts at Idifu village have rendered land inexpensive, if not free. Advocacy of simple land management practices that lead to improvement of land utilization is crucial so that expansion of land activities and income contribution can be realized. Consequently, improving land utilization and building the capacity of the rural labor force may positively affect capital accumulation, allowing more enterprises to be established, thus accelerating economic growth in a rural setting.

The Factor of Production Shares across Sectors
Distribution of the factors of production among the farmers, such as land, capital, and labor, has a great role in both implementing policy interventions and understanding the resource allocation criteria of farmers. Results shown in table 2 highlight the distribution of the factor shares across sectors in Idifu village. These indicate that labor has the highest share of 25.0% in VA for millet and sorghum, followed by livestock, which has a share of about 14.7%. In addition, natural resource extraction has a share of about 14.5%, while other crops are about 17.4% and non-agricultural activities make up 17.3% of the labor force. Results further indicate sorghum and millet have the largest proportion of capital share at about 38.7%, consuming about 26.2% of the agricultural land. On the other hand, sunflower and maize have almost the same land allocation shares of about 13.2% and 13.3%, respectively, while other crops occupy about 37.1%.
In contrast, Nerman [13] finds that own crop preferences are not crucial for the allocation of a factor of production for rural households in Tanzania. However, findings from this study indicate that farmers allocate their resources to crops that they are sure of harvesting: first are food crops followed by cash crops that grow well in the area, although specific crop preferences may vary. The second source of household income is labor (30%), where household members work for wages in agricultural or non-agricultural activities, either within or outside the village. As pointed out earlier, household members are mostly engaged in low or unskilled labor, as there are fewer opportunities for skilled jobs. In addition, Tiffen, [12] also notes that the rural economy in Sub-Saharan Africa has ceased growing because the rural labor force is poorly educated. This implies that approaches focusing on promoting rural labor force knowledge and skills development would be helpful at expanding opportunities for the rural labor in developing countries like Tanzania.
Other sources include land rent (10%), remittances (6%), capital (5%), and enterprises (2%). One would expect a high share of income from land rent, however, it is worth noting that persistent droughts at Idifu village have rendered land inexpensive, if not free. Advocacy of simple land management practices that lead to improvement of land utilization is crucial so that expansion of land activities and income contribution can be realized. Consequently, improving land utilization and building the capacity of the rural labor force may positively affect capital accumulation, allowing more enterprises to be established, thus accelerating economic growth in a rural setting.

The Factor of Production Shares across Sectors
Distribution of the factors of production among the farmers, such as land, capital, and labor, has a great role in both implementing policy interventions and understanding the resource allocation criteria of farmers. Results shown in Table 2 highlight the distribution of the factor shares across sectors in Idifu village. These indicate that labor has the highest share of 25.0% in VA for millet and sorghum, followed by livestock, which has a share of about 14.7%. In addition, natural resource extraction has a share of about 14.5%, while other crops are about 17.4% and non-agricultural activities make up 17.3% of the labor force. Results further indicate sorghum and millet have the largest proportion of capital share at about 38.7%, consuming about 26.2% of the agricultural land. On the other hand, sunflower and maize have almost the same land allocation shares of about 13.2% and 13.3%, respectively, while other crops occupy about 37.1%. In contrast, Nerman [13] finds that own crop preferences are not crucial for the allocation of a factor of production for rural households in Tanzania. However, findings from this study indicate that farmers allocate their resources to crops that they are sure of harvesting: first are food crops followed by cash crops that grow well in the area, although specific crop preferences may vary.

