ESSP WORKING PAPER 62 Determinants and Impact of Sustainable Land and Watershed Management In- vestments: A Systems Evaluation in the Blue Nile Basin, Ethiopia Emily Schmidt*, Paul Chinowsky**, Sherman Robinson‡ and Ken Strzepek‡‡ * Country Program Coordinator / GIS Specialist, Ethiopia Strategy Support Program, Development Strategy and Governance Division, International Food Policy Research Institute **Professor, Department of Civil, Environmental, and Architectural Engineering, University of Colorado ‡ Senior Research Fellow, International Food Policy Research Institute, Development Strategy and Govern- ance Division, International Food Policy Research Institute ‡‡ Research Scientist, MIT Joint Program on the Science and Policy of Global Change, Center for Global Change Science February 2014 TABLE OF CONTENTS Abstract .................................................................................................................................................................................................... 3 1. Introduction ......................................................................................................................................................................................... 4 2. Literature Review ................................................................................................................................................................................ 4 3. Multi-Market Model ............................................................................................................................................................................. 5 3.1 Modeling production .................................................................................................................................................................... 6 3.2 Modeling income and demand ..................................................................................................................................................... 7 3.3 Model data ................................................................................................................................................................................... 7 4. Model Equations ................................................................................................................................................................................. 7 4.1. Price equations ............................................................................................................................................................................ 7 4.2. Supply equations ......................................................................................................................................................................... 8 4.3. Consumption equation ................................................................................................................................................................. 9 4.4. Income equation .......................................................................................................................................................................... 9 4.5. Market clearing equation ............................................................................................................................................................. 9 5. Results ................................................................................................................................................................................................ 9 5.1. Production effects of SLWM ...................................................................................................................................................... 11 5.2. Price effects of SLWM investments ........................................................................................................................................... 13 5.3. Income effects of SLWM investments ........................................................................................................................................ 13 6. Policy options to increase SLWM uptake and maintenance ............................................................................................................. 15 7. Conclusion and further research ....................................................................................................................................................... 17 References ............................................................................................................................................................................................. 18 LIST OF TABLES Table 1—Average production, prices, and income by slope in representative watershed...................................................................... 13 Table 2—Middle slope simulation results ............................................................................................................................................... 15 Table 3—Benefits and costs of policy options to incentivize SLWM sustainability (billion birr) .............................................................. 16 LIST OF FIGURES Figure 1—Area in Blue Nile basin with similar land characteristics to Fogera woreda ............................................................................. 6 Figure 2—Average monthly surface flow (2009-2030) ........................................................................................................................... 11 Figure 3—Average monthly sediment yield (2009-2030) ....................................................................................................................... 11 Figure 4—Total maize production in rural representative watershed ..................................................................................................... 12 Figure 5—Average household income ................................................................................................................................................... 14 Figure 6—Rural household income composition .................................................................................................................................... 14 Appendix Table A1—Benefits and costs of policy options to incentivize SLWM sustainability (5% discount rate) ................................................. 21 Table A2—Area and yield of five major crops in highlands Ethiopia by slope ........................................................................................ 21 3 ABSTRACT Ongoing debate over water resource management in the Nile basin and continuing land degradation in agricultural areas of Ethiopia suggest a need for efficient mechanisms to improve agricultural output in the Blue Nile basin in Ethiopia. Nu- merous econometric and hydrological models have been developed to assess the effects of sustainable land and water- shed management (SLWM) investments, however these models fail to address the trade-offs faced by rural farmers in maintaining such structures. This study combines household survey data that evaluates the economic determinants of program sustainability with a detailed hydrological model that explores location specific effects of SLWM structures. Household survey analysis suggests that households that invested in SLWM infrastructure on their agricultural plots be- tween 1992 and 2002 and subsequently maintained those structures had a 24 percent higher value of production in 2010 than farming households that did not make such SLWM investments. The location specific hydrological model analysis suggests that terraces on middle and steep slope areas have the largest benefit in terms of decreased runoff and sedi- ment and increased agricultural yields. Utilizing the results from the econometric and hydrological model, a systems model is constructed to analyze investment packages. Results suggest that the benefit of implementing only terracing on steep and mid-slope terrain does not outweigh the cost of foregone off-farm labor opportunities nor compensate for a fall in the price of agricultural output (due to increased supply). However, more comprehensive investments (such as in- creased fertilizer use with SLWM) show economically significant increases in household income, suggesting that a pack- aged investment approach is needed to reap welfare benefits from investments in SLWM infrastructure at farm level. 4 1. INTRODUCTION Land degradation in the Ethiopian highlands is due to a variety of factors including climatic variability, dependence on rain-fed agriculture, and a predominantly mixed crop-livestock farming system, which often leaves soil free of ground cover in the rainy season. In addition, high population pressure in rural areas (55 people per km2 in rural Ethiopia, and 86 people per km2 in the rural highland cereal area; CSA, 2010), deforestation due to farmland expansion and energy needs, fragile soils, undulating terrain, and heavy seasonal rains make the highlands vulnerable to soil erosion and gully formation. The SLWM programs implemented thus far are designed to decrease erosion and increase agricultural yields in the high- lands of Ethiopia, thereby improving rural household welfare. Since the overall intention of these programs is to improve environmental outcomes and reduce poverty, it is important to understand the linkages among production increases, cor- responding output price changes, and household income effects in program areas. In addition, given that research sug- gests that maintenance of SLWM structures in rural Ethiopia is uncommon, understanding the private opportunity costs to farmers of maintaining such structures is important in designing effective SLWM programs that ensure sustainability. Linking engineering models (hydrological models) with economic analysis of household survey data provides a more comprehensive framework for evaluating land management programs because doing so takes into account both biophys- ical changes due to SLWM investments and economic components that shape program uptake and success. Thus far, however, combining engineering and economic analysis into a systems framework remains in a nascent stage in the lit- erature. From an engineering standpoint, many project assessments are situated within an engineering economics framework where a majority of inputs are fixed and evaluation results reflect detailed direct effects of investments. Con- versely, economic analysis of project impact takes into account indirect effects in a more stochastic framework, but often lacks detailed quantitative estimates of actual soil losses and water flows. Given that a variety of factors influence the adoption and maintenance of SLWM, including socio-economic factors and biophysical determinants, a systems evalua- tion allows for a wider inclusion of influencing factors and potential outcomes. This study combines a hydrological model that explicitly estimates benefits of SLWM adoption on different slope types (steep hills, midlands, and flat areas) within a watershed of the highland cereals region; a detailed household survey analysis from nine population clusters in the same region of the country that evaluates the economic determinants of pro- gram sustainability; and a road investment model to explore diverse investment scenarios. In doing so, the analysis ex- plores the benefits and costs of providing a packaged rural investment strategy that includes improved market access, better access to fertilizer, and efficient maintenance of SLWM on a variety of slope types within a representative highland cereal watershed. Using a multi-market (MM) systems model computed using GAMS (General Algebraic Modeling Sys- tem) software, we examine the impact of a variety of SLWM investments on agricultural production, producer and con- sumer prices, and household income. The study focuses on five simulations: 1) calibrated base, 2) terraces built on steep and mid-slope terrain, 3) increased fertilizer application with no SLWM intervention, 4) terraces built on steep and mid-slope terrain and increased fertilizer application, and 5) terraces built on steep and mid-slope terrain and increased fertilizer, combined with road investments that decrease transportation costs. Results suggest that the benefit of implementing only terracing on steep and mid- slope terrain does not outweigh the cost of foregone off-farm labor opportunities, nor compensate for a fall in the price of agricultural output (due to increased supply). However, packaged investments, such as increased fertilizer with SLWM, do show economically significant increases in household income compared to the base. Finally, the study explores a va- riety of policy options and their respective costs to determine how adequate incentives might be provided for the mainte- nance and sustainability of SLWM investments in the future. The remainder of this paper is organized as follows. Section 1 provides a literature review on systems modeling and MM models in assessing environmental and economic objectives. Section 2 describes the model and the data used to simu- late SLWM investment options. Section 3 provides a detailed explanation of the model equations. Section 4 provides results from the MM model simulations and discusses tradeoffs in investments. Section 5 reports benefits and costs of a variety of policy options aimed to ensure improved maintenance of SLWM structures. Section 6 discusses future re- search and concludes. 2. LITERATURE REVIEW Given that economic growth and improved welfare is paired with environmental and agricultural sustainability in Ethiopia, analyses that are able to simulate long-term effects of complex ecological-economic systems are necessary in order to inform policy and investment decisions. A variety of software programs, including the General Algebraic Modeling Sys- 5 tem (GAMS) have been developed to identify and optimize synergies between economics, computer science and opera- tions research in the development of a systems model (Bussieck and Meerhaus 2004). This study uses a multi-market (MM) model framework in GAMS whereby various simulations model the impact of household supply and demand deci- sions and household investment choices on the prices of selected commodities and on household income (Braverman and Hammer 1986). The MM model allows ex-ante estimates of the impacts of investments by taking into account both direct and indirect effects in order to better understand the set of investments available to policy makers and what the effects of such investments might be. For example, an investment in SLWM aims to increase agricultural production, but may have an indirect effect of reducing prices of agricultural goods, thereby partially offsetting the effect of increased production on household incomes. The MM framework is useful for evaluating agricultural policy reform. It identifies the network of important relationships and effects of an external shock or policy change (in this case, increased investment in SLWM leading to increased agri- cultural output) on a specified economy. The primary advantage of specifying an MM model in GAMS is the ability to in- corporate both inter-temporal linkages and systems of simultaneous equations within a period. Many systems models in diverse software programs have been used to evaluate environmental and ecological outcomes with economic results. Guo et al. (2009) used a systems model to research a variety of policy scenarios related to socio- economic and environmental tradeoffs and implications for regional planning in the Lake Erhai region of China. Parsons et al. (2011) developed an integrated crop-livestock model to study sheep farming systems in Mexico. Scenarios simu- lated specialized sheep farming systems versus mixed farming and found that mixed systems suggested a higher rate of return. Fiddaman (2007) and Ford (1999) reviewed a variety of environmental analyses using systems models including hydrological modeling, pesticide application and climate policy. Purnomo et al. (2011) analyzed forest management in Indonesia taking into account socio-economic variables that influence behavior and decision-making choices at the household level. MM models have been used for a variety of agricultural policy analyses.1 Dorosh and Haggblade (1997) used a MM model to simulate the impact of switching from a food-for-work (FFW) delivery of wheat food-aid program to a cash-for- work (CFW) program in Bangladesh, and found that the proposed CFW program would improve welfare in poor house- holds compared to the more commonly practiced FFW program. Minot and Goletti (2000) utilized a spatial multi-market analysis based on agro-ecological zones in Vietnam to evaluate market liberalization of the rice sector. Similarly, Goletti and Rich (1998) simulated a variety of policy options for agricultural diversification in different agro-ecological zones in Viet Nam. Lundberg and Rich (2002) built a generic multi-market model that could be adapted to various African coun- tries in order to evaluate agricultural reforms. Stifel and Randrianarisoa (2004) draw from Lundberg and Rich (2002) to analyze alternative agricultural policies on the well-being of households in Madagascar, but also included simulations that evaluated infrastructure improvements (improving market access and lowering transaction costs). The MM model employed in this study utilizes the ability to set equilibrium conditions in order to introduce economic out- put derived from the household survey into a systems framework. In doing so, the MM model includes explicit price-sen- sitive supply and demand equations for a set of commodities, and commodity prices must adjust to equate supply and demand. Household agricultural income is calculated each year based on the value of production supplied to the market by each household type, taking into account costs of production. A comparison of different MM model investment simula- tions facilitates an analysis of optimal investment strategies by quantifying the impacts of these investments on produc- tion, producer and consumer prices, household income, income distribution, and other variables. 3. MULTI-MARKET MODEL In this study, we simulate the effects of SLWM investments on agricultural production, producer and consumer prices, and household income in the short to medium term, taking into account substitution effects in production and consump- tion, as well as opportunity costs of investing in a certain technology (Braverman and Hammer 1986). The model devel- oped for this evaluation is grounded in detailed empirical data and technical parameters related to Ethiopian highland agricultural systems. First, the model takes into account different production systems based on slope type (agriculture on steep hills, midlands, and flat areas) in order to understand how investments in SLWM infrastructure on different slopes affect agricultural production and household welfare. Second, the model utilizes detailed information derived from house- hold surveys that is representative of similar geographic and topographic areas in the rural, highland cereal agro-ecologi- cal zone. In particular, specific data on cropping patterns and hydrological processes for each slope type are drawn from house- hold survey data (Schmidt and Tadesse 2012) and a hydrological model (Schmidt and Zemadim 2013) in the Mizewa 1 Croppenstedt et al. (2007) provide a review of selected examples of MM models 6 watershed located in the southeast of Fogera woreda (district) in Amhara region. Utilizing a relief roughness analysis, land types are classified within the Blue Nile basin using Geographic Information Systems (GIS) by calculating the differ- ence between the maximum and minimum elevation at a 90 meter resolution, divided by half the cell length connecting the center of each grid cell, taking into account latitude and the direction of water flow (Meybeck et al. 2001; Vörösmarty et al. 2000). The Mizewa watershed has a relief roughness classified as hilly or mountainous with areas of low and mid plateau. This characterization is similar to field descriptions of the area in Zemadim et al. (2012). After identifying the re- lief roughness of Fogera, rural kebeles (sub-districts) that have similar relief roughness scores (within 0.5 standard devia- tions of the relief roughness score for Fogera woreda) are included in the MM model analysis. Taking into account these areas, the MM model represents approximately 30 percent of both the agricultural area and the production of teff and maize in the highland cereal agro-ecological zone (Figure 1). Figure 1—Area in Blue Nile basin with similar land characteristics to Fogera woreda Source: Author’s calculations The model specifies the supply of seven consumer goods: teff, maize, barley, sorghum, and wheat (the five main cereals in Ethiopia), fertilizer, and other activities (including non-agricultural goods and services). Households are divided into four groups: three types of rural households in the highland cereal agro-ecological zone farming on (1) steep slopes, (2) midlands / rolling hills, (3) flatlands, and (4) a representative households for the rest of Ethiopia (ROE). 