Village Macroeconomic Indicators
The 2013 Idifu village Macro SAM indicates various linkages of the economic sectors within and outside the village. Results show that the village gross domestic product (GDP) is 2,221,020 USD, with a GDP per capita income (constant 2010 purchasing power parity) of 1356 USD. Income from government expenditures through transfers to households amounts to 434.76 in USD and is the highest share among the households' income sources, indicating that public support plays a crucial role in supporting household consumption in the village. Government income sources are from the central government, which is outside the village, and amounts to 322,630 USD, with levies from sold commodities in the village 104,870 USD, while indirect taxes from the factors of production is 7260 USD. Moreover, households in the village receive income from factors of production, amounting to 129,940 USD, while transfers from household member 8050 USD, and 4540 USD remittances from outside the village. On the other hand, a small proportion of about 2020 USD of households' income is saved and invested in agriculture commodities.
Commodities consumed by households in the village normally come from own production and market purchases, which amount to 28,710 USD and 538,500 USD, respectively. The village imports more commodities than it exports, amounting to 370,910 USD and 26,840 USD, respectively. Thus, the village is a net importer, especially for non-agricultural commodities and maize, which are not sufficiently produced in the village (Appendix A). Our findings, especially on the GDP per capita for the Idifu village, are relatively close to the 1654 USD observed in the 2014 Tanzania human development report [67]. This implies that, although Tanzania has shown continuous economic growth for the last decade, the growth is not shared among the large rural population.

The Village Agricultural Sector
First, we simulate the effects of farmers participating in a more integrated high-value market outside the village, where they accrue lucrative prices for their production. Therefore, we double export prices (PrES1) for the agricultural sector commodities, then observe its effects on the domestic, export sales, and consumption of commodities for other economic sectors in the village. Results indicate that PrES1 reduces domestic marketed commodity sales for almost all economic sectors in the village, except for a 100% increase in sales of natural resource extraction commodities, such as fish and timber products. Results further indicate an increase in export commodity sales for sorghum and millet (58.3%), sunflower (22.7%), livestock (12.1%), other crops (53.5%), and non-agricultural commodities (84.5%). These findings resonate well with the findings of Biénabe and Sautier [43] and Kaganzi et al. [50], who find that high market values offer higher incomes to smallholder farmers, especially when access is maintained to these markets with significant upgrading in terms of product quality and quantity, as per market demand. Herrmann, et al. [8] also find that linking farmers to markets leads to higher rates of crop commercialization and an increase in crop prices. This implies that the village economy depends on agriculture; therefore, linking farmers to markets helps farmers participate in high-value markets and reduces the obstacles due to a lack of investment in value addition facilities and the exploitation of middlemen.
Secondly, output quantities are increased twofold for the agricultural sector in the productivity increase scenario (QuIS 2 ) as a result of introducing tied ridges technology to farmers. The technology is assumed to increase yields; consequently, we observe what would be the effects on the village domestic sales, export sales, and commodities' consumption for all economic sectors. Results indicate that doubling the output increases the quantities of production sold in and outside the village for almost all economic sectors, except the sunflower sub-sector, which had a decrease in export sales and consumption of 19.2%. Other studies in Tanzania [8,9,57,68] observe that an increase in farmers' output, if coupled with provision of export or high-value markets linkages, not only increases consumption but also sales of the surplus. This means that an increase in crop yields and linking farmers to markets outside the village will increase crop sales as a result of lucrative prices; hence, increasing farmers' profits (Table 3). Table 3. Agricultural sector productivity and farmer-market linkages simulation.

The Village Sunflower Sub-Sector
Additionally, we simulate the same scenarios PrES 1 and QuIS 2 for the sunflower sub-sector in the village so that we can identify the potential effects of the sub-sector on the village economy. Results indicate increasing sunflower exports: 22.18% and 151.89% for PrES 1 and QuIS 2 , respectively. It is worth noting that the sub-sector affects the other economic sectors in the village, as there are also increases in the quantities of consumed commodities. The noted increases are for sorghum and millet (0.42%), maize, non-agricultural, and natural resource extraction (0.62%), as well as livestock (0.3%). These multiplier effects could be enhanced by mixed cropping, which is the farming system commonly practiced by farmers in the village. A study by Mkonda et al. [10] reports a decreasing trend in the production of maize, sorghum, and millet in the semi-arid regions of Tanzania. Therefore, promoting sunflower crops where the intercropping system is dominant could help revitalize the production of these other crops. In addition, since sunflower has low soil nutrient requirements, expanding the crop can be accomplished by making use of the abundant marginal land in the village, where other food crops cannot thrive. Promoting sunflower crops, both in terms of market linkages and increased productivity, leads to increased outputs and quantities sold from other sector commodities. Furthermore, there is also an increase in income for the village households, which is then channeled to purchase other commodities, ultimately reducing government dependency (Table 4). Moreover, an increase in income is noted for all scenarios, for labor, capital, land, households, and the government. Comparing the two upgrading strategies scenarios, QuIS 2 has slightly higher income effects than PrES 1 does (Table 5). However, combining the two upgrading strategies results in greater income increases for all the observed indicators than for the upgrading strategies individually. Similarly, References [8,9,57,68] observe increases in economic returns to farmers' households when these upgrading strategies are combined. Combining upgrading strategies underscores a win-win scenario due to crop yield and price increases, which results in higher economic returns to farmer households in the village.