3.1 Modeling production Production is modeled as a function of producer prices.2 Supply elasticities are estimates from Abrar (2003). The model assumes the same price response across all households regardless of slope types; however the productivity effect from SLWM varies by slope type. Each producing region is endowed with fixed labor and land at the farm level which is allocated across crops. Allocation decisions assume price sensitive behavior by households, whereby farmers allocate more resources to higher value crops.3 In the MM framework, this is captured by a land demand function by crop which is a function of the output prices of all crops, and a land clearing condition that specifies that a farmer cannot allocate more land than is available. Yield is a function of input prices and the price of output production. 2 This specification assumes that the farmers’ expected price is the actual price that ultimately prevails in the market (perfect foresight). 3 For example, analysis by Abrar et al. (2004) suggest a statistically significant and positive response of fertilizer demand to output price. 7 3.2 Modeling income and demand Income in the four producing regions is derived from the net value of agricultural production and off-farm receipts. Off- farm wage income is held exogenous in the model. However, labor allocated to SLWM investments (maintaining terrac- ing or bund structures, gully rehabilitation, etc.) is included in order to take account of the opportunity costs of other off- farm wage activities. Based on household survey data, construction of SLWM structures is estimated to take approxi- mately 5 weeks per year. The model assumes 5 weeks of labor demand for the first year of SLWM construction and 2 weeks per year of maintenance in the following years.4 The demand system used is the Linear Expenditure System (LES). Price and income elasticities are from econometric analysis based on the Household Income Consumption Expenditure Survey (HICES) of 2004/2005 and reported by Ber- hane et al. (2012) for individual crops. The demand system is calibrated to known information about own-price elasticities and expenditure elasticities (Abrar 2003; Berhane et al. 2012). Teff, maize, sorghum and barley are modeled as non- tradable goods produced and sold on free markets within Ethiopia but isolated from world markets. Thus, the prices of these commodities vary endogenously in order to equate supply and demand. Wheat is produced as well as imported, thus the price of wheat is exogenously determined by a fixed world price. The domestic price of wheat clears the domes- tic market, and imports respond to the price of wheat endogenously. For non-imported commodities, a large share of pro- duction is dedicated to own-household consumption. To take into account distribution of commodities from producer to the rest of Ethiopia (ROE, which include urban areas), a price wedge is introduced in order to distinguish prices in ROE from the producing regions. This price wedge can be thought of as representing a marketing margin, comprised primarily of transport costs in the teff market in highland Ethiopia (see Minten et al. 2013a). 3.3 Model data Defining 2009 as the base year, the model incorporates data from a variety of sources including the 2009 Ethiopia Re- gional Social Accounting Matrix (SAM) (EDRI 2009), the Ethiopia Agricultural Sample Survey (AgSS) 2004/5-2007/8, and household-level survey data collected in 2010. In addition, data on fertilizer price and transport costs were adopted from analysis by Minten et al. (2013a, 2013b). Per capita consumption data for the five major crops were obtained from the national household survey data used to build the regional SAM, adjusting for the share of the study region in the Ethio- pian rainfall-sufficient highland cereals agro-ecological zone of the SAM. Similarly, data on prices, output, and area for the ROE region were obtained from the Agricultural Sample Survey (CSA 2004/5 -2007/8). For the regions defined by slope type, the household survey data provided total production and area by slope type. Price and income elasticities of demand, and supply elasticities with respect to the price were obtained from previous econometric analyses (Abrar 2003; Berhane et al. 2012). 4. MODEL EQUATIONS There are five core equation groups in the multimarket model: prices, supply, consumption, income, and equilibrium (Lundberg and Rich 2002).5 The price equations define relationships between producer and consumer prices in the do- mestic market, and the relationship of consumer and world prices of fertilizer and wheat (imported commodities). The supply equations represent domestic production of the five cereals and off-farm labor. The consumption category reports household demand for commodities. Income is defined as the sum of the value of agricultural output and exogenous off- farm production. End stocks are related to initial stocks, and equilibrium conditions are set to specify market clearing. 4.1. Price equations The price equations relate producer prices to consumer prices through an exogenously determined marketing margin, which reflects transportation costs and other fees incurred when moving commodities from the farm-gate to the market. Detailed price data on the five cereals are obtained from the AgSS data, while the prices of other commodities in the market are collapsed into an ‘other’ good category and fixed at one, becoming the numeraire of the model. The net price to producers is then computed taking into account fertilizer use and prices. Producer price: 𝑃𝑃𝑉(𝑗) = ∑ 𝑗𝑐𝑟𝑎𝑡𝑖𝑜𝑗,𝑐 ∗ 𝑃𝑋𝑉𝑐 / (1 + 𝑡𝑟𝑠𝑗)𝑐 4 Woldehanna and Oskam (2001) found that an average household used only 47 percent of their time working on-farm and off-farm due to Ethiopian Orthodox church holidays and lack of off-farm labor opportunities, thus we reduce maintenance work to 2 weeks per year to take into account intermit- tent labor patterns. 5 Import or export prices are fixed, so trade clears the markets, rather than prices. 8 [ 𝑃𝑟𝑜𝑑𝑢𝑐𝑒𝑟 𝑃𝑟𝑖𝑐𝑒 ] = [ 𝑆ℎ𝑎𝑟𝑒 𝑜𝑓 𝑐𝑜𝑚𝑚𝑜𝑑𝑖𝑡𝑦 𝑐 𝑖𝑛 𝑎𝑐𝑡𝑖𝑣𝑖𝑡𝑦 𝑗 𝑜𝑢𝑡𝑝𝑢𝑡 ] ∗ [ 𝐷𝑜𝑚𝑒𝑠𝑡𝑖𝑐 𝑠𝑢𝑝𝑝𝑙𝑦 𝑝𝑟𝑖𝑐𝑒 ] / [ 𝑆ℎ𝑎𝑟𝑒 𝑜𝑓 𝑡𝑟𝑎𝑛𝑠𝑝𝑜𝑟𝑡𝑎𝑡𝑖𝑜𝑛 𝑐𝑜𝑠𝑡𝑠 𝑓𝑜𝑟 𝑒𝑎𝑐ℎ 𝑎𝑐𝑡𝑖𝑣𝑖𝑡𝑦 𝑗 ] The net producer price is equivalent to producer price net of input (fertilizer) prices. Net producer price: 𝑃𝑁𝐸𝑇𝑉𝑗,𝑓𝑝𝑢,𝑙𝑛𝑑 = 𝑃𝑃𝑉𝑗,𝑓𝑝𝑢,𝑙𝑛𝑑 − ∑ 𝑖𝑜𝑚𝑎𝑡𝑐,𝑗,𝑓𝑝𝑢,𝑙𝑛𝑑 ∗ 𝑃𝑄𝑉𝑐𝑐 [ 𝑃𝑟𝑜𝑑𝑢𝑐𝑒𝑟 𝑝𝑟𝑖𝑐𝑒 𝑛𝑒𝑡 𝑜𝑓 𝑖𝑛𝑝𝑢𝑡𝑠 ] = [ 𝑃𝑟𝑜𝑑𝑢𝑐𝑒𝑟 𝑝𝑟𝑖𝑐𝑒 𝑜𝑓 𝑎𝑐𝑡𝑖𝑣𝑖𝑡𝑦 𝑗 𝑜𝑢𝑡𝑝𝑢𝑡 ] − [ 𝑆𝑢𝑚 𝑜𝑓 𝑖𝑛𝑝𝑢𝑡 𝑜𝑢𝑡𝑝𝑢𝑡 𝑟𝑎𝑡𝑖𝑜 𝑜𝑓 𝑖𝑛𝑝𝑢𝑡 𝑐 𝑡𝑜 𝑝𝑟𝑜𝑑𝑢𝑐𝑒 𝑎𝑐𝑡𝑖𝑣𝑖𝑡𝑦 𝑗 ] ∗ [ 𝐶𝑜𝑚𝑝𝑜𝑠𝑖𝑡𝑒 𝑝𝑟𝑖𝑐𝑒 𝑜𝑓 𝑐𝑜𝑚𝑚𝑜𝑑𝑖𝑡𝑦 𝑐 ] 4.2. Supply equations Supply equations consist of output supply and input demand. The model differentiates production of agricultural crops and demand for fertilizer by region, allowing for targeted simulations of SLWM investments, agricultural productivity, and input use. For the five major cereals, production is a function of the quantity of land used for a specific crop and its asso- ciated yield. Output is modeled as a multi-product farm using the ‘jc ratio’, whereby agricultural output is translated from a producer good (or activity j) to a commodity good (commodity c) in order to accommodate producer activities that may produce more than one commodity (i.e. chicken produces chicken meat as well as eggs, which have separate prices at the market). Total land under cultivation is fixed, but can be reallocated among crops depending on changes in relative prices. Total commodity supply to the market is defined as production by commodity net of household own consumption. Total commodity supply: 𝑄𝑆𝑈𝑃𝑉(𝑐) = ∑ 𝑗𝑐𝑟𝑎𝑡𝑖𝑜𝑗,𝑐 ∗ 𝑄𝑆𝑉(𝑗)𝑗 − ∑ 𝑄𝐻𝐴𝑉(𝑗, ℎ)ℎ [ 𝐶𝑜𝑚𝑚𝑜𝑑𝑖𝑡𝑦 𝑆𝑢𝑝𝑝𝑙𝑦 ] = [ 𝑆ℎ𝑎𝑟𝑒 𝑜𝑓 𝑐𝑜𝑚𝑚𝑜𝑑𝑖𝑡𝑦 𝑐 𝑖𝑛 𝑎𝑐𝑡𝑖𝑣𝑖𝑡𝑦 𝑗 𝑜𝑢𝑡𝑝𝑢𝑡 ] ∗ [ 𝑃𝑟𝑜𝑑𝑢𝑐𝑡𝑖𝑜𝑛 𝑜𝑓 𝑎𝑐𝑡𝑖𝑣𝑖𝑡𝑦 𝑗 ] − [ 𝑂𝑤𝑛 ℎ𝑜𝑢𝑠𝑒ℎ𝑜𝑙𝑑 𝑐𝑜𝑛𝑠𝑢𝑚𝑝𝑡𝑖𝑜𝑛 𝑜𝑓 𝑎𝑐𝑡𝑖𝑣𝑖𝑡𝑦 𝑗 ] The area, land market equilibrium, and yield equations are defined below. The shadow rental price of land and the total area is held constant; however, area allocated to specific crops changes depending on relative prices. In addition, the wage rate of labor is fixed. Area demand: 𝐴𝑅𝐸𝐴𝑉𝑗,𝑓𝑝𝑢,𝑙𝑛𝑑 = 𝐴𝑟𝑒𝑎𝐼𝑛𝑡𝑗,𝑓𝑝𝑢,𝑙𝑛𝑑 ∗ 𝑊𝐹𝑉𝑓𝑝𝑢,𝑙𝑛𝑑 𝑊𝐹𝐸𝑙𝑎𝑠𝑗,𝑓𝑝𝑢,𝑙𝑛𝑑 ∗ ( 𝑃𝑃𝑉𝑗 𝑃𝑃00𝑗 ) 𝐴𝑟𝑒𝑎𝐸𝑙𝑎𝑠𝑗,𝑓𝑝𝑢,𝑙𝑛𝑑 [ 𝐴𝑟𝑒𝑎 𝑐𝑟𝑜𝑝𝑝𝑒𝑑 𝑜𝑓 𝑎𝑐𝑡𝑖𝑣𝑖𝑡𝑦 𝑗 𝑏𝑦 𝑠𝑙𝑜𝑝𝑒 𝑎𝑛𝑑 𝑙𝑎𝑛𝑑 𝑡𝑦𝑝𝑒 ] = [ 𝐴𝑟𝑒𝑎 𝑖𝑛 𝑏𝑎𝑠𝑒 𝑦𝑒𝑎𝑟 𝑜𝑓 𝑎𝑐𝑡𝑖𝑣𝑖𝑡𝑦 𝑗 ] ∗ [ 𝑅𝑒𝑛𝑡𝑎𝑙 𝑝𝑟𝑖𝑐𝑒 𝑜𝑓 𝑙𝑎𝑛𝑑 ] 𝐸𝑙𝑎𝑠𝑡𝑖𝑐𝑖𝑡𝑦 𝑜𝑓 𝑙𝑎𝑛𝑑 𝑟𝑒𝑛𝑡𝑎𝑙 𝑝𝑟𝑖𝑐𝑒 ∗ [ 𝐶ℎ𝑎𝑛𝑔𝑒 𝑖𝑛 𝑝𝑟𝑜𝑑𝑢𝑐𝑒𝑟 𝑝𝑟𝑖𝑐𝑒 𝑜𝑓 𝑐𝑟𝑜𝑝𝑠 𝑓𝑟𝑜𝑚 𝑏𝑎𝑠𝑒 𝑦𝑒𝑎𝑟 ] 𝐴𝑟𝑒𝑎 𝐸𝑙𝑎𝑠𝑡𝑖𝑐𝑖𝑡𝑦 Land market equilibrium: ∑ 𝐴𝑅𝐸𝐴𝑉𝑗,𝑓𝑝𝑢,𝑙𝑛𝑑𝑗 = 𝑞𝑓𝑠𝑓𝑝𝑢,𝑙𝑛𝑑 [ 𝐴𝑟𝑒𝑎 𝑐𝑟𝑜𝑝𝑝𝑒𝑑 𝑜𝑓 𝑎𝑐𝑡𝑖𝑣𝑖𝑡𝑦 𝑗 𝑏𝑦 𝑠𝑙𝑜𝑝𝑒 𝑎𝑛𝑑 𝑙𝑎𝑛𝑑 𝑡𝑦𝑝𝑒 ] = [ 𝑇𝑜𝑡𝑎𝑙 𝑎𝑣𝑎𝑖𝑙𝑎𝑏𝑙𝑒 𝑙𝑎𝑛𝑑 𝑏𝑦 𝑠𝑙𝑜𝑝𝑒 𝑎𝑛𝑑 𝑙𝑎𝑛𝑑 𝑡𝑦𝑝𝑒 ] Yield: 𝑌𝐿𝐷𝑉𝑗,𝑓𝑝𝑢,𝑙𝑛𝑑 = 𝑌𝑙𝑑𝐼𝑛𝑡𝑗,𝑓𝑝𝑢,𝑙𝑛𝑑 ∗ ∏ 𝑊𝐹𝑉𝑓𝑝𝑢,𝑝𝑓𝑎𝑐𝑝𝑓𝑎𝑐 𝑌𝑙𝑑𝐸𝑙𝑎𝑠𝑊𝐹𝑗,𝑝𝑓𝑎𝑐 ∗ ( 𝑃𝑁𝐸𝑇𝑉𝑗 𝑃𝑁𝐸𝑇00𝑗 ) 𝑌𝑙𝑑𝐸𝑙𝑎𝑠𝑗,𝑓𝑝𝑢,𝑙𝑛𝑑 [ 𝑌𝑖𝑒𝑙𝑑 𝑜𝑓 𝑎𝑐𝑡𝑖𝑣𝑖𝑡𝑦 𝑗 𝑏𝑦 𝑠𝑙𝑜𝑝𝑒 𝑎𝑛𝑑 𝑙𝑎𝑛𝑑 𝑡𝑦𝑝𝑒 ] = [ 𝑌𝑖𝑒𝑙𝑑 𝑖𝑛 𝑏𝑎𝑠𝑒 𝑦𝑒𝑎𝑟 𝑜𝑓 𝑎𝑐𝑡𝑖𝑣𝑖𝑡𝑦 𝑗 ] ∗ [ 𝑅𝑒𝑛𝑡𝑎𝑙 𝑝𝑟𝑖𝑐𝑒 𝑜𝑓 𝑙𝑎𝑏𝑜𝑟 ] 𝑌𝑖𝑒𝑙𝑑 𝑒𝑙𝑎𝑠𝑡𝑖𝑐𝑖𝑡𝑦 𝑜𝑓 𝑙𝑎𝑏𝑜𝑟 𝑟𝑒𝑛𝑡𝑎𝑙 𝑝𝑟𝑖𝑐𝑒 ∗ [ 𝐶ℎ𝑎𝑛𝑔𝑒 𝑖𝑛 𝑝𝑟𝑜𝑑𝑢𝑐𝑒𝑟 𝑝𝑟𝑖𝑐𝑒 𝑜𝑓 𝑐𝑟𝑜𝑝𝑠 𝑓𝑟𝑜𝑚 𝑏𝑎𝑠𝑒 𝑦𝑒𝑎𝑟 ] 𝑌𝑖𝑒𝑙𝑑 𝐸𝑙𝑎𝑠𝑡𝑖𝑐𝑖𝑡𝑦 The demand for fertilizer is a function of output prices, yield, and area: 9 Fertilizer demand: 𝑄𝐼𝑁𝑇𝑉𝑐 = ∑ 𝑖𝑜𝑚𝑎𝑡𝑗,𝑐 ∗ 𝑌𝐿𝐷𝑉𝑗,𝑓𝑝𝑢,𝑙𝑛𝑑 ∗ 𝐴𝑅𝐸𝐴𝑉𝑗,𝑓𝑝𝑢,𝑙𝑛𝑑𝑗 [ 𝐼𝑛𝑡𝑒𝑟𝑚𝑒𝑑𝑖𝑎𝑡𝑒 𝐼𝑛𝑝𝑢𝑡 𝐷𝑒𝑚𝑎𝑛𝑑 ] = [ 𝑆𝑢𝑚 𝑜𝑓 𝑖𝑛𝑝𝑢𝑡 𝑜𝑢𝑡𝑝𝑢𝑡 𝑟𝑎𝑡𝑖𝑜 𝑜𝑓 𝑖𝑛𝑝𝑢𝑡 𝑐 𝑡𝑜 𝑝𝑟𝑜𝑑𝑢𝑐𝑒 𝑎𝑐𝑡𝑖𝑣𝑖𝑡𝑦 𝑗 ] ∗ [ 𝑌𝑖𝑒𝑙𝑑 𝑜𝑓 𝑎𝑐𝑡𝑖𝑣𝑖𝑡𝑦 𝑗 ] − [ 𝐴𝑟𝑒𝑎 𝑜𝑓 𝑎𝑐𝑡𝑖𝑣𝑖𝑡𝑦 𝑗 ] 4.3. Consumption equation Given that the model simulates the behavior of rural households, we do not differentiate between rural and urban de- mand. However, the model does take into account household own-consumption demand. Thus, household demand is the product of the market price of the commodity, minimal level of subsistence consumption based on income of house- hold ℎ and price of crop 𝑐, and available household income after taking into account subsistence spending at the market and consumption from own-stock. Household consumption is modeled based on the ‘linear expenditure system’ (LES) which has been used widely in economic models. Household demand: 𝑃𝑄𝐻𝑉𝑐,ℎ ∗ 𝑄𝐻𝐷𝑉𝑐,ℎ = 𝑃𝑄𝐻𝑉𝑐 ∗ 𝛾𝑚 𝑐,ℎ ∗ 𝛽𝑚 𝑐,ℎ * (𝐸𝐻𝑉ℎ − ∑ 𝑃𝑄𝐻𝑉𝑐𝑐𝑐𝑐 ∗ 𝛾𝑚 𝑐𝑐,ℎ − ∑ 𝑃𝑃𝑉𝑗𝑗𝑗𝑗 ∗ 𝛾ℎ 𝑗𝑗,ℎ ) [ 𝐻𝑜𝑢𝑠𝑒ℎ𝑜𝑙𝑑 𝑐𝑜𝑛𝑠𝑢𝑚𝑝𝑡𝑖𝑜𝑛 𝑠𝑝𝑒𝑛𝑑𝑖𝑛𝑔 𝑜𝑛 𝑚𝑎𝑟𝑘𝑒𝑡 𝑜𝑓 𝑐𝑜𝑚𝑚𝑜𝑑𝑖𝑡𝑦 𝑐 ] = [ 𝑀𝑎𝑟𝑘𝑒𝑡 𝑝𝑟𝑖𝑐𝑒 𝑜𝑓 𝑐𝑜𝑚𝑚𝑜𝑑𝑖𝑡𝑦 𝑐 𝑡𝑜 ℎ𝑜𝑢𝑠𝑒ℎ𝑜𝑙𝑑 ℎ ] ∗ [ 𝑆𝑢𝑏𝑠𝑖𝑠𝑡𝑒𝑛𝑐𝑒 𝑐𝑜𝑛𝑠𝑢𝑚𝑝𝑡𝑖𝑜𝑛 𝑜𝑓 𝑚𝑎𝑟𝑘𝑒𝑡𝑒𝑑 𝑐𝑜𝑚𝑚𝑜𝑑𝑖𝑡𝑦 𝑐 𝑓𝑜𝑟 ℎ𝑜𝑢𝑠𝑒ℎ𝑜𝑙𝑑 ℎ] ∗ [ 𝑀𝑎𝑟𝑔𝑖𝑛𝑎𝑙 𝑠ℎ𝑎𝑟𝑒 𝑜𝑓 𝑐𝑜𝑛𝑠𝑢𝑚𝑝𝑡𝑖𝑜𝑛 𝑠𝑝𝑒𝑛𝑑𝑖𝑛𝑔 𝑜𝑓 𝑚𝑎𝑟𝑘𝑒𝑡𝑒𝑑 𝑐𝑜𝑚𝑚𝑜𝑑𝑖𝑡𝑦 𝑐 𝑓𝑜𝑟 ℎ𝑜𝑢𝑠𝑒ℎ𝑜𝑙𝑑 ℎ ] ∗ [ 𝐴𝑣𝑎𝑖𝑙𝑎𝑏𝑙𝑒 ℎ𝑜𝑢𝑠𝑒ℎ𝑜𝑙𝑑 𝑖𝑛𝑐𝑜𝑚𝑒 𝑓𝑜𝑟 𝑛𝑜𝑛 − 𝑠𝑢𝑏𝑠𝑖𝑠𝑡𝑒𝑛𝑐𝑒 𝑔𝑜𝑜𝑑𝑠 ] 4.4. Income equation Household income is calculated as the sum of agricultural income (net of fertilizer inputs), off-farm income, and transfers, plus residual savings, taking into account household own consumption. While agricultural income varies according to agricultural prices, off-farm income and transfers are treated as exogenous. Household Income: 𝑌𝐻𝑉ℎ = 𝑌𝑟𝑒𝑠𝑖𝑑ℎ + ∑ 𝑠ℎℎ,𝑙𝑛𝑑 ∗ 𝐴𝑅𝐸𝐴𝑉𝑗,𝑓𝑝𝑢,𝑙𝑛𝑑 ∗ (𝑌𝐿𝐷𝑉𝑗,𝑓𝑝𝑢,𝑙𝑛𝑑 ∗𝑗,𝑙𝑛𝑑,𝑓𝑝𝑢 𝑃𝑃𝑉𝑗 ∗ 𝑃𝑁𝐸𝑇𝑉) + 𝑤𝑙𝑏𝑟ℎ ∗ 𝑛𝑙𝑏𝑟ℎ ∗ (𝑞𝑙𝑏𝑟𝑛𝑓ℎ − 𝑞𝑙𝑏𝑟𝑆𝐿𝑀ℎ) + 𝑇𝑅𝑁𝑆𝐹𝑅ℎ ∗ 𝑛𝑙𝑏𝑟ℎ [ 𝐻𝑜𝑢𝑠𝑒ℎ𝑜𝑙𝑑 𝐼𝑛𝑐𝑜𝑚𝑒 ] = [ 𝑅𝑒𝑠𝑖𝑑𝑢𝑎𝑙 𝐼𝑛𝑐𝑜𝑚𝑒 ] + ∑ [ 𝑆ℎ𝑎𝑟𝑒 𝑜𝑓 𝑙𝑎𝑛𝑑 𝑜𝑤𝑛𝑒𝑑 𝑏𝑦 ℎ𝑜𝑢𝑠𝑒ℎ𝑜𝑙𝑑 ℎ ] ∗ 𝑗,𝑙𝑛𝑑,𝑓𝑝𝑢 [ 𝐴𝑟𝑒𝑎 𝑜𝑓 𝑎𝑐𝑡𝑖𝑣𝑖𝑡𝑦 𝑗 𝑏𝑦 𝑠𝑙𝑜𝑝𝑒 𝑎𝑛𝑑 𝑙𝑎𝑛𝑑 𝑡𝑦𝑝𝑒 ] ∗ [ 𝑁𝑒𝑡 𝑝𝑟𝑜𝑓𝑖𝑡 𝑝𝑒𝑟 ℎ𝑒𝑐𝑡𝑎𝑟𝑒 𝑦𝑖𝑒𝑙𝑑 ] + [ 𝑊𝑎𝑔𝑒 𝑜𝑓 𝑜𝑓𝑓 − 𝑓𝑎𝑟𝑚 𝑙𝑎𝑏𝑜𝑟 ] ∗ ([ 𝑄𝑢𝑎𝑛𝑡𝑖𝑡𝑦 𝑜𝑓 𝑜𝑓𝑓 − 𝑓𝑎𝑟𝑚 𝑙𝑎𝑏𝑜𝑟 ] − [ 𝑄𝑢𝑎𝑛𝑡𝑖𝑡𝑦 𝑜𝑓 𝑆𝐿𝑊𝑀 𝑙𝑎𝑏𝑜𝑟 ]) + [ 𝐺𝑜𝑣𝑒𝑟𝑛𝑚𝑒𝑛𝑡 𝑇𝑟𝑎𝑛𝑠𝑓𝑒𝑟𝑠 ] 4.5. Market clearing equation A market equilibrium equation closes the system. Equilibrium is achieved by setting total supply (domestic supply plus imports) equal to household demand for each commodity, 𝑐, plus demand for intermediate inputs (fertilizer). For non- tradeable goods, this implies that as quantity demanded changes, the price of commodity 𝑐 adjusts, whereas for tradea- ble goods (fertilizer), as the quantity demanded increases the quantity imported of other tradable commodities de- creases. Commodity market equilibrium (supply = demand): 𝑄𝐷𝑉𝑐 + 𝑄𝑀𝑉𝑐 = ∑ 𝑄𝐻𝐷𝑉𝑐,ℎ + 𝐼𝑁𝑇𝑉𝑐ℎ [ 𝑇𝑜𝑡𝑎𝑙 𝑑𝑜𝑚𝑒𝑠𝑡𝑖𝑐 𝑠𝑢𝑝𝑝𝑙𝑦 𝑜𝑓 𝑐𝑜𝑚𝑚𝑜𝑑𝑖𝑡𝑦 𝑐 ] + [ 𝐼𝑚𝑝𝑜𝑟𝑡 𝑠𝑢𝑝𝑝𝑙𝑦 𝑜𝑓 𝑐𝑜𝑚𝑚𝑜𝑑𝑖𝑡𝑦 𝑐 ] = [ 𝑆𝑢𝑚 𝑜𝑓 ℎ𝑜𝑢𝑠𝑒ℎ𝑜𝑙𝑑 𝑑𝑒𝑚𝑎𝑛𝑑 𝑓𝑜𝑟 𝑐𝑜𝑚𝑚𝑜𝑑𝑖𝑡𝑦 𝑐 ] + [ 𝐷𝑒𝑚𝑎𝑛𝑑 𝑓𝑜𝑟 𝑖𝑛𝑡𝑒𝑟𝑚𝑒𝑑𝑖𝑎𝑡𝑒 𝑖𝑛𝑝𝑢𝑡𝑠 (𝑓𝑒𝑟𝑡𝑖𝑙𝑖𝑧𝑒𝑟) ] 5. RESULTS In analyzing the impacts of SLWM, separate simulations model supply effects of an exogenous increase in maize and teff yields in the primary growing season (meher season) on market prices and household income. Market prices adjust in order to balance supply and demand, and ensure market equilibrium in the five food markets specified in the model. 10 In this study, five simulations are reported and each of the simulations (modeled in GAMS) use yield data output by the hydrological model analysis using the Soil and Water Assessment Tool (see Arnold et al. 1998 for details on the SWAT model). The simulations are as follows: 1) A base simulation (calibrated using household survey data and the hydrological model); 2) Constructing and maintaining terraces on middle and steep slope areas; 3) Doubling fertilizer application from 50 kg to 100 kg per hectare, but not investing in SLWM; 4) Constructing and maintaining terraces on middle and steep slope areas and doubling fertilizer application; 5) Constructing and maintaining terraces on middle and steep slope areas, and doubling fertilizer application, and improving in transportation infrastructure. Results are disaggregated into three categories based on slope type and corresponding to the specific household defini- tions used in the MM model. The base simulation represents the current (calibrated in SWAT) state of the representative watershed, whereby no SLWM is constructed or maintained. A cumulative land productivity decrease of 1 percent per year is imposed in the base simulation to capture long-term soil productivity declines in land without SLWM investments as modeled by the econometric analysis (following results reported by Schmidt and Tadesse 2012). The MM model is run on an annual time step from 2009 to 2030; simulations are compared and expressed as a percentage deviation from the base simulation. In order to quantify the yield shock of SLWM on agricultural output, a water model was constructed using the Soil and Water Assessment Tool (SWAT) to quantify water runoff, erosion, and agricultural yield changes given a variety of SLWM investments.6 The SWAT model uses data from recently installed gauge stations in Mizewa watershed in order to understand more localized impacts of sustainable land management practices on runoff and erosion. The model is cali- brated at a daily, weekly, and monthly time step. Parameter calibration was completed using one year of data collected at the outlet of the Mizewa watershed. The weekly and monthly simulated flow accurately depicts runoff, and reveals that hydrologic processes and flow regimes in SWAT have a good fit with observed flow data (monthly simulated to observed R2 values were .94). The SWAT model has a built-in crop model in order to track the growth cycle and yield of specified agricultural crops. Utilizing the calibrated SWAT model, yield shocks are derived based on different parameter modifica- tions in SWAT to represent SLWM investments. Fertilizer application is also manipulated in the simulations in order to calculate yield shocks due to increased fertilizer application. SWAT simulations suggest that the most effective SLWM investments in terms of decreasing surface flow and erosion (which has a direct effect on agricultural yields) in the representative watershed are 1 meter tall terraces on steep agricul- tural areas (greater than 20 degree slope) and 0.5 meter tall terraces on midland (5-20 degree slope) agricultural areas (Figures 2 and 3).7 Results suggest that implementing SLWM on steep, middle and flat slopes results in marginal im- provements of runoff and sediment capture compared to terracing on only steep and middle slopes (Schmidt and Zemadim, 2013). However, SLWM on flat slopes provided mixed results when analyzing agricultural yield data due to increased flooding or waterlogging of agricultural crops during high rainfall years. Thus, this study focuses on middle and steep slope terrace construction as the primary SLWM intervention, paired with other supply side shocks (i.e. fertilizer application, transport cost decreases) . Each SWAT simulation is run on a daily time-step and aggregated to evaluate annual yields. Computed agricultural yields are then used in the MM model to compare SLWM simulations taking into account indirect effects (price changes, off-farm labor opportunity costs). 6 A commonly used hydrological model in Ethiopia is the Soil and Water Assessment Tool (SWAT) developed by the US Department of Agriculture (Arnold et al. 1998). Previous analyses in Ethiopia include: Tesfahunegn et al. (2011), Bewket and Sterk (2005), Legesse et al.(2003), Nyssen et al. (2010), Betrie et al. (2011), Schmidt and Zemadim (2013). 7 Agricultural area is defined as any terrain that is not under forested landcover or within a wetland area 11 Figure 2—Average monthly surface flow (2009-2030) Source: Author’s calculation Figure 3—Average monthly sediment yield (2009-2030) Source: Author’s calculation 5.1. Production effects of SLWM The main effect of SLWM is to increase production by decreasing runoff and erosion, as well as increasing water capture in the soil. On average, given ongoing degradation in the watershed, middle and steep slope terrace investments im- prove agricultural production compared to the base by impeding the 1 percent decrease in production each year (Figure 4).8 Productivity shocks due to land degradation are calculated using household survey data stratified by households that participated in an SLWM program and a control group in order to allow for robust estimates of a single-difference in out- comes analysis. Analysis suggests that the marginal difference between plots that receive SLWM investments and those that do not invest, increase at an increasing rate. For example, if SLWM structures are maintained on a plot for 7 years, the value of production is estimated to be 2 percent greater than non-SLWM plots, whereas if a household continues to maintain SLWM for 15 years the expected value of production is 12 percent greater (Schmidt and Tadesse 2012). These results echo analysis by Holden and Shiferaw (2002) that estimated productivity declines (using the universal soil loss equation) of 1.1 percent per year due to land degradation. 8 Fluctuations in agricultural production reflect varying yearly climate conditions reported in historical data and analyzed in the hydrological model. 