Conclusion and Policy Implications
The paper highlights the economic situation of rural households in the semi-arid and arid regions of Tanzania, where there is persistent food insecurity as a result of prolonged drought. We achieve this by analyzing the economic activities of Idifu village, which is located in the arid region of Dodoma, an area prone to food shortages and food insecurity. Various interventions are implemented, including, among others, introducing tied ridges as a simple moisture conserving technology designed to improve crop yields and curb food shortage in the area. Another intervention links farmers to markets where farmers can sell their crops, especially sunflowers, at a higher price.
A computable general equilibrium (CGE) model is applied to understand the economic situation of the village prior to and after the implementation of the interventions, helping to determine the impacts on the village economy. Findings from the study underscore the use of the CGE model, first, for analyzing and understanding the economic sectors at a village level. Second, the effects of various upgrading strategies promoted toward improving rural farming communities by the government and non-governmental development partners at the micro-scale can be analyzed and potential agricultural commodity value chains identified Results from the simulation of the two scenarios indicate the impacts on the village economy; for instance, if farmers in Idifu village would use simple moisture conserving techniques, like tied ridges, their yields would increase for all the crops grown in the village. In addition, when farmers are linked to markets, crop yields and prices increase, especially for sunflower, which is used as a cash crop. Furthermore, when sunflower is intercropped with other crops, promoting sunflower crops ultimately results in increasing the yields of the other crops.
The paper arrives at two major conclusions. First, government investment in agricultural infrastructure, such as irrigation in developing countries like Tanzania, is low. Already, small investments in technologies like tied ridges that increase moisture retention and, consequently, increase agricultural output would help increase food security and overall welfare of semi-arid areas in Tanzania. Second, for the case of market integration strategy, farmer-processor linkages can enhance the opportunities for small farmers to participate and benefit from emerging high-value commodities market chains, especially for those in isolated rural villages. As the results from the simulated market linkage scenario for Idifu village farmers indicate, linking sunflower farmers to processors would promote the production of sunflower and other crops, and would also promote welfare.
Therefore, governments should be encouraged to use policies that promote the use of such upgrading strategies among low-income groups in rural settings, not just in Tanzania, but also in other countries with similar endowments. This can be achieved through integrated policies that promote yield increase, such as the use of oxen and ridging tools, via microcredit to farmers' groups aiming at promoting use of tied ridges as a management practice. Moreover, farm field schools and field days can be used as a platform for demonstration and promotion of simple farming techniques such as tied ridges. Moreover, the government of Tanzania, through its agriculture extension system, should teach rural farmers in rural areas multiple skills, thus building knowledge that enhances agriculture as a business in order to generate income and improve their livelihood. Ultimately this will reduce the amount of government financial support required while also making the farmers resilient to shocks.

Acknowledgments:
The authors thank the German Federal Ministry of Food and Agriculture (BMEL) under the Federal Office for Agriculture and Food (BLE) for the grant that supported this work and support from the Leibniz Centre for Agricultural Landscape Research (ZALF) for the workspace. The authors thank the reviewers and editors for their helpful comments and valuable suggestions, which substantially improved this study.

Conflicts of Interest:
The authors declare no conflict of interest.

Equation Description
PE i = ER·PWEZ i Price export in local currency