0 5 10 15 20 25 Ja n Fe b M ar A p r M ay Ju n Ju l A u g Se p O ct N o v D ec Su rf ac e Fl o w ( m m ) Base (mm) Terrace (slope >20°) Terrace (slope >5°) Terrace (slope >5°) Bund (1-5 slope°) Residue management (< 5°) Terrace (>5°) Residue management (5-20° slope) Terrace (slope >20°) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 Ja n Fe b M ar A p r M ay Ju n Ju l A u g Se p O ct N o v D ec M t / h e ct ar e Base (mm) Terrace (slope >20°) Terrace (slope >5°) Terrace (slope >5°) Bund (1-5 slope°) Residue management (< 5°) Terrace (>5°) Residue management (5-20° slope) Terrace (slope >20°) 12 Given calculations of land productivity due to land degradation, teff production in the base simulation decreases from approximately 1.7 million tons in 2009 to 1.3 million tons in 2030, whereas in the middle and steep terraces investment (MST) simulation, land degradation is less significant and overall maize production values fluctuate around 1.7 million tons throughout the simulation period (Figure 4). A similar pattern is reflected in teff production (Table 1). Figure 4—Total maize production in rural representative watershed Source: Author’s calculation * MST = Middle and steep slope terraces Fertilizer application greatly improves production. Currently, rural households apply approximately 50 kg/ha of DAP and urea to agricultural plots. When doubling fertilizer application to 100 kg/ha (recommended use of fertilizer application in Ethiopia varies from 100-150 kg/ha), production increases 1.5 times that of the base simulation immediately.9 This is sim- ilar to previous field trials of maize production in Ethiopia where data suggest that efficient adoption of seed-fertilizer technologies could as much as double cereal yields (Howard et al. 2003). However, the fertilizer base simulation reflects the ongoing degradation due to a lack of SLWM. Although fertilizer application increases yields initially, sheet erosion and gulley formation continue to wash away topsoil and fertilizer over time, and the impact of fertilizer on production eventually begins to decline by 2015. Pairing increased fertilizer application with MST investments reaps the highest pro- duction gains, given that fertilizer application is fully efficient (it is not running down the hill slopes), and runoff and ero- sion are not causing greater land degradation. Evaluating production gains disaggregated by slope type, simulations suggest that implementing MST investments in- creases production in steep areas by 26 and 24 percent after 20 years for teff and maize, respectively, compared to the base (Table 1). Similarly, middle slope agricultural production increases by 24 and 23 percent after 20 years for teff and maize, respectively, compared to the base. In flat areas, maize production decreases slightly due to the decline in market prices for teff and maize resulting from the overall increase in supply. Comparing agricultural production gains across all simulations, results suggest that increased fertilizer application is im- portant to achieving large gains in productivity. Increasing fertilizer application without investing in SLWM increases maize yields by approximately 60 percent in flat, middle and steep slope areas (Table 1). Applying fertilizer and investing in MST almost doubles production of maize in middle and steep terrain (increasing production by 95 and 99 percent, re- spectively), and has similar effects for teff production. Similar production gains of teff and maize result from investments in MST, increased fertilizer use, and improved transportation infrastructure (Simulation 5). 9 This productivity shock due to fertilizer is based on the SWAT calculation of a harvest index which takes into account the potential nitrogen and phos- phorus uptake available to agricultural crops 0 0.5 1 1.5 2 2.5 3 M ill io n T o n s Base MST Fertilizer Base MST and Fertilzer MST, Fertilizer, and -50% transport cost * 13 Table 1—Average production, prices, and income by slope in representative watershed Simulation (Values in 2030) Base 1 MST 2 Fertilizer base 3 MST and fertilizer 4 MST, fertilizer and -50% transport cost 5 Teff (million tons) Steep 0.04 26.3% 68.4% 107.9% 110.5% Mid 0.52 23.9% 55.7% 91.6% 93.3% Flat 0.08 0.0% 56.4% 55.1% 56.4% Maize (million tons) Steep 0.07 23.9% 61.2% 98.5% 100.0% Mid 1.08 22.8% 59.9% 94.8% 96.3% Flat 0.15 -0.7% 60.0% 57.9% 59.3% Teff price (1000 birr/ton) Fogera 2.23 -3.7% -9.8% -14.5% -7.3% Maize price (1000 birr/ton) Fogera 1.11 -8.8% -22.8% -30.7% -25.2% HH income (1000 birr/household) Steep 2.21 -4.4% 7.2% 9.5% 14.3% Mid 2.69 0.1% 10.1% 20.9% 28.7% Flat 2.21 -3.1% 5.8% -1.0% 2.0% Source: Author’s calculation *MST = Middle and Steep Terraces 5.2. Price effects of SLWM investments The increase in production given investments in SLWM and fertilizer application has an indirect effect of lowering pro- ducer prices of both teff (by 14.5 percent) and maize (by 30.7 percent) due to increases in supply (Table 1). Given that teff has a high income elasticity of demand, productivity increases do not result in as large a price fall as for maize. A large share of teff consumption occurs in the highland cereals region, where the representative watershed modeled in these simulations accounts for 20 percent of national teff consumption. As teff farmer incomes rise, their demand for teff increases, preventing a large price decline. Investments in roads that improve market access (through a 50 percent de- crease in transportation cost) attenuate such a price drop, but only by 5 percent compared to no improvements in road infrastructure (maize prices drop 25 percent when investing in transport, MST, and fertilizer compared to the base). 5.3. Income effects of SLWM investments Given that market prices decrease as production increases, percentage gains in producer income are less than the per- centage gains in quantity produced. Moreover, households that invest in SLWM must devote labor time to construction and maintenance that could otherwise be allocated to off-farm income activities. Although off-farm income is less than agriculture income (accounting for 27 and 35 percent of overall income for households with land on middle and steep slopes, respectively), the opportunity cost of foregone off-farm labor due to SLWM maintenance has a large impact on income over time. Overall, simulations that pair fertilizer and SLWM result in an overall increase in producer incomes compared to the base (Figure 5). Investing in SLWM without fertilizer investments results in lower producer incomes compared to the base sim- ulation, although annual household income approaches the base simulation values after 25 years. This lackluster perfor- mance of SLWM investments without fertilizer is due mainly to the relatively low yield increases resulting from SLWM investments. In addition, the foregone wage opportunity from off-farm labor in lieu of constructing (in the first year) and maintaining SLWM decreases off-farm income significantly.10 Households on middle and steep slopes dedicate time to constructing and maintaining their SLWM infrastructure, and in doing so, they earn almost 25 percent less off-farm wage income per year of maintenance compared to not adopting SLWM infrastructure. Finally, an increase in yields with limited market access causes prices to fall on specific commodities, thus farmers are receiving less revenue per kilogram of maize or teff. 10 63 percent of off-farm labor is lost during the first year of SLWM construction. 14 Figure 5—Average household income Source: Author’s calculation *Note: Exchange rate in 2005 (birr/dollar) was 8.67 It is important to note that income levels and composition vary depending upon the location of the household in the wa- tershed.11 Most agricultural land in the Ethiopia highlands is on middle slope terrain – plot level household survey data suggest that agricultural land area on mid-slope terrain is 6 times that of agricultural land on steep and flat slopes. Household survey data also suggest that households located on mid-slope areas (slopes ranging from 5-20 degrees) on average derive a greater share of income from agriculture than households living in steep or flat areas. Given that house- holds in mid-slope areas derive a large share of their income from agriculture (Figure 6), returns from SLWM investments represent a high share of their total income. In addition, middle slope areas dedicate 580 thousand hectares to teff pro- duction (a high value crop in rural Ethiopia) compared to flat and steep areas that dedicate 70 thousand and 60 thousand hectares respectively. Agricultural yields are also higher for middle slope areas growing maize compared to flat and steep areas; and yields are higher for teff in middle areas compared to steep slopes (Appendix 2). Figure 6—Rural household income composition Source: Author’s calculation *Other income is derived from a variety of sources including remittances and other transfers 11 The breakdown of income is derived as follows: Total income (assumed to be equal to total expenditure as obtained in the regional SAM data) equals the sum of agricultural income (derived from the watershed household survey), off-farm labor income (estimated using the product of off-farm employ- ment and the wage rate), other income (difference between total income, labor income and agricultural income). 2 2.2 2.4 2.6 2.8 3 3.2 3.4 2009 2012 2015 2018 2021 2024 2027 2030 Th o u sa n d b ir r / H H Base MST Fertilizer Base MST and Fertilizer MST, Fertilizer, and -50% transport cost 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 Steep Middle Flat Th o u sa n d B ir r/ P e rs o n Agriculture Non-Agriculture Other 15 At the household level, pairing SLWM investment with fertilizer application nearly doubles teff and maize production and increases income in both the middle and steep terrain areas by 9.5 and 20.9 percent, respectively (Table 1). Although investing in SLWM with no extra fertilizer application increases maize yields by 23 percent in the steep terrain area, sim- ulations suggest that income is 4 percent lower than the base scenario where land is degrading over time (Table 1). This result is due to price decreases, as well as foregone off-farm labor opportunities, which represent a larger share of in- come in steep areas compared to mid-slope areas. In comparison, mid-slope households increase teff and maize production by 24 percent under the MST simulation, and income increases (albeit minimally) by 0.1 percent (Table 2). When comparing results between fertilizer simulations (ap- plying fertilizer with no SLWM compared to applying fertilizer with SLWM investments), greater income benefits are achieved with SLWM and fertilizer application on mid-slope areas (21 percent increase in income). Mid-slope areas re- ceive higher gains from SLWM compared to steep slopes because a larger share of income is based on agriculture, thus increases in productivity have a greater effect on overall income gains. Table 2—Middle slope simulation results Simulation Teff Production Maize Production Teff Price Maize Price Income Calibrated Base 0.52 1.08 2.23 1.11 2.69 MST 23.9% 22.8% -3.7% -8.8% 0.1% Fertilizer base 55.7% 59.9% -9.8% -22.8% 10.1% MST and fertilizer 91.6% 94.8% -14.5% -30.7% 20.9% MST, fertilizer, and -50% transport cost 93.3% 96.3% -7.3% -25.2% 28.7% Source: Author’s calculation Simulations suggest that investing only in SLWM without increasing fertilizer use is not profitable for individual house- holds. On average, incomes are lower in 2030 after investing only in SLWM compared to the base simulation. Although on average the watershed area gains from investing in SLWM and fertilizer, mid-slope households gain considerably more than steep-slope households, and flat slope households derive less income (compared to the base) due to de- creases in producer prices. Although flat areas are not investing in SLWM and so do not forego off-farm labor opportuni- ties, they are still affected by market prices and are selling the same amount of crop output for a lower price.12 These discrepancies in income gains suggest a need to rethink SLWM program implementation. Focusing only on SLWM in- vestments without increased access to fertilizer (which is the current method of SLWM programming in the highland ce- real area) does not provide the necessary household income increases that would attract sustained farmer maintenance of SLWM structures in the long term. The next section explores policy options and cost-benefit analysis for incentivizing greater farmer uptake and maintenance of SLWM. 6. POLICY OPTIONS TO INCREASE SLWM UPTAKE AND MAINTENANCE It is often the case that larger projects that have an immediate cost (in this case, construction in terms of labor costs of SLWM) at the beginning of the project rarely reap clear, quick benefits in the short run. In order to evaluate different op- tions for SLWM investments, we compare the net present value of changes in household consumption with the net pre- sent value of costs, using a 3 percent discount rate (sensitivity analysis using a 5 percent discount rate is presented in Appendix Table A1). The benefit-cost analysis compares the following program packages: 1) Terraces on middle and steep slopes (MST) 2) Increased fertilizer application 3) Increased fertilizer application and MST 4) Government transfers to households maintaining MST (with increased fertilizer application) 5) Decrease in transport costs of: 50, 30, and 20 percent The total net present value of production from 2010 to 2030 for highland cereal households that invest in middle and steep terraces is about 92 billion birr, 2.7 percent lower than the base calibrated simulation (Table 3). These results are consistent with other cost-benefit studies found in the literature, whereby benefits of soil and water conservation alone do not provide large income benefits to implementing households. Shiferaw and Holden (2001) analyzed experimental trials 12 It is important to note that the price fall due to increased supply does yield a benefit to consumers that are buying goods at the market. Further work should evaluate private profitability of farm households and the social benefits of decreased prices for consumers. 16 of bunds and terraces in west and east Amhara and found insufficient economic incentives for investment in such struc- tures. Hengsdijk et al. (2005) underlined the tradeoffs of soil and conservation investments in Tigray region where soil bunds slightly increased crop productivity during drier periods when yields were low, but decreased productivity during moist seasons because overall cropped area was reduced for the construction of bunds. Table 3—Benefits and costs of policy options to incentivize SLWM sustainability (billion birr) Simulation Discount rate 3% Household Consumption (NPV) Change in Household Consumption (NPV) Marginal benefit of government investment (NPV) Government cost (NPV of cost) Marginal Benefit : Cost 1) Base 94.2 0 - - - Land Investments 2) Terraces on middle and steep slopes (MST) 91.66 -2.54 - - - 3) MST with government transfer 101.70 7.50 10.04 6.95 1.44 4) Fertilizer increase 117.11 22.91 - - - 5) Fertilizer increase and MST 114.25 20.05 - - - Road Investment Scenarios (with Fertilizer and MST) 6) Transport cost decrease 50% 117.06 22.86 2.81 2.08 1.35 7) Transport cost decrease 30% 115.86 21.66 1.61 0.88 1.83 8) Transport cost decrease 20% 115.31 21.11 1.06 0.88 1.20 Source: Author’s calculation Simulation results of increased investment in fertilizer application (Table 3, simulation 4) compared to increased invest- ment in fertilizer application with MST investments (simulation 5) suggest similar lackluster results for MST investments. Whereas fertilizer application without investing in MST has a net present value of about 117 billion birr, a comprehensive investment in fertilizer and MST yields a net present value of approximately 114 billion birr. Although the net present value of investing in both MST and fertilizer is greater than the base, it is more profitable (within the 20 year timeframe) to invest in only fertilizer and allow the land to continue to degrade. Thus, other policy instruments may need to be introduced to offset the initial construction and labor costs of MST invest- ments. In simulation 3 in Table 3 we model the effects of providing SLWM adopting households with a government trans- fer of 450 birr per household in the first year (2005 US$ 56) and 180 birr per household during the following 20 years to compensate for SLWM maintenance labor inputs. The government transfer compensates producers for approximately 80 percent of their lost off-farm labor income (due to time allocated to construct and maintain SLWM) and increases the net present value of MST to 101.7, or 7.5 billion birr more than the base simulation. Although this policy increases govern- ment costs of implementing SLWM investments, the marginal benefits are almost 1.5 times that of costs.13 It is important to note that limited access to credit, risk averseness, and higher private discount rates raise the costs of farmer investments in such infrastructure, suggesting that even larger government transfer may be necessary to encour- age farmers to maintain private SLWM investments during the medium and long term. Another option of increasing profitability and improving sustainability of SLWM programs is through a more comprehen- sive rural development approach incorporating increased fertilizer application, investments in SLWM, and improvements in rural road infrastructure. The impacts of road infrastructure investments on agricultural output and productivity are par- ticularly important given that a large share of Ethiopia’s population lives in rural areas and long average travel time to markets result in high transaction costs for sales of agricultural outputs (Dorosh et al. 2012; Stifel and Minten 2008). Im- proving market access to rural households decreases the cost of transporting agricultural goods to markets and poten- tially increases market demand for agricultural goods through agglomeration economies (see Henderson 2001). 13 These calculations of benefits and costs take into account only private benefits of investment to farmers (gains in their net incomes). The calculations do not include the social benefits to consumers arising from lower prices of teff and sorghum. Thus, the social benefits shown in the table are equal to the private benefits. 17 In order to evaluate the net profitability of improved market access (decreased transport cost), a range of estimated road infrastructure upgrading costs are evaluated. Following Chinowsky et al. (2011), an annual maintenance cost is imposed for earth, gravel, and paved roads, respectively. Annual maintenance takes into account rainfall in the area, which is es- pecially important on earth roads given that these surfaces are more prone to erode and disintegrate over time. Approxi- mate annual costs of maintaining an earth and gravel road are US$ 750 and US$ 3,000 per kilometer respectively (COWI 2009 and Skorseth and Selim 2000). In addition, more comprehensive rehabilitation work is needed on earth roads every 7 years (US$ 50,000 per kilometer) and gravel roads (grading, re-graveling, and/or stabilization of soils) every 12 years (US$ 90,000 per kilometer) in order to maintain drivability. These costs are taken into account and evalu- ated in the benefit-cost analysis. The simulations that take into account improvements in market access simulate several benefit streams. In order to have a 50 percent reduction in transport costs, the simulation assumes that 25 percent of earth roads are upgraded to gravel road status (and receive annual maintenance and rehabilitation) in the representative watershed area, while the remain- ing road stock receives planned annual maintenance and 7 or 12-year road rehabilitation. Assuming a 50 percent reduc- tion in transport costs, the benefits to improving market access are over a third greater than the costs (Table 3). Although the benefits outweigh the costs in this scenario, neither the costs nor the time needed to improve this large share of the rural unpaved roads in Ethiopia are trivial. Thus, more modest simulations are also undertaken assuming that 10 percent of earth roads are upgraded to gravel status, resulting in a 30 percent decrease in transport costs. This analysis is re- evaluated for sensitivity purposes assuming a 20 percent reduction in transport costs. In both simulations, the benefits of increased agricultural productivity, leading to higher income, outweigh the costs of investing in road upgrades (benefit- cost ratios of 1.8 and 1.2, respectively).14 7. CONCLUSION AND FURTHER RESEARCH The trade-off between short-term welfare gains and long-term agricultural investment planning in the highland region of Ethiopia represents a large challenge that requires understanding the linkages between socio-economic factors and bio- physical determinants of agricultural production. This study employs a MM systems model that incorporates a detailed hydrological model to explicitly estimate benefits of SLWM adoption on different slope types and detailed household sur- vey data that evaluates the economic determinants of program sustainability. A variety of investment simulations are run over a 20 year time horizon (2009 to 2030), including simulations to understand benefits of more comprehensive rural investment packages (SLWM, fertilizer, and improved transportation infrastructure investments). Results suggest that the benefits of implementing only SLWM do not outweigh the cost of foregone off-farm labor opportunities, nor increase out- put enough to compensate for price falls in the agricultural commodity market due to increased supply. Previous research, however, notes that the prevailing agricultural land use, in combination with seasonally heavy rainfall and advanced deforestation, threatens to decrease agricultural productivity to unsustainable levels in future years. Thus, a more comprehensive investment approach may be necessary to ensure that SLWM provides tangible benefits to rural households in a timelier manner. A benefit-cost analysis of a variety of investment packages suggests that government transfers during the first phase of SLWM investment would provide farmers the needed incentive to maintain SLWM structures on their land during the slack labor season. Similarly, upgraded transportation infrastructure paired with in- creases in fertilizer application and SLWM significantly improve household welfare results. This analysis could be strengthened through collection of additional biophysical data on rainfall, soil characteristics, and water flow to supplement the data from the watershed used in the hydrological model. Increasing the number of water- sheds modeled would also provide a more nuanced view of the agricultural economy in the highland cereal area. In addi- tion, given current projections of climate change, simulations that take into account variations in weather patterns pre- dicted by global climate models may produce significantly different results. Future research could look into these climate uncertainties and build upon historical data to model simulations to understand different sequences and scenarios of complementary investment packages to measure potential synergies in policy and programming. Given ongoing land degradation and its continuing effect on agricultural productivity (loss of 2-6 percent of agricultural GDP per year), it is vital that land management is not only socially beneficial in future years, but also provides the necessary private income gains to incentivize rural farmers to maintain SLWM structures tomorrow. 14 Estimates of marketing margins for Ethiopia using CSA data for 2007/8 suggest a 50 percent difference between rural and urban markets (Dorosh et al., 2012), and assume that 80 percent of marketing costs are due to factors related to transportation. Gabre-Madhin (2001) reported an average transport share of 31 percent in 1996 due to transportation costs for the cereal trade. 18 REFERENCES Abrar, S. 2003. Estimating Supply Response in the Presence of Technical Inefficiency Using the Profit Function: An Ap- plication to Ethiopian Agriculture. Discussion Papers in Economics 03/4, Department of Economics, University of Leicester. Abrar, S., O. Morrissey and T. Rayner. 2004. Aggregate agricultural supply response in Ethiopia: a farm-level analysis. Journal of International Development. 16: 605–620. Arnold, J.G., R. Srinivasan, R.S. Muttiah, and J.R. Williams. 1998. Large‐area hydrologic modeling and assessment: Part I. Model development. Journal of American Water Resources Association. 34(1): 73‐89. Berhane, G., L. McBride, K.T. Hirfrfot, and S. Tamiru. 2012. Patterns in foodgrain consumption and calorie intake, Chap- ter 7 in Dorosh, P. and S. Rashid (eds.), Food and Agriculture in Ethiopia: Progress and Policy Challenges. Uni- versity of Pennsylvania Press. Philadelphia, Pennsylvania. Betrie, G.D., Y.A. Mohamed, A. van Griensven, and R. Srinivasan. 2011. Sediment management modelling in the Blue Nile Basin using SWAT model. Hydrology and Earth Systems Science. 15: 807–818. Bewket, Woldeamlak and Geert Sterk. 2005. Dynamics in land cover and its effect on stream flow in the Chemoga water- shed, Blue Nile basin, Ethiopia. Hydrological Processes. 19: 445-458. Bojö, J. and D. Cassells. 1995. Land degradation and rehabilitation in Ethiopia: a reassessment. AFTES Working Paper 17. Washington, DC: World Bank. Braverman, A. and J.S. Hammer. 1986. Multimarket Analysis of Agricultural Pricing Policies in Senegal, Chapter 8 in Singh, I., L. Squire and J. Strauss (eds.), Agricultural Household Models: Extensions, Applications, and Policy. Baltimore, MD.: The Johns Hopkins University Press. Bussieck, M.R. and A. Meeraus. 2004. General Algebraic Modeling System (GAMS) Chapter 8 in Kallrath, J. (ed.), Mod- eling Languages in Mathematical Optimization, Norwell, Massachusetts: Kluwer Academic Publishers. Central Statistical Agency of Ethiopia (CSA). 2010. Population and Housing Census Atlas of Ethiopia. Addis Ababa: Cen- tral Statistical Agency. Central Statistical Agency of Ethiopia (CSA). 2008. Agricultural Sample Survey 2007/2008 (2000 E.C.): Volume I - Re- port on Area and Production Crops (Private Peasant Holdings, Meher Season). Statistical Bulletin 417. Addis Ababa: Central Statistical Agency. Central Statistical Agency of Ethiopia (CSA). 2007b. Agricultural Sample Survey 2006/2007 (1999 E.C.): Volume I - Re- port on Area and Production Crops (Private Peasant Holdings, Meher Season). Statistical Bulletin 388. Addis Ababa: Central Statistical Agency. Central Statistical Agency of Ethiopia (CSA). 2006. Agricultural Sample Survey 2005/2006 (1998 E.C.): Volume I - Re- port on Area and Production Crops (Private Peasant Holdings, Meher Season). Statistical Bulletin 361. Addis Ababa: Central Statistical Agency. Central Statistical Agency of Ethiopia (CSA). 2005. Agricultural Sample Survey 2004/2005 (1997 E.C.): Volume I - Re- port on Area and Production Crops (Private Peasant Holdings, Meher Season). Statistical Bulletin 331. Addis Ababa: Central Statistical Agency. Chinowsky, P., S. Hayles, C., A. Schweikert, and N. Strzepek. 2011. Climate Change As Organizational Challenge: Comparative Impact On Developing And Developed Countries. Engineering Project Organization Journal. 1(1). COWI. 2009. Making Transport Climate Resilient, Report to the World Bank, Document P-70922A_02, Washington DC. Croppenstedt, A., L. Giovanni Bellu, F. Bresciani, and S. DiGiuseppez. 2007. Agricultural policy impact analysis with mul- timarket models: A primer. ESA Working Paper 07-26. Rome: Agriculture Economic Development Division, Food and Agriculture Organization (FAO) of the United Nations. Dorosh, P.A. and S. Haggblade. 1997. Shifting Sands: The Changing Case for Monetizing Project Food Aid in Bangla- desh. World Development. 25 (12): 2093-2104. Dorosh, P., H. Wang, L. You, and E. Schmidt. 2012. Road connectivity, population and crop production in Sub-Saharan Africa: Spatial correlations and analysis. Agricultural Economics 43 (1): 89-103. Fiddaman, T. 2007. Dynamics of climate policy. System Dynamics Review. 23: 21–34. 19 Food and Agriculture Organization of the United Nations (FAO). 1986. Highlands Reclamation Study: Ethiopia. Final Re- port. Volume I and Volume II. Rome: FAO. Ford, A. 1999. Modeling the Environment: An Introduction to System Dynamics Modeling of Environmental Systems. Is- land Press, Washington, DC. Ethiopian Development Research Institute (EDRI). 2009. Ethiopia: Input Output Table and Social Accounting Matrix for 2005/06. (In collaboration with the Institute of Development Studies at the University of Sussex.) Addis Ababa: EDRI. Gabre-Madhin, E. Z., 2001. Market institutions, transaction costs, and social capital in the Ethiopian grain market. IFPRI Research Report 124. International Food Policy Research Institute, Washington, DC. Goletti, F. and K. Rich. 1998. Policy simulation for agricultural diversification. Report prepared for the UNDP project on Strengthening Capacity Building for Rural Development in Viet Nam, Washington, D.C.: International Food Policy Research Institute. Guo H. C., L. Liu, G.H. Huang, G.A. Fuller, R. Zou, Y.Y. Yin. 2001. A system dynamics approach for regional environ- mental planning and management: A study for the Lake Erhai Basin. Journal of Environmental Management. 61: 93-111. Henderson, J. V., Z. Shalizi, A.J. Venables. 2001. Geography and Development. Journal of Economic Geography. 1: 81– 10. Hengsdijk, H, G.W. Meijerink, and M.E. Mosugu. 2005. Modeling the effect of three soil and water conservation practices in Tigray, Ethiopia. Agriculture, Ecosystems and Environment. 105: 29–40. Holden, ST, H. Lofgren, B. Shiferaw. 2005. Economic Reforms and Soil Degradation in the Ethiopian Highlands: A Micro CGE Model with Transaction Costs. Paper presented at the ECOMOD conference, Istanbul. Howard, J., E. Crawford, V. Kelly, M. Demeke, and J.J. Jeje. 2003. Promoting high-input maize technologies in Africa: The Sasakawa-Global 2000 experience in Ethiopia and Mozambique. Food Policy 28: 335–348. Legesse, D., C. Vallet-Coulomba, F. Gassea. 2003. Hydrological response of a catchment to climate and land use changes in Tropical Africa: case study South Central Ethiopia. Journal of Hydrology 275: 67-85. Lundberg, M. and K. Rich. 2002. Multimarket Models and Policy Analysis: An Application to Madagascar. Development Economics Research Group/Poverty Reduction Group, Environment and Infrastructure Team, mimeo. Washing- ton, D.C.: World Bank. Meybeck, M., P. Green, and C. Vörösmarty. 2001. A New Typology for Mountains and Other Relief Classes: An Applica- tion to Global Continental Water Resources and Population Distribution. Mountain Research and Development. 21 (1): 34–45. Minot, N., and F. Goletti. 2000. Rice Market Liberalization and Poverty in Vietnam. Research Report 114, Washington, D.C.: International Food Policy Research Institute. Minten, B., B. Koro, and D. Stifel. 2013a. The last mile(s) in modern input distribution: evidence from northwestern Ethio- pia. Ethiopia Strategy Support Program-2 Working Paper # 51. IFPRI – ESSP2. Addis Ababa, Ethiopia. Minten, B., S. Tamru, E. Engida, T. Kuma. 2013b. Using evidence in unraveling food supply chains in Ethiopia: the sup- ply chain of teff from major production areas to Addis Ababa. Ethiopia Strategy Support Program-2 Working Pa- per # 54. IFPRI – ESSP2. Addis Ababa, Ethiopia. Nyssen, J., W. Clymans, K. Descheemaeker, J. Poesen, I. Vandecasteele, M. Vanmaercke, A. Zenebe, M. Van Camp, M. Haile, N. Haregeweyn, J. Moeyersons, K. Martens, T. Gebreyohannes, J. Deckers, K. Walraevens. 2010. Im- pact of soil and water conservation measures on catchment hydrological response—a case in north Ethiopia. Hydrological Processes. 24(13): 1880–1895. Parsons, D., C.F. Nicholson, R.W. Blake, Q.M. Ketterings, L. Ramírez-Aviles, D.G. Fox, L.O. Tedeschif, J.H. Cherneyg. 2011. Development and evaluation of an integrated simulation model for assessing smallholder crop–livestock production in Yucatán, Mexico. Agricultural Systems. 104 (1): 1-12. Purnomo H., G. Mendoza. 2011. A system dynamics model for evaluating collaborative forest management: a case study in Indonesia. International Journal of Sustainable Development & World Ecology. 20 Schmidt, E. and F. Tadesse. 2013. Sustainable Agriculture in the Blue Nile Basin: Land and Watershed Management Practices in Ethiopia. Environment and Development Economics. forthcoming. Schmidt, E. and B. Zemadim. 2013. Hydrological modeling of sustainable land management interventions in the Mizewa watershed of the Blue Nile Basin. Ethiopia Strategy Support Program-2 Working Paper #61. IFPRI – ESSP2. Addis Ababa, Ethiopia. Shiferaw, B., and T.S. Holden. 1998. Resource Degradation and Adoption of Land Conservation Technologies in the Highlands of Ethiopia: A Case Study of Andit Tid, North-Shewa. Agricultural Economics 21: 53–67. Shiferaw, B., and T.S. Holden. 2001. Farm-level benefits to investments for mitigating land degradation: Empirical evi- dence for Ethiopia. Environment and Development Economics. 6: 336–359. Skorseth, K. and A. A. Selim. 2000. Gravel roads: Maintenance and Design Manual. U.S. Department of Transportation Federal Highway Administration. http://www.epa.gov/owow/nps/gravelman.pdf Sonneveld, B. G. J. S. 2002. Land Under pressure: The Impact of Water Erosion on Food Production in Ethiopia. PhD Thesis, Vrije Universiteit, Amsterdam, The Netherlands Sterman JD. 2000. Business dynamics: systems thinking and modeling for a complex world. Boston (MA): McGraw-Hill Higher Education. Stifel, D. and B. Minten. 2008. Isolation and Agricultural Productivity. Agricultural Economics. 39 (1): 1–15. Stifel, C., and J.-C. Randrianarisoa. 2004. Rice prices, Agricultural Input Subsidies, Transactions Costs and Seasonality: A Multi-Market Model Approach to Poverty and Social Impact Analysis for Madagascar. mimeo. Lafayette Col- lege, Easton, PA. Sutcliffe, J. P. 1993. Economic assessment of land degradation in the Ethiopian highlands: a case study. National Con- servation Strategy Secretariat. Addis Ababa: Ministry of Planning and Economic Development, Transitional Gov- ernment of Ethiopia. Tadesse, M., and K. Belay. 2004. Factors Influencing Adoption of Soil Conservation Measures in Southern Ethiopia: The Case of the Gununo Area. Journal of Agriculture and Rural Development in the Tropics and Subtropics 105 (1): 49–62. Tesfahunegn, G.B., L. Tamene, P.L.G. Vlek. 2011. Evaluation of soil quality identified by local farmers in Mai-Negus catchment, northern Ethiopia. Geoderma. 163(3):209-218. Vennix, J.A.M. 1996. Group Model Building: Facilitating Team Learning Using System Dynamics. John Wiley & Sons: New York, NY, USA. Vörösmarty C.J., B.M. Fekete, M. Meybeck, R.B. Lammers. 2000. The global system of rivers: Its role in organizing conti- nental landmass and defining land-to-ocean linkages. Global Biogeochemical Cycles 14: 599–622. Warren, K. 2006. Improving strategic management with the fundamental principles of systems dynamics. System Dy- namics Review 21: 329–350. Woldehanna, T. and A. Oskam. 2001. Income diversification and entry barriers: evidence from the Tigray region of north- ern Ethiopia. Food Policy 26: 351-365. World Food Programme. 2005. Report on the Cost-Benefit Analysis and Impact Evaluation of Soil and Water Conserva- tion and Forestry Measures. Managing Environmental Resources to Enable Transitions to More Sustainable Livelihoods (MERET) Project. Addis Ababa, Ethiopia: World Food Programme. Zemadim, B., M. McCartney, B Sharma, A. Wale. 2011. Integrated rainwater management strategies in the Blue Nile Ba- sin of the Ethiopian highlands. International Journal of Water Resources and Environmental Engineering. 3 (10): 220-232. http://www.epa.gov/owow/nps/gravelman.pdf http://www.researchgate.net/researcher/79224438_Lulseged_Tamene/ http://www.researchgate.net/researcher/874470_Paul_L_G_Vlek/ http://www.researchgate.net/publication/251729924_Evaluation_of_soil_quality_identified_by_local_farmers_in_Mai-Negus_catchment_northern_Ethiopia?ev=auth_pub http://www.researchgate.net/publication/251729924_Evaluation_of_soil_quality_identified_by_local_farmers_in_Mai-Negus_catchment_northern_Ethiopia?ev=auth_pub 21 APPENDICES Table A1—Benefits and costs of policy options to incentivize SLWM sustainability (5% discount rate) Simulation Household Consump- tion (NPV) Change in Household Consump- tion (NPV) Marginal Benefit of Govt. Invest- ment (NPV) Government Cost (NPV of cost) Marginal Benefit : Cost Base 78.70 0.00 Land Investments Terraces on middle and steep slopes (MST) 76.28 -2.42 - 0.00 - MST with government transfer 84.46 5.76 8.18 6.95 1.18 Fertilizer increase 97.81 19.11 - 0.00 - Fertilizer increase and MST 95.07 16.37 - 0.00 - Road Investment Scenarios (with Fertilizer and MST) Transport cost decrease 50% 97.41 18.71 2.34 1.74 1.34 Transport cost decrease 30% 96.41 17.71 1.34 0.74 1.81 Transport cost decrease 20% 95.94 17.24 0.87 0.74 1.18 Table A2—Area and yield of five major crops in highlands Ethiopia by slope Agricultural area by crop (millions of hectares) Steep Mid-slope Flat Teff 0.06 0.58 0.07 Barley 0.04 0.12 0.01 Wheat 0.00 0.00 0.00 Maize 0.05 0.49 0.09 Sorghum 0.00 0.00 0.00 *Steep is defined as agricultural plots on slopes >20 degrees, Mid-slope is 5-20 degrees, and Flat is less than 5 degrees. Yield by crop (tons / hectare) Steep Mid-slope Flat Teff 0.78 1.21 1.52 Barley 0.28 0.49 0.41 Wheat 0.00 0.92 0.54 Maize 1.78 2.91 2.15 Sorghum 0.00 1.44 0.00 INTERNATIONAL FOOD POLICY RESEARCH INSTITUTE 2033 K Street, NW | Washington, DC 20006-1002 USA T: +1.202.862.5600 | F: +1.202.457.4439 Skype: ifprihomeoffice | ifpri@cgiar.org | www.ifpri.org ETHIOPIAN DEVELOPMENT RESEARCH INSTITUTE P.O. Box 2479, Addis Ababa, Ethiopia T: +251.11.550.6066; +251.11.553.8633 | F: +251.11.550.5588 info@edri-eth.org | www.edri-eth.org IFPRI–ESSP ADDIS ABABA P.O. Box 5689, Addis Ababa, Ethiopia T: +251.11.617.2000 | F: +251.11.646.2318 mahlet.mekuria@cgiar.org | http://essp.ifpri.info The Ethiopia Strategy Support Program (ESSP) is financially supported by the United States Agency for International Development (USAID) and the Department for International Development (DFID) of the government of the United Kingdom and is undertaken as part of the CGIAR Research Program on Policies, Institutions, and Markets (PIM) led by the International Food Policy Research Institute (IFPRI). This publication has been prepared as an output of ESSP and has not been independently peer reviewed. Any opinions expressed here belong to the author(s) and do not necessarily reflect those of IFPRI, the Ethiopian Development Research Institute, USAID, DFID, PIM, or CGIAR. Copyright © 2014 International Food Policy Research Institute. All rights reserved. To obtain permission to republish, contact ifpri-copyright@cgiar.org. About ESSP II The Ethiopia Strategy Support Program II is an initiative to strengthen evidence-based policymaking in Ethiopia in the areas of rural and agricultural development. Facilitated by the International Food Policy Research Institute (IFPRI), ESSP II works closely with the government of Ethiopia, the Ethiopian Development Research Institute (EDRI), and other development partners to provide information relevant for the design and implementation of Ethiopia’s agricultural and rural development strategies. For more information, see http://www.ifpri.org/book-757/ourwork/program/ethiopia-strategy-support-program or http://essp.ifpri.info/ or http://www.edri-eth.org/. About these working papers The ESSP Working Papers contain preliminary material and research results from IFPRI and/or its partners in Ethiopia. The papers are not subject to a formal peer review. They are circulated in order to stimulate discussion and critical comment. The opinions are those of the authors and do not necessarily reflect those of their home institutions or supporting organizations. mailto:ifpri@cgiar.org http://www.ifpri.org/ mailto:info@edri-eth.org http://www.edri-eth.org/ mailto:mahlet.mekuria@cgiar.org http://essp.ifpri